Musk Aims for Orbital Nvidia Launch Next Year Amid Physics and Timing Concerns
SpaceX CEO Elon Musk recently announced on X that he is 'highly confident' that SpaceX will launch Nvidia VR NLV72 AI computers in space next year. This announcement has reignited a debate over whether orbital data centers are closer to reality or still mostly a pitch deck.
The claim refers to a space-adapted version of Nvidia's Vera Rubin NVL72 rack, which is built around 36 Vera CPUs and 72 Rubin GPUs. Musk has called the design 'a radical simplification of the normal NVL72 rack' and said SpaceX would deploy versions both on the ground and in orbit.
However, SpaceX's own investor materials paint a more cautious picture. The company aims to launch initial orbital AI-computing demonstrations by late 2027, with its IPO filing citing 'as early as 2028' for broader deployment. This means Musk's public 'next year' framing sits well ahead of that internal roadmap.
The physics of space also pose an obstacle. In space, there is no atmosphere, so conduction and convection are unavailable; heat can only be shed through radiators, and a rack pulling the power of dozens of Rubin GPUs generates far more waste heat than any system flown to date.