Google is set to send its first AI chips into space. The move is part of Project Suncatcher, a plan to check if AI computing can work in low Earth orbit. A prototype satellite carrying four Google Tensor Processing Units will launch aboard SpaceX's Transporter-18 rideshare mission on October 1.
Google built the satellite with Planet, a space data firm. The goal is not to build a full space data center right away. It is a technology test to see how the chips hold up outside Earth.
The mission will study how the TPUs cope with the shake and force of launch. Engineers also want to see how the chips react to radiation and extreme heat in orbit. Radiation can harm electronic parts or cause errors in chip function. Heat control is also a tough problem for powerful computer hardware in space.
To manage this, Google added a thermal system that moves heat away from the chips to radiators. These radiators release the heat into space. The system is still experimental, though, and cannot support nonstop computing. Reports say the TPUs will run in short bursts of about 15 minutes, followed by a cooling break. During tests, Google plans to run its Gemini models on the chips.
Also Read: 10 Space Technologies That Could Redefine the Future of Humanity
Project Suncatcher was first announced in 2025. The idea centers on a network of solar-powered satellites fitted with TPUs. Google sees value in placing computing power in orbit, where sunlight is constant and land-based limits do not apply. Plans include linking many satellites through fast optical connections, forming a kind of computing web above Earth.
For now, the immediate aim stays simple: find out if Google's AI hardware can survive and function in space. The data gathered from this flight will guide decisions on future missions and shed light on the real challenges of running AI systems beyond the planet.
If the test goes well, it could mark a small but telling step toward computing that stretches past Earth's surface and into orbit itself.