Google Prepares First Project Suncatcher Satellite to Test AI Computing in Orbit
Google plans to launch the first experimental Project Suncatcher satellite on October 1, carrying four Tensor Processing Units to run Gemini workloads in orbit. The refrigerator-sized spacecraft, developed with Planet, is scheduled to fly on SpaceX’s Transporter-18 rideshare mission and will gather data on how TPUs handle launch stresses, radiation, and temperature swings. Google says the test will inform a larger Suncatcher milestone planned for 2027.
Google's first Suncatcher spacecraft, named MVP, is scheduled for an October 1 flight on a Falcon 9 through SpaceX's Transporter-18 rideshare. It will carry four Google-designed AI chips and solar panels meant to generate around one kilowatt. The effort follows Suncatcher's November 2025 introduction and brings forward an earlier plan to test two Planet-built prototypes by early 2027.
The mission aims to produce operational evidence ahead of a bigger 2027 goal. Google is also entering a nascent orbital-computing contest: SpaceX has outlined an AI1 satellite design with about 120 kilowatts of continuous computing capacity and has asked the FCC about a potentially vast satellite network.
If orbital AI computing advances, cloud customers, satellite operators, and communities near terrestrial data centers could be affected. Moving some workloads off Earth may reduce demand for land, power, and water in some regions, though launch costs, space debris, and orbital operations may create new challenges. Google's early test may shape how quickly such systems become practical and who can access them.