Google Launches First Suncatcher Orbital AI Data Center Test on October 1

Google is launching its first experimental satellite for Project Suncatcher on October 1, marking a significant step toward its vision of orbital AI data centers. The satellite, dubbed MVP, will ride to space on a SpaceX Falcon 9 rocket as part of the Transporter-18 missions. Unlike custom space-rated hardware, this test integrates four standard Google TPU AI accelerators into a spacecraft built by Planet Labs, aiming to validate if off-the-shelf AI chips can survive and function in orbit.

Technical Constraints and Cooling

The primary challenge for orbital AI is heat dissipation. The MVP satellite relies on solar panels generating only 1 kilowatt of power. To manage the heat produced by the TPUs, Google uses a thermal interface material connected to aluminum and copper heat pipes that project heat into space via radiators. This system is currently limited: the TPUs can only run Gemini models for brief spurts of about 15 minutes before shutting down to allow the radiators to cool down.

Future Roadmap

This initial test is designed to evaluate how standard Tensor chips handle radiation, vibration, and thermal stress in space. While Google plans to launch two custom satellites in 2027, the company acknowledges that Suncatcher will remain a research project for several years before becoming a viable product. The long-term goal is a constellation of satellites communicating via high-speed laser links to provide scalable AI infrastructure beyond terrestrial data center limitations.