The AI trade's credit check

Apparently, “all the money in the world” gets a little harder to find when the debt you already issued is flashing distress.

CoreWeave hopes to eventually earn an investment-grade credit rating, giving it access to much deeper pools of insurance money, co-founder Brannin McBee told the Financial Times this week in a quote he probably wishes he could take back.

The AI cloud-computing giant has relied heavily on debt to fund its breakneck expansion, just as investors are questioning how much more leverage the funding model can support. CoreWeave shares plunged as much as 17% Wednesday, while bonds due in 2032 sank to roughly 78 cents on the dollar. The cost of insuring its debt now implies about a 63% chance of default within five years, finds S&P Global data.

The result isn’t your ordinary AI stock selloff. Instead, it’s turned one of AI’s fastest-growing infrastructure companies into a stress test for a much bigger question hanging over the boom.

Debt keeps the GPUs coming

Renting out Nvidia-powered computing capacity to customers like OpenAI and Microsoft, CoreWeave’s business model requires enormous upfront spending on GPUs and data centers and relies on long-term contracts generating enough revenue to cover the bill.

That math gets tougher as borrowing costs rise. CoreWeave carries roughly $50 billion of debt, could spend as much as $39 billion on capex this year, and is still expected to lose more than $2 billion. The company says its backlog comfortably covers obligations, and Nvidia CEO Jensen Huang said he’s “optimistic” about its financial position.

But investors are less convinced. If financing gets too expensive, CoreWeave may have to slow its expansion even while demand for AI computing remains strong. Given CoreWeave is a major customer, Nvidia was also pulled into the selloff, falling 1.84% today. The Philadelphia Semiconductor Index has lost more than $1.6 trillion in market value during the recent rout.

Competing for everyone’s money

Although CoreWeave is just one company, AI infrastructure requires enormous amounts of capital. Gartner estimates that AI providers will spend nearly $1 trillion this year and another $6 trillion between 2027 and 2030, while warning that their revenue may not justify that spending.

Citi analyst Stephanie ​von Friedeburg says competition for capital is already showing up in sovereign debt markets, with some borrowers turning more aggressively to non-dollar bonds as funding tightens. That doesn’t mean your mortgage rate is rising because someone built another data center—but it does mean AI is becoming large enough to influence borrowing costs.

For investors, that changes what matters in the AI trade. Companies have to prove they can finance the buildout without crushing returns, giving cash-rich giants a different risk profile from debt-heavy players like CoreWeave. The bottom line: The next phase of the AI boom could come down to who can afford to keep spending.—HC