Google is set to launch its first experimental satellite for “Project Suncatcher” on October 1, marking a significant step toward its goal of building orbital AI data centres powered by near-constant sunlight.
The refrigerator-sized satellite, named MVP, will carry four of Google’s custom Tensor Processing Units (TPUs) and launch aboard a SpaceX Falcon 9 as part of the Transporter-18 rideshare mission. The spacecraft was built in partnership with satellite imagery firm Planet Labs.
Project Suncatcher was announced last year, aimed at investigating whether space can host scalable machine learning infrastructure. In low Earth orbit, satellites can access up to eight times more solar power than on Earth, potentially sidestepping the energy constraints facing terrestrial data centres.
The mission will assess how Google’s TPUs withstand launch forces, radiation, and thermal extremes. Google said its Trillium TPUs held up well in ground tests, surviving radiation doses greater than what a five-year space mission would deliver.
But cooling remains a critical challenge. In space, there’s no airflow; heat must be diffused via radiators. Google’s design combines heat pipes and radiators, but the system can only run in roughly 15-minute bursts before the TPUs must shut down to let radiators catch up.
The satellite will run Gemini models for testing. Future designs envision clusters of satellites carrying dozens of TPUs each, communicating via high-bandwidth lasers. Google plans to launch two satellites in 2027 to test those laser links.
Experts caution the concept remains years from commercial viability due to high launch costs and engineering constraints. Google stressed the first mission is about gathering data and recognising and identifying failure points, not showcasing an operational orbital data centre.
2026-09-24 23:06:00
