Compute-for-Capital: End of The Offshored Empire (Part 2)
The old bargain was goods-for-debt. The new one is compute-for-capital. America is asking the Gulf to pay for it.
Grüezi!
In the dying days of the Biden presidency, America wrote its worldview into compute.
The AI Diffusion Rule, finalised on 15 January 2025, divided the world by access to advanced chips.
The inner circle — most NATO members, plus Japan, South Korea and Taiwan — kept broad access. China, Russia and Iran were outside the system.
Gulf states sat in the middle: rich, useful, strategically important, but not fully trusted with the crown jewels.
The rule was due to take effect on 15 May. It never did. On 13 May, Commerce scrapped it.
That did not mean America had stopped controlling chips. Control changed form. Biden wanted rules. Trump wanted deals.
The week Commerce killed the Diffusion Rule, Saudi Arabia unveiled HUMAIN — a new AI company owned by the kingdom’s sovereign wealth fund and chaired personally by Mohammed bin Salman.
HUMAIN became the vehicle for planned purchases of Nvidia’s top-end GPUs, an AWS AI Zone and an AMD partnership. The UAE was promised a 5-gigawatt AI campus near Abu Dhabi, linked to a $1.4 trillion investment framework announced two months earlier.
The standard reading of Trump’s second term is that the postwar order is being dismantled — tariffs against allies, retreat from Ukraine, self-enrichment replacing strategy.
Not entirely wrong. Not the whole story.
Something is being built as well as broken.
Part 1 described the bargain America is now dismantling: an offshore industrial empire in which East Asian states manufactured, exported, accumulated dollars and recycled them into American finance.
Washington got cheap goods, low long rates, strategic alignment and deniability. Tokyo, Seoul, Taipei, Singapore and eventually Beijing got market access, technology transfer, security protection and permission to run industrial systems America could not have built at home.
That was the old bargain.
Part 2 is about the substitute. America is trying to rebuild industrial capacity at home while asking the Gulf to supply the capital, land and electricity that East Asia once supplied indirectly through goods, surpluses and Treasury purchases.
The old bargain was goods-for-debt. The new bargain is compute-for-capital.
1. The Rule That Never Took Effect
The Diffusion Rule was the end point of “small yard, high fence” — the doctrine Jake Sullivan laid out in April 2023.
Draw a narrow perimeter around frontier technologies. Defend it at all costs. Let trade in everything else continue.
The rule sorted the world into three levels of trust. At the top were states inside the technological perimeter. At the bottom were states to be denied.
In the middle sat the awkward category: countries America wanted onside, but not as owners of frontier compute. Saudi Arabia. The UAE. India. Singapore. Malaysia. Countries with money, ambition, useful geography and varying degrees of Chinese exposure.
The rule was meant to prevent advanced AI chips leaking sideways into China through nominally friendly states. It was also meant to prevent a new class of compute powers emerging outside the American alliance system.
Trump kept the premise but changed the method.
The old bargain was goods-for-debt.
The new bargain is compute-for-capital.
Chips were still controlled. Nvidia could not simply ship whatever it wanted, wherever it wanted. Gulf buyers still needed Washington’s approval. Data centres still sat inside national-security review.
What changed was that access was no longer governed by rules and tiers. It was governed by deals.
A country that wanted chips now had to offer Trump something: capital, energy, alignment, China divestment, American procurement, diplomatic support.
Biden had tried to force compute access into a rules-based order.
Trump made it pay-to-play.
2. Why Repatriation Is Hard
The old offshore bargain worked because Asia supplied America with goods and recycled the proceeds.
Japan ran surpluses and bought Treasuries. Korea ran surpluses and bought Treasuries. Taiwan accumulated reserves. China took the model to continental scale.
American consumers bought the goods. American borrowers got the capital back. American politicians called it globalisation.
The bargain was not cost-free. It hollowed out some American factory jobs. It handed over some technology. It created corporate rivals like Huawei who would go to America’s wars to sell its clients telecoms systems under gunfire.
It turned China into the world’s greatest manufacturing power and one of the largest foreign creditors of the American state.
