Google Suncatcher TPUs confirm contact after Transporter-18
Google’s Project Suncatcher prototype reached orbit on SpaceX’s Transporter-18 rideshare and confirmed contact, the company said. The Planet-built satellite carries four Trillium TPUs to test whether AI chips can run on near-constant sunlight. Households and labs still feeling Earth-side power and cooling pinch watch whether space compute is more than a moonshot.
Google confirmed contact with its Project Suncatcher prototype after the satellite launched Oct. 1, 2026, on SpaceX’s Transporter-18 Falcon 9 rideshare from Vandenberg Space Force Base. Senior Director Travis Beals said the craft, built with Planet, is operating as expected. Morning Overview and TechRadar reporting put four Trillium tensor processing units aboard the small bus — Google’s first in-orbit test of whether data-center AI chips can survive launch stress, radiation, and vacuum heat.
The stake is power and cooling. In the right low Earth orbit, Google says panels can collect up to eight times more solar energy than the same area on the ground. That only matters if the chips keep working. Ground tests at UC Davis’s Crocker Nuclear Laboratory showed Trillium hardware tolerating a total ionizing dose greater than a five-year mission, Beals’s team wrote. Vibration tests shook components at forces up to 50–100 g. In orbit, heat must leave through heat pipes and radiators with no air to help. TechRadar’s account of the prototype says the TPUs run about 15 minutes before shutting down while radiators catch up — a limit that marks this as an experiment, not a flying data center.
Backstory: Google framed Suncatcher last year as a long research moonshot toward AI constellations, with a 2027 follow-up to put two Planet satellites up and try high-bandwidth laser links between them. Engaging the work means watching Google Research posts and the peer-reviewed Joule paper for early telemetry, not treating one rideshare contact as proof of orbital cloud economics.
What we know:
Contact is confirmed after the Oct. 1 Transporter-18 launch; four Trillium TPUs are the widely reported payload count; cooling and radiation are the stated test goals.
What we don’t:
How long the TPUs ran after contact; whether the 15-minute thermal limit holds in flight; when 2027 laser-link sats will fly.
Agree (full): Google’s blog and Morning Overview agree the Suncatcher prototype launched on Transporter-18 and that Google confirmed spacecraft contact.
Differ (full): Google stresses a research first step; trade coverage leans harder on the four-chip count and short thermal run window as the practical limits of this flight.
Where outlets agree Google’s research blog and Morning Overview agree the Suncatcher prototype launched on Transporter-18 and that Google confirmed contact with the satellite.
Where they differ Google frames a long research moonshot while trade desks emphasize the four-chip payload and short thermal run window as the flight’s hard limits.