The encyclopedia · Trading & Investing · Financial decision · 2025
CoreWeave brought $13B of debt to its IPO — the market cut the price first
The biggest US tech IPO since 2021 had to price at $40 against a $47–55 range, opened below that, and raised $1.5B instead of $2.5B.
CoreWeave · 2025-03-28
What happened
CoreWeave rents out hundreds of thousands of Nvidia GPUs to AI companies, growing from a crypto-mining operation into a supplier to OpenAI, Meta and IBM. The growth ran on borrowed money: by listing day the company carried almost $13 billion of debt, and a single customer — Microsoft — accounted for 62% of revenue.
On 28 March 2025 CoreWeave went public on the Nasdaq, the largest US tech IPO since 2021. It marketed shares at $47–55, but demand forced the price down to $40 and the offering down from roughly $2.5 billion to $1.5 billion. CEO Michael Intrator called it a rightsizing for macro headwinds; the Nasdaq itself fell nearly 3% that day.
The first trade was worse than the pricing: CoreWeave opened at $39, below its already-cut IPO price, and traded under water for most of the session before closing flat at $40. Public markets had marked down the debt load and the customer concentration before they marked up the AI growth story.
Why it happened
- The company asked public markets to fund a balance sheet built for private-market exuberance: nearly $13B of debt against concentrated AI infrastructure revenue.
- One customer generated 62% of revenue. IPO buyers priced that concentration risk, which the growth narrative had not addressed.
- The listing landed in a market swoon, just as appetite for debt-funded AI capital expenditure was thinning.
The lesson
An IPO window prices the balance sheet you bring — heavy debt and one dominant customer get marked down before any growth story gets marked up.
Aftermath
CoreWeave's shares recovered within months as the AI buildout accelerated, and the debut's discount was forgotten. The listing remains the reference point for how public markets price debt-funded AI infrastructure.
Sources
spotted an error? The club wants to know.
More like this
Signature Bank was the crypto-friendly bank — and it failed two days after SVB
Silvergate was the crypto industry's bank — and the crypto winter killed it in a month
Silicon Valley Bank bet on long-term bonds — and a bank run destroyed it in 48 hours
Somewhere, someone solved the problem this company failed at. 2nd Opinion →
Comments · 0
Sign in to join the comments.