The only real test
Google says it has conducted ground testing, but the satellite launch will provide a more complete picture of the current challenges. For example, the team will need to evaluate how the Tensor chips handle space conditions. On many space missions, the hardware is designed specifically for extreme temperatures or radiation, but these are the same TPUs that Google integrates into servers on the ground. The radiation can damage the chips or simply interfere with calculations by reversing the bits. But vibrations and high G-forces during launch also pose problems. Still, as we’ve seen with missions like the Ingenuity Mars helicopter, off-the-shelf hardware can hold up pretty well.
Testing AI chips to survive in space.
Chip cooling may be the biggest problem, something many observers have pointed out as the AI industry has gone wild over the idea of orbital data centers. Radiator systems exist in space, but they are designed to remove a relatively small amount of heat. In comparison, AI accelerators generate much more heat. Google’s solution uses a layer of malleable “thermal interface material” that connects the chips to aluminum and copper heat pipes. These components conduct heat to a radiator that can project it into space.
Google will run Gemini models on the TPUs for testing, but they won’t be able to run continuously. The cooling system will only be able to operate in short periods of about 15 minutes. After that, the TPUs must be turned off to allow the radiators to catch up.
Making sure AI chips don’t overheat in space.
With this first test, Google hopes to understand what works and what doesn’t, allowing it to improve its designs for future missions. 2027 releases are still on the agenda, but the team expects it will be years before Suncatcher evolves from “project” to “product.”
