Google Launches TPU Satellite to Test Orbital Data Center Ambitions
Google's prototype orbital compute satellite launched today onboard a SpaceX rocket from California, marking the first time the tech giant sends one of its advanced chips into space. Built by Planet Labs, the satellite will prove that Google's Tensor Processing Unit, its competitor to Nvidia's GPUs, can function in orbit. That means supplying a kilowatt of continuous power, cooling the chip, and running a series of models to see if anything goes wrong. "We've done testing on the ground, but there's no test that's completely as good as the real thing," says Travis Beals, the Google executive managing Project Suncatcher, the company's plan to develop large-scale compute clusters in orbit. Once commissioned, the satellite will fire up its TPU in 15-minute bursts to avoid straining the satellite's power and thermal management systems.
This satellite rides on a standard Planet Labs platform, but the two companies are working on a demo expected to launch next year featuring two purpose-built satellites capable of running more substantial workloads. Those future versions will attempt to collaborate via a laser communications link. Suncatcher is not the only space AI payload on this SpaceX rocket, which carries more than 100 payloads, including missions from Satlyt and Cowboy Space Company. What sets Google's initiative apart from those startups — and from SpaceX itself — is that it is a long-term project focused on building for the space infrastructure and AI workloads of the future.
Google envisions an orbital data center consisting of a network of 81 satellites flying in close formation and processing in parallel. "The bandwidth and the latency between TPUs really matters when you're trying to run a multi-rack workload," Beals says, explaining that the company looks ahead to where workloads will be in five years. That long view is largely because the rockets required to scale up orbital data centers cost-effectively do not yet exist — Google's white paper on orbital data centers, published in Joule this week, suggests SpaceX's Starship may need to launch 1,600 times before such projects get off the ground. The peer-reviewed paper offers one of the most rigorous analyses available of how compute reaches orbit.