Project Suncatcher First Tests Whether AI Hardware Can Survive Orbit
Project Suncatcher is Google's long-term research effort to explore machine learning infrastructure in space. Its first prototype satellite is planned for SpaceX's Transporter-18 rideshare mission, developed with Planet, and will collect real-world data on Google Tensor Processing Units (TPUs) during launch and in orbit.
Launch vibration and acceleration create substantial mechanical stress for the chips, while solar events and cosmic rays can cause electronic errors in orbit. Google says ground-based proton-beam tests indicate its Trillium TPUs can withstand a total ionizing dose greater than the expected exposure of a five-year space mission, but only an orbital test can show how that result holds in practice.
Cooling and Connectivity Are the Hard Parts of Space AI
Sunlight in orbit could offer an attractive power source, but vacuum eliminates air convection, so heat from densely packed TPUs must be removed through systems such as heat pipes and radiators. Google has tested cooling approaches in a thermal-vacuum chamber and plans to refine them with the first mission. A longer-term concept would place dozens of TPUs on satellite clusters linked by high-bandwidth laser connections.
Project Suncatcher Is Still an Early Feasibility Test
This mission will not prove that space-based data centers are commercially viable, nor that large AI workloads can run reliably there. Precise satellite pointing, sustained thermal management, radiation tolerance, and maintenance costs remain unresolved. Google plans to test laser communication between two satellites in 2027; until those orbital results arrive, space compute is best understood as an engineering research direction rather than a ready replacement for terrestrial infrastructure.