Above all, it solved Washington’s political problem. America wanted the benefits of industrial policy — strategic production, disciplined capital allocation, export platforms and technological upgrading inside its alliance system — without creating the kind of planning state Congress and American business would have resisted at home.
So the planning happened offshore. The bureaucracies were in Tokyo, Seoul and Taipei. The goods came back cheap. The dollars came back as capital. The costs fell on American manufacturing regions, but the bargain itself remained hard to see.
Wall Street didn’t invent the bargain, but it found itself one of the biggest beneficiaries once Asian surpluses began flowing through the dollar system at scale.
That bargain is now being hunted down.
China is being cut out. Japan, South Korea and Taiwan remain inside the perimeter, but the terms have changed. Their firms are now expected to build capacity in America while limiting exposure to China.
TSMC is no longer just Taiwan’s strategic asset. It is being pulled into Arizona. Samsung is no longer just Korea’s champion. It is being pulled into Texas.
The old offshore system is being brought home. But repatriation creates a problem.
Factories cost money. Data centres cost more. AI infrastructure consumes power at a scale America’s ageing grids cannot deliver quickly. Permitting is slow. Transmission is slow. Nuclear is slow. Gas turbines are backlogged. Public finance is picked apart politically.
America wants to rebuild at home. But it does not want to pay for it alone.
That’s where the Gulf comes in.
East Asia supplied cheap manufactured goods and recycled the dollars into American debt. The Gulf is being asked for something different — capital, land, electricity and sovereign speed.
Not to become the next China.
To help finance and host the supporting infrastructure for the next American technological order.
3. Europe: Inside, Outside
Europe was inside the American order, but outside the offshore bargain that made East Asia rich.
Western Europe got reconstruction, security and access to markets. It did not get the same role in America’s industrial machine. The Marshall Plan revived damaged industrial economies; it did not create export-surplus developmental states.
NATO gave Europe security; it did not make Europe strategically autonomous. Europe sat in the senior institutional tier of the postwar order, but it never became the kind of offshore platform America built in East Asia. NATO’s nuclear, command and control capabilities stayed in Washington.
Germany was a partial exception.
Inside the euro, it ran something close to the East Asian playbook — wage restraint after the Hartz reforms, an effectively undervalued currency, and a manufacturing surplus machine. But it ran that inside Europe, not within a larger American bargain.
For two decades, German industry treated China as the answer to Europe’s low-growth problem. Sell cars, chemicals and machinery into the world’s fastest-growing market. Treat Chinese industrial upgrading as an opportunity rather than a threat. Assume that technological distance would hold.
It did not.
Volkswagen spent a quarter-century as China’s largest foreign carmaker. Then Chinese firms caught up. BYD overtook it in 2024. Geely followed in 2025. By then German brands held just 5% of the Chinese electric-vehicle market — the segment meant to be the future.
For every Porsche Taycan sold in China in 2024, Xiaomi sold 74 of its rival SU7. Volkswagen’s China joint-venture profits fell from over €3 billion in 2022 to less than €1 billion two years later. Between 2016 and 2023, Germany supplied 58% of all EU foreign direct investment in China. Europe’s China policy had been a German car and chemicals policy with no plan B.
The second break was Russia.
Gerhard Schröder signed Nord Stream 1 in September 2005, days before leaving office, then took a paid position with Gazprom-controlled Nord Stream AG that December. Russia invaded Georgia in 2008 and took two pieces of it. Germany kept buying.
In 2011, Angela Merkel decided to phase out nuclear power, closing the only large-scale alternative to imported gas. Russia annexed Crimea in 2014. The next year, Merkel approved Nord Stream 2. By 2021, Germany was importing more than half its natural gas from Russia.
Both legs failed at the same time.
The China model broke when Chinese firms became competitors. The Russia model broke when Moscow invaded Ukraine in February 2022.
The sanctions that followed — SWIFT cuts, €300 billion of Russian central-bank assets frozen at Euroclear, energy restrictions — only worked because they sat inside a dollar-centred financial system underwritten by Washington. Brussels could sanction Moscow. The measures that mattered most still relied on American payments plumbing.
On defence, Europe deferred too.
