Google launches first satellite to test AI infrastructure in orbit

Google has launched its first experimental satellite for Project Suncatcher, a multi-year research initiative to explore whether artificial intelligence infrastructure can be deployed and operated in space. The satellite, named MVP and roughly the size of a refrigerator, launched on October 1 as part of SpaceX's Transporter-18 rideshare mission aboard a Falcon 9 rocket. MVP carries four of Google's custom Tensor Processing Units (TPUs)—AI accelerators used to train models and run inference—powered by solar panels generating approximately one kilowatt of power.

The mission aims to validate whether Google's AI hardware can survive the extreme conditions of spaceflight and low Earth orbit. During launch, spacecraft components experience vibration and acceleration forces up to 10 times Earth's gravity, with individual TPU chips potentially facing 50 to 100 g-forces. Once in orbit, the hardware must contend with intense solar radiation and cosmic rays. Google conducted ground testing at UC Davis's Crocker Nuclear Laboratory using a proton beam, where Trillium TPUs demonstrated resilience to radiation levels exceeding what they would experience during a five-year mission. The company also tested thermal management systems in vacuum chambers, as heat dissipation in space requires radiators rather than conventional air cooling.

Google announced Project Suncatcher last year as a "moonshot" to eventually create constellations of solar-powered orbital data centers capable of handling AI workloads. In low Earth orbit, satellites can access near-constant sunlight, potentially generating up to eight times more solar power than ground-based installations. The company plans to launch additional custom satellites in 2027 with dozens of TPUs each, designed to communicate via high-speed laser links. This initial MVP test will operate for a few months and provide crucial in-orbit data to inform future designs.

The effort reflects broader industry interest in orbital computing as a response to terrestrial data center constraints. AI systems consume vast amounts of electricity—according to the International Energy Agency, natural gas supplied over 40 percent of US data center electricity in 2024, while renewables accounted for approximately 24 percent. Space-based solar power could address these energy demands, though the technology remains unproven at scale. MVP was developed in partnership with Planet Labs, a satellite imagery company that provided the spacecraft while Google supplied the AI hardware. Google has not disclosed project costs, though economics depend on declining launch expenses and achieving cost parity with traditional data centers.