SpaceX commits to ground and orbital AI data centers built on NVIDIA Vera Rubin GPUs, formalizing major partnership

CEO Elon Musk announced that SpaceX will rely exclusively on Nvidia systems to power its artificial intelligence infrastructure, building all future AI services on Nvidia's Vera Rubin architecture, which Musk cited as the most advanced AI computing solution available. The announcement came during SpaceX's inaugural earnings call following its recent public offering.

SpaceX plans to scale its computing capacity from over 2 gigawatts this year to nearly 10 gigawatts by the end of next year.

The partnership will incorporate Nvidia's Vera Rubin NVL72 rack-scale system into both ground-based data centers and orbital platforms as part of SpaceX's Starmind satellite initiative. The first launches of these AI-capable satellites are scheduled to begin next year. Musk highlighted the advantages of space-based servers, including reduced land and cooling requirements, while acknowledging the operational complexities of deploying such infrastructure in orbit.

SpaceX reported AI-driven revenue of $2.6 billion for the most recent quarter, representing triple-digit year-over-year growth driven by expanded cloud service agreements and increased subscription sales in Grok and X businesses.

The announcement triggered sharp market divergence. AMD's stock rose 7% during Tuesday's trading to $518.58 but fell approximately 9% to $472.94 in after-hours trading, despite the company's recent gains and ongoing AI ambitions. Nvidia, by contrast, extended its advance, gaining 2.16% to $216.52 per share, reflecting investor recognition of the company's commanding position in advanced AI hardware.