The Question: Is Peace a Delusion?

“Every gun that is made, every warship launched, every rocket fired signifies, in the final sense, a theft from those who hunger and are not fed, those who are cold and are not clothed.”

Dwight D. Eisenhower, 1953. A man who, having spent his career building the machine, understood precisely what it cost

In the spring of 2023, I was on a project call with my team. Half of them were in Kyiv. The other half were scattered across London, Chennai, and Malaga. It was an unremarkable Tuesday. Except that it was not, because a year had passed sine Russia invaded Ukraine and my Kyiv colleagues were doing what people do when their city has just been bombed. Working, because working was one of the things that still made sense. One of them, a quietly formidable analyst named Oksana, apologised for the intermittent audio. She had moved to her mother’s flat on the western side of the city. The internet was more reliable there, she explained. She did not mention the other reasons.

I thought about Oksana last week when I opened my investment portfolio. It had declined by ~12% since the US and Israel bombed Iran. In normal times this would be a stern conversation with my fund manager.

These two experiences. The call with Oksana, the portfolio on a bad day are connected. The same system that makes war structurally convenient for certain powerful actors also makes it financially legible to investors. A spike in defence stocks, a sell-off in regional equities, a flight to gold. Markets, it turns out, have priced in forever wars. They have built them into the model.

Do we have to?

Humanity is not uniquely wicked nor do shadowy cabals orchestrate conflict from underground bunkers. Though the bunkers are perfectly real; Raytheon’s revenue was $68 billion in 2023, and I am told they build very good ones. The argument that the economic, political, and military systems we have constructed over the past two centuries are architectures with embedded incentives is both simple and troubling. Those incentives make sustained, universal peace structurally inconvenient for the most powerful actors within them.

Humanity spends roughly $2.5 trillion on defence annually. Money that could theoretically fund the end of global hunger and energy poverty roughly 6 to 8 times over every single year. This figure has risen every year for a decade. Our planet simultaneously possesses the agricultural capacity to feed all eight billion of its inhabitants, the medical knowledge to prevent most of its leading causes of death, and the renewable technology to power itself without burning anything.

An alien civilisation observing from a safe distance would conclude that the dominant species of this planet has made a choice. Not a conscious one, perhaps. But a structural one.

In this essay I attempt to define that structure, trace its origins, map its geography, and argue that it can be changed. Not easily. Not soon. But deliberately, if enough people understand what they are actually dealing with.

That requires, first, a frank look at the machine.

The Machine: How Modern Capitalism Built Institutional Actors Whose Prosperity Depends on Dysfunction

Three American asset management firms are collectively the largest shareholder in approximately 90 per cent of S&P 500 companies. They are BlackRock, Vanguard, and State Street. Think Raytheon, Lockheed Martin, Boeing, Northrop Grumman, Pfizer, Johnson & Johnson, Exxon, and Chevron. When you own the entire playing field, you’re not interested in any player winning. You’re interested in the game continuing indefinitely.

This is not sinister in the way conspiracy theories require sinister to be, like villains in rooms making decisions. Systems are sinister in a more enduring way: automatically, invisibly, through the aggregation of individually rational decisions that lead to collectively catastrophic outcomes. No one at BlackRock is cackling about perpetual war. They are optimising a portfolio. The portfolio happens to include the companies that profit from it.

Nor is this purely an American pathology. The City of London. The square mile of ancient privilege and modern financial engineering has for centuries specialised in the profitable management of other people’s conflicts. British arms manufacturers like BAE Systems, the world’s seventh-largest defence contractor with annual revenues of around £23 billion, have sold equipment to both sides of conflicts with the impartiality of a well-mannered arms dealer. BAE has supplied fighter jets to Saudi Arabia throughout its bombardment of Yemen. A campaign that has caused hundreds of thousands of deaths and led to one of the worst humanitarian crises in modern history. The UK government has approved over £23 billion in arms export licences to Saudi Arabia since the Yemen war began in 2015. Parliamentary challenges to these sales have been consistently defeated, suggesting a deliberate decision.

The Western liberal capitalist model, prevalent in the US, UK, EU, and their dependencies, has created a three-layered Dysfunction Preservation System. It’s not a conspiracy, but a system.

Layer One is financial capital. Layer Two is the lobbying and political donation architecture that transmits capital’s preferences into policy. Layer Three is what is loosely and often hysterically called the ‘deep state’, better understood as institutional inertia: the tendency of large bureaucracies to optimise for continuity over change, because continuity keeps it funded, staffed, and relevant.

Consider the lobbying architecture.

In the United States, the defence industry spends approximately $110 -$150 million annually on lobbying Congress.

In the United Kingdom, the system is technically cleaner. British MPs are not supposed to accept direct payments from defence companies. But the revolving door rotates with identical efficiency. The Ministry of Defence is staffed by officials who came from BAE Systems, Rolls-Royce, and who will return to them. The pipeline flows in both directions with such regularity it has ceased to be newsworthy. This is a form of institutional capture.

Germany’s version of this story was most vivid and catastrophic. For nearly two decades, successive German governments, led by politicians from both the centre-left SPD and the centre-right CDU, built an energy system of extraordinary dependence on Russian natural gas. Nord Stream 1 was completed in 2011. Nord Stream 2. A second pipeline, running directly under the Baltic Sea to Germany, was completed in 2021, despite warnings from the US, protests from Ukraine, and Russia’s annexation of Crimea in 2014. The pipeline project championed by the former German Chancellor Gerhard Schröder subsequently joined the board of the Russian state energy company Rosneft and accepted a seat on the supervisory board of Gazprom. He has not returned his Russian salary. Germany’s energy dependency did not happen by accident. Politicians and corporate interests, with shared incentives, built the system. When Russia invaded Ukraine in 2022, Germany couldn’t respond decisively, despite its rhetoric. You cannot sanction the country that heats your homes.

This money and these relationships primarily buy access, agenda-setting, and the removal of disruptive alternatives from serious political consideration. Meaningful defence spending reform, real antitrust enforcement in the energy sector, and a pharmaceutical system focused on prevention become structurally impossible to implement regardless of electoral outcomes. Prime Ministers come and go, but the revolving door keeps spinning.

The ‘deep state’ myth, popular among political movements on both sides of the Atlantic, is not entirely inaccurate, but its portrayal is exaggerated. There is no coordinated shadow government. Instead, there are permanent institutions with their own cultures, career incentives, and operational doctrines that persist across administrations and actively resist radical structural change that could threaten their budgets, relevance, or worldview. These include MI6, the MoD, the Foreign Office, the Federal Reserve, the ECB, the Bank of England, and major regulatory agencies. As a result, genuinely transformative policy is almost impossible to implement, even when elected governments want it, as they rarely do due to funding.

The processed food industry is a prime example of the system at work. What does a prepackaged food company want? It has systematically eroded the conditions that made home cooking natural, creating a world where the extended family, unhurried afternoons, and proximity to fresh markets are no longer as common. Remember these. The food company didn’t cause urbanisation or the sixty-hour working week. But it lobbied vigorously against nutritional labelling reform, funded research that muddied dietary science for decades (the sugar industry’s corruption of cardiovascular research in the 1960s is now thoroughly documented), and in the United Kingdom successfully resisted meaningful action on junk food marketing to children for the better part of thirty years. The dysfunction was not designed. It was exploited, maintained, and quietly subsidised.

This template is the system’s plan. It doesn’t need to plan dysfunction; it only needs to prevent its resolution.

The Doctrine: How Geopolitical Architecture Became a Tool of Resource Extraction

On December 2, 1823, President James Monroe addressed Congress and made a clear anti-colonial statement. Europe should refrain from interfering in the Americas, and the United States should avoid interfering in Europe. The two hemispheres should be separate, with distinct spheres of influence, and all nations should return to their respective homes.

Historians believe that Monroe didn’t fully grasp the importance of his words. John Quincy Adams, his Secretary of State, was the intellectual architect of the Monroe Doctrine. He was cleverer and more calculating than his predecessor. However, the practical enforcement of the doctrine in its early years was provided by the British Royal Navy, not the US Navy, due to the latter’s limited size. The British had commercial reasons for favouring an independent Latin America over a Spanish colonial one. The original Monroe Doctrine was largely a bluff backed by the British fleet. The British preferred trade over territory, and an independent Latin America was beneficial to their trade interests. Anti-colonialism can be an elegant position when it serves your commercial interests.

By 1904, the United States had a stronger navy and a President extremely keen to use it. Theodore Roosevelt. Soldier, hunter, imperialist, and relentless self-publicist added the Roosevelt Corollary to Monroe’s original declaration. A debt crisis provided the perfect foil. British, German, and Italian gunboats had blockaded Venezuelan ports in 1902 to recover debts owed by the Venezuelan government to foreign bondholders. Roosevelt, haunted by nightmares of European naval bases in the Caribbean, threatening his Panama Canal project and American regional dominance, declared the US had the right and duty to intervene in Latin America to prevent such meddling.

The corollary confidently stated that ‘chronic wrongdoing’, specifically defaulting on debts to American and European banks, would necessitate ‘intervention by a civilised nation’. Roosevelt didn’t specify which nation he meant. As a result, the Monroe Doctrine, initially an anti-colonial declaration, became the legal foundation for American imperialism in the Americas. Over the next century, the US sent troops into Latin America over forty times, with US Marines becoming known as ‘State Department Troops’. The Panama Canal was built, controlled, and profitable, and the debts were collected.

Meanwhile, Europe was doing its own version of the same thing, with the extra problem that the colonial powers were often fighting each other over the territories they were exploiting. The Sykes-Picot Agreement of 1916. In a secret meeting between Britain’s Mark Sykes and France’s François Georges-Picot, with Russian approval, the Ottoman Empire’s Arab territories were divided between the two countries using a ruler and a map, disregarding the ethnic, tribal, religious, and cultural realities on the ground. These borders, with some modifications, are the borders of modern West Asia. The conflicts that arise from these borders are, in essence, the legacy of a 1916 lunch meeting between two diplomats who had never visited most of the territory they were dividing.

This is the master grammar of modern geopolitical conflict, and it operates in both directions of the Atlantic.

The trick and it is a very good trick indeed, is that the wrapper is always sincere. The people waving the flag genuinely believe in the flag. The people profiting from the war are merely grateful that they do.

The Laboratory: West Asia (aka The Middle East) as the World’s Most Expensive Proof of Concept

If you wanted to design a regional theatre that perfectly demonstrated the relationship between great power resource interests, proxy conflict architecture, and the systematic production of the very enemies you are fighting, head to modern West Asia. Over the past century, it has been subjected to British colonial carve-ups, American Cold War proxy battles, Israeli security operations that consistently produce more dangerous successors, Saudi cultural warfare that radicalises communities, and Iranian strategic patience that turns every Western military intervention into a gift. It is a laboratory, but unfortunately, the people of the region are the experimental subjects.

The British origins of this theatre deserve emphasis, because they are too often excused from the story by commentators who begin their analysis of West Asian instability at a conveniently post-imperial date. Britain drew the borders of the modern West Asia. It also made multiple, contradictory, incompatible promises to every significant party in the region simultaneously. And then managed the consequences with a military and diplomatic combination, quite impressive were it not quite so catastrophic.

To the Hashemites, Britain promised an independent Arab kingdom for launching the Arab Revolt against the Ottomans in 1916. To the French, it promised to share the Arab territories under Sykes-Picot. To the Zionist movement, it issued the 1917 Balfour Declaration promising ‘a national home for the Jewish people’ in Palestine. A territory it had already effectively promised to the Arabs. The subsequent British Mandate in Palestine attempted to honour all three commitments simultaneously. Roughly equivalent to promising the same house to three different families and then expressing surprise when they argue about the furniture.

Britain assembled Iraq from the three Ottoman provinces of Mosul, Baghdad, and Basra. Combining Kurds, Sunni Arabs, and Shia Arabs under a Hashemite king from the Hejaz, who had never visited the country he was to rule. This Iraq required constant authoritarian oversight to prevent fragmenting along the fault lines Britain had chosen to ignore. When Saddam Hussein provided that management, Britain and America were content to supply him with weapons, intelligence, and diplomatic cover. Most notoriously, during his use of chemical weapons against Kurdish civilians at Halabja in 1988. When he became inconvenient, they removed him, expressing subsequent surprise at the chemical weapon fuelled sectarian civil war that filled the vacuum.

Begin with Lebanon specifically, and the sequence of events that produced Hezbollah. Currently the world’s most heavily armed non-state military force, and a creation that nobody planned, everybody contributed to, and which has since made the entire region considerably more dangerous than the problems it was notionally created to solve.

After Jordan’s expulsion in 1970–71, the Palestine Liberation Organisation (PLO) relocated to Lebanon and used its southern territories as a base for operations against Israel. In March 1978 a Palestinian coastal road attack killed 34 Israelis. Israel launched Operation Litani, invading southern Lebanon with 25,000 troops. The stated aim was to push the PLO north of the Litani River. The outcome: destruction of 82 Lebanese villages, more than 200,000 refugees, and mass displacement of Lebanon’s Shia population into the already overcrowded slums of Beirut. These displaced, impoverished, politically marginalised Shia communities would later form the recruitment pool for Hezbollah. The 1978 invasion, designed to solve the PLO problem, created the human substrate for a far more durable successor.

The 1982 invasion was more ambitious and more revealing. The official “just cause” was the attempted assassination of Israel’s ambassador to Britain by gunmen from the Abu Nidal Organisation. A group that was, ironically, at war with the PLO it was used to justify attacking. Israeli Prime Minister Menachem Begin blamed the PLO, while Defence Minister Ariel Sharon had been planning the invasion since August 1981 and said so publicly. The strategic objective was to destroy the PLO’s military infrastructure in Lebanon, install a pro-Israeli Maronite Christian government in Beirut, and reshape the region’s political geography in Israel’s favour.

In reality: the PLO was expelled to Tunisia, effectively removing it from the immediate theatre. Israel’s chosen Lebanese president, Bachir Gemayel, was assassinated before he could take office. The subsequent Sabra and Shatila massacre destroyed whatever moral legitimacy the operation had retained.

And from the ruins of this particular geopolitical adventure, funded and armed by the United States to the tune of nearly $218 million in military equipment in the first quarter of 1982 alone, emerged Hezbollah.

The Houthi story in Yemen is even more comic and ironic. The Houthis, a branch of Shia Islam from northern Yemen, were displaced by a republican revolution in 1962. Egypt backed the revolution, while Saudi Arabia supported the displaced Zaydi monarchy. The Saudis armed the Zaydi tribes, including the Houthi families, to fight against Egyptian influence in the region.

Saudi Arabia, using the Zaydis as convenient proxies, systematically erased their religious and cultural identity over the following decades. From the early 1980s, the Saudi government financed Wahhabi seminaries in the Zaydi heartland, training clerics to convert the community to Wahhabi Sunni Islam.Faced with this existential threat, the Zaydi community reacted with the force of any community that believes its survival is at stake. The Houthi movement emerged as a Zaydi cultural revival, political movement, military insurgency, Iranian-aligned military force, and the most effective Yemeni fighting force in modern history after Saudi Arabia launched a bombing campaign against it in 2015.

Saudi Arabia created its own enemy. It armed the tribes, marginalised them, culturally invaded them, and then bombed them. Createing a conflict that has killed an estimated 377,000 people and given Iran a dagger pointed directly at the Bab-el-Mandeb strait, through which roughly 10 per cent of global trade passes. The Houthi attacks on Red Sea shipping since 2023 have disrupted global supply chains, increased shipping costs, and forced vessels to reroute around the Cape of Good Hope. Adding two weeks and significant cost to every journey. The consequences land, eventually, in the prices that ordinary people pay for ordinary things in ordinary supermarkets in Birmingham and Bristol and Bordeaux.

The Iraqi Shia militia story completes the trilogy. In 2003, the US invaded Iraq with British support, including 45,000 troops. The intelligence dossier, which was later found to be unreliable, claimed Iraq had weapons of mass destruction. Saddam Hussein, who had been the main counterbalance to Iranian regional ambitions, was removed from power. The Coalition Provisional Authority then disbanded the Iraqi Army, putting 400,000 armed, trained, and now unemployed men onto the streets. Iran moved with extraordinary speed. Within months, Iranian-backed militia groups formed across Iraq. The US and Britain had spent $2 trillion sacrificing thousands of lives to hand Tehran the dominant position in Iraqi politics. A feat Iran failed to achieve in eight years of conventional warfare during the Iran-Iraq War. The Chilcot Report published after a seven-year inquiry in 2016 concluded diplomatically that the decision to invade had been made ‘before the peaceful options for disarmament had been exhausted’. This sort of sentence might be described as damning.

The pattern, across all these cases, is identical.

A military intervention designed to solve a security problem produces a power vacuum. A patient, strategically sophisticated regional actor fills the vacuum. The resulting force is more dangerous, more durable, and more difficult to address than the original problem.

Is this serendipity OR is it a system producing its predictable outputs at every stage? Defence contractors file quarterly results, reconstruction contracts are awarded, and the conditions for the next intervention are quietly assembled.

The Models: Different Systems, Same Grammar

One instructive feature of the modern world is its diverse political-economic systems. Western liberal capitalism, Gulf sovereign capitalism, Russian patrimonial state capitalism, Chinese developmental state capitalism. All express the same underlying resource logic through different institutional architectures. The pathologies differ. The grammar is identical.

The Gulf model. Operating in Saudi Arabia, the UAE, Qatar, Kuwait, and Bahrain is sovereign capitalism. A system in which the distinction between state, ruling family, and economy is blurred, and private accumulation is permitted provided it does not challenge political authority. Saudi Aramco is a publicly listed instrument of Saudi state power. This eliminates the lobbying dysfunction of the Western model. There is no need to lobby the government when you are the government. But replaces it with the complete absence of accountability for catastrophic decisions. The Yemen bombing campaign, launched in 2015 on the assumption it would last ‘a matter of days’, has continued for a decade. There was no parliament to challenge it, no independent press to sustain scrutiny, no electoral consequence for its architects.

Russia presents patrimonial state capitalism. A system in which wealth is a reward for political loyalty rather than an independent power base. Mikhail Khodorkovsky, once Russia’s richest man, discovered the limits of this arrangement in 2003 when he began financing political opposition. He spent ten years in prison. The lesson was not lost on his colleagues. In this model, war serves primarily as a domestic legitimacy tool, though the resource logic underneath is identical to any other system’s resource calculus. Control of energy infrastructure, pipeline geopolitics, buffer territory. The invasion of Ukraine is dressed in the language of deNATOfication and historical reunification. Its content is power, territory, and gas.

Dependency is a form of vulnerability and that vulnerability is eventually exploited. Several generations of European policymakers had chosen not to learn, despite it being rather clearly demonstrated by every decade of the twentieth century.

Germany is a perfect case study. She outsourced her energy sovereignty for the sake of cheap gas and the profits of its industrial sector. It subsequently discovered that sovereignty, once outsourced, is not easily retrieved. For two decades, Berlin built an energy dependency on Russia, believing trade relationships were a substitute for strategic thinking. When Russia invaded Ukraine, Germany’s first instinct was to continue gas payments because the alternative was turning off its own heating. In effect, fund the invasion of Ukraine. The economic cost of the energy transition following Nord Stream’s destruction was €500 billion to €800 billion on excess energy costs between late 2021 and 2022 alone.

China is the most intellectually interesting case, because it has achieved something that western economic theory spent decades insisting was impossible. Rapid economic development, 800 million people lifted from poverty in four decades, and technological advancement, all within a political system that maintains Leninist party control over capital rather than allowing capital to capture the party. The Chinese model is developmental state capitalism. The party sets strategic direction, private capital operates within that direction, and those who challenge the boundaries are dispatched on a sabbatical. Jack Ma’s $34 billion Ant Group IPO was cancelled in November 2020 after a speech mildly critical of Chinese financial regulators. He subsequently took an extended voluntary break from public life.

China’s Belt and Road Initiative. Infrastructure investment across 140 countries is the most explicit expression of the ‘trade interdependency is the structural foundation of peace’ thesis.

This is exactly the logic that underpinned European economic integration after 1945. The theory that countries sufficiently entangled in each other’s prosperity will find war prohibitively expensive. It worked tolerably well for Western Europe. The Franco-German relationship. Two countries that had fought three catastrophic wars between 1870 and 1945 was stabilised by fifty years of economic integration to a degree that no treaty or military alliance could have achieved alone. The European Union, for all its considerable bureaucracy, is the most successful peace project in modern history. It is worth stating this plainly, particularly as the political forces most committed to dismantling it are often the same forces most animated by the prospect of European conflict.

The conclusion that emerges from comparing these models is: no existing political-economic system is architecturally aligned with peace as a primary output. The Western model is captured by institutional actors whose prosperity depends on dysfunction.The Gulf model achieves stability through accountability’s absence. The Russian model uses conflict as a domestic legitimacy tool. The Chinese model has genuine peace incentives in its economic architecture but territorial revisionist ambitions in its political one. Each has a different pathology. All are running the same underlying code: control resources, maintain dominance, define security in terms of relative rather than absolute advantage.

The Resource Map: The Coming Conflicts Are Already Visible

The resource extraction thesis makes the future legible. If modern conflict is fundamentally about the control of energy, water, trade routes, and the critical materials that underpin economic and military power, then the next generation of wars are not mysterious. They are already structured. They are already funded. In several cases they have already begun. Conducted, for the moment, with treaty filings and economic coercion rather than missiles. The missiles come later.

The South China Sea is the immediately visible theatre. China claims sovereignty over virtually the entire sea.

The 2016 international arbitration ruling against China’s claims was rejected by Beijing. The South China Sea is the Monroe Doctrine replayed in Asia, with China as the rising power and the United States, for the first time in a century, being challenged in someone else’s backyard.

The Arctic is the conflict that has not yet begun but whose shape is perfectly clear. Climate change is melting Arctic sea ice, opening new shipping lanes and exposing an estimated 30 per cent of the world’s undiscovered natural gas reserves and 13 per cent of its undiscovered oil. Russia has been aggressively militarising the region since 2014. The United States, Canada, Norway, and Denmark are staking competing territorial claims. Norway and Russia share the longest Arctic border of any two countries. The lawyers are very busy. The admirals are watching.

Britain has its own stake in this. The United Kingdom is a member of the Arctic Council and maintains territorial interests through its Overseas Territory of South Georgia, its relationship with the Falkland Islands, and its position as a major Atlantic maritime power.

The geopolitical implications of Arctic ice melt are directly linked to the geopolitical stability of the Gulf. Another example of the resource map’s interconnections.

Water deserves more attention than it receives in mainstream conflict analysis. It is more fundamental than energy and less amenable to technological substitution. You can build a solar panel. You cannot synthesise a river. The Grand Ethiopian Renaissance Dam (GRED) the largest dam in Africa continues to be a source of acute tension between Ethiopia, Egypt, and Sudan since 2011. Egypt depends on the Nile for 97% of its fresh water. Hence a constitutional provision declaring it a matter of national security backed by military options. GRED is a conflict between a country’s right to electricity and another country’s right to water. There is no clean resolution. There will be consequences.

Europe is not immune to water politics. The Danube River. Shared by nineteen countries across a basin that covers 8% of the European continent. It is already under climate stress. Alpine glaciers, which feed several of Europe’s major rivers including the Rhine, are retreating at rates that will reduce summer river flows by 20 to 30% by 2050 per current modelling. The Rhine is the arterial waterway of Germany’s industrial heartland. A 20% reduction in its navigable flow is an economic and political crisis in slow motion.

Rare earth minerals and critical supply chains represent the newest iteration of resource dependency politics. The Democratic Republic of Congo (DRC) supplies approximately 70% of the world’s cobalt, an essential component of the lithium-ion batteries that power electric vehicles and grid-scale energy storage. The DRC has been in a state of near-permanent conflict since the 1990s, with a death toll estimated at between 5-6 million people. The connection between cobalt demand, militia financing, and the artisanal mining sector is direct and documented. The electric vehicle sitting in a London driveway has a Congolese conflict embedded in its battery.

Into this landscape of competing resource interests sits India. The world’s most populous nation, its largest democracy, and perhaps the most strategically sophisticated practitioner of what it calls ‘strategic autonomy’. India imports roughly 85% of its oil, sources its defence equipment from Russia, Israel, France, and the United States simultaneously, maintains trade relationships with Iran, navigates a contested border with China, and has declined to join western sanctions on Russia despite considerable pressure to do so. India’s refusal was widely described in the Western press as moral failure. More accurately energy dependency is political dependency, and the only reliable foundation for genuine political independence is the ability to meet your own energy needs without a single patron’s permission.

The Human Constant: Why This Isn’t New And Why That Matters

At some point it is necessary to confront the uncomfortable possibility that the structures are not the deepest cause of conflicts. Underneath the military-industrial complex, the lobbying architecture, the resource competition, and the geopolitical chess, there is something older and less tractable. The human impulse to dominate, accumulate, and expand. An impulse that predates capitalism by several thousand years and expressed in every political system that has existed.

Mahmud of Ghazni raided the Indian subcontinent seventeen times between 997 and 1030 AD. He was not motivated by a shareholder return requirement. He was motivated by the fact that the great Hindu temples of northern India were essentially the banks of medieval South Asian civilisation. The religious wrapper was sincere enough. The content was gold. He came seventeen times because each time it worked.

Alexander the Great is the philosophically interesting case. Alexander conquered the Persian Empire not primarily for its resources, though resources followed. He conquered it because his tutor was Aristotle, his role model was Achilles, and his psychological architecture was organised around the Greek concept of kleos — ‘eternal glory through deeds that outlast the man.’ When his generals refused to march further east, at the banks of the Hyphasis River in what is now Pakistan, Alexander wept. Not for territory. Not for resources. For the limits of the possible. An attempt to resolve the terror of insignificance through the accumulation of dominion.

The British Empire deserves honest examination rather than the nostalgia that occasionally clouds it.

At its peak, the Empire governed a quarter of the Earth’s land surface and approximately 412 million people. It was a commercial enterprise, a strategic project, and a civilisational ideology so thoroughly internalised by its administrators that many of them genuinely believed they were doing the colonies a favour. The East India Company. A private corporation that governed India for nearly a century is history’s most dramatic example of the fusion of commercial interest and military power. It had its own army, its own foreign policy, and its own courts. It was, in a very real sense, BlackRock with bombs and muskets.

What changes across history is not the impulse but the instruments and the scale. The ancient conquerors were at least honest about their violence. The modern system has developed the management of conflict as a recurring revenue stream rather than a one-time conquest. You do not need to occupy a country to extract its resources. You need to ensure that its government is sufficiently indebted to your financial institutions, sufficiently dependent on your military technology, and sufficiently integrated into your trade architecture. The alternative, genuine independence is economically catastrophic. This is empire without the administrative burden. It is, from a purely commercial perspective, a significant improvement.

The military-industrial complex does not want to conquer Iran. It wants an Iran threat sufficiently credible to justify a $50 billion missile defence procurement cycle. It does not want to occupy Venezuela. It wants a Venezuelan government sufficiently destabilised that its oil infrastructure remains accessible on favourable terms. The ancient kings needed territory. The modern institutional actors need only the conditions that make dependency rational. It is a more sinister innovation than Alexander’s kleos. Alexander at least had the courage or the madness to ride at the front of his cavalry.

And yet.

The human impulse toward domination, for all its catastrophic expressions, built every cathedral, composed every symphony, mapped every coastline, and sent the first humans to the moon. The terror of insignificance is the same force that drives every scientist to ask a question nobody has asked before, every artist to attempt a form nobody has attempted. The question is not how to eliminate the impulse. The question is what institutional architecture channels it toward creation rather than destruction.

That is the question that has defeated every utopian project in history. It is also the only question worth asking.

The Architecture of Peace: What Would a System Actually Built for Peace Look Like?

Let us be honest. It is difficult.

The changes required to build a world structured around peace, health, and human flourishing rather than resource extraction and managed dysfunction are institutional operating models.

We largely have the technology. Solar panels work. Desalination plants work. Agricultural science has eliminated the caloric basis for famine.

The challenge is that the tools threaten the business models of the institutions that control the systems those tools would replace. These institutions have invested a lot of money to make sure that the political systems that could force their adoption are sympathetic to the argument that it’s not the right time.

The first and most fundamental reform required is in how we measure economic success. Gross Domestic Product. The dominant metric of national performance since its invention by Simon Kuznets in 1934, who immediately warned that it should not be used as a measure of welfare goes up when there is a war. It goes up when there is a natural disaster that requires reconstruction. It goes up when a community that previously grew its own food begins buying it from a supermarket. It is a measure of economic activity rather than human flourishing. Optimising for it produces predictable results. Activity over wellbeing, throughput over sustainability, and growth over resilience. Scotland, New Zealand, Iceland, Wales, and Finland have been experimenting with wellbeing economy frameworks that measure health outcomes, environmental sustainability, and social cohesion alongside economic activity. These experiments are promising, politically fragile, and vigorously resisted by the interests that benefit from GDP optimisation.

The second reform is energy sovereignty. For the first time in the history of energy, the technology exists to generate power at the point of consumption rather than extracting it from concentrated deposits that must be controlled, transported, and defended. Solar panels and wind turbines cannot be embargoed. They do not require a Strait of Hormuz. A village with a solar microgrid is energy sovereign in a way that no oil-importing nation has ever been. The renewable transition, properly structured is the first genuine opportunity in the industrial era to decouple political sovereignty from energy dependency. Around local generation, distributed storage, and community ownership rather than the green-washing of centralised fossil fuel monopolies.

The critical caveat is that the transition as currently structured is not building energy sovereignty. It is building new dependencies. Solar panels are manufactured overwhelmingly in China. Battery supply chains run through the Democratic Republic of Congo and Chile. Rare earth processing is concentrated in Chinese facilities. A Europe that replaces Russian gas dependency with Chinese solar panel dependency has not achieved energy sovereignty. It has changed creditors. A genuinely peace-positive energy transition would invest in domestic manufacturing capacity, diversified supply chains, and community-owned generation. It must resist the pressure from large energy corporations to replicate, under renewable branding, the centralised extraction models that made fossil fuel geopolitics so lethal.

The third reform concerns food sovereignty. A nation that cannot feed itself cannot set its own foreign policy. The movement toward food sovereignty is a political project. Local agriculture, fresh markets, cooking education, meaningful nutritional policy must graduate beyond a lifestyle preference. The Common Agricultural Policy was at its core an attempt to ensure that Europe could feed itself. A lesson drawn directly from the experience of two World Wars in which food supply proved as decisive as artillery.

The fourth reform is the redesign of capital accountability structures to extend the time horizon of economic decision-making beyond the quarterly earnings cycle. The most destructive feature of the financial capitalism that dominates the Western world is not its greed but its time horizon. Greed by the way is unremarkable and universal. The defence industry does not invest in peace because peace does not appear on a quarterly earnings call. The food industry does not invest in nutrition because malnutrition does not appear on a quarterly earnings call. The fossil fuel industry spent four decades funding climate denial because the costs of climate change appeared on a 50-year horizon and the profits of climate denial appeared on a quarterly one.

Patient capital operates on different time scales and produces different decisions. Sovereign wealth funds, long-horizon institutional investors, public development banks, and cooperatively owned enterprises. The Norwegian Government Pension Fund, the world’s largest sovereign wealth fund at over $1.7 trillion, has divested from tobacco companies and coal producers, incorporated climate risk into its investment framework, and used its ownership stakes to push corporate governance reforms. It demonstrates that capital can be structured to operate on time horizons that align with human wellbeing rather than quarterly performance metrics, and that the political will to do so is achievable. Although it helps considerably to have a century’s worth of oil revenue to work with.

The fifth reform is international institutional architecture. The United Nations Security Council was designed in 1945 for a world that no longer exists. One in which five victorious powers would jointly manage a fragile post-war order. Those five permanent members each hold a veto, which means that any conflict in which any of them is directly or indirectly involved can be blocked from meaningful collective response. The Russian invasion of Ukraine, China’s activities in the South China Sea, American military interventions in West Asia, British military action in Iraq. None of these can be meaningfully addressed by the Security Council because the perpetrators can block the Council’s action. The institution designed to prevent great power conflict is constitutionally incapable of restraining great powers. This is a design feature, installed by the great powers in 1945. Reforming it requires the consent of those who benefit most from its current architecture, which is, admittedly, a significant obstacle.

The sixth reform is decentralisation of energy, food, manufacturing, and political decision-making. The concentration of essential resources and capabilities in centralised systems creates the dependencies that make resource-based conflict rational. A world in which communities, regions, and nations can meet their basic needs without external dependency is a world in which the leverage that enables resource-based coercion does not exist. Decentralisation is not autarky. Global trade is genuinely beneficial. The interconnection of economies has been one of the stronger inhibitors of interstate conflict. But there is a crucial difference between trade in goods and services that communities choose to exchange because of genuine comparative advantage, and dependency on external suppliers for basic survival requirements. The former enriches. The latter controls.

The Question That Cannot Be Deferred

Let me return, at the end, to where this began: to a Tuesday morning in 2023, and Oksana apologising for intermittent audio from a city that was being bombed. And to a portfolio statement from last month, 12% down since the beginning of a conflict in which I have no stake and no say, in a region whose instability I am financing, indirectly, through a hundred small decisions about where to invest and what to buy and which companies to trust with my money.

These connections between my portfolio and the conflict, between the conflict and the architecture that sustains it, between that architecture and the decisions made in 1916 over a map of the Middle East and in 1904 over a debt crisis in Venezuela and in 1982 over a plan Ariel Sharon had been preparing for a year are structural, not metaphorical. The financial system has priced in forever wars. It has incorporated them as a variable. Should we, as citizens, investors, voters and human beings with some residual interest in our planet’s continued habitability, have done the same. And if so, whether we are comfortable with that.

To alien observers, patiently watching the full earth story unfold:

The Monroe Doctrine and its forty military corollaries; the Sykes-Picot lines drawn over lunch; the Balfour Declaration’s incompatible promises; the 1982 Lebanon invasion and the Hezbollah it inadvertently created; the Saudi Wahhabi programme that radicalised the Zaydi tribes it had previously armed; the Iraq invasion that handed Iran the regional dominance it had failed to achieve in eight years of conventional war; Germany’s decision to build its energy policy around the assumption that Vladimir Putin’s interests were reliably commercial; the renewable transition that is building new dependencies even as it dismantles old ones.

Their conclusion, one imagines, would be something like this: the species is not uniquely wicked. It is architecturally misaligned. Its economic systems reward extraction over sustainability, dependency over sovereignty, conflict over resolution. Its political systems are captured by the interests those economic systems generate. Its international institutions are designed by the powerful to preserve the advantages of the powerful. It’s ideological frameworks. Democracy, freedom, civilisational values, the rules-based international order are sincere in their expression and systematically deployed in the service of material interests. The people waving the flags believe in the flags. The people profiting from the conflicts are grateful that they do.

And yet.

The alien observers would also note: the information environment that sustained the monopolies on which war has historically depended is fragmenting in ways that make large-scale popular consent for military adventure harder to manufacture and easier to withdraw. The 2003 Iraq War launched on falsified intelligence, supported by a British government that silenced its own legal advisers, and prosecuted with a catastrophic absence of post-conflict planning could not be replicated today with the same ease. The Chilcot Report, published thirteen years late, ensures that. The economic interdependency that the liberal international order built, for all its asymmetries, has made the direct costs of great power conflict visible in ways that the rulers of 1914 could not have imagined. The renewable energy transition, however imperfectly structured, is creating the technical possibility for the first time in the industrial era of genuine energy sovereignty at the community and national level.

These are not sufficient. They are necessary.

Peace is not the natural equilibrium of the systems we have built. It never has been and it will not become so without deliberate intervention. The interests that benefit from the current system will resist that intervention with every instrument available to them. They are very good at this, its their muscle memory.

But the deepest history lesson is also, paradoxically, its most hopeful one. Every system that humanity has built. The slave economies, the feudal orders, the colonial empires, the apartheid states was, at the time of its operation, described by its beneficiaries as natural, inevitable, and permanent. None of them were. All of them ended not through the spontaneous moral awakening of those who benefited from them, but through the organised political agency of those who did not.

The question is not whether peace is possible. It is whether the political will to build its architecture can be generated before the current system produces a catastrophe large enough to foreclose the option. That is the question that our children will judge us by, in whichever world we bequeath to them.

We know what a system built for peace would require. We have the tools. We have the knowledge. What remains is the hardest thing in politics: the willingness to build something that the most powerful interests in the world have structured everything to prevent.

Oksana, for what it is worth, is still working. The internet in Kyiv is more reliable now. The war is not over.

That, at least, is an honest place to start.

Recommended Reading:

Core Belief: Discomfort is not a warning to exit. It is information about where the market is testing your conviction.

Investor Application: The moments that feel most unbearable. A 30% drawdown, a position held in silence while peers rotate out are precisely the moments that separate investors who compound wealth from those who merely participate.

What it looks like in practice:

1. You hold a position through volatility without rushing to “do something” to relieve the tension

2. You sit with an unpopular thesis. Declining revenues, cyclical headwinds. You don’t need the market to validate you immediately

3. You recognise that the discomfort of inaction is not the same as the risk of being wrong

ATTITUDE 2: Detach from the Market’s Daily Verdict

Core Belief: Seeking moment-to-moment approval from price action is the investor’s equivalent of adjusting your opinions to match the room.

Investor Application: The market offers a new verdict on your intelligence every trading day. Internalising that verdict as truth rather than noise corrodes long-term positioning. Conviction cannot be crowd-sourced.

What it looks like in practice:

1. You stop checking your portfolio as a measure of whether you were right today

2. You resist the urge to post, share, or explain your thesis every time it faces a drawdown

3. You measure your investment decisions by the quality of the reasoning at entry, not by the immediate price response

ATTITUDE 3: Slow Down Before You Act

Core Belief: The impulse to react instantly to market events. Earnings misses, macro surprises, analyst downgrades, geopolitical upheavals. This is not decisiveness. It is emotionally-driven impulsiveness dressed in urgency.

Investor Application: Tempo is leverage. The investor who delays a sell decision by 48 hours, who pauses before chasing a momentum name, who breathes before capitulating. That investor controls the quality of their own decision-making.

What it looks like in practice:

1. You implement a personal rule: no portfolio changes within 24 hours of a major macro announcement

2. You write down your reaction before acting on it, creating a gap between stimulus and response

3. You recognise that most of the “urgent” decisions you’ve ever made in markets were driven by the discomfort of uncertainty, not by genuine new information

ATTITUDE 4: Be Selectively Generous with Your Conviction

Core Belief: Not every idea, sector, or narrative deserves equal allocation of your capital or attention. Indiscriminate diversification, like indiscriminate generosity, flattens hierarchy and destroys edge.

Investor Application: The investor who has a strong view on everything has a strong view on nothing. Selective concentration reserved for your highest-conviction, most thoroughly stress-tested positions is what gives a portfolio both structure and authority.

What it looks like in practice:

1. You reserve meaningful position sizing for ideas you have genuinely stress-tested, not ideas you simply find interesting

2. You triage your research time: not every macro theme earns deep diligence; you decide consciously where your analytical energy is deployed

3. When a position no longer meets your standards, you reduce exposure calmly and without drama. Not as punishment, but as repositioning

ATTITUDE 5: Withhold the Need for Narrative Closure

Core Belief: The market rarely offers clean endings. Insisting on a tidy resolution through a confirmed thesis, a catalyst that finally arrives exposes a need for certainty that markets will exploit.

Investor Application: Some of the best investments in history looked unresolved, ambiguous, and narratively incomplete for years before the thesis crystallised. Mystery, in a position, is not a flaw. It is often where the premium lives.

What it looks like in practice:

1. You hold a position through a period of narrative confusion — no clear catalyst, no clear timeline — without forcing an exit to regain a sense of resolution

2. You resist the urge to over-explain your thesis to others, which often leads to over-committing to a specific outcome

3. You allow the investment to develop without needing the story to be finished before it has earned its ending

ATTITUDE 6: Treat Every Market Cycle as a Power Exchange

Core Belief: Bull markets redistribute confidence; bear markets redistribute ownership. Every cycle is an exchange. Between the impatient and the patient, the leveraged and the liquid, the reactive and the deliberate.

Investor Application: The question in every market environment is not “what is happening?” but “who is giving up what, to whom, and why?” Framing markets as an exchange of positioning not just a reflection of economic reality sharpens the analytical lens considerably.

What it looks like in practice:

1. During sell-offs, you ask who is forced to sell and why, rather than simply asking whether the fundamentals have changed

2. You track the asymmetry in a room: when consensus is crowded on one side of a trade, the exchange of leverage is already underway

3. You calibrate your own energy expenditure. If a position is draining your mental bandwidth disproportionately relative to its size, that imbalance is data worth acting on

ATTITUDE 7: Design Your Investment Identity Intentionally

Core Belief: If you do not define your investment philosophy with precision, the market and the narratives surrounding it will define it for you. A vague mandate invites drift; a clear identity creates structure.

Investor Application: Your investment reputation to yourself as much as to others is built through repeated, consistent behaviour under pressure. Every panic sale, every momentum chase, every capitulation to consensus writes a line in that biography. The investor who operates without a deliberate identity accumulates contradictions rather than compounding returns.

What it looks like in practice:

1. You can articulate your investment philosophy in three sentences, and those sentences are consistent with how you actually behave under stress

2. You audit your recent decisions: are they aligned with the investor you intend to be, or are they shaped by approval-seeking. From the market, from peers, from financial media?

3. You treat each decision not only as a capital allocation but as a precedent. Because the standards you enforce with yourself now determine the quality of decisions you make when the stakes are highest​​​​​​​​​​​​​​​​

The Central Argument

The 2026 letter is Fink’s most explicitly political in several years, which is notable given how aggressively BlackRock has retreated from visible political positioning since the ESG backlash of 2022 and 2023. The core argument runs like this:

Globalisation created more wealth than any previous era, but the gains were captured by asset owners rather than wage earners. Since 1989, a dollar in the S&P 500 has grown more than fifteen times the value of a dollar tied to median wages. AI threatens to replay this pattern at greater speed. The answer is not to slow down the capital markets but to bring more people into them.

This is, at its core, an ownership argument rather than a redistribution argument, and that is a deliberate choice. Fink is not calling for higher taxes, stronger unions, or wealth transfers. He is calling for broader market participation. It is a position that is simultaneously pro-capital and pro-inclusion, which is why he can argue it without offending the institutional clients who pay BlackRock’s fees. Whether that makes it less sincere or simply more durable is worth asking.

The argument deserves to be taken seriously on its merits. The empirical case is strong. Countries with deep capital market participation tend to generate better retirement outcomes, broader wealth distribution, and more domestic investment. Japan’s NISA reform is a striking example in the letter: after contribution limits were increased in 2022, nearly ten million new investors entered Japanese markets in three years, the Nikkei rose from roughly 28,000 to over 50,000, and companies began investing more in their own futures. That is a genuine policy success story, not a marketing anecdote.

The Four Big Claims

Claim one: self-reliance is the defining trend of this era, and it requires more long-term capital.

This is the letter’s strongest section and probably its most original contribution relative to prior years. Fink is right that governments everywhere are reconfiguring away from interdependence and toward resilience. Europe is rebuilding its defence industry. America is trying to re-shore semiconductor manufacturing. Emerging markets are securing domestic energy supply. Every one of these transitions is expensive, and the old financing channels, banks and governments, are constrained. Banks face capital requirements. Governments carry record debt.

His conclusion, that capital markets must fill the gap, is logical. The global capital markets approach $300 trillion. Governments cannot write these cheques alone. BlackRock, which manages $14 trillion and is now embedded in infrastructure, private credit, and data centres through GIP and HPS, is obviously positioned to channel this capital. The letter is honest about this alignment without being embarrassed by it. That is the right posture.

What is underdeveloped in this section is the tension between self-reliance and capital market efficiency. Markets allocate capital toward the highest risk-adjusted return regardless of national flags. If British pension funds are encouraged to invest in British infrastructure, they may be giving up global diversification for political solidarity. Fink gestures at this, noting that “money should still move freely toward opportunity,” but he does not resolve the tension. In practice, governments pushing for domestic investment mandates face genuine trade-offs that this letter does not fully reckon with.

Claim two: wealth inequality is primarily a market participation problem.

Here is where the letter starts to narrow its frame in ways that deserve pushback. Fink’s diagnosis is that most of the wealth created since the Berlin Wall fell flowed to the capital markets, and too few people were invested in them. The solution is therefore to get more people invested. This is not wrong, but it is incomplete in a way that matters.

The wealth gap in advanced economies is driven by multiple structural forces simultaneously: decades of wage stagnation, the collapse of collective bargaining, the hollowing out of middle-skilled manufacturing jobs, housing affordability crises in major cities, and the concentration of corporate power in winner-takes-most technology sectors. Market participation helps people who already have disposable income. It does not address the families, mentioned by Fink himself, who live paycheck to paycheck and cannot save in the first place.

Fink acknowledges this: “You can’t invest if you’re not sure you can afford next month’s rent.” But the letter treats this as a prelude rather than the central problem. The emphasis quickly shifts to mechanisms for expanding investment, emergency savings accounts, birth investment accounts, NISA-style tax wrappers, rather than the harder question of why so many households have no surplus to invest. A more complete letter would have spent more time on wage growth, the cost of housing, and the distributional structure of corporate profits.

None of this invalidates the market participation argument. Getting more people invested is genuinely good policy. But it is not a substitute for addressing the conditions that prevent participation in the first place.

Claim three: AI will concentrate wealth unless ownership broadens alongside it.

This is the sharpest and most intellectually honest passage in the letter. Fink is not promoting AI as a universal good. He is describing it as a force that will generate enormous value, most of which will accrue to the companies building it and the investors who own those companies. The K-shaped economy he describes, where Walmart reaches its highest-ever valuation in the same fortnight that Saks files for bankruptcy, is a real phenomenon, and AI risks accelerating it.

His suggestion, that broader market participation is the mechanism through which more people share in AI gains, is reasonable but limited. The families most likely to be displaced by AI in entry-level white-collar roles are also the families least likely to hold equity portfolios. The correlation between AI disruption risk and investment portfolio size runs the wrong direction for Fink’s argument to fully land.

The section on skilled trades is refreshing and somewhat unexpected from a Wall Street executive. Fink quotes Jensen Huang on the dignity of work and argues that electricians and data centre technicians represent some of the most secure careers in the AI economy. The $100 million BlackRock Foundation pledge to train 50,000 skilled trades workers in America has the feel of genuine belief rather than corporate window dressing. Whether a philanthropic initiative at that scale moves any needle in a labour market of 160 million is a separate question.

