Google plans first test of AI chips in space under Project Suncatcher

Google plans to launch the first prototype satellite for Project Suncatcher, a research initiative exploring whether large-scale AI computing infrastructure could eventually be deployed in low Earth orbit (LEO). The mission will fly on SpaceX’s Transporter-18 rideshare launch, in partnership with Planet Labs. The prototype satellite will carry Google Tensor Processing Units (TPUs) and will test whether the AI chips can operate reliably in the harsh space environment, including exposure to radiation, extreme temperatures and launch forces. Radiation can cause electronic errors such as bit flips, making in-orbit testing important. Google will also evaluate a new thermal management system for cooling power-intensive AI chips in space. Because conventional airflow cooling is not possible in a vacuum, the design combines heat pipes and radiators to transfer and release heat. The first mission is primarily designed to collect in-orbit performance data and identify potential failure points, rather than demonstrate a fully operational orbital data center. Google has already tested its TPUs running AI workloads at a facility at the University of California, Davis, but says real-world orbital testing is necessary to understand how the hardware performs in space. Google plans to launch two additional satellites in 2027 to test the high-bandwidth laser communication links that could eventually connect multiple space-based computing clusters. The concept is part of a broader industry effort, with companies such as SpaceX and Starcloud exploring space-based data centers that could use near-continuous sunlight as an energy source and potentially avoid some of the electricity constraints faced by terrestrial AI data centers. However, experts say commercial deployment is still years away, mainly because of high launch costs, engineering challenges, radiation protection, thermal management and satellite manufacturing capacity.