Fed Names AI the Top Financial Stability Risk Amid a $680 Billion Computing Debt Boom

2026-11-15 • Technology

Fed Names AI the Top Financial Stability Risk Amid a $680 Billion Computing Debt Boom

The Federal Reserve asked Wall Street what could break the financial system, and 87% said artificial intelligence. Not the robots, the receipts. Companies have issued more than $680 billion in debt to build the machines, which is about five times the 2023 figure. The Fed even compared it to the 1990s telecom boom, a period best remembered for how it ended. Your chatbot is polite, helpful and heavily mortgaged.

87%Market contacts naming AI as a risk
$680BComputing debt issued by October 2026
31%Share of top grade corporate bond issuance

Why Is the Fed Suddenly Worried About AI?

Twice a year the Federal Reserve asks a group of Wall Street contacts a simple question, which is what could break the financial system. In the spring the answers were wars and oil, with geopolitical risks first and an oil shock second (Board of Governors of the Federal Reserve System [Fed], 2026). This fall the top answer was the same technology that writes your emails.

In the latest survey, 87% of contacts named artificial intelligence as a significant risk, up from 50% in May. It beat a broad correction in risk assets at 52%, higher long term interest rates at 48%, and cyberattacks and fiscal debt sustainability at 43% each (Investing.com, 2026).

Share of market contacts citing each risk in the fall 2026 survey in percent
Artificial intelligence87%
Risk asset correction52%
Higher long term rates48%
Cyberattacks43%
Fiscal debt sustainability43%

The worry is not a robot uprising. Contacts fear that lofty stock valuations could fall if earnings disappoint, that advances in AI are lowering the barrier to sophisticated cyberattacks, and that circular financing and heavy reliance on debt could amplify losses (Fed, 2026). If the cyber part sounds familiar, it is the same threat we covered when most organizations got hacked last year.

The $680 Billion Tab Behind the Chatbots

The numbers explain the nerves. Debt issuance tied to advanced computing rose to more than $680 billion by October 2026, about five times the $130 billion issued in 2023 (Fed, 2026). It now makes up about 31% of all nonfinancial investment grade corporate bond issuance, up from under 6% in 2023.

Bonds are only the visible part. The Fed estimates more than $1.1 trillion of public advanced computing debt outstanding, more than $1 trillion in leases that have not yet started, and more than $2 trillion in other obligations such as take or pay contracts (Fed, 2026). Large US bank holding companies have close to $1 trillion of exposure to large technology firms, data center developers and related borrowers.

This is a real change in behavior. The report notes that the strong cash flows of hyperscalers reduced their need to issue debt until this year (Fed, 2026). We looked at that turn when Big Tech started spending faster than it earns.

Is This the 1990s Telecom Boom All Over Again?

The Fed makes the comparison itself. It says the current buildout often draws comparisons to the broadband telecom investment cycle of the late 1990s and early 2000s. It adds that advanced computing investment is accelerating and may surpass the pace of that era relatively soon (Fed, 2026).

For anyone who missed that episode, telecom companies borrowed heavily to lay fiber for an internet boom, and many ran out of money before the traffic showed up. The internet turned out to be real. The balance sheets did not all survive to enjoy it.

That is the uncomfortable lesson for today. A technology can be transformative and still be a bad investment at the wrong price. The question is never only whether AI works, it is who borrowed how much to bet on when it pays off.

A technology can be transformative and still be a bad investment at the wrong price.

Twenty Three People and a Very Loud Alarm

Now for the skeptical part. The fall survey reflects the views of 23 contacts at broker dealers, banks, investment funds and advisory firms, gathered in September and October (Fed, 2026). That is a small room, and it measures what professionals worry about, not what will happen.

Fear also follows fashion. The same survey ranked AI third in the spring, behind geopolitics and oil, and fewer participants cited private credit this time than six months ago (Fed, 2026). A list that reshuffles this fast tells you as much about the mood as about the risk.

Still, the Fed is not calling AI a bubble. Its point is narrower and more useful, which is that the money behind the boom has become large, debt funded and tangled together. Because the lenders include banks and bond funds, a disappointment in one corner could travel to places where ordinary savings live.

What You Can Actually Do About It

Look at how much of your index fund or pension sits in a handful of AI linked companies, because many broad funds hold more of them than people expect. Spread your money across sectors and regions, and check whether your bond funds hold a lot of technology debt. If you work in tech, keep an emergency fund that does not depend on your employer's share price. Do not sell in a panic because of one survey. Then ask yourself, if AI stocks dropped sharply next year, how much of your savings would feel it?

References

Board of Governors of the Federal Reserve System. (2026, November 13). Financial stability report. https://www.federalreserve.gov/publications/files/financial-stability-report-20261113.pdf

Investing.com. (2026, November 13). AI risks top financial stability concerns, Fed reports. Investing.com. https://www.investing.com/news/stock-market-news/ai-risks-top-financial-stability-concerns-fed-reports-93CH-5026543

If AI stocks dropped sharply next year, how much of your savings would feel it?
Enjoyed this? Get the next one in your inbox

New articles on finance, technology, trade and real estate, written in plain English.

← Back to articles

Comments (0)

No comments yet — be the first to comment.


Leave a comment