Google tests space-based AI data centers via Project Suncatcher

Google is advancing its Project Suncatcher, an ambitious initiative to establish AI data centers in space to meet the growing computational demands of artificial intelligence. The project aims to utilize solar energy in low Earth orbit, where solar panels can produce up to eight times more energy than on Earth, while bypassing terrestrial constraints such as land, water, and power availability.

To test the feasibility of orbital computing, Google recently launched a refrigerator-sized satellite via a SpaceX rocket. Developed in collaboration with Planet Labs, the satellite carries four of Google’s custom Tensor Processing Units (TPUs) to evaluate how they handle the extreme radiation, temperature fluctuations, and vibrations of space.

Due to the lack of air in space for traditional cooling, the hardware currently operates in short 15-minute intervals to prevent overheating. Future iterations will require large-scale radiators to manage heat dissipation. For the project to become economically viable compared to terrestrial facilities, researchers suggest launch costs to low Earth orbit must drop below $200 per kilogram.