The warning came thirty years earlier. Three weeks into the Yugoslav wars, Luxembourg’s foreign minister declared: “This is the hour of Europe, not the hour of America.”
Four years, 100,000 dead and one Srebrenica massacre later, the war ended not in Brussels, Paris or Berlin, but at Wright-Patterson Air Force Base in Ohio, after American airpower and American diplomacy imposed a settlement.
Europe could not stop a war in the Balkans without American help. Ukraine has shown how much has changed — and how much has stayed the same.
Now Europe is stuck. The gas it used to get from Russia it now buys dearer from America. The batteries, solar panels and electric cars it needs to electrify are mostly built in China. The nuclear and defence capabilities it cannot afford to replace are mainly American, and Washington is openly transactional about the cost of keeping them.
Mario Draghi’s Future of European Competitiveness, published in September 2024, put a price on buying out of that trap: €750–800 billion a year, a higher proportion of GDP than the Marshall Plan absorbed.
Europe’s politicians and publics ignored its implications. Europe remained inside the old American order.
It was not central to the new bargain.
4. The Redoubt America Built
America did build one high-tech client in the Middle East: Israel.
Israel received the closest regional version of the East Asian package: preferential market access, defence and civilian technology transfer, procurement support, and extraordinary capital backing.
The 1985 US-Israel Free Trade Agreement was the first FTA America ever signed, three years before Canada and nine before NAFTA. Foreign Military Financing was structured so that most of it had to be spent on US weapons, but with enough room for Israel’s own defence industry to build an export base while subsidising Lockheed and Raytheon.
Intel’s first overseas R&D centre opened in Haifa in 1974. Its first overseas fab opened in Jerusalem in 1981.
By the mid-2020s, Israel had the highest R&D intensity in the OECD, above 6% of GDP. High-tech accounted for over half of exports. Defence sales reached roughly $15 billion a year. Israel had run a current-account surplus since 2003.
But Israel was never a production platform. It was a redoubt.
From the 1970s onwards, it sold defence technology to clients Washington could not always supply directly — apartheid South Africa, Argentina under the junta, Pinochet’s Chile, Taiwan after de-recognition, Iran under the Shah and, covertly, after.
It ran its own strategic agenda when the alliance allowed and sometimes when it did not: the Lavi fighter programme, the Phalcon AWACS deal with China that Washington forced it to cancel in 2000, intelligence cooperation outside US channels.
A junior partner with its own foreign policy, operating with American financial weapons and protection.
The Iran campaign became one prototype for the financial-warfare state that Washington later turned on Russia, North Korea, Venezuela and China. Edward Fishman’s Chokepoints traces the decisive shift to Stuart Levey’s Treasury team in the mid-2000s, when Levey realised that access to the dollar system could be used to force foreign banks to choose between Iran and the United States.
Israel didn’t build that weapon. But its supporters advocated for it, and its strategic priority — keeping Iran under maximum pressure — helped keep it pointed at Tehran long enough for it to do damage.
But for Israel, sanctions were necessary, but not sufficient.
By 2024, it had concluded that economic measures alone could not contain Iran. In June 2025, it struck Iran’s nuclear facilities. America joined on day nine. In February 2026, the US and Israel struck again — together this time — a 39-day air war currently paused by an April ceasefire brokered by Pakistan.
America had been persuaded that this would be fast and furious. It was not prepared for a prolonged, disruptive conflict. By the time of the second Iran attack, according to a CSIS assessment, the United States had burned through roughly half its Tomahawks, half its THAAD interceptors, half its Patriot interceptors, and almost half of its Precision Strike Missile stockpile.
Replacement munitions had to come from somewhere.
South Korea sent 330,000 artillery shells to backfill American magazines already emptied by Ukraine. Hanwha is now building a $1 billion propellant plant on American soil because the United States cannot ramp up production fast enough itself.
To arm its Middle Eastern ally, America had to rely on the very Asian countries it was simultaneously putting the squeeze on.
Pull one thread and another one unravels.
Israel never became the centre of a productive regional network. It became a highly capable exclave whose conflicts increasingly consumed the diplomatic capital and military inventories of the power that built it.
The Gulf bargain is different.