Claim four: Social Security should be partially invested in the capital markets.

This is the letter’s most politically charged section and the one most likely to generate controversy, as Fink acknowledges he expects it to. His position is careful and specific: he is not calling for privatisation. He is pointing to the Cassidy-Kaine proposal, which would create a new investment fund parallel to the existing trust fund, not replacing it, invested in a diversified mix of stocks and bonds, with a 75-year horizon. He draws the analogy to Australia’s superannuation system and to the federal Thrift Savings Plan for government employees.

The intellectual case is genuine. The Social Security trust fund is invested almost entirely in US Treasury bonds. At current contribution and benefit levels, it is projected to be unable to pay full benefits by 2033. A diversified long-term portfolio would, historically, have generated meaningfully higher returns than Treasuries. Norway’s sovereign wealth fund, Australia’s superannuation system, and Canada’s CPP Investment Board all demonstrate that government retirement funds can be invested in diversified markets without compromising the underlying guarantee.

The criticism is equally genuine. Market timing is brutal for government guarantees. A retiree in 2009 or 2020 would have been exposed to sharp declines precisely when they needed income. The transition cost of the Cassidy-Kaine proposal, $1.5 trillion of initial investment with 75 years to mature, is an enormous commitment requiring Treasury to cover benefits throughout. And there is an obvious conflict in the CEO of the world’s largest asset manager publicly endorsing a policy that would potentially channel enormous flows of new capital into financial markets that BlackRock manages. Fink does not address this conflict. He probably should.

The Forecasts and Prognoses

Fink makes a series of specific-enough claims to be held accountable to.

He forecasts that private markets will grow significantly enough for the 50/30/20 portfolio allocation (public equities, bonds, private assets) to become a mainstream target. This is consistent with Preqin’s own research projecting private credit at $4.5 trillion by 2030. It is also the forecast most favourable to BlackRock’s recent acquisition strategy, which should be noted. However the recent bad press on private credit redemptions makes it a questionable allocation.

He predicts that tokenisation will transform market infrastructure, comparing it to where the internet was in 1996. This is a high-conviction claim. The analogy is provocative but plausible: settlement efficiency, fractional ownership, and broadened access are genuine use cases. The timeline is uncertain. Fink and Rob Goldstein wrote about tokenisation in The Economist; the fact that BlackRock has already built the world’s largest tokenised treasury fund ($150 billion in digital asset AUM) gives this claim operational credibility rather than just theoretical weight.

He projects BlackRock reaching $35 billion in revenue by 2030, with 30% or more from private markets and technology. At $24.2 billion in 2025 revenue, that requires roughly 7.5% annual growth, which is achievable given the trajectory. The private markets and technology mix shift is more ambitious: it requires these segments to roughly double their share of revenue. This is the single most important financial forecast in the letter, and it is where execution risk is highest.

What Is Missing

The letter is notably silent on three things that thoughtful shareholders will notice.

First, there is no mention of the HLEND withdrawal restriction. In March 2026, BlackRock’s flagship HPS private credit fund gated withdrawals at 5% when redemption requests reached 9.3%. This is the most significant operational event at the firm in the reporting period and it is entirely absent. A letter that discusses the growth of private markets and the democratisation of alternatives without mentioning that one of its own funds is currently restricting redemptions is incomplete.

Second, CEO succession is not addressed. Three senior executives widely seen as succession candidates have departed in two years. Fink is 73. The letter references expanding the Global Executive Committee and welcoming leaders from the acquired firms, but it does not speak to how the firm will manage a transition of the founder-CEO who has run it for 38 years.

Third, the regulatory and antitrust environment is unaddressed. The Texas lawsuit alleging that BlackRock’s index fund voting suppressed fossil fuel output has survived legal challenge. The DOJ filed in support of the states. These are material risks that a letter addressed to shareholders should probably acknowledge.

My Honest Verdict

Larry Fink writes well and thinks carefully. This is a genuinely interesting letter, not corporate boilerplate. The central argument, that prosperity concentrated in capital markets requires capital markets to be more accessible, is correct and important. The case studies from India, Japan, Germany, and the United States are instructive and specific.

But the letter is also the work of a man whose firm stands to benefit enormously from the policies he advocates: more retirement assets in diversified funds, more private markets allocation, more capital market participation globally, more tokenisation. That does not make the arguments wrong. It does mean the reader should hold them up to independent scrutiny rather than treating them as disinterested analysis.

The best test of Fink’s sincerity is whether BlackRock’s actions match its words. On private markets democratisation, they broadly do: the JioBlackRock venture, the LifePath target-date innovation, and the 401(k) private markets push are genuine product-level commitments. On workforce investment, the $100 million foundation pledge is real. On tokenisation, the firm is building rather than just talking.

A letter from the CEO of the world’s largest asset manager is always partly a vision statement, partly a shareholder communication, and partly a position paper for policymakers. This one does all three reasonably well. It is just worth knowing which hat is on at any given moment.

If you are new to my Substack Read This First

BlackRock delivered record results in 2025 yet the stock trades 21% below its October 2025 high. The disconnect stems from a March 2026 private credit scare (the HPS Lending Fund gated withdrawals) and broader concerns about digesting three transformational acquisitions worth $27.7 billion in nine months. This creates a compelling entry point.

Price at analysis: ~$960 | AUM: $14.04 trillion | Adjusted P/E: ~20x trailing / ~17.5x forward / Frameworks applied: Nick Sleep · Peter Lynch · Anthony Bolton · Howard Marks / March 2026

On an adjusted basis, BLK trades at ~20x trailing earnings and ~17.5x forward, close to its 5-year average despite a fundamentally upgraded business mix with fee yields on new flows running 6–7x higher than two years ago. The bear case deserves scrutiny. BlackRock overpaid for private markets at the cycle top. But the structural shift from passive-only to an integrated public-private platform with a $2 billion technology moat is the most significant transformation in asset management since Vanguard popularised indexing.

FY2025 earnings reveal a business in transformation

BlackRock’s full-year 2025 results tell two stories depending on whether you read GAAP or adjusted figures. GAAP revenue rose 19% to $24.2 billion, but operating income actually fell 7% and GAAP EPS dropped to $35.31 from $42.01. An 800-basis-point margin compression driven by $2.2 billion in non-cash acquisition-related charges(contingent consideration markups, intangible amortisation, and deal-linked compensation). Adjusted for these items, the picture is entirely different: operating income grew 18%, margins held at 44.1% (down just 40 bps), and EPS climbed 10% to $48.09, beating consensus by $0.74.

The flow picture was extraordinary. Record net inflows of $698 billion. More than many asset managers’ total AUM. $527 billion into iShares ETFs, $40 billion into private markets, and $131 billion into cash management. Q4 alone contributed $342 billion. Revenue composition shifted meaningfully: technology services hit $1.98 billion (+24%), alternatives base fees nearly doubled to $3.02 billion, and performance fees grew 27% to $1.25 billion. Investment advisory fees on the core business rose $3.1 billion year-over-year.

The critical forward-looking metric is BlackRock’s run-rate base fee trajectory. Entering 2026, base fees are approaching $21 billion, roughly 35% higher than 2024 and 50% higher than 2023.

The Business in Plain English

Imagine you own a motorway. Every car that passes pays a toll. You do not care whether it is a Ford or a Ferrari, a commuter or a lorry driver. You collect a small slice from all of them, automatically, every day. Now imagine that motorway handles not hundreds of thousands of journeys but the equivalent of the entire gross domestic product of every country on earth bar two. That is roughly what BlackRock does.

BlackRock manages money. $14.04 trillion of it. On behalf of pension funds, sovereign wealth funds, insurance companies, and ordinary savers worldwide. It is the largest asset manager the world has ever seen, and it has just completed the most ambitious three-deal transformation in the history of its industry: buying a global infrastructure empire (GIP), the leading private markets data company (Preqin), and one of the world’s premier private credit lenders (HPS) in the space of nine months.

So what do we actually own if we buy the shares? Three things running in parallel.

1. First, an ETF factory that manages $5.5 trillion through its iShares brand, roughly 29% of the global ETF market.

2. Second, a risk management software platform called Aladdin, used by over 1,500 institutions to run their portfolios, generating $2 billion a year in subscription fees with 98% client retention.

3. Third, a rapidly growing private markets operation spanning airports, energy pipelines, data centres, and direct corporate loans.

Each of these three engines is exceptional on its own. Together, they form a business that is genuinely difficult to compete with. The question for a long-term investor is whether the price is right, and whether the transformation from passive giant to integrated platform is as durable as it looks.

The Numbers: What the Scorecard Says

Before applying any qualitative lens, it is worth grounding ourselves in the financials. What has actually happened to this business over the past five years?

A few things stand out. Revenue fell in 2022 as markets dropped and took AUM with them, recovering strongly in 2023 and then accelerating. The adjusted operating margin, which strips out acquisition-related charges, has stayed in a tight 42–47% band across a full market cycle. That is a mark of genuine operating leverage: revenues fluctuate with markets but the cost base is disciplined.

The per-share earnings story is quietly impressive. Adjusted EPS has compounded at roughly 9% per year from 2021 to 2025 despite the HPS dilution (around 5% more shares outstanding). The dividend has grown from around $16 annualised in 2020 to $22.92 today, a CAGR of approximately 9.1%, funded by a payout ratio sitting comfortably below 50% of adjusted earnings.

The most important trend is the fee mix shift. Alternatives AUM has grown at a 22% five-year CAGR, rising from 2% to 3% of total AUM. That sounds modest, but alternatives carry fee rates of 60–100 basis points versus 4–6 basis points for passive ETFs. As this slice grows, it pulls average fee yield upward even as the headline passive book commoditises. The 2025 private markets net inflow of $40 billion, generating fee revenue at 7–8x the rate of an equivalent passive flow, is a structural tailwind the market is still fully pricing in.

Three acquisitions that reshaped the entire firm

Global Infrastructure Partners closed in October 2024 for $12.5 billion ($3 billion cash, remainder in stock), adding $170 billion in infrastructure AUM spanning airports, data centers, energy pipelines, and ports across 100+ countries. GIP’s flagship Fund V closed above its $25 billion target, the largest private markets raise in BlackRock history. The AI Infrastructure Partnership. A consortium with Microsoft, NVIDIA, and MGX has already committed to acquiring Aligned Data Centers at a $40 billion enterprise value, and GIP was central to the $22.8 billion CK Hutchison ports deal.

Preqin, acquired for $3.2 billion in March 2025, is the strategic lynchpin most investors overlook. As the premier private markets data provider covering 210,000 funds and 220,000 users, Preqin transforms Aladdin from a public markets operating system into a whole-portfolio technology platform. Fink called it the “unlock” for democratising alternatives. Its data is essential for creating private market indices, enabling private assets in 401(k) plans, and providing the transparency fiduciaries require. Aladdin ACV grew 31% including Preqin (16% organic) in Q4 2025.

HPS Investment Partners closed in July 2025 for approximately $12 billion in all-stock consideration, adding $148 billion in private credit AUM. The deal created a new Private Financing Solutions unit with $190 billion in combined assets, led by HPS founder Scott Kapnick. HPS contributed approximately $230 million in Q4 base fees and $158 million in performance fees. Immediately accretive. Combined, the three deals added roughly $320 billion in fee-paying AUM and shifted over 20% of BlackRock’s revenue to long-dated, less market-sensitive products.

The total capital deployed was significant: $6.2 billion in cash and up to 25.7 million shares of BlackRock stock (representing ~15% dilution at maximum). Share count rose from 154.9 million to approximately 163 million. BlackRock’s CFO has signaled that these deals “round out near-term to intermediate-term priorities for large M&A,” suggesting the digestion phase takes priority.

Aladdin and iShares form a structural competitive moat

Aladdin is the operating system for institutional investing. More than 130,000 users across 1,500+ institutions manage roughly $21–25 trillion on the platform, approximately 3x BlackRock’s own AUM. Client retention stands at 98% over three years. Technology services revenue reached $1.98 billion in 2025, and with Preqin integrated, BlackRock now possesses a SaaS franchise covering both public and private markets with no true competitor. The addressable market is estimated at $12.5 billion, of which Aladdin has captured approximately 11%. The newly announced Aladdin on AWS partnership brings the platform to Amazon Web Services with Amazon using Aladdin to manage its own treasury. This validates the platform’s enterprise-grade positioning.

iShares remains the world’s dominant ETF franchise with $5.47 trillion in AUM across more than 1,700 products. 6x the lineup of the next largest issuer. Full-year 2025 net inflows of $527 billion were a record. The platform’s evolution beyond passive indexing is the underappreciated story: active ETFs drove over $50 billion in inflows, nearly tripling their AUM to surpass $100 billion. IBIT, the iShares Bitcoin Trust, became the third-highest revenue generator among BlackRock’s ~1,200 funds, attracting $25.4 billion in 2025 inflows. In Europe, iShares holds 40.4% market share and recorded $136 billion in net inflows, more than double the nearest competitor.

The fee compression narrative, while real, is moderating. Industry-wide passive fund fees declined just 1.5% year-over-year in 2024. Far slower than prior years. BlackRock’s asset-weighted ETF expense ratio of ~0.22% remains well above Vanguard’s 0.07%, but this reflects product mix rather than competitive weakness. BlackRock actually increased its effective fee rate by 0.7–1 basis point in 2024 through mix shift toward higher-fee active, thematic, and alternatives products.

Private markets: the $4.5 trillion opportunity and its risks

BlackRock’s private markets transformation is the central investment thesis. Fee-paying private markets AUM reached $322.6 billion at year-end 2025, with broader alternatives client assets of $675 billion up from $288 billion just one year earlier. Infrastructure leads at $112 billion, private credit at $145 billion (overwhelmingly from HPS), private equity at $31 billion, and real estate at $25 billion. These assets represent just 3% of total AUM but generate 16% of base fees and a disproportionate share of performance fees.

The private credit market, projected by both BlackRock and Preqin to reach $4.5 trillion by 2030 (from approximately $1.7 trillion today), is driven by three structural forces: Basel III capital requirements pushing banks out of lending, insurance companies seeking yield for long-dated liabilities, and the nascent “democratisation” of alternatives through semi-liquid vehicles. Morgan Stanley’s broader definition estimates $3 trillion today growing to $5 trillion by 2029. BlackRock’s combined private credit franchise of ~$220 billion positions it as one of the largest players alongside Apollo, Blackstone, and Ares.

Fink’s 2025 annual letter articulated the strategic vision: replace the traditional 60/40 portfolio with a 50/30/20 allocation (stocks, bonds, private assets). He called infrastructure “a generational investment opportunity” requiring $68 trillion globally by 2040, and advocated for private credit in 401(k) plans. An extra 0.5% annual return over 40 years yields 14.5% more retirement savings.

However, the March 2026 HLEND episode is a genuine warning signal. BlackRock’s $26 billion HPS Corporate Lending Fund received 9.3% in redemption requests but could only pay out 5% under its quarterly cap. The stock fell 7.5% on the news. This is not unique to BlackRock. Blackstone’s BCRED faced 7.9% redemption requests. Blue Owl restricted withdrawals. Morgan Stanley gated its North Haven fund. It tests the thesis that semi-liquid private credit can serve retail investors without liquidity mismatch. The industry-wide “hidden” default rate (including selective defaults and liability management exercises) may approach 5%, well above the reported 2% headline figure.

Is This a Destination Business?

Nick Sleep and Qais Zakaria at Nomad Investment Partnership spent 12 years studying what they called ‘destination companies’: businesses where scale creates such durable advantages that customers are drawn inexorably toward them, and where the firm reinvests those advantages back into customers rather than hoarding them as margin. Their great examples were Amazon and Costco, businesses that used scale to lower prices, which attracted more customers, which generated more scale.

The question for BlackRock is whether it exhibits the same self-reinforcing logic. On first inspection, it seems unlikely. Asset management looks nothing like retail. But look again.

Scale Economics Shared: the passive engine

When BlackRock reaches $5.5 trillion in iShares ETF assets, its cost to run each fund falls dramatically. Regulatory overhead, technology, liquidity management, and index licensing costs are mostly fixed. The more assets, the lower the per-unit cost. Critically, BlackRock passes much of this saving back to investors through lower expense ratios. The iShares Core S&P 500 ETF (IVV) charges just 0.03% per year. That is not charity. It is Sleep’s scale economics shared in action: as the business grows, so do the benefits to customers, which attracts more customers, which funds more growth. Vanguard pioneered this, but BlackRock has matched it whilst running a far broader product range.

Aladdin is an even purer example. The platform processes data on $21–25 trillion of assets, roughly three times BlackRock’s own AUM. Every new institution that joins Aladdin adds to the risk dataset, which improves the models for everyone, which attracts more institutions. The switching cost is enormous: years of data, workflows, and institutional habit live inside the system. Over 98% of Aladdin clients renew each year. This is not a toll road. This is a city that people choose never to leave.

The Destination Test

Sleep asked a simple question: where will this business be in ten or twenty years if everything goes roughly right? For BlackRock, the answer is compelling. The global savings pool is not shrinking. Pension systems are underfunded everywhere. Governments increasingly outsource their investment decisions. Retail investors are only beginning to access private markets through semi-liquid vehicles. The demographic wave of retirement in developed economies ensures capital will be seeking management for decades. BlackRock sits at the centre of all of these flows.

More specifically, the Preqin acquisition is a Sleep-style move. Preqin owns the data infrastructure of private markets. As private markets grow from $1.7 trillion today to a projected $4.5 trillion by 2030, Preqin’s data will become more valuable, not less. BlackRock is not just a manager in this market. It is becoming the operating system for the market itself. No competitor has this combination.

Per-Share Value Creation

Sleep cared deeply about per-share value, not just total value. When companies grow by issuing stock, they can create headline numbers that look impressive but deliver nothing to existing shareholders. How does BlackRock measure up?

The HPS deal was all-stock, adding approximately 7.6 million Class B units (roughly 5% dilution). The GIP deal was partly stock. The share count has risen from around 155 million to 163 million over 18 months. This is genuine dilution and deserves to be weighed honestly against the earnings accretion. On the HPS deal alone, management estimates the acquisition contributes $200–230 million in quarterly base fees. At 16x earnings multiple, that represents approximately $12–15 billion in value created on a $12 billion purchase price, suggesting the dilution is being compensated.

The adjusted EPS of $48.09 in 2025 compares to $43.61 in 2024, a 10% increase despite the dilution. That is per-share value creation. The dividend growing at 9% per year whilst the payout ratio stays below 50% suggests the firm is generating more earnings per share than it is distributing, reinvesting the difference.

Sleep’s verdict on BlackRock

A destination business is one where scale compounds the advantage. BlackRock does not just have scale. It has built a system that other institutions cannot function without.

Where Sleep might hesitate is on the acquisition pace. Nomad was famously sceptical of companies that grew primarily through deals rather than organic reinvestment. Three $10+ billion acquisitions in nine months is aggressive. The risk is that management attention, culture, and talent are stretched, and what looks like value creation on paper unravels in execution. The BGI acquisition in 2009 is the precedent: it was transformational, it was managed well, and it created the modern BlackRock. But it took years to digest.

The Sleep score: strongly favourable on structure, watch the integration. BlackRock is almost certainly a destination company. The pace of the build is the genuine concern.

Would Lynch Buy It?

Peter Lynch’s greatest gift was making investing accessible. His central argument was that ordinary people can identify great investments by understanding what they see around them. Buy what you know. Understand what you own before you buy it. Favour boring businesses that Wall Street ignores over glamorous ones that everyone already owns.

BlackRock is not boring in the way that Lynch’s favourite tyre or shoe company was boring. It is not a business most people encounter daily. But it has several qualities Lynch would have found deeply attractive.

Business category: slow grower with an accelerator attached

Lynch categorised stocks into six types: slow growers, stalwarts, fast growers, cyclicals, turnarounds, and asset plays. BlackRock is a stalwart in its passive ETF business, and a fast grower in private markets and technology. The combination is unusual and powerful.

A stalwart with a fast-grower attached is not something Lynch saw often, but when he did, he liked it. The stalwart side provides the earnings base and the dividend, the fast-grower side provides the earnings acceleration. BlackRock’s passive AUM grows at roughly 8–10% per year, largely driven by market appreciation and net flows. Its alternatives AUM is growing at 22% per year. The technology revenue is growing at 15–25% per year. The blended result is a business that looks like it is compounding at 10–15% per year in earnings power.

The Lynch checklist

Lynch famously listed the characteristics of a perfect tenbagger: low analyst coverage, boring product, no institutional ownership, depressing name, no growth industry association. BlackRock fails most of these tests. It is the most widely covered financial stock on earth. Institutional ownership is near-universal. But Lynch also applied a second checklist for larger, more established businesses: does it have a niche? Does it have consistent earnings? Is the balance sheet manageable? Does management own a meaningful stake?

On the niche question, BlackRock has three: the world’s largest ETF platform, the only integrated public-private risk operating system, and now a top-5 alternatives manager.

On consistency, adjusted earnings have grown in every year since the pandemic with a single exception (2022, when markets fell globally).

The balance sheet carries $12.8 billion in long-term debt, significant given $63.3 billion in goodwill and intangibles from the acquisitions, but supported by $7.5 billion in adjusted operating cash flow, giving a debt-to-operating-cash-flow ratio of approximately 1.7x. Manageable.

On management alignment, Larry Fink and Robert Kapito, BlackRock’s President, each owned tens of millions of dollars of stock at last count. The new Executive Carry Programme ties the senior leadership team’s compensation explicitly to the performance of the private markets business, meaning the people running the transformation have real skin in the game.

Lynch’s classic traps: does BlackRock fall into any?

Lynch warned against what he called ‘whisper stocks’ (too much hype), companies that diversify through acquisition (the ‘diworsification’ problem), and businesses where the most exciting growth is in unrelated areas. It is worth checking each honestly.

Diworsification risk: Lynch reserved his sharpest criticism for companies that bought unrelated businesses to disguise stagnant core growth. Is that what BlackRock is doing? Partially. GIP, Preqin, and HPS are not random diversifications. They all serve the same clients (institutional investors), fit the same platform logic (Aladdin), and address the same structural shift (alternatives allocation). This is strategic extension, not diworsification. But the pace and scale require discipline to execute. If the private credit market sours badly and HPS becomes a write-down, Lynch’s warning will look prescient.

Too much coverage risk: Lynch worried about stocks that every analyst already owned. BlackRock has 16 analyst buy ratings, an average price target of $1,254, and is a core holding in virtually every financial sector fund. There is no ‘undiscovered gem’ premium here. But Lynch also recognised that the best businesses eventually become well-known, and the question then is whether you are paying a fair price for quality. At 17–18x forward earnings, the answer is: roughly, yes.

What the business actually does: Lynch said if you cannot explain in two minutes why you own a stock, you do not own it properly. For BlackRock: the world’s largest asset manager is transforming from a passive ETF company into a full-service capital markets platform. It earns fees on $14 trillion, provides the technology that 1,500 institutions use to run their portfolios, and is building the infrastructure business that will fund the energy transition and AI data centres. Revenue is growing at 15–20% per year, earnings at 10%, and the dividend at 9%.

Lynch’s verdict

If you like the business, you will love it in three years if the story plays out. If it doesn’t, you will wish you had waited.

Lynch would probably classify BlackRock as a stalwart trading at a fair-to-modest premium. He would like the earnings consistency, the dividend growth, the management alignment, and the niche dominance. He would worry about the acquisition pace, the integration risk, and the political controversy around ESG. His instinct would be: this is worth owning, but be honest about what you are paying for, and do not overpay for the private markets story before it has delivered.

The Lynch score: cautiously positive. Own it if you understand the thesis; add aggressively only if the price gives you a meaningful margin of safety.

What Anthony Bolton Would Notice

Anthony Bolton ran Fidelity’s Special Situations fund for 28 years, producing one of the most consistent records in British fund management history. His approach was contrarian by instinct: he looked for stocks that the market misunderstood, disliked, or feared, where sentiment was worse than the underlying reality. He was particularly attracted to inflection points, governance changes, and situations where a catalyst could close the gap between price and value.

Bolton would approach BlackRock with a set of questions that are different from Lynch’s or Sleep’s. Not ‘is this a great business?’ but ‘why is the market wrong today, and what would close the gap?’

Sentiment and the Loneliness Premium

Is BlackRock unpopular? Not obviously. With 16 Buy ratings and an average price target 31% above the current price, it is hardly a contrarian pick in the traditional sense. But Bolton understood that within a widely-owned stock, there can be a specific thesis that the market is mispricing, even if the headline view is positive.

The market is currently pricing BlackRock as though the HLEND gating episode is the beginning of a structural problem in private credit. The stock fell 7.5% on the news, and it has fallen approximately 21% from its October 2025 peak. That correction is doing a lot of valuation work. At $960, BlackRock is trading at ~17.5x forward earnings, well below the 24–26x it commanded six months ago.

Bolton would ask: is the HLEND episode a signal or noise? His instinct would be to distinguish the business quality from the market’s temporary anxiety. A gating on a semi-liquid credit fund, in an environment where multiple managers simultaneously gated, is more consistent with a liquidity management event than a credit quality crisis. If the underlying loans in HLEND are performing (the default rate for direct lending funds remains below 2% on a headline basis), the gating is a structural feature of how semi-liquid products work, not evidence of impairment.

The Special Situation: an underpriced technology premium

Bolton was masterful at identifying assets within companies that the market was valuing at zero or near-zero. In BlackRock, that asset is Aladdin-plus-Preqin as a combined technology franchise.

Aladdin generates roughly $2 billion in annual recurring revenue, growing at 15–25%. By any reasonable SaaS comparable, a business with those characteristics, 98% retention, and no credible competition deserves a 25–35x revenue multiple. Implying a standalone value of $50–70 billion. The market is currently pricing all of BlackRock at approximately $150 billion. That means it is implicitly valuing the entire $14 trillion asset management business, GIP, HPS, iShares, and $2 billion in technology revenue together for $80–100 billion. That looks like a significant undervaluation of the technology component.

Preqin deepens this. Private markets data is subscription-based, recurring, defensible, and growing as the alternatives industry expands. There is no market comparator because no one else does what Preqin now does inside BlackRock. If Preqin were a standalone company, it would likely attract a 20–30x revenue multiple from growth investors.

Management quality and governance

Bolton placed heavy emphasis on management character: did they do what they said they would? Were they honest about problems? Did they allocate capital well?

Larry Fink’s record across 37 years is extraordinary by any standard. The BGI integration, the post-2008 pivot to iShares dominance, the early recognition of the ETF structural shift, and now the private markets repositioning are all evidence of genuine strategic foresight. His annual letters read less like PR and more like a coherent multi-year thesis playing out. They have addressed infrastructure, retirement security, and capital markets democratisation with increasing precision.

The less comfortable governance question is succession. Bolton would note that three potential CEO successors have departed in two years: Mark Wiedman, Salim Ramji (now at Vanguard), and Mark Wiseman. That is not a clean record. It suggests either that the internal talent pipeline is thinner than the board acknowledges, or that Fink’s dominance makes the organisation difficult for ambitious executives to thrive within. Either interpretation creates genuine succession risk.

Capital allocation has been bold but defensible. The decision to pay $3.2 billion for Preqin was visionary; $12.5 billion for GIP, the largest infrastructure asset manager in the world, was expensive but strategically coherent; $12 billion for HPS was an all-stock deal that diluted shareholders by approximately 5% but acquired a franchise generating over $800 million per year in fee-equivalent revenue. Bolton would apply his nose-test: does this feel like a company building something lasting, or a company doing deals for the sake of it? The answer here leans toward the former, but the pace is testing the boundaries.

Where the contrarian case is strongest

Bolton’s most compelling observation would be this: BlackRock is currently being penalised for risks that are real but manageable, at a valuation that assumes those risks materialise in their worst form. The ESG political controversy, the HLEND gating, the succession question, and the integration complexity are all legitimate concerns. But none of them, individually or together, challenges the core earnings power of a business that manages $14 trillion and grows the fee rate every year.

A Bolton-style position would be: the discount is real, the reasons for the discount are understood, and the fundamental case has not changed. This is precisely the profile he found most attractive.

The contrarian angle that the market is truly missing: the private markets franchise, fully integrated and at scale, will generate $3–5 billion in fee revenue by 2028, up from under $1 billion in 2023. That is an earnings change of approximately 25–30% from a single business line that did not exist in its current form two years ago.

Bolton’s verdict

The best contrarian investments are not the ones where no one has heard of the company. They are the ones where a great company has a problem that the market is treating as permanent but which is actually temporary.

Bolton would likely view the current entry point as more attractive than twelve months ago. The stock is cheaper, the strategic transformation is more advanced, and the specific concern (private credit stress) is industrywide rather than BlackRock-specific. He would want to see evidence that HLEND redemption pressure has stabilised before adding aggressively, but he would not be selling.

The Bolton score: constructive with a specific catalyst watch. The underpriced technology franchise is the genuine special situation. Succession and integration are the legitimate worries.

Howard Marks: Price Versus Value

Howard Marks, co-founder of Oaktree Capital and one of the most rigorous thinkers in investment management, builds his entire approach around second-level thinking: not ‘is this a good company?’ but ‘does the market’s price for this company reflect the correct assessment of risk and return?’ His other great contribution is cycle awareness: understanding where we are in the cycle, and recognising that most investors err by extrapolating the present into the future.

Marks would approach BlackRock with a specific set of questions. What is the risk of permanent capital loss? What does the price imply? Where are we in the cycle? What cannot be known? And is the consensus broadly right or broadly wrong?

Risk of permanent capital loss

This is the first question Marks always asks. Not ‘can this stock go down?’ but ‘can this business be permanently impaired?’

For BlackRock, permanent impairment would require one of the following: a structural collapse in global savings behaviour, a regulatory breakup of the firm, a catastrophic failure in private credit that generated legal liability rather than just fee income losses, or a technology disruption that made Aladdin obsolete. None of these are implausible in isolation. None of them look probable in the next five to ten years.

The regulatory and antitrust risk deserves the most serious treatment. The Texas lawsuit alleging that BlackRock, Vanguard, and State Street used their common ownership of energy companies to restrict coal output has survived a motion to dismiss. The DOJ filed a Statement of Interest supporting the states, a significant escalation. If the common ownership theory were to gain broad legal traction, it could constrain the voting behaviour of index funds, potentially reducing their value proposition. This is a long-tail risk, not an immediate one, but Marks would insist it belongs in the scenario set.

What is the probability of permanent impairment? Honestly assessed: low but non-trivial. The regulatory path would take years to resolve. The private credit stress, even if it deepens, would reduce fee income rather than create liabilities. The business is not leveraged in the way that a bank is, so a credit event does not threaten solvency.

Price versus value: the cycle check

Marks is famous for his pendulum metaphor. Markets swing between fear and greed, pessimism and optimism. The highest-quality investments are made when the pendulum has swung toward fear, pricing assets below their intrinsic value. The worst are made when everyone is optimistic, the story is obvious, and the price reflects perfection.

Where is the pendulum for BlackRock today? In October 2025, at $1,220, the stock was pricing considerable optimism: three transformational deals closed, private markets growing explosively, AUM at record levels, and the market willing to pay 26x adjusted earnings for what had become a structurally more valuable business. That was a moment of mild greed.

Today, at $960, the pendulum has moved toward mild anxiety. The HLEND gating, the succession uncertainty, and broader market nervousness about private credit have reduced the multiple to approximately 17.5x forward earnings. That is not fear territory, but it is meaningfully more attractive than six months ago. Marks would note that the consensus is still positive (31% upside to the average analyst target), which means this is not yet the point of maximum despondency. But it is certainly closer to the value zone than it was.

What is in the price?

At $960 and 17.5x forward earnings, the market is implying roughly 12–14% earnings growth in 2026, broadly in line with what the business should deliver if private markets continue to ramp and no major credit event materialises. There is no embedded optimism premium. There is, equally, no meaningful pessimism discount.

Marks would call this ‘priced for a normal outcome.’ The question is whether normal outcomes are what happen here, or whether the genuine option value of the private markets and technology transformation deserves a higher price than normal.

His answer would probably be: neither cheap nor expensive. Priced appropriately for the risks visible today, with asymmetric upside if the technology franchise is properly recognised.

The limits of knowledge

This is where Marks is at his most valuable. He insists that investors be honest about what they genuinely cannot know, and that intellectual humility is a competitive advantage.

What cannot be known about BlackRock?

First, whether the private credit cycle has peaked. The HLEND gating is the first genuine test of the semi-liquid private credit model at scale. Whether it is a liquidity management blip or the leading edge of a broader credit stress event is not knowable yet. The answer will become clearer over the next two to three quarters.

Second, whether Larry Fink stays or goes, and when. If he leaves abruptly, the stock will de-rate. By how much, and for how long, depends on who succeeds him and how the market views the transition. This is genuinely binary.

Third, whether the regulatory backlash against common ownership index fund practices has any substance in law. Most securities lawyers believe the common ownership theory is legally weak. But the DOJ’s intervention suggests political motivation may drive the outcome rather than legal merit.

Fourth, whether the integration of GIP, Preqin, and HPS creates the synergies that management projects, or merely the complexity that sceptics fear. Three transformational deals in nine months is unprecedented. The track record of large financial services M&A is mixed at best.

Marks’s verdict

Investing is not about being right. It is about calibrating how much you are paying for what you are getting, and whether the risks you are taking are compensated.

Marks would say: BlackRock is an exceptional business trading at a fair price with a slightly better-than-average risk/reward profile for patient capital. It is not obviously cheap, so it demands patience. It is not obviously expensive, so there is no urgency to sell. The margin of safety is thin but present at current prices. For investors already owning the stock, the case for holding is strong. For new buyers, the ideal entry is either on further weakness (a market correction to $850–900 would represent genuinely attractive value) or after evidence that the private credit stress is contained.

The Marks score: fair value, lean positive, sized accordingly. Not a maximum conviction bet, but a deserving position in a long-term quality portfolio.

The Synthesis: Weaving the Four Lenses Together

Four intelligent people looking at the same business ask different questions and find different things. That is the point. Let us now bring them together.

Three of the four frameworks arrive at the same conclusion by different routes: BlackRock is a high-quality business trading at a price that is reasonable but not heroic. The fourth framework, Sleep’s, would assign a higher structural value to the destination characteristics but would share Lynch’s, Bolton’s, and Marks’s concern about execution risk over the next 12–24 months.

What does it take for this to be a great long-term investment from here? The risk register: what could break the thesis. The thesis rests on four things being true simultaneously.

One: the private markets transformation delivers. If HPS, GIP, and Preqin generate the fee income and synergies that management projects, private markets will represent 25–30% of total revenue by 2028, shifting the earnings quality decisively upward. Fee income from illiquid, long-dated assets is more durable and less market-sensitive than passive AUM management. The business mix shift is the single most important driver.

Two: Aladdin’s expansion into private markets accelerates. The integration of Preqin data into Aladdin’s operating platform is a two-to-three year project. If it succeeds, it creates a product with no competitor and an addressable market that grows every time a pension fund increases its alternatives allocation. The technology revenue could reach $3–4 billion by 2030.

Three: the private credit market does not experience a structural default cycle. The HLEND episode showed the liquidity risk inherent in semi-liquid credit products. If defaults remain below 3–4%, the market stabilises, and institutional flows into alternatives resume, the HPS acquisition will look inspired. If defaults spike to 6–8% and redemption pressure persists, the acquisition becomes a weight on earnings and the strategic rationale frays.

Four: management stability is maintained through the transition. The next two to three years will likely see Fink begin a formal succession process. If handled well, the transition will validate the depth of the organisation. If handled poorly, it creates the kind of uncertainty that keeps institutional investors on the sidelines.

What could cause permanent capital loss?

Being honest: very few scenarios. The most plausible route to permanent impairment is a combination of events: a deep private credit default cycle impairing HPS’s reputation and economics, simultaneously with a Fink departure, a regulatory constraint on voting behaviour that reduces the value of the index franchise, and a market correction that compresses AUM by 30%. Even in this scenario, the technology platform, the iShares franchise, and the distribution infrastructure retain most of their value. BlackRock does not use leverage the way a bank does. It does not have proprietary trading positions. It manages other people’s money for a fee and owns a software platform. Neither of those things go to zero in any realistic scenario.

The more genuine risk is not permanent capital loss but structural disappointment: a business that grows at 6–8% rather than 10–12%, where the private markets thesis takes twice as long as expected, and the stock de-rates from 18x earnings toward 14x. That would produce a poor outcome for anyone paying today’s prices.

Valuation, dividends, and capital return

At approximately $960, BLK trades at ~20x adjusted trailing earnings and ~17.5x 2026 consensus EPS of $54.42. While the GAAP P/E of ~27x appears stretched versus the 5-year average of ~22x and 10-year average of ~20x, this reflects $2.2 billion in noncash charges that will diminish over time. On an adjusted basis, the forward multiple implies roughly 14% earnings growth in 2026. Achievable given the fee rate uplift from private markets and continued AUM growth. EV/EBITDA stands at ~16.4x.

• Analyst consensus: Strong Buy across 16 analysts, average price target of $1,254 (31% upside), range $980–$1,486

• Key bulls: Morgan Stanley ($1,486), BofA ($1,467), Deutsche Bank ($1,380)

• Key bears: Wells Fargo ($1,170), Evercore ISI ($1,235)

The dividend story is compelling for income-oriented investors. BlackRock has raised its dividend for 16–17 consecutive years, most recently a 10% increase to $5.73 per quarter ($22.92 annualized) in January 2026. The current yield of ~2.4% is well-covered at a 43% adjusted payout ratio. Total capital returned in 2025 was approximately $5 billion ($1.6 billion in buybacks, $3.4 billion in dividends), with the 2026 buyback plan increased to $1.8 billion. The 10-year dividend CAGR of ~9.1% is exceptional for a mega-cap financial.

How to think about position sizing

A thoughtful, patient investor would recognise three things. First, BlackRock is not a ‘bet the portfolio’ holding. The margin of safety is thin at $960. The right position size reflects the quality without concentrating on the execution risk. Second, adding on weakness is structurally sensible. Every time BlackRock has de-rated sharply, it has been because of temporary concerns (2020 pandemic, 2022 bear market, now the private credit scare) rather than fundamental business impairment. Dollar-cost averaging into these dislocations has rewarded patience. Third, the dividend provides a genuine annual return of 2.4% that compounds the total return, and the 16-year growth streak is a credible signal that management is confident in future earnings.

A staged entry, owning a meaningful core position now and adding if the stock tests the $850–900 zone, is the pragmatic approach. It captures the upside from the technology re-rating whilst managing the downside from any worsening in the private credit backdrop.

Competitive positioning has fundamentally changed

BlackRock’s competitive landscape has bifurcated. In public markets, it competes with Vanguard ($12T AUM) and State Street ($5.7T) on the passive front, where its product breadth (1,700 ETFs) and geographic reach provide structural advantages. In private markets, the comparison set shifts to Blackstone (~$1T+), Apollo (~$750B), KKR (~$550B+), and Carlyle (~$420B). BlackRock’s $675 billion in alternative client assets now positions it as a top-5 global alternatives platform, though with shorter track records than specialists.

The differentiated strategy is integration. Apollo and KKR have pursued insurance balance sheet models (Athene, Global Atlantic) to create permanent capital. Blackstone remains fee-based and capital-light. BlackRock’s approach is unique: combining the world’s largest public markets franchise, the dominant technology platform, and a rapidly scaling private markets capability into a single operating system for institutional and retail investors. The Executive Carry Program (tying executive compensation to private markets performance) adopted in January 2026 signals commitment to competing directly with pure-play alternatives managers on talent.

The contrarian case centers on timing and entry price

The market is pricing two legitimate fears: that private credit is cracking and that BlackRock bought at the top. Both deserve respect. But the structural case is stronger. BlackRock has transformed from a passive index manager with a ~20 basis point average fee into an integrated platform where incremental flows generate 6–7x higher fee yields. Aladdin’s $2 billion recurring revenue base with 98% retention creates technology lock-in no competitor can replicate. The retirement market (Fink’s 50/30/20 vision, private assets in 401k’s) represents a multi-decade addressable market expansion.

At ~17.5x forward earnings with a PEG ratio near 1.0, a 2.4% dividend yield growing at 9%+ annually, and consensus calling for $54+ in 2026 EPS, the risk-reward skews favorably for patient capital. The most underpriced element is the Aladdin + Preqin technology convergence. A private markets data and analytics monopoly being valued as a traditional asset management fee stream. The most overpriced risk is ESG politics, which has cost ~$13 billion against $14 trillion in AUM. The genuine risk to monitor is whether the HLEND gating episode is an isolated liquidity management event or the leading edge of a broader private credit downturn that could impair BlackRock’s transformational thesis. Next earnings on April 10, 2026 will be the critical data point.

A Closing Thought

There is a version of the financial world in 2030 where every major pension fund runs its portfolio on Aladdin, allocates 20% to private markets managed by BlackRock, and indexes its public holdings through iShares. In that world, BlackRock is worth substantially more than today. There is another version where private credit defaults mount, Fink retires unexpectedly, the Texas lawsuit damages the indexing business, and integration costs erode margins for three years. In that world, BlackRock is worth less than today.

The task of the investor is not to predict which version occurs. It is to assess the probability-weighted range of outcomes, compare them to what the current price implies, and decide whether the risk/reward is worth taking.

At $960, the maths says: reasonably. The four frameworks say: yes, with patience and awareness.

BlackRock is the kind of business you want to own for decades. The price you pay determines how much of that quality you actually capture. Today, the gate is open, and the entry price is better than it was six months ago. The question, as always, is how long you are willing to wait for the story to complete itself.

Related Essays:

The Wrong Clock

Every morning I watch the same thing happen.

IGLG, my iShares physical gold ETC opens, and within the first hour it drops. Not dramatically. Not in a way that would alarm anyone who wasn’t paying attention. Just a quiet, steady bleed of one, sometimes two percent. Then, sometime after lunch, it begins to recover. By the afternoon close it has often regained everything it lost, and occasionally a little more.

For weeks I watched this pattern repeat and asked myself the obvious question. Is something wrong with gold? Is the market telling me something I am missing? Is sterling moving in a way that is quietly undermining my position?

The answer, as it turned out, was structural and entirely mechanical. The GBP hedge in IGLG resets daily using the prior day’s closing exchange rate. If sterling has moved overnight as it frequently has in recent months, the ETC is briefly mis-hedged at the London open, creating a small but visible drag on the sterling price that resolves through the day. The LBMA’s morning gold fix brings concentrated commercial selling pressure into the early session. And New York institutional money, broadly bullish on gold for reasons I will come to, provides the afternoon bid that lifts prices back. The morning dip is not a signal. It is plumbing. It is arithmetic. It tells you nothing about the direction of gold over the next decade.

But here is what stayed with me after I understood the mechanics. I had been asking the wrong question. I had been watching the clock on the wall when I should have been reading the calendar on the desk.

This is, I have come to believe, the central error most investors are making right now. Not just with gold. With everything.

We are living through one of those rare moments in financial history when the entire frame of reference shifts. When the rules that governed the previous era quietly stop applying, and the rules of a new era begin. Slowly, then all at once. These transitions are almost never obvious in real time. They are only obvious in retrospect, when historians can draw the clean line and say: there. That is where the old world ended and the new one began.

The last time this happened at scale, the year was 1968.

Lyndon Johnson’s Great Society was in full spending mode. Medicare, Medicaid, anti-poverty programmes, education spending. All of it financed not by taxation but by debt and monetary expansion, while simultaneously funding a ground war in Vietnam that was consuming billions with no end in sight. The Bretton Woods system, which had anchored the global economy since 1944 by fixing the dollar to gold at $35 an ounce, was quietly breaking under the strain. Too many dollars had been printed. Too little gold remained to back them. France, ever alert to American financial sleight of hand, was already sending ships to New York to redeem its dollar reserves for physical bullion.

In August 1971, Nixon closed the gold window. The dollar was cut loose from gold entirely, becoming a fiat currency backed by nothing but faith in the United States government. The last link between money and something real was severed. What followed was thirteen years of monetary disorder: inflation that rose from 1.4% in 1960 to 13.5% by 1980, two catastrophic oil shocks, a stock market that lost nearly half its real value over the decade, and bond markets that were quietly destroyed as yields climbed from 6% to over 15%.

The investors who navigated that period well were not the ones who reacted to every headline. They were not the ones watching their portfolios tick up and down on a daily basis and adjusting accordingly. They were the ones who had understood, early enough, that the regime had changed. That the portfolio which had served them through the postwar boom. Heavy in financial assets, light in real ones. Now precisely the wrong portfolio for the world they were actually living in.

The investors who did not understand this held on to their growth stocks and their long bonds and watched inflation eat them quietly, year after year, until by 1980 the Dow Jones had gone nowhere in nominal terms for a decade and absolutely nowhere in real ones.

I think about that investor often. Not as a historical curiosity. As a warning.

Because right now, today, the average investor’s portfolio looks remarkably similar to the one that failed between 1968 and 1981. It is built around financial assets. Technology stocks, growth equities, long-duration bonds. In a world where those assets are priced for a low-inflation, low-interest-rate, US-dominated, unipolar future. A world, in other words, that may already be behind us.

Energy, materials, and industrials make up just 14% of the S&P 500 today. In the early 1980s, at the peak of the commodity supercycle, they made up close to 50%. That gap is not a curiosity. It is the setup.

This essay is an attempt to understand it. Not to predict the future. No one can. But ask the question that I think deserves more serious attention than it is currently receiving: if we are in the early stages of a regime change comparable to 1968, what does a sensible portfolio look like for the next twenty years? And how different is it from the one most of us are currently holding?

The answer, I suspect, will be uncomfortable. Let us get into it.

The Largest Shock the Oil Market Has Ever Seen

Most people, when they think about the oil price, think about the number on the pump. They think about what it costs to fill the car, or what the headline says when Brent crude moves above a round number. They do not think about what happens when the oil stops moving altogether.

That is what is happening now.

The Strait of Hormuz is a narrow channel of water. At its tightest, just twenty-one miles wide. This sits between Iran and the Arabian Peninsula.

Through it passes somewhere between ten and fifteen million barrels of crude oil every single day. Saudi crude bound for Asia. Emirati oil heading to European refineries. Qatari liquefied natural gas destined for power stations in Japan, South Korea, Germany. The strait is not just a shipping lane. It is the circulatory system of the global industrial economy. And right now, it is closed.

