CoreWeave shares plunged 16% on Wednesday as investors grew wary of the heavy borrowing used to finance artificial intelligence infrastructure, a move that pushed yields on the company's debt toward 15%. The selloff highlights a widening gap between the capital required for AI build-outs and the risk compensation investors are now demanding.
The pressure on CoreWeave, a provider of AI cloud services, was visible across both equity and fixed income markets. The yield to maturity on its 9.625% bond due in 2032 reached 14.8%, while the yield on its 9.75% bond due in 2031 climbed to approximately 14.9%. Because bond yields move inversely to prices, these rising figures indicate that investors are selling the debt, pricing in higher risk.
Luis Alvarado, co-head of global fixed income strategy at Wells Fargo Investment Institute, described a shift in sentiment where the "era of easy money" experienced in the first nine months of the year is dissipating. Alvarado noted that investors are now scrutinizing the concessions issuers must offer to secure marginal dollars, rather than accepting the lower risk premiums that characterized the earlier part of the year.
The concern was amplified by Chamath Palihapitiya, founder and CEO of Social Capital and CEO of 8090, who highlighted the soaring yields on X late Tuesday. Palihapitiya shared data showing the 2032 bond yield touched 15% before closing at a new high. He also referenced an analysis suggesting lenders are charging surprisingly little extra for AI-specific risk, raising questions about whether a private credit bubble is forming. Elon Musk, CEO of Tesla and SpaceX, responded to the post with the single word "yikes."
The selling did not stop at CoreWeave. Oracle shares fell nearly 7% on Wednesday, with the yield to maturity on its 6.25% bond due in November 2032 rising to 7.24%. Dell shares also dropped sharply, falling by 10%. The broader tech sector dragged down the Nasdaq Composite, which lost 1.4% on the day.
Analysts suggest the debt concerns are hitting companies outside the so-called "Magnificent Seven" hardest. As firms race to build AI infrastructure, they have turned to credit markets, increasing the supply of bonds and straining investor appetite. JPMorgan estimates that hyperscalers are on track to spend $800 billion this year, with annual spending expected to exceed $1 trillion by 2027.