Story
Oct 18, 2026
CoreWeave’s Bond Rout Puts AI’s Debt Boom on Trial
CoreWeave’s soaring borrowing costs have become a stress test for the AI infrastructure race. Bears see a debt-fueled model cracking under pressure; backers point to contracted revenue from major customers and resilient demand.
The alarm first sounded in the bond market, where investors began questioning whether the borrowing behind the AI infrastructure boom would pay off. Credit analysts warned that debt markets tend to expose the weak spots in speculative cycles before equities do, with one researcher calling neocloud operators such as CoreWeave “the subprime tranche of the AI trade.”1
That unease quickly became a CoreWeave-specific rout. The company borrows heavily to build data centers and lease computing capacity, but had roughly $35 billion of debt and $16 billion in lease liabilities at the end of June. Its negative free cash flow reached $5.7 billion in its latest quarter. Bernstein analyst Madison Rezaei’s verdict was blunt: CoreWeave needs lenders willing to finance its hardware “on terms that leave something attractive for shareholders.”2
As bond yields approached 16% and five-year credit-default swaps topped 1,000 basis points, the market’s message was stark: insuring CoreWeave debt had become extraordinarily expensive. One estimate put the implied five-year default odds at about 60%, even though a default remains far from certain.3
The bearish case rests on a simple fear: demand could soften before the company’s obligations do. Accountant and journalist Francine McKenna disputed the idea that Microsoft would rescue CoreWeave if the strain intensified, saying the software giant would act in its own interest.4
But CoreWeave’s defenders argue that its backlog changes the equation. Investor Evan Schlossman pointed to $103.7 billion in remaining performance obligations, much of it tied to non-cancellable, take-or-pay contracts with customers including Microsoft and OpenAI. CoreWeave said 98% of revenue comes from committed contracts and that customer cash flows are structured to repay associated debt.5
Nvidia chief Jensen Huang also called the company’s position “stable” and “improving every day,” while David Sacks amplified the view that the selloff reflected volatility around vast AI-growth bets rather than collapsing demand.
6 The next test is whether credit costs retreat—or force CoreWeave to slow the buildout that made it central to AI’s ambitions.