The International Energy Agency has confirmed that global oil output fell by approximately eight million barrels a day in March. To put that in context: the two OPEC oil embargoes of the 1970s (the events that defined a decade of inflation and economic misery for the Western world) disrupted supply by a fraction of this magnitude. The COVID demand collapse of April 2020, when oil briefly went negative, was a demand shock, not a supply one. Nothing in the recorded history of the global oil market compares to what is currently unfolding in the Strait of Hormuz.

Adam Rosenwag, managing partner at Garing and Rosenwag, put it plainly in a recent conversation: somewhere in the order of 230 million barrels of oil have already been impacted since the closure began. And the strait remains shut as this is written. When it does reopen. And it will. Those barrels will not be retroactively produced. They simply will not exist.

The IEA’s coordinated release of 400 million barrels from member nations’ strategic petroleum reserves sounds substantial. It is less so when you understand the plumbing. The realistic maximum flow rate from the combined OECD reserves is somewhere around four million barrels a day. Against a daily shortfall of ten to fifteen million barrels, that leaves an enormous hole. Every single day.

The Second-Order Effects Nobody Is Talking About

The Strait of Hormuz does not just carry oil. It carries the raw material inputs for the entire real economy. Urea fertiliser. The nitrogen base for a significant portion of global crop production passes through the strait. Phosphates, another critical agricultural input, pass through the strait. Sulphur, aluminium, liquefied natural gas, petrochemical feedstocks. All of it moves through that twenty-one-mile channel.

The disruption to these flows is already registering. Urea fertiliser prices have risen approximately 28% since the closure began. Diesel has crossed $5 a gallon just as American farmers are entering planting and seeding season. Larry McDonald, founder of the Bear Traps Report, framed the second-order effects precisely: the disruption to supply chain trust across the Gulf region will take sixty to ninety days to begin normalising even after the physical flows resume.

Bunkering fuel. The heavy fuel oil that powers cargo vessels is becoming scarce in key Asian ports. If bunkering fuel becomes broadly unavailable, the secondary disruption to global trade could rival the supply chain seizure of the COVID era. That disruption was inflationary for the better part of three years.

The oil shock is the headline. The food inflation shock, the fertiliser shock, the shipping disruption. These are the second chapter. They are arriving now. Most people have not yet noticed them at the supermarket till.

The Hidden Fire

There is a particular kind of danger that arrives not with a bang but with a series of small, seemingly unconnected events that nobody is paid to join together.

In 2007, oil was rising. The headlines were about oil. Meanwhile, in the unglamorous back offices of mid-tier American mortgage lenders, something else entirely was beginning to come apart. Nobody was talking about it. Everyone was watching oil. By September 2008, Lehman Brothers was gone.

I do not think we are in 2008. The structural differences are real. But I do think we are in a moment that rhymes with the eighteen months before 2008 in one very specific and very important way: the thing most people are watching is not the thing most likely to break the system.

Everyone is watching the oil price. The real risk is in private credit.

To understand why, you need to understand how large this market has become and who is sitting inside it without knowing it.

The global private credit market reached $3.5 trillion in assets under management by the end of 2024, up 17% from the end of 2023. At the start of 2025 it stood at $3 trillion, compared to roughly $2 trillion in 2020, and is estimated to grow to approximately $5 trillion by 2029. To put those numbers in context: the entire US subprime mortgage market at the peak of the 2008 crisis was approximately $1.3 trillion. Private credit is already more than twice that size, growing faster, and significantly less transparent.

Private credit firms can report the value of their loans according to their own internal models, rather than the market price. Marking to model instead of marking to market. It is nearly impossible to verify that the loans are worth what they are claimed to be.

Now consider who is holding this $3.5 trillion. It is not a small circle of sophisticated hedge funds. Large institutional investors such as pension funds and insurance companies hold substantial exposure. Nearly half of the increase in average Level 3 assets among selected pension funds, representing 17.5% of global pension assets, is linked to growing allocations to private credit since 2016. Private credit now accounts for about 35% of the investment portfolios of North American insurance companies. Your pension. Your annuity. Your life insurance policy. These are the vehicles through which ordinary people have been quietly and unknowingly allocated to the riskiest corner of an unregulated shadow banking system that is now under strain.

Jeffrey Gundlach, the billionaire bond investor, has predicted that the next financial crisis will come from private credit. Jamie Dimon warned that when you see one cockroach, there are probably more.

The oil shock is tightening financial conditions. Tightening financial conditions are stressing private credit borrowers. Stressed borrowers are triggering redemption gates. Redemption gates are forcing institutions to sell liquid assets. Selling liquid assets depresses public equity markets. Depressed markets reduce consumer wealth and spending. Reduced spending deepens the recession that was already forming before the first missile was fired.

Each step in that chain is already under way. We are not predicting a catastrophe. We are describing a transmission mechanism that is operating in real time, largely out of sight, while everyone watches the oil price.

The oil shock is the fire everyone can see. Private credit is the fire in the engine.

Gold’s Two Clocks

Let me start with the question everyone is asking. If the world is coming apart at the seams. A war closing the most strategically important waterway on earth, a private credit market of $3.5 trillion under quiet but accelerating stress, a financial system running on debt that it cannot plausibly repay. Then why is gold not at $8,000 an ounce?

The answer is both more instructive and more interesting than either the bulls or the bears typically acknowledge.

Gold runs two clocks simultaneously. They are set to completely different time zones. And the single greatest source of confusion among investors is the habit of checking the wrong one.

The short-term clock measures sentiment, positioning, leverage, and liquidity. It responds to margin calls, to momentum reversals, to the behaviour of fast money that entered the trade for the wrong reasons and will exit it just as quickly. The long-term clock measures something entirely different. It measures the accumulating weight of monetary debasement, fiscal recklessness, geopolitical fragmentation, and the gradual erosion of institutional trust.

The Short-Term Case: The Flush Before the Run

Gareth Soloway, one of the more rigorous technical analysts operating in the precious metals space, has made what he freely acknowledges is an uncomfortable call. He expects gold to pull back to approximately $3,500 an ounce by year end. He holds this view not because he is bearish on gold’s long-term fundamentals, but because he is honest about what the short-term chart is telling him.

His argument: the final surge in gold’s price had the hallmarks of momentum-driven, emotionally-fuelled buying. When financial conditions tighten and a credit event forces deleveraging, fear does not discriminate. Gold, as one of the most liquid assets an investor holds, gets sold first. We saw this in March 2020. Gold fell sharply. Not because its fundamentals had deteriorated, but because it was liquid and everything else was not. It recovered within weeks and went on to new highs.

Soloway’s response to the bullish structural arguments is characteristically direct: he agrees with them entirely. What he disputes is the assumption that institutional money is immune to fear in a genuine deleveraging event. Both positions can be simultaneously correct. That is the point.

The Long-Term Case: $10,000 and the Paper Coverage Argument

Adam Rosenwag offers the most intellectually rigorous framework for understanding where gold goes over the next decade. All the gold ever mined in human history is worth approximately $30 trillion at current prices. Roughly half the market capitalisation of the S&P 500. Historically, major gold bull markets have ended with gold’s total value somewhere between one and a half and two times the S&P 500. On a flat S&P, that implies a gold price of around $10,000 per ounce.

The second data point is the coverage ratio. At current prices, every US dollar in circulation is backed by approximately 25 cents worth of gold in US reserves. Historically, gold bull markets have ended when that coverage ratio reaches or exceeds 100%. Getting from 25 cents to full coverage requires roughly a fourfold increase in the gold price. Taking you, from approximately current levels, to somewhere north of $10,000.

Larry McDonald adds the policy dimension that gives the long-term gold case its most uncomfortable edge. With $38 trillion in US government debt growing by approximately $2 trillion a year, the only realistic exit from this debt trap is to allow inflation to run persistently above the real return on government debt. Gradually eroding the real value of the liability. This is financial repression. Gold is the asset that financial repression cannot touch. It has no coupon to inflate away. It has no maturity date. It is nobody’s liability.

The short-term clock may be flashing caution. The long-term clock is pointing somewhere most investors are not yet looking.

What the Last Great Inflation Regime Can Teach Us

In the summer of 1968, the average American investor had no particular reason to worry. The postwar economic boom had delivered two decades of extraordinary prosperity. The conventional investment portfolio had performed brilliantly for a generation. There was no obvious reason to question a portfolio heavy in equities, supported by bonds, denominated in dollars.

What that investor could not see was the slow accumulation of forces that were about to make that portfolio precisely the wrong one for the world they were about to inhabit.

What Actually Happened to Portfolios

The S&P 500, adjusted for inflation, lost nearly half its value over the 1970s. The Dow Jones Industrial Average had a nominal gain of thirty points across an entire decade. An investor who held a balanced 60/40 equity and bond portfolio from 1968 to 1981 and did nothing else would have ended the period with roughly the same nominal wealth and substantially less real wealth.

The bonds were destroyed by rising yields. 10-year Treasury went from 6% in 1970 to over 15% by 1981. The equities were destroyed by valuation compression. The the premium growth stocks of the late 1960s, the Nifty 50 saw their prices cut by 60 to 90% in the 1973-74 bear market alone.

What Worked

Gold rose from approximately $35 in 1970 to approximately $850 at its January 1980 peak. An insignificant gain (😂) of over 2,300%. In real, inflation-adjusted terms, gold delivered approximately 9% annual returns throughout the stagflation period. Nothing else came close.

Broad commodities delivered a total return of approximately 586% across the 1970s. Oil prices rose over 1,000%. Agricultural commodities. Wheat, corn, soybeans more than tripled. Copper and industrial metals posted large gains. Farmland appreciated from $137 per acre in 1970 to $737 per acre by 1980. Real estate delivered approximately 13% nominal annual returns.

Within equities, value stocks outperformed growth stocks significantly. Defensive sectors of consumer staples, healthcare, utilities held up far better than economically sensitive areas. Energy equities were among the best performers in the stock market. The sector that the broader investment community had ignored and underfunded through the 1960s became the dominant return driver of the following decade.

The Translation: What the 1968 Playbook Looks Like Today

The structural similarities between 1968 and now are substantial. Both began with prolonged fiscal excess. Both involved a monetary system under strain from too much debt and too much printed money. Both were characterised by geopolitical fragmentation and the erosion of US dominance. Both featured energy shocks. And both arrived after a long period of capital starvation in real assets.

Gold and silver remain the foundational monetary hedge. Gold is not a trade. It is insurance against the long-term consequences of a financial system that has borrowed more than it can repay and will choose inflation over default as its preferred exit route.

Energy is the most direct translation from the 1970s playbook. Oil producers, natural gas companies, and offshore drilling operators are priced for $70-75 oil over the long term. The structural supply deficit that predates the Hormuz closure suggests long-term prices will settle considerably above that level.

Uranium represents the cleanest long-duration energy thesis available to a retail investor. Nuclear power plants will pay almost any price to keep their reactors fuelled. New mine supply is structurally insufficient. The price would need to roughly double to incentivise sufficient new mine construction.

Coal is the investment nobody wants to discuss in polite company and one of the most compelling valuations in the real asset complex. Global coal demand is growing. Coal companies have been effectively debanked. In every commodity bull market, the most hated sector at the bottom tends to produce the largest gains by the peak.

Agricultural commodities are the second-order consequence of the Hormuz closure that most investors have not yet fully priced.

Short-duration fixed income rather than long bonds. The destruction of long-duration bond holders was one of the defining features of the 1968-1981 period. Long bonds in a financial repression regime are not a safe haven. They are slow-motion wealth destruction dressed up in the language of safety.

Bitcoin warrants a small but considered allocation as a new-era monetary hedge. Over a ten-year horizon, in a world of persistent monetary debasement and growing institutional adoption, a modest allocation — in the region of 3 to 5% — provides asymmetric upside that the traditional real asset portfolio cannot replicate.

The Scarcity Thesis

There is a framework I have been developing through the pages of this Substack for some time now, and it sits at the heart of everything this essay has been building towards.

The world is dividing into two categories of things. Things that are becoming abundant, and things that remain scarce. Stubbornly, structurally, irreversibly scarce.

Software intelligence is becoming abundant. The cost of a unit of computation has fallen by orders of magnitude over the past decade and continues to fall. But intelligence runs on atoms. And atoms are not becoming abundant. They are becoming scarcer.

Every data centre that trains a large language model consumes enormous quantities of electricity, water, copper, and rare earth elements. Every electric vehicle requires lithium, cobalt, nickel, and manganese in quantities the current mining system is nowhere near equipped to supply at scale. The AI era is not a dematerialisation of the economy. It is a rematerialisation. A massive, accelerating increase in the demand for physical infrastructure at precisely the moment when capital investment needed to supply it has been systematically withheld for the better part of a decade.

Where Scarcity Lives

Uranium. The world’s nuclear fleet requires a steady supply of enriched uranium. Data centres are turning to nuclear as the only baseload clean energy source with the density and reliability they need. The demand signal is unmistakable. The supply response is structurally constrained.

Copper. The energy transition requires copper in quantities that have no historical precedent. Every electric vehicle uses roughly three to four times the copper of its combustion equivalent. The world’s copper mines are ageing, their ore grades are declining, and new mine permitting is subject to political risk that is structurally increasing.

Silver. Unlike gold, silver is industrial in ways directly tied to the technologies the world is building. Solar panels require silver. Every semiconductor requires silver. Defence electronics require silver. Silver is largely a byproduct of lead and zinc mining, meaning its supply cannot respond quickly to demand signals.

The abundance of intelligence is being constructed on a foundation of scarce atoms. And those scarce atoms are controlled, in significant proportion, by the very countries and regimes that the West is simultaneously trying to contain.

The AI buildout is the single largest driver of demand for the physical commodities that the technology sector spent the previous decade telling us we no longer needed. Natural gas to power the data centres. Copper to wire them. Silver for the semiconductors. Uranium for the nuclear plants. This is not a temporary tension that technology will eventually resolve. It is a structural feature of the physical world that no amount of software can engineer away.

Conclusion: The Cards I Was Dealt

I want to be honest with you about something before we part ways on this essay.

Everything I have written here. The oil shock, the private credit unwind, the 1968 parallel, the scarcity thesis. I did not arrive at from the comfort of a detached analytical position. I arrived at it from the very specific discomfort of watching my own savings move up and down on a screen every morning and asking myself whether I understood what I was actually holding and why.

That is where this essay began. Not in a library. Not in a trading room. At my London desk watching IGLG dip one percent before 9am and wondering whether the pattern meant something I was missing.

It did not. But the process of investigating it led me somewhere more important than the answer to that specific question. It led me to ask the bigger one: is my portfolio built for the world that exists, or for the world that used to exist?

The answer, when I looked honestly, was both. And that is probably true for most of you reading this.

Here Is Where I Actually Stand

Roughly half of what I hold. Just over 51%. Expresses the thesis I have spent this essay building. Gold in two currencies, across two continents. Uranium. Global mining. Commodities. Infrastructure. Coal India. The Multi Commodity Exchange. Silver, in modest but growing size. Bitcoin as the new-era monetary hedge. The FTSE 100 as the most underappreciated expression of real asset equity available to a British retail investor. These positions were not all assembled with the 1968 framework consciously in mind. Some of them were instincts that the framework has since given language to. Others were deliberate. All of them, viewed through this lens, belong.

About 14% sits in the middle. Partially aligned, carrying some real asset characteristics alongside enough contradictions to make me honest about their limitations. Indian equities with genuine infrastructure and demographic tailwinds but financial sector drag. Semiconductors priced for an AI capex cycle that may be later in the innings than the market believes.

And then there is the remaining 34%. The 2010-2020 portfolio. The NASDAQ exposure. The global financial sector ETF. The all-world trackers. The megatrends fund. The asset management equity. These are not bad investments. They are investments built for a world of falling inflation, cheap capital, and uncontested American financial dominance. A world I am no longer certain we live in.

I am not selling the misaligned positions tomorrow. Panic-driven portfolio reconstruction is one of the most reliable ways to crystallise losses and miss recoveries. Being early and being wrong look identical for long, uncomfortable stretches. The 1968 investor who rotated too aggressively in the first year would have faced several years of doubt before the thesis vindicated itself.

What I am doing is something more deliberate and less dramatic. I am letting the misaligned positions run their natural course and directing new capital exclusively towards the aligned bucket. Towards the scarce atoms. Towards the assets that have no maturity date to inflate away, no coupon to erode in real terms, no sovereign’s balance sheet underpinning their value.

I am also sitting with the uncertainty of it openly, which is something I think financial writing rarely does honestly enough. I do not know when the private credit market tips from stress into crisis. I do not know whether the Hormuz closure lasts three weeks or three months. I do not know whether gold’s short-term flush reaches $3,500 or finds a floor well above it. Nobody does. The analysts whose voices have shaped this essay — Rosenwag, Soloway, McDonald, Skovdal — disagree with each other on timing even as they converge on direction. That disagreement is not a weakness in the thesis. It is the honest condition of anyone thinking seriously about a future that has not yet arrived.

The dam is higher than it has ever been. The water behind it is deeper. The pilot is still speaking calmly over the intercom. And the long-term clock, if you are willing to look at it honestly, is pointing somewhere that the daily price chart cannot yet see.

Now I want to ask you something directly.

Look at your own portfolio. Not the performance numbers. Those are the short-term clock, and we have established what that clock is and is not measuring. Look at the composition. Look at what you actually own, sector by sector, asset class by asset class. Ask yourself what world that portfolio was built for. Ask yourself whether that world still exists in the form it did when you assembled those positions.

If the answer gives you pause. If there is a gap between the portfolio you hold and the thesis you believe. Then the question is not whether to act. It is how deliberately, how patiently, and how honestly you are willing to act on what the evidence is telling you.

The 1968 investor who asked that question early enough and answered it honestly did not just preserve their wealth through one of the most destructive monetary periods in modern history. They built it. Not because they were lucky. Not because they had access to information the rest of the market lacked. But because they were willing to read the long-term clock, hold their conviction through the noise, and position for the world that was arriving rather than the one that had already passed.

That option is still available. The clock is still readable. The question is whether you are looking at it.

Nothing in this essay constitutes financial advice. I am a retail investor and pattern-watcher writing from London, not a licensed financial adviser. The views expressed here represent my own analytical framework, informed by the thinking of the investors and analysts cited throughout, and are offered as a starting point for your own research rather than a prescription for action.

References:

Glossary of Terms

Before diving into the essay itself, I want to flag some of the economic and geopolitical language I have had to learn to make sense of this analysis. These are not terms I grew up with. If they are new to you too, this section is worth reading first.

K-shaped economy

An economy where the recovery or growth after a shock splits into two diverging paths. Those who own assets (property, stocks, businesses) see their wealth rise. Those who rely on wages see their living standards stagnate or fall. The letter K captures the shape: one arm pointing up, one pointing down. The pandemic accelerated this split dramatically, but the structural causes go back decades.

Petrodollar

The system established in the 1970s whereby global oil is priced and traded in US dollars. This arrangement was negotiated between America and Saudi Arabia after the US came off the gold standard in 1971. It means that any country wanting to buy oil must first acquire dollars, creating a permanent global demand for American currency and, by extension, American Treasury bonds. It is the foundation of dollar supremacy in the post-Bretton Woods era.

Financial-Industrial Complex (FIC)

Simon Dixon’s term for the network of global investment banks, asset managers, private equity houses, sovereign wealth funds, and the policy institutions they influence. Distinct from the Military-Industrial Complex in that its primary profit motive is not war itself but the financing of war, the management of capital flows, and the reconstruction and privatisation that follows conflict. Dixon argues the FIC now sits above the MIC in the hierarchy of global power.

Military-Industrial Complex (MIC)

A term coined by US President Dwight Eisenhower in 1961 to describe the relationship between the defence industry, the military, and the political establishment. The MIC profits from weapons production, military contracts, and geopolitical conflict. Its economic model requires ongoing threat, ongoing spending, and ongoing war. Dixon uses it to describe the faction currently being displaced by the FIC in the global power hierarchy.

Technical-Industrial Complex (TIC)

This is Dixon’s term for the network of big technology companies, surveillance infrastructure builders, data brokers, and algorithmic media platforms that have grown into a third pillar of global power alongside the MIC and FIC. Think of the companies building AI systems, facial recognition, predictive policing tools, social media algorithms, and central bank digital currency infrastructure. The TIC profits not from war or capital flows directly, but from data, attention, and control. Its product is the architecture of a managed population. Dixon sees it as increasingly intertwined with both the FIC and the state, providing the surveillance and behavioural control layer that sits on top of the financial and military layers beneath it.

Proof of Weapons Network

Dixon’s framework for understanding how the FIC, the MIC, and the TIC operate together as an interlocking system. He uses the analogy of Bitcoin’s proof of work as the resistance mechanism against this network, arguing that financial self-sovereignty is the individual’s defence against a system built on debt, surveillance, and coercion.

Financialisation of the economy

The process by which financial instruments, financial institutions, and financial motives come to dominate economic activity that was previously driven by production, manufacturing, or trade. In a financialised economy, the most profitable activity is not making things but managing, trading, and leveraging claims on things. America’s economy is now roughly 70 per cent services, with a large portion of that being financial services. This hollowing out of the productive base is a central theme in understanding why the US is in long-term relative decline.

Multipolar world

A global order in which power is distributed among multiple major actors rather than concentrated in one (unipolar, as under US dominance post-1991) or two (bipolar, as during the Cold War). The emerging multipolar world features the US, China, Russia, India, and the Gulf states as distinct power centres with their own spheres of influence, currencies, and trade arrangements. Much of the current geopolitical turbulence, from tariffs to the Iran conflict, can be understood as the managed or contested transition from unipolarity to multipolarity.

Fiscal dominance

A condition in which a government’s debt levels become so large that monetary policy is effectively subordinated to fiscal needs. Rather than the central bank independently setting interest rates to control inflation, it is forced to keep rates low (or to print money) in order to make the government’s debt serviceable. The US is argued by many macro analysts to be approaching, or already in, a state of fiscal dominance. It implies that inflation is a policy choice, not an accident.

Force majeure

A legal clause in a contract that allows one or both parties to suspend their obligations in the event of extraordinary circumstances beyond their control, such as wars, natural disasters, or pandemics. In the context of the Iran conflict, private credit funds invoked force majeure to suspend redemptions for retail investors, effectively locking their money in place. The term literally means superior force in French.

Strategic petroleum reserve (SPR)

Government-held emergency stockpiles of crude oil, maintained by major consuming nations to buffer against supply disruptions. The US SPR, stored in underground salt caverns in Louisiana and Texas, is the largest in the world. Releasing SPR supplies into the market is one of the few tools governments have to quickly suppress an oil price spike. The Biden administration drew the SPR down significantly during the Russia-Ukraine energy crisis, leaving it at historically low levels heading into the Iran conflict.

De-dollarisation

The gradual process by which countries reduce their dependence on the US dollar for trade, reserves, and debt. This includes settling bilateral trade in local currencies, accumulating gold instead of Treasury bonds, and building alternative payment systems to SWIFT. Russia and China have accelerated their de-dollarisation since Western sanctions froze Russian reserves in 2022, demonstrating to the rest of the world that dollar-denominated assets could be weaponised.

Private credit markets

The ecosystem of non-bank lending, where institutional investors (pension funds, endowments, insurance companies, and wealthy individuals) lend directly to businesses or projects through private funds, bypassing traditional banks. Private credit grew enormously after the 2008 financial crisis as banks retrenched from business lending. It now funds significant portions of the AI infrastructure buildout, leveraged buyouts, and real estate development. Unlike public bond markets, private credit is illiquid and opaque, making it a source of hidden systemic risk.

Proof of work / self-custody

Bitcoin concepts with broader philosophical implications in Dixon’s framework. Proof of work refers to the computational process by which Bitcoin transactions are validated, requiring real energy expenditure and making the system resistant to manipulation. Self-custody means holding your own Bitcoin private keys rather than trusting an exchange or institution to hold them for you. Dixon uses both as metaphors for financial sovereignty: the idea that individuals can opt out of a system designed to keep them in perpetual debt and dependency.

Genocide-as-a-service

A deliberately provocative term used in some alternative media analysis to describe the outsourcing of military violence by major powers to client states or proxy forces, with the sponsoring power maintaining plausible deniability. In the context of Gaza, critics use it to describe the role of US military and diplomatic support in enabling Israeli operations, arguing that the destruction served specific strategic objectives for multiple parties beyond the stated aims. I include it here not as an endorsement but because it appears in the analysis I am drawing on and readers deserve to know what it means.

How We Got Here: A Brief History of the System

Dixon’s analysis does not exist in a vacuum. It draws on a specific reading of economic and imperial history that is worth sketching briefly. Understanding this arc makes the current moment considerably less confusing.

1600s to 1700s: The East India Companies

The British East India Company and the Dutch East India Company were the first truly global corporations. They were privately owned, state-backed, and militarised. They did not just trade; they colonised, taxed, and governed entire territories. Their model, extracting resources from the periphery and concentrating wealth at the centre, became the template for all subsequent imperial expansion. Crucially, they privatised the profits of empire while socialising the costs through state military power. Dixon sees direct continuity between this model and the modern FIC.

1800s: The British Empire and the Bank of England

Britain’s industrial revolution and naval supremacy made sterling the world’s reserve currency. The Bank of England, nominally private, sat at the centre of a global financial web that channelled the wealth of empire into London. When the costs of empire became unsustainable, the same financial institutions that had profited from expansion found ways to profit from contraction, financing the wars that ended British hegemony and positioning themselves to survive the transition to American dominance.

1913 to 1945: The Federal Reserve, Two World Wars, and the Transfer of Power

The creation of the US Federal Reserve in 1913 established a privately owned central bank at the heart of American monetary policy. The same network of European banking families that had financed the British Empire helped establish it. Both World Wars accelerated the transfer of global financial power from London to New York, with America emerging in 1945 as the world’s dominant military and economic power, holding the majority of the world’s gold reserves.

1944 to 1971: Bretton Woods and the Dollar Standard

The Bretton Woods agreement made the US dollar the anchor of the global monetary system, with all other currencies pegged to the dollar and the dollar convertible to gold at $35 per ounce. This gave America extraordinary privilege: it could print the world’s reserve currency. When the costs of the Vietnam War and the Great Society programmes made the gold peg unsustainable, Nixon closed the gold window in 1971. This was, in effect, a default. It also freed America from any external constraint on money creation.

1973 to 1980s: The Petrodollar and Financialisation

With gold backing gone, the dollar needed a new anchor. The petrodollar system, negotiated with Saudi Arabia in 1973 to 1974, provided it. Oil priced in dollars meant permanent global demand for American currency. Simultaneously, the deregulation of financial markets in the late 1970s and 1980s unleashed the financialisation of Western economies. Manufacturing moved offshore. Capital markets grew. Debt became the primary driver of growth.

1990s to 2008: Globalisation, China, and the Hollowing Out

China’s entry into the World Trade Organisation in 2001 accelerated the offshoring of Western manufacturing at scale. China absorbed the industrial base, building the world’s largest manufacturing economy. Western financial institutions profited from the intermediation of this process. By 2008, the US economy was so financialised that the failure of a single class of mortgage-backed securities nearly brought down the entire global system. The bailout that followed transferred enormous wealth from taxpayers to financial institutions.

2008 to present: The End of the Unipolar Moment

Since 2008, the trajectory has been clear even if the pace has been debated. China’s GDP has grown from roughly 30 per cent of America’s to approaching parity. Russia, after the 2022 sanctions shock, has deepened its pivot towards Asia. The Gulf sovereign wealth funds have shifted from buying US government debt to buying US and global equities, changing their relationship with American power from creditor to co-owner. BRICS has expanded. The dollar’s share of global reserves has declined. The Iran conflict, on this reading, is not an aberration. It is the latest episode in a long and largely predictable story.

The War Behind the War

There is a version of the Iran conflict that most people are watching. Missiles. Drone strikes. Oil markets in freefall. Trump posting on Truth Social at 2am. It is dramatic, confusing, and designed to be both.

Then there is another version. Quieter. More deliberate. Harder to see unless you stop watching the missiles and start watching the money.

That second version is what macro analyst Simon Dixon has been mapping in real time. His framework is not comfortable. It will frustrate people who have strong feelings about any of the parties involved. But as an investor trying to understand what is actually driving markets right now, it is the most coherent lens I have found. This essay breaks it down, challenges it honestly, and draws out what it means for how we position our money.

Stop watching the missiles. Start watching the money.

Watch the entire Simon Dixon podcast here.

I. Two Industrial Complexes, One War

Dixon’s starting point is this: the real war in the Middle East is not being fought between Iran, Israel, and America. It is being fought between two factions of global power that use nation states as proxies.

The first is the Military-Industrial Complex. Its economic model depends on permanent conflict. War is not a failure state for the MIC. War is the product. It wants a forever war. The second is the Financial-Industrial Complex. Its model is different. It profits from financing wars, yes, but its real prize is reconstruction, privatisation, and the renegotiation of energy and capital flows that follows. The FIC wants a managed transition to a stable, profitable new order. It does not want forever war. Forever war is bad for asset prices.

Dixon argues the FIC is currently winning. And the Iran conflict is less a war than a controlled demolition of the old order, with the rubble already being priced into contracts before the dust settles.

The case for

The market behaviour during week two of the conflict is difficult to explain under any other framework. If this were a genuine all-out war, you would expect gold surging towards $6,000/oz, oil well above $150, and equity markets in severe decline globally. None of that happened. Oil was capped at $115 and crashed back to $89. Gold barely moved. The S&P 500 pulled back modestly. That is not how markets behave when they believe they are looking at World War Three. It is how they behave when the sophisticated money already knows, broadly, how the story ends.

Furthermore, the decision by risk insurers to pull coverage for ships but not for planes travelling through the same airspace is extraordinary. It implies coordination, not random market reaction. Insurance markets do not make that distinction by accident.

The case against

The MIC versus FIC framework is elegant but risks being unfalsifiable. Almost any outcome can be retrofitted into it. If the war escalates, Dixon can say the MIC is fighting back. If it de-escalates, he can say the FIC won. A framework that can explain everything explains nothing. Furthermore, attributing coordinated intent to thousands of independent actors across insurance markets, sovereign governments, and private institutions requires a level of centralised control that strains credibility. Markets can produce outcomes that look coordinated simply through shared incentives, without any single hand on the tiller.

My verdict: The MIC versus FIC lens is the most useful organising framework I have found for this conflict, but I hold it loosely. It explains the market signals better than any alternative narrative I have encountered. What it cannot do is prove intent. I treat it as a map, not a photograph.

II. The Strait of Hormuz: A Financial Weapon, Not a Military One

Roughly 20 per cent of all global oil, as well as significant volumes of liquefied natural gas, passes through the Strait of Hormuz. When it closes, energy markets convulse. When it stays closed, the global economy begins to fracture.

But the Strait was not physically blocked by naval force. What closed it was insurance. War risk underwriters withdrew coverage, making it commercially impossible for any vessel to pass through. At the same moment, thousands of flights were cancelled across the region, but aviation insurance was not pulled. Planes kept flying. Ships stopped sailing.

Dixon reads this asymmetry as deliberate. The financial system, he argues, did what missiles could not. And the insurance market, controlled by institutions at the centre of the FIC, made a strategic choice.

The initial market reaction was violent. Oil futures hit $115 from a pre-war price in the mid-fifties. Asian equity markets triggered circuit breakers. South Korea was particularly hard hit given its dependence on both energy imports and the semiconductor supply chains that underpin the global technology economy. Then came the reversal. Within days, oil crashed back to around $89, driven by Trump announcements, G7 SPR release pledges, and most significantly, Russian oil shipments moving again after quiet sanctions relief from the US Treasury.

America went to war with Iran and within days quietly lifted sanctions on Russia to keep oil prices down.

The case for

The insurance asymmetry is genuinely hard to explain without invoking deliberate choice. The risk profile for planes flying over the region was not materially different from the risk profile for ships sailing through it. Yet one market pulled coverage and the other did not. If this were purely a risk-based commercial decision, you would expect symmetry. The fact that it was asymmetric suggests something else was operating.

The Russia sanctions relief is equally telling. If the conflict were a genuine ideological confrontation with the axis of authoritarian states, lifting Russian oil sanctions within days would be political suicide. It happened anyway. The price ceiling was the priority, not the narrative.

The case against

Aviation and marine insurance are entirely separate markets with different risk models, different underwriters, and different regulatory frameworks. Ships travel slowly through a defined chokepoint and cannot be rerouted. Planes can change altitude, routing, and timing in real time. The asymmetry may reflect genuine operational risk differences rather than coordination. Lloyd’s of London withdrawing marine cover is not evidence of a conspiracy. It may simply be evidence that the marine risk was quantifiably catastrophic while the aviation risk, though elevated, remained manageable.

Similarly, the Russia sanctions relief may reflect pragmatic crisis management rather than a pre-arranged script. Governments under acute pressure use whatever tools are available. Russia happened to be a useful lever. That does not require advance coordination.

My verdict: The insurance asymmetry is the single most compelling piece of evidence Dixon presents. I cannot fully dismiss it. But I am cautious about reading deliberate orchestration into what may be the emergent outcome of many actors responding to shared incentives under pressure. The Russia sanctions relief, however, does require explanation. That decision was made at speed, at the highest level of the US Treasury. It is difficult to reconcile with the stated rationale for the conflict.

III. Russia is Quietly Winning

Of all the actors in this conflict, Russia’s position is the most straightforwardly advantageous. Before the war, Russian oil was selling at a heavy discount due to sanctions, roughly $35 to $40 per barrel. Now it is approaching $100. That is a near-tripling of revenue, with no additional costs and no military involvement required.

That windfall is not being recycled into US Treasury bonds. A de-dollarised Russia accumulates its gains in gold, in commodities, and in deeper partnerships with China and the broader BRICS bloc. Meanwhile, Russia uses its position as a swing oil supplier to exert direct influence over European energy security. Europe, having pivoted away from Russian gas, found itself dependent on Qatari LNG. Qatar is now constrained. That leaves Europe choosing between American LNG on predatory terms or renegotiating with Russia. Either outcome strengthens someone else’s leverage.

The case for

The numbers here are straightforward and verifiable. Russia’s oil revenues have risen dramatically. European energy vulnerability is real and documented. The structural logic of Russia’s position, sanctioned but indispensable, is not a conspiracy theory. It is the outcome of a decade of policy choices by Western governments that chose narrative over energy security. Germany shutting its nuclear plants while remaining dependent on Russian gas was not an accident of geography. It was a policy choice with entirely predictable consequences.

The case against

Russia’s gains are real but not without constraint. Elevated oil revenues do not easily translate into broader economic strength when your financial system is still largely cut off from Western capital markets, when brain drain is accelerating, and when the costs of the Ukraine war continue to compound. Russia is winning a narrow energy leverage game but losing a longer demographic and technological competition. And higher oil revenues for Russia do not necessarily mean Russia is directing events. A pickpocket who benefits from a crowd’s panic did not cause the panic.

My verdict: Russia is the clearest short-term winner from this conflict regardless of whether the managed transition thesis is correct. The revenue windfall is real. The leverage over Europe is real. I find the case for Russia’s strategic advantage more compelling and more verifiable than some of Dixon’s other arguments. For the investor, this does not mean buying Russian assets. It means recognising that the energy infrastructure bet is structural, not cyclical.

IV. The Multipolar Transition: What This Is Really About

To understand why Dixon frames this as a managed transition rather than a genuine war, you need the broader arc. The American empire, like the British Empire before it, was built on military dominance, financial control, and the ability to force subordinate nations to price their resources in dollars. That system required the Middle East to remain in perpetual tension. Unified, resource-rich nations in the Gulf or Africa were a threat to dollar supremacy. Divided, war-dependent ones were not.

China changed the equation. By becoming the manufacturing base for the entire world, including for the weapons systems of its rivals, China quietly made itself indispensable to everyone. American F-35 jets require Chinese rare earth minerals. European energy infrastructure requires Chinese components. The US military cannot replenish its munitions without Chinese supply chains.

As China’s weight grew, the Gulf sovereign wealth funds began redirecting capital. Rather than buying US Treasury bonds, they started buying American equities. Saudi Aramco took a board seat at BlackRock. Gulf capital flooded into AI infrastructure and asset management. In doing so, the Gulf states moved from being subordinates of American power to co-investors in it. And what they want from that leverage is not more war. It is regional stability, the exit of US military bases, the normalisation of Iran, and a resolution to the Palestinian cause.

China made itself indispensable to everyone, including its rivals. That was not an accident.

The case for

The structural shift in Gulf capital flows is documented and public. The Saudi Aramco IPO, the establishment of Jared Kushner’s Affinity Partners with Gulf sovereign wealth fund backing, and the broader pivot of Gulf money from US Treasuries to equities and co-investments are matters of public record. China’s normalisation deal between Iran and Saudi Arabia in 2023 was a genuine diplomatic earthquake that received far less Western media attention than it deserved. These are not speculative claims. They are verifiable facts that support the broad direction of Dixon’s argument.

Furthermore, the demographic and manufacturing reality of China’s position is hard to argue with. America’s share of global GDP has declined from over 40 per cent at the peak to around 25 per cent today. China now outproduces the US in shipbuilding, electric vehicles, solar panels, and increasingly in advanced electronics. The unipolar moment is ending regardless of anyone’s intentions.

The case against

Dixon’s version of the multipolar transition can veer into teleology: the sense that all of this was planned and is proceeding to a predetermined destination. History rarely works that way. The Gulf states have multiple and conflicting interests. Saudi Arabia and Iran were enemies for decades and remain deeply suspicious of each other despite the China-brokered normalisation. The idea that the Gulf sovereign wealth funds are operating as a unified bloc with a clear strategic vision overstates the coherence of a group of rival monarchies with very different domestic political pressures.

China’s rise is real, but China has its own structural vulnerabilities. An ageing population, a property sector in crisis, a youth unemployment problem, and a political system that increasingly concentrates risk in a single leader’s judgement. The multipolar world may be coming, but it is not arriving on schedule or according to plan.

My verdict: The direction of the multipolar transition is not seriously in doubt among serious macro analysts. Raoul Pal, Jordi Visser, and many others who do not share Dixon’s more conspiratorial framings arrive at essentially the same structural conclusion: the unipolar dollar era is ending and a more fragmented world order is emerging. Where I am cautious is in Dixon’s confidence about the pace and the coordination. Transitions of this scale are messy, contested, and full of unintended consequences. The destination may be correct even if the map of how we get there is wrong.

V. The Private Credit Warning

Alongside the oil market drama, Dixon flags a second stress signal that has received far less attention. Private credit markets are showing signs of acute distress. The mechanism is worth understanding.

Over recent years, private credit was opened up to retail investors who were told they could access institutional-grade returns with quarterly liquidity windows. Force majeure clauses were buried in the small print. When the war began, those clauses were invoked. Retail investors who expected to redeem in 90 days found their money locked. The funds, unable to meet redemptions, began selling assets at steep discounts. Dixon cites bids as low as 14 cents on the dollar for some positions.

The connection to the broader macro theme is this: a significant portion of the assets inside these funds are companies being disrupted by AI. SaaS businesses that once generated reliable cashflows are seeing those cashflows eroded as AI automates their core functions. The underlying assets are deteriorating at the same moment that redemption pressure is rising.

The case for

The private credit stress is not Dixon’s invention. Reports of major institutions restricting withdrawals and invoking force majeure clauses have appeared across financial media. The structural mismatch between illiquid assets and retail investor expectations of quarterly liquidity is a genuine and well-documented risk that regulators have flagged repeatedly. Raoul Pal and Jordi Visser’s analysis supports the broader point: the S&P 500 is roughly 50 per cent technology and communications. Materials and energy together account for around 5 per cent. That disproportion reflects years of capital flowing towards abundance assets over scarcity assets, and the rebalancing, when it comes, will be painful for those on the wrong side.

The case against

Dixon’s framing of the private credit stress as part of a deliberate FIC strategy to write down losses under cover of war is the weakest element of his analysis. The stress in private credit markets predates this conflict and has multiple mundane explanations: overleverage, duration mismatch, and the very ordinary consequence of a decade of cheap money funding unprofitable businesses. Blaming force majeure invocations on coordinated war-cover overstates the sophistication of what may simply be panicked institutional behaviour under stress.

My verdict: The private credit stress is real and worth monitoring. Its connection to AI disruption of SaaS business models is the insight I find most useful here, and it aligns with the broader scarcity versus abundance framework I have been developing. Whether it is being deliberately managed by the FIC or simply reflects market dysfunction under stress, the investment implication is the same: be cautious about assets whose cashflows depend on business models that AI is actively undermining.

VI. What the Markets Are Actually Saying

Dixon uses market behaviour as his primary evidence base, and this is where his methodology is strongest. In a genuine World War Three scenario, you would expect gold surging well above $6,000/oz. oil above $150/barrel with no ceiling in sight, equity markets in severe global decline, and the dollar weakening sharply. None of that has materialised.

His interpretation is that markets are pricing in a managed resolution. The sophisticated money is not positioned for armageddon. It is positioned for an offramp. And the timing of that offramp, Dixon believes, centres on the April China summit where Trump is expected to meet Xi Jinping.

The markets are not pricing in World War III. They are pricing in a deal.

The case for

Reading geopolitical risk through market positioning is a legitimate and well-established methodology. The gold signal is particularly hard to dismiss. Gold is the canonical flight-to-safety asset. In every genuine systemic crisis of the past fifty years, gold has spiked. Its relative flatness during this conflict, combined with Bitcoin’s resilience and the S&P’s modest pullback, is consistent with the idea that large, well-informed capital is not panicking. And large, well-informed capital tends to know things that retail observers do not.

The case against

Markets can be wrong. They can be manipulated. And they can be rational in the short term while missing longer-term discontinuities. Markets did not predict the 2008 financial crisis until it was already happening. They did not price the full severity of COVID until weeks in. Sophisticated investors using derivatives to suppress oil prices, as Dixon himself describes, can create market signals that look like calm but reflect managed suppression rather than genuine confidence. A market reading that rests partly on interventions by the very actors Dixon says are coordinating the outcome is, at minimum, circular.

My verdict: I find the market signals genuinely informative but not conclusive. They are consistent with the managed transition thesis. They are also consistent with a market that is simply not yet pricing in the full downside risk. What I take from the signals is a probability-weighted view: the most likely outcome, as priced by sophisticated capital, is some form of negotiated resolution. That is useful for positioning, even if it is not certainty.

VII. Where Does This Leave the Investor?

I want to be clear about what I am doing here. I am not endorsing every element of Dixon’s framework. What I am doing is taking the question seriously: if the broad structural logic is correct, what does that imply for how a retail investor should be positioned?

The answer aligns closely with the scarcity thesis I have been developing in my own writing. If we are in a multipolar transition, with energy flows being rerouted, supply chains being restructured, and physical infrastructure becoming the primary battleground of geopolitical competition, then the assets that benefit are those tied to things that are hard to replicate and impossible to print.

Energy Infrastructure

The Hormuz shock has reminded the world that energy security is not abstract. Countries that control physical energy flows hold structural leverage in a multipolar world. Integrated energy companies, midstream infrastructure, and LNG terminal operators are not exciting in a world of abundant cheap energy. They become essential in a world of contested supply chains. The Western race to rebuild energy independence creates long-duration demand for physical energy infrastructure that is not going away when this conflict ends.

Critical Minerals and Industrial Metals

If AI is the defining race of this decade, the physical inputs to AI are the scarcest assets in the world. Copper for power transmission. Silver for solar and electronics. Rare earth elements for semiconductors and defence systems. China’s dominance of rare earth refining capacity is a structural fact that will take years and enormous capital to change. Pal and Visser frame this well: the S&P 500 is structurally underweight in materials and energy relative to where capital flows need to go. This is not a cyclical trade. It is a decade-long structural squeeze.

European Defence

European nations can no longer rely on American security guarantees, and they are spending accordingly. The WisdomTree European Defence UCITS ETF and its 32 constituents represent the most direct liquid exposure to what may be the most durable budget cycle in a generation. The Hormuz crisis has added energy security to the list of strategic vulnerabilities Europe is racing to address, broadening the defence spending narrative beyond hardware into dual-use technology, intelligence infrastructure, and secure supply chains.

Gold

Gold has not spiked dramatically, which is itself a signal. But in the medium term, every tool being used to manage this crisis, reserve drawdowns, debt monetisation, fiscal dominance, dollar debasement, is structurally bullish for gold. Central banks in the Global South and BRICS nations have been accumulating gold at record rates precisely because they no longer trust the dollar system as a store of value. Gold is not a trade here. It is the hedge against the endgame.

Bitcoin

Dixon’s investment conclusion is that Bitcoin is ultimately a self-custody trade. Not because of its price today, but because the logical endpoint of the system he describes is a world in which the ability to hold an asset that no institution can freeze or confiscate becomes profoundly valuable. Private credit funds invoking force majeure to block redemptions are a mild preview of what capital controls look like in a financialised system under stress. Self-custody is not ideological in that context. It is rational portfolio construction.

The common thread across all five is physical scarcity. In a world where digital and software assets are being rapidly commoditised by AI, and where the physical infrastructure of the intelligence economy is becoming the real prize, the direction of capital rotation seems clear. The question is not whether to be positioned in scarcity. It is how much, and how patiently.

A Final Thought

I want to end where I began. I am not presenting this as established fact. The managed conflict thesis is a framework, not a verified history. Wars are chaotic. Factions defect. Plans go wrong. Dixon himself has said this conflict is moving faster and more unpredictably than he anticipated.

What I take from his analysis, held cautiously and challenged at every step, is a structural orientation rather than a specific prediction. The unipolar dollar era is ending. Physical scarcity is the investment theme of the coming decade. And the gap between what the headlines are saying and what the money is doing is, as always, where the most useful signal lives.

The fog of war is real. The fog of financial misdirection is thicker still. Following the money does not guarantee you will always be right. But it is a far better compass than following the headlines.

Credit: Framework and analysis drawn from Simon Dixon, Simon Dixon Hard Talk Live, March 2026, and BTC Sessions interview. Additional macro context from Raoul Pal and Jordi Visser. Historical context draws on publicly available sources.

Subscribe now

Related essays:

Right.

Where to begin?

The world this week has been simultaneously conducting a geopolitical crisis, a software power grab, a boardroom pay controversy, a private markets wobble, an EU policy fight, and because apparently that was not enough, a fifteen-day war in the Middle East that has closed the Strait of Hormuz and sent oil prices to a 22-month high.

I have read more FT &Bloomberg articles this week than is medically advisable. The good news is that all of it connects and that is what this newsletter is about.

