Google is developing plans to place artificial intelligence datacentres in space. The company aims to launch its first trial equipment into orbit in early 2027. According to The Guardian, scientists and engineers believe tightly packed groups of about 80 solar-powered satellites could operate about 400 miles above Earth. These satellites would carry powerful processors needed to meet rising AI demand.
Project Suncatcher Takes Shape
Google released research on Tuesday showing that falling space launch prices could make orbital datacentres cost-competitive with Earth-based ones by the mid-2030s. The project, called Suncatcher, would use solar panels up to eight times more productive than those on Earth. Once in orbit, the datacentres would send results back through optical links using light or laser beams.
The company plans to launch two prototype satellites by early 2027. Google said the research marks a first milestone toward scalable space-based AI. The satellites would minimize impact on land and water resources needed to cool existing datacentres.
Challenges and Competition
Google noted that significant engineering challenges remain. These include thermal management, high-bandwidth ground communications, and on-orbit system reliability. Each rocket launch emits hundreds of tonnes of carbon dioxide. Astronomers have raised concerns that growing numbers of satellites in low orbit interfere with observations of the universe.
Industry-Wide Space Push
Major technology companies are projected to spend $3 trillion on earthbound datacentres from India to Brazil. This spending has raised concerns about carbon emissions if clean energy is not found. Elon Musk said last week his companies would start scaling up to create datacentres in space. He runs Starlink satellite internet and SpaceX rocket programme.
Nvidia AI chips will also launch into space later this month with startup Starcloud. Philip Johnston, co-founder of Starcloud, said space offers almost unlimited low-cost renewable energy. He claimed the datacentres would save 10 times the carbon dioxide over their lifetime compared with Earth-based facilities. The only environmental cost would be the initial launch.
Google stated that space may be the best place to scale AI computers in the future. The company uses TPUs, processors optimized for training and daily use of AI models. Free-space optical links would provide wireless transmission between satellites.
