Will the AI Bubble Burst? Harvard Faculty Weigh In

Three Harvard faculty said they think fears that an artificial intelligence bubble will burst — leading stock prices to collapse in the wake of soaring investments into AI companies — are overblown and a recession is unlikely.

HKS senior fellow Paulo Carvão, who specializes in AI regulation, said that he has been warning of an AI bubble “before it became fashionable” — and he believes that markets are now showing signs of “bubble conditions.”

Over the last year, the value of many AI and tech companies have skyrocketed due to widespread expectations about the potential of new AI technology to multiply companies’ profits. But many experts worry that companies’ valuations are far too high, as many investors are betting eye-popping sums on companies that are still far from making significant returns.

An AI bubble, Carvão said, could present two possible scenarios: one where the bubble “pops,” resulting in a sharp economic downturn, and another where it “deflates.”

“A balloon exploding – that has some catastrophic implications,” Carvão said. “The other one is you’re slowly letting the air out of the balloon, and you have time to adjust.”

The second case is more likely, he said.

“It doesn’t mean that we’re not going to have a few failures. It doesn’t mean that we couldn’t have a small correction in the stock market. But these are normal events that we navigate through without major economic consequences,” he added.

According to Carvão, what happens in the next few months will dictate which of these two scenarios the economy will face. However, he sees the deflating scenario as more likely, where risk is concentrated in individual companies such as OpenAI rather than the industry as a whole.

Carvão argued “accelerating demand” from firms, governments, and consumers for AI products is the best way to avoid a bubble and fuel economic growth. Modernizing government platforms and investing in high-tech defense contracts could help keep the AI economy aloft, he said, pointing to the Trump administration’s push for AI research at national laboratories.

But if demand doesn’t rise, Carvão said, the data centers and other infrastructure that AI firms are currently making major investments in could go unused — leading to conditions like those leading up to the dot-com bubble burst at the turn of the century.

Economics professor David I. Laibson ’88 said that he has concerns regarding AI companies’ profitability.

“You take a firm like OpenAI, their current revenue is $13 billion, but they’re making capital expenditure commitments over the next eight years that are, at this point, $1.4 trillion,” he said.

“So one might worry that down the road, the profits that they’re anticipating might not be sufficiently large to justify the enormous expenditures that they’re undertaking right now.”

Given these figures, which Laibson said are common throughout the industry, he is worried that a slowdown in the growth of AI stocks — which account for an ever-increasing share of the S&P500’s value — could drag down the market.

But much of the investment in AI infrastructure comes from private equity as venture capitalists plow vast sums into tech firms. Laibson said that a broader collapse remains unlikely, unless banks become a “major funding source” for AI firms and take on more AI-related debt, which would tie the financial system more closely to their performance.

“I would get more worried if banks became more and more directly connected to financing these investments. Then we would have the possibility of a systemic financial crisis,” he said.

“We’re not there yet. However, I’m watching it closely. We are expanding the vulnerability of the banking sector week by week,” Laibson added.

Economics professor Jason Furman ’92 struck a more optimistic note, saying that he is “not betting on a bubble.” Even if the value of AI stocks is inflated, the economic impacts of a crash may still be relatively contained, Furman said.

“These stocks are not an important part of the wealth of most people in the economy,” he said. “It’s not the same thing as if the value of your house falls by a lot.”

“I don’t think it’s likely that will translate to a financial crisis or deep recession,” Furman added.

—Staff writer Evan H.C. Epstein can be reached at [email protected]. Follow him on X at @Evan_HC_Epstein.

—Staff writer Graham W. Lee can be reached at [email protected]. Follow him on X @grahamwonlee.