This week’s MonkStuff covers:

1. four short essays on the themes shaping markets right now;

2. a note on the European Defence trilogy Ipublished this month; and

3. a six-video curated playlist on the Iran conflict from White House war advisors to ancient Indian palm-leaf manuscripts. Yes, really.

Grab something warm. Let’s get into it.

THIS WEEK IN THE WORLD

Day 15 of The Iran War

On 28 February, the United States and Israel launched coordinated strikes on Iran, killing Supreme Leader Ali Khamenei and triggering a regional conflict now entering its third week. Iran has responded with waves of drones and ballistic missiles across nine countries. At least 1,450 civilians are reported killed in Iran. The Strait of Hormuz through which roughly 20 per cent of the world’s oil passes has been severely disrupted. Dubai Airport has been struck. A UK military base in Cyprus was hit by an Iranian drone. The IEA released a record 400 million barrels of strategic reserves (mathematically this is not even a drop in the ocean because humanity consumes 100 million barrels per day). US gas prices have hit a 22-month high.

Mojtaba Khamenei the former Supreme Leader’s son has been named his successor. He does not currently hold the clerical rank of Ayatollah, which raises a genuinely interesting constitutional question about the man now running one of the most consequential theocracies on earth. More on that below. The experts on YouTube say he is far more inclined to having the nuclear weapon than his dad. So much for regime change.

I will not pretend to have a clean take. It is too fast-moving and too serious for neat conclusions.

What I can do is point you to interesting material I have found covering multiple perspectives via a curated reading and viewing list. You make your own take.

THIS WEEK’S ESSAYS

Four Short Essays. One Thread.

Four pieces published this week on Substack. They look like separate topics. They are not. The connecting thread running through all of them: intelligence and software are becoming abundant and cheap. Physical things remain scarce. Infrastructure, liquidity, atoms. Every one of these stories is, at its core, about that shift.

Essay 1

Microsoft and Anthropic announced a deal this week to integrate Anthropic’s AI agent Cowork into Microsoft’s Copilot platform. It was reported as a partnership. I think it is something more interesting: a slow-motion battle for the most valuable real estate in the coming decade of AI. The workplace itself. Copilot has only 3 per cent of Microsoft’s Office users paying for it. That is not a triumph. That is a company calling for reinforcements.

Meanwhile, Anthropic’s annualised revenue has doubled from $9 billion to $19 billion in three months. And Microsoft is currently riding the Elizabeth Line looking for a new London headquarters between Paddington and Canary Wharf, somewhere north of 200,000 square feet. Fitting, really. They are looking for a home at precisely the moment they are inviting a rival to share it.

The essay asks: who is actually inside whose ecosystem? The Trojan horse is already inside the gates.

Essay 2

Google gave Sundar Pichai a pay package worth up to $692 million this week. In the same breath, Alphabet announced it is lining up a 100-year sterling bond. A century. As in: we fully expect to be a going concern in 2126. Bold.

What caught my attention is that $350 million of Pichai’s incentives are tied not to search or cloud, but to Waymo and Wing. A self-driving car service and a drone delivery business. Physical infrastructure. Atoms. The old Google was built on abundance: search costs almost nothing to replicate. The new Google is quietly betting on scarcity. That pivot, hidden in a pay document, tells you more about where Alphabet thinks the world is going than any earnings call.

The essay argues: sometimes the most revealing document in a company’s filing is not the income statement. It is the CEO’s pay packet.

Essay 3

BlackRock sent a letter to investors in its $26 billion HPS Corporate Lending Fund this week. The letter said: we received $1.2 billion in withdrawal requests. We are paying out $620 million. For the rest, the gate is closed. BlackRock’s shares fell 5.1 per cent on the day. I opened my first position the previous day. Ouch! That hurt! KKR, Blue Owl and Ares all dropped more than 5 per cent.

This is the first significant crack in the everything-is-fine private markets narrative. Private credit was sold to retail investors as the sophisticated asset class they had previously been locked out of. High returns, quarterly liquidity. What was never emphasised is that the liquidity was always conditional. True liquidity is genuinely scarce. This week, that became visible.

The essay asks: did you read the small print? The gate was always in the documents. Most people just did not read that far.

Essay 4

The European Commission is preparing a tech sovereignty package. A plan to reduce Europe’s dependence on American cloud, software, and AI.

The problem: European businesses are quietly telling Brussels it cannot be done quickly, cheaply, or without competitive damage. Commerzbank, Deutsche Bank, Thyssenkrupp, and a major unnamed carmaker all said essentially the same thing: we depend on Microsoft and Google because nothing comparable exists in Europe, and switching now would hurt us.

They are right. But the deeper point is that Europe does not need to out-hyperscale Silicon Valley. It needs to out-apply it. Europe’s real competitive advantage was always atoms, not algorithms. Manufacturing, defence hardware, energy infrastructure, industrial automation. And in the AI era, that is precisely where value is moving.

The essay argues: Brussels is solving the wrong problem. Europe did not miss the AI revolution. It is sitting directly in the path of the most important part of it.

The European Defence Trilogy

My three-part series that goes considerably deeper on the macro thesis underpinning everything above. It covers the WisdomTree European Defence UCITS ETF and all 32 of its constituents.

PART ONE

The Genghis Khan bit. Where did the current world order come from, who built it, and why is Europe now rearming for the first time in a generation? Context first. Without it the rest does not make sense. If you have ever thought geopolitics was someone else’s problem, this is your entry point. It will not stay that way for long.

PART TWO

The speculative fiction piece. All 32 ETF constituents appear in a single dramatic narrative set on the AI-enabled battlefield of the near future. Rheinmetall, Thales, Leonardo, BAE and the rest. Each company’s technology translated into a scene. It is, I think, the most unusual thing I have written. Whether that is a compliment is entirely up to you.

PART THREE

The Goldman Sachs bit. A full financial analysis of all 32 ETF constituents. Operating margins, free cash flow yields, EV multiples, BUY / HOLD / MONITOR ratings, and five structural investment themes. If Part One is why and Part Two is what, Part Three is the how much and should I. Written for retail investors, not for fund managers though I suspect a few of those have found their way here.

CURATED READING

Iran: Read Both Sides

Before the playlist, a quick reading list. Two minutes of context on each side of the argument. The war looks very different depending on which news sources you trust.

THE WESTERN / US-ISRAEL PERSPECTIVE

→ CNN: What we know

→ UK Parliament House of Commons Library: Full briefing on the strikes

→ Wikipedia: 2026 Iran War

THE IRAN / REGIONAL PERSPECTIVE

→ Al Jazeera: Iran War

→ Al Jazeera: Death toll and casualties live tracker

THE SLIGHT DETOUR: WHO IS AN AYATOLLAH AND CAN I BECOME ONE?

Not unless you are a Shia Male Muslim.

Mojtaba Khamenei the new Supreme Leader does not currently hold the rank of Ayatollah. Which raises the question: what does it take to earn one? The answer is at least 30 years and a long arduous journey. There is no ceremony, no formal conferral, no graduation day. You study for decades at a seminary in Qom (Iran) or Najaf (Iraq). You master Islamic jurisprudence, theology and philosophy, write books, attract students, and gather followers. Eventually, if enough respected clerics and ordinary believers treat you as an Ayatollah, you become one. It is less like becoming a bishop and more like becoming a respected elder in an unwritten tradition. The title literally means Sign of God. Khamenei’s father was elevated quickly and controversially after the revolution. His son is starting without the title at all.

→ Slate: How do you become an Ayatollah?

→ Business Standard: The Shia clerical system explained

IRAN: AN INVESTOR’S DILEMMA

Six Perspectives. One Conflict.

Six videos that approach the Iran war from six completely different angles. They do not agree with each other. That is the point. The Carnegie video is the institutional lens. Simon Dixon is the financial-flows lens. Professor Jiang is the historical-patterns lens. McCormack and Modad is the power-structures lens. Robert Pape is the simulation lens. Craig Hamilton-Parker is, well, the dark arts lens.

THE ESTABLISHMENT LENS

Who’s Running Iran? — Carnegie Endowment for International Peace (1hr 3min)

Recorded on 5 March. A week after Khamenei was killed. This is Jon Bateman interviewing Karim Sadjadpour, I believe one of the most credible western Iran analysts alive. The question is exactly what it says: with the Supreme Leader gone after 37 years, who is actually in charge?

Sadjadpour describes the assassination as the political equivalent of a bunker buster. The conversation covers internal regime dynamics, the Revolutionary Guards, the possibility of popular uprising, and whether any exile opposition figures are credible alternatives.

Start here. Sober, well-sourced, and built on decades of serious research.

THE FINANCIAL FLOWS LENS

This Iran War Is the Global Reset — Simon Dixon — BitcoinHardTalk (3hr 7min)

Simon Dixon, Bitcoin investor and macro thinker, argues the conflict is not a conventional war but a pre-scripted geopolitical and financial transition, driven by transnational capital rather than national interest. His thesis: Iran gets absorbed into China’s orbit, Israel into the Gulf Cooperation Council, and the whole thing is a managed reset towards a multipolar world order.

Dixon is sharp on capital markets and genuinely provocative on geopolitics. This is the follow-the-money lens. Three hours long. Worth watching with one eyebrow firmly raised. Best for investors trying to understand who benefits economically and how energy markets, dollar hegemony and Bitcoin intersect with the conflict.

THE HISTORICAL PATTERNS LENS

Game Theory #9: The US-Iran War — Predictive History — Professor Jiang Xueqin (45min)

In May 2024, Professor Jiang made three predictions to his students: Trump would win the election, the US would go to war with Iran, and the US would lose. The first two have already come true.

Using game theory and historical pattern analysis, he argues the US faces a catastrophic war of attrition. Iran has been preparing for this specific conflict for twenty years. In their eschatology, this is literally a war against the Great Satan. He draws parallels to Athens in Sicily, Vietnam, Afghanistan, and Iraq. Wars where tactical precision masked strategic collapse.

Critics note his analysis relies on selective historical analogies. He now has 3.8 million views on this video. His third prediction, US defeat remains to be manifested. Watch it anyway.

THE POWER STRUCTURES LENS

The Elite Oligarchy Profiting from the Middle East War — Peter McCormack & Firas Modad (1hr 35min)

Firas Modad argues the real consequences of the Iran war are economic rather than military. Spreading through energy markets, shipping lanes, and commodity prices far beyond the battlefield. The conversation examines the role of donor networks, corporate interests, and what Modad describes as an oligarchy that disregards citizens of the Middle East and the West alike.

McCormack is a good interviewer. He pushes back and keeps Modad accountable. Most uncomfortable watch on the list, in a productive way. Key timestamps: Energy Shock Risk at 12 minutes, Iran Endgame Scenarios at 20 minutes, Energy Decides It at 54 minutes.

THE SIMULATION LENS

The Iran War Expert: I Simulated the Iran War for 20 Years — Diary of a CEO — Steven Bartlett & Robert Pape (1hr 29min)

Professor Robert Pape has advised every White House on military strategy since 2001 and has run a war simulation on Iran at the end of every strategy class for twenty years. He is the man who has rehearsed this moment more times than anyone alive.

His central warning: the US is falling into a smart bomb escalation trap, where tactical precision is producing strategic failure. He identifies three stages of the conflict and puts a 75 per cent probability on the US escalating to Stage 3. Boots on the ground. He also raises the nuclear question nobody wants to answer: Iran had the material for 16 nuclear bombs, and the US does not know where that material is now.

Pape notes that assassinating Khamenei who was personally opposed to nuclear weapons may have cleared the path for a more aggressive successor who is not.

The most credible, most alarming strategic assessment on this list. Do not watch this one late at night.

THE DARK ARTS LENS

Iran War Predictions: Regime Change, Oil Crisis, Japan’s Future — Craig Hamilton-Parker — Coffee with Craig (39min)

Right. And now for something completely different. Craig Hamilton-Parker is a British psychic, author of 31 books on New Age topics, and known to the popular press as the Prophet of Doom and the new Nostradamus. He predicted Brexit, Trump’s 2016 victory, Covid, and the Queen’s death. He also predicted the post-Brexit economy would thrive with a strong pound. So, as ever, a mixed scorecard.

For those who believe in the ancient arts of astrology, palm reading, and 2,000-year-old Indian palm-leaf manuscripts, Hamilton-Parker draws on the Nadi Astrology texts to interpret current and future events. His prediction for Iran: very targeted strikes, rapid escalation, then a sudden stop, with no boots on the ground, leaving the outcome to the Iranian people. Which, so far, is closer to what has happened than some of the generals managed.

He additionally foresees a new Iran with freedom and rights for women, a catastrophic gas cloud over Japan in August, a major cyber event disrupting crypto and AI markets, and the possibility of a Trump third term. The man is not running short of material. Watch with the same energy you bring to reading your horoscope on a Monday morning. Lightly, with one eye open, and the slight uncomfortable awareness that sometimes these things are oddly on the money.

Finally, It’s Not Only About Oil

Naphtha, Sulphuric Acid and Caustic Soda are three important raw materials impacted by this war. . This can ripple across industries such as automobiles, consumer goods, construction, and clothing, pushing up costs and creating supply shortages. Read Ashna’s essay.

Ashna's Substack

The War at Sea: The Supply Chains You Didn’t Know Were at Risk

On 2 March 2026, a senior official of Iran confirmed what shipping markets had already priced in: the Strait of Hormuz was closed. Any vessel attempting transit would be fired upon. Within 72 hours, tanker traffic had collapsed by an estimated 70 percent. Over 150 ships sat anchored at either end of the strait, waiting…

Read more

a month ago · 51 likes · 20 comments · Ashna Dhuper

That’s it for this week.

A war, four essays, a trilogy, six videos, a brief education in Shia clerical hierarchy, and a psychic who may or may not know something about Japan. Not a bad week’s reading, all things considered.

The world is moving fast. The scarcity thesis: physical things hold value as digital things become cheap keeps proving itself in unexpected places. I will keep pointing them out.

As always: I am not a financial adviser. Nothing here is financial advice. Do your own research. And if you are sitting in a private credit fund with quarterly redemption rights, do re-read the small print.

Satishan

It is the largest asset manager on the planet.

Its entire brand proposition rests on institutional competence, stability, and scale.

So when it quietly sends a letter to investors in one of its flagship private credit funds telling them it will only honour 54 per cent of their withdrawal requests, that is a signal.

Not a routine operational adjustment.

The fund in question is the HPS Corporate Lending Fund.

It is a $26 billion vehicle that BlackRock acquired when it bought private credit specialist HPS Investment Partners last year for $12 billion.

Investors submitted withdrawal requests totalling $1.2 billion in the first quarter of 2026, roughly 9.3 per cent of the fund’s net asset value. BlackRock paid out just over half, around $620 million and invoked a 5 per cent redemption cap to block the rest.

BlackRock’s shares fell 5.1 per cent on the day.

KKR, Blue Owl, and Ares Management all dropped more than 5 per cent.

The market understood immediately what many retail investors are only now beginning to grasp.

The Everything-Is-Fine Narrative Has a Crack In It

Private credit has been the investment story of the last decade.

As banks retreated from corporate lending after the 2008 financial crisis, asset managers stepped in. They offered businesses loans. They offered investors typically 8 to 12 per cent returns that public markets could not match in a low-rate world. Hundreds of billions of dollars flooded in.

The pitch to retail investors and wealthy individuals was compelling.

You get the returns of private markets with the access of a semi-liquid fund. You can get your money out quarterly. It is not quite as flexible as a stock you can sell on a Tuesday afternoon, but it is far more accessible than a traditional private equity lockup.

What was never fully explained or perhaps never fully understood is what happens when everyone wants out at the same time.

Private credit funds hold loans. Corporate loans to mid-sized businesses. These loans do not trade on an exchange. They are not liquid in any meaningful sense. When an investor in a semi-liquid fund asks for their money back, the fund cannot simply sell a position the way a conventional fund manager sells shares. It has to wait for loans to mature, generate new capital through inflows, or hold cash in reserve.

This is the structural mismatch that HPS acknowledged in its letter to investors. The candour is notable. Most firms in this position dress the language up. HPS called it foundational. They were essentially admitting that the redemption cap is load-bearing architecture, not an emergency measure.

Why Now? Follow the Stress Fractures.

The timing is canny.

A cluster of events has eroded investor confidence in private credit over the past year. Two auto parts suppliers failed, raising uncomfortable questions about the due diligence standards being applied to corporate borrowers. A string of write-downs followed at funds managed by KKR, Apollo, Blackstone, and a separate BlackRock vehicle.

Federal Reserve began cutting interest rates.

That sounds counterintuitive, lower rates should help borrowers. But for private credit funds, lower rates mean lower yields on new loans, which squeezes returns and, in some cases, has forced dividend cuts. Investors who signed up for 10 per cent returns started receiving less. Some started asking for their money back.

Some started permanently halting redemptions.

Blue Owl made the situation worse last month by permanently halting redemptions at one of its funds entirely. Not capping them. Halting them. The distinction matters enormously. A cap says: you can have some of your money, just not all of it. A halt says: you cannot have any of it, indefinitely. Blue Owl’s decision sent a chill through the entire sector.

BlackRock, to its credit, has not gone that far. It paid out over half of what investors requested and drew $840 million in new commitments in the same quarter, suggesting the fund is not in freefall. But the direction of travel is clear. And in markets, direction of travel matters as much as the current position.

The Retail Investor Always Finds Out Last

Here is where I find myself returning to a framework I come back to repeatedly: the K-shaped economy. (Thinking of it, I will share a piece on the K-shaped economy. Its being bandied quite a lot these days)

Private credit, for most of its history, was an institutional product.

Pension funds, sovereign wealth funds, endowments. Entities with long time horizons, sophisticated risk teams, and the ability to model illiquidity into their portfolios. They understood the trade-off. Lock up your capital, accept illiquidity, receive a premium return.

The democratisation of private credit.

The semi-liquid fund structure that opened this asset class to retail investors and high-net-worth individuals was sold as a levelling of the playing field. Finally, ordinary investors could access the same returns as the big institutions. What they also got, it turns out, is the same risks. Including the risk of being gated when they want to leave.

This is the K-shape in action. The institutions that built the private credit market understood exactly what they owned and when to reduce exposure. The retail investors who arrived later, attracted by the yield in a low-rate world, are now discovering that the liquidity they thought they had was always conditional. The gate was always there. It just had not been closed yet.

The Real Scarcity Nobody Discussed

There is a second lens worth applying here: scarcity.

I write often about avoiding abundance and seeking scarcity as an investment principle.

Private credit was sold as though liquidity were abundant. As if the mechanics of turning illiquid corporate loans into a quarterly-redeemable product had somehow resolved the underlying constraint. It had not. It had merely deferred it.

True liquidity is genuinely scarce. The ability to sell an asset at a fair price, quickly, when you choose to, is one of the most undervalued properties in investing, precisely because it is invisible when you do not need it and priceless when you do.

What BlackRock’s gate reveals is that the private credit boom packaged an illiquid asset inside a liquid-adjacent wrapper and sold it as though the wrapper changed the underlying nature of the asset. It did not. The loans inside the HPS Corporate Lending Fund are as illiquid today as they were on the day they were made. The only thing that has changed is that more investors now want their money back than the structure can accommodate.

For retail investors sitting in similar vehicles, the question is whether you understand what you actually own. Not whether to panic. Semi-liquid is not liquid. Quarterly redemptions are not a right. They are a feature that can be switched off.

BlackRock’s gate is not the end of private credit. The asset class has genuine merit for investors who understand what they are buying and can genuinely afford to wait. But this moment is worth taking seriously.

This quiet letter, this 54 per cent payout, this 5.1 per cent share price fall.

The first crack rarely announces itself loudly. It just appears one morning in a letter you almost did not read.

Subscribe now

Someone types a question.

Google shows an answer alongside an advert.

Multiply that by a billion people, every day, and you have a money machine of almost incomprehensible scale.

The magic of that model is abundance.

Search costs almost nothing to replicate.

Each additional query does not require a new factory, a new mine, or a new fleet of vehicles.

You just add servers.

Since Sundar Pichai became CEO in 2015, Google’s market capitalisation has 7x’d from $535 billion to $3.6 trillion. Almost entirely on the strength of that abundance flywheel.

Now Alphabet has handed Pichai a pay package worth up to $692 million over three years.

The headline number is extraordinary. But what is truly revealing is where the money is pointed.

Up to $350 million of it is tied not to search, not to cloud, not to AI models.

But to Waymo and Wing.

A self-driving taxi service. A drone delivery start-up.

That is strategy. And it tells you more about where Google thinks the world is heading than any earnings call ever could.

The Price of Fear

Pichai has had a complicated relationship with AI.

When OpenAI released ChatGPT in late 2022, Google was caught flat-footed. The product that should have been Google’s was built by a startup. The company that arguably invented the modern AI era through its research scrambled to respond.

It has since recovered ground.

Gemini is competitive. AI Overviews are now embedded in search.

But the existential anxiety has never fully lifted. Because the deeper threat to Google is not that a rival builds a better chatbot. It is that AI makes the search box itself redundant.

Think about it.

Why do you go to Google? To find something.

What does an AI agent do? It finds things for you.

If Anthropic’s Cowork or OpenAI’s agents become the front end through which people navigate information and complete tasks, Google’s search dominance becomes a legacy asset.

The entire foundation of its $3.6 trillion valuation.

The $692 million package is, in part, the board telling Pichai:

Do not let that happen on your watch.

The performance stock units are benchmarked against the S&P 100. Underperform and you get nothing. It is a gun pointed at the future, with Pichai’s net worth as the incentive to keep firing.

Abundance Built the Empire. Scarcity May Save It.

Here is the investment framework I keep returning to:

Avoid abundance, buy scarcity.

Software is becoming abundant at a pace few people have fully digested. The cost of building a competitive AI product is collapsing.

What took Google teams of hundreds can now be replicated by a handful of engineers with the right tools.

The moat around software businesses including search is narrowing.

Deflationary pressure is already visible in the share prices of SaaS companies, and it will reach further up the stack.

Waymo and Wing operate in a completely different world. You cannot vibe-code a fleet of autonomous vehicles. You cannot train a language model into existence and suddenly have drone delivery infrastructure across a continent. These businesses require physical assets, regulatory approvals, proprietary mapping data built over years of real-world driving, and hardware that takes time to manufacture and deploy.

That is scarcity.

And Alphabet, whether it has articulated it this way or not, is now betting a significant portion of its CEO’s compensation on the scarcity thesis.

Waymo is already operating commercially in multiple US cities. It has logged tens of millions of autonomous miles. That data is a competitive moat of the old-fashioned kind. Accumulated through time and physical presence, not compute.

Wing is similarly positioned. The right to operate drone delivery corridors is not something you can conjure overnight. Regulators, airspace agreements, local partnerships: these are frictional barriers that protect early movers.

Two Googles in One Pay Packet

What makes this pay deal worth talking about is that it captures both versions of Google in a single document.

The performance stock units tied to Alphabet’s total shareholder return represent the old Google. Defend the search empire, keep the cloud growing, do not cede the AI race. That is an abundance business fighting to preserve its position in a world where its core product is under structural threat.

The Waymo and Wing equity represents something else entirely. It is a bet that the next great platform is not built on clicks and queries but on physical infrastructure. Robots, vehicles, and airspace. Things that take years to build, cannot be easily replicated, and become more valuable precisely because they are hard to copy.

As a retail investor, I find that split instructive.

The board is incentivising Pichai to protect what Google has and to build what Google needs. The implicit acknowledgement is that search alone is not enough to anchor a $3.6 trillion valuation through the next decade of AI disruption. However dominant it is today

What This Means If You Own Alphabet

The honest answer is that Alphabet is one of the most genuinely complex investment cases in the market right now.

It is not simply a software company under threat.

It is also, quietly, one of the largest investors in physical-world AI infrastructure on the planet.

Most investors own Alphabet through index funds and think of it as a search and advertising business.

They are not wrong.

That is still where the vast majority of revenue comes from. But the pay deal signals that the company’s own board views Waymo and Wing as the strategic bets that will define the next chapter. The incentive structure is doing the talking that the press releases are not.

The question worth sitting with is this:

If AI agents erode search meaningfully over the next five years, does Waymo’s progress arrive quickly enough and at sufficient scale to fill the gap? That is the bet embedded in this pay package. It is a bet on scarcity replacing abundance as Alphabet’s primary engine of value.

I do not know the answer. Neither does Sundar Pichai, he definitely knows more than me.

But I find it telling that Alphabet’s board has decided $350 million of motivation is the appropriate way to find out.

Sometimes the most revealing document in a company’s filing is not the income statement. It is the pay packet.

Subscribe now

Microsoft’s AI assistant, Copilot, has been a quiet embarrassment. Launched in 2023 with enormous fanfare, it has attracted just 15 million paid users. Roughly 3 per cent of Microsoft’s Office base.

For a company that bet its future on AI, that is a sobering number.

Enter Anthropic’s Cowork.

The general-purpose AI agent launched earlier this year and immediately rattled software stocks. Within two months it had become the poster child for what AI can actually do inside a workplace: clean up your calendar, prepare your briefings, manage your context.

The things Copilot promised but never quite delivered.

So Satya Nadella in true Microsoft spirit did what it usually does when a challenger threatens its turf. He invited the challenger inside.

The deal integrates Cowork into Copilot.

On the surface, both sides win. Microsoft gets a credibility boost for its AI stack. Anthropic gets access to Microsoft’s vast corporate customer base, potentially turbocharging an annualised revenue run-rate that has already doubled from $9 billion to $19 billion in just three months.

But look closer.

This is not a partnership of equals.

It is a positioning game.

The prize is control of the most valuable piece of real estate in the coming decade of AI.

Whoever Owns the Workplace, Wins

Think about how you actually use software at work. You open an app. You type something in. The app does something back. Multiply that interaction by millions of workers and billions of hours, and you begin to understand what platform power really means.

Now imagine that interface disappears. Instead of opening apps, you just ask. An AI agent does the work. It books the meeting, pulls the data, writes the summary, flags the anomaly. The question then is not which app you use. It is which agent you use. And whoever controls that agent controls everything beneath it.

Microsoft built its empire by doing exactly this.

Office did not just sell word processing and spreadsheets.

It became the layer through which an entire generation of workers interacted with their professional lives. Rivals got obliterated. Not because Microsoft’s products were always superior, but because the platform itself became the habit.

That battle is now being refought. The new Office is not a suite of apps. It is an AI agent layer. And Anthropic, quietly and without much fanfare, is building exactly that.

Anthropic Is Playing a Different Game

Anthropic’s public positioning is modest. Its executives insist it has no desire to compete with the specialised workflow software that most businesses depend on. It will stay in its lane, they say.

I do not entirely believe them.

They are not being dishonest. The economics will not allow it.

Anthropic needs enormous, sustained revenue to fund the infrastructure underpinning its models. That pressure pushes every AI lab towards the same place: horizontal software. The tools used by everyone, in every kind of company.

Cowork already has plug-ins.

Third-party developers are already building on top of it. The shape of a platform is beginning to emerge. If adoption accelerates and Microsoft’s distribution deal makes that far more likely, Anthropic will not need to invade software deliberately. Software will simply reorganise itself around what Anthropic has built.

That is precisely what Microsoft is trying to prevent by pulling Cowork inside its own ecosystem. By setting the terms of integration, Microsoft shapes what Cowork can and cannot do. Inside Copilot, Cowork gains cloud access and enterprise-level security.

And most importantly it operates on Microsoft’s ground, under Microsoft’s rules.

The Scarcity Hidden Inside the Software War

Here is the investment lens I keep coming back to:

Avoid abundance, buy scarcity.

The software wars are ultimately a battle being fought in the abundance zone.

Copilot versus Cowork, Salesforce versus agents, old SaaS versus new AI-native tools.

Intelligence, at the software layer, is becoming cheap. Vibe coding means any developer can build a rival product almost instantly. Competitive moats that took decades to construct are being tunnelled through in months.

What cannot be replicated overnight is distribution at scale, enterprise trust, and the institutional inertia of a Microsoft 365 relationship. That is the real scarcity here.

Microsoft is precisely trying to defend that and Anthropic is feverishly trying to acquire that. Match made in heaven. Only time will tell.

As a retail investor watching this unfold, I take two things from it.

First, be deeply sceptical of software companies whose moat is built on headcount and legacy contracts. The disruption is real and it is accelerating. Second, watch who controls the agent layer. Whoever wins that fight inherits the platform economics that Microsoft enjoyed for thirty years.

An Uncomfortable Truce

The FT called this an uncomfortable truce. That feels right. Both parties need each other today. Both know they are rivals tomorrow.

The deal will hold as long as the interests align.

Microsoft needs Cowork to make Copilot credible. Anthropic needs Microsoft’s distribution to fund its infrastructure.

But the moment Cowork achieves genuine mass adoption, the moment it becomes a habit for millions of white-collar workers, the relationship will fracture. At that point, Anthropic will not need Microsoft’s platform. It will have become one.

History offers a clear precedent. Microsoft itself was once a scrappy software company invited inside IBM’s hardware empire. We know how that ended.

Watch Anthropic carefully. Not because it is a good or bad company. But because it is quietly assembling the pieces of something that, if it succeeds, will be one of the most consequential platform shifts of the last thirty years.

The Trojan horse is already inside the gates.

Subscribe now

The stated goal is to reduce Europe’s dependence on American technology.

Its cloud infrastructure, its office software, its AI services.

The ambition is understandable.

The diagnosis, however, is wrong.

We know this because European businesses are saying so directly.

Commerzbank told the FT that the range and technological maturity of US platforms simply cannot be matched by European alternatives right now. A major European carmaker said switching away from existing infrastructure would cost too much and take too long. Thyssenkrupp’s chief executive said bluntly that Europe is not in a position to substitute its IT solutions with European ones.

These are not small companies hiding behind American coattails. These are the industrial backbone of the continent. And they are telling Brussels, as clearly as corporate diplomacy allows, that the tech sovereignty drive risks making European businesses less competitive at precisely the moment they can least afford it.

The question worth asking is not whether Europe should have a tech strategy.

It clearly should.

The question is whether trying to rebuild the cloud is the right one.

The Truth Nobody In Politics Wants to Say

There is a version of this debate that is entirely rational. Trump’s foreign policy has introduced genuine uncertainty into transatlantic technology relationships.

The kill switch fear is driving European leaders paranoid.It has become a recurring topic in boardrooms across the continent.

Kill switch is a scenario in which US cloud and software providers are forced to withdraw services from European customers through sanctions or executive action.

But the response to that risk matters as much as acknowledging it. And the response currently taking shape in Brussels is likely to produce neither sovereignty nor competitiveness. Build European clouds, develop European foundation models, replicate at vast expense what Silicon Valley built over three decades is a pipe dream worth burying beneath the NordStorm pipelines.

A German government insider quoted in the FT put it well, describing it as a trade-off between sovereignty and competitiveness. European companies, they said, need to focus on applying the best available AI models over the next three to four years or risk being overwhelmed by both Chinese and American competition.

That framing is more honest than most of what emerges from Brussels. You cannot out-hyperscale Google, Microsoft, or Amazon. You cannot out-model OpenAI and Anthropic by regulatory fiat.

What you can do and what Europe is actually positioned to do is something entirely different.

Own the Application Layer, Not the Foundation

There is a distinction in technology that rarely makes it into political speeches. The foundation layer is where the Americans and Chinese are engaged in a capital war that will cost trillions of dollars over the next decade. The models, the cloud infrastructure, the compute are a race defined by scale, by the ability to concentrate resources, by the kind of winner-takes-most dynamics that favour the largest players.

The application layer is different. This is where AI gets applied to specific industries, specific workflows, specific problems.

Healthcare diagnostics. Industrial automation. Precision agriculture. Advanced manufacturing. Defence systems. Energy grid management.

These are domains where deep domain knowledge, regulatory relationships, and physical infrastructure matter as much as algorithmic horsepower.

Europe has genuine competitive advantages in almost every one of those domains.

ASML makes the machines that manufacture the chips that run the AI. Siemens and Schneider Electric run the industrial infrastructure of the modern world. Airbus builds the aircraft. Rheinmetall and BAE Systems are rapidly becoming the architects of the AI-enabled battlefield. Novo Nordisk is demonstrating what happens when you apply world-class science to a specific biological problem at global scale.

None of these companies needs to build a cloud. They need to apply the best available AI to the domains they already own. And crucially, they are already doing exactly that. Largely using American platforms, because those platforms are currently the best tools available.

Atoms, Not Algorithms

Here is the investment framework I keep returning to:

Europe’s economy is overwhelmingly built on atoms. It manufactures things. It builds things. It moves things. It has deep expertise in the physical industries that AI is about to transform most dramatically. That is not a weakness to be apologised for. It is a structural advantage to be exploited.

The irony is rich.

For two decades, Europe was mocked for failing to build the next Google or Facebook. Its tech sector was small, fragmented, and over-regulated. Meanwhile America and China built the software platforms that came to dominate the global economy. In that software-first world Europe fell behind.

But the world is shifting.

The next decade is not about who builds the best model. It is about who applies intelligence most effectively to the physical economy.

Energy transition. Autonomous logistics. Smart manufacturing. AI-enabled defence. Precision medicine.

These are European domains. Europe did not miss the AI revolution. It is sitting directly in the path of the part of it that matters most.

What Brussels Should Actually Do

The smart European tech policy is a strategy built around three honest realities, not a sovereignty package that tries to replicate Silicon Valley.

First, use the best available tools. If American AI models are currently superior, European companies should use them and gain the productivity benefits. Handicapping your own businesses in a global competition to score a political point is self-harm, not sovereignty.

Second, build leverage rather than walls. The answer to dependency risk is negotiating power, not isolation. Europe has real leverage over American tech companies: its market, its regulatory authority, its data. Use that leverage to secure contractual guarantees, data residency commitments, and continuity protections. That is far more practical than trying to build a European version of AWS from scratch.

Third, invest aggressively in the physical layer. Europe’s real technology strategy should be pumping capital and policy support into the industries where it already has structural advantages. Defence technology, energy infrastructure, advanced manufacturing, critical minerals processing, and industrial AI applications. These are the sectors where European companies can build genuine global leadership over the next decade.

The businesses speaking to the FT are not arguing against European ambition. They are arguing for European pragmatism. There is a difference between a strategy and a slogan. Tech sovereignty, as currently conceived, risks being the latter.

Europe’s advantage was never algorithms. It was always the world those algorithms need to operate in.

Subscribe now

Read my DISCLAIMER

If you are new to my Substack Read This First

Prologue

This is the third essay in a trilogy I did not entirely plan.

It began with history. Part One, “The Man Who Created the Modern World,” traced the deep roots of European conflict and cooperation, from the steppe empires that shaped the continent’s eastern borders to the political architecture that emerged from the ruins of the twentieth century. The argument was that Europe’s instinct for collective defence was memory rather than idealism. The continent had learned, at enormous cost, what happens when it forgets how to protect itself.

Then the world intervened.

I published Part One on the day American and Israeli forces launched their offensive on Iran. The timing though not deliberate, was clarifying. History, it turned out, was something that was still evolving, on live television, in real time, with oil prices moving and gold absorbing every dollar that fled the uncertainty.

Part Two followed from a question that would not leave me alone. Not why the war was being fought. Not who would win. But something more practical and, I think, more interesting: what does modern warfare actually look like? What happens in the space between the political decision and the physical consequence? Who makes the systems, and how do they work together?

I spent weeks researching that question with the help of Claude, Gemini and Perplexity, going deep into the technology, the doctrine, the supply chains. What emerged was not an analysis. It was a story. “The Thinking Battlefield: War at Machine Speed” is a fictional engagement set in 2036, constructed entirely around real companies, real systems, and real capabilities that exist today. Every constituent of the WisdomTree European Defence ETF appears in the narrative. None of the technology required invention.

The story also introduced a framework that has shaped everything that follows. The world is dividing into two kinds of things. Things that artificial intelligence can replicate at near-zero cost: software, content, code, analytical output. And things it cannot: physical infrastructure, manufacturing capacity, institutional knowledge built over decades, the hard-won ability to produce a precision munition or a naval vessel or a battlefield sensor network. The AI revolution deflates the first category and inflates the second. Defence sits entirely in the second. That is a geopolitical economic observation.

Part Three which is this essay is where that observation meets the numbers.

What follows is a financial analysis of all thirty-two constituents of the WisdomTree European Defence UCITS ETF. It is written from the same perspective as the two essays that preceded it: not an expert, not an analyst, not someone with access to institutional research or proprietary data. A retail investor living in South East London, by the Thames, trying to understand what he owns and why he owns it.

The history came first because context matters. The fiction came second because sometimes a story teaches you what a spreadsheet cannot. The numbers come last because that is, in the end, what an investment decision requires.

Read Parts One and Two if you have not already. They will make Part Three considerably more interesting.

If you have read them, you already know what these companies do.

Now let us look at whether the price is right.

The Machine That Prints Contracts

How €800 Billion Gets Unlocked

On 4 March 2025, European Commission President Ursula von der Leyen stood before the cameras and announced the ReArm Europe Plan. Formally titled Readiness 2030. The number attached to it was almost incomprehensible: up to €800 billion in defence spending, to be mobilised over four years. It arrived six days after Donald Trump suspended all military aid to Ukraine. The timing was coincidental.

The mechanism itself is worth understanding precisely, because this is what turns a political speech into a structural investment cycle. It works in two layers.

The first layer is the SAFE (Security Action for Europe) instrument which the Council of the EU adopted on 27 May 2025. This is a €150 billion loan facility, raised by the EU on capital markets and disbursed to member states on request, based on national procurement plans. It covers air and missile defence, drones and counter-drone systems, ground combat capabilities, ammunition, missiles, cyber, electronic warfare, military mobility, and space assets. The loans carry long maturities and competitive rates. SAFE entered into force on 29 May 2025.

The second layer is the activation of the national escape clause in the EU’s Stability and Growth Pact. The rules that normally force governments to keep deficits below 3% of GDP. With the clause triggered, member states can increase defence spending by up to 1.5% of GDP per year for four years without regulatory penalty. Across the whole of the EU, the Commission estimates this creates €650 billion in additional fiscal space. Add both numbers together and you have the €800 billion headline figure.

The SAFE regulation has legal force. Poland, Germany, Italy, Denmark, Finland, Greece, Estonia, Bulgaria, Belgium, Croatia, and others have already submitted national plans. The money is real, and it is arriving in one direction: towards the European defence industrial base.

The Moat: Buy European

Here is what makes the investment thesis structurally durable rather than merely cyclical. SAFE comes with strings. Specifically, an explicit preference for European suppliers.

For Category 2 systems (which include air and missile defence, drones, AI, electronic warfare, and space), SAFE imposes stricter eligibility conditions: contractors must be able to make decisions on the definition and development of products independently of non-EU third parties. In plain language: American primes need not apply, at least not as lead contractors for the most sensitive capabilities. Procurement must be joint. At minimum two member states collaborating. Ukraine and EEA-EFTA countries also eligible.

This “buy European” mandate flows directly from the lesson that three years of war in Ukraine made impossible to ignore. Strategic dependence on a single ally is a form of vulnerability. When the US suspended aid in February 2025, European capitals felt exposed and abandoned. The policy response is to close that exposure permanently. Not by rejecting the United States, but by ensuring that European armies can be equipped, resupplied, and sustained without needing Washington’s permission.

For investors, this creates something rare: a structural moat that is written into law. European defence primes are not competing on price against Lockheed Martin and Raytheon for these contracts. They are the only ones eligible for the most valuable tranches of spending. The competitive landscape has been reorganised by legislation.

The New Battlefield - The Death of Exquisite

Before we look at the companies, we need to understand what the war in Ukraine has changed about warfare itself. This context is the reason the ten companies in the WisdomTree European Defence ETF are positioned so differently to a decade ago.

Jordi Visser, speaking at a Global Macro Investor event, made a point that cuts to the heart of the matter.

“Think about the fact that Europe is now going to increase spending on defence. Well, that defence is not going to be tanks. It’s going to be drones.”

A modern Leopard 2 main battle tank costs approximately €12-13 million. A first-person view drone capable of destroying it costs around €500. Ukraine has been demonstrating this equation at scale every day since February 2022. The frontier of modern warfare has shifted from the expensive and exquisite to the cheap and numerous.

This shift has two profound implications. First, mass matters again. Countries that can manufacture in volume not just design brilliance win. Second, the entire sensor-to-shooter chain has collapsed in time and cost. Intelligence, target acquisition, strike, and battle damage assessment, tasks that once required years of training and millions in equipment, can now be performed by a twenty-three-year-old with a modified commercial drone and a laptop.

But here is the critical nuance that the simple narrative misses. The dominance of cheap drones has not made expensive platforms obsolete. It has transformed them. The new generation of armoured vehicles, aircraft, and warships being procured under ReArm Europe are themselves drone-integrated systems. Launching swarms, defeating incoming threats, fusing sensor data across platforms in real time. The €800 billion is the purchase of a networked, intelligent, layered military machine.

Every major company in the WisdomTree European Defence ETF has a role in building that machine. Let me introduce them to you.

The Wisdom Tree European Defence Fund

The WisdomTree Europe Defence UCITS ETF (ticker: WDEF on the London Stock Exchange, EUDF on XETRA) launched on 4 March 2025 and gathered €4.7 billion in assets under management within months. A pace of inflows that itself tells a story. It tracks the WisdomTree Europe Defence UCITS Index, which weights companies by free-float market capitalisation multiplied by their defence revenue exposure score. The expense ratio is 0.40%. It holds 32 companies in total, but the top ten account for over 85% of the fund.

The portfolio aggregates $314bn in revenue and $1.02 trillion in market capitalisation across 32 companies. The 9.5% operating margin understates quality: strip out Airbus (80% civil aviation) and Safran (65% civil) and the defence-pure margin approaches 12-13%. Cash from operations of $42.6bn substantially exceeds reported operating profit of $29.7bn. A 1.43x conversion ratio that signals genuine cash generation power across the portfolio. The $9.5bn FCF figure is compressed by front-loaded capital expenditure at Safran, Airbus, Kongsberg and Saab, all of which are investing heavily in new programmes. This is a sign of a sector in its early build phase.

The Five ReArm Europe Pillars

The €800 billion ($925 billion) headline number represents the total potential defence spending unlocked across Europe through these mechanisms.

Strategically, the plan signals something far more important:

Europe is preparing for a world in which the United States may no longer guarantee its security.

The suspension of US military aid to Ukraine by Donald Trump accelerated a shift that had already begun.

For investors, it effectively marks the birth of a European military-industrial cycle that could run for a decade or longer.

The ETF exhibits Five Structural Themes

The Backlog is the Business

Backlog-to-revenue ratios are the single most important metric in defence. Rheinmetall at 5.6x, BAE at 3.0x, and Kongsberg at a recent 3.22x quarterly book-to-bill mean revenues are largely pre-sold for years ahead. In any other sector, backlog of this magnitude would command extreme premium valuations. European defence has not yet fully re-rated.

Defence Purity Commands Premium

Companies with 80%+ government defence revenue. BAE (98%), Saab (91%), Rheinmetall (80%), Chemring (100%) exhibit fundamentally different risk profiles to aerospace conglomerates. They are not exposed to airline demand cycles, consumer discretionary, or commercial aviation financing risk. The ETF deliberately overweights purity relative to revenue scale.

Margin Convergence Is Incomplete

European defence companies trade at 8-15% operating margins. US peers have historically run ahead. Lockheed Martin (13%), Northrop Grumman (15%), RTX (15%). As European governments lock in multi-year contracts at higher prices, and as volume scales against fixed cost bases, European margins will converge upward. Rheinmetall at 14.6% and already tracking US levels is the leading indicator.

Physical Complexity Is the Moat

You cannot vibe-code a submarine. You cannot prompt-engineer a frigate fire control system. The AI deflationary wave that is compressing software margins and erasing SaaS business models has no equivalent in defence manufacturing. A Kongsberg Naval Strike Missile requires physical production infrastructure, supply chains, test facilities and regulatory certification that took decades to build. This is Jordi Visser’s scarcity principle applied to industrial capability rather than raw materials.

ETF Construction Creates Alpha Opportunities

WisdomTree weights by defence revenue, not market cap. This creates systematic mispricings. Take the example of Kongsberg. The highest-quality small-mid cap in the portfolio with 3.22x book-to-bill, net cash, and a 17% EBITDA target sits at only 5.02% weight. Exosens at 21.8% operating margin and GPS-independent targeting technology sits at 0.73%. Investors who understand the individual constituents can complement ETF exposure with targeted direct holdings.

ETF Constituents

Airbus Distorts the Revenue Picture

Airbus contributes 27.1% of all portfolio revenue but holds only 6.6% ETF weight. This is intentional. WisdomTree weights by defence revenue, not total revenue. Strip Airbus from the revenue aggregate and the remaining 31 companies generate $228bn with a markedly higher blended operating margin. Airbus is a diversification anchor, not a defence pure play.

Thales: The Hidden FCF Champion

Thales reports $2.05bn operating profit but $4.03bn free cash flow. A 1.96x FCF conversion ratio that is extraordinary. This reflects significant working capital inflows, advance payments on long-term contracts, and non-cash charges depressing reported earnings. At 6.0% FCF yield on market cap, Thales is the most attractively priced large-cap in the portfolio on a cash basis.

Dassault: Fortress Balance Sheet, Ignored by the Market

Dassault carries $1.87bn in cash against only $276m in debt. A 0.15x debt-to-cash ratio. The business generates CFO of $2.18bn on operating profit of only $617m, a 3.54x conversion driven by massive Rafale export advance payments. At EV of $27bn against a net-cash-adjusted equity value, the stock appears structurally undervalued. Family-controlled governance is the only constraint on re-rating.

Melrose: The One Red Flag

Melrose Industries is the portfolio’s only company with negative operating profit on the current period’s figures. The transition from GKN Aerospace acquisition to a focused aerospace structures business is still in progress. Debt of $2.19bn against cash of $118m (18.6x ratio) is the highest leverage in the portfolio by a wide margin. This is a recovery trade, not a defence thesis.

Kongsberg: The Most Underweighted Quality Name

On every quality metric Kongsberg sits at or near the top of the portfolio. Book-to-bill (3.22x), net cash ($1.14bn), operating margin (14.9%), CFO conversion (2.72x), EBITDA margin target (17%). Yet its 5.02% ETF weight underrepresents this quality. Investors seeking to tilt toward the highest-conviction names should consider overweighting Kongsberg directly.

