Fed Warns AI Investment Boom Is Emerging as Top Threat to Financial Stability
November 15, 2026
The IMF warned that global growth and financial stability are increasingly exposed to a downturn in AI investment, as confidence and spending in the sector become more important to the economy.
One concern is more than $1 trillion in AI-related circular financing, roughly two-thirds of it debt-funded: technology companies invest in AI developers, which then buy chips, computing capacity and services from those same investors.
The risks are harder to assess because some obligations sit off balance sheet, including energy contracts, data-center and computing leases, and guarantees on asset resale values.
Survey respondents also flagged a possible asset-price correction, higher long-term interest rates, cyberattacks and government-debt sustainability as risks, though they ranked below AI; the assessment covered developments through October 28, 2026.
The commentary characterizes the moment as a serious stress test, not a debt crisis, and argues that lasting winners will be judged by sustainable returns rather than GPU counts or announced data-center capacity.
The infrastructure buildout faces constraints in electricity, land, water, equipment and skilled labor, with backlogs for some imported inputs estimated at two to 10 years; local concerns about power prices, environmental effects and quality of life could also limit construction.
Political pressure is adding to those constraints: voters backed limits on data-center development in 40 of 41 local ballot measures, while newly elected Democrats have signaled support for policies addressing electricity costs, independent testing of advanced models and company liability.
The House passed the Ratepayer Protection Act (H.R. 9340), which would require state utility regulators to consider assigning higher energy costs to large-load customers, but a Senate cloture vote failed.
At the same time, high semiconductor profits could draw in competitors, and the substantial chip and data-center capacity expected over the next two years could weaken pricing power and squeeze margins.
Cheaper open-weight models are also challenging frontier labs: a September survey found that 53% of 200 enterprises had partly replaced frontier models with open-weight alternatives.
The competitive landscape remains hard to predict: OpenAI and Anthropic rose quickly, but Amazon, Microsoft, Alphabet, Meta and Oracle have the resources to challenge them, and today’s leaders may not remain dominant over the next five to 10 years.
Overall, investors have raised near-term commercial expectations without giving up on longer-term technology bets; companies will need to show that their products can be delivered reliably and that orders turn into collected revenue.
Summary based on 82 sources
Get a daily email with more Macroeconomics stories
Sources

The New York Times • Nov 16, 2026
Opinion | The Financial Bubble Cycles Are Getting Worse
CNBC • Nov 13, 2026
AI risks are the biggest threat to the financial system, Fed survey finds
Axios • Nov 16, 2026
The AI boom faces a dot-com reckoning