Saudi Arabia and the UAE are not small military-technology clients. They are capital states, with balance sheets, land, cheap energy, centralised decision-making and rulers willing to transact at scale.
America underwrote Israel as a Middle Eastern high-tech redoubt.
The Gulf is now being asked to underwrite the next American industrial revolution. And it wants in.
5. The Gulf Substitute
The major Arab sovereign funds hold trillions between them. ADIA, PIF, Mubadala, KIA and Qatar’s investment vehicles are no longer passive wells of oil rents. They are strategic instruments deployed with intent.
But sovereign wealth is not the same as spare cash.
The Gulf’s current-account surpluses are thinner. Saudi Arabia is borrowing more. Construction, defence, domestic subsidies, sports, tourism, logistics, mining, aviation, renewables and AI all compete for capital.
The UAE is more financially flexible, but its own ambitions are not small. Kuwait has money but weaker political execution. Qatar has gas and diplomatic clout, but less capacity for scale.
The money exists. The question is whether it can be mobilised fast enough.
The second complication is rivalry.
The Gulf is not one buyer. Rivalry is a feature, not a footnote. Riyadh wants HUMAIN to be the Arab world’s compute champion. Abu Dhabi already has G42, an AWS partnership and a 5-gigawatt campus.
Both want the same Nvidia allocations, the same hyperscaler partnerships, the same status as Washington’s preferred regional interlocutor. Qatar is its own pole. The politics of who gets chips, who hosts data centres, who owns operating companies and who becomes Washington’s preferred partner are not clear cut.
The third problem is China.
The Gulf wants American chips without becoming an American dependency. It wants access to Chinese markets, Chinese engineers, Chinese contractors and Chinese technology where useful.
Washington wants that access narrowed. It wants Gulf capital, but it does not want China inside the bargain.
That is why the Microsoft-G42 deal of April 2024 set a template. Microsoft committed $1.5 billion to G42 and a senior American executive joined its board. The condition was clear: G42 had to distance itself from China, including breaking ties with Huawei and ByteDance.
A Gulf national-security-adjacent technology firm aligns structurally against China. An American hyperscaler takes a strategic position. Washington permits access to advanced chips and cloud infrastructure. The Gulf supplies capital, political backing and physical sites. America supplies the software stack, the licences and the strategic seal of approval.
Trump’s deals scaled that template.
Saudi Arabia announced $600 billion in commitments to American companies. HUMAIN became the vehicle through which Riyadh would enter the AI infrastructure race. Nvidia announced a first phase of 18,000 GB300 Grace Blackwell systems. Saudi ambitions stretched to several hundred thousand chips over time. AWS announced an AI Zone in Riyadh. AMD announced a partnership. The UAE-US AI campus near Abu Dhabi was announced at 5 gigawatts.
The numbers were impressive. But announcements are not deliveries.
Chips still have to clear Washington’s export-control process. Data centres can be announced faster than turbines can be installed, grids expanded, cooling secured and licences approved.
The Gulf can break ground faster than America can make up its mind.
6. Compute-for-Capital
The new bargain is not oil-for-security.
That was the old Gulf arrangement: the US Fifth Fleet protected the sea lanes, the Gulf priced oil in dollars, recycled rents through American banks, bought American weapons and held American assets.
That bargain still exists. But it is no longer enough.
Oil is no longer the strategic resource frontier. Compute is.
The Gulf gets access to chips, cloud infrastructure, American AI firms, model partnerships and a seat at the technological centre of the compute economy.
America gets sovereign capital, energy infrastructure, land for data centres, and a way to finance industrial renewal without putting the full bill on the federal balance sheet.
The richest country in the world is asking hydrocarbon monarchies to finance part of its technological future.
That sounds strange. But America has done versions of it before.
East Asia bankrolled American consumption and deficits by exporting goods and buying Treasuries. The Gulf enabled American financial depth by recycling oil rents through banks, bonds, arms purchases and asset managers.
What’s different now is the object.
The money isn’t just going into Treasuries. It is going into AI infrastructure, chip supply chains, power generation, hyperscale campuses, sovereign AI companies and strategic stakes in the firms that sit between compute and state power.
