On October 1, 2026, Google reported that its Project Suncatcher prototype had launched aboard SpaceX’s Transporter-18 rideshare mission, made contact and was operating as expected at the time of the update. Planet developed the satellite with Google; SpaceX provided the launch. Suncatcher is a long-term research effort testing whether solar-powered satellites carrying Tensor Processing Units (TPUs), Google’s machine-learning chips, could eventually support computing in orbit.
The update follows NeoTeo’s earlier report on the Suncatcher launch. The focus here is what the prototype is designed to test.
What the prototype is testing
Google says the satellite is designed to gather data on how chips handle launch vibration, radiation and orbital thermal conditions. In pre-launch testing, the team shook hardware on a vibration table along three axes to simulate launch forces. It also exposed TPUs to proton beams while running AI workloads, and identified silent data corruption during those tests.
The prototype is reported to carry four AI chips. Google identifies Trillium TPUs in its radiation-testing program.
Cooling and radiation tests
Space has no air to carry heat away from electronics. Google is investigating heat pipes and pumped-fluid loops to move heat from the chips to radiators, which release heat through radiation. The prototype’s chips have been reported to run for 15 minutes at a time before being shut down to cool.
Google conducted proton-beam tests at UC Davis’s Crocker Nuclear Laboratory. The company says initial tests showed Trillium TPUs tolerating a total ionizing dose greater than the dose expected over a five-year space mission.
The next step: a planned laser test
Google plans to test laser communications between two satellites in 2027. The company likens the precision needed to hitting a coin-sized target from miles away while both endpoints are moving.
Why Suncatcher remains research
Google estimates that satellites in low Earth orbit could generate up to eight times more solar power than on Earth. Its longer-term research also explores satellite clusters, with future designs that could carry dozens of TPUs per satellite.
Scaling the idea would mean tackling heat rejection in a vacuum, radiation-related chip errors, the cost and mass of transporting solar arrays and other equipment, and the difficulty of repairing or replacing hardware in orbit. The planned two-satellite laser test is one step in exploring how a network might work.
