The High Cost of AI: Google Looks to Space for Sustainable Scaling

The rapid scaling of artificial intelligence comes at a high cost, both financially and environmentally. With $3 trillion in datacentre spending projected, Google is exploring a sustainable alternative by looking to space, where resources like solar energy are virtually unlimited.
Google’s “Project Suncatcher” envisions a future where AI computation is moved off-planet. The plan involves constellations of satellites powered by solar panels, which in orbit are eight times more effective than on Earth. This offers a path to scaling AI without a proportional increase in carbon emissions from terrestrial power grids.
This orbital approach also directly addresses the land and water impact of AI. Terrestrial datacenters require massive footprints and use immense amounts of water for cooling. By moving this infrastructure into space, Google hopes to find a more sustainable model that doesn’t deplete precious resources on Earth.
The timing of this “moonshot” is tied to the falling cost of rocket launches, which could make space-based operations economically comparable to Earth-based ones by the mid-2030s. Other companies, like Starcloud, are making similar calculations, claiming a potential 10-fold CO2 saving over the datacentre’s life.
However, this “clean” solution has a dirty entry cost: the rocket launches themselves emit hundreds of tonnes of CO2. Furthermore, astronomers worry that these constellations will add to the “light pollution” in low orbit. Google’s first 2027 prototypes will be the first step in balancing these costs and benefits.