The FCF Destroyers Are Investing, Not Struggling

Safran ($8.6bn implied capex), Airbus ($10.7bn), Saab ($1.32bn), and Kongsberg ($1.63bn) all report negative FCF. In each case, cash from operations is strongly positive. The gap is capital expenditure: new engine programmes, Gripen E production ramp, facility expansion, and naval weapons system development. This is not financial stress. It is the front end of the demand curve being built out in physical infrastructure.

The ETF Covers the Five Conflict Domains of 2038

Military planners now speak of five domains of conflict: land, sea, air, space, and the electromagnetic spectrum. By 2038, all five will be contested simultaneously in any serious conflict, and the electromagnetic spectrum will be the most contested of all. Whoever controls it, controls every other domain.

Think of the electromagnetic spectrum as the nervous system of a modern military. Drones navigate by it. Missiles home in by it. Commanders communicate through it. Radar sees through it. The moment you blind or jam or corrupt it, you have paralysed the entire organism. This is why electronic warfare has moved from a niche capability to the central prize. The side that can suppress, deceive, and destroy enemy electronics while protecting its own will set the terms for everything else that follows.

Above the electromagnetic battlefield sits space. By 2038, low-earth orbit will be saturated with military satellites: surveillance constellations watching every square metre of earth’s surface in near-real time, communications relays bouncing encrypted signals between units at the speed of light, early-warning sensors detecting missile launches within seconds of ignition. Space will not be a sanctuary. Both sides will field anti-satellite weapons. Kinetic missiles, directed energy lasers, co-orbital interceptors. Blinding the enemy’s space layer is the fastest way to make their entire military force deaf and blind simultaneously. A modern army without satellites cannot navigate, cannot target, cannot communicate beyond line of sight. It is, essentially, a very expensive mob.

On land, the human soldier will not disappear. The ratio of humans to autonomous systems will invert. The front line of 2038 will be held predominantly by robots: ground vehicles with sensors and weapons, loitering munitions orbiting above positions for hours waiting for a target, static sensor networks feeding AI systems that track every movement within kilometres. Humans will be further back, making decisions at the mission level while the machines execute at the tactical level. The bottleneck will be decision speed, rather than firepower. Whoever’s AI can process sensor data, identify threats, and execute responses faster than the adversary’s AI will have a decisive advantage. Because they completed the kill chain in seconds rather than minutes.

At sea, the surface ship is increasingly a liability. Surface ships are large, slow, visible from space, and exquisitely vulnerable to hypersonic anti-ship missiles travelling at Mach 8. Fast enough to cross the North Sea in under four minutes. The future of naval power lies underwater and above the water simultaneously: submarine fleets hunting and denying, unmanned surface vessels acting as expendable sensor pickets and decoys, and overhead, anti-ship missiles and naval drones working in coordinated swarms. The navy of 2038 will look less like a fleet of grey warships and more like a distributed network of threats that can appear from any direction, in any medium.

And in the air, the manned fighter jet will still exist. But it will not fly alone. It will fly at the centre of a constellation: loyal wingman drones on its flanks carrying sensors and weapons it can direct without risking its own airframe, AI systems processing radar returns and electronic emissions to build a real-time picture of the battlespace, electronic warfare pods actively jamming and deceiving enemy radar. The pilot’s role shifts from stick-and-rudder warrior to mission commander: a human brain at the centre of a machine network, setting objectives and constraints, letting the AI and the drones execute. The jet’s value is in its ability to command the swarm along with its speed or its cannon.

The ETF’s Architecture reflects these five domains

The Ground Battle: Rheinmetall and Leonardo

The Air Battle: Saab, Thales, Dassault, Safran, Airbus

The Full-Spectrum Prime: BAE Systems

The Maritime and Electronic Battle: Rolls-Royce, Kongsberg, Hensoldt

The Speed Problem and Why It Changes Everything

There is one constraint that runs through all five domains and shapes everything else: time. A hypersonic missile in 2038 travels at Mach 8. That is roughly 2.7 kilometres per second. A missile fired from the Kaliningrad (Russia’s military outpost on the Baltic) reaches Warsaw in under three minutes. Human reaction times operate on a scale of seconds for a single decision, minutes for a command chain, hours for a national government. An adversary who launches a coordinated hypersonic salvo at eighteen targets simultaneously has already hit them before the relevant committee has convened.

This is the fundamental dilemma of future warfare: the speed at which threats move has outrun the speed at which human beings can respond. Which means that, increasingly, the machines must be allowed to respond autonomously. Not because anyone is comfortable with that. The alternative which is waiting for a human to authorise each intercept means accepting that most missiles get through.

This is not a technology problem. The technology is largely solved. It is a doctrine problem, a legal problem, and ultimately a political problem. The nations willing to delegate lethality to algorithms will have a significant battlefield advantage over those who insist on keeping a human hand on every trigger. This implies accepting that a machine will sometimes be wrong, in order to ensure that the right answer comes fast enough. It is an uncomfortable truth, and it will define the strategic debates of the 2030s in the way nuclear deterrence defined those of the 1950s.

The Arsenal Being Built By This ETF

Every single technology described above requires hardware. The drone swarms, the electronic warfare systems, the sixth-generation fighters, the submarine networks, the space-based sensor layers are physical hardware. Hardware that must be designed, manufactured, tested, and delivered in quantities that no single company, and no single nation, can provide alone. The companies inside the WisdomTree European Defence ETF are building this arsenal of 2038, right now.

Hensoldt is building the electronic warfare and radar systems that will determine who controls the electromagnetic spectrum. Its naval radar tracks hypersonic threats. Its electronic warfare systems deploys counter-drone capability at scale, gives the Eurofighter Typhoon the ability to see, jam, and deceive simultaneously. In the five-domain war described above, Hensoldt’s technology is what decides whether your side is blind or sighted.

Thales is building the integration layer. The AI-driven sensor fusion and command systems that allow machines from different nations, operating in different domains, to share a single coherent picture of the battlefield. Its radar is already the brain of the Rafale fighter. Its counter-drone systems are firing on active targets in Ukraine today. In the war of 2038, Thales is the nervous system.

Saab’s Gripen is designed from the ground up for distributed operations. It can operate from motorways, be maintained by a small ground crew, and survive the loss of centralised air bases that an adversary will target first. Its counter-drone system is already deployed with NATO forces in Poland. Saab has built the aircraft and the countermeasures for both sides of the drone equation.

Leonardo’s sensors flying on the Gripen, the Osprey radars on maritime patrol aircraft are the eyes of the future battlefield. Its joint venture with Rheinmetall is designing the armoured vehicles of the 2030s. Leonardo is the optics of a network that is otherwise blind.

Rheinmetall is building the kinetic layer. The ammunition, the armoured vehicles, the drone-integrated battle tanks that turns sensor intelligence into physical effect. Its Panther tank launches reconnaissance drones directly from its turret. It is simultaneously the largest artillery shell manufacturer in Europe and an emerging drone developer. When the AI identifies the target and assigns the response, Rheinmetall’s ammunition is what arrives.

Kongsberg’s Naval Strike Missile skims the surface below radar horizon, making it nearly invisible to ship-based defences. In a maritime environment where surface ships are increasingly vulnerable to hypersonic threats, Kongsberg’s missiles are how you reach a target without putting a crew in harm’s way.

Rolls-Royce’s turbines power the warships that will carry NATO’s naval presence through the 2030s and 2040s. No engine, no ship. No ship, no presence.

BAE Systems, spanning the entire stack from sixth-generation fighter design through to nuclear submarine construction and space-based missile tracking satellites, is building the infrastructure for all five domains simultaneously. The GCAP programme will produce the fighter jet that commands the swarm. The Dreadnought submarines will carry Britain’s nuclear deterrent through the 2060s. The Ball Aerospace satellites will provide the space layer that makes everything else possible.

And Airbus, through its A400M transport fleet and its role in FCAS (Future Combat Air System) — the Franco-German sixth-generation fighter programme — is building both the logistics backbone and the next generation of air combat architecture, wiring the datalinks that allow all of these systems to talk to each other across national borders.

The war of 2038 will by the side that achieves decision superiority. Less by the largest army or the richest treasury. The ability to sense, process, decide, and act faster than the adversary, across all five domains, simultaneously. Every company in the WisdomTree European Defence ETF is building a piece of that capability. None of them is building it fast enough, because European defence industry has spent thirty years in hibernation. But they are building it now, with a speed and a funding certainty that did not exist three years ago. The question for the investor is simple: who holds the industrial contracts to build the arsenal of 2038? The answer is sitting in the ETF.

Why This Is Not a Trade

The Structure Beneath the Story

Commodity cycles end. Defence spending cycles, particularly those driven by structural geopolitical shifts, do not behave like commodity cycles. Consider the data. European NATO members underspent the 2% GDP target by a cumulative $828 billion between 2014 and 2024. Russia increased its defence budget by over 300% in the same decade. China increased its budget by over 600%. The gap between what Europe spends and what the threat environment requires is a structural deficit that was building for thirty years. Closing it will take at least as long.

Germany’s Zeitenwende. The “watershed moment” Chancellor Scholz declared in February 2022 is a €400 billion rearmament programme, now combined with a €100 billion Sondervermögen (special fund) that bypasses normal constitutional debt limits. Poland is spending 4.12% of GDP on defence. Far more than any other NATO member. France has locked in a Military Programming Law running to 2030. The UK is on a trajectory towards 2.5% of GDP. These are multi-year legislative commitments, not annual budget lines.

Jordi Visser’s framework applies here with precision. The things that take a long time to build. Jets, radars, missiles, submarines, trained personnel are in insatiable demand. You cannot manufacture a Saab Gripen in six months. You cannot train a Rafale pilot in a year. You cannot scale naval propulsion capacity in a single budget cycle. This is the opposite of software, where a competitor can appear overnight. The lead times in this industry are measured in years and decades. That is a structural tailwind, not a cyclical one.

The buy-European mandate reinforces this. Every contract funded under SAFE — and remember, SAFE is just the first instrument of a multi-pillar €800 billion programme — flows first to European primes. Rheinmetall, Leonardo, Thales, Saab, BAE Systems, Kongsberg, Hensoldt: these companies are not just beneficiaries of increased spending. They are the designated recipients of a legally mandated industrial strategy.

The Downside Risks

I said this essay would be analytical as well as personal. That requires me to be honest about the downside.

The first risk is geopolitical resolution. If a durable ceasefire in Ukraine is reached and credibly maintained, the urgency driving procurement could diminish. History suggests that peace dividends are real. Spending fell sharply after 1991 and after 2001. Against this, I would note that the structural argument does not depend on ongoing conflict. The threats of Russian revanchism, Chinese expansionism, the fragility of the American security guarantee do not disappear with a ceasefire. European governments have already committed in law to spending trajectories that run to 2030 and beyond.

The second risk is industrial capacity. These companies have order books that are stretching their factories. Delivery timelines are slipping. Skilled labour is constrained. The risk of disappointment between orders and revenues is real in the near term. This is not a reason to avoid the sector. Capacity bottlenecks in a demand-driven cycle ultimately mean pricing power and margin expansion. It is a reason to think in years, not months.

The third risk is currency. If you are investing from outside the eurozone, movements in EUR/GBP, EUR/USD, or EUR/SEK will affect your real returns. WDEF on the London Stock Exchange is EUR-denominated. The EUR-unhedged class (EUDF on XETRA) gives you clean eurozone exposure.

The fourth risk is valuation. These stocks have re-rated sharply. Rheinmetall traded at roughly €70 in early 2022 and has since moved dramatically higher. You are not buying at the bottom. But in a structural multi-year cycle driven by legally committed government spending, re-rating can persist for longer than seems rational. And earnings are growing into valuations that once looked stretched.

Where I Stand

I want to end plainly, without the hedges that financial caution often demands of honest writing.

The ReArm Europe plan is real. The SAFE mechanism is law. The buy-European mandate is structurally embedded. The companies in the WisdomTree European Defence ETF are building the weapons, electronics, aircraft, ships, and drones that Europe’s governments have already committed, in binding legislation and multi-year budgets, to purchasing. The war in Ukraine is the accelerant that revealed a structural deficit that had been accumulating for thirty years.

I believe the single cleanest expression of this thesis for a private investor is the WisdomTree Europe Defence UCITS ETF on the London Stock Exchange, or EUDF on XETRA. In one trade you own Rheinmetall’s tanks and drones, Thales’s radars and anti-drone rockets, Saab’s fighters and submarines, Leonardo’s avionics and armoured vehicles, BAE Systems’s submarines and aircraft, Rolls-Royce’s naval propulsion, Kongsberg’s naval strike missiles, Hensoldt’s sensors and electronic warfare, Safran’s engines, and Airbus’s military transport. You own the arsenal of the European rearming.

This is not a trade. Genghis Khan’s horsemen rode for thirty years before the Pax Mongolica remade the world. The spending commitments being made in Brussels, Berlin, London and Warsaw today are denominated in decades, not quarters. The companies in WDEF are building the wall, not just riding a wave.

References and Sources

• ReArm Europe / Readiness 2030 — European Commission, March 2025. Available at: commission.europa.eu/topics/defence/future-european-defence_en

• SAFE Regulation (EU) 2025/1106 — Council of the EU, 27 May 2025. Available at: consilium.europa.eu

• WisdomTree Europe Defence UCITS Index Methodology — WisdomTree, 2025. Available at: wisdomtree.eu

• WisdomTree: “Welcome to the Era of European Strategic Rearmament” — Blog post, July 2025. Available at: wisdomtree.com/investments/blog

• ETF holdings data (WDEF.L / EUDF.DE) — Yahoo Finance, February 2026. Available at: finance.yahoo.com

• Rheinmetall KF51 Panther specifications and procurement updates — Rheinmetall AG. Available at: rheinmetall.com

• Leonardo Q1 2025 Results Presentation — Leonardo S.p.A., May 2025.

• BAE Systems Annual Report 2024 — BAE Systems plc, 2025.

• Saab Year-End Report 2025: Record Order Bookings — Saab AB, February 2026. Available at: saab.com

• “The Loke Counter-Drone Concept Debuts in NATO Mission” — Saab newsroom, September 2025. Available at: saab.com

• “Thales Anti-Drone Rockets Now Being Used in Ukraine” — The War Zone, October 2025. Available at: twz.com

• “Kongsberg Secures €100 Million Contract to Equip Denmark with Advanced NSM” — Army Recognition, June 2025.

• “Romania Evaluates German Rheinmetall KF51 Tank Proposal” — Army Recognition, 2025.

• Euronews: “How Can the EU Unlock Up to €800bn for Its Rearmament Plan?” — 5 March 2025. Available at: euronews.com

• European AESA Radar Market 2026: Systems, Operators and Contracts — Grosswald.org, February 2026.

• Pal, Raoul and Visser, Jordi. “Macro, Crypto & AI Converge” — Global Macro Investor transcript, 2025.

• European Parliament Research Service: “ReArm Europe Plan/Readiness 2030” — EPRS Briefing, March 2025.

• NATO Defence Expenditures and 5% Commitment — NATO, June 2025. Available at: nato.int

• CSIS: European NATO underspend data 2014-2024 — Centre for Strategic and International Studies.

• Fortune: “Defence Giants in the ReArm Europe Race” — 9 March 2025. Available at: fortune.com

IMPORTANT NOTE: This essay is written for informational and analytical purposes only. It does not constitute financial advice. The author may hold positions in securities mentioned. Past performance of any investment does not guarantee future results. Always conduct your own research and consult a qualified financial adviser before making investment decisions.

Subscribe now

I published Part One of this trilogy on a day that made the subject feel suddenly, uncomfortably real.

American and Israeli forces had launched their offensive on Iran. The news was moving fast and the analysis was moving faster, filling every channel with a version of events shaped by whoever was doing the filling. Mainstream media focused on what was verifiable: strikes, responses, positions, casualties. Social media offered something wilder. The Epstein distraction theory, the third temple prophecy, regime change as the hidden agenda, oil as the real motive. Every corner of the internet had a different war going on.

I am not a geopolitical analyst. I am not a defence expert. I approached this the only way I know how. As an investor trying to understand why markets crash, why oil shoots north, and why gold gets all the money when the world decides it is frightened.

But somewhere in this past week, the investment question started giving way to a different one. Not why is this war being fought. Not even who benefits. But something more fundamental: how is war itself changing? What does it actually look like now, in 2025, when the weapons think for themselves and the battlefield is electromagnetic before it is physical?

That question sent me down a rabbit hole. I spent weeks with Claude, Gemini and Perplexity, pushing into the technology, the doctrine, the companies, the systems.

What I found kept pointing back to a single organising idea: that the world is splitting into two kinds of things. Things that can be replicated at near-zero cost. Code, software, AI models, digital content. And things that cannot. Minerals, manufacturing capacity, physical infrastructure, institutional knowledge accumulated over decades. The AI revolution is simultaneously deflating the first category and making the second category more valuable, more contested, and more scarce.

Defence is entirely in the second category. You cannot vibe-code a missile guidance system, maybe sometime in the near future but not now. You cannot prompt-engineer a submarine. You cannot spin up a competitor to Rheinmetall’s artillery production line with a few GPUs and a good idea. These capabilities took thirty years to build and they will take thirty years to replicate. That is a moat. And it is the reason the companies in this index deserve serious examination rather than a passing mention in a geopolitics column.

That framework became the lens through which I researched this essay. I gave Claude the parameters. A fictional engagement, a real index, thirty-two constituent companies, the technology they actually make and asked it to assemble a scenario in which all of it was present and working. The result surprised me. Not because the technology was invented, but because none of it was. Every system in the story exists. Every company is real. The fiction was only in the date.

I want to be clear about one thing before you read further. I am, philosophically, against war. I believe in a world where people coexist, where borders are honoured and disagreements are resolved without the young dying for the decisions of the old. That is not a political position. It is a human one.

But we do not get to choose the hand we are dealt. We only get to decide how we play it.

This essay is not a case for conflict. It is an attempt to understand the world as it is. Not as we would prefer it to be. If you are going to hold the WisdomTree European Defence ETF, or any exposure to this sector, you should understand what these companies actually do. Not in the language of quarterly earnings calls, but in the language of the thing itself.

Part Two is that language. Part Three is the investment case that follows from it.

Read them in order, and make up your own mind.

Dawn. 04:47. Brussels. 14th March 2036.

The city is asleep.

In the Atomium district, a baker is pulling his first loaves from the oven. On the E40 motorway, a single autonomous freight carrier is making good time toward Liège. In the Parc du Cinquantenaire, a fox is crossing the empty path between the arches, unhurried, indifferent to history.

Everything is normal.

Except that three floors beneath the NATO Strategic Command Centre on the Avenue du Bourget, a bank of screens has just turned amber.

The Room

They call it the Atrium. Not because it has a skylight. It has no windows at all. But because everything flows through it. Every signal, every satellite feed, every encrypted channel from every forward position between Tallinn and the Aegean. It is, in the military parlance of the time, the nervous system of a continent.

The operator stations are ringed in tiers around the central display. Each one is equipped with a Mildef Group hardened tablet terminal.A ruggedised Swedish military computing built to function at forty below, or in the electromagnetic chaos of an active electronic warfare environment.

The screens are matte black, unreadable at any angle except direct. On each one, a small blue logo: a Swedish crown above a circuit board. Nobody in the room has ever thought about who made the terminals. That is precisely the point.

Air Marshal Diana Hartley of the Royal Air Force is already at her station when the alert comes in. She has been here since 03:00, which is when her instinct woke her and her discipline brought her in. She has learned, over thirty-one years, to trust both.

To her left, General Klaus Brandt of the Bundeswehr is studying a topographic overlay of the Suwalki Gap.

The narrow land corridor between Poland and Lithuania that military planners have described, with characteristic understatement, as the most dangerous eighty kilometres in Europe.

Across the circular table, Admiral Sophie Moreau of the French Navy is on a secure channel with the Charles de Gaulle carrier group. She has the particular stillness of someone who has already run the scenarios and found them finite.

And at the far end, General Tomasz Wisniewski of the Polish Armed Forces is standing. He never sits in this room. He says it is because he thinks better on his feet. Everyone in the room knows it is because he has been waiting for this moment for his entire career, and sitting down would feel like submission to it.

On the central screen the map of eastern Europe is lit in gradations of amber. It is built and maintained by Thales, the French defence and electronics giant that forms 12% in the WisdomTree index. The system is called SYNAPSE. It is, at this moment, watching seventy-three thousand square kilometres of contested territory simultaneously.

Then, at 04:51, the amber turns orange.

Then, at 04:53, it turns red.

“Talk to me,” says Hartley.

The First Move

The voice that answers is not human.

MINERVA (Multi-domain Integrated Network for European Response and Vigilance Architecture) has been processing the sensor data for nineteen minutes. It was built by a consortium, as most important things in Europe are. The threat-assessment algorithms were developed by QinetiQ, the UK government-owned defence research organisation that has spent forty years teaching machines to understand what danger looks like before human eyes can see it.

The data fusion layer was built by Cohort PLC, the British defence technology specialist whose electronic intelligence products had been stitching together fragmented battlefield signals since the late 2020s. Where QinetiQ taught MINERVA to think, Cohort taught it to listen. To pull coherent meaning from the electromagnetic noise of a continent in tension.

The ground vibration sensors buried along the Polish border were supplied by Chemring Group, the British specialist in energetic materials and sensor systems whose engineers had spent three years perfecting how to make something invisible to metal detectors, impervious to frost, and capable of distinguishing a T-14 Armata tank from an agricultural vehicle at four hundred metres.

Thermal imaging is streaming from two MALE surveillance drones. Powered by Rolls-Royce turbines,

carrying sensor packages designed by Hensoldt, the German optronics and radar specialist it has been tracking movement along three separate approach routes since 02:30. The signatures match no known civilian pattern.

MINERVA assembles all of it in under a second.

“Confirmed mechanised column movement. Bearing two-seven-zero. Estimated strength: two armoured brigades, one electronic warfare battalion. Estimated time to the Polish border: sixty-four minutes.”

“I am also detecting pre-launch signatures consistent with Iskander-M ballistic missile systems at three sites. Confidence: 91 percent.”

General Wisniewski finally sits down.

“So,” says Brandt quietly. “Here we are.”

Act One: The Invisible Layer

What happens in the next eleven minutes is, to anyone watching from the outside, entirely invisible.

There are no explosions. No aircraft scrambling from runways. No columns of soldiers moving to contact. To a satellite observer, to a civilian news drone, to the baker in the Atomium district, everything is still normal. The fox has found something under a bench in the park and is investigating it with the methodical attention of a creature with no geopolitical concerns.

But the battlefield has already begun.

The missile coordination network activates automatically the moment MINERVA’s confidence threshold crosses 85 percent. Kongsberg Gruppen‘s Naval Strike Missiles loaded on three frigates in the Baltic and two land-based launcher vehicles in northern Poland receive target packages.

The missiles do not move yet. They simply know where they are going. Inside each launch canister, Kitron, the Norwegian electronics contract manufacturer, had assembled the circuit boards that run the weapon’s navigation and fuzing systems. Kitron’s name appears on no press release, no defence contract announcement, no stock tip. It appears on a quality certification label, inside a sealed compartment, that no one will ever read. It is exactly where it needs to be.

Simultaneously, the drone layer deploys.

They rise from sites so distributed and unremarkable. Farm buildings in eastern Poland, a disused industrial estate outside Vilnius, three converted shipping containers on a forest track in Lithuania. No pre-planned strike could have targeted them all. The platforms are Rheinmetall autonomous combat drones, each one carrying an electro-optical sensor package and a precision munition. They are not glamorous. They are not supersonic. They are simply very numerous, very quiet, and very patient.

They do not attack. Not yet. They disperse into a holding pattern so wide and so low that radar cannot distinguish them from terrain clutter. They are waiting.

“We are at DEFCON Europa Two,” says Hartley. “I am recommending we move to Three.”

“Seconded,” says Moreau. “Agreed,” says Brandt.

Wisniewski says nothing. He is watching the red markers move.

Act Two: The Human Question

The protocol requires a human decision before any kinetic action. This was the argument that consumed three years of NATO committee work and two near-collapses of the alliance’s command framework. The machines could see further, react faster, and calculate better than any human. Everyone agreed on this. The disagreement was about what it meant.

The question was simple and almost impossibly hard: at what point does a machine have permission to defend Europe?

The answer arrived at was characteristically European. A compromise that satisfied nobody completely and everybody sufficiently. MINERVA could act unilaterally on electronic warfare, counter-battery radar, and missile interception below a defined yield threshold. Everything else required four human signatures or a two-of-four majority in the event of communication failure.

In practice, at this speed, four people have approximately ninety seconds to make a decision that previous generations had the luxury of debating for months.

“Iskander pre-launch signatures are now at three additional sites. I am detecting GPS spoofing activation across the Polish corridor. Source: ground-based emitter network, estimated location — “ MINERVA gives a grid reference. Brandt marks it on the overlay without looking up.

“Do it,” says Hartley.

Thales’ SYNAPSE electronic warfare network distributed across forty-seven mobile platforms, three airborne assets, and the MINERVA core begins suppression. The GPS spoofing signals are identified, characterised, and jammed in eleven seconds. In the background, Cohort‘s intelligence layer is already cross-referencing the emission signatures against its library of known Russian electronic warfare systems, flagging two emitter types that have not been seen outside Russian territory before tonight.

Supporting the broader electronic warfare picture, Indra Sistemas. Spain’s defence electronics specialist. It is running real-time spectrum analysis across forty-three frequency bands, identifying the spoofing signatures and building the electronic order of battle. They had learned, in the war games of the early 2030s, that the first battle of any modern conflict is the spectrum battle. If you lose that, you lose everything.

“Their artillery solutions are degraded. Estimated targeting error: 340 metres. Insufficient for precision strike. Sufficient for area denial.”

“They will recalibrate.”

“Yes. Estimated time to recalibration: twenty-two minutes.”

“Then we have twenty-two minutes,” says Wisniewski.

He is standing again.

Act Three: The Engagement

At 05:34, the first Iskander launches.

MINERVA tracks it from ignition. The missile is fast. Mach 6. Depressed trajectory, designed to defeat theatre missile defence by arriving faster than the reaction window. It was designed in the 1980s for a world where no one had built a counter-battery system fast enough.

The outer intercept layer belongs to Saab. The Swedish company’s BAMSE next-generation surface-to-air system.

Upgraded in the 2031 batch contract to carry a new seeker with three times the original engagement envelope. It fires a single missile at an intercept solution calculated to within four metres. The warhead proximity fuse detonates at optimal stand-off distance. There is a flash above the Mazurian lakes that briefly outshines the still-dark horizon.

To the people on the ground below, it looks like a sudden star. Brilliant white. Gone in a second.

Four more Iskanders follow. The inner intercept layer ASTER 30 Block II missiles integrated onto naval platforms built by Fincantieri, the Italian shipbuilder whose frigates form the backbone of the European naval strike group, accounts for three more.

Leonardo‘s AULOS passive radar network, operating in complete emissions silence, is tracking the fifth missile on acoustic and thermal signature alone, feeding the intercept solution to BAMSE without a single active emission that could be backtracked to a launcher.

One missile is not intercepted.

It lands in an open field twelve kilometres east of Bialystok. The warhead is conventional. The field is empty. The crater is four metres deep and smells of scorched earth and burnt propellant. A herd of cattle in the adjacent field do not even look up.

“They are testing the intercept rate,” says Moreau.

“Yes.” “And?” “Their assessment will be that our intercept capability is between 78 and 84 percent.”

The room is quiet for a moment.

“Good,” says Hartley.

“Our actual intercept capability is 94 percent.”

The Ground

While the missile exchange plays out above, the armoured column is still moving.

It crosses the border marker at 05:41, forty-seven minutes ahead of MINERVA’s initial estimate. Someone has pushed the accelerator. Someone on the other side made a decision, and the column is now committed.

The forward observation teams that have been watching it from concealed positions since 04:15 are wearing Avon Rubber respirators and sealed protective suits under their camouflage.

It is standard issue for any forward deployment in a theatre where chemical precursors have been detected in the logistical supply chain. The masks are British-made, military-certified, and have the particular intimacy of equipment worn against skin in the dark. The soldiers behind them see the column moving and say nothing into their radios. They already sent the data. MINERVA has it. The humans are now spectators to the decision they made earlier.

What the column does not know is that the Rheinmetall drone swarm has been repositioning for the last thirty minutes. INVISIO Communications‘ tactical headsets worn by every ground commander in the forward positions have been receiving a continuous encrypted datastream telling them exactly where the swarm is and what it is doing, without producing a single electromagnetic emission that could be detected and localised.

Babcock International’s forward support teams are pre-positioned at three logistics nodes along the Polish corridor. Not fighting. Not in the engagement zone. Doing the work that no one films and no one celebrates: managing fuel, ammunition resupply routes, vehicle recovery, communications relay infrastructure. The unglamorous arithmetic that determines whether a battle can be sustained beyond its first twenty minutes.

RENK Group’s advanced transmission systems keep the heavier Rheinmetall ground platforms manoeuvring at speeds the column’s tactical planners did not model. They had assumed European armoured vehicles would be slower. They had used old data.

Chemring’s battlefield sensor network — the same buried ground sensors that heard the column forming up in the dark — is now providing real-time positional updates with fifteen-metre accuracy. The data is routed through Serco‘s battlefield management system, which is coordinating six forward operating bases, running simultaneous logistics threads across four national force contributions without a single human dispatcher.

It is, in its quiet way, one of the most complex real-time resource allocation operations ever run in a combat environment. Nobody will write about it.

The drone swarm activates at 05:44. Sixty platforms from the forward holding pattern drop from altitude in a coordinated attack sequence that MINERVA has been calculating since 05:12. The lead elements of the column, three T-14 Armata tanks and two IFVs, are disabled by precision munition strikes in the first eleven seconds. Not destroyed. Disabled. The distinction is intentional.

The lead elements stop moving at 05:47.

The column behind them stops too.

Not because they are destroyed. Because they can see what happened to the front.

The Air

Dassault Aviation’s Rafale F5 multirole fighters are holding at the edge of Russian radar coverage over northern Poland. Two aircraft, four crew. They have been airborne since 04:30.

Inside the cockpit, the flight management system runs on avionics supplied by Safran.

The fuel flow regulators and hydraulic line assemblies threading through the airframe are Senior PLC components — the British aerospace structures specialist whose precision engineering appears in the fluid systems of almost every front-line European combat aircraft without appearing on any front page.

The Rafale is a French machine. But it is also German bearings, British tubing, Swedish electronics, and Norwegian targeting data. It is, like the alliance that flies it, a coalition made of parts.

Safran’s AASM Hammer precision-guided munitions, loaded on the outer pylons of three aircraft, have been tracking target packages since 04:55. The weapons do not need GPS. Their inertial reference units — Safran again, vertically integrated from sensor to munition — navigate by the accumulated physics of their own movement through space. In an environment of total GPS denial, they remain accurate to within three metres.

The Airbus A400M tactical transport, circling at altitude over Lithuania, is acting as a communications relay node.

Its military satellite uplink connecting the tactical picture in Poland with the strategic command in Brussels without routing through any ground-based infrastructure that could be disrupted. Smiths Group‘s electronic systems integration package manages the A400M’s communications suite, keeping the command picture coherent across seven simultaneous encrypted channels.

BAE Systems’ electronic warfare aircraft operating under the call sign PHANTOM RELAY is doing something the Russians will spend considerable effort trying to understand after tonight.

It is transmitting, on seventeen separate frequencies, the electromagnetic signature of an aircraft type that does not exist. MINERVA designed the spoofed signature eighteen months ago from a library of observed emissions. The Russians are, right now, attempting to coordinate a response to an adversary that was never there.

The Sea

In the Baltic, the Charles de Gaulle has not moved. It does not need to. Its Dassault-built carrier aircraft are within strike range of every target in the engagement zone. The carrier group is the last argument, and it has not yet needed to be made.

Two hundred kilometres north, a Kongsberg-equipped frigate has its Naval Strike Missiles locked on three Russian surface combatants that moved into the Baltic seven hours ago under a routine exercise declaration. The exercise notification was filed correctly. The weapons are real. The missiles wait in their canisters with the patience of machines.

The primary satellite communications backbone provided by SES SA, the Luxembourg-based operator whose constellation provides the secure beyond-line-of-sight datalink connecting every asset in the theatre suffers a partial jamming attack at 05:51. Three transponders go dark. For four seconds, the Atrium loses its feed from the carrier group.

MINERVA reroutes automatically. The secondary path runs through Eutelsat Communications the French satellite operator whose civilian-band Ka-frequency transponders would not normally carry military traffic. But MINERVA had been given emergency access codes to seven Eutelsat transponders in 2034, after war-game exercises demonstrated that purely military satellite infrastructure was the most predictable and therefore most targetable element of the communications architecture.

Eutelsat does not know, on this morning, that its civilian bandwidth is carrying the command traffic of a continent in crisis. The signal looks, to any external observer, like broadband internet for a shipping company in the North Sea. It will be three years before that detail is declassified.

Exail Technologies the French specialist in inertial navigation and underwater robotics has its autonomous underwater vehicles operating in the shipping lanes south of Gotland, tracking the acoustic signatures of two Russian submarines that have been running silent since 05:20. The submarines are not a threat yet. The fact that MINERVA knows exactly where they are is the only reason they remain not a threat.

From the deck of the Italian frigate Caio Duilio, operating sixty nautical miles south of the Kongsberg-equipped group, a single loitering munition is released at 05:53. It carries an Avio SpA solid-fuel propulsion module — the same Italian propulsion technology that had been scaling from space launch vehicles down to tactical stand-off weapons since the early 2030s.

The munition climbs to observation altitude, circles its designated target area for eleven minutes, and then descends on a radar signature that MINERVA has been tracking since before Hartley walked into the Atrium. The target is a ground-based GPS jamming station. After 06:04, the jamming stops.

Colt CZ Group’s forward-positioned small arms and ammunition stockpiles are being drawn down in real time across the Polish forward operating bases. The consumption data feeds directly into Serco‘s logistics management system. Nobody runs out of anything.

The Table

At 06:12, a secure channel opens from Moscow to Brussels for the first time in nineteen months.

The conversation lasts four minutes. Its contents will never be made public.

What is known is that by 07:30, the armoured column has reversed its bearing. By 09:00, the Iskander batteries have been stood down and three of the six launch sites have been abandoned. By 11:00, the Baltic surface group has filed a return course to Kaliningrad.

The four people around the table do not celebrate. Wisniewski pours coffee. Brandt finally drinks his. Moreau calls the Charles de Gaulle and says two words. Hartley stands at the back of the room for a long time, looking at the map.

“They will try again,” she says. Not a question.

“Yes,” says MINERVA. “Different approach?” “Almost certainly. I am already modelling the variants.”

Outside, somewhere above the clouds, the sun is coming up over Brussels.

The baker in the Atomium district is pulling his second batch of loaves from the oven. The autonomous freight carrier has reached Liège. The fox in the Parc du Cinquantenaire found what it was looking for and has moved on.

The city never knew.

The Companies Behind the Story

Every system that appeared in the narrative above was built by a real company. Every company below is a confirmed constituent of the WisdomTree European Defence UCITS ETF. This story was constructed around the index not the other way around.

The Heavyweights — Top 10 Holdings

Thales SA (France) — 12.20%: The brain of the battlefield. Command systems, radar, SYNAPSE electronic warfare network, the central display of the Atrium. The company that sees everything.

BAE Systems PLC (UK) — 12.39%: Electronic warfare aircraft (PHANTOM RELAY), radar warning receivers, combat systems integration. Felt without being seen.

Rheinmetall AG (Germany) — 10.41%: Autonomous combat drones, ammunition, armoured vehicle platforms. The muscle of the ground response.

Leonardo SpA (Italy) — 8.41%: AULOS passive radar network, helicopter platforms, defence electronics. Teaching machines to hear what human ears cannot.

Saab AB (Sweden) — 8.32%: BAMSE surface-to-air missiles, Gripen fighters, electronic warfare. The outer intercept layer that turned Mach 6 to light.

Safran SA (France) — 8.19%: AASM Hammer precision munitions, aircraft engines, optronics, inertial navigation. Weapons that find their targets without GPS.

Rolls-Royce Holdings PLC (UK) — 7.95%: The engines that power European air and naval capability. Surveillance drone turbines in the night sky over Poland.

Dassault Aviation (France) — 6.67%: Rafale F5 carrier and multirole aircraft. The strike capability held in reserve — whose presence shaped the outcome without a weapon being fired.

Airbus SE (pan-European) — 6.60%: Defence satellites, A400M tactical transport relay platform. The eyes in orbit and the communications bridge over Lithuania.

Kongsberg Gruppen ASA (Norway) — 5.02%: Naval Strike Missiles, precision ground strike systems. The message sent without a word — locked, armed, waiting.

The Specialists — Mid-Tier Holdings

Hensoldt AG (Germany) — 2.27%: Radar, optronics, KALAETRON electronic intelligence. Cataloguing Russian radar emissions in real time, building the electronic order of battle.

Babcock International Group PLC (UK) — 1.86%: Defence engineering and forward support logistics. The unglamorous work that keeps everything else possible.

Melrose Industries PLC / GKN Aerospace (UK) — 1.61%: Structural aerospace components. The airframes that carry everything else.

RENK Group AG (Germany) — 1.36%: Advanced transmission systems for armoured vehicles. Keeping the heavy platforms faster than expected.

Smiths Group PLC (UK) — 0.85%: Electronic systems integration. Managing the A400M’s multi-channel communications suite.

Fincantieri SpA (Italy) — 0.84%: Naval shipbuilding. The hulls that carry ASTER 30 and the Baltic strike capability.

Indra Sistemas SA (Spain) — 0.84%: Defence electronics, real-time spectrum analysis. Identifying spoofing frequencies faster than any human signals officer.

QinetiQ Group PLC (UK) — 0.76%: Defence technology research and AI-assisted threat assessment. The institutional intelligence that taught MINERVA to think.

Exosens SA (France) — 0.73%: Photonics and optical sensors. Targeting data that works when GPS does not.

Serco Group PLC (UK) — 0.57%: Battlefield management and logistics coordination. Six forward operating bases, running simultaneously, nobody runs out of anything.

Chemring Group PLC (UK) — 0.38%: Energetic materials and sensor systems. The buried sensors that heard the column moving before anyone could see it.

INVISIO Communications (Denmark) — 0.32%: Tactical communications and hearing protection. Keeping human operators connected without producing a single detectable emission.

SES SA (Luxembourg) — 0.19%: Primary satellite communications backbone. The thread that held the coalition together — until it was jammed.

Colt CZ Group SE (Czech Republic) — 0.20%: Small arms and ammunition. Stockpiles drawn down in real time, consumption data feeding directly into the logistics management system.

Exail Technologies SA (France) — 0.17%: Inertial navigation and autonomous underwater vehicles. Tracking Russian submarines south of Gotland throughout the engagement.

The Smaller Specialists — Expanded Index

Cohort PLC (UK) — 0.15%: Defence technology — Electronic intelligence. The listening layer that taught MINERVA to extract signal from noise. Where QinetiQ taught the machine to think, Cohort taught it to hear.

Kitron ASA (Norway) — 0.15%: Electronics contract manufacturing. The circuit boards inside every Kongsberg Naval Strike Missile — precision components assembled in Norway for Norwegian weapons, bearing a quality certification that no one will ever read in the field, and without which the missile would be an expensive steel tube.

Avon Rubber PLC (UK) — 0.14%: Protective equipment — military respirators and sealed suits. The forward observers watching the column wore Avon-made masks against the dark. Unglamorous, essential, human.

Mildef Group AB (Sweden) — 0.13%: Ruggedised military computing. The hardened terminals at every operator station in the Atrium — built to function in electromagnetic chaos, readable only at direct angle, bearing the crown-and-circuit-board logo that nobody in the room ever thought about.

Eutelsat Communications (France) — 0.12%: Satellite communications — civilian-band infrastructure repurposed under emergency protocols. When SES was jammed, MINERVA rerouted through Eutelsat’s Ka-band transponders. The signal looked like North Sea shipping broadband. It was carrying the command traffic of a continent.

Senior PLC (UK) — 0.10%: Aerospace structures and fluid systems. The fuel flow regulators and hydraulic line assemblies inside the Rafale airframe — invisible components that make a coalition aircraft out of national parts.

Avio SpA (Italy) — 0.10%: Space and tactical propulsion. The solid-fuel propulsion module in the loitering munition launched from the Italian frigate Caio Duilio — Italian propulsion technology, scaled from launch vehicles to precision weapons, that put a GPS jamming station out of commission at 06:04.

Note on MBDA: Several systems in the narrative reference MBDA missiles. ASTER 30, SCALP, and related weapons. MBDA is the European missile systems consortium owned jointly by BAE Systems (37.5%), Airbus (37.5%), and Leonardo (25%). It is not a direct constituent of the index, but its principal shareholders. Three of the top four holdings provide the investment exposure. The alliance’s missile capability is present in the portfolio whether or not MBDA is named.

Part Three: The Investment Case

Epilogue: A Word of Thanks, and Mild Astonishment

I should confess something before I close.

I am a retail investor living in South East London, by the Thames. My formal qualifications in military strategy are, to put it generously, non-existent. My experience of armed conflict extends to an unfortunate incident with a neighbour’s cat and a territorial dispute over a recycling bin. I have never set foot in a NATO command centre, never spoken to an Air Marshal, and would not recognise an Iskander-M ballistic missile if one landed on the riverbank outside my window. Given recent geopolitical developments, I am choosing not to dwell on that possibility.

And yet, somehow, I wrote a story about all of these things.

I did not write it alone.

The research behind this trilogy involved hundreds of exchanges with Claude, Gemini and Perplexity. Patient, tireless, occasionally brilliant AI systems that indulged my questions without once suggesting I was out of my depth, even when I clearly was. But it was Claude that assembled the fiction. I gave it the parameters: a date, an index, thirty-two companies, and a request to make it feel real. What came back stopped me.

I do not say this lightly, because I am constitutionally suspicious of hyperbole. But the scene in the Atrium, the fox in the park, the city that never knew. I did not write those. I shaped the conditions for them. Claude wrote them. And when I read them back, I felt something I was not expecting to feel whilst researching an investment vehicle. I felt moved.

There is something genuinely remarkable happening in the space between human curiosity and machine capability. I came with an investor’s question and left with a story. The research became richer for the collaboration. The investment case became clearer for the fiction that preceded it. I am not sure I could have written Part Three without first having written Part Two. The story taught me what the numbers meant.

So: thank you to Claude, for constructing something I could not have constructed alone, and for having the literary decency not to name a character after itself.

And thank you to you, the reader, for following me down this particular rabbit hole. It is an unusual place for an investment essay to end up. I am aware of that. But then again, the world is an unusual place at the moment, and perhaps unusual times call for unusual frameworks.

We were dealt these cards.

I hope this helped you read the hand.

In February 2026, Amazon quietly dethroned Walmart as the world's largest company by revenue. The announcement barely made headlines. Wall Street was too busy obsessing over the $200 billion the company planned to spend on data centres and AI infrastructure in 2026 alone. And that, in a single moment, tells you everything you need to know about why Amazon is not merely the world's biggest company today. But why it will almost certainly become the first in history to post a trillion dollars in annual revenue.

This is not a simple story about a retailer that got very big. It is a story about a company that has quietly built the most diversified, most scalable, and most structurally advantaged revenue machine the world has ever seen. It is a story about three distinct but deeply interconnected businesses. Retail, cloud, and advertising. All compounding simultaneously, each feeding the others, each operating at a scale that most rivals cannot begin to imagine. And it is a story about timing. Amazon sits at the precise intersection of two of the most powerful economic forces of our era: the physical build-out of AI infrastructure, and the rapid displacement of human labour by intelligent machines.

The numbers make the ambition feel less audacious than it sounds. Amazon reported $717 billion in revenue for 2025. A trillion dollars represents an increase of roughly 40% from here. Given that Amazon has been growing revenues at 10-12% annually in recent years, and that the tailwinds behind its core businesses are accelerating rather than fading, a trillion-dollar revenue run-rate by the early 2030s is not a fantasy. It is a forecast.

The Machine Behind the Numbers

To understand why Amazon will reach a trillion dollars, you first need to understand what Amazon actually is. Most people still think of it as a shop. A very big, very fast shop that sells you things and delivers them the next day. That picture is outdated.

Amazon is better understood as a three-engine aircraft.

1. The first engine is retail. Both first-party sales and its vast third-party marketplace where millions of independent sellers pay Amazon fees to reach its customers.

2. The second engine is Amazon Web Services, or AWS, the cloud computing division that provides the digital infrastructure on which vast swathes of the modern economy run.

3. The third engine is advertising. A business that barely existed a decade ago, which now generates over $50 billion a year and is growing at roughly 20% annually.

Each of these engines is large enough to be a Fortune 500 company on its own. Together, they create a flywheel of compounding scale that is extraordinarily difficult to disrupt.

The Bloomberg piece rightly notes that Amazon's revenue without cloud computing was $588 billion in 2025. Still enormous, but over $100 billion short of Walmart's total. It is AWS that pushed Amazon over the top. That is the crucial insight. Amazon is not winning a retail war against Walmart. It is winning a different war entirely: the war for the digital and physical infrastructure of the twenty-first century economy. And it is winning that war from a position of near-unassailable strength.

AWS and the $85 Trillion Tailwind

Here is where the macro picture becomes inseparable from the Amazon investment thesis. At Davos earlier this year, Jensen Huang, the chief executive of Nvidia, put a number on the global build-out of AI infrastructure: $85 trillion over the next fifteen years. That is the estimated cost of wiring intelligence into every machine, factory, vehicle, hospital, and home on the planet. It is a number so large it strains comprehension. But what matters for our purposes is this: the overwhelming majority of that spend will flow through or be hosted on cloud infrastructure. And in cloud infrastructure, Amazon is the dominant player.

AWS currently commands roughly a third of the global cloud market. Microsoft Azure and Google Cloud are serious competitors, but AWS has a multi-year head start. The deepest enterprise relationships. The broadest portfolio of services. More importantly, the shift from AI experimentation (software models running in labs) to AI deployment (intelligent systems embedded in the physical world) is dramatically increasing the complexity and scale of cloud workloads. Training large language models, running inference at scale, orchestrating fleets of AI agents: all of this is extraordinarily compute-intensive. And compute-intensive workloads are high-margin workloads for AWS.

