Google is preparing to take artificial intelligence beyond Earth by launching two prototype satellites equipped with its Tensor Processing Unit (TPU) AI chips to test whether advanced computing systems can operate effectively in space.
The initiative, known as Project Suncatcher, explores the possibility of building future AI computing infrastructure in orbit. The concept would use satellites powered by solar energy and connected through high-speed optical links, potentially allowing large-scale AI workloads to be processed outside conventional data centres on Earth.
A major focus of the experiment is determining whether Google’s AI chips can withstand the radiation, temperature changes and other extreme conditions in space. Engineers will also evaluate communication performance and the ability of multiple satellites to operate together as part of a distributed computing system.
One potential advantage of orbital AI infrastructure is access to abundant solar energy, which could reduce some of the electricity and cooling demands associated with terrestrial AI data centres. However, challenges including launch costs, maintenance, radiation exposure and reliable high-speed communication would need to be addressed before the concept could be deployed on a larger scale.
The prototype mission represents an early step towards determining whether space-based AI data centres could eventually become technically and economically practical as global demand for computing power continues to increase.
