The company’s scientists and engineers plan to launch around 80 solar-powered satellites about 400 miles above Earth’s surface.
Each satellite will carry advanced processors to handle the surging demand for AI computing.
According to Google research published Tuesday, falling space launch costs could make orbital data centers as economical as those on Earth by the mid-2030s.
The move could also reduce pressure on land and water resources currently used to cool terrestrial data centers.
Once in orbit, the centers will run on solar panels that are up to eight times more efficient than ground-based ones.
However, launching a single rocket still emits hundreds of tons of CO₂.
Big technology companies are projected to invest about $3 trillion in land-based data centers worldwide, from India to Texas and from Lincolnshire to Brazil.
That scale of spending has fueled concerns about rising carbon emissions if clean energy sources are not secured.
Elon Musk, chief executive of SpaceX and satellite internet provider Starlink, said last week that his companies also plan to build data centers in space.
AI data centers are planned to be deployed in space to meet the growing demand for computing power and to lessen environmental strain on Earth.
Stronger and more efficient solar energy is expected to be harnessed through the use of solar-powered satellites in orbit, eliminating the need for land and water resources typically required to cool data centers.
With launch costs falling, the plan is viewed as increasingly feasible, and a sustainable, cost-effective infrastructure for AI operations beyond Earth’s surface is being pursued as the long-term objective.