Google Sends AI Processors to Space to Assess Feasibility of Orbital Data Centers
Google is preparing to launch a satellite with four of its AI chips into low Earth orbit to test their resilience in space. The experiment, part of Project Suncatcher, will gather data on how the hardware handles radiation and launch stresses. Google hopes the results will advance its vision of building AI data centers in space, which could tap into abundant solar energy.
The prototype satellite will ride on SpaceX’s Transporter-18 rideshare mission, developed alongside Planet Labs. Prior to launch, Google subjected its Trillium TPUs to proton beam testing at UC Davis’s Crocker Nuclear Laboratory, simulating radiation exposure exceeding what a five-year space mission would typically deliver. The chips reportedly performed well despite the bombardment.
Cooling presents a distinct hurdle, as vacuum conditions eliminate airflow-based thermal management. Google is experimenting with heat pipes and radiators, validated in thermal vacuum chambers that replicate space’s combined heat and vacuum conditions. Looking ahead, the company plans twin satellite launches in 2027 to test high-bandwidth laser links intended for future orbital clusters.
This experiment could reshape how society balances growing AI demand against environmental costs. If orbital data centers prove viable, they may reduce terrestrial energy strain and land-use conflicts, potentially lowering the carbon footprint of cloud computing. However, space-based infrastructure raises concerns about orbital debris, launch costs, and concentrated control of computing resources. The outcome may influence whether future AI expansion happens in orbit or remains grounded, affecting energy markets, tech investment, and environmental policy worldwide.