For a stock that has climbed more than 70% since its initial public offering this year, CoreWeave isn’t feeling a lot of love. The one thing that can change that? Profit.
CoreWeave went public in March at $40 share as a former crypto miner now focused on artificial-intelligence infrastructure. The company works with partners to construct data centers, buys Nvidia chips to fill the facilities, and leases out cloud-computing capacity to enterprises, AI labs, and hyperscalers.
CoreWeave stock more than quadrupled to $183.58 a share on June 20, rising alongside cloud names both familiar, like Oracle, and new, like Nebius Group and IREN.
Things have flipped for the stock in the second half of the year. CoreWeave shares have tumbled more than 60% since that June high, weighed down by construction delays, concerns over financing costs, and skittishness about whether the benefits of AI will justify the costs.
Shares dropped 7.9% on Monday and were down another 3.9% to $69.50 on Tuesday.
While the promised gains from AI might be realized over decades, investors are losing patience with high-growth AI stocks that haven’t yet proved they can produce a net profit. Fair or not, CoreWeave has emerged as the poster child of that trend, and it faces a make-or-break 2026 less than a year after going public.
The selloff from CoreWeave’s post-IPO peak was perhaps inevitable. The bull case for the stock was—and still is—that demand for AI compute outstrips supply, leaving CoreWeave with a seemingly endless growth horizon.
But investors seemed to realize that standing up data centers and getting them firing at full capacity is difficult, expensive work.
“The pace and scale of the AI cloud buildout has put unprecedented stress on the supply chain over the past several years,” a CoreWeave spokesperson told Barron’s. “Over that time period, CoreWeave has navigated these challenges, bringing online 41 active data centers.”
CoreWeave carried $14 billion in debt at the end of September, with recent unsecured senior notes priced at a 9% interest rate. Weather-driven construction delays, meanwhile, sent shares plummeting after the company’s third-quarter earnings report.
The cost of insuring against a CoreWeave default has also climbed. Spreads on the company’s five-year credit default swaps have risen to 803.635 as of Monday’s close, according to Bloomberg, up from 368.395 on Oct. 6. Default swaps can be volatile and thinly traded, but the upward trend on spreads has remained consistent for months.
CoreWeave has long insisted that it makes capital expenditures only when it has the customer contracts in hand to pay for them. “We’re not buying infrastructure and hoping that people come and use it,” chief development officer Brannin McBee told Barron’s in May.
Most analysts seem comfortable with that idea. Of the 33 firms polled by FactSet, just over half rate CoreWeave stock a Buy or equivalent. Analysts’ average target price is $126. But they also want results—and fast.
CoreWeave is profitable on an operating basis, but it reported a net loss of $863 million in 2024, a number analysts expect to rise to $1.09 billion in 2025. Next year is when the rubber will meet the road. Analysts forecast a loss of $337 million for 2026, most of which will come in the first half of the year. They expect a net profit by the first quarter of 2027.
“The company is in hypergrowth mode and the management team is choosing to invest,” the CoreWeave spokesperson said.
Put simply, CoreWeave has five quarters to prove it can turn AI spending into AI returns. And as the second half of 2025 has taught us, any unexpected supply shocks or signs of trouble at the company’s 40-plus data centers will be met with Wall Street’s wrath.
That’s the bargain not just for CoreWeave but also seemingly every player in the AI trade. Oracle stock, for example, has tumbled ever since it reported second-quarter earnings that blew past projections. Even for a company as big and storied as Oracle, capital expenditures are a concern.
CoreWeave has a path to becoming an AI darling once again. But it has almost no margin for error in the minds of investors.
Comments