SpaceX will use Nvidia GPUs exclusively, citing their quality. SpaceX's Vera Rubin AI supercomputer will be launched next year.

SpaceX has struck an exclusive agreement to use Nvidia GPUs across its AI compute buildout, with CEO Elon Musk saying the chips will power the company's next-generation infrastructure "because they are the best". The centerpiece is Nvidia's Vera Rubin NVL72 rack-scale system, which SpaceX plans to launch into orbit next year as part of the Starmind AI1 satellite compute payload.

Musk has set a target of roughly 2 gigawatts of space-based compute capacity online by the end of 2026, scaling toward 10 gigawatts by the end of 2027. The exclusivity commitment also surfaced on SpaceX's own earnings call, underscoring how central [NVDA](/stocks/NVDA)'s architecture has become to the company's AI ambitions beyond its core launch and Starlink businesses.

The news lifted Nvidia shares as investors weighed the added demand signal from a customer building compute infrastructure in space, a market with essentially no established competitors. For SpaceX, betting its AI roadmap on a single supplier ties its Vera Rubin timeline to Nvidia's own production and orbital deployment progress, a dependency with no track record since no AI supercomputer has previously been launched into space.

SpaceX Forms Exclusive Partnership with Nvidia for AI and Space Compute

Related Stocks