Google’s Project Suncatcher to put ML infrastructure in space
Points and comments are a snapshot, not live.
Google's Project Suncatcher will launch a prototype satellite to test TPUs in orbit.
Project Suncatcher, announced last year, is a long-term research moonshot exploring whether space can host scalable machine learning infrastructure. In low Earth orbit, satellites can access near-constant sunlight, generating up to eight times more solar power than on Earth. A first test on SpaceX's Transporter-18 rideshare mission, developed with Planet, will gather in-orbit data on how TPUs handle spaceflight. Vibration tests showed the hardware held up, and Trillium TPUs survived a radiation dose exceeding a five-year mission at UC Davis's Crocker Nuclear Laboratory. Cooling uses heat pipes and radiators, and satellites will communicate via lasers. A 2027 milestone will put two satellites in orbit.
What commenters are saying
Skepticism dominates, with several commenters calling space data centers uneconomical and pointing to the cooling problem. One detailed calculation claims a 1 MW workload needs roughly 3,300 square meters of solar panel and 1,200 to 1,500 square meters of radiator, calling it an orbital megastructure. Others counter that it is hard but not impossible, citing liquid droplet radiators and high-temperature radiators with heat pumps. A separate thread notes SpaceX's ten-year lead in launch and satellite manufacturing, and one commenter frames the project as moving compute beyond national control.