Former US Securities and Exchange Commission (SEC) chairman Gary Gensler has warned that the artificial intelligence (AI) boom has created a bubble that could burst before the 2028 US presidential election and trigger a recession, though the impact is expected to be less severe than the 2008 global financial crisis.
Gensler, who chaired the SEC from 2021 to 2025, told CNN that companies are pouring vast sums into AI infrastructure, with annual investment estimated at as much as US$800 billion (about S$1.03 trillion), but earnings growth is falling far short of the pace of investment.
Gensler, now a professor at the Massachusetts Institute of Technology, said tech giants’ spending this year is five times what it was three years ago — a staggering scale of investment. Citing past booms in railways, electricity and the internet, he said technological breakthroughs were often accompanied by overinvestment, ultimately leading to market corrections.
He said: “History tells us that when new technologies attract large amounts of investment, there always comes a point when growth stalls and contraction begins. The only question is whether the economy will experience a mild recession or a more severe shock.”
Gensler is particularly concerned about financing models in the AI industry, including circular financing and complex off-balance-sheet arrangements that make it difficult for investors to assess the actual risks. He believes some lenders are lending too much, too quickly, potentially setting the stage for another bubble to burst.
He predicted that if the bubble bursts, the first to be hit would be a new breed of cloud computing companies that rely heavily on borrowing and rent computing power to AI developers. Some could go bankrupt, while large technology companies would be better placed to weather the crisis.
Gensler believes the potential economic impact could be comparable to the bursting of the dotcom bubble in the early 2000s, but would not be as severe as the 2008 subprime mortgage crisis. He expects the US economy could see assets repriced and growth slow before the presidential election in November 2028.
Wall Street, however, is not uniformly bearish on AI’s prospects. Ed Yardeni, president of Yardeni Research, believes the recent pullback in AI-related stocks could help cool the market and prevent valuations from becoming excessively inflated.
He said chipmakers’ price-to-earnings ratios have now fallen below that of the broader market. Unlike the period before the dotcom bubble burst in 2000, there are no signs of irrational exuberance.
Gensler stressed that he remains optimistic about AI’s long-term potential, but worries that capital markets are underestimating the associated risks.