CoreWeave’s sell-off puts the cost of AI’s computing boom in focus

CoreWeave’s shares fell 25% in a week as investors worried about rising borrowing costs and the debt used to build its AI data centres. The sell-off puts a practical question behind the AI infrastructure boom: can demand for computing keep pace with the cost of financing it?

CoreWeave Watch analysis

What happened

The Taipei Times, carrying Bloomberg’s report, says CoreWeave’s shares fell 20% in the first three sessions of last week, recovered on Thursday after Nvidia chief executive Jensen Huang expressed confidence, then dropped 12% on Friday. The shares finished the week down 25%.

The report says CoreWeave finances data-centre construction by issuing bonds, leaving it exposed to higher US Treasury yields and borrowing costs. On Thursday morning, the cost of protecting its bonds against default reached a record intraday high; credit-default swaps implied a greater than 60% chance of default within five years. That market pricing is a measure of investor concern, not a prediction that default will happen.

Why it matters

CoreWeave sits near the centre of the AI computing build-out, but its growth requires substantial financing. If debt becomes more expensive or harder to raise, that could complicate plans to expand capacity even while demand for AI computing remains strong.

The report also sets out the counter-case: demand is still said to outstrip supply, and some analysts point to CoreWeave’s customer base and backlog. A sharp share-price fall is not proof that the business is failing. It is a sign that investors are weighing the growth story against the cost and risk of funding it.

Our read

This is a useful stress test for the AI infrastructure boom, not a verdict on CoreWeave. A company can have eager customers and still face a tougher financing equation when interest rates rise. The key distinction is whether demand becomes dependable cash flow, rather than simply more buildings and more borrowing.

What to watch

• Whether CoreWeave can keep financing data-centre growth at manageable cost.
• Whether customer demand and backlog translate into revenue and cash flow.
• How the company’s borrowing costs and credit-market signals move from here.

Discussion spark: If AI computing demand remains strong, is that enough to justify the debt used to build capacity, or should investors demand proof of cash flow first?

not affiliated with or endorsed by CoreWeave