Google's Project Suncatcher Seeks Scalable Machine Learning in Space
Google has confirmed that its Project Suncatcher prototype satellite is in orbit and working as expected. The satellite, built in partnership with Planet, launched aboard SpaceX's Transporter-18 rideshare mission and is designed to test the thermal performance of Tensor Processing Units (TPUs) in space.
The project is a long-term research moonshot that aims to explore whether space could one day host scalable machine learning infrastructure. The team will gather in-orbit data on how TPUs handle physical stress, radiation, and thermal extremes in space over the coming weeks.
Google's own material states that some things can only be tested in space, highlighting the importance of this mission. The satellite is a thermal experiment before it is a compute one, testing approaches such as heat pipes and radiators to solve the thermal radiator problem.
The 2027 milestone marks the next readout for Project Suncatcher, which is not announcing a product but rather stress-testing a physical assumption that could change the infrastructure equation for AI compute at scale. Google's design of its own AI accelerator silicon in the TPU and collaboration with Planet and SpaceX are key aspects of this research programme.