Wall Street Is Shaking Off Fears of an A.I. Bubble. For Now.

 

Wall Street Is Shaking Off Fears of an A.I. Bubble. For Now.

 

The valuations of some artificial intelligence companies are approaching those of the dot-com boom. But investors worry that pulling money from today’s market risks future gains.

 

Joe Rennison

By Joe Rennison

Dec. 9, 2025

Updated 1:58 p.m. ET

https://www.nytimes.com/2025/12/09/business/wall-street-valuation-ai-bubble.html

 

The stock market has broken multiple records this year, brushing past tariffs and signs of a strained American consumer, and recovering from a recent dip, largely because of the promise of artificial intelligence.

 

Share prices of some A.I. companies have risen dramatically in a short time, and tech companies are spending billions to build data centers and microchip plants to power the boom.

 

While investors and analysts see good reasons to justify the exuberance behind the nearly 50 percent increase in the S&P 500 over the past two years, some warn that current valuations still rest on a big bet on the future.

 

The stock market found itself at a similar moment 29 years ago, when Alan Greenspan famously warned of “irrational exuberance” fueling a bubble in internet stocks. The S&P 500 had risen more than 60 percent in the two years before Mr. Greenspan’s speech.

 

Policymakers are again sounding alarm bells.

 

“On some measures, equity valuations in the U.S. are approaching levels not seen since the dot-com bubble,” Andrew Bailey, the governor of the Bank of England, noted in a recent speech.

 

But when Mr. Greenspan delivered his warning, the dot-com bubble had only just begun inflating. The S&P 500 would rise more than 100 percent, meaning it doubled in value, before peaking in March 2000. After that, the index fell for the next two years, halving in value and hitting a bottom at roughly the same level as when Mr. Greenspan gave his warning six years earlier.

 

Some investors are worried we may be in a period akin to 1999, just before the crash. But more seem to believe we are closer to 1996, and that pulling money out of the market risks giving up hefty gains still to come.

 

“It’s uncertain but still early,” said Paul Christopher, head of global investment strategy at Wells Fargo. “We are not at 1999 — we think we can say that for sure.”

 

A.I. vs. Dot-Com

There are some important differences between the current excitement over A.I and the dot-com bubble, traders and analysts say.

 

One way to look at this is using the price-to-earnings ratio — which is calculated by dividing a company’s share price by their earnings, or how much money a company makes. The higher the number, the more speculative the bet.

 

The current P/E ratio of the S&P 500 is higher than usual. Over the past 10 years the index has averaged a ratio of around 22. That same ratio is now roughly 27, closer to 29, the 1999 peak just before the dot-com bubble burst.

 

“It’s hard for the equity market to perform well when you get to these valuations,” said James Masserio, head of global equities at RBC Capital Markets.

 

One important difference is that the companies driving the rally today are also bringing in the most money.

 

In the 1990s and early 2000s, many of the companies leading the stock rally were not making much money, if any. This led to very high P/E ratios for some companies because share prices kept going higher, even when earnings were lagging well behind.

 

 

Cisco Systems was the largest company in the S&P 500 when the dot-com bubble burst in March 2000. Its P/E ratio peaked at over 200, meaning its share price was more than 200 times its earnings over the preceding year and without further growth, even if all the company’s profits were passed back to shareholders, it would take two centuries for an investor to break even.

 

A high P/E ratio is therefore also a sign of investors betting on a company growing substantially, meaning it will earn more in the future, and when it does that will make its price appear more reasonable. That means P/E ratios can also give us a window into how dependent investors are today on the future panning out as they expect.

 

It never panned out for Cisco. The company’s earnings grew from 15 cents per share in 1996 to 36 cents per share in 2000, paling in comparison to its rampant price appreciation. Today, Cisco is valued at roughly $300 billion, still less than investors had valued the company at all those years ago.

