Concern about artificial intelligence has climbed to the top of the list of risks to the financial system, according to the Federal Reserve’s semi-annual Financial Stability Report, released on Friday.
The survey found that 87% of those questioned now see AI as a major risk to the financial system. In the Fed’s May survey, only 50% did. Respondents pointed to three main worries: cyberattacks, complicated financing linked to AI, and the chance that slowing AI earnings could set off a broad market fall.
Other worries tied to AI
Several other concerns in the survey also had strong links to AI. A broad correction in risky assets came second, cited by 52% of respondents. Cyberattacks were flagged by 43%. Higher long-term interest rates worried 48%, while 43% pointed to doubts about whether government debt can be sustained.
The survey was carried out in September and October. That means respondents answered before the recent upheaval at chipmaker Nvidia, where chief executive Jensen Huang stepped down on Thursday following reports of extramarital affairs. The New York Times has reported that one of the women involved is under suspicion of being a Chinese spy. The survey was also completed before recent midterm elections, where voters’ growing unease about AI was a major issue.
Borrowing for data centres
The Fed devoted a large part of the report to borrowing tied to data centres and other advanced computing infrastructure. Debt issued for this purpose reached $680 billion by October 2026. That is already five times the debt issued in all of 2023.
Such borrowing now makes up nearly a third of all investment-grade debt issued by non-financial companies. Three years ago, it was less than 6%.
The Fed compared the current boom with the broadband telecom investment wave of the late 1990s. It added that the present cycle could soon outdo that earlier period.
Complex financing arrangements
The central bank also looked closely at how this spending is being financed. Long-term contracts, leases and guarantees allow large technology firms, known as hyperscalers, and other big companies to back projects run by smaller third parties. Graphics card makers and hyperscalers often add to these investments by taking equity stakes.
Data gathered by the Fed suggests that debt, guarantees and leases not yet begun on this infrastructure could exceed $3 trillion.
The Fed said this circular web of financing could help the largest firms absorb losses if a downturn hits, limiting the damage to the wider financial system. But it also warned that these firms’ exposure could be far larger than their balance sheets show. The report said the links between data centre builders, public debt markets, private lenders and special purpose financing vehicles create weak spots that need close watching.
Wider risks in the financial system
Elsewhere, the Fed found that asset prices across many categories remain high, partly because investors are excited about AI. The report noted that AI investment has been a significant driver of US economic growth this year. An unexpected drop in corporate earnings could trigger a market decline, tighter lending and reduced investment, all of which would weigh on the economy.
Such a slump could be made worse by high borrowing levels in parts of the financial sector. The Fed noted that leverage among hedge funds, particularly the largest ones, is at the highest level recorded since the central bank began collecting such data in 2013.