Google has announced Project Suncatcher, a research initiative that explores the concept of solar-powered AI data centres orbiting in space. The concept proposes compact constellations of satellites equipped with Google’s Tensor Processing Units (TPUs), which would draw continuous energy from the Sun and communicate via high-speed optical links.
Survey
✅ Thank you for completing the survey!
In a recent preprint paper titled “Towards a Future Space-Based, Highly Scalable AI Infrastructure System Design,” the company provided an overview of the project. The study claims that launching AI computing infrastructure into orbit could unlock enormous energy potential while reducing reliance on terrestrial resources. Reports state that solar panels in orbit can run almost continuously and generate up to eight times as much energy as those on Earth, eliminating the need for substantial battery backups.
To make this logical, Google engineers are addressing a number of key challenges, including maintaining tight satellite formations and ensuring high-bandwidth, low-latency data transfers on par with Earth-based data centres. Early lab tests have shown 1.6 Tbps transmission speeds over optical inter-satellite links.
However, radiation tolerance poses yet another challenge. Tests on Google’s Trillium Cloud TPUs show that the chips are surprisingly resilient, withstanding radiation levels nearly three times those expected for low-Earth orbit missions.
Economically, falling launch costs could make the concept more feasible by the mid-2030s. According to Google’s analysis, if launch prices fall below $200 per kg, the cost of operating space-based compute systems could match the energy costs of terrestrial data centres.
Furthermore, Google now aims to work with Planet Labs and launch two prototype satellites by 2027. If successful, Project Suncatcher will be the first step towards testing space-based AI computing and the future of scalable, sustainable AI infrastructure that is literally powered by the Sun.
Ashish Singh is the Chief Copy Editor at Digit. He's been wrangling tech jargon since 2020 (Times Internet, Jagran English '22). When not policing commas, he's likely fueling his gadget habit with coffee, strategising his next virtual race, or plotting a road trip to test the latest in-car tech. He speaks fluent Geek. View Full Profile