SpaceX is taking a big step toward turning a long-rumored idea into something more concrete: an orbital data center built from satellites. Following earlier hints from Elon Musk, the company has now filed with the U.S. Federal Communications Commission (FCC) for permission to deploy a massive new satellite constellation designed to support high-demand computing workloads, potentially tied to Musk’s artificial intelligence ambitions.
The proposal describes an “Orbital Data Center” system operating in Low Earth Orbit, with satellites placed roughly 500 to 2,000 kilometers above the planet. That’s a busy neighborhood in space, and SpaceX appears to be aware of the added congestion. The filing notes that satellites would be spaced out to “deconflict” operations—especially in cases where control issues occur between satellites, or if other companies launch similar orbital data center networks in the future.
The timing is notable, because SpaceX has also had to deal with recent satellite incidents. The company previously reported a satellite anomaly and worked to reassure that its trajectory would stay below the International Space Station, after the spacecraft behaved unpredictably before ultimately reentering Earth’s atmosphere and burning up. While such events aren’t unprecedented in large constellations, they highlight why traffic management and safety planning matter even more if orbital infrastructure expands dramatically.
So what exactly is SpaceX pitching? In essence, a next-generation satellite swarm that behaves less like a traditional communications network and more like compute infrastructure in space. The satellites would use optical inter-satellite links—similar in concept to the laser-based connections used by Starlink—to move data between spacecraft. However, these orbital data center satellites are described as much smaller than typical broadband satellites, with different hardware configurations depending on the orbital “shell” they’re assigned to and the role they perform.
The big promise is AI compute capacity without the same strain on Earth-based power grids. Space-based systems can tap solar energy continuously, and they could, at least in theory, reduce the burden created by energy-hungry data centers on the ground. SpaceX’s filing makes a bold case for scale, suggesting that launching large amounts of satellite mass each year—paired with high compute density—could add enormous AI computing capability annually, while requiring minimal ongoing operations and maintenance once deployed.
Of course, turning orbit into a practical home for computing isn’t as simple as adding more satellites. Even if solar power is abundant, challenges remain: latency compared to ground data centers, thermal management and cooling in space, the increasing complexity of collision avoidance, and the long-term risk of debris in already crowded Low Earth Orbit. SpaceX’s proposal acknowledges the need for coordination and spacing, but the sheer size of the constellation being discussed will likely intensify debate over sustainability in LEO.
Still, the FCC filing signals that this isn’t just a thought experiment anymore. If SpaceX follows through, an orbital data center could become one of the most ambitious attempts yet to move AI infrastructure off-planet—reshaping not only space operations, but also how and where the world powers the next wave of artificial intelligence computing.
