Google Launches First Orbital AI Data Center Satellite
What happened
Google launched its Project Suncatcher MVP satellite on SpaceX's Transporter-18 rideshare Oct. 1, carrying four Trillium TPUs, built with Planet Labs, to test AI compute in orbit. The fridge-sized satellite draws about 1kW of solar power and runs Gemini queries in short bursts, powering down every 15 to 20 minutes to manage heat. Google plans two continuously operating follow-up satellites in 2027 and envisions future clusters holding 50-100 kW each. Two other Transporter-18 payloads, from Cowboy Space and Star Catcher, separately tested power-beaming to feed similar orbital-compute hardware.
Read-across for SES
Google becomes the sixth disclosed backer of orbital AI computing, alongside Starcloud, Orbital, China's Three-Body/Xingshu, Chaozhisuan and Astro Digital. SES's own space-data push rests on narrower DoD-program bets — the Starlab O3b mPower relay and K2 Space's optical-terminal work. A hyperscaler-funded entrant with its own satellite manufacturer validates the market's scale. It also crowds the field SES is trying to enter from a smaller vendor base, on a 3-to-5-year horizon.
As the brief filed it
[AGE: 9h] Google launched its first orbital AI-computing satellite on SpaceX's Transporter-18 rideshare, built with Planet Labs. It's the sixth disclosed entrant in the space-data-center race SES's own DoD optical-relay bet competes alongside.