Bank of England and WTO Warn: Iran War Threatens the Fragile Economics of the AI Boom

The war around Iran could hit more than gasoline and oil prices. It could hit the economics of the artificial intelligence boom itself: the industry depends critically on cheap energy, builds its infrastructure with borrowed money, and has not yet proven it can quickly pay back its gigantic investments.

The logic is straightforward. If the conflict drags on and disruptions in the Strait of Hormuz keep pushing oil prices up, it will not only be fuel that gets more expensive, but also electricity, logistics and component manufacturing. That is especially sensitive for AI, because training models, running clouds and operating data centers all require enormous amounts of energy. Even if the US, as an oil exporter, weathers such a shock better than many other countries, American tech companies still cannot fully insulate themselves from a global rise in costs. And a significant share of the investment frenzy around generative AI is concentrated in the US.

There is an additional risk: the sector's financial model looks fragile. In its latest review of risks to the financial system, the Bank of England noted that even before the escalation around Iran, investors had begun to ask whether the expected returns from very large AI investments would pay off. Against that backdrop, any new cost can add pressure on share prices. The regulator separately pointed out that the war could heighten anxiety because of the energy intensity of supply chains for key components and of data center operations. In other words, the market already doubted how quickly AI would pay for itself, and costlier energy deepens those doubts.

The same concerns are being voiced at the level of global trade. WTO Chief Economist Robert Staiger warned that a prolonged period of high energy prices could noticeably slow AI investment. That matters not only for tech companies but for the entire US economy: according to the WTO, 70% of the growth in American investment over the first three quarters of last year came from goods and infrastructure tied to artificial intelligence. If that flow begins to slow, the blow will reach beyond a single industry. It is no longer only about demand for chips and servers, but also about construction, capacity leasing, lending and stock market expectations.

Another problem is the financing structure. In a recent review, the law firm Quinn Emanuel showed how skewed the sector's basic numbers are: the industry's revenue last year was about $60 billion, while capital expenditure reached roughly $400 billion. The gap is covered by debt and complex financial arrangements.

The biggest players and infrastructure providers such as CoreWeave borrow colossal sums to build new data centers quickly. Part of the obligations moves off balance sheets into special structures that own the facilities, receive the future rental stream and borrow against it. Such debts can then be pooled, split up and resold to pension funds and asset managers.

Over the past two years, according to the same analysts, about $120 billion of data center debt has already been moved off balance sheets.

This structure makes the sector especially vulnerable to an external shock. If energy stays expensive for long, operating costs rise, investors grow more cautious about new loans, and consumer demand and interest rates behave worse than expected. In a tightly interconnected AI ecosystem, trouble at one node spreads easily — through capacity tenants, cloud providers, lenders and holders of securitized debt. That is why even a moderate rise in energy costs could trigger a repricing of the entire AI boom.

The main takeaway is that the debate over AI now hinges not only on model quality or adoption speed, but also on the old economy: oil, electricity, the cost of capital and the transparency of debt. As long as the industry's revenue lags far behind the scale of investment, any prolonged energy shock can turn a story about technological growth into one about financial vulnerability. For the market, this is a signal: AI remains not only a big bet on the future, but also one of the sectors most sensitive to expensive energy here and now.