Fed puts AI first among financial risks, and the survey behind it asked just 23 people

The Federal Reserve's November Financial Stability Report, released Friday, says equity valuations remain high, partly because of "market enthusiasm for super intelligence," and that borrowing to build AI data centers has grown about fivefold since 2023. CNBC reported that AI topped the list of risks in a Fed survey, with 87% of respondents naming it. Here is what the report itself says, including the part the headlines left out.

The survey is 23 people, and it is not the Fed's view

The "survey of salient risks" is Box 5.1 of the report. Staff at the Federal Reserve Bank of New York asked market contacts for their views during September and October, and got responses from 23 contacts at broker-dealers, banks, investment funds and advisory firms. The report states the box "should not be interpreted as representing the views of the Federal Reserve Board or the Federal Reserve Bank of New York."

So "the Fed says AI is the biggest threat" overstates it. A small, self-selected group of market professionals ranked AI first, and the Fed published their answers. Their three concerns were stretched equity valuations, AI-enabled cyberattacks and complex AI-related financing. They also widely cited a risk-asset correction, higher long-term Treasury yields, cyberattacks, fiscal debt sustainability and higher energy prices, the last tied to a possible escalation in the Middle East.

The number behind the number: $680 billion of debt

The Fed's own analysis is in Box 2.2, "Leverage in the Data Center Buildout," and it is the more useful read for anyone watching credit.

• Debt issued to finance advanced computing infrastructure rose to more than $680 billion by October 2026, about five times the $130 billion issued in all of 2023.
• That borrowing was about 31% of investment-grade nonfinancial corporate debt issuance in 2026, up from less than 6% in 2023.
• Outstanding public debt tied to advanced computing is estimated at more than $1.1 trillion. The Fed says private credit deals make the true total hard to measure.
• Beyond debt, large tech firms have taken on leases that have not yet started (over $1 trillion, undiscounted) and other commitments such as take-or-pay contracts and residual value guarantees (over $2 trillion, undiscounted). The Fed cautions these are not directly comparable with discounted balance-sheet debt.
• The largest US bank holding companies have close to $1 trillion of exposure to large tech firms, data center developers, related special purpose entities and private credit vehicles, and utilities. Those exposures are tested in the annual stress test.

The report also notes that the build-out is tracking the late-1990s telecom investment cycle relative to GDP, and that spending on advanced computing has been accelerating, suggesting it "may surpass" that cycle relatively soon. Hyperscalers' strong cash flow kept their debt needs down, "at least until this year."

Leverage elsewhere in the system

The report's leverage section is blunt. Hedge fund gross notional leverage, measured as exposure divided by net asset value, reached its highest level since the data began in 2013, as of the first quarter of 2026 (the latest comprehensive data, which run on a two-quarter lag). It is concentrated in a small number of large funds, with record or near-record leverage in equity, relative value and macro strategies. Leverage at the largest life insurers is also historically high. Banks, by contrast, are described as sound, with capital near historic highs, though fair-value losses on fixed-rate assets remain sizable and sensitive to long-term yields.

On valuations, the forward price-to-earnings ratio of the S&P 500 has fallen over the past year as earnings expectations rose, but it remains well above its historical median, and the equity premium sits near the low end of its historical range. Treasury yields "rose markedly" in the six months since the May report. The index closed Friday near 7,562, about 3.3% below its Oct. 6 peak, according to Yahoo Finance data.

Who this actually touches

For a trader, the practical reading is that the Fed's stress points are the same ones already moving this week: AI-linked equities, long yields and crowded hedge fund positioning. Our reports on the chip-stock reversal and the 10-year yield above 5% cover those moves.

For a business owner, the exposure is indirect but real. If lenders pulled back from data-center lending or a correction hit AI-linked equities, the Fed says the damage could spill over to business investment and household consumption. A restaurant owner or contractor with a floating-rate line would feel that through tighter credit and weaker customer spending, not through any AI exposure of their own. Nothing in the report says that is happening: the Fed calls business and household debt vulnerabilities "moderate" and funding risks "moderate." It is a map of where the pressure points are, as of Oct. 28.

Sources: Federal Reserve Board, Financial Stability Report, November 2026 (data as of Oct. 28, 2026; survey of 23 contacts, Sept.-Oct.); CNBC; Yahoo Finance for S&P 500 levels. This is market information, not investment advice.