Fed’s Bowman warns of a sharp AI-driven market correction as AI debt tops $800 billion

On Tuesday, the Federal Reserve’s top banking supervisor put the AI boom’s financing on the table — and the numbers she cited are the starkest a senior U.S. official has attached to the buildout so far. In a speech in Washington, Vice Chair for Supervision Michelle Bowman said debt tied to artificial intelligence is on track to pass $800 billion this year, while the revenues meant to service it still don’t cover the cost of building and running the systems. If that gap doesn’t close, she warned, the market could correct sharply.

Her remarks, delivered to the Council on Foreign Relations, frame the correction as a risk, not a prediction. But the Fed’s supervision chief detailing the financing structures behind the boom — and noting that markets have already begun repricing AI bets — makes this the most direct official warning about the AI trade yet.

The numbers behind the warning#

The headline figure is the debt: AI-related borrowing across corporate bonds, private credit, and project finance is on track to exceed $800 billion in 2026. What makes it bigger is what sits beside it — long-term leases, contract commitments, and residual-value guarantees whose potential payments she estimated at more than three times the debt itself. The headline borrowing is less than a quarter of the obligations the buildout has stacked up.

The revenue side is the problem. Bowman said AI revenues are rising quickly but do not yet cover the cost of maintaining existing models and developing new ones, or of building the data centers to run them. As spending has surged, the big builders have increasingly turned to credit and equity markets, while financing costs have risen notably as investors grow warier.

The buildout is also feeding through to prices: memory chips, semiconductors, and computers have risen sharply this year, in some categories by double digits, as data-center demand outruns supply. AI-related demand accounted for roughly a quarter of above-target core inflation over the year to September, by her account, alongside tariffs and energy.

Circular financing, in the Fed’s own words#

The part of the speech that will travel furthest concerns how the boom is financed, not how big it is. Bowman said the largest tech firms have taken on off-balance-sheet liabilities that smaller, riskier companies pledge as collateral to borrow for data centers and chips. At the same time, hyperscalers and chip designers have been buying equity stakes in smaller companies that purchase their products — “creating the appearance of circular financing.”

That is not an accusation of wrongdoing; it is a structure that makes risk harder to see. Bowman said the biggest U.S. banks have some AI-financing exposure, checked in annual stress tests, and that insurers, pension funds, private credit funds, and mutual funds are exposed too. Most are well capitalized — but the complex structures make counterparty risk hard to assess.

The risk is concentrated in a small number of large firms, she said: the AI labs and the hyperscalers. If revenues fall short of sharply rising costs, the sequence she sketched is plain — a sharp correction in AI companies’ share prices, higher financing costs, reduced access to credit — and a shock that would slow the buildout and soften both economic activity and inflation.

Why the warning lands now#

The timing is hard to miss. Bowman said markets may already be moving: in recent weeks, equity valuations dropped, corporate bond spreads widened, and credit default swaps rose for some companies in the AI buildout. Financial conditions have tightened, she noted, as investors reconsider the returns on AI projects funded with outside money.

The warning also lands in a crowded week for the AI trade. Anthropic is scheduled to price its initial public offering on November 18, according to Semafor — the first listing of a frontier lab, and a live test of what investors will pay for AI revenues still catching up with AI spending. It comes the same day Nvidia’s board announced Jensen Huang’s resignation, after a week of board investigations.

Bowman is not a critic of the technology. She credited AI investment with boosting growth and productivity, said business investment is growing at a pace reminiscent of the 1990s tech boom, and noted that about a quarter of businesses and a tenth of workers now use AI daily. Her point is narrower: the financing has outrun the revenue, and the structures hiding the gap make the system harder to read.

What to watch#

First, the December meeting of the Federal Open Market Committee. Bowman backed holding rates at 3.75 to 4 percent in October and stressed policy is “not on a preset course” — the October inflation report and November jobs data land first, and she said clearer AI financing conditions would raise her confidence. Second, whether AI revenues start closing the gap with costs — the variable the whole warning turns on. Third, the credit markets: spreads, default swaps, financing costs. And fourth, Anthropic’s IPO pricing next week, the first market verdict on a frontier lab since the warning.

Sources: Federal Reserve — “The U.S. Economy and Considerations for Monetary Policy,” remarks by Vice Chair for Supervision Michelle W. Bowman (Nov. 10, 2026); Bloomberg Law — Fed’s Bowman Warns of Stock Correction If AI Bets Don’t Pan Out (Nov. 10, 2026); JBizNews — AI-Related Debt on Track to Exceed $800 Billion This Year, Fed’s Bowman Says (Nov. 10, 2026). Prior coverage: Anthropic prices its IPO at $70–$78 a share, targeting up to $1.4 trillion; Jensen Huang resigns as Nvidia CEO; CFO Colette Kress named interim chief.