Jordi Visser and Raoul Pal, in their recent discussion of macro, crypto and AI convergence, made a point that deserves particular attention here. We are crossing an inflection point, they argued, where AI moves from being a software phenomenon to becoming embedded in the physical world. In drones, humanoids, vehicles, factories. This transition requires a colossal build-out of both digital infrastructure (data centres, chips, cloud services) and physical components (silver, copper, transformers, turbines). Amazon is extraordinarily well-positioned to capture the digital side of that build-out. Every company that wants to deploy AI at scale needs cloud compute. And cloud compute means, more likely than not, AWS.

The implication for revenue is straightforward. AWS grew at roughly 19% in 2025, reaching approximately $108 billion in revenue. If that growth rate is sustained. And the $85 trillion build-out thesis suggests it could accelerate. AWS alone could be generating $300 billion or more in annual revenue by the early 2030s. Add that to a retail and advertising business that will itself have grown substantially, and a trillion dollars stops looking like a stretch target. It starts looking like a conservative estimate.

The Advertising Engine Nobody Talks About

There is a third revenue stream that rarely features in discussions of Amazon's long-term potential, but which may ultimately prove to be its most powerful: advertising. Amazon's advertising business generated over $56 billion in revenue in 2025, growing at around 18-19% annually. That makes it, in revenue terms, already larger than the entire market capitalisations of most companies in the S&P 500.

The reason Amazon's advertising business is so formidable is structural. Unlike Google or Meta, which serve ads to people who are browsing or socialising, Amazon serves ads to people who are actively shopping. This is the highest-value advertising context that exists. A consumer searching for a coffee machine on Amazon is not passively consuming content. They are actively in the market for a product. The conversion rates are dramatically higher, the purchase intent is unambiguous, and brands will pay a premium for that access. As AI improves product discovery, personalisation, and ad targeting within Amazon's ecosystem, the monetisation potential of that intent-rich audience is enormous.

There is also a longer-term dynamic at work here. AI agents will increasingly handle the tedious aspects of shopping. Price comparisons, product research, reordering of household essentials. The role of the platform that those agents transact through becomes more valuable, not less. Amazon is actively positioning itself to be the transactional layer for AI-assisted commerce. If an agent is buying on your behalf, it is almost certainly buying through Amazon. And every transaction that passes through the platform generates fees, data, and advertising revenue.

The Retail Business Is Not Standing Still

It would be a mistake to view Amazon's retail business as mature and therefore unexciting. It is true that Walmart's e-commerce operation is growing faster than Amazon's in the United States this year. But Amazon's retail advantage is not primarily about domestic e-commerce market share. It is about global scale, logistics infrastructure, and the flywheel effects of Prime membership.

Amazon's Prime membership programme with over 200 million subscribers globally is one of the most valuable loyalty ecosystems ever created. Prime members spend dramatically more than non-members. They buy more frequently, they are less price-sensitive, and they are significantly more likely to start product searches on Amazon rather than Google. This makes Prime a moat around the retail business that is genuinely difficult to replicate. Walmart is trying, with its Walmart+ programme, and it is making progress. But Amazon has a decade-long head start and the brand loyalty that comes with it.

Beyond Prime, Amazon is expanding its physical retail presence and its logistics network in ways that will support sustained revenue growth. The push into rural America, which the Bloomberg article notes will be increasingly competitive with Walmart, represents a meaningful expansion of Amazon's addressable market. And Amazon's logistics capabilities, built at enormous expense over the last two decades, now constitute a competitive advantage that is effectively impossible to replicate at speed. The company can now deliver more than half of its own parcels, reducing its dependence on third-party carriers and lowering unit costs at scale.

The Structural Shift: From Software Abundance to Physical Scarcity

There is a broader macro framework that makes the Amazon thesis even more compelling. As Visser argued persuasively, we are living through a fundamental shift in where economic value is created. For fifteen years, the dominant investment thesis was software abundance: companies that could scale with near-zero marginal cost, generating extraordinary profit margins and compounding revenue at rates that physical businesses could never match. That era is ending, not because software becomes less important, but because AI makes software itself abundant. When any reasonably competent developer can build a functioning application in hours using AI tools, the moat around any individual software product narrows dramatically.

The new scarcity is physical. It is the silver and copper required to build the infrastructure of AI. It is the data centres that host the models. It is the power grid that feeds the compute. It is the logistics networks that move goods in the real world. And it is the trust relationships and data assets that AI cannot replicate overnight. Amazon, uniquely among the major technology companies, has invested as heavily in physical infrastructure as in digital. Its warehouse network, its delivery fleet, its data centres: these are the new oil wells. They are hard to build, expensive to maintain, and they generate enormous value for anyone who controls them.

This is why the comparison to Walmart, while interesting, misses the point. Walmart is a physical retail business attempting to build digital capabilities. Amazon is a digital-physical hybrid that has already built the infrastructure for both. In a world where AI is driving demand for both cloud compute and physical delivery, Amazon sits at the crossroads.

The Path to a Trillion

Let us be concrete about the arithmetic. In 2025, Amazon generated $717 billion in revenue. For context, that is approximately the GDP of the Netherlands. To reach $1 trillion requires roughly 40% revenue growth from current levels. At a 10% compound annual growth rate which by the way is below Amazon's recent average that milestone arrives in 2029. At 12%, which is closer to the recent trend, it arrives in 2028. If AWS continues to grow at 19% or above, as the AI infrastructure build-out accelerates, the consolidated growth rate could be higher still.

The more interesting question is what happens to the mix of revenue as Amazon approaches and then exceeds a trillion dollars. The high-margin AWS and advertising businesses are growing faster than the lower-margin retail business. This means that even as headline revenue scales impressively, the profit margins of the combined business are likely to expand. Amazon in 2030 or 2032 may not just be the first trillion-dollar revenue company in history. It may also be generating operating income of $200 billion or more annually. The combination of scale and margin improvement makes Amazon one of the most compelling long-term investment stories in the global equity market.

There are risks, of course. Regulatory pressure particularly in Europe, but increasingly in the United States remains a genuine threat to Amazon's ability to operate its marketplace on its own terms. The capital intensity of the AI build-out is enormous, and there is a legitimate debate about whether the returns on $200 billion in annual capex will be sufficient to justify the expenditure. Competition in cloud from Microsoft and Google is real and intensifying. And a severe global recession could compress consumer spending in ways that hurt the retail business.

But these risks need to be weighed against the structural tailwinds. The $85 trillion AI infrastructure build-out is not optional. It is the central geopolitical and economic project of the next decade. Every nation, every corporation, every institution that wants to remain competitive needs to deploy AI at scale. That means they need cloud compute. And the odds that AWS is not a major beneficiary of that spend are very low indeed.

The Quiet Revolution

Amazon's ascent to the top of the global revenue rankings passed without fanfare. Wall Street, fixated on AI spend and margin compression, barely noticed that the world's largest company by revenue had just changed for the first time in over a decade. This is, in many ways, is a metaphor for Amazon's entire history. The company has consistently been underestimated, its investments consistently misunderstood as recklessness rather than recognised as the construction of durable competitive advantages.

Jeff Bezos spent years being accused of running a non-profit. He was investing in warehouses. Andy Jassy is now accused of the same crime. He is investing in data centres. The pattern is identical. And if the past is any guide to the future, the investors who focus on the long-term structural advantages being built will be the ones who benefit most from what comes next.

Amazon will be the first company in history to generate a trillion dollars in annual revenue. It will not happen in a single dramatic moment. It will happen gradually, then all at once. The way all genuinely transformative things do. The machine is already running. The flywheels are already spinning. The only question is whether you are paying attention.

Sources: Bloomberg (Spencer Soper, February 2026); Raoul Pal / Jordi Visser, Global Macro Investor Davos discussion (January 2026); Jensen Huang, Davos interview with Maria Bartiromo (January 2026); Amazon full-year 2025 earnings.

As I am doing the final edits of this essay America attacked Iran and the Ayatollah was no more.

I want to be clear about what this piece is not. It is not political commentary. It is not military strategy. I have no particular axe to grind about who is right or wrong in any of the conflicts that form the backdrop to what you are about to read. My purpose has not changed from the day I started writing: to understand how Europe is rearming itself, why it matters, and what it means for investors.

But this moment. Sitting at my desk. Watching breaking news scroll across my screen. Knowing I was putting the finishing touches on an essay about European defence spending. It felt like the world making a point I had been trying to articulate for weeks.

This is how it always goes. You think you are writing about one thing. Then the pattern reveals itself.

I am a retail investor. That is the honest starting point. Not a professional analyst. Not an academic. Not someone with institutional resources behind him. Just someone who watches how money moves and cannot stop asking the deeper questions underneath the obvious ones. Why this? Why now? Why does this shape keep repeating?

Four weeks ago, I sat down to write what I imagined would be a clean 3,500-word essay on the WisdomTree European Defence Fund. It’s constituents, the investment case, the macro backdrop. Straightforward enough.

It did not stay straightforward.

Because once you start pulling the thread. Once you ask why is Europe spending €800 billion on defence right now. You cannot stop at the press release. You end up in the 1990s, watching the post-Cold War order collapse in slow motion. Then you end up in the ruins of the Soviet Union, asking who benefited from the chaos. Then, somehow, you end up on the Central Asian steppe in the 12th century, staring at a Mongol army, realising that the pattern you are looking at is ancient.

That is what this trilogy became. Three essays.

Part one - this essay - lays the foundations. From Genghis Khan to the €800 billion package, tracing the long history of how external pressure forces European transformation.

Part two looks forward. AI-enabled warfare, autonomous systems, and how the battlefield of 2035 is being built today.

Part three returns to where I started. The WisdomTree European Defence Fund, its constituents, and the investment case as I see it.

I want to be transparent about my method, because I think it matters. I have tried hard not to be conspiratorial. Every argument I make is rooted in historical record, published research, and sourced analysis. Where I have reached for deeper patterns like the financial architecture of intervention, the political economy of reconstruction, the resource logic underneath geopolitical competition, I have tested those arguments relentlessly. My research companions throughout this process have been Claude and Gemini, which I have used not to generate conclusions but to spar, to stress-test, to push back, to find the holes.

What you are reading is the result of four weeks of intense reading, questioning, and rewriting. It started as an ETF note. It became something I could not have predicted when I began.

The world has a way of doing that.

Where it all started

I am reading A History of the World in 47 Borders: The Stories Behind the Lines on Our Maps, by Jonn Elledge. It looked like the sort of book you’d browse on a lazy Sunday afternoon. It turned out to be the sort of book you can’t put down.

Each chapter is short, five pages, maybe six and covers a single border or boundary somewhere in the world. Some are famous. Most aren’t. All of them turn out to matter more than you’d expect. Elledge writes the way a very clever friend talks: warm, funny, occasionally outraged, never dull.

I got to the chapter on Genghis Khan. It’s titled The Open Borders Policies of Genghis Khan, with a subtitle that made me gasp: The Man Who Created the Modern World. I read it twice. Then I sat back contemplating. Something about Europe today, about Russia and Ukraine, about the strange way history keeps rhyming with itself even when no one is paying attention.

What follows is where that train of thought ended up. It starts in 1206. It ends in 2025. And somewhere in the middle, it raises a question I think matters enormously for the next two decades.

Is Europe finally, irreversibly, growing up?

The Horsemen at the Gate

It is 1206.

On the vast steppe north of the Gobi Desert, a gathering of tribes has just given one man a name that will echo for eight centuries: Genghis. It means, roughly, great ruler. The man who received it was born Temujin, the son of a chieftain murdered by rivals when his boy was nine years old. He had spent his childhood as an outcast, hunted across the grasslands, learning early that in a world of warring clans, survival is the only currency that matters.

Within two decades of being given that name, Temujin had unified the Mongol tribes, crushed the Uyghurs and the Tanguts, and toppled the Jin dynasty in northern China. By 1220, he controlled most of central Asia. By 1223, his generals were raiding deep into Russia. The Mongol war machine moved at a speed that nothing in the medieval world had seen before. It left pyramids of skulls in its wake. One English monk, writing from the safety of his monastery, which the Mongols never came within 1,500 miles of, called them a detestable nation of Satan that poured out like devils from Tartarus.

That reputation was earned. The invasions Genghis Khan led killed an estimated 40 million people roughly one in ten of every person alive at the time. The Tanguts, a Buddhist civilisation in what is now central China, were systematically exterminated. Their culture, their cities, their records. Historians have called it the first recorded attempt at genocide. Entire regions of Persia and Central Asia were depopulated. The Islamic world, then the intellectual centre of the planet, never fully recovered its political or intellectual leadership.

And yet. And yet.

Genghis Khan died in 1227. The empire he built kept growing. It swallowed Persia, Tibet, Korea, Crimea. His grandson Kublai Khan established the Yuan dynasty in China in 1271. The empire eventually fragmented into four successor states. The Golden Horde in eastern Europe, the Ilkhanate in the Middle East, the Chagatai Khanate in central Asia, and the Yuan in China. Even in fragmented form it remained the largest contiguous land empire in history.

What it also became, almost accidentally, was the first greatest free-trade zone the world had ever seen.

The Peace After the Terror

To understand what the Mongol Empire did to the world, you need to understand what the world looked like before it.

Eurasia in the twelfth century was a patchwork of competing kingdoms, warring confederations, petty rulers demanding tribute, and bandits preying on anyone foolish enough to travel far.

The Silk Road which was the network of caravan routes stretching 4,000 miles from China to the Mediterranean had been operating since the second century BCE. But it was dangerous, fragmented, and riddled with tolls. Merchants rarely travelled its full length. They handed goods off in stages, each middleman taking his cut.

The Mongols changed this almost overnight. Their nomadic and pastoral lifestyle meant they produced almost nothing that required skilled manufacturing. No pots, no clothes, no weapons. They depended entirely on trade to acquire these things. Where Chinese civilisation looked down on merchants as a low-status class who created nothing of value, the Mongols prized them as the people who brought them stuff. This was not a philosophical position. It was practical hunger.

Once the conquering was done, the Mongols built. They established a postal relay system called the Yam stretching from Eastern Europe to the Pacific, a sophisticated network of stations providing food, lodging and fresh horses to messengers and merchants alike. They standardised tariffs (yes even them, not just Donald) across the empire. They created trade associations called ortoghs, pooling merchant resources and providing something like insurance schemes and low-interest loans. They patrolled the roads. The historian Jack Weatherford has written that it was said a girl could walk from one end of the empire to the other with a gold coin in her hand and not be troubled.

This period became known as the Pax Mongolica. The Mongol Peace. It lasted roughly from 1250 to 1350. Scholars now regard it as the moment when globalisation truly began, when, for the first time in history, the civilisations of Europe and China came into direct contact rather than knowing each other only through intermediaries.

What flowed along those suddenly safe roads was extraordinary. From east to west: silk, spices, porcelain, paper money, printing techniques, gunpowder, the compass, advanced mathematics, Islamic astronomy, pharmacology, banking instruments and the idea of insurance. From west to east: silver, fine cloth, horses and linen. Advanced Islamic medicine and Chinese pharmacology cross-pollinated. The Venetian merchant Marco Polo made his famous journey to China during this period: the first time an Italian could plausibly travel between Venice and Beijing along safe, well-maintained roads and return to tell the tale.

The city of Genoa signed a treaty with the Golden Horde giving it exclusive trading privileges in the Black Sea. Genoese merchants bought a port city from the Mongols and established Kaffa. It became a flourishing settlement and the gateway through which the goods and technologies of Asia flooded into Europe. The wealth that cascaded into the northern Italian city-states of Venice, Genoa and Pisa from eastern markets became the financial foundation for something that would reshape Western civilisation entirely. Historians now argue that the Pax Mongolica helped kick off the Renaissance.

In 1200, Western Europe was, by any measure, a backwater. Politically fragmented, technologically behind, and struggling to feed itself. By 1400, it had acquired a critical mass of new technologies, new ideas, new financial instruments, and a burning appetite for eastern luxuries. When the Pax Mongolica collapsed, those luxuries stopped arriving. And so some Europeans decided to go looking for them. They built boats. Those boats eventually reached India in the east and the Americas in the west.

Genghis Khan was the man who created the conditions for this Age of Discovery. Indirectly, inadvertently, and with his characteristic method of organised violence.

The Aggressor Strengthens What It Attacks

Here is the central irony of the Mongol story, and the one most relevant to our own time. The region that benefited most from the Pax Mongolica was the one the Mongols never conquered.

Western Europe escaped the full force of Mongol invasion almost by accident. The Golden Horde had penetrated Poland and Hungary by 1241. Nothing stood between the Mongol cavalry and the rest of the continent. Then, in 1242, Genghis’s son and successor, Ogedei Khan died unexpectedly. The Mongol commanders turned back. Internal succession politics mattered more to them than the conquest of a cold, soggy peninsula at the western edge of Eurasia. Europe could wait. In fact, it waited forever. The blow never came.

Europe was therefore uniquely positioned: close enough to the Mongol empire to benefit from the trade routes and technology transfer it enabled, but spared the devastation that those who resisted received. The Islamic world, which had been the intellectual and commercial centre of the medieval world, was hammered. China was occupied and altered. Russia or rather the Kievan Rus’, the principality from which Russia, Ukraine and Belarus all claim descent was conquered and forced to pay tribute to the Golden Horde for two and a half centuries.

Western Europe got the upside without the downside. It received gunpowder, banking, printing and the compass. It did not receive the conquerors. The threat that somewhere beyond the eastern horizon, something vast and terrible existed was enough to shift European minds towards trade, exploration and eventually expansion.

External threat, whether it arrives fully or merely hovers at the frontier, forces transformation on those who face it. The alternative of comfortable fragmentation, the endless quarrels of petty kingdoms with no common enemy to unite against produces stagnation. Europe’s great weakness throughout the medieval period was precisely this: it was a continent of warring micro-powers, each too busy fighting its neighbours to build something larger. The Catholic Church alone provided any semblance of shared identity, and even that was deployed as often to fragment as to unify.

What the shadow of the Mongols provided and what its successor threats have provided ever since was a reason to grow up.

Europe’s Eight Centuries of Comfortable Forgetting

After the Mongol threat receded and the Renaissance bloomed, Europe spent the next several centuries doing what it does best when not under existential pressure: fighting itself.

The Thirty Years War. The Wars of Religion. Napoleon. Two world wars that killed roughly one hundred million people, mostly Europeans killing other Europeans. The continent that had been accidentally unified by Mongol trade proved constitutionally incapable of maintaining that unity once the threat disappeared.

What finally forced a different approach was exhaustion with a sprinkling of wisdom. After 1945, the continent lay in ruins. The project that became the European Union was born from the cold calculation of nations that had destroyed each other twice in thirty years and could not afford a third attempt, bolstered with a mild seasoning of idealism. It was also born in the shadow of a new threat: the Soviet Union, which had acquired nuclear weapons by 1949 and now sat on Europe’s eastern border with an army numbering in the millions.

NATO was the military expression of this new unity. The Marshall Plan was its economic one. Europe, for the first time in its history, began to act as something approaching a coherent entity to avoid extinction.

Then the Soviet Union collapsed. The Berlin Wall came down in 1989. The threat dissolved. And Europe did what it always does without a threat to focus the mind. It relaxed.

Defence spending fell across the continent throughout the 1990s and 2000s. The peace dividend, as it was called, was spent on welfare states, infrastructure and consumption. By 2014, when Russia annexed Crimea, the most blatant violation of European territorial sovereignty since 1945, the response was muted. Sanctions, yes. Strongly worded statements, certainly. But no fundamental reappraisal of the security architecture that had been allowed to atrophy since 1989. In 2021, only seven of twenty-one EU member states that were also NATO members spent the agreed 2% of GDP on defence. European NATO members had collectively underfunded their own defence by an estimated $850 billion since the Wales Summit commitment was made in 2014.

Meanwhile, Russia increased its defence budget by 300% over the same decade. China increased its by 600%.

And then there is the peculiar case of Britain which, by 2016, had decided that the entire European project was not for it. The Brexit vote was portrayed at the time, and by many since, as an act of national self-harm: a wealthy, globally connected country voluntarily cutting itself off from its largest trading partner, motivated by a politics of grievance and nostalgia. It was messy. It was expensive. It fractured British politics for the better part of a decade.

Yet when the Russian tanks rolled into Ukraine in February 2022, it was Britain. Outside the EU, unburdened by the institutional caution and consensus-seeking that the Brussels machinery requires. Britain moved fastest and most decisively. She was the first Western country to supply Ukraine with anti-tank missiles, doing so in January 2022, weeks before the invasion. She was the first to provide long-range strike capabilities. She launched Operation Interflex, training over 51,000 Ukrainian soldiers on British soil. Prime Minister Boris Johnson flew to Kyiv in April 2022, a visit that reportedly stiffened Ukrainian resolve at a moment when some in the West were quietly urging Zelensky to negotiate. Britain committed military aid at a level that punched well above its weight.

The irony is: a country that had rejected the political union of Europe became, in the hour of Europe’s greatest post-Cold War crisis, one of its most reliable military defenders. Brexit Britain, derided for turning inward, turned outward when it mattered. Whatever one thinks of Brexit as an economic proposition and the costs have been real, as a security actor in 2022, Britain did not behave like a country that had walked away from Europe. It behaved like a country that had never left.

Europe was, in short, doing exactly what it always does in the absence of existential pressure. Being comfortable, fragmented and optimistic about things it had no particular reason to be optimistic about. Britain, for all its political chaos of the preceding decade, was doing what island nations on the edge of continents have historically done best, moving quickly when the threat became undeniable.

The System Behind the Horsemen

There is something the Mongol story does not tell you. It tells you about conquest, about trade, about the unintended consequences of organised violence. What it does not tell you is who profits most from the aftermath. How, over centuries, those who profit most have become very good at engineering the conditions for their own enrichment.

This is not a conspiracy. It is something more mundane and more durable than a conspiracy. It is a system. Forgive me if I am drifting, it’s an important segue to understand the system we are living in.

To understand the world that Europe now finds itself in. Th world that produced the Ukraine war. The rearmament. The €800 billion defence plan. You need to understand how that system was built. It was built incrementally, after 1945, with remarkable consistency of purpose. And once you see its architecture, the events of the last three years look less like a crisis and more like a logical outcome.

The Architecture of Managed Sovereignty

After the Second World War, the United States and its allies built a set of international institutions that would, ostensibly, prevent the conditions that had produced two catastrophic wars: the International Monetary Fund, the World Bank, and a network of central banking arrangements anchored by the Bank for International Settlements in Basel. The stated purpose was stability. Prevent the competitive currency devaluations and economic nationalism that had destabilised the 1930s and helped bring fascism to power.

The actual mechanism was more precise. An IMF loan came with two non-negotiable conditions. The borrowing country had to install a central bank subordinate to the wider international banking architecture. Thus effectively ceding monetary sovereignty. And it had to privatise its natural resources, opening them to foreign investment. Once both conditions were met, the country’s assets became accessible to Western capital. Its debt, denominated in dollars, could be rolled over indefinitely, keeping the borrower perpetually dependent. Its oil, its minerals, its land, its public infrastructure. The country remained nominally sovereign. Its government held elections. Its flag flew over its parliament. But the commanding heights of its economy were no longer fully its own.

This is not a radical interpretation. It is the documented analysis of economists including John Perkins, whose memoir Confessions of an Economic Hit Man detailed how this process worked in practice across Latin America, Africa and Asia. It is the framework that Jeffrey Sachs. He helped design shock therapy programmes in Poland and Russia in the early 1990s before becoming one of their most prominent critics. He has spent decades trying to explain to Western audiences. It is what Yanis Varoufakis discovered when, as Greek Finance Minister in 2015, he attempted to negotiate with the IMF and the ECB on behalf of a country that had technically been bankrupt for years but was being kept alive on rolling debt in order to service loans that were, in practice, bailing out French and German banks.

The system is not secret. It operates in daylight. It is just not the story that gets told in most Western newspapers.

Russia: The Country That Refused

When the Soviet Union collapsed in 1991, the American led financial-industrial complex moved quickly. This was the opportunity of a generation: a vast country of immense natural resources, a demoralised population, a political class that had watched its entire ideological framework disintegrate overnight. Jeffrey Sachs and a team of Western economists arrived in Moscow to advise on the transition to a market economy. The prescription was shock therapy: rapid privatisation, currency liberalisation, removal of price controls, all at once.

What followed was one of the largest transfers of public wealth into private hands in recorded history. State enterprises. Oil companies, mineral resources, industrial assets built over seven decades of Soviet investment. Sold at fractions of their value to a small group of well-connected insiders who became, overnight, some of the wealthiest individuals on earth. These were the oligarchs. Many of them moved capital offshore. Many had close relationships with Western financial institutions. The Russian economy contracted by roughly 40% between 1990 and 1998. Male life expectancy collapsed. The country was a superpower a decade earlier. By the late 1990s, it effectively became a client state deeply in debt. It’s resources accessible to Western capital. It’s central bank integrated into the international system.

Then Vladimir Putin arrived.

Whatever one thinks of Putin. There is no shortage of evidence for his authoritarianism, his brutality, and his willingness to use violence as a political instrument. What he did to the oligarchs is a matter of record. He arrested the most prominent ones. He renationalised the oil assets that had been most flagrantly stripped. He paid down Russia’s foreign debt. The IMF loan was repaid in full by 2005, years ahead of schedule. He kept the Rouble as a genuinely sovereign currency, not pegged to the dollar. He built up a sovereign wealth fund from oil revenues. He withdrew Russia, in other words, from the architecture of managed dependency that the 1990s had imposed on it.

This is not a defence of Putin. It is an explanation of why he became the enemy. A country that controls its own resources, runs its own currency, owes nothing to the IMF, and is not subordinate to the Bank for International Settlements is, by definition, outside the system. And a country outside the system is, from the system’s perspective, a problem to be solved.

Ukraine: The Asset

Ukraine, in this light, is primarily a story about which economic architecture the country would operate within. Though democracy versus autocracy, or NATO expansion versus Russian security interests are real and matter.

The 2014 Maidan uprising, which removed President Yanukovych and installed a pro-Western government, was a genuine popular revolt against a corrupt government. It was also the moment at which Ukraine began its decisive integration into Western financial structures. IMF loans, privatisation programmes, association agreements with the EU. Within months of the new government taking power, an American investment firm had begun acquiring Ukrainian agricultural land. BlackRock, the world’s largest asset manager, would later be formally appointed to advise on the reconstruction of Ukraine’s economy. A role that comes with an intimate knowledge of which assets are available and at what price.

None of this required coordination or conspiracy. The system does not need a puppet master. It needs aligned incentives. In a country receiving billions in Western financial support, whose government depends on continued Western backing, whose reconstruction will be financed by Western capital, the incentives align automatically. The financial-industrial system does not need to install leaders. It needs to ensure that whoever leads makes decisions compatible with the system’s continued operation.

What the Ukraine war produced, in financial terms, was a bonanza for the Western defence industry. Primarily American. NATO countries committed hundreds of billions in military aid, the overwhelming majority of which was spent on weapons manufactured by Lockheed Martin, Raytheon, General Dynamics, and Boeing. The war, in other words, simultaneously served as a massive stimulus programme for the US defence-industrial base. A mechanism for drawing Ukraine firmly into the Western financial orbit. And finally a demonstration to Europe that its own security dependency on Washington remained total.

The Pivot and Why It Matters

This is where the story turns. The €800 billion ReArm Europe plan is the first serious attempt by Europe to build sovereign defence-industrial capacity outside the American system. Whether or not its architects fully intend it this way.

When European governments buy from Rheinmetall instead of Raytheon, the money stays in Europe. When they fund Leonardo’s radar systems instead of Lockheed Martin’s, the jobs, the intellectual property, and the supply chains are European. The SAFE mechanism’s buy-European mandate requires that defence spending go to companies with independent European decision-making. This is structurally, a partial economic declaration of independence.

It will not transform everything overnight. Europe remains deeply integrated into the Western financial system, and its governments are not about to replicate Putin’s move of paying off the IMF and floating their own currencies. But in the specific domain of defence, which is where sovereignty is ultimately enforced, something is shifting.

The question this raises — the question Part 2 and 3 of this essay attempts to answer — is: which companies sit at the centre of that shift? And what does it mean for the investor who understands what is actually happening, rather than the version of events that makes it onto the front page?

The Horsemen Return

24th February 2022.

Russian armoured columns crossed the Ukrainian border from multiple directions. The stated aim was a special military operation. Kremlin’s bureaucratic euphemism for what was, in practice, an attempt to erase a sovereign nation of forty-four million people.

The invasion did not go as planned. Ukraine’s armed forces, backed by Western weapons and intelligence, stopped the advance on Kyiv within weeks. The war settled into a grinding attritional struggle in the east and south of the country. It became the largest land war in Europe since 1945. It also became the most consequential geopolitical event in Europe since the Berlin Wall came down and its effects on European defence thinking have been, in a very precise historical echo, analogous to the effect the Mongol shadow had on medieval Europe.

The transformation has been rapid and, by European standards, remarkable.

Germany, which had not supplied lethal weapons to a conflict zone since 1945. A taboo so deeply embedded in post-war German identity that it seemed structural rather than political. She reversed course within days of the invasion. Chancellor Olaf Scholz stood before the Bundestag and declared a Zeitenwende: an epochal shift. Germany committed €100 billion in a special defence fund. He promised to meet and then exceed NATO’s 2% spending target. For a country whose entire post-war strategic culture had been built around never again becoming a military power, this was a civilisational reorientation.

Germany was not alone. Neutral for two centuries, having sat out both world wars Sweden applied to join NATO in 2022 and became a full member in 2024. Finland, sharing a 1,340-kilometre border with Russia, did the same. Poland, which occupies the geographical position that has made it the perpetual victim of European great-power struggles, increased defence spending by 50% in 2023 and a further 17% in 2024. By 2025, Poland was spending 4% of GDP on defence, more than any other NATO member, including the United States.

Across the EU, collective defence spending rose from €218 billion in 2021 to €326 billion in 2024, an increase of more than 30% in three years. In March 2025, the European Commission published its ReArm Europe Plan, unlocking a potential €800 billion in additional defence funding through exemptions to debt limits, a new €150 billion loan mechanism called SAFE, and mobilisation of private capital. For the first time in EU history, a Commissioner for Defence and Space was appointed. The European Parliament’s security subcommittee was elevated to a full standing committee. By 2025, all NATO allies were expected to meet the 2% spending target for the first time, compared with just three allies meeting that target in 2014.

The pattern is identical to the one the Mongols set in motion eight centuries ago. A fragmented continent, comfortable in its divisions, jolted into coherence by an external threat it could no longer pretend did not exist. The aggressor, by attacking, has inadvertently strengthened the thing it sought to destroy.

The Nature of the Lesson

History does not repeat.

But it rhymes, and sometimes it rhymes loudly enough to hear from across the centuries.

The Mongol invasion teaches three things that are directly relevant to Europe in 2025 and beyond.

The first is that threat unifies. Europe’s chronic problem is fragmentation. It is a continent of nations with different languages, histories, interests and threat perceptions. The further a country sits from Russia’s border, the less urgently it has historically felt the need to spend on defence. Spain’s defence budget in 2022 was, in real terms, exactly equal to what it had been in 2008. Italy moved slowly. Western Europe, broadly, treated the 2% target as an aspiration rather than a floor. What changed this was not argument, not diplomacy, not the patient institutional work of Brussels. What changed it was tanks crossing a border in the night.

The second lesson is that the aggressor inadvertently strengthens what it attacks. Russia’s strategic goal in February 2022 was to prevent Ukraine from joining NATO. To keep it in Moscow’s sphere of influence. To demonstrate that NATO expansion was dangerous and that the Western alliance was too weak and divided to respond. Within a year of the invasion, Sweden and Finland had joined NATO, two countries that had been determinedly neutral for generations. Within two years, Germany had undergone the most significant strategic transformation in its post-war history. Within three years, Europe was spending at a scale that would have been politically inconceivable in January 2022. Russia’s invasion produced the exact opposite of its intended effect.

This, too, is the Mongol lesson.

Genghis Khan did not intend to give Europe banking, printing and the compass. He did not plan the Renaissance. He was doing what he always did: conquering, killing and moving on. The consequences: the Pax Mongolica, the trade routes, the technology transfer, the wealth that flooded into northern Italy and eventually financed the voyages of Columbus were entirely unintended. The monster built the world it was trying to destroy.

The third lesson is the most important, and the most uncomfortable. The transformation does not last without the threat. The period of forced coherence that the Mongol shadow enabled eventually gave way, once the empire fragmented and the threat receded, to the same European divisions that had always existed. The Cold War unity of NATO was real and effective and it eroded almost immediately once the Soviet threat disappeared. Europe’s current rearmament and integration is driven by fear. Fear is a powerful motivator. It is also temporary. The question is whether the structures being built now: the spending commitments, the industrial co-operation, the joint procurement, the SAFE mechanism, the new defence Commissioner can outlast the immediate crisis and become durable.

The Mongols did not give Europe permanent unity. They gave it a window. Europe used that window to produce the Renaissance, the Age of Exploration and eventually the modern world. Whether the current window produces something equally transformative depends on choices that are being made right now.

Coda: The Shape of What Is Coming

Jonn Elledge’s chapter on Genghis Khan is subtitled The Man Who Created the Modern World. It is a provocation, but not an unfair one. The argument is not that Genghis Khan was good. He was, by any conventional measure, one of the most destructive individuals in human history. The argument is that the consequences of his actions were far larger and stranger than his intentions and that the largest of those consequences fell on people he never reached.

Vladimir Putin is no Genghis Khan. The Russian military is not the Golden Horde. But the structural dynamic is the same: an authoritarian power with expansionist ambitions launching an attack that it expected to be quick and decisive, producing instead a prolonged struggle that is reshaping the continent it sought to dominate.

European defence spending is rising at a rate not seen since the Cold War. European defence industry, which had been hollowing out for three decades, is being rebuilt. New companies are emerging. New capabilities are being funded. The continent that spent thirty years cashing its peace dividend is now spending at a scale that suggests it has genuinely internalised the lesson, at least for now.

What Genghis Khan accidentally gave Europe was a connected world: a world in which ideas, technologies and capital could move across borders that had previously been impermeable.

What Putin has accidentally given Europe is a common purpose: a shared recognition that the continent’s security cannot be permanently outsourced to Washington, and that the fragmentation which has been Europe’s defining political pathology for a thousand years is a luxury it can no longer afford.

Whether Europe uses this moment as cleverly as the Italian merchants used the Pax Mongolica remains to be seen. History suggests the window will not stay open indefinitely.

The horsemen always eventually ride away.

The question is what you build while they are still at the gate.

Part 2: The Ungiven War - A work of speculative fiction describing the battleground of the future

Part 3: The European Defence Fund: who is building it, what it is buying, and where the opportunity lies.

References & Sources

1. Primary Source for the Genghis Khan Narrative: Elledge, Jonn. A History of the World in 47 Borders: The Stories Behind the Lines on Our Maps. Wildfire / Headline Publishing Group, London, 2024. ISBN: 9781472298508. Chapter: ‘The Open Borders Policies of Genghis Khan — The Man Who Created the Modern World’, pp. 58–65. The Genghis Khan narrative, the Pax Mongolica framing, and the argument that western Europe benefited most from an empire it never directly encountered are all drawn from and credited to this chapter.

2. Historical — Pax Mongolica & the Mongol Empire: Weatherford, Jack. Genghis Khan and the Making of the Modern World. Crown Publishers, New York, 2004. The foundational scholarly argument that the Pax Mongolica represented the beginning of globalisation, and that direct Europe-China contact began during this period.

3. O’Rourke, Kevin H. and Ronald Findlay. Power and Plenty: Trade, War, and the World Economy in the Second Millennium. Princeton University Press, 2007. The argument that ‘globalisation began with the unification of the central Eurasian landmass by the Mongol conquests.’

4. Rossabi, Morris. The Mongols and Global History. W.W. Norton, 2011. On the first direct personal contact between Europe and China in the thirteenth century.

5. Wikipedia: Pax Mongolica. Consulted February 2025. For trade goods, relay systems, banking and insurance instruments introduced to Europe during this period.

6. Federal Reserve Bank of Richmond. ‘Genghis Khan, Trade Warrior.’ Econ Focus, Q4 2021. On the asymmetric benefits of the Pax Mongolica and the stimulus it gave to European exploration.

7. Bank for International Settlements. About the BIS. BIS.org. The BIS was established in 1930 and serves as the central bank for central banks, coordinating monetary policy among member institutions.

   International Monetary Fund. Conditionality. IMF.org, updated 2023. On the standard conditions attached to IMF loan programmes, including fiscal adjustment, structural reforms, and institutional governance requirements.

8. Perkins, John. Confessions of an Economic Hit Man. Berrett-Koehler Publishers, San Francisco, 2004. ISBN: 9781576753019. The foundational first-person account of how large infrastructure loans to developing countries were structured to ensure capital flows back to US corporations, with debt dependency as the mechanism of political compliance. Spent more than 70 weeks on the New York Times bestseller list.

9. Perkins, John. The New Confessions of an Economic Hit Man. Berrett-Koehler Publishers, 2016. Updated edition expanding the original thesis to include more recent case studies and the broader corporatocracy model. Varoufakis endorsed this edition directly: “Perkins has, once again, made a substantial contribution to a world that needs whistle-blowers.”

10. Varoufakis, Yanis. Adults in the Room: My Battle with Europe’s Deep Establishment. The Bodley Head, London, 2017. ISBN: 9780374101008. First-person account of the 2015 negotiations between the Greek Syriza government and the Troika (European Commission, IMF, ECB). Provides documented evidence of the structural constraints placed on elected governments operating within the Western financial architecture. Sunday Times Number One bestseller.

11. Sachs, Jeffrey D. Poland’s Jump to the Market Economy. MIT Press, 1993. Sachs’s own account of the shock therapy approach applied in Poland, and the model he believed should have been replicated in Russia with adequate Western financial support.

12. Sachs, Jeffrey D. “Russia — IMF Gives Too Little, Too Late.” Personal View, Financial Times, 31 March 1994. Sachs’s contemporaneous public criticism of the IMF’s failure to provide adequate stabilisation support for Russia, written during his time as adviser to the Yeltsin government.

13. Sachs, Jeffrey D. “What I Did in Russia.” Jeffrey D. Sachs personal website, 2012. Sachs’s retrospective clarification of his role as adviser to the Russian government 1991–1994, distinguishing his recommendations from the “loans for shares” privatisation scandal he explicitly opposed. Available at jeffsachs.org.

14. NPR Planet Money. “The Fall of Russia.” Episode transcript, May 2022. Includes Sachs’s own account of advising the Russian government, his lobbying of the IMF and US government for a ruble stabilisation fund and debt standstill, and the US government’s rejection of these measures.

15. Sachs, Jeffrey D. “A Front Row Seat to the Cold War That Never Ended.” Drop Site News, September 2024. Sachs’s most recent long-form reflection on Russia, Ukraine, and Western policy — arguing that the failure to support Russia’s democratic transition in the 1990s and NATO’s subsequent eastward expansion set the conditions for the 2022 war.

16. IMF. “The IMF and Russia in the 1990s.” By John Odling-Smee. International Monetary Fund Working Paper, 2004. Official IMF account of its engagement with Russia during the transition period, noting total IMF funding of $22.7 billion between 1992 and 1999.

17. Putin, Vladimir. “Russia at the Turn of the Millennium.” Published in Nezavisimaya Gazeta, December 1999. Putin’s first major policy statement as incoming President, explicitly addressing the economic damage of the 1990s and signalling a turn toward state-directed economic management.

18. BlackRock Financial Markets Advisory and Ministry of Economy of Ukraine. Memorandum of Understanding. Signed Washington DC, 10 November 2022. Published by BlackRock press release, 16 November 2022. Available at blackrock.com/corporate/newsroom.

19. Ministry of Economy of Ukraine. “The Government of Ukraine Announces its Plan to Launch a Ukraine Development Fund.” Press release, May 2023. Confirming BlackRock FMA’s formal appointment to advise on the Ukraine Development Fund design, on a pro bono basis.

20. CNBC. “Zelenskyy, BlackRock CEO Fink Agree to Coordinate Ukraine Investment.” December 28, 2022. Reporting on the December meeting between Zelenskyy and Fink formalising the reconstruction investment framework.

21. Kyiv Post. “BlackRock Backs Out of Ukraine Rebuild Fund Initiative.” July 2025. Reporting that BlackRock ended its advisory role in 2024 following reduced investor interest post-Trump election and continued uncertainty over Ukraine’s future.

22. Jacobin Magazine. “Ukraine’s Postwar Reconstruction Has Big Business Licking Its Lips.” January 2023. Critical analysis of the BlackRock-Ukraine MoU and the privatisation and deregulation framework being constructed for post-war reconstruction, including documentation of labour law changes advised by the British Foreign Office.

23. European Defence Spending & NATO :European Parliament Think Tank. ‘EU Member States’ Defence Budgets.’ EPRS_ATA(2025)772846. May 2025. Source for EU defence spending figures 2021–2024, country-by-country breakdowns, and the ReArm Europe / SAFE mechanism details.

24. Atlantic Council. ‘Who’s at 2 Percent? Look How NATO Allies Have Increased Their Defense Spending Since Russia’s Invasion of Ukraine.’ July 2024. Source for the 18% spending surge figure and Poland’s 4.12% GDP allocation.

25. NATO. ‘Funding NATO.’ NATO Topic Page, updated 2025. Source for the Wales Summit 2014 Defence Investment Pledge and the Vilnius 2023 commitments.

26. CSIS. ‘Chapter 13: Defense Budgets in an Uncertain Security Environment.’ Defence Programs Report, 2025. Source for the ReArm Europe Plan detail and the €800 billion potential funding figure.

27. Intereconomics. ‘The Fog of War: Russia’s War on Ukraine, European Defence Spending and Military Capabilities.’ Vol. 57, No. 3, 2022. On Germany’s Zeitenwende and the broader paradigm shift in European defence thinking.

28. Council on Foreign Relations. ‘As NATO Countries Reach Spending Milestone, Is 2 Percent Enough?’ Expert Brief, 2024. On the $828 billion cumulative underspend by European NATO members since 2014.

29. Britain, Brexit & Ukraine: House of Commons Library. ‘Military Assistance to Ukraine (February 2022 to January 2025).’ Research Briefing CBP-9477, updated February 2025. Source for UK commitments, Operation Interflex figures (51,000+ personnel trained), and the EU’s European Peace Facility (€11.1 billion committed).

30. Medieval European Fragmentation: Blaydes, Lisa, and Christopher Paik. ‘The Impact of Holy Land Crusades on State Formation.’ American Political Science Review, 2016. On the feudal fragmentation of medieval Europe and the role of the Church in sustaining it.

31. Tilly, Charles. Coercion, Capital and European States. Blackwell, 1990. Background on warfare and state formation in Europe.

Read my DISCLAIMER

If you are new to my Substack Read This First

Tickers Referenced:

Fear, Uncertainty, and Doubt: The AI Sell-Off Cycle

On the 20th of January 2025, a seismic tremor ran through the technology markets of the world. DeepSeek, a largely unknown Chinese artificial intelligence laboratory headquartered in Hangzhou, quietly released its R1 reasoning model into the public domain. Within hours, it had topped the Apple App Store download charts, displacing ChatGPT. Nvidia lost nearly 17%, roughly $600 billion in a single session, the largest single-day loss in market history for any company. The Nasdaq tumbled 3%. The Philadelphia Semiconductor Index collapsed almost 8%. It was, a rout.

The panic was fuelled by a simple proposition: if a Chinese startup could train a frontier-grade AI model for less than six million dollars using older, export-restricted Nvidia chips, then what justified the hundreds of billions being poured into data centres, GPU clusters, and proprietary large language models by America’s technology giants? The logic of the AI supercycle, the grand narrative that had propelled them to stratospheric valuations over the previous two years, suddenly seemed vulnerable. Were the hyperscalers building cathedrals to a god that might not come? (None of the gods ever come by the way, but that’s besides the point).

The investment thesis for much of the AI boom had rested on the assumption of scarcity: scarce compute, scarce talent, scarce frontier models. DeepSeek punctured all three in one afternoon. Its model was efficient. It was open-source. It was free. And it had been built, apparently, with a fraction of the resources that OpenAI, Google DeepMind, or Anthropic had consumed. Marc Andreessen called it a Sputnik moment. He was not entirely wrong, though whether it heralded American decline or merely a recalibration remained deeply contested.

The sell-off was, in retrospect, characteristic of the kind of FUD (Fear, Uncertainty, and Doubt) that has periodically gripped technology markets throughout history. The internet bubble collapse of 2000, the cloud computing scepticism of the early 2010s, the crypto crashes of 2018 and 2022 (we are in one now too): each saw a genuine technological advancement become the subject of irrational exuberance followed by equally irrational despair. Bank of America’s senior analyst Vivek Arya captured the essential absurdity of the DeepSeek-era panics in a note that called the indiscriminate sell-offs ‘internally inconsistent.’ The paradox, as BofA articulated it, is a logical impossibility: AI cannot simultaneously be so expensive as to produce no return on investment and so powerful as to devastate every established software business model. Both outcomes cannot occur at once.

Yet the fear has proven remarkably persistent. Through late 2025 and into 2026, fresh waves of anxiety have buffeted the Nasdaq. Speculation about DeepSeek V4, whispers about a second ‘DeepSeek moment,’ concerns about Palantir’s AI revenue trajectories, and broader unease about whether the hundreds of billions committed to data centre infrastructure by Amazon, Microsoft, Meta, and Alphabet could ever generate commensurate returns. All these have kept investors in a state of nervy vigilance. The SaaS sector suffered what Jason Lemkin, its self-styled godfather, termed a ‘crash.’ Semiconductor stocks lurched. The AI arms race, once a guaranteed source of upside, became a potential liability.

One company however, curiously, seemed largely immune. Not because it had solved AI. Not because it had conquered the frontier model challenge. But because it had, largely, opted out of the arms race altogether.

The Apple Exception: Decoupling from the Nasdaq

The Bloomberg headline of the 25th of February 2026 said it plainly: Apple had become an ‘accidental haven’ from AI fears, for now. As the Nasdaq fretted and the hyperscalers braced for scrutiny, Apple’s stock demonstrated a striking decoupling from the index’s AI-driven volatility. The company had, somehow, managed to be a technology giant without being a technology liability.

This is a remarkable thing to say about the world’s most valuable company — a company that, not so long ago, was being lampooned for the inadequacy of its AI ambitions. Siri, the virtual assistant that Apple introduced in 2011 and has been improving at a distinctly measured pace ever since, had become something of a cultural punchline. While ChatGPT rewrote the rules of human-computer interaction, while Google’s Gemini impressed with its multimodal capabilities, while Microsoft embedded Copilot across its entire enterprise ecosystem — Apple seemed to be hosting a rather more sedate party. Its Apple Intelligence suite, unveiled at WWDC in June 2024, received respectful notices but few raves. The revamped Siri — the one that would supposedly change everything — kept getting delayed, pushed from late 2025 into early 2026.

