Google Launches Project Suncatcher to Test Orbital AI Data Centers
Google launched a prototype satellite on October 1 to test if AI chips can operate in space to bypass terrestrial energy and land constraints.
On October 1, Google launched a prototype satellite from Vandenberg Space Force Base via a SpaceX Falcon 9 rocket as part of Project Suncatcher. Developed in partnership with Planet Labs, the mission tests the feasibility of hosting scalable machine learning infrastructure in low Earth orbit. The satellite carries four Trillium Tensor Processing Units (TPUs) to evaluate how AI silicon withstands radiation, thermal extremes, and the physical stress of launch.
To address the lack of air for cooling in a vacuum, Google implemented a system of heat pipes and radiators using aluminum and copper. Despite these measures, the hardware currently requires a cooldown period every 15 to 20 minutes. Prior to launch, the chips underwent vibration testing in San Francisco and radiation exposure at the University of California Davis Crocker Nuclear Laboratory, where they survived doses exceeding those of a five-year space mission.
Google aims to leverage near-constant sunlight in space, which can provide up to eight times more solar power than on Earth, to reduce the environmental impact and power grid strain of terrestrial data centers. The company plans to launch two additional satellites in 2027 to test high-bandwidth laser interconnectivity, with a long-term goal of deploying fleets of over 80 satellites.
Parallel efforts are emerging across the industry. SpaceX is developing its own orbital data center swarms, with COO Gwynne Shotwell targeting supercompute capabilities in space by 2027. Additionally, the startup Satlyt recently raised $8 million to develop software that allows existing satellites to process AI applications locally.
