Meet Project Suncatcher, a research moonshot to scale machine learning compute in space

One fundamental technology, artificial intelligence, has the potential to assist us in addressing the greatest challenges facing humanity. Now, we want to know where we can take it to realize its full potential. Project Suncatcher, our new research moonshot to one day scale machine learning in space, is announced today. Utilizing our Tensor Processing Unit (TPU) AI chips, we are investigating how an interconnected network of solar-powered satellites, working backward from this possible future, might maximize the Sun’s power. We have begun the foundational work required to one day make this future possible, inspired by other Google moonshots like autonomous vehicles and quantum computing. We’re excited that this is a growing area of exploration, and our initial research, shared today in a preprint paper, describes our approach to satellite constellation design, control, and communication, and also our initial learnings from radiation testing Google TPUs.

Our next step is a learning mission with Planet to launch two prototype satellites to test our hardware in orbit by early 2027, laying the groundwork for massively scaled space computation in the future.