Google Plans to Put Datacentres in Space to Meet AI Demand

TL;DR: Google is developing plans to put AI datacentres into space through Project Suncatcher, with first trial equipment launching early 2027. Solar-powered satellite constellations could match Earth-based running costs by the mid-2030s whilst minimising impact on terrestrial water and land resources, though each rocket launch emits hundreds of tonnes of CO₂.

Google is hatching plans to put artificial intelligence datacentres into space, with its first trial equipment scheduled for orbit in early 2027. The initiative, dubbed Project Suncatcher, represents a radical approach to addressing the infrastructure demands of rapidly expanding AI systems.

The Orbital Architecture

Google’s scientists and engineers envision tightly packed constellations of approximately 80 solar-powered satellites arranged in orbit about 400 miles above Earth’s surface. These satellites would carry the powerful processors required to meet rising AI demand, with results beamed back through optical links using light or laser beams.

According to Google research released on Tuesday, falling space launch prices mean that by the middle of the 2030s, the running costs of a space-based datacentre could be comparable to terrestrial facilities. Once in orbit, solar panels can be up to eight times more productive than Earth-based equivalents.

“In the future, space may be the best place to scale AI computers,” Google stated. “Working backward from there, our new research moonshot, Project Suncatcher, envisions compact constellations of solar-powered satellites, carrying Google TPUs and connected by free-space optical links.”

Environmental Trade-Offs

The space-based approach offers potential advantages for terrestrial resource conservation. Using satellites could minimise impact on the land and water resources needed to cool existing datacentres—a growing concern as major technology companies project $3 trillion spending on AI datacentres from India to Texas and from Lincolnshire to Brazil.

However, the environmental calculus isn’t straightforward. Whilst orbital solar collection offers dramatic efficiency gains, launching a single rocket into space emits hundreds of tonnes of CO₂. The net environmental impact depends on comparing launch emissions against the lifetime carbon footprint of equivalent terrestrial infrastructure.

Philip Johnston, co-founder of startup Starcloud (which is launching Nvidia AI chips into space this month), claims “10 times carbon dioxide savings over the life of the datacentre compared with powering the datacentre terrestrially.” However, this calculation requires assumptions about facility lifetimes, launch frequency for maintenance and upgrades, and the carbon intensity of alternative terrestrial power sources.

Competing Initiatives and Concerns

Google isn’t alone in pursuing orbital computing. Elon Musk announced last week that his companies (Starlink satellite internet and SpaceX rockets) would scale up to create datacentres in space. This suggests the concept is moving from research curiosity to competitive reality.

However, astronomers have raised concerns about rising numbers of satellites in low orbit, describing them as “like bugs on a windshield” when attempting to peer into the universe. Google’s constellation would add to this crowding.

Technical Challenges

Google acknowledged significant engineering hurdles remain:

  • Thermal management: Dissipating heat in the vacuum of space
  • High-bandwidth ground communications: Maintaining reliable data transfer
  • On-orbit system reliability: Ensuring processors function in harsh space environment

The company described its research results as a “first milestone towards a scalable space-based AI” whilst sounding a cautionary note about the work ahead.

Looking Forward

The 2027 prototype launch represents a test of technical feasibility rather than commercial deployment. Success would validate the approach, but questions remain about scalability, maintenance, and long-term economics.

For organisations dependent on AI infrastructure, Google’s space ambitions highlight the scale of computing demand projected for coming decades. Whether that demand is ultimately met through orbital constellations, terrestrial facilities, or a hybrid approach, the willingness of major technology companies to explore radical alternatives suggests conventional datacentre expansion alone may not suffice.

The initiative also raises questions about the governance of commercial computing infrastructure in space—a domain with limited regulatory frameworks compared to terrestrial facilities.

Source Attribution:

Share this article