Nvidia Halts AI Compute Revenue-Sharing Deals and Cuts OpenAI Data Center Backstop

Nvidia has suspended some agreements under an artificial intelligence cloud revenue-sharing program. The initiative, called the AI Compute Partner Program, was announced in July 2026 and was designed to provide credit support to cloud service providers in exchange for a share of revenue. Under the program, if a cloud provider could not find other customers, Nvidia would lease its GPU capacity itself, providing guaranteed revenue and helping the provider secure financing.

This week, Nvidia disclosed that such agreements, typically spanning six years, involved commitments of $36 billion. Under the proposed terms, once a cloud provider's revenue exceeded a base threshold covering costs such as chip depreciation, data centers, and personnel, Nvidia would receive 50% of the excess amount.

The related collaborations were paused last week, less than two months after the program was announced, and the exact reasons for the decision remain unknown. An Nvidia spokesperson said the new business model introduced in July 2026 was still being implemented, adding that it provided computing access to the rapidly growing artificial intelligence ecosystem.

Separately, Nvidia has recently scaled back its plan to provide a $250 billion financing backstop for OpenAI's large data center project in Ohio.