And yet. The iPhone 17 cycle produced what Tim Cook called demand that was ‘off the chart.’ The holiday quarter for the three months ending December 2025 delivered $144 billion in revenue, a 16% year-on-year increase and a figure that stunned Wall Street.

Had Apple won the AI race or had it redefined what winning might look like?

Was this immunity from the AI sell-off the product of strategic genius or fortunate accident?

Was Apple’s relative calm during the storms that battered Nvidia, Microsoft, and the broader semiconductor sector the consequence of deliberate positioning?

OR

Was it simply the serendipitous result of being late to a party that turned out to be in an unstable building?

Apple’s AI Investments: Quiet Accumulation Spanning the Spectrum of Machine Learning Capability

The narrative that the company has done nothing in artificial intelligence is wrong, though Apple has, characteristically, done things differently.

Apple has been acquiring AI companies for years, almost always quietly and almost always for technology or talent rather than for market positioning — in 2023 alone it acquired 32 AI startups. The acquisitions span the spectrum of machine learning capability:

DarwinAI, a Canadian startup focused on AI model efficiency, was absorbed in early 2024.

Pointable AI, an enterprise knowledge-retrieval startup that links data feeds to large language model workflows, was snapped up in January 2026.

Kuzu, a graph database optimised for query speed and scalability, was acquired in October 2025.

Q.ai, an Israeli startup which specialises in technology that reads facial movements and understands silent communication in 2026.

The company reportedly explored purchasing Perplexity AI before walking away over concerns about its web-scraping practices, which sat uneasily alongside Apple’s privacy principles.

Apple’s R&D spending reached $33.4 billion, an 8% increase year-on-year. This isn’t the behaviour of a company that’s choosing to ignore the AI era, but one that’s choosing to participate on its own terms.

Those terms are defined by the Private Cloud Compute architecture — Apple’s proprietary system for handling complex AI queries that cannot be processed on-device. Rather than routing sensitive user data to third-party cloud services, Apple built its own infrastructure, powered by M-series chips and purpose-built servers. A new 250,000-square-foot server manufacturing facility opened in Houston in 2026 to support this infrastructure, part of Apple’s broader pledge to invest more than $500 billion in the United States over four years — a commitment that encompasses AI infrastructure, silicon engineering, and advanced manufacturing.

The silicon strategy is perhaps the most under appreciated element of Apple’s AI positioning. The M5 chip series, launched in October 2025, includes neural accelerators specifically optimised for on-device machine learning workloads. This matters: Apple’s AI runs, wherever possible, on the device itself — your iPhone, your MacBook, your iPad. No data leaves the ecosystem. No query is processed in a data centre unless necessary. The privacy architecture is a genuine technical differentiator.

The partnership strategy completes the picture. Apple partnered with OpenAI to integrate GPT-5 into iOS 26. It has reportedly struck a deal with Google to test a custom Gemini model as the intelligence layer behind the revamped Siri. It selected Alibaba’s AI for its Chinese market rollout, a pragmatic accommodation of Beijing’s data sovereignty requirements. It has deepened its collaboration with Anthropic, integrating Claude into Xcode as a developer assistant. And it has expanded its partnership with GlobalFoundries to secure domestic semiconductor capacity.

Apple’s financial chief Kevan Parekh summarised the philosophy

‘I don’t see us moving away from this hybrid model, where we leverage both first-party capacity as well as leverage third-party capacity.’

Translation: Apple will rent what it needs and build only what it must own.

Why Apple Has Not Taken the Hyperscaler Route

The contrast between Apple and the hyperscalers — Amazon, Microsoft, Alphabet, and Meta — now sounds deliberate. Amazon plans $200 billion in capital expenditure for 2026. Alphabet has committed between $175 and $185 billion. Meta between $115 and $135 billion. Microsoft around $145 billion. Together, the four are spending approaching $700 billion in a single year on AI infrastructure. Apple’s capital expenditure for fiscal 2025 was $12.7 billion — less than a tenth of Alphabet’s projected spend, and barely a rounding error in the context of the hyperscaler arms race.

Is this restraint a consequence of poverty? I don’t think so. Apple sits on over $130 billion in cash and marketable securities. It returned $106 billion to shareholders in fiscal 2025 alone. It has the financial firepower to compete in the data centre arms race if it chose to. It has simply decided not to.

The reasons are multiple and mutually reinforcing.

First, and most fundamentally, Apple’s business model does not require it. The hyperscalers are infrastructure businesses: they need vast compute because their entire commercial proposition is the provision of computing services to others. Amazon sells cloud capacity. Microsoft sells enterprise software and cloud services. Google sells advertising, indexed by AI-powered search. All three need to be at the frontier of AI capability because their revenues depend directly on it.

Apple’s revenues depends on selling hardware — iPhones, Macs, iPads, Apple Watches, AirPods — and the services that run atop that hardware. Its AI investment is justified by the need to make its devices sufficiently compelling that consumers upgrade every two or three years rather than every four or five. Apple Intelligence, in this framing, is not a product in itself. It is a reason to buy the latest iPhone.

Second, Apple has made arguably a very shrewd bet that foundational AI models will become commoditised. The key observation is borrowed from cloud computing history: owning infrastructure during a technological arms race rarely produces the highest long-term returns. The companies that built mainframes did not dominate the PC era. The companies that built the physical infrastructure of the early internet — fibre optic cables, server farms — did not capture most of the internet’s commercial value. That value accrued to the companies that built on top of the infrastructure: Google, Amazon, Facebook. If AI models become, as Apple appears to believe they will, interchangeable commodities — if the gap between GPT-5, Gemini, Claude, and whatever emerges from DeepSeek narrows to near-irrelevance — then owning the infrastructure offers little competitive advantage. The moat lies elsewhere.

For Apple, the moat lies in three places: the hardware, the ecosystem, and the privacy architecture.

The hardware because Apple’s A-series and M-series chips provides performance and efficiency that no third-party device can match for it’s specific workloads.

The ecosystem because a billion-plus active devices represents a distribution network of unparalleled reach: Apple can deploy AI features to hundreds of millions of users through a software update, without the capital expenditure of building data centres.

And

Privacy because, in a world increasingly suspicious of AI and the companies that build it, the ability to offer genuinely on-device AI where your data never leaves your phone is a genuine differentiator.

Third and this is where strategy morphs into something more like luck — Apple’s restraint has spared it from the valuation dynamics that made other technology companies so vulnerable to the AI FUD cycle. A company that has spent $145 billion on AI data centres, as Microsoft has, is deeply exposed to any narrative that questions whether those data centres will generate returns. A company that spent $12.7 billion and outsourced its model development to OpenAI, Google, and Anthropic has very little skin in that particular game. When DeepSeek panic hits the Nasdaq, the question investors ask is: who has overinvested? Apple’s answer: not us.

So should Apple become an AI Company or remain a Hardware Company?

Very interesting question: the answer is that it is both and that the distinction matters less than it once did.

The binary between ‘AI company’ and ‘hardware company’ was in my mind always somewhat artificial, but it feels particularly irrelevant in Apple’s case. Apple is, at its heart, a systems integration company. It does not manufacture components. It designs chips and has them fabbed by TSMC. It writes operating systems. It creates the user experience that unifies hardware and software into a seamless whole. The magic of Apple’s products has always been the integration — the way the hardware and software talk to each other in ways that Android or Windows ecosystems cannot quite replicate.

AI is, in this sense, simply the latest layer of integration. Apple Intelligence is an operating system feature — as fundamental, in ambition, as the introduction of multi-touch in 2007 or the App Store in 2008. The revamped Siri, whenever it fully arrives, will not be Apple competing with ChatGPT for enterprise AI contracts. It will be Apple making the iPhone more indispensable to its existing users and more attractive to those who haven’t yet joined the ecosystem.

The iPhone 17’s unprecedented commercial success (16% revenue jump in the holiday quarter) suggests this strategy is working. Consumers are upgrading specifically because of Apple Intelligence. The on-device AI features — improved auto-complete, smarter photo editing, more capable summarisation — are apparently sufficient to motivate purchase decisions, even without a frontier-grade Siri. If the revamped Siri delivers on its promise, the next upgrade cycle could be even more dramatic.

So should Apple become an AI company? Not in the way that OpenAI or Anthropic or even Google DeepMind is. It is not building frontier models. It is not publishing research papers that advance the science. It is not attempting to create artificial general intelligence. It is, instead, using AI — other people’s AI, primarily — to make its hardware more compelling. That is a very different ambition, but it is not a lesser one.

The comparison that feels most apt is not with the AI labs, but with the early smartphone era. When Apple launched the iPhone, it did not build the mobile networks that carried its data. It did not manufacture the memory chips inside its devices. It did not write the web standards that its browser relied upon. It integrated, curated, and delivered. The result was the most successful consumer electronics product in history. Apple appears to be attempting the same trick with AI: integrate the best available models, curate the experience, deliver it through hardware that no one else can match, and capture the value at the point of sale.

Does It Matter That Apple Is Not a Hyperscaler or LLM Builder?

The short answer is: probably not, at least not in the way the conventional narrative suggests.

The conventional narrative says that AI is a winner-takes-all race, that frontier model capability determines market outcomes, and that the companies which build the biggest models with the most compute will dominate the AI economy. Under this narrative, Apple’s decision to outsource its model development to OpenAI and Google is a strategic failure — a surrender of the crucial battleground.

There are serious reasons to doubt this narrative. The AI landscape of 2026 looks very different from the AI landscape imagined by the bull case of 2023. Models have proliferated. Open-source alternatives from Meta, Mistral, and DeepSeek have demonstrated that frontier-grade capability is no longer the exclusive preserve of those with billions to spend. The gap between the leading closed models and the best open-source alternatives has narrowed substantially. Inference costs have fallen dramatically. The commodity thesis — that AI models will become interchangeable utilities, much as cloud storage became a utility — appears increasingly plausible.

If that thesis is correct, then Apple’s positioning looks not merely adequate but genuinely clever. The company that owns the interface — the hardware through which a billion people experience AI — is likely to capture more value than the company that trains the underlying model. The utility analogy is instructive: electricity companies are essential, but the valuable companies are those that build the appliances that run on electricity.

Apple is building the appliances. And it has $130 billion in reserve to make acquisitions if the landscape shifts in ways that require it to own more of the stack. The exploration of Mistral AI — a French open-source AI developer valued at around $14 billion — suggests that Apple is watching the market carefully and is prepared to act when valuations become attractive. A DeepSeek-style panic, which deflates AI startup valuations across the board, could actually be in Apple’s interest: it creates acquisition opportunities at prices the company can easily afford.

What Apple cannot afford or rather, what its brand cannot afford is to be left entirely behind in the consumer AI experience. The delayed Siri revamp has been damaging to the company’s reputation for innovation. If the new Siri, powered by Gemini or Claude or some combination of models, fails to deliver a materially better experience than asking Google Assistant or opening ChatGPT, then the AI premium embedded in iPhone pricing may come under pressure. The consumer risk is real, even if the financial risk from hyperscaler-style overinvestment is absent.

The deeper question may be philosophical rather than commercial.

Apple has always derived its identity from the belief that the best products are those that integrate hardware and software into experiences of exceptional elegance. It has historically been suspicious of open platforms, of fragmentation, of the kind of messy, competitive market dynamics that define the Android ecosystem. Its approach to AI reflects this same instinct: bring AI inside the ecosystem, control the experience, protect the user’s privacy, and charge a premium for the privilege.

Whether this instinct is sufficient to win in the AI era remains to be seen. But the evidence of early 2026 suggests that it is, at minimum, not losing and that not losing, when your competitors are spending $700 billion in a single year on an uncertain bet, may be the most profitable position of all.

By Design or by Serendipity? A Verdict

The honest answer is: both, but more by design than the sceptics would have you believe.

Tim Cook and his team made a deliberate decision not to compete for frontier model primacy. They chose, instead, to pursue a hybrid architecture — on-device AI powered by proprietary silicon, supplemented by selective partnerships with the leading AI labs — that preserved capital, protected the privacy brand, and kept the company’s options open. They watched as their rivals committed hundreds of billions to data centres and then, when the FUD cycle arrived, found themselves sitting on an enviable combination: significant AI capability, minimal AI-related financial exposure, and over $130 billion in cash.

The serendipity lies in the timing. Apple’s Siri delays, much mocked in 2024 and 2025, inadvertently positioned the company as the one tech giant that had not over-committed to any particular AI vision. The very inadequacy that drew criticism also drew a kind of immunity: you cannot be accused of over-investing in something you have visibly under-invested in. When the AI sell-off arrived, Apple had little to sell off.

There is also something structural about Apple’s relationship with the markets that insulates it from AI panic cycles in ways that pure-play tech companies cannot enjoy. Apple is, at some level, a consumer goods company. Its fortunes depend on whether people want to buy the next iPhone — a question that has, for the past seventeen years, been answered with a reliable yes. That reliability provides a bedrock of investor confidence that even the most fevered AI panic cannot entirely erode. Warren Buffett, whose Berkshire Hathaway holds Apple as its largest single investment, bought it because he understood the power of a brand that generates loyal, repeat customers who upgrade every few years regardless of the macroeconomic environment.

The Bloomberg characterisation of Apple as an ‘accidental haven’ from AI fears is seductive, but it is also slightly unfair. It was not entirely accidental. The caution was deliberate. The privacy architecture was deliberate. The silicon strategy was deliberate. What was accidental or at least unforeseen was that the very characteristics which drew criticism from AI enthusiasts (the slow Siri, the partnership model, the absence of a frontier lab) would turn out to be precisely what made Apple look prudent when the winds changed.

Apple has not won the AI race. It may never win the AI race, in the sense that OpenAI or Google DeepMind might understand winning. But it has demonstrated something rather more valuable: that you do not need to run a race to benefit from the terrain it covers. The company has positioned itself as the delivery mechanism for a technology it did not primarily build, distributed through hardware it uniquely controls, to a customer base of extraordinary loyalty. In a market periodically convulsed by fear about who is spending too much on AI, Apple’s answer — that it is spending just enough, on the right things, in the right places — looks increasingly like wisdom rather than weakness.

The sanctuary may have been accidental. But the architecture that created it was not.

The headlines are everywhere. America is pulling up the drawbridge. Europe is rebuilding its factories. China is hoarding rare earth metals. And all around the world, politicians are giving speeches about bringing supply chains back home. It sounds like the end of an era. It sounds, frankly, like the end of globalisation.

But here is the thing. They are wrong. Or, at the very least, they are asking the wrong question. The real question is not whether globalisation is dying. The real question is what kind of globalisation is being born. To understand the answer, you have to go back — a really long way back — to the 1600s when a group of ambitious merchants, a handful of wooden ships, and the smell of black pepper.

Globalisation 1.0: The Age of Extraction

Once upon a time there were no free-trade agreements or the internet. To me globalisation begins in the sixteenth century, with a simple and ruthless idea: go somewhere far away, take what they have, bring it home, turn it into something more valuable, and sell it back to them.

Europe’s great trading companies — the Dutch East India Company (founded in 1602), the British East India Company (founded in 1600), and their French and Portuguese counterparts — were the original architects of globalisation. They were not governments, exactly. But they were not ordinary businesses either. They had armies. They had navies. They had the legal power to sign treaties and wage war. In short, they were empires that wore the clothes of commerce.

What they built was a global system of extraction. Spices from the Spice Islands (aka Indonesia). Cotton and indigo from India. Tea from China. Sugar from the Caribbean. Silk from Persia. None of these goods were processed where they were found. They were shipped raw to Europe, where they were refined, manufactured, woven, or blended, and then sold back to the very people who had grown or mined them in the first place — at enormous profit.

The East India Companies built the first architecture of global supply chains — extraction at the periphery, value creation at the centre.

Britain won this game more decisively than any other nation. The British East India Company eventually became the de facto ruler of India, controlling territory larger than Europe. Its success was logistical as well as military. The British understood, earlier than anyone else, that the real power in globalisation lay in controlling the journey from raw material to finished product not owning the raw material itself. Whoever controlled that journey controlled the wealth.

By the nineteenth century, this model had formalised into something called industrial capitalism. Britain was the workshop of the world. Raw cotton arrived from India and America; finished textiles left for markets everywhere. Iron ore arrived from Wales and the Midlands; steam engines and railway tracks left for every corner of the empire. The pattern was consistent and deliberate: extract, manufacture, sell. The periphery supplied the ingredients. The centre added the value. The periphery paid the bill.

This was Globalisation 1.0. It was brutal, often violent, and structurally designed to keep power concentrated in a small number of European capitals. But it was also, unmistakably, a global system. The world was deeply interconnected in spite of its inequality, and one can argue because of it.

Globalisation 2.0: The Great Inversion

For most of the twentieth century, the old colonial model creaked along in various forms. But after the Second World War, something began to shift. Decolonisation swept through Asia and Africa. New nations claimed sovereignty over their own resources. And the West, rather than fighting every battle to keep control, began to realise that there was a smarter way to stay on top.

The new model was investment. Instead of sending soldiers and ships to take raw materials, Western governments and corporations sent capital. They built factories in Mexico, Korea, Taiwan, and eventually China. They moved manufacturing to wherever labour was cheapest. And they kept the high-value work — design, branding, finance, intellectual property — firmly at home.

This was Globalisation 2.0. And its defining feature was not the movement of raw materials, as in the colonial era. It was the movement of labour. The engine was cheap hands, not scarce earth.

Globalisation 2.0 was powered by an arbitrage — the gap between what a worker in Detroit or Dortmund was paid, and what a worker in Shenzhen or Dhaka would accept.

China was the great beneficiary of this shift. After Deng Xiaoping opened the Chinese economy in the late 1970s, Western multinationals flooded in. By the 1990s and 2000s, China had become the world’s factory. It made the trainers, the televisions, the smartphones, the furniture, the steel, the solar panels. Western brands supplied the designs and the marketing. China supplied the manufacturing. And the goods came back to Western consumers at prices that would have seemed impossible a generation before.

India found its own niche in this system. Rather than manufacturing goods, India manufactured services. The liberalisation of the Indian economy in 1991 unleashed an enormous wave of English-speaking, technically educated graduates into the global labour market. Western companies discovered that software could be written, customer calls could be answered, and back-office accounts could be processed in Bangalore or Hyderabad for a fraction of what it cost in London or New York. India became the back office of the world.

For roughly thirty years, this system worked spectacularly well — at least by the metrics that economists preferred to measure. Global trade volumes grew faster than GDP. Hundreds of millions of people were lifted out of poverty in China, India, and across Southeast Asia. Consumer prices in the West fell dramatically. Inflation seemed tamed, perhaps permanently. Economists coined the phrase ‘hyperglobalisation’ to describe the extraordinary pace at which the world’s economies had become intertwined.

Thomas Friedman’s “The World Is Flat” explained that digital technology and globalisation have leveled the economic playing field, allowing individuals and companies from anywhere to compete and collaborate as if geography didn’t matter, creating new opportunities but also intense competition, with key “flatteners” being the internet, workflow software, and outsourcing. It’s about the rapid removal of barriers, enabling a single, interconnected global marketplace where talent and services can flow freely, making the world flatter and fairer for many.

But cracks were forming. The model had assumed that labour arbitrage would last forever. It had assumed that supply chains could stretch to the ends of the earth without breaking. And it had assumed — above all — that the political and economic interests of the world’s major powers would remain broadly aligned.

None of those assumptions would survive the 2020s intact.

The Crack in the System

The financial crisis of 2008 was the first major warning. Trade growth, which had outpaced GDP for decades, stalled. Economists began using a new word: ‘slowbalisation.’ The era of explosive integration, they suggested, might be over.

Then came Donald Trump’s trade war with China in 2018, which replaced the polite fiction of mutual benefit with explicit economic nationalism. Then came COVID-19 in 2020, which broke supply chains. Hospitals in Britain could not get masks. Car factories in Germany sat idle for want of a single microchip made in Taiwan. Supermarket shelves in the United States ran bare.

The world discovered, all at once, the hidden vulnerability in Globalisation 2.0. The system had been optimised ruthlessly for efficiency. It had never been designed for resilience. When something went wrong at one node — a factory in Wuhan, a cargo ship blocking the Suez Canal, a drought in Taiwan affecting a semiconductor plant — the entire network shuddered.

COVID revealed how fragile the system always was.

The political reaction was swift and, in some quarters, shrill. Politicians across the spectrum reached for the same diagnosis: we have become too dependent on other countries. We must bring production home. ‘Reshoring,’ ‘friend-shoring,’ ‘nearshoring’ — the vocabulary multiplied as fast as the trade barriers. The narrative of deglobalisation took hold in newspapers, think tanks, and cabinet meetings alike.

But the narrative, for all its political convenience, missed something important. The problem was never that the world was too interconnected. The problem was that the interconnections had been built on a single fragile premise — cheap labour in a small number of countries. When that premise was disrupted, the whole structure wobbled.

What was actually needed was not less interdependence, but better interdependence. And the AI revolution, arriving precisely at this moment of anxiety, was about to make a different kind of interdependence not just possible, but unavoidable.

Globalisation 3.0: The Age of Scarcity

Here is the central fact that changes everything: building artificial intelligence at scale is a materials problem and not a software problem.

Every AI chip needs gallium. Every data centre needs copper. Every electric motor in a robot or a drone needs neodymium — a rare earth element whose name most people have never heard. Every smart device uses silver, not as jewellery or currency, but as the conductor that makes its circuits function. These are not optional ingredients. They cannot be replaced by writing better code. They must be dug out of the ground.

And here is the crucial detail: no single country has all of them. Not the United States. Not China. Not the European Union. The materials needed to build the AI-enabled world of the next thirty years are distributed across the planet in a pattern that is geologically indifferent to political borders.

China dominates the processing of rare earth elements, accounting for over 90 per cent of the downstream value chain — the separation, refining, and magnet production that turns raw ore into usable components. The Democratic Republic of Congo holds roughly 70 per cent of the world’s cobalt reserves. Chile and Argentina sit atop vast deposits of lithium. Brazil, Indonesia, and the Philippines are among the world’s largest producers of nickel. The deep sea off the coasts of Africa and the Pacific may hold more manganese and copper than the entire known land-based supply.

The AI revolution has made geography more important than at any point since the age of empires.

This is the structural foundation of Globalisation 3.0. The axis of interdependence has shifted. In the colonial era, the engine was the trade of raw commodities for finished manufactured goods. In the post-war era, the engine was labour arbitrage — the movement of manufacturing to wherever wages were lowest. Now, the engine is mineral scarcity. The world must cooperate — not because of any altruistic virtuous motive, but because no nation possesses everything it needs to build the future alone.

Consider silver. It is the best natural conductor of electricity on Earth, and it is present in virtually every intelligent device ever made. Every solar panel. Every electric vehicle. Every smartphone. Every server in every data centre. As AI expands from software into physical machines — robots, drones, autonomous vehicles, humanoids — the demand for silver does not grow incrementally. It explodes. And silver is not like oil. You cannot frack it into abundance. The world’s mines produce roughly 800 million ounces per year. Demand is already beginning to outstrip supply. That gap will widen dramatically over the next decade.

The same logic applies to copper, to lithium, to the rare earths. These are not temporary bottlenecks that clever engineering will soon resolve. They are structural constraints built into the physics of the technology itself. The world needs these materials. The materials are in specific places. To get them, nations must deal with each other. Globalisation, far from ending, is being reordered around an entirely new set of physical necessities.

The New Map of Interdependence

If Globalisation 3.0 is driven by mineral scarcity, then its geography looks very different from its predecessors. The old centres of power — London, New York, Tokyo — built their dominance on controlling finance and manufacturing. The new centres of power will be defined, in large part, by who controls the ground beneath their feet.

This reordering is already visible in the geopolitical manoeuvring of the major powers. The United States has launched the Mineral Security Partnership with fourteen nations and the European Union, explicitly designed to diversify critical mineral supply chains away from Chinese dominance. The Trump administration pursued bilateral resource deals, including highly publicised overtures towards Greenland — an island whose geological profile makes it one of the most important untapped mineral repositories on the planet. A peace agreement between Rwanda and the Democratic Republic of Congo, brokered in part by American diplomatic pressure in 2025, was partly motivated by US interest in securing access to the DRC’s vast cobalt and copper reserves.

China, for its part, has not been passive. In October 2025, Beijing expanded its rare earth export controls to their most sweeping extent yet, effectively turning critical minerals into a geopolitical weapon. Gallium, germanium, antimony, and five additional rare earth elements were added to the restricted list. Shipments of rare earth magnets to the rest of the world fell by 74 per cent in a single month. The message was unmistakable: China intends to use its mineral dominance as leverage in the broader contest for AI supremacy.

The moves on Greenland, Venezuela, and the DRC are about mineral deposits more than energy or ideology. The new scramble for resources is the old scramble for resources — just with a different list of materials.

But it would be a mistake to reduce Globalisation 3.0 to a simple US-China confrontation. The emerging map is far more complex. It is not deglobalisation. It is, as some scholars have begun to call it, ‘bloc-based globalisation’ — deeper integration within aligned groups, and strategic decoupling between rivals.

Europe offers an interesting case study. For much of the AI debate, Europe has appeared to be falling behind. It has no hyperscaler technology companies to rival Google, Amazon, Microsoft, or Alibaba. Its data protection laws have slowed the rollout of AI applications. Its labour regulations make rapid restructuring difficult. By the logic of the software era, Europe is a loser.

By the logic of the materials era, Europe may be a winner. Germany’s heavy industrial base, the same factories that look obsolete in a software world is precisely what the physical AI buildout requires. Motors, transformers, actuators, precision manufacturing equipment: these are the things that go into the robots and data centres of tomorrow. Europe is also betting heavily on renewable energy. Solar and wind require enormous quantities of copper, silver, and rare earth magnets. By accelerating the green energy transition, Europe is simultaneously building the infrastructure that AI-enabled industrial systems will eventually run on.

India presents a different kind of pivot. The narrative of doom around India’s IT sector — that vibe coding and AI will eliminate software jobs — is not entirely wrong in the short term. India’s NITI Aayog has warned that tech service headcount could fall from eight million workers to six million by 2031. That is real disruption for real people.

But the longer arc points in a different direction. As the physical AI buildout accelerates, the world will need millions of people who can install, maintain, repair, and operate robots and AI-enabled machinery. These are not desk jobs that can be eliminated by a language model. A robot that breaks down in a factory in Chennai needs a human engineer to fix it. An autonomous vehicle that loses its GPS signal in a tunnel needs a remote human operator to guide it safely. Waymo, the self-driving car company, already employs teams of remote operators — reportedly including staff in the Philippines — who wear virtual reality headsets and take over vehicles when the AI reaches the limits of its competence. This is not the end of global labour. It is labour globalised into an entirely new shape.

Brazil deserves special attention in this new geography. Sitting atop vast reserves of iron ore, nickel, lithium, copper, and rare earth elements, and with abundant solar and wind energy resources, Brazil is structurally positioned as one of the most important nations in the Globalisation 3.0 era. It has historically been undervalued by markets that were focused on software and manufacturing. As the axis shifts to materials, that undervaluation is unlikely to persist.

The Thesis Unpacked: Same Game, New Rules

Step back and consider the full arc of the argument. Five hundred years ago, the British East India Company extracted raw materials from the periphery, processed them at home, and sold finished goods back to the world. The raw materials were spices, cotton, and tea. The finished goods were textiles, ceramics, and trade goods. The power resided in controlling the journey between the two.

The AI era is structurally similar but with a crucial difference. The raw materials are now silver, copper, lithium, and rare earths. The ‘finished good’ is not a textile or a machine. It is intelligence itself embedded in chips, robots, autonomous vehicles, data centres, and the vast digital infrastructure that will run the economy of the twenty-first century. And the power, once again, resides in controlling the journey from raw material to finished product.

The commodity has changed. The structure of power has not. Whoever controls the journey from mine to machine will write the rules of the new era.

This is why the proclamations of deglobalisation are so misleading. They identify real trends — rising tariffs, reshoring of manufacturing, strategic decoupling between the US and China — and mistake them for the whole picture. Yes, some specific supply chains are becoming more local. Semiconductor manufacturing, in particular, is being deliberately duplicated across multiple geographies for reasons of national security. But the system as a whole is not becoming less global. It is becoming global in a different way.

Consider the scale of what is being built. Jensen Huang, the chief executive of NVIDIA, has estimated the total cost of embedding intelligence into every machine, factory, vehicle, and device on earth at roughly 85 trillion US dollars over the next fifteen years. That is not a national project. It is not even a continental project. It is a planetary project, and it requires planetary coordination. The materials are in the ground in dozens of countries. The processing happens in a handful of others. The chip design happens in America. The chip fabrication happens in Taiwan. The servers are assembled in China and Malaysia. The data centres are being built on every inhabited continent. No nation, however powerful, can do this alone.

There is one important counter-argument that must be honestly acknowledged. A 2025 report from the United Nations Development Programme, building on analysis from Brookings, warned that AI could accelerate a ‘Great Divergence’ — a growing gap between the countries that can access AI infrastructure and those that cannot. Electricity. Computing power. High-speed connectivity. These are the prerequisites for participating in the AI economy. Much of the Global South currently lacks them. If the benefits of AI concentrate in wealthy nations and a handful of resource-rich middle-income countries, the result could be a kind of hyper-globalisation for some, and renewed marginalisation for others.

This is a genuine risk, and it complicates the optimistic version of the hyperglobalisation thesis. The new interdependence is not seamless or equitable. It is structured around strategic blocs, shaped by geopolitical rivalries, and capable of excluding as well as including. The world of Globalisation 3.0 will not look like the frictionless borderless market that idealists once dreamed of. It will look, in some ways, more like the world of the East India Companies: intensely global in its reach, deeply unequal in its rewards, and governed by the strategic interests of the powerful.

The difference and it matters is that the axis of that power has shifted. In the colonial era, power came from controlling trade routes and manufacturing capacity. In the labour arbitrage era, power came from controlling access to capital and consumer markets. In the materials era, power increasingly flows from controlling access to the minerals that intelligence requires. Countries that were peripheral in earlier eras — the DRC, Bolivia, Chile, Greenland, Brazil — now sit atop strategic assets that the most powerful nations on earth urgently need.

That is a profound change in the underlying structure of global power. It does not guarantee justice or equality. But it does mean that the geography of relevance has changed. The old periphery is becoming, in at least some respects, strategically central.

A Warning and an Opportunity

The narrative of deglobalisation is, at its core, a story about the end of Globalisation 2.0. And that story is largely true. The model built on cheap labour, frictionless supply chains, and the assumption of shared economic interest between the US and China is over. COVID cracked it. Trade wars widened the crack. The AI revolution is building something new in its place.

But ‘building something new’ is not the same as ‘ending globalisation.’ The world of the next three decades will be more interdependent than ever — not less. It will simply be interdependent in a different way. The ties that bind nations together will no longer be primarily the movement of manufactured goods from low-wage factories. They will be the movement of critical minerals from mine to data centre, of energy from solar panel to server, of robots from assembly plant to shopfloor, of AI-enabled services from wherever the talent happens to be to wherever the problem happens to exist.

Globalisation is moulting not dying. The old skin built on cheap labour and frictionless trade is being shed. What emerges will be harder, stranger, and built around physical scarcity in ways that will surprise almost everyone.

The East India Company model has returned. The commodity is different. The ships are different. The weapons are chips and export controls rather than cannon and treaty. But the underlying logic — extract, process, control the journey, capture the value — is eerily familiar to anyone who has read their history.

The countries and companies that recognise this earliest will write the rules of the new era. Those that remain fixated on the deglobalisation narrative — retreating inward, protecting old industries, treating every foreign dependency as a vulnerability rather than a negotiating asset — risk becoming the colonies of the next chapter. Not by force of arms, as in the sixteenth century. But by being cut out of the supply chains, the mineral deals, and the technological alliances that will determine who prospers and who does not.

The spices have changed. The game has not.

Key Sources and Further Reading

• Foreign Policy Analytics: ‘Artificial Intelligence and the Critical Minerals Crunch’ (2025).

• Council on Foreign Relations: ‘It’s Not Deglobalisation, It’s Regionalisation’ (2025).

• UNDP / Brookings: ‘The Next Great Divergence’ (December 2025).

• SFA Oxford: ‘China’s Rare Earth Export Controls and Their Impact on Global Supply Chains’ (2025).

• World Economic Forum Future of Jobs Report 2025.

• NITI Aayog: ‘Roadmap for Job Creation in the AI Economy’ (October 2025).

• GQG Partners: ‘Critical Dependence on Rare Earth Minerals’ (2025).

• Japan Times: ‘From Hyperglobalisation to Thin Globalism’ (2024).

• Raoul Pal / Jordi Visser, Global Macro Investor Think Piece (January 2025).

Read my DISCLAIMER

If you are new to my Substack Read This First

Tickers Referenced: ,

Netflix piqued my interest following its audacious bid to buy Warner Brothers Discovery at an all cash $82billion purchase price. I wrote about this in a detailed essay in December 2025.

I followed this up with my own views (however myopic, but heavily drawn from my experience of integrating acquired businesses and technologies) from an integration perspective on what the challenges were and why I would stay clear of investing in Netflix.

This deep dive stemmed from a personal quest to gauge Netflix’s financial fortitude and the courage it possesses to successfully integrate Warner Bros’ assets into its operations.

As usual, my analysis will be framed through the lenses of Peter Lynch's "know what you own" discipline, Sleep's destination-compounder framework, Howard Marks's focus on permanent capital loss and price versus value, and Anthony Bolton's contrarian checklist of sentiment, governance, and balance sheet strength. I’ll refer to them where relevant throughout this essay.

TLDR;

1. Netflix has quietly evolved from a cash-burning disruptor into a cash machine, generating around $9.5bn of free cash flow in 2025 and guiding to roughly $11bn in 2026.

2. Over a decade, revenue has grown from about $6.8bn to $45.2bn, while operating margins have expanded from roughly 4–5% to nearly 30%, turning scale into genuine economic strength.

3. The investment case has shifted from “story stock” to “show me the numbers”: margins, free cash flow, and returns on capital are now robust enough to underwrite the thesis on hard data, not just narrative.

4. Per-share value creation is real: FCF per share has swung from deeply negative to around 2.17 (2025), and management has been retiring stock with over $9bn of buybacks in 2025.

5. Netflix increasingly resembles a long-duration “destination” compounder —using global scale and better product to feed a virtuous loop of engagement, pricing power, and reinvestment.

6. The main hard risk: large, largely fixed content obligations and a major Warner Bros. acquisition layered on top of existing debt, which could bite if growth or execution stumbles.

7. A familiar mix of comfort and discomfort: a widely owned winner, but with genuine uncertainty around ads, live, games and a transformational deal —exactly the sort of tension where mispricing can persist.

8. Netflix is now a global entertainment utility with a direct billing relationship, not a speculative tech punt; the traps are overpaying for content, overreaching on strategy, and assuming the moat is permanent.

9. The Warner Bros. deal is a live swing factor: financed with a large bridge and new credit facilities, buybacks paused to accumulate cash, and management promising to preserve investment grade — huge upside if executed well, real downside if not.

10. A thoughtful investor doesn’t need faith so much as patience: if engagement holds, ads scale, and capital allocation stays disciplined, Netflix can keep compounding per share; the art is sizing it so you can sit through the inevitable sentiment swings.

Once upon a time Netflix was a story stock. Investors flocked to it drawn by its narrative of a total addressable market disruption and blue-sky optionality. The idea was proposed to Hastings during a commute, sparked by a $40 late fee from Blockbuster for Apollo 13. This cultish journey of its founders Marc Randolph and Reed Hastings further captivated investors, despite weak or unproven economics.

In its 2015–2019 phase, Netflix fit that description much more closely:

The investment case was largely “streaming will take over the world, Netflix will be the winner, and the cash will come later” — classic story‑stock logic.

By contrast, over the last several years the numbers look much more like a mature, cash‑generative subscription platform:

That combination — large, recurring free cash flow, double‑digit operating margins, and explicit FCF guidance — moves the centre of gravity of the thesis away from “someday this might be huge” and towards “this is already throwing off cash; the question is how durable and how well it’s allocated”. On top of that, the company is now using that cash to reduce net share count (over $9bn of buybacks in 2025 alone, with diluted shares edging down), rather than issuing large amounts of equity to fund growth as in its earlier years.

That’s why I am analysing it more like a mature entertainment utility than a pure story stock.

It doesn’t mean the narrative has disappeared (there is still a lot of story around ads, live, games, and the WBD acquisition); it means the investment case can now be underwritten by demonstrated unit economics and cash generation, not just by faith in the end‑state.

Knowing what you own

At its core, Netflix sells a simple promise: pay a monthly fee (or watch some advertisements) to access a constantly refreshed library of video entertainment that is easy to discover and reliable to watch. The company operates as one segment, and revenues are primarily monthly membership fees, with a growing advertising contribution via its ad-supported plan launched in recent years. As of Q4 2025, Netflix crossed the 325 million paid memberships milestone and serves an audience "approaching one billion people globally".

In Peter Lynch's terms, Netflix is no longer an "early fast-grower" that can forgive financial indiscipline because the market is being created; it is closer to a "stalwart" with optionality.

Revenue has scaled, EBIT has expanded and Operating Margins have substantially improved.

Netflix is often described as a technology company, a media studio, or an advertising platform; in reality it is a distribution-and-portfolio business where content is the raw material and the interface is the shopfront. The operating truth is: if Netflix consistently supplies programming that feels worth the bill, people stay; if it doesn't, they leave, and the fixed-ish content cost base bites quickly. Subscribers cancel for many reasons — perception of insufficient use, need to cut household expenses, dissatisfaction with content (including advertisements), preference for competitive services, or unresolved customer service issues — and that adverse macroeconomic conditions can further pressure retention.

So what are the "favourable traits" and what are the "traps"?

Favourable traits include the direct billing relationship (no intermediary), global scale that spreads content risk, a business model where customers pay upfront for content amortised over time, and a product that benefits from network effects in discovery and word-of-mouth. The traps include the temptation to chase growth via over-spending on content or poorly conceived acquisitions, the risk of regulatory capture (quotas, levies, investment obligations described in the 10-K), and the danger of assuming competitive advantage is permanent when the barrier to entry for distribution of video is low, even if building a compelling content portfolio at scale is hard.

Is Netflix a Destination Compounder?

Two details matter for durability.

First, Netflix has explicitly centred its internal scorecard on revenue and operating margin (rather than membership counts alone), implying management believes the model is now mature enough that monetisation efficiency is at least as important as pure subscriber accumulation. The Q4 2025 letter states clearly: "Our primary financial metrics are revenue for growth and operating margin for profitability. Our goal is to sustain healthy revenue growth, expand operating profit and margin, and deliver growing free cash flow."

Second, capital allocation has become a first-order lever.

This is crucial through Nick Sleep's lens: long-duration compounders are not just "good businesses", they are machines that turn operating advantage into per-share value. In 2025, free cash flow per share is approximately $2.17, up sharply from prior years, and the combination of rising FCF and falling share count means owners increasingly receive the business's compounding in a concentrated way.

Another question is does Netflix use scale to improve the customer proposition — better service, better value — which then drives more scale in a self-reinforcing loop.

Netflix's version of this loop is:

The Q4 2025 letter explicitly frames scale as serving an audience approaching one billion people globally and describes product improvements (TV UI redesign, discovery features, live programming, games, video podcasts) and content breadth as the path to stickiness.

The moat question is whether Netflix's combination of global distribution, recommendation and merchandising technology, and content portfolio breadth produce a robust advantage that persists even when competition is fierce and budgets are large. Netflix itself emphasises that it competes for leisure time against streaming, linear television, social media, gaming, and other activities, and that the market is "intensely competitive and subject to rapid change". The 10-K notes that competitors include traditional broadcasters, cable operators, internet-based entertainment providers, technology companies, and local media companies, many with "long operating histories, large customer bases, strong brand recognition, exclusive rights to certain content, large content libraries, and significant financial, marketing and other resources".

On balance, Netflix’s moat looks real but not invincible. The defence is strongest where

(a) discovery and merchandising matter because the library is huge and heterogeneous,

(b) global scale spreads content risk across many markets and amortises fixed technology and localisation costs, and

(c) the direct billing relationship and data on viewing habits create a feedback loop that competitors without scale cannot easily replicate.

The offence is the advertising business, which management says will exceed $1.5bn in 2025 and double in 2026. This is important because advertising can monetise engagement that subscription pricing alone can’t capture, and ad-supported tiers lower entry prices and broaden the addressable market.

Sleep also asks if Netflix shares scale economics with customers enough to remain a destination rather than a churned-out utility. Periodic price increases and password-sharing enforcement can improve monetisation, but excessive measures risk weakening the shared-benefit loop by making Netflix feel like a toll booth rather than a great value. The 10-K explicitly acknowledges:

Sleep’s concept of the “long end of the equity yield curve” refers to businesses that can reinvest internally at high returns for extended periods, creating compounding that is hard to replicate via external capital allocation. However, earnings yield remain below treasury yields.

For Netflix, this means the question is not just “can it grow?”, but “can it keep deploying capital into content, technology, and international expansion at returns meaningfully above its cost of capital, whilst still returning surplus cash to owners?”

Avoiding Permanent Capital Loss

The future is unknowable in detail, so the job is to understand what can go wrong and ensure you are paid for bearing those risks.

For Netflix, permanent capital loss would come from a structural impairment of the model — loss of pricing power, a step-down in engagement that makes content spending less productive, or balance sheet stress that forces suboptimal decisions at the wrong time.

The most concrete “hard risk” today is leverage in a scenario where strategic ambition outruns the cash engine.

Critically, the Q4 2025 letter discusses the pending acquisition of Warner Bros. Discovery’s Warner Bros. business, noting that Netflix has obtained commitments for a $59bn senior unsecured bridge facility, subsequently reduced to $42.2bn via a $5bn revolving credit facility and a $20bn delayed draw term loan facility, with further increases to $42.2bn following the shift to an all-cash transaction structure.

Netflix announced in December 2025 that it would acquire Warner Bros., including film and television studios, HBO Max, and HBO, for $27.75 per WBD share in an all-cash transaction. The deal is fraught with risks

The filing also explicitly warns: “If we are unable to service our debt and other obligations from cash flows, we may need to refinance or restructure all or a portion of such obligations prior to maturity. If the financial markets become difficult or costly to access, including due to rising interest rates, fluctuations in foreign currency exchange rates or other changes in economic conditions, our ability to raise additional capital may be negatively impacted, and any refinancing or restructuring could be at higher interest rates and may require us to comply with more onerous covenants, which could further restrict our business operations.”​

Even without the acquisition, Netflix would focus on the fixed costs of content commitments. Netflix disclosed its total streaming content obligations of about $24 billion as of December 31, 2025, including content liabilities on the balance sheet and obligations not yet recognised under accounting standards. The 10-K warns that if business performance doesn’t meet expectations, margins may be adversely impacted due to the multiple-year duration and largely fixed cost nature of content commitments. Further, Netflix may be unable to reduce streaming content obligations in the near term if cash flows from operations decrease, even during economic downturns.

This is the core risk mechanism: if revenue growth stalls unexpectedly, content and operating costs cannot instantly shrink, so margins and cash flow can compress quickly. The 10-K notes that payment terms for content Netflix produces directly typically require more upfront cash than licensed content, and that revenue growth not meeting expectations could adversely affect liquidity and results of operations.

On the other hand, Netflix’s recent cash generation is a genuine risk-reducer. In 2025, Netflix reported net cash provided by operating activities of about $10.15 billion and free cash flow of $9.46 billion, with management’s 2026 FCF guidance of roughly $11 billion (assuming no material FX swings) implying continued strength. In Marks’s language, stronger liquidity and internal funding reduce the odds of being forced to sell equity or borrow at punitive rates in a downturn.

Netflix’s price has changed from being priced like a distressed turnaround to trading like a quality compounder.

As the business looks better, expectations in the price must be checked because good news becomes fully priced and then fragile. Netflix’s enterprise value per share has risen from about $10 in 2015 to approximately $119 in 2025, while EBIT per share has grown from $0.07 to $3.07 over the same period. This shows that valuation expansion (multiple re-rating) has been a significant component of total shareholder returns, not just fundamental improvement.

Sentiment, Discomfort, and the “Loneliness Premium”

Netflix today is widely followed and widely owned; that usually means less loneliness premium is available than in smaller, messier companies. Yet discomfort can still exist in large caps when the debate shifts from “will it survive?” to “can it keep compounding at scale?”, particularly with uncertainties around advertising execution, live programming, games, and major M&A integration risk.

Bolton would likely divide the current setup into two opposing sentiment forces. One is confidence in a proven cash-generating model with expanding margins and ongoing buybacks, allowing investors to expect steady compounding. The other is fear that streaming is mature, attention is fragmenting (YouTube, social platforms, open content ecosystems), content inflation is structural, and strategic overreach, especially through acquisitions, could limit returns. The Q4 2025 letter notes that Netflix’s share of TV time in major markets remains below 10% (9.0% in the US in December 2025, an all-time high but still a minority), and that linear TV still accounts for over 40% of US TV screen time, which can be interpreted as either a ‘huge runway’ or ‘entrenched incumbents are sticky’.

The pending Warner Bros. deal is a prime example of a “special situation” characteristic. It could strengthen the bundle, expand the library, improve multi-tier pricing, and accelerate advertising time-to-scale. However, it also poses integration, regulatory, and financing risks that could be detrimental if mismanaged. Bolton’s focus on governance and balance sheet management suggests monitoring capital allocation discipline, including buybacks when undervalued and restraint when not. A clear plan to maintain leverage within a comfortable range is essential. Pausing buybacks to accumulate cash for the acquisition is prudent from a balance sheet perspective, but it will slow near-term per-share value creation until the deal closes and post-merger buybacks resume.

Bolton would also examine management quality and incentives. Netflix is co-CEO-led (Greg Peters and Ted Sarandos) and has a unique ‘Netflix Culture Memo’ that emphasises excellence, transparency, and a ‘dream team’ approach to talent. Employees are paid at their ‘personal top of market’ and can choose between cash and stock options, which management believes aligns long-term interests.

However, past returns don’t guarantee future returns and can lead to overconfidence.