And BlackRock, Microsoft, GIP and MGX are not random counterparties. They are the shape of the bargain: asset management, cloud infrastructure, physical infrastructure and Gulf capital, bound together as co-investors in a system none of them could finance alone.
Nvidia remains the bottleneck.
Without the chips, there is no bargain. That gives Washington the controlling position. It can promise access, delay access, condition access, threaten access and withdraw access.
The Gulf can supply money.
But America provides permission.
This is why scrapping the Diffusion Rule did not end the hierarchy.
It just made it negotiable.
7. The Bargain in Plain Sight
The old bargain lasted because it was barely visible.
American voters saw cheap goods, low inflation, low mortgage rates and globalisation. They did not see exchange-rate management, procurement flows, technology licences, Asian reserve accumulation, security guarantees or the quiet transfer of industrial capacity.
The costs were dispersed and delayed. The benefits were immediate and anonymous.
A microwave got cheaper. A mortgage rate stayed lower. A pension fund rose. A factory closed. A town declined. A supply chain moved. A Treasury auction cleared.
No one had to call it imperial policy.
That was the political genius of the old system. Its victims could not easily identify the bargain that had hurt them.
The substitute is different.
It is visible from the start.
The old bargain hid inside the operating system of globalisation. The new one appears as chip allocations, sovereign investment pledges, hyperscale campuses, export-control waivers and photographs with Gulf rulers.
That makes it easier for Trump to sell.
It also makes it easier to attack, renegotiate and break.
The bargain depends on Trump signing chip allocations, Gulf rulers deploying capital, American agencies accepting the security risk, Nvidia’s supply chain, the availability of power, oil prices, Saudi willingness to remain a swing producer, and the UAE staying close enough to Washington without cutting itself off from China.
It also depends on the Middle East not plunging into conflict.
For decades, the Gulf’s accumulation of wealth depended on an American security order that kept oil moving and suppressed the region’s most disruptive powers. Iran and Iraq were periodically sanctioned, invaded, contained, isolated or degraded. Saudi Arabia, the UAE, Kuwait and Qatar picked up the profits of a system whose security was mostly managed by the US.
Now those same states are being asked to become strategic co-investors in America’s AI future.
If America is asking the Gulf to supply capital and electricity, then Gulf stability is no longer just an energy issue.
It is an AI infrastructure issue.
A war that threatens Gulf power systems, shipping lanes, insurance markets, sovereign balance sheets or political confidence now threatens the substitute bargain itself.
That’s where Israel complicates the picture.
Israel is no longer just a small ally with large influence in Washington. Its wars now risk colliding with America’s wider technological bargain. Iran is not only Israel’s strategic problem. It is one of the places where America’s Gulf bargain can fail.
That doesn’t mean the Gulf will abandon Washington. It has no better security provider.
China can buy oil and build ports. It cannot yet replace the US Navy, the dollar system, American air defence, Nvidia, Microsoft and AWS.
But the Gulf will price American unreliability more aggressively.
More chips. More guarantees. More technology transfer. More freedom to hedge.
That is the new bargain’s weakness.
The Gulf is not Japan in 1945, Korea in 1965, or Taiwan in 1987. It is not defeated, poor, dependent and locked into American tutelage.
It is rich, transactional, politically centralised and fully aware that Washington needs something from it.
The old offshore empire worked because America could grant access to its market and call it free trade.
The new one requires America to grant access to its most sensitive technology and call it partnership.
That is tougher.
The old bargain was bureaucratic, slow and deniable.
The new one is personal, transactional and exposed.
It may still work. The Gulf has the capital. America has the chips. Both sides would like a deal. China is too large to ignore and too dangerous for Washington to leave uncontained. Europe is too slow. East Asia is too exposed. Domestic American capacity cannot be rebuilt quickly enough without outside money.
So the substitute has a certain logic. It does not yet have durability.
And it is not the end of American global power.
But it is the end of that power going unquestioned.
The old bargain hid inside globalisation. The new one sits there in plain sight, in one of the world’s most combustible regions.
The Iran war is the test of whether compute-for-capital can survive the realities of conflict in the region that is supposed to finance it.
Thanks for reading,
Bis bald,
Adrian