Home / USA News / Business and Finance
U.S. FINANCE · FEDERAL RESERVE · AI INFRASTRUCTURE · NOVEMBER 15, 2026AI Tops Fed’s Financial-Stability Survey as Data-Center Debt Surges
The Federal Reserve’s November 2026 Financial Stability Report says 87% of 23 surveyed market contacts cited AI as a risk, while data-center borrowing and off-balance-sheet commitments expanded.
Artificial intelligence was the most frequently cited potential threat to U.S. financial stability in the Federal Reserve's November report, while borrowing tied to data centers and advanced computing expanded sharply. The report's survey found that 87% of 23 market contacts named AI as a salient risk, up from 50% in the spring survey. The result is a warning about exposures and financing links, not a prediction that an AI crash or banking crisis is imminent.
What the Fed survey actually measured
Federal Reserve Bank of New York staff collected responses from 23 professionals at broker-dealers, banks, investment funds and advisory firms during September and October. AI was the leading concern for the next 12 to 18 months. Other frequently cited risks were a correction in risk assets (52%), higher long-term interest rates (48%), cyberattacks (43%) and fiscal debt sustainability (43%). The report explicitly says the summary reflects respondents' views and should not be interpreted as the views of the Federal Reserve Board or the New York Fed.
That distinction matters. The 87% figure is a small panel's assessment of possible shocks, not a probability assigned by the Fed that an AI downturn will occur. It is not a poll of the public or a survey of all financial institutions.
Data-center borrowing has grown quickly
The report's separate analysis of the data-center buildout found that debt issuance to finance advanced-computing infrastructure exceeded $680 billion by October 2026, roughly five times the $130 billion issued in all of 2023. Such borrowing grew from less than 6% of nonfinancial investment-grade corporate debt issuance in 2023 to about 31% in 2026. The Fed estimated outstanding public debt related to advanced computing at more than $1.1 trillion, while cautioning that private-credit transactions make the full amount difficult to measure.
The report also described large obligations that do not look like ordinary outstanding bonds or loans. Undiscounted values of leases that had not yet commenced exceeded $1 trillion. Other obligations, including take-or-pay contracts and residual-value guarantees, exceeded $2 trillion. These categories are not directly comparable with on-balance-sheet debt and should not be casually added together as if they were one conventional debt balance.
The largest U.S. bank holding companies had close to $1 trillion in exposure to large technology firms, data-center developers, special-purpose financing vehicles, private-credit vehicles and utility companies, according to the report. The Fed said these exposures are evaluated in the annual supervisory stress test.
How the financing could transmit losses
Large cloud and technology companies often support data-center construction through long-term leases, purchase commitments, guarantees or equity stakes in smaller developers and computing firms. Those arrangements can help projects obtain financing and bring capacity online. They can also create links between hyperscalers, chip suppliers, project companies, lenders, bond investors and private-credit funds.
If expected demand or AI-related earnings fall short, a project developer could face difficulty meeting its obligations. A slowdown could affect the value of data-center assets, the firms that lease them, lenders that financed construction and investors holding the debt. The Fed also highlighted cybersecurity concerns, including attacks on financial-market infrastructure and possible unintended damage caused by advanced AI models.
The report does not say these losses would automatically spread through the banking system. It notes that the largest technology firms could absorb some losses, which may limit spillovers, while also warning that their economic exposure could be larger than their reported balance-sheet liabilities suggest.
The wider financial backdrop
The Fed described the banking system as sound and resilient overall, with high regulatory capital ratios. At the same time, it said hedge-fund leverage had reached its highest level since comprehensive data began in 2013 and remained concentrated among the largest funds. The report also noted elevated equity valuations and rising Treasury term premiums. Those conditions could make markets more sensitive to a sharp change in expectations or a rise in long-term borrowing costs.
The report generally reflects information available as of October 28, 2026. It is a financial-stability assessment, not a new monetary-policy decision, a warning that banks are insolvent or a directive to slow AI investment.
What investors, lenders and companies should watch
Investors can monitor how AI companies fund capital spending, the share of projects financed through private credit and structured vehicles, and whether expected revenue growth is keeping pace with data-center construction. Lenders and counterparties may focus on lease commitments, guarantees, collateral values and the credit quality of smaller operators. Technology companies may need to assess whether their long-term capacity commitments remain viable if demand or pricing changes.
For households, the report does not signal an immediate change to bank deposits, credit-card rates or mortgage rates. Its significance is broader: AI infrastructure has become large enough and interconnected enough that regulators are tracking its financing as a potential source of financial-system vulnerability.
Frequently asked questions
Does the Fed say AI is likely to cause a financial crisis?
No. The survey records concerns from 23 market contacts, not a forecast or official Fed probability. The report identifies vulnerabilities that regulators say should be monitored.
Can the leases and guarantees be added to the debt estimate?
No. The Fed reports leases that have not started and other commitments such as guarantees as separate, undiscounted categories. They are not directly comparable with outstanding debt and should not be treated as a single debt total.
Sources and documents
Editorial note: This article distinguishes official data, reported figures and editorial interpretation. Daily fuel trackers and weekly federal estimates use different methods and reporting periods; market-risk surveys do not represent official policy forecasts.
Related coverage: Latest U.S. news · Business and finance · Technology and AI