The company's 70 meter AI1 orbital data center prototype, which would run compute workloads in space rather than relay internet, is targeted for 2027, and no regulator has yet claimed it.
SpaceX President Gwynne Shotwell said on CNBC on June 12 that the company plans to launch its first orbital data center satellites in 2027, with smaller "canary" satellites carrying compute payloads flying earlier on Starlink buses. The full-scale AI1 spacecraft, described in company materials as roughly 70 meters long and 20 meters tall when deployed, would run continuous compute workloads in orbit rather than relay internet traffic. Astronomers, as reported by SpaceNews, are now asking which regulator should set the rules before that hardware reaches orbit.
The concern is specific to optics. At a June 4 National Academies meeting, Tony Tyson, distinguished research professor at the University of California, Davis, and chief scientist at the Vera C. Rubin Observatory, said orbital data centers would compound the brightness problems astronomers already face with Starlink. Tyson noted that the AI satellites would be "extremely bright" during low-orbit checkout before raising to operational altitude, and that the high launch rates required to build out the constellation would create continuous "bright lanes" in low Earth orbit. At the projected scale — SpaceX has an FCC application for up to 1 million satellites — he estimated sky brightness would become comparable to the glare of a half-moon, precluding most science programs. In higher operational orbits, glints from the spacecraft can reach magnitude zero, roughly as bright as Venus, interfering with time-domain astronomy that monitors dynamic events like supernovae and gamma-ray bursts.
The hardware proposal is unusual. AI1 is being framed as an orbital data center rather than a communications satellite, with power and thermal budgets closer to a small space station than to a current Starlink bus. The "canary" path matters more than it sounds: compute payloads riding Starlink hardware would put test workloads into the sky years before the 2027 flagship, while the regulatory question is still open.
That regulatory question is the part that makes this a story now. The U.S. Federal Communications Commission, which licenses radio transmissions and orbital slots for commercial space, has not announced a category for orbital data centers. The International Telecommunication Union, the U.N. body that coordinates radio spectrum across borders, has existing rules for active satellite services and for passive science bands, and it has not said how a compute satellite would be slotted. The International Astronomical Union, which sets voluntary brightness and interference norms for observatories, has not been asked. The Inter-Agency Space Debris Coordination Committee, which tracks orbital debris, has not been asked either.
SpaceX shares began trading on the Nasdaq on June 12, the same day as the CNBC interview. The combination matters: an orbital infrastructure announcement and a public-market debut landed together, with no clear rule book for the hardware in between. Shotwell, in that same broadcast, characterized the spacecraft as essentially solar panels and a radiator. That framing is the company's preferred simplification. The constructive question is not whether the simplification is wrong, but which regulator writes the checklist the spacecraft has to meet.
What to watch before 2027: an FCC filing for AI1, an ITU coordination request, a statement from NSF NOIRLab (the U.S. National Science Foundation's astronomy center) or the International Astronomical Union naming brightness or thermal limits for orbital data centers, and any move by SpaceX to publish brightness, thermal, or spectrum mitigation figures for the canary path.