 

Not Just Blue Skies and Unicorns

Nvidia, the largest company in the S&P 500 today and the first company to reach a value of $5 trillion in the stock market, has also flirted with some extreme valuation metrics. Its P/E ratio peaked in 2023 at over 200, the same as Cisco’s, during the early enthusiasm for A.I. after ChatGPT’s introduction to the world in November 2022.

 

But since then, Nvidia has made a lot more money. It still hasn’t sated demand for its high powered computer chips. Today, its P/E ratio is around 45, and if you base the calculation on expected future earnings, rather than its earnings over the past 12 months, that ratio drops to around 25.

 

The ability for Nvidia to continually meet lofty earnings expectations has been “remarkable,” said Alex Altmann, head of equities tactical strategies at Barclays. “It’s almost unbelievable.”

 

“It’s not just blue skies, rainbows and unicorns, but so far these businesses have demonstrated repeat excellence,” he added, “which has to be acknowledged.”

 

While Nvidia’s stock price has risen roughly 1,000 percent over the past three years, from $17 to $180, its earnings — the actual money it is making — have increased even faster. This means the stock is arguably cheaper today than it was three years ago, said Stacy Rasgon, a stock analyst at AB Bernstein.

 

Mr. Masserio of RBC Capital Markets remains wary of the path ahead for the stock market. “There is always a camp that says this time is different,” he said while still acknowledging that “the ice is thicker under these more expensive stocks in 2025 than in the early 2000s.”

 

The Weak Links in A.I.

Bubbles tend to start with a reasonable thesis before investors’ bets on the future become divorced from reality. Even when the market crashes, the thesis often proves correct. The internet didn’t disappear after the dot-com bubble burst.

 

While investors are less worried about the big companies driving the rally than they were 25 years ago, they are worried about some of the spillover fervor that has seen unprofitable companies getting financed, or companies using debt to fund A.I. investments, as well as the market’s concentration in just a few mammoth businesses.

 

Those concerns helped propel a sell-off last month that traders say prompted investors to move out of some of the marquee A.I. stocks and into other areas of the market.

 

Oracle lost a third of its value after revealing that it would lean on debt markets for its A.I. build-out, borrowing $18 billion in September. Luke Yang, an analyst at Morningstar, predicted Oracle’s overall debt outstanding could grow from $100 billion today to $300 billion by 2030 as a result of the company’s A.I. plans, noting a “very high risk if the A.I. demand doesn’t realize as people are expecting now.”

 

CoreWeave has fallen about 42 percent since October. The new A.I. cloud computing company is heavily dependent on big contracts with companies like Nvidia and Microsoft, but it still isn’t making any profit and has built up a large amount of debt.

 

“You do have more of a systemic risk because of the entangled nature of these companies, and you have heavy borrowing to fund” capital expenditures, Mr. Masserio said. “They are not all going to be winners. Some will be losers, and when they lose how does that knock on to other companies?”

 

There could also be new competitive pressures that alter the outlook for some of the biggest companies driving the rally. Many investors are worried about the potential for inflation to accelerate next year, which could prompt the Federal Reserve to keep interest rates elevated. Or there could simply be delays in the time it takes to build the necessary infrastructure and secure the required energy to power A.I.’s increasing use.

 

Timing if or when the bubble might burst is very difficult. Last month’s market dip was turned around again by strong earnings and greater confidence that the Fed would continue cutting rates.

 

Michael Burry, the investor heralded for forecasting the 2008 financial crisis, recently likened Nvidia’s role in the A.I. mania to Cisco’s during the dot-com boom and described OpenAI as the next Netscape, “doomed and hemorrhaging cash.”

 

“If you are worried about valuation today, then you would have been worried about valuation all year and that has cost you 17 percent performance,” Mr. Altmann of Barclays said, referring to the S&P’s rise since January.

 

“I have personally pooh-poohed the sense we are in a bubble,” he added.

 

Still, he said, “there are plenty of reasons to be nervous.”

 

Eshe Nelson contributed reporting.

 

Joe Rennison writes about financial markets, a beat that ranges from chronicling the vagaries of the stock market to explaining the often-inscrutable trading decisions of Wall Street insiders.