AI Becomes the Top Financial Stability Worry in Fed Survey, Cited by 87% of Contacts

The twice-yearly report, released on Friday, November 13, showed the share of respondents flagging AI jumping from 50% in the spring survey. Coverage of the report put the figure at 30% in the autumn of 2025, which means AI has gone from a minority concern to a near-consensus one in about a year. The respondents include market participants, economists, analysts and institutional sources.

The result matters because of how far AI now sits ahead of the risks that usually dominate the survey. A correction in risk assets was the next most cited concern at 52%, followed by higher long-term interest rates at 48%, and cyberattacks and the sustainability of government debt, each at 43%, according to CNBC's account of the report.

The report groups the AI concerns into three channels. The first is stock valuations: equity prices are elevated and could fall sharply if AI-related earnings disappoint. The second is cybersecurity, where contacts worry about attacks by state actors and about threats arising from frontier AI models themselves. The third is the increasingly complex financing behind AI infrastructure, where borrowing tied to data centres and computing capacity has risen steeply.

The survey period closed on October 28, so the responses were collected before Jensen Huang resigned as Nvidia's chief executive and before the scandal over restricted Nvidia chips reaching China that has since rattled chip stocks. The worries recorded in the report were therefore building before the latest turmoil in the sector's most valuable company.

The Fed's overall assessment remained that the financial system is fairly stable. Beyond AI, the report pointed to hedge fund leverage, which it said is at its highest level since 2013 and concentrated among a relatively small group of funds, a structure that can amplify selling if markets turn.

The survey is a gauge of sentiment rather than a forecast. It records what a panel of contacts is worried about, not whether or when a market correction will happen, and the Fed does not attach probabilities to the risks it lists.

The findings land in a crowded week for AI and monetary policy. Fed Chair Kevin Warsh is hosting a public forum on AI's economic impact in San Francisco on Monday, Nvidia reports quarterly results on Tuesday under interim chief executive Colette Kress, and Anthropic is expected to price its initial public offering on Wednesday ahead of its Nasdaq debut on Thursday. Each will give investors a fresh read on whether the earnings behind AI valuations are holding up.