CoreWeave’s early rally faded Monday as investors continued to question the cost of financing the AI infrastructure boom.
CoreWeave Inc (NASDAQ: CRWV) got a bullish call from Goldman Sachs on Monday, but the upgrade did little to settle nerves around the debt-heavy AI infrastructure company.
Goldman raised its rating on CoreWeave to Buy from Neutral, even as it cut its price target to $106 from $139. Shares jumped in premarket trading before giving back the gains. By midday, CRWV had slipped below $61, down roughly 1% on the session and about 8% from its intraday high as trading volume picked up.
Thomas Hayes, chairman of Great Hill Capital, questioned the timing of Goldman’s call during an interview with Joel Elconin on Stock Trader Network’s Think Tank.
“How do you come out and upgrade CRWV?” Hayes asked. “Could they run this thing back up? Sure. If you’re right, maybe you get back to recent all-time highs. If you’re wrong, you could face some serious capital impairment.”
Goldman analyst Gabriela Borges expects CoreWeave’s annual recurring revenue per megawatt to increase over the next two years as tight supply and the company’s performance advantages support higher returns. Goldman also expects borrowing costs to rise by 100 to 200 basis points, but believes the impact can be managed.
The risks are harder to ignore after last week’s selloff though.
CoreWeave lost roughly a quarter of its market value as investors focused on its approximately $50 billion in debt and the rising cost of financing its expansion. Yields on some of its bonds approached 16%, signaling growing concern among creditors. The company reported a $1.2 billion net loss on $5.1 billion in revenue in 2025, while its capital spending is expected to climb sharply as it builds out data center capacity.
The pressure has spread beyond CoreWeave to other neocloud providers and semiconductor stocks, as investors question whether the returns from AI infrastructure will justify the enormous upfront investment.
Dennis Dick at Stock Trader Network described the selling in semiconductor stocks at the open as a market where these names are “guilty until proven innocent.”
Hayes sees a potential rotation within the AI trade itself, expecting companies applying AI to their existing businesses, rather than supplying the infrastructure, to outperform over the next 12 to 24 months.
“I think you’re going to see many of these stocks in the other 490 that have been beaten down double and triple over the next 12 to 24 months,” he said, referring to the S&P 500 excluding its 10 largest AI-related names.
This next phase could favor businesses using AI to improve margins and generate cash, according to Hayes, while investors scrutinize the returns on capital spent by hyperscalers and infrastructure providers.
