Skip to content
MustHave.ai Musthave.ai
AI Startups & Business

AI topped the Fed’s November risk survey. What changed since May?

Compare the respondents, percentages and approximate counts in the Fed’s spring and fall 2026 AI risk surveys, and see what the report says about data center debt.

AI topped the Fed’s November risk survey. What changed since May?
Illustration of AI appearing more frequently in the fall survey.

AI moved from 10 mentions among 20 market contacts in spring to about 20 among 23 in fall. The report also tracks changes in data center financing.

Artificial intelligence was the most frequently cited risk in a survey of market contacts published in the Federal Reserve’s November 13 Financial Stability Report. The Federal Reserve Bank of New York gathered 23 responses in September and October. About 87% cited AI, compared with 50% in the spring.

Respondents included people working at financial institutions. They identified possible shocks over the next 12 to 18 months. The shares tell us how often each risk came up in this group. They do not estimate the probability that a shock will happen.

Put the percentages back into counts

The November report shows both survey rounds in Box 5.1. Multiplying the published percentages by each sample size recovers the approximate count behind the bars.

Survey roundMarket contactsShare citing AIApproximate count
Spring 20262050%10
Fall 20262387%20

In the fall, 20 divided by 23 is about 86.96%, which rounds to 87%. The report’s 52% for risk asset correction works out to 12 of 23. On those figures, the AI category drew roughly eight more mentions than the next category in the fall. In the May report, geopolitical risk and an oil shock ranked ahead of AI.

Each fall response accounts for about 4.35 percentage points. The headline increase of 37 percentage points is meaningful in describing these two sets of replies. The survey does not provide a matched panel showing which individuals changed their minds.

The financing figures answer a different question

The survey records concerns. Box 2.2 in the same report describes measurable changes: who is borrowing, how they borrow, and what obligations they take on. The report estimates that outstanding public debt tied to advanced computing exceeded $1.1 trillion as of October 2026. It also describes private debt, leases and guarantees that can make the full exposure harder to measure.

The debt figure is a stock at a point in time. The Fed separately tracks how much new debt companies issue over a period. Keeping those measures separate matters: a company can borrow more this year while still owing substantial amounts from earlier years. A future lease or guarantee can also expose a company to costs outside the simple bond total.

What to watch in the next report

The next survey can tell readers whether AI remains near the top of respondents’ concerns. The financing analysis can show whether new borrowing is slowing or shifting into harder-to-track structures. Those questions can have different answers. Mentions can fall while lease commitments rise, or market contacts can grow more cautious while the amount of debt firms issue slows. Reading the survey and financing box together makes it easier to spot that difference.

Abdessalam Alaoui

I'm Abdessalam Alaoui, founder of Musthave.AI. I write practical guides to AI models, tools, generative media.

Join the discussion

Your email address will not be published. Required fields are marked *

Keep reading

All news