Google Launching Prototype Satellite to Test Space-Based AI Compute
Google is launching an experimental prototype satellite, named MVP, aboard a SpaceX Falcon 9 rocket under its Project Suncatcher initiative. Equipped with four custom Tensor Processing Units (TPUs), the satellite will test whether AI hardware can endure harsh space environments and process simple queries using solar power. As terrestrial data centers encounter growing power grid constraints, physical space limits, and environmental opposition, orbital data centers powered by limitless solar energy present a potential long-term alternative. If successful, orbital inference could enable new paradigms for space-based edge computing and scalable AI infrastructure. The satellite operates on just 1 kW of solar power—roughly equivalent to a household hair dryer—and packs compute capacity comparable to a single server node in a ground data center. The mission will evaluate hardware survival and query processing over one year, with the satellite remaining in orbit for up to six years before eventually re-entering and burning up in the atmosphere.
## BACKGROUND
Tensor Processing Units (TPUs) are Google's custom-developed application-specific integrated circuits (ASICs) optimized for accelerating machine learning workloads. Project Suncatcher is Google's research moonshot aimed at exploring whether scalable, solar-powered machine learning infrastructure can be deployed into Earth's orbit.