Fed Financial Stability Report: AI Is Top Risk for 87% of Contacts
Six months ago half of the Fed's market contacts named AI as a risk. Now almost all of them do, and the report puts new numbers on the debt behind the build-out.
·6 min read
The Fed financial stability report published on Friday, November 13, says artificial intelligence is now the risk its market contacts mention most. 87% of the people surveyed named AI as a salient risk to the US financial system, up from 50% in the spring, CNBC reported from the document. No other risk on the list was cited by much more than half.
The report also contains the central bank's most detailed look so far at how the AI build-out is being paid for. Debt issued to fund advanced computing infrastructure passed $680 billion by October 2026, about five times the $130 billion issued in all of 2023, and the largest technology companies carry more than $3 trillion of leases and contract commitments that do not yet sit on their balance sheets as debt.
- 87%
- Survey contacts citing AI as a risk, up from 50% in spring
- $680B+
- Advanced computing debt issued by October 2026
- 31%
- Share of investment-grade corporate issuance in 2026
- $3T+
- Leases and commitments not yet counted as debt
What the survey found
The numbers come from the Survey of Salient Risks, which staff at the Federal Reserve Bank of New York run twice a year. This round drew responses from 23 contacts at broker-dealers, banks, investment funds and advisory firms during September and October. The Fed stresses that the box summarizes their views, not those of the Board. With a sample that small, each answer moves the result by about four percentage points.
Respondents described three separate worries about AI: high equity valuations that could fall if earnings come in below expectations, cyberattacks made easier or caused by frontier AI models, and what the report calls complex forms of AI-related financing. A correction in risk assets was the second most cited risk. Many respondents tied it to the same source, pointing to stretched valuations, circular financing arrangements and heavy reliance on debt to fund the build-out.
Most cited risks to US financial stability, fall 2026
Percent of 23 market contacts surveyed in September and October
Source: Federal Reserve Financial Stability Report, as reported by CNBC, November 13, 2026
In the spring survey of 20 contacts, AI sat third on the Fed's chart, behind geopolitical risks and an oil shock. Mentions of AI, higher long-term interest rates and cyberattacks rose the most between the two surveys, the report says.
The Fed's count of AI debt: $680 billion and rising
A box titled Leverage in the Data Center Buildout does the arithmetic. Advanced computing borrowing made up less than 6% of investment-grade corporate debt issuance outside the financial sector in 2023. In 2026 it is about 31%. Public debt outstanding tied to advanced computing is estimated at more than $1.1 trillion, and the Fed notes that its figures miss part of the private market, so the true total is harder to measure.
The mix is changing as well, from mostly corporate bonds to a combination of public and private debt, structured credit and project finance. We tracked the same shift from the lenders' side in our analysis of the AI debt bubble.
More than $3 trillion that is not on the balance sheet
Debt is the smaller part of the story. For seven companies, Amazon, Alphabet, Broadcom, Meta, Microsoft, Nvidia and Oracle, the Fed counts leases that have not yet started with an undiscounted value above $1 trillion, plus more than $2 trillion of other obligations such as take-or-pay contracts and residual value guarantees. Because those sums are undiscounted, the report cautions, they are not directly comparable with debt.
These promises let a company with a strong credit rating support projects built by weaker borrowers. A hyperscaler signs a long lease, and the developer uses that lease to raise construction money. Chip makers guarantee the resale value of the processors that back a customer's loan. When equity stakes are added on top, the report says, the result is a circular financing dynamic, the pattern we mapped in AI circular deals.
“these firms' economic exposure to such a downturn may be substantially larger than suggested by their on-balance-sheet liabilities”
The report sees a cushion too: the largest technology firms may absorb losses in a downturn and soften the effect on everyone else. On who holds the paper, it says the big companies' own debt sits mainly with mutual funds, pension funds and insurers, while the riskier slices of project financings go to high-yield funds, life insurers and private credit vehicles. The largest US bank holding companies have close to $1 trillion of exposure to large technology firms, data center developers, related financing vehicles and utilities.
Faster than the 1990s telecom build-out
The Fed sets the spending of five hyperscalers since 2020 against the whole telecom industry from 1995, both as a share of GDP. The two lines tracked each other closely, but AI investment has been accelerating and, in the report's words, may surpass the pace of the 1990s cycle relatively soon. One difference stands out: strong cash flows meant the hyperscalers needed less debt to fund similar levels of investment, at least until this year. Our dot-com and AI comparison covers that history.
What the rest of the report says
Outside AI, the picture is steadier. The banking system remained sound and resilient with high capital ratios, the Fed says. Asset valuation pressures stayed elevated: the S&P 500's forward price-to-earnings ratio is well above its historical median, and the extra return investors earn for holding stocks over bonds is near the low end of its historical range. Hedge fund leverage reached its highest level since 2013 and is concentrated in the largest funds.
The report reflects data through October 28. That is before the worst of the chip sell-off and before a week in which the fallout from Jensen Huang's resignation cut more than $600 billion from Nvidia's market value, according to the Associated Press.
What the AI Bubble Index says
The AI Bubble Index read 61.7 after Friday's close, in its Normal band, down from 63.9 a week earlier. Its picture is close to the Fed's in one respect: Fragility, at 68.6, sits above Euphoria at 62.3, and the systemic risk family reads 68.9. Sentiment and attention is the hottest family at 99.9, while the demand family is the coolest at 40.7. Each family is updated daily on the dashboard.
The index and the survey measure different things. One is built from market and economic data, the other from the opinions of 23 professionals. Our guide to AI investment risks explains how financing links can spread a loss beyond the company that takes it.
What to watch next
Nvidia's earnings on Tuesday, November 17, the first test of the earnings worry the survey describes. New bond and private credit deals for data centers, and what lenders charge for them. The 10-year Treasury yield, which ended Friday at 5.15%. And the Fed's next report in the spring, which will show whether AI stays at the top of the list.Questions people ask
- What does the Fed financial stability report say about AI?
- The November 2026 report says AI was the risk most often cited by the Fed's market contacts, named by 87% of them according to CNBC, up from 50% in the spring. They worry about high equity valuations, AI-enabled cyberattacks and complex AI-related financing.
- How much debt has been issued to build AI data centers?
- The Fed estimates that debt issued for advanced computing infrastructure passed $680 billion by October 2026, about five times the $130 billion issued in all of 2023. It was roughly 31% of investment-grade corporate issuance outside the financial sector in 2026.
- Does the Fed think AI is a bubble?
- The report does not use that label. It says equity valuations are elevated and compares the pace of AI investment with the 1990s telecom build-out, but it also says banks are sound and that the largest technology firms could absorb losses in a downturn.
- Who took part in the Fed's risk survey?
- Staff at the Federal Reserve Bank of New York collected responses from 23 contacts at broker-dealers, banks, investment funds and advisory firms in September and October 2026. The results are their views, not the Fed's.
Sources
- Financial Stability Report, November 2026 · Board of Governors of the Federal Reserve System · November 13, 2026
- AI risks are the biggest threat to the financial system, Fed survey finds · CNBC · November 13, 2026
- Financial Stability Report (publication page) · Board of Governors of the Federal Reserve System · November 13, 2026
- Stocks fall but Wall Street notches a winning week after several turbulent days · Associated Press (via BNN Bloomberg) · November 13, 2026