Nvidia’s CoreWeave Capacity Deal Stops Short of a Debt Guarantee
CoreWeave had $35.6 billion in total indebtedness as of June 30; Nvidia’s $6.3 billion order form covers qualifying unused capacity.

Nvidia’s agreement to buy certain unused CoreWeave computing capacity gives the AI cloud provider a buyer for qualifying spare capacity, but does not guarantee its $35.6 billion in total indebtedness.
At an Oct. 15 event in Stockholm, Nvidia CEO Jensen Huang praised CoreWeave’s improving strength and falling cost of capital. “CoreWeave’s strength is improving every day,” Huang said. “Their cost of capital is actually decreasing,” he added, then said, “I’m optimistic.”
The companies’ Sept. 9, 2025, order form set an initial value of $6.3 billion and ties Nvidia’s purchase obligation to capacity that CoreWeave’s customers do not use. The arrangement applies to qualifying residual capacity, provided contractual delivery requirements are met and the agreement remains active.
Either party can end the arrangement after giving notice of a breach, or if specified insolvency proceedings have not been resolved within 90 days. The order commits Nvidia to purchase computing services under defined conditions; it does not assure payment to CoreWeave’s lenders.
As of June 30, 2026, CoreWeave had $35.6 billion in total indebtedness and $10 billion in undrawn availability under specified credit facilities. The undrawn amount is borrowing capacity that had not been used.
CoreWeave’s revenue backlog stood at about $104 billion on June 30. That figure excluded more than $25 billion in net new customer commitments added early in the third quarter. The backlog reflects amounts expected from customer agreements after services are provided, rather than cash already received. Delivery and service availability are conditions tied to those amounts.
Nvidia also has an equity stake in CoreWeave and a role in its expansion plans. On Jan. 23, CoreWeave issued Nvidia 22,935,780 Class A shares at $87.20 apiece for $2 billion. A collaboration framework is intended to help CoreWeave build more than five gigawatts of AI factories by 2030.
At the time of its second-quarter filing, every GPU CoreWeave used in its infrastructure was an Nvidia product. Nvidia’s capacity purchase commitment extends through April 13, 2032, with earlier termination possible under the contract’s conditions.