Just Announced: SpaceX and Nvidia Teaming Up on New Orbital AI Data Center

Space Exploration Technologies, trading as SPCX and better known as SpaceX, disclosed a new partnership with Nvidia as part of Elon Musk’s Starmind initiative, a plan to place AI data center capacity in orbit. The announcement arrived around the company’s Q2 earnings news and centered on Starmind AI1, described as the first satellite in a broader network intended to handle AI workloads in outer space. According to SpaceX’s post on X, the two companies are collaborating on the compute payload for Starmind AI1. The stated goal behind the orbital design is to improve power efficiency and reduce compute costs, addressing one of the biggest expense categories for conventional data centers: electricity. SpaceX’s thesis is that space offers potential advantages for energy use. Solar energy is more abundant in orbit than at the Earth’s surface, and satellite mounted solar panels that can be oriented toward the sun are presented as potentially more efficient than terrestrial solar panels. The company also points to cooling. On Earth, 30% to 40% of a terrestrial data center’s power is used to operate large cooling systems needed to keep servers from overheating while processing AI workloads. In orbit, where space is typically extremely cold, the theory is that cooling an orbital computer could require no electricity at all. Even so, the company’s vision remains subject to major engineering constraints. The article notes that these advantages are still theoretical. A computer cannot simply be placed into the vacuum of space without consequences, because it would freeze. Sunlight may be stronger in orbit, but even a large orbital solar panel produces only a fraction of the electricity generated by a solar farm. Space systems also face the lack of gravity, the absence of an air supply, and the reality that there is no technician available to replace a burnt-out fuse. Starmind will have to be designed around each of those issues. Nvidia’s role is central to that design. SpaceX said the first Starmind satellite will use Nvidia’s Vera CPUs and Rubin GPUs, along with the Vera Rubin NVL72 rackscale system. On the Q2 earnings call, Musk said, “Going forward, we've decided to build exclusively on Nvidia, because we think the Vera Rubin architecture is the best architecture. We think it's the best AI computer, and we greatly value our close cooperation and partnership on many levels with Nvidia. We're exclusive to Nvidia.” He also said, “We think the design of the NVL72 VR computer is a much better design than, say, having a standard rack-style design. We expect to actually deploy this on the ground as well as in orbit, because we think it's going to be a radical simplification of the normal NVL72 rack. It will cost less. It'll be more effective. If we're going to put it in space, why not put it on the ground? I think that's going to be pretty cool.” That exclusivity appears to be a newer development. At press time, the Starmind website still said, “We are AI chip vendor agnostic. Our system architecture supports compute modules from any provider.” Nvidia had not issued official comment on how broad its involvement will be. Musk said the first launch is not far off. On the earnings call, he said, “With respect to the Starmind AI satellite, which will be essentially an optimized Vera Rubin NVL72 computer, this is not some sort of far future distant thing. We expect to start launching this next year.” Investors, however, may want to temper expectations. Musk has a history of setting aggressive deadlines that slip, and the article characterizes the timing as aspirational. SpaceX has filed with the Federal Communications Commission to seek approval for a megaconstellation of up to 1 million Starmind satellites. That would make the system about 100 times larger than the company’s existing Starlink fleet, which took 7 years to deploy. As a result, while Nvidia’s participation in the design process is notable, the article indicates that the project will likely take years to become operational and even longer to make a meaningful contribution to SpaceX’s bottom line. For that reason, investors likely should not include Starmind as a near term driver in their thesis for SPCX stock.