In its quarterly update on October 8, the Bank of England’s Financial Policy Committee (FPC) issued a stark warning over the feverish investor enthusiasm surrounding AI, saying that “equity valuations appear stretched,” especially in certain backward-looking metrics in U.S. stocks, and especially technology companies focused on AI. When combined with increasing concentration within market indices, the FPC added, equity markets find themselves “particularly exposed should expectations around the impact of AI become less optimistic.” Since its last meeting in June, the FPC noted that risky asset valuations had increased as credit spreads had compressed, as it questioned these stretched valuations

Nvidia CEO Jensen Huang has defended the large—and, some would say, “circular“—deals at the center of the increasing talk about an AI bubble, during a September 25 appearance on the Bg2 podcast with Brad Gerstner and Clark Tang. He said Nvidia’s $100 billion deal with OpenAI was an “opportunity to invest” in a company that Nvidia believes will be “the next multitrillion hyperscale company.” He said OpenAI will pay Nvidia back through its offtake/future revenues, which he pointed out out are “growing exponentially,” as well as capital it raises via future equity and debt sales, underscoring his own high levels of optimism around AI in general and the OpenAI example in particular.

AI mania and market valuations

The Bank of England FPC noted that backward-looking metrics in the U.S. are one particular place to look for stretched valuations, and offered a striking comparison. “For example, the earnings yield implied by the Cyclically-Adjusted Price-to-Earnings (CAPE) ratio was close to the lowest level in 25 years—comparable to the peak of the dot com bubble.” Fortune‘s Shawn Tully has repeatedly argued in a similar vein that multiples are stretched and the S&P 500 is overconcentrated, most recently writing on September 23 that the index, after flirting with a price-to-earnings ratio of nearly 30, actually crossed the line around 3:00 pm ET on September 22. It’s a “terrible omen for investors,” he added.

Regarding concentration, the FPC highlighted how the price appreciation among the largest tech players has propelled concentration within U.S. indices, with the top five members of the S&P 500 now commanding nearly 30% of market share, a record high for any point over the last 50 years. Forward-looking price-to-earnings ratios do not rival the dotcom boom of the 2000s, the FPC added, although they remain strikingly elevated.

Morgan Stanley Wealth Management’s chief investment officer Lisa Shalett previously told Fortune she was bracing for a “Cisco” moment, when the dotcom bubble peaked and that stock went on to lose 80% of its value.

Risks of market correction

The FPC’s message comes amid mounting global uncertainties—from geopolitical tensions and trade fragmentation to rising sovereign debt risks—that elevate the likelihood of a sharp market correction. If investor sentiment around AI sours, or if progress stalls due to technological bottlenecks or supply constraints, equity prices could tumble—and, given the degree of market concentration, such an adjustment would ripple rapidly through broad market indices, affecting millions of investors. “The risk of a sharp market correction has increased,” the FPC said.

The FPC stressed that asset price corrections could adversely impact the cost and availability of credit for households and businesses. A sudden shift in AI market sentiment, or crystallization of wider global risks, would not only affect tech heavyweights but could spill over into broader financial stability concerns, including for the UK as a leading global financial center.

The FPC did not comment on potential aftershocks in the U.S., other than to note the “continued commentary about Federal Reserve independence.” A sudden or significant change in perceptions of the Fed’s credibility could result in the U.S. dollar undergoing a sharp repricing, and the FPC flagged the potential for higher volatility and global spillovers.

For this story, Fortune used generative AI to help with an initial draft. An editor verified the accuracy of the information before publishing.