(Bloomberg) — As artificial intelligence trades continue to drive stock markets to new highs, investors are increasingly wondering whether we are experiencing a new financial bubble destined to burst someday.
The answer, at least according to history, is not that simple.
Most Read Articles on Bloomberg
The S&P 500 index (^SPX) rose 16% in 2025, with AI winners Nvidia Corp. (NVDA), Alphabet Inc. (GOOG, GOOGL), Broadcom Inc. (AVGO), and Microsoft Corp. (MSFT) contributing the most. But at the same time, concerns are growing over the hundreds of billions of dollars Big Tech companies have pledged to spend on AI infrastructure. Capital spending at Microsoft, Alphabet, Amazon.com (AMZN) and Meta Platforms (META) is expected to rise 34% next year to a total of about $440 billion, according to data compiled by Bloomberg.
OpenAI (OPAI.PVT), on the other hand, has pledged to spend more than $1 trillion on AI infrastructure, an impressive number for a private company that doesn't make a profit. But perhaps even more troubling is the cyclical nature of many of the arrangements, with investments and spending going back and forth between OpenAI and a handful of publicly traded tech giants.
Throughout history, overinvestment has been a common theme when society-transforming technological advances occur, said Brian Levitt, chief global market strategist at Invesco, who pointed to developments in railroads, electricity and the internet. This time may be no different.
“At some point, infrastructure development may exceed what is needed by the economy in a short period of time,” he said. “But that doesn't mean the railroad wasn't built or the internet didn't take off, right?”
Still, with stock valuations creeping higher and the S&P 500 just posting its third straight year of double-digit gains, it's understandable that investors are growing concerned about how much upside remains and how much market value could be lost if AI fails to live up to the hype. Nvidia, Microsoft, Alphabet, Amazon.com, Broadcom, and Meta Platforms make up nearly 30% of the S&P 500, so a decline in AI would hit the index hard.
Gene Goldman, chief investment officer at Cetera Financial Group, doesn't think AI stocks are in a bubble, saying, “The bubble is likely to burst in a bear market.” “It's hard to imagine a bear market starting in the near future.”
Here's how today's AI boom overlaps with past market bubbles.
One easy way to gauge whether AI-powered tech stocks are rallying too much, or too quickly, is to compare them to past bull markets. A study by Michael Hartnett, a strategist at Bank of America, found that 10 stock bubbles around the world since 1900 lasted on average for just over two-and-a-half years and rose 244% from bottom to top.
By comparison, the AI-driven bull market is in its third year, with the S&P 500 index up 79% since the end of 2022 and the tech-heavy Nasdaq 100 (^NDX) index up 130%.
Although it's difficult to draw conclusions from the data, Hartnett cautions investors not to run away from the stock market, even if they think it's a bubble. This is because the last part of the upswing is usually the steepest and the losses are greater if you miss. One way to hedge, he says, is to buy cheap value plays such as British stocks and energy companies.
The top 10 stocks in the S&P 500 now account for about 40% of the index, a level of concentration not seen since the 1960s. As a result, some investors, including Wall Street research veteran Ed Yardeni, said in December that it no longer makes sense to recommend overweighting tech stocks.
Market historians argue that while the concentration seems extreme compared to recent memory, there is precedent. According to Paul Marsh, a professor at London Business School who has studied global asset returns over the past 125 years, the share of top stocks in the U.S. market was similar in the 1930s and 1960s. In 1900, 63% of U.S. market capitalization was tied to railroad stocks; by the end of 2024, 37% was tied to technology, Marsh said.
Asset bubbles tend to be much harder to spot in real time than after the fact, said TS Lombard economist Dario Perkins. Fundamentals are at the center of the discussion, and the indicators investors pay attention to are fluid.
“It's easy for tech enthusiasts to argue that things are different now and that fundamental valuations will never be the same again,” he said.
But some basics are always important. For example, compared to the dot-com bubble, today's AI giants have lower debt returns than, say, WorldCom Inc. And while companies like Nvidia and Meta Platforms are already reporting significant profit growth from AI, that wasn't necessarily the case in the speculative era 25 years ago.
The potential for credit risk in AI trading has made some investors nervous. Oracle sold $18 billion in bonds on September 24, but its stock price plummeted 5.6% the next day and has fallen 37% since then. Societe Generale estimates that Meta, Alphabet and Oracle will need to raise a combined $86 billion in 2026 alone.
The S&P 500's valuation is at an all-time high, with the exception of the early 2000s, at least according to the economy-adjusted price-to-earnings ratio (a measure created by economist Robert Shiller that measures stock prices divided by the average inflation-adjusted earnings over the past 10 years).
Bullish investors say that while technology is driving up market valuations, they are rising much more slowly than in the dot-com era. At one point in 2000, Cisco Systems Inc.'s stock price was more than 200 times trailing 12-month earnings; today, Nvidia's stock price is less than 50 times trailing 12-month earnings.
In an environment where there is no valuation debate, stock prices are decoupled from earnings growth, said Richard Claude, a fund manager at Janus Henderson. “We just don’t see it at this point,” he said.
Discussion of a potential stock bubble had permeated throughout the year, but picked up significantly in November and December following warnings from investor Michael Barry and the Bank of England. According to data compiled by Bloomberg, the term “AI bubble” was mentioned in more than 12,000 articles in November, almost as many as in the previous 10 months combined.
Investors see an AI bubble as the biggest “tail risk” event, according to a December Bank of America poll. More than half of respondents said the Magnificent Seven's tech stocks were the most heavily traded on Wall Street.
This is in contrast to the dot-com bubble, when there was “absolute excitement that the internet was going to revolutionize everything,” said Venu Krishna, head of U.S. equity strategy at Barclays. And as bond issuance increases, so do questions about whether AI investments will pay off.
“I don't discount it, but I think surveillance in general is healthy,” he said. “In fact, that scrutiny prevents extreme movements like crashes.”