So What Must Be True for Long-Term Success

• The core subscription product must keep feeling like good value so retention stays healthy, because Netflix itself notes that cancelled memberships can rise for reasons ranging from perceived low usage to macro pressure to dissatisfaction with content or advertisements. The Q4 2025 letter reports that engagement in the second half of 2025 was up 2% year-on-year (driven by a 9% rise in viewing of branded originals), but also that non-branded view hours declined year-on-year, primarily reflecting lower licensed second-run content following elevated 2023–2024 licensing during the WGA strike. Sustaining engagement as the licensed content normalises will test whether Netflix’s originals slate is strong enough to stand alone.

• Operating margin expansion must continue gradually (management targets 31.5% in 2026, up from 29.5% in 2025) without starving the content engine that sustains engagement. The 2026 margin forecast reflects content amortisation growth of 10%, with higher growth in the first half than the second half due to title launch timing, and management expects “plenty of room to increase our margins” with the intent to grow operating margin each year, though “the magnitude of margin expansion will vary year-to-year as we balance reinvesting in our business with improving profitability”.​

• Advertising must scale from “promising” to “meaningful”, because management reports 2025 ad revenue of over $1.5bn (more than 2.5× versus 2024) and expects it to roughly double in 2026 — which, if achieved, would approach $3bn and add a second monetisation engine that can lift ARPU without pure subscription price rises. The letter also notes Netflix is using AI to help advertisers create custom ads, automate campaign planning, and improve targeting, which could lower friction and accelerate adoption.​

• Free cash flow must remain structurally positive and increasingly directed towards per-share compounding, as shown by rising FCF and substantial repurchases (recognising buybacks may pause for M&A funding needs). The 2026 FCF guidance of approximately $11bn assumes a cash content spend to content amortisation ratio of 1.1×, implying disciplined content investment that does not outrun revenue growth.

And What Could Cause Permanent Disappointment

• Content becomes less productive: If the slate fails to create “fandom” and must-see moments often enough, Netflix may need to spend more merely to stand still. The Q4 2025 letter explicitly discusses fandom as “a powerful engine for our business” and notes that some titles “spark a level of passion and fandom that delivers outsized impact for both our members and our business”, translating into “greater member satisfaction, higher retention, and increased word-of-mouth and acquisition, fueling both audience growth and brand loyalty”. Conversely, if the hit rate declines, the fixed-cost content base becomes a drag.

• Monetisation backlash: If pricing, ad load, or password-sharing enforcement is pushed beyond what customers tolerate, Netflix explicitly acknowledges it could harm acquisition and retention. The 10-K warns: “If consumers do not perceive our service offering to be of value, including if we introduce new or adjust existing features, adjust pricing or service offerings, or change the mix of content in a manner that is not favorably received by them, we may not be able to attract and retain members, and accordingly, our revenue and results of operations may be adversely affected.”​

• Balance sheet strain: Large debt-funded moves — particularly a poorly executed acquisition — could increase indebtedness materially and reduce flexibility, especially if a cyclical downturn or higher interest rates coincide. The Warner Bros. transaction will be a live test of management’s M&A discipline, integration capability, and ability to navigate regulatory scrutiny whilst maintaining an investment grade rating.

• Regulatory and competitive friction: Netflix notes expanding regulation (content quotas, levies, investment obligations, potential restrictions on content review and ownership rights) and intense competition (traditional media, technology platforms, piracy). None alone is fatal, but they can compress the economic moat at the margin and raise the cost of doing business in key markets. The 10-K states: “Piracy also threatens to damage our business, as its fundamental proposition to consumers is so compelling and difficult to compete against: virtually all content for free.”​

In Marks’s terms, these are the paths to “permanent capital loss”: not volatility, but a lasting impairment of cash generation and strategic freedom.

How to Think About Sizing and Behaviour Through Cycles

A thoughtful investor would view Netflix as a high-quality business with a volatile stock price due to sentiment and valuation fluctuations. The 10-K states that the stock price is subject to wide fluctuations driven by quarterly results, guidance changes, announcements, financial estimates, and market conditions. Therefore, position sizing should allow for holding through drawdowns without forced exits, as the best buying opportunities often arise during uncomfortable near-term news but stable long-term fundamentals.

Behaviourally, the practical discipline is to separate “business progress” from “share price mood”. If operating margin and free cash flow keep growing while the share count decreases, the per-share value story improves even if the market temporarily undervalues the stock. However, if the market is overly optimistic and valuations suggest perfection, for example, if the EBIT/EV earnings yield compresses to levels that imply flawless execution for many years, Marks would advise caution: great companies can still be poor investments when bought at the wrong price, and the market will eventually correct itself.

My Final Verdict (for now): A Compounder with Execution Risk

Netflix has travelled an extraordinary distance: from a DVD-by-post service to a global streaming leader, from persistent cash burn to nearly $10bn of annual free cash flow, and from a speculative growth stock to a profitable stalwart with reinvestment optionality. The per-share metrics tell a story of genuine value creation, not just revenue growth.

The investment case rests on a few core beliefs:

1. That Netflix is a "boring brilliant" stalwart whose habitual, global, direct-to-consumer model can sustain modest growth and high incremental margins, provided management avoids the traps of overpaying for growth or assuming competitive advantage is permanent.

2. That Netflix can remain a destination compounder by sharing scale economics with customers (great content, easy discovery, reasonable value), reinvesting in content and product at high returns, and returning surplus cash to shrink the share count.

3. That the Netflix balance sheet is strong enough to absorb a major acquisition and a cyclical downturn without forced selling or value-destructive financing, provided the Warner Bros. integration is executed with discipline and the company does not stretch leverage beyond investment grade comfort.

4. That whilst Netflix is no longer a “lonely” contrarian bet, the combination of advertising scale-up, international growth, and a major M&A event creates genuine uncertainty that could reward patient owners if fundamentals improve whilst sentiment remains cautious.

What must be true is that engagement stays healthy, pricing power holds, advertising scales, and management allocates capital with discipline. What could go wrong is that content productivity declines, monetisation is pushed too hard, the balance sheet is over-leveraged for an acquisition that disappoints, or regulatory and competitive pressures compress the moat faster than innovation can widen it.

A long-term owner should size the position to withstand volatility, prioritise per-share progress over quarterly fluctuations, and be ready to invest more if the business continues to grow but the market loses interest. Netflix is no longer a risky investment; it’s a proven success. The question is whether it can maintain its compounding growth at scale, or if pursuing growth through mergers and acquisitions and new business ventures marks the transition from a compounding company to a conglomerate.

Read my DISCLAIMER

If you are new to my Substack Read This First

All Code Blocks in this essay were generated using Perplexity AI and are not tested. They are only used to substantiate or expand on a relevant assertion.

Tickers Referenced:

This essay is a follow-up on my early three essays:

The Big Three (Salesforce, ServiceNow, Workday) emerged precisely because they replaced monolithic, on-premise, human-centric platforms with cloud-based speed and simplicity. Yet they are now facing the same existential choice their predecessors faced, whether to cannibalise their own business model to remain relevant in the next era.

Let me work through the architectural, data, and commercial implications systematically.

From Procure-to-Production: The Deployment Inversion

My historical framing is essential.

Legacy enterprise software operated on 12-24 month cycles: procurement debate (1-12 months), infrastructure setup, data migration, business process mapping, UAT, go-live theatre, then hyper-care support. SaaS compressed this to 1-6/12 months by simplifying configuration, hosting infrastructure, and offering pre-built workflows.

Agentic AI inverts this entirely. A well-architected agentic platform should deliver initial value in 2-4 weeks, not months.

Here’s how:

• Week 1: Goal alignment (what workflow should this agent handle?), process discovery (parse existing data to understand current state), agent selection (use pre-trained agents)

• Week 2: Data integration (connect to tracking system, Web Portals, document parsers), agent configuration (define criteria, scheduling rules)

• Week 3-4: Pilot execution, human-in-the-loop review, iterate on agent decisions

The critical difference: agents learn from domain data rather than requiring customisation. Traditional SaaS needed business analysts to translate workflows into configuration. Agentic AI platforms ingest historical data, infer the workflows, and propose agent behaviours to match past practice.

Implication for the Big Three: If deployment becomes a 4-week problem rather than a 4-month problem, the economics of professional services delivery collapse. They cannot sustain high-margin implementation revenue by selling 200 hours at $300/hour when customers self-service agents in weeks. This pushes them toward platform-delivered outcomes rather than implementation-led revenue.

Architecture: Process-Centric vs. User-Centric

This is where the architectural reckoning begins. Legacy SaaS is fundamentally user-centric: humans trigger workflows through dashboards, forms, and buttons. The system is stateless — each request is independent; the database is the only persistent truth.

Agentic AI requires process-centric architecture, which is architecturally alien to SaaS companies trained on stateless request-response models.

The Three-Tier Agentic Architecture

Foundation Tier (Language models, embeddings, vector databases)

• The “intelligence” substrate

• Salesforce’s Atlas (hybrid reasoning model), ServiceNow’s Now Assist LLMs, Workday’s custom models

• Critically: organisations need to swap models at runtime depending on cost/latency tradeoffs (GPT-4 for complex decisions, smaller models for simple classification)

Workflow Tier (Orchestration, planning, multi-agent coordination)

• This is the real competitive moat, not the LLMs

• Central orchestration engine that decomposes goals into agent tasks

• Handoff patterns: Agent A (recruiter triage) → Agent B (technical screening) → Agent C (offer generation) → Human (final sign-off)

• Agents route dynamically based on confidence, capability gaps, or policy violations

• Data structure required: Workflow state machine tracking current stage, which agents have touched the workflow, why each handoff occurred

Autonomous Tier (Agent execution, tool invocation, memory management)

• Agents act against APIs: ATS, email, calendar, Slack

• Durable execution: if an agent pod crashes mid-workflow, the system resumes exactly where it left off (requires frameworks like Temporal or Akka; stateless servers won’t work)

• Long-term memory: agents need to remember “we screened 200 candidates from MIT and 80% advanced” to inform future screening thresholds

The Architectural Requirement All Three Must Build

Enterprise Orchestration Layer (newly required)

• Acts as the control plane for all agentic work

• Enforces governance policies: no agent can do X without Y approval

• Builds comprehensive process models from agent activity telemetry

• Enables continuous optimisation: “Agent identified this screening step takes 3x longer than industry benchmark; codify the shortcut as a reusable playbook”

• Maintains compliance audit trail: every action logged with context, approval, rationale

This is perhaps the most important architectural insight: in a multi-agent world, you don’t just need agents; you need a meta-orchestrator that coordinates them, enforces policy, and learns from their behaviour.

Data Model Transformation: The Tokenised, Action-Centric Inversion

This is the deepest shift and the one least discussed. Current SaaS data models are fundamentally user-centric:

sql
CREATE TABLE Orders (
  order_id INT,
  user_id INT,          -- "Who created this order?"
  created_by INT,       -- Human user ID
  created_date TIMESTAMP,
  amount DECIMAL
);

Revenue is tied to this user_id: more users = more revenue. The entire business model hangs on this relationship.

Agentic data models must become action-centric and tokenised.

sql
CREATE TABLE Agent_Actions (
  action_id INT,
  agent_id INT,         -- "Which agent performed this?"
  workflow_id INT,
  action_type VARCHAR,  -- e.g., "candidate_screen", "schedule_interview"
  tokens_consumed INT,  -- The new unit of value
  cost_usd DECIMAL,
  confidence_score FLOAT, -- Did the agent make this with high certainty?
  human_approval_required BOOL,
  approval_status VARCHAR,
  business_outcome VARCHAR  -- "Candidate moved to Stage 2"
);

The implications are staggering:

1. Metering becomes core infrastructure: Every API call, every decision, every data retrieval must be logged with token count. This is not an afterthought; it’s foundational.

2. Pricing data is embedded in operational data: You can no longer separate the “operational database” from the “billing database.” Every action is simultaneously an operational event and a billable event.

3. Approval workflows are part of the data model: Human governance is not an audit layer on top; it’s embedded in Agent_Actions (approval_required, approval_status, approver_id). This ensures every organisation can see who approved what and why.

4. Memory becomes a queryable entity:

sql
CREATE TABLE Agent_Memory (
  memory_id INT,
  agent_id INT,
  semantic_embedding VECTOR(384),  -- From embeddings model
  context TEXT,
  relevance_score FLOAT,
  created_timestamp TIMESTAMP
);

Agents query this not just by matching IDs, but by semantic similarity: “Find all workflows where we screened candidates from software companies.”

5. Workflow tracking is explicit and observable:

sql
CREATE TABLE Workflows (
  workflow_id INT,
  orchestrator_id INT,
  goal_statement TEXT,
  agent_sequence VARCHAR[],  -- [Triage Agent, Technical Agent, Offer Agent]
  current_stage INT,
  status VARCHAR  -- IN_PROGRESS, BLOCKED, ESCALATED, COMPLETED
);

What This Means for Tokenisation

A “token” is not an abstract concept; it’s a first-class entity in the database, tied to every action:

• Recruiting example: Screening one candidate = 50 tokens (call ATS API, run classifier, generate summary, queue for human review)

• Pricing: $0.005/token = $0.25 per candidate screened

• Billing: Automatic; customer’s bill accumulates in real-time as agents execute

• Forecasting: “You’ve consumed 400k of 500k baseline tokens; at current velocity, you’ll hit overage in 8 days”

This is fundamentally different from per-seat licensing because:

• A recruiter seat might process 5 candidates/day = 1,250 tokens/day → $6.25/day

• But if organisations deploy agents to do the same work, they’ll process 50 candidates/day = 12,500 tokens/day → $62.50/day

• The vendor’s revenue grows with customer value, not with customer employee count

But this requires architectural discipline. Salesforce, ServiceNow, and Workday must redesign their core databases. They cannot bolt tokenisation on top of user-centric schemas; they must migrate to action-centric schemas. This is a multi-year engineering commitment.

Human-Agentic Orchestration: The Approval Patterns

One of the most underestimated challenges is designing where humans intervene in agentic workflows and how that intervention is tracked, auditable, and doesn’t become a bottleneck.

There are three primary patterns:

Pattern 1: Handoff Orchestration (Dynamic Specialisation)

text
Recruitment Triage Agent (confidence: 92%)
  ↓ Determines: "This candidate needs technical screening"
Technical Screening Agent
  ↓ Runs code challenge, calculates score
  ↓ Confidence: 87% (edge case not in training data)
Senior Engineer Human
  ↓ Reviews code, makes judgment call
  ↓ Approves; sends back to system
Offer Generation Agent
  ↓ Creates offer with approved parameters

Data model requirement: Track the entire handoff chain in a single workflow record. When the senior engineer reviews the candidate, the system must provide context: what did the triage agent observe? Why did it pass? What’s the technical agent’s rationale?

Pattern 2: Confidence-Based Escalation

Instead of pre-defining who approves what, let confidence levels drive approval routing:

text
Confidence > 95% → Execute autonomously; log decision with high-confidence flag
Confidence 70-95% → Execute, but flag for human post-review within 24 hours
Confidence < 70% → Escalate to human BEFORE executing; wait for approval

Advantage: Reduces the number of decisions humans must make. High-confidence decisions rarely need review. Low-confidence decisions get priority attention.

Data model: Agent_Actions must track confidence_score + approval_type (PRE vs. POST).

Pattern 3: Policy-as-Code Escalation

Organisations have policies that should automatically trigger human intervention if violated:

sql
-- Policy: No offer can exceed 40% above market rate
IF offer_amount > market_rate * 1.4 THEN
  escalate_to_human_for_approval = TRUE
  policy_violation_type = "Offer Above Threshold"
END;

-- Policy: No candidate from Region A can be rejected without technical screen
IF region = 'A' AND rejected_without_screen = TRUE THEN
  escalate_to_human_for_review = TRUE
  compliance_flag = TRUE
END;

Critical requirement: Policy violations must be detectable in real-time and escalate asynchronously without blocking the workflow. The agent should not freeze; instead, it logs the action as “pending policy review,” executes provisionally, but the human is alerted immediately.

The Governance Architecture All Three Must Build

Action-Level Approvals (ServiceNow is leading here)

Instead of broad role-based permissions (”Sarah can do X because she’s a hiring manager”), assign permissions at the action level:

text
Agent Action: Send offer to Candidate X
Requires Approval From: Manager (approval_type = "offer_authority")
Context: 
  - Market rate: $120k
  - Proposed offer: $135k (+12.5%)
  - Agent confidence: 88%
  - Candidate qualifications: [matches all criteria]
Approval Route: Slack message to candidate's hiring manager
Approval Timeout: 4 hours (escalate if no response)
Audit Trail: Every approval logged with timestamp, approver identity, rationale

Why this matters:

• Compliance teams can prove audit trails (who approved, when, why)

• Organisations can tune approval thresholds (if 95% of offers are approved, lower the threshold)

• Agents can learn what was approved and what wasn’t, improving future decisions

• Humans are not bottlenecks; they’re directing high-stakes decisions, not reviewing routine ones

License Models: The Uncomfortable Transition

You’ve touched the nerve. The path from per-seat to tokenised is commercially and politically fraught.

The Hybrid Transition (Today → Q1 2026)

ServiceNow’s model (closest to working example):

• Base subscription: $50k/year (traditional CRM or IT service seat bundle)

• Now Assist Packs: $500 each (= ~10M tokens for the year)

• Customer buys 5 packs: $2,500

• Total: $52,500/year

Advantage: Predictability (base is fixed) + flexibility (variable component for agents)
Disadvantage: Vendor doesn’t fully capture AI upside; customer’s agent spend is capped

The Consumption Escalation (Q2 2026 → 2027)

As customer adoption of agents matures, the pricing model should evolve:

1. Usage-based with guardrails:

   • Base: $X/year (includes Y million tokens)

   • Overage: $Z per million tokens above Y

   • But offer monthly cap: “You won’t pay more than $N/month, guaranteed” (protects customer finance team)

2. Enterprise flat-fee ALAs (Agentic License Agreements):

   • Multi-year contract (3 years, e.g.)

   • Unlimited tokens for fixed annual fee

   • Advantage for vendor: Predictable revenue; advantage for customer: Cost certainty

3. Outcome-based (speculative, 2027+):

   • Vendor takes % of value created

   • Example: “We’ll run your recruitment agents; you pay us 5% of salary savings from better hiring”

   • Requires shared data access and trust; early attempts will fail

The Business Model Transition Peril

Here’s why this is genuinely difficult for the Big Three:

Year 1: Gross margin may compress

• Current SaaS gross margins: 70-75% (cost of goods sold is mostly cloud infra + support staff)

• Agentic AI introduces new costs: LLM API fees (OpenAI, Anthropic), token metering infrastructure, governance tools

• If Salesforce’s gross margin drops to 65% because of agent infrastructure costs while analysts expected 72%, stock will dip

Year 2: Revenue unpredictability increases

• With seat-based pricing, you can forecast: “1,000 customers × 50 seats × $150/seat × 3-year contract = predictable ARR”

• With consumption-based pricing: “1,000 customers × ?? tokens/customer × $X/token = ???”

• The variance is high, making guidance difficult

• Wall Street dislikes surprises; stock downside as guidance range widens

Year 3: Sales team resistance

• Salespeople are compensated on ACV (annual contract value) closed

• With per-seat: “Close a 500-seat deal at $150/seat = $75k ACV” (one-time commission bump)

• With consumption: “Close a 50-token-pack deal at $500/pack = $2.5k ACV” (much lower immediate commission; rely on expansion over time)

• Sales teams will resist and may churn to competitors with simpler licensing

Why it’s necessary anyway:

• Competitors will move first (some already have; Salesforce has Flex Credits, ServiceNow has Assist Packs)

• If vendors don’t transition, they’ll watch customers deploy agent platforms from competitors

• The upside is unlimited TAM (revenue grows with customer automation, not customer headcount)

Pricing Models: What the Big Three Should Watch Each Other Doing

New “North Stars”: What Success Actually Means in Agentic Era

This is perhaps the most important section. Traditional SaaS metrics (ARR, NDR, CAC) will persist but become insufficient.

Customer-Focused North Stars

1. Autonomous Task Completion Rate (70-80% target)

   • % of workflows completing without human intervention

   • Example: “85% of candidates screened by agent; 15% escalated for human review”

   • Why this matters: Reflects true automation value; low % means agents aren’t trusted

2. Mean Time to Resolution (MTTR) (50-70% reduction target)

   • How fast workflows complete end-to-end

   • Example: “Recruitment cycle: 14 days (human-only) → 3 days (with agents)” → 79% improvement

   • Why this matters: Speed is the true value prop for customers

3. Human Time Reclaimed (20-30% of workforce)

   • Hours per week freed from manual work

   • Example: “5 recruiters × 40 hours/week = 200 hours freed”

   • Monetised: 200 hours × $50/hr × 52 weeks = $520k annual value

   • Why this matters: Justifies the investment; shows real ROI

4. Process Error Reduction (>90% accuracy target)

   • Decrease in rework, escalations, manual corrections

   • Example: “Candidate screening errors: 15% → 2%”

   • Why this matters: Quality metrics matter more than volume metrics

5. Agent Learning Velocity (5-10% monthly accuracy improvement in Year 1)

   • How fast agents improve over time

   • Measured: Accuracy gain per feedback cycle, per 100k tokens processed

   • Why this matters: Proves platform is improving, not decaying

6. ROI (3x-6x Year 1, 5x-10x Year 2)

   • (Cost Savings + Revenue Gains - Implementation Costs) / Investment

   • Why this matters: Ultimate business justification; drives renewal and expansion

Vendor-Focused North Stars

1. Tokens-as-Revenue (TAR) (replacing ARR)

   • Annual tokens consumed × price per token

   • Example: “1,000 customers consuming 50B tokens/year at $0.005/token = $250M ARR equivalent”

   • Why this matters: Direct proxy for customer value; TAR growth >50% YoY signals healthy adoption

2. Blended Net Expansion Rate (BNER) (130%+ target, vs. 120% for traditional SaaS)

   • (Starting ARR + Expansion - Churn) / Starting ARR

   • Should be higher for agentic platforms because token consumption expands naturally

   • Why this matters: Shows customers increasing automation investment over time

3. Agent Adoption Velocity (50%+ of customer base within 12 months of GA)

   • % of customers deploying agents from zero to first agent in production

   • Why this matters: If agents aren’t adopted, the entire business model transition fails

4. CAC Payback Period (8-12 months target, vs. 12-18 for traditional SaaS)

   • Months to recover customer acquisition cost

   • Should improve because deployment time is shorter, time-to-value is faster

   • Why this matters: Indicates efficiency of sales & implementation model

5. Gross Margin (65-75%, despite inference costs)

   • Revenue - (COGS + LLM inference fees + hyperscaler costs)

   • Key challenge: As scale increases, LLM API costs can squeeze margins

   • Why this matters: Proves the unit economics work at scale

6. Escalation Rate (<5% target)

   • % of agent decisions requiring human intervention

   • Low escalation = high agent autonomy = customers trust the system

   • Why this matters: Reflects maturity; high escalation = governance is too heavy

7. Multi-Agent Orchestration Adoption (60-70% of workflows)

   • % of automated workflows involving 2+ agents coordinating

   • Why this matters: Signals evolution from isolated agents to enterprise-scale orchestration

The Uncomfortable Truth: These North Stars Are Not Comparable to Traditional SaaS Metrics

A CEO or CFO looking at:

• ARR: $50M

• NDR: 125%

• CAC: $150k

• Churn: 5%

Can easily benchmark against competitors. But if the same vendor reports:

• TAR: $50M (tokens-as-revenue)

• Agent Adoption: 45%

• MTTR Improvement: 65%

• ROI: 3.8x

Investors and analysts won’t know how to evaluate it. This is why the transition is risky: you’re asking the market to re-learn how to value your company just as you’re transforming it.

The Implementation Partner Extinction

One final implication worth calling out: implementation partners (Deloitte, Accenture, Cognizant etc.) will lose revenue in agentic-first architectures.

• Legacy SaaS model: 30% software licensing revenue, 70% implementation services revenue (data migration, process mapping, customisation)

• Agentic model: 90% software (platform + agents), 10% implementation services (mostly training and governance setup)

This creates an incentive misalignment: implementation partners want slower, manual transitions (more billable hours). Agentic platforms want self-service, rapid deployment.

Salesforce, ServiceNow, and Workday will increasingly self-deliver implementation through dedicated agentic platform teams, reducing their reliance on partners. This is actually healthy for customers (faster, cheaper) but represents a fundamental shift in go-to-market.

Synthesis: The “North Star” Question

So what should be the North Stars for these companies. Based on this analysis:

In 12 months, the Big Three’s earnings calls should be dominated by these metrics:

1. Agent Adoption Rate: “X% of our customer base now has agents in production”

2. Tokens-as-Revenue trajectory: “TAR up YoY by X%; blended unit economics of tokens consumed per customer”

3. Deployment time: “Average time from contract to first agent in production: X days” (should be <30 days)

4. Customer ROI: “Average customer realising 4.2x ROI on AI agents; Y% of customers report >100% ROI”

5. Escalation and autonomy rates: “Z% of agent decisions execute without human approval; escalation rate below X%”

6. Gross margin on agentic products: “Despite LLM costs, agentic gross margin holding at 68% (vs. 72% legacy SaaS)”

If these metrics are not being cited, it means the companies haven’t genuinely transformed. They’re layering AI features on top of legacy business models, which will eventually lose to companies that embrace the agentic transition fully.

The Disconfirming Evidence

Salesforce’s own stance on AI‑agent pricing provides strong disconfirming evidence for a hard, fast cannibalisation of seat‑based models. Far from embracing pure tokenisation, Salesforce has moved back toward seat‑based AI licensing, with the Agentic Enterprise License Agreement (AELA) as the preferred model. AELA is tied to the number of seats, and customers can reuse credits within a capped envelope rather than facing open‑ended consumption costs. This is the opposite of the unconstrained token‑driven world outlined in your thesis: it preserves predictability for buyers and control for Salesforce while leaving the core per‑seat revenue model intact. That shift suggests that investors may be over‑projecting how quickly or how fully AI agents will replace human‑centric seating.

There is also real pushback on the agent‑first UX in Salesforce, with customers complaining that Agentforce can be slower and more frustrating than classic search, and that Salesforce is forcing AI interactions into places where users simply want fast, deterministic answers. This is symptomatic of a broader tension: enterprise buyers are willing to pay for AI that augments human‑centric workflows, not one that replaces them. Salesforce’s own marketing for Agentforce 360 repeatedly frames agents as elevating human potential, not obsoleting it, which underlines that the platform is still rooted in the same user‑centric CRM layer you critiqued.​

For ServiceNow, while AI Agent Fabric and the Control Tower sound like the archetypal agentic orchestration stack, early‑adopters are signalling that implementation is heavy, multi‑month, and over‑engineered for many use cases. Third‑party analysts describe the Workflow Data Fabric as “complex, multi‑month project, not a plug‑and‑play solution,” and note it is overkill for agile, focused automation. That complexity is a structural brake on the kind of lightweight, token‑driven agent economy you envision; it keeps projects in the purview of IT and integration teams, not business users, and reinforces the traditional SaaS implementation cycle rather than collapsing it into weeks.

Workday’s Agent System of Record is similarly framed as a governance and control layer, not a full agentic re‑platforming of the core HCM and finance workflows. It assumes that agents are add‑ons to an existing human‑centric system of record, and that the primary value is risk‑managed deployment, not a fundamental re‑architecture of how work gets done. Reports show that adoption is gated by data‑quality and role‑definition issues, and that the focus is on “real‑time visibility and cost tracking” rather than autonomous workflow execution. This is consistent with the view that agentic AI will augment and extend SaaS, not rewrite it from the ground up.

Across the board, disconfirming evidence clusters around three themes:

• Customers are not yet operating at the token‑level: they want predictable per‑seat or hybrid pricing, not open‑ended usage‑based bills.

• Agent adoption is slower than marketers claim, constrained by data‑quality, integration complexity, and trust barriers, not by a lack of architectural vision.

• The core units of value are still human‑centric workflows, with agents bolted on at the edges rather than structurally re‑defining the product’s data model and interaction model.

From an investor’s perspective, the key questions are:

• Pricing signalling: is the company leaning into token‑driven, usage‑based models, or reinforcing per‑seat economics via AELAs and hybrid caps?

• UX orientation: is the product shifting from human‑centric dashboards and forms to agent‑centric conversational and autonomous flows, or merely adding AI copilots inside legacy screens?

• Adoption evidence: what are the real‑world usage and retention stats for agentic modules, and what are the explicit blockers that customers cite?

• Architectural stamina: is the company re‑investing heavily in new data models, orchestration layers, and observability that can support true multi‑agent workflows, or is it layering AI on top of existing stacks?

Investors should treat the agentic‑AI thesis as a multi‑year evolution, not a near‑term disruption. The Big Three will experiment with tokens and consumption, but are equally likely to fall back on hybrid models that preserve the core seat‑based business while monetising AI at the margin. The real signal will be when their North Stars explicitly shift from ARR and seat counts to TAR (tokens‑as‑revenue), autonomous task completion rates, and MTTR reductions, and when their pricing and architecture have genuinely re‑centred around agents rather than users.

Reference Material:

1. https://promethium.ai/guides/enterprise-ai-implementation-roadmap-timeline/

2. https://www.ruh.ai/blogs/enterprise-framework-safe-fast-ai-deployment-guide

3. https://engineering.leanix.net/blog/saas-are-moving-to-agentic-centric/

4. https://architect.salesforce.com/fundamentals/agentic-enterprise-it-architecture

5. https://www.linkedin.com/pulse/human-in-the-loop-ai-governance-design-choice-vicente-carvalho-kfqsc/

6. https://www.runonatlas.com/blog-posts/how-to-launch-your-first-action-based-pricing-model

7. https://lorojournals.com/index.php/emsj/article/view/1568

8. https://dcai.csail.mit.edu/2024/data-centric-model-centric/

9. https://agentic-design.ai/patterns/multi-agent/handoff-orchestration

10. https://www.theregister.com/2026/01/27/salesforce_ai_gartner/

11. https://www.theregister.com/2025/12/12/ai_agents_salesforce_pricing/

12. https://www.weforum.org/stories/2025/12/3-obstacles-to-ai-adoption-and-innovation-and-how-to-overcome-them/

Subscribe now

Read my DISCLAIMER

If you are new to my Substack Read This First

Tickers referenced:

For years, the trillion-dollar valuation club has resembled an exclusive Silicon Valley dinner party. Entry required one or more of the following: an obsession with artificial intelligence, a fondness for subscription models, and a corporate culture that encourages employees to wear hoodies to earnings calls. Yet, in a twist that would have sounded improbable a decade ago, the newest entrant into this gilded circle is not a cloud computing titan or a semiconductor wizard.

Welcome Walmart — the undisputed monarch of discount retail and a business more closely associated with bulk toilet rolls than machine learning breakthroughs.

Walmart’s ascent past the $1 trillion market capitalisation milestone is remarkable and quietly inevitable. It represents a victory for a retailer, and a symbolic moment for the broader evolution of commerce. The modern retail battleground has become a sophisticated technological arms race where logistics, data, and artificial intelligence increasingly determine who wins the weekly grocery shop.

More importantly, Walmart’s arrival in this rarefied club reveals something rather profound about the changing nature of “technology companies”. Increasingly, every successful large enterprise is becoming one.

The Unexpected Tech Giant in Plain Sight

To understand why Walmart reaching a trillion-dollar valuation is such a significant moment, it helps to appreciate the company’s origins. Walmart built its empire through relentless operational discipline. Its founder, Sam Walton, famously focused on efficiency, supply chain mastery, and a simple yet brutally effective promise: everyday low prices.

For decades, Walmart’s competitive advantage rested on scale and logistics. It perfected inventory management long before it became fashionable to label such systems as “data-driven”. Walmart’s supply chain has often been described as one of the most efficient in human history, allowing it to squeeze costs out of distribution in ways that competitors could barely comprehend.

Yet, what began as operational excellence has quietly transformed into technological sophistication. Walmart’s modern growth engine is powered by e-commerce infrastructure, automated fulfilment centres, sophisticated data analytics, and increasingly, artificial intelligence.

The company’s AI-powered chatbot, Sparky, now assists customers in navigating the labyrinth of online product catalogues, recommending alternatives and improving search relevance. Partnerships with technology leaders such as Google and OpenAI suggest Walmart recognises that the next frontier of retail will be mediated by digital assistants and platform ecosystems. The shopping basket is becoming more of an algorithmic outcome.

In short, Walmart has embedded technology into its commercial DNA.

The Amazon Rivalry: Retail’s Heavyweight Bout

No discussion of Walmart’s rise can ignore its enduring rivalry with Amazon. If Walmart represents retail’s physical dominance evolving into digital strength, Amazon represents the inverse — a digital empire attempting to anchor itself in physical commerce.

Amazon remains comfortably ahead in market value, sitting above $2.5 trillion, and continues to dominate e-commerce. Yet, Walmart’s progress reveals that the supposed inevitability of Amazon’s retail supremacy was perhaps overstated.

The competitive dynamics between these two giants now resemble a fascinating convergence. Walmart is investing heavily in digital infrastructure, online subscriptions, and advertising platforms, while Amazon experiments with physical retail formats, including large superstores and continued grocery expansion through Whole Foods.

This convergence highlights a simple but critical truth: the distinction between online and offline retail is fading. Consumers expect seamless integration across channels. They may browse online, purchase via mobile, and collect in store — or vice versa. Retail success increasingly depends on controlling the entire customer journey.

Walmart’s vast network of more than 4,600 US stores gives it a structural advantage that is difficult for Amazon to replicate quickly. Physical presence enables last-mile delivery efficiencies, click-and-collect services, and a tangible brand familiarity that remains powerful despite the digital revolution.

Meanwhile, Amazon’s relentless innovation continues to challenge Walmart’s technological ambitions. The competition is a mutual escalation that benefits consumers through lower prices, faster delivery, and improved service quality.

The Subscription Economy Comes to the Supermarket

One of Walmart’s most strategically important initiatives has been the expansion of its Walmart+ membership programme. With approximately 28 million subscribers, the service aims to replicate the loyalty and recurring revenue model that has made Amazon Prime such a formidable force.

Subscription services transform retail economics. They create predictable revenue streams, deepen customer engagement, and encourage behavioural loyalty. Once consumers are embedded within an ecosystem offering free delivery, exclusive discounts, and bundled services, switching becomes psychologically and economically inconvenient.

The success of Walmart+ suggests that the retailer has learned from the most powerful lesson of modern technology platforms: long-term value lies in relationships, not merely in transactions.

From an investor perspective, subscription revenue commands higher valuation multiples because it offers stability and growth visibility. Walmart’s premium valuation — trading at over 43 times forward earnings — reflects this transformation from a cyclical retailer into a hybrid technology-enabled platform.

The Valuation Question: Has Walmart Become Too Techie for Its Own Good?

Walmart’s lofty valuation naturally invites scepticism. Retail has historically been a low-margin, capital-intensive industry vulnerable to economic cycles and changing consumer behaviour. Assigning technology-style multiples to a retailer might appear optimistic, if not faintly reckless.

However, valuation ultimately reflects expectations about future growth, profitability, and competitive positioning. Investors increasingly view Walmart as a logistics and data platform with retail as its front-end interface and less of a traditional retailer.

The company’s advertising business is a particularly intriguing growth driver. By monetising customer data and offering targeted marketing opportunities to suppliers, Walmart is replicating Amazon’s highly profitable advertising model. Advertising carries significantly higher margins than product sales, which could materially enhance Walmart’s overall profitability profile.

Nevertheless, risks remain. Maintaining technological competitiveness requires enormous ongoing investment. Retail margins remain thin, and economic downturns can quickly shift consumer spending patterns. Walmart must also balance its low-price identity with its aspiration to become a high-margin digital platform — a delicate strategic balancing act.

The Macro Tailwind: Bargains in an Age of Economic Anxiety

Walmart’s success is not occurring in isolation from broader economic forces. The company’s positioning as a value retailer makes it particularly attractive during periods of consumer uncertainty. As inflationary pressures linger and consumer confidence wavers, shoppers often gravitate towards retailers offering price leadership.

This phenomenon is historically consistent. During economic downturns or periods of financial stress, discount retailers frequently outperform premium brands. Walmart’s ability to capture both lower-income and increasingly middle-income shoppers has strengthened its resilience.

Interestingly, economic anxiety has coincided with technological adoption. Consumers are not merely seeking lower prices; they are seeking convenience and efficiency. Walmart’s integration of digital tools into its value proposition aligns neatly with these evolving preferences.

The Blurring Lines Between Retail, Technology, and Infrastructure

Walmart’s trillion-dollar milestone reflects a broader structural shift across industries. Technology is no longer a sector; it is an enabling layer across the entire economy. Retailers, banks, healthcare providers, and manufacturers are increasingly valued based on their technological capabilities rather than their traditional business classifications.

Walmart’s transformation illustrates how scale businesses can leverage technology to defend and expand competitive advantages. The company’s physical infrastructure, once perceived as a legacy burden, has become a strategic asset when combined with data analytics and automated fulfilment.

This evolution mirrors developments in other industries where incumbents successfully reinvent themselves through technological integration. The lesson for investors is increasingly clear: disruption does not always destroy incumbents. Sometimes, it forces them to evolve into something stronger.

The Consumer: The Quiet Winner

Perhaps the most understated outcome of the Walmart-Amazon rivalry is the extraordinary benefit delivered to consumers. Competitive intensity between these giants has produced faster delivery times, broader product selection, and aggressive pricing strategies.

From an economic standpoint, this competition resembles a deflationary force within retail. Efficiency gains, logistics innovation, and scale economics continue to suppress price inflation for consumer goods.

Consumers may not consciously appreciate the technological sophistication underpinning their shopping experience. They simply notice that groceries arrive quicker, prices remain competitive, and product choices appear endless. Behind the scenes, however, an intricate dance of algorithms, warehouse robotics, and predictive analytics is working tirelessly to fulfil those expectations.

The Nasdaq Moment: Symbolism Matters

Walmart’s recent inclusion in the Nasdaq 100 index represents more than a bureaucratic reshuffling of stock market categories. It symbolises a shift in how investors perceive the company’s identity.

Historically associated with technology companies, Nasdaq inclusion signals recognition of Walmart’s transformation into a digitally driven enterprise. It also reflects the broader market trend of rewarding companies capable of blending operational scale with technological innovation.

The Broader Investment Implications

Walmart’s trillion-dollar valuation raises intriguing questions for investors. Should traditional sector classifications be reconsidered? Are logistics networks the new cloud infrastructure? And perhaps most importantly, which other “boring” industries might secretly be undergoing technological metamorphosis?

Investors often focus disproportionately on emerging disruptors while overlooking incumbents quietly reinventing themselves. Walmart demonstrates that sustained competitive advantage can arise from combining legacy strengths with modern technology.

The company’s journey also underscores the enduring value of execution. While technological innovation captures headlines, consistent operational excellence often determines long-term success. Walmart’s ability to integrate technology without abandoning its core identity has been central to its transformation.

The Trolley Rolls Into the Future

Walmart’s entry into the trillion-dollar club is not merely a corporate milestone. It represents a subtle but profound redefinition of what constitutes a technology company. The future of commerce will not be dictated solely by software platforms or physical retailers, but by hybrid enterprises capable of mastering both domains.

In a sense, Walmart’s achievement is a triumph of pragmatic evolution over glamorous disruption. The company did not attempt to reinvent itself overnight. Instead, it gradually layered technological capability onto an already formidable operational foundation.

The image of a discount retailer standing alongside Silicon Valley titans may still feel faintly surreal. Yet, it also feels oddly appropriate. After all, in an era where algorithms decide what we buy and warehouses resemble science fiction laboratories, the humble shopping trolley has become just another interface in the digital economy.

And if Walmart’s rise teaches us anything, it is that sometimes the future of technology does not arrive in a sleek gadget or futuristic headset. Occasionally, it arrives quietly — in aisle seven, next to the discounted breakfast cereal, and rather conveniently priced.

Subscribe now

My Contrarian Take

Walmart’s trillion‑dollar valuation is an undeniably impressive story, but it is not quite the bullet‑proof, tech‑platform‑fairy‑tale the market seems to be pricing it as. There is a perfectly reasonable set of counter‑arguments that suggest this trolley‑pulling giant might be trading a bit more like a Silicon Valley darling than it really is.

The valuation looks a little too excited for a grocer

Let’s start with the numbers. The stock is sailing along at a forward P/E in the 40–45× range, which is closer to the pricing of a high‑growth software outfit than a discount‑grocery‑heavy retailer. That’s effectively a bet that Walmart can keep growing earnings at a healthy clip while quietly turning its low‑margin business into something more platform‑like.

The problem is that if the ad‑driven growth slows, the membership ramp pauses, or the consumer spends more cautiously, the Street could just as easily slap a lower retail multiple on the same business. In that case, the share price might not be as thrilled as it is today.

Thin‑margin foundations versus boutique‑style multiples

Walmart’s net margins are still around 2.5–3%, with gross margins hovering in the low‑to‑mid‑20% range, even after all that scale and logistics wizardry. That is a world away from the kind of businesses that happily trade at 40× earnings, where gross margins often sit above 70% and operating margins are comfortably in the 20–30% bracket.

Put another way: Walmart’s valuation is assuming it will morph into a tech‑style platform, yet most of its revenue still comes from selling baked beans and batteries at minimal mark‑ups. If the higher‑margin streams — advertising, membership, and data — don’t grow faster than the Street hopes, the premium could start to look a bit like wishful thinking.

Tech‑platform or glorified in‑house toolkit?

Walmart has been busily investing in AI, automation, and data systems, but a lot of that technology is being used inside the business, not sold as a standalone product. This makes it more of an operational accelerator than a genuine external platform like AWS or Azure, which sit outside the core product and sell directly to other companies.

Walmart Connect is growing at a brisk low‑30%+ year‑on‑year pace, but it is still tethered to Walmart’s own shopper base and sales. It’s not an open ecosystem that can monetise traffic from anywhere in the economy; it is a retail‑media arm built on Walmart’s own store, online and otherwise. That’s a useful asset, but not quite the same as a scalable, horizontally‑oriented tech platform.

Heavy tech spend that could eat into margins

All this tech‑driven ambition comes with a hefty price tag. Walmart’s capex runs in the low‑tens of billions of dollars annually, pouring money into automation, AI, and omnichannel infrastructure. That spending is necessary to keep costs under control and e‑commerce competitive, but it also sits right on top of an already thin‑margin business.

If the returns on those investments are slower or smaller than expected — or if adoption of AI‑driven tools disappoints—investors might start to wonder whether Walmart is funding a future platform or a very expensive overhaul of its trolley‑pushing operations.

Macro and regulatory risks are still very real

Walmart’s crown jewel is also its Achilles’ heel: it is the shop of choice for low‑ and middle‑income shoppers who care deeply about price. That’s a defensive advantage in bad times, but it also exposes the company to policies that hit their wallets, like changes to SNAP benefits or food‑stamp programmes.

Estimates suggest a pause or cut in SNAP benefits could knock out roughly $2 billion in annual sales for Walmart, which is not trivial. Add on tariff and trade risks on imported goods, and you have a business that is still very much at the mercy of both politics and consumer nervousness. That’s not how classic “tech” stocks usually behave in the market’s imagination.

What investors should keep in mind when looking at Walmart now

Given all this, investors should treat the trillion‑dollar story as a promising hypothesis, not a done deal. To get a clearer picture, they should focus on a few key questions:

Is the valuation still in love with the story?

Keep an eye on the P/E spread between Walmart and the broader market, as well as pure‑tech names. If earnings growth slows or guidance gets modest, the Street could quickly swap out the “tech‑platform” narrative for a more modest retail‑style multiple.

It also helps to watch how quickly Walmart+ membership and Walmart Connect move the needle on both revenue and margins. If high‑margin streams grow but not at an accelerating pace, the justification for a perpetual‑growth premium starts to thin out.

How convincing is the platform‑style story?

Look at whether Walmart can turn its data and AI into something more than internal tools. Genuine platform‑style businesses usually sell their infrastructure, software, or services to third parties, not just use them to keep their own shelves full. If Walmart’s tech remains a closed ecosystem, the stock may be better thought of as a tech‑enabled retailer than a true platform.

At the same time, assess how Walmart Connect scales relative to Amazon’s ads business. Rapid growth is nice, but if it plateaus because it is constrained by Walmart’s own sales and traffic, the fair value will likely sit somewhere between a grocer and a full‑blown platform.

Is the tech‑driven capex delivering real returns?

Track whether Walmart’s automation and AI investments are clearly improving productivity, lowering costs, and lifting returns on capital. If the capex is just a recurring drain on margins without visible efficiency gains, the hybrid‑tech narrative will start to look a bit wobbly.

Comparing Walmart’s debt and leverage ratios with peers and the broader sector is also useful. The company needs to keep its balance sheet healthy enough to fund digital transformation without sacrificing credit quality, especially as tariffs and inflation pressures keep nibbling at margins.

How resilient is its position in a twitchy economy?

Watch how Walmart holds up in tougher consumer‑spending environments. Strong footfall and sales when households are tightening their belts support the idea that Walmart is a defensive, every‑day‑low‑price safe haven. But if traffic weakens or price‑sensitive shoppers start trading down further, the low‑end exposure could start to look more like a risk than a moat.

Keep an eye on competition from Amazon, Costco, and discount rivals, as well as any new threats from dollar stores or online‑only players. If the industry slides into a price war, Walmart’s famous low‑price advantage could morph into a margin‑squeeze that undercuts the platform‑oriented story.

Is management balancing the act well?

Finally, ask whether Walmart’s leadership is actually executing a disciplined balancing act between tech‑enabled growth and core retail discipline. The company must keep its prices low while gradually shifting more of its revenue toward higher‑margin services; if management over‑commits to tech‑style ambitions at the expense of its everyday‑low‑price promise, the whole story could start to unravel.

Watch how guidance evolves over time, especially anything the company says about margin targets, reinvestment plans, and the pace of AI‑driven change. If executives start hinting that the tech‑driven path is slower, more expensive, or less profitable than expected, the Street will likely recalibrate its view of where this discount‑retailer‑cum‑platform really belongs in the valuation league table.

In short, Walmart’s trillion‑dollar moment is a triumph of pragmatic, tech‑assisted evolution, but it sits on a foundation of low margins, heavy capex, and macro risks. Investors should treat the current valuation as a bet on continued high‑quality growth, margin‑mix improvement, and very clean execution — not as a one‑way ticket to Silicon‑Valley‑style multiples that never come down for a reality check.

Share