.png){kind=logo}
Google has taken a major step in its effort to explore space-based artificial intelligence infrastructure. The company successfully placed a prototype satellite carrying four Tensor Processing Units (TPUs) into orbit as part of Project Suncatcher, its long-term research initiative to examine whether AI computing could eventually operate in space.
The refrigerator-sized satellite launched on October 1 aboard SpaceX’s Transporter-18 rideshare mission. Google developed the spacecraft in partnership with Planet. After launch, the company confirmed contact with the satellite and said it was operating as expected.
The first mission focuses on testing hardware rather than building a functioning space data centre. Google will collect data on how its TPUs respond to the physical stress of spaceflight, radiation and extreme thermal conditions. These tests will help engineers identify potential problems before attempting larger deployments.
The TPUs will also run AI workloads during the experiment. However, thermal management remains a significant challenge. Unlike terrestrial data centers, spacecraft cannot rely on conventional air-based cooling. Google therefore needs specialized systems to transfer heat from the chips and radiate it into space.
Google sees solar power as one of the biggest potential advantages of orbital computing. Satellites in low Earth orbit can receive sunlight for much of their time in orbit, creating an opportunity to generate substantial power for energy-intensive AI workloads. Google said future satellite networks could eventually use TPUs and high-bandwidth optical links to distribute machine-learning workloads in space.
The company already outlined follow-up experiments. Future missions are expected to test satellite-to-satellite communication, including high-speed laser links. The broader objective is to determine whether clusters of AI-equipped satellites could eventually operate as an orbital computing infrastructure.
