Google’s Project Suncatcher is exploring whether AI computing infrastructure could eventually be deployed in orbit. On October 1, the project’s first prototype satellite was launched aboard SpaceX’s Transporter-18 mission. Google later confirmed that it had established contact with the satellite and that it was operating as expected.
The idea is to build a network of solar-powered satellites equipped with Google’s Tensor Processing Units (TPUs), which are designed for AI workloads. In the long term, these satellites could potentially be connected through high-speed optical links and work together as a distributed computing system.
According to Google, satellites in low Earth orbit could receive up to eight times more solar energy than equivalent solar infrastructure on Earth. This could provide a potential source of power for large-scale AI computing in space.
However, putting AI chips into orbit is only the beginning. Keeping them operational for long periods presents much bigger engineering challenges. The TPUs must withstand radiation, extreme temperature changes and the vibrations experienced during launch. Cooling is another major issue, as conventional air-based cooling does not work in the vacuum of space. Google is testing heat pipes and radiators to manage the heat generated by the computing hardware.
The next major step is planned for 2027, when Google aims to launch two satellites and test high-speed optical communication between them. In the longer-term concept, each satellite could carry dozens of TPUs.
Project Suncatcher is not yet a space-based data center. Google is still testing whether the underlying technology can work reliably in orbit and whether multiple satellites can be connected into a practical computing system.
If these challenges can be solved, AI infrastructure could eventually extend beyond traditional data centers on Earth to computing systems operating in orbit.
