Wall Street eyes AI bubble as skepticism grows over trillion-dollar bets

It’s been three years since OpenAI set off euphoria over artificial intelligence with the release of ChatGPT. And while the money is still pouring in, so are the doubts about whether the good times can last.

From a recent selloff in the shares of Nvidia Corp., to Oracle Corp.’s plunge after reporting mounting spending on AI, to souring sentiment around a network of companies exposed to OpenAI, signs of skepticism are increasing. Looking to 2026, the debate among investors is whether to rein in AI exposure ahead of a potential bubble popping or double down to capitalize on the game-changing technology.

“We’re in the phase of the cycle where the rubber meets the road,” said Jim Morrow, chief executive officer of Callodine Capital Management. “It’s been a good story, but we’re sort of anteing up at this point to see whether the returns on investment are going to be good.”

The queasiness about the AI trade involves its uses, the enormous cost of developing it, and whether consumers ultimately will pay for the services. Those answers will have major implications for the stock market’s future.

The S&P 500’s three-year, $30 trillion bull run has largely been driven by the world’s biggest tech companies like Alphabet Inc. and Microsoft Corp., as well as firms benefiting from spending on AI infrastructure like chipmakers Nvidia and Broadcom Inc., and electricity providers such as Constellation Energy Corp. If they stop rising, the equities indexes will follow.

“These stocks don’t correct because the growth rate goes down,” said Sameer Bhasin, principal at Value Point Capital. “These stocks correct when the growth rate doesn’t accelerate any further.”

Of course, there are still plenty of reasons for optimism. The tech giants that account for much of the AI spending have vast resources and have pledged to keep pumping in cash in the years ahead. Plus, developers of AI services, like Alphabet’s Google, continue to make strides with new models. Hence the debate.

Here’s a look at the key trends to watch while navigating through these choppy waters.

Access to Capital

OpenAI alone plans to spend $1.4 trillion in the coming years. But the Sam Altman-led company, which became the world’s most valuable startup in October, is generating far less revenue than its operating costs. It expects to burn $115 billion through 2029 before generating cash in 2030, The Information reported in September.

The company has had no problem with fundraising so far, collecting $40 billion from Softbank Group Corp. and other investors earlier this year. Nvidia pledged to invest as much as $100 billion in September, one of a series of deals the chipmaker has made that funnel cash to its customers, which is causing fears of circular financing in the AI industry.

OpenAI could run into trouble if investors start to balk at committing more capital. And the consequences would spiral to the companies in its orbit, like computing-services provider CoreWeave Inc.

“If you think about how much money — it’s in the trillions now — is crowded into a small group of themes and names, when there’s the first hint of that theme even having short-term issues or just valuations get so stretched they can’t possibly continue to grow like that, they’re all leaving at once,” said Eric Clark, portfolio manager at the Rational Dynamic Brands Fund.

Plenty of other companies are reliant on external funding to pursue AI ambitions. Oracle shares soared as it racked up bookings for cloud computing services, but building those data centers will require massive amounts of cash, which the company has secured by selling tens of billions of dollars in bonds. Using debt puts pressure on a company because bondholders need to be paid in cash on a schedule, unlike equity investors, who mostly profit as share prices rise.

Oracle’s stock got pounded on Thursday after the company reported significantly higher capital expenditures than expected in its fiscal second quarter and cloud sales growth missed the average analyst estimate. On Friday, a report that some data center projects it’s developing for OpenAI have been delayed sent Oracle’s shares down further and weighed on other stocks exposed to AI infrastructure. Meanwhile, a gauge of Oracle’s credit risk hit the highest level since 2009.

An Oracle spokesperson said in a statement that the company remained confident in its ability to meet its obligations and future expansion plans.

“The credit people are smarter than the equity people, or at least they’re worried about the right thing — getting their money back,” said Kim Forrest, chief investment officer at Bokeh Capital Partners.

Big Tech Spending

Alphabet, Microsoft, Amazon.com Inc. and Meta Platforms Inc. are projected to spend more than $400 billion on capital expenditures in the next 12 months, most of it for data centers. While those companies are seeing AI-related revenue growth from cloud-computing and advertising businesses, it’s nowhere near the costs they’re incurring.

“Any plateauing of growth projections or decelerations, we’re going to wind up in a situation where the market says, ‘Ok, there’s an issue here,’” said Michael O’Rourke, chief market strategist at Jonestrading.

Earnings growth for the Magnificent Seven tech giants, which also includes Apple Inc., Nvidia and Tesla Inc., is projected to be 18% in 2026, the slowest in four years and slightly better than the S&P 500, according to data compiled by Bloomberg Intelligence.

Rising depreciation expenses from the data center binge is a major worry. Alphabet, Microsoft and Meta combined for about $10 billion in depreciation costs in the final quarter of 2023. The figure rose to nearly $22 billion in the quarter that just ended in September. And it’s expected to be about $30 billion by this time next year.

All of this could put pressure on buybacks and dividends, which return cash to stockholders. In 2026, Meta and Microsoft are expected to have negative free cash flow after accounting for shareholder returns, while Alphabet is seen roughly breaking even, according to data compiled by Bloomberg Intelligence.

Perhaps the biggest concern about all the spending is the strategy shift it represents. Big Tech’s value has long been premised on the companies’ ability to generate rapid revenue growth at low costs, which resulted in immense free cash flows. But their plans for AI have turned that upside down.

“If we continue down the track of lever up our company to build out for the hopes that we can monetize this, multiples are going to contract,” said Jonestrading’s O’Rourke. “If things don’t come together for you, this whole pivot would have been a drastic mistake.”

Rational Exuberance

While Big Tech’s valuations are high, they’re nowhere near excessive compared to past periods of market euphoria. Comparisons to the dot-com bust are common, but the magnitude of the gains from AI are nothing like what happened during the development of the internet. For example, the tech-heavy Nasdaq 100 Index is priced at 26 times projected profits, according to data compiled by Bloomberg. That figure exceeded 80 times at the height of the dot-com bubble.

Valuations during the dot-com era were far in excess of where they are now partly because of how far the stocks had run, but also because the companies were younger and less profitable — if they had profits at all.

“These aren’t dot-com multiples,” said Tony DeSpirito, global chief investment officer and portfolio manager of fundamental equities at BlackRock. “This isn’t to say there aren’t pockets of speculation or irrational exuberance, because there are, but I don’t think that exuberance is in the AI-related names of the Mag 7.”

Palantir Technologies Inc., which trades at a multiple of more than 180 times estimated profits, is among the AI stocks with nosebleed valuations. Snowflake Inc. is another, with a multiple of almost 140 times projected earnings. But Nvidia, Alphabet and Microsoft are all below 30 times, which is relatively tame considering all the euphoria surrounding them.

All of which leaves investors in a quandary. Yes, the risks are right on the surface even as investors keep pouring into AI stocks. But for now, most companies aren’t priced at panic-inducing levels. The question is which direction the AI trade goes from here.

“This kind of group thinking is going to crack,” Value Point’s Bhasin said. “It probably won’t crash like it did in 2000. But we’ll see a rotation.”

Wittenstein writes for Bloomberg.