Google is preparing to launch its Tensor Processing Units (TPUs) into low Earth orbit on October 1, 2026, as part of Project Suncatcher. The prototype satellite will be carried by SpaceX’s Transporter-18 rideshare mission, launching from California. Planet Labs is collaborating with Google on the spacecraft effort.
The primary goal of Project Suncatcher is to test how AI hardware performs in the harsh environment of space. The TPUs will be subjected to radiation, extreme temperatures, and the intense vibrations of a rocket launch. Cosmic rays and solar activity can disrupt electronics and create errors known as bit flips. Google has already conducted ground tests, including proton beam radiation testing at UC Davis, where AI workloads were run while researchers monitored radiation-related errors. Vibration tests across all three satellite axes were also performed, and the hardware reportedly held up during these stress tests.
One of the biggest challenges for space-based AI computing is cooling. In the vacuum of space, conventional fans cannot dissipate heat. Google plans to use heat pipes and radiators to transfer heat away from the TPUs, a method already tested in thermal vacuum chambers.
Another advantage of orbital computing is access to near-constant solar power. Satellites in low Earth orbit can receive up to eight times more solar energy than ground-based installations, making space an attractive location for energy-intensive AI workloads.
Looking ahead, Google envisions multiple satellites working together as orbital computing clusters, connected by high-bandwidth laser links. The company plans to launch two satellites in 2027 to test these laser communication systems.
The upcoming launch is not a full-scale AI data center in space but a testbed to gather real-world data on hardware reliability and operational limits. The findings will inform the design of future space-based AI systems, with commercial deployment still a long-term possibility.