The artificial intelligence revolution has an environmental dark side: a projected $3 trillion spend on power-hungry datacenters that threaten to derail climate goals. Google is now championing a potential solution, “Project Suncatcher,” which aims to move this infrastructure to space.
The core of the problem on Earth is power. Running AI models requires vast amounts of electricity, and if that power isn’t from a clean source, it generates massive carbon emissions. Google’s plan is to bypass this problem by using solar panels in orbit, which are up to eight times more productive and offer a 24/7 source of renewable energy.
This orbital approach also eliminates the “land and water” problem. Terrestrial datacenters use enormous quantities of water for cooling, stressing local resources. A space-based datacentre, operating in a vacuum, would avoid this entirely, offering what Google calls a minimal “impact on terrestrial resources.”
While the long-term operation could be carbon-negative, the setup is not. Each rocket launch required to deploy the satellite constellations emits hundreds of tonnes of CO2. This initial environmental cost is a significant drawback that must be weighed against the long-term benefits.
Other companies, like Starcloud, which is launching Nvidia chips into space, argue the trade-off is worth it. Starcloud’s co-founder claims a “10 times carbon dioxide savings over the life of the datacentre.” Google’s 2027 prototype launch will be the first step in testing whether this gamble pays off for the planet.