Insider Brief
- Eric Schmidt’s acquisition of Relativity Space signals plans to deploy AI data centers in orbit, addressing the growing energy and cooling demands of artificial intelligence workloads.
- Space-based data centers could leverage solar power and the vacuum of space for heat dissipation, offering a sustainable alternative to terrestrial infrastructure.
- With Relativity’s heavy-lift Terran R rocket in development, Schmidt is positioning for launch control in a market with few viable options for large-scale orbital deployment.
- Image: Relativity Space
Former Google CEO Eric Schmidt appears to be laying the groundwork for an ambitious new frontier in artificial intelligence, one that extends beyond Earth’s surface, according to media reports. Less than two months after acquiring Relativity Space, Schmidt’s broader aim is coming into focus: building the capacity to launch AI-focused data centers into orbit.
While Schmidt has made few public statements about his purchase, his testimony before the U.S. House Committee on Energy and Commerce in April offers a revealing window into his thinking. As reported by Ars Technica, Schmidt spoke about the rapidly escalating energy demands of AI systems and the looming strain they are expected to place on national power infrastructure. He noted that average energy consumption growth has been minimal — roughly 0.5 percent annually over the past decade — yet demand from AI could require tens of gigawatts of new capacity in just the next few years.
AI models, particularly large language models, require enormous amounts of computational power. A single query to ChatGPT, for example, can use up to 10 times more energy than a standard web search. With expectations that AI-related data centers will demand an additional 29 gigawatts of electricity by 2027 and another 67 gigawatts by 2030, the U.S. power grid may not be ready.
With both electricity and water usage expected to surge to unsustainable levels, Schmidt appears to be exploring an alternative: relocating energy-hungry computing operations to space. In theory, orbiting data centers powered by solar energy could bypass terrestrial constraints. In space, systems could more easily radiate excess heat, and solar panels could generate continuous power in sun-synchronous orbits.
The plan seems like a bit of a stretch, but, following his congressional testimony, Schmidt responded affirmatively to speculation on social media that his acquisition of Relativity was connected to these plans.
Eric Berger is the senior space editor at Ars Technica, writes: “‘This probably helps explain why Schmidt bought Relativity Space,’ I commented on the social media site X after Schmidt’s remarks. A day later, Schmidt replied with a single word, ‘Yes.'”
Relativity Space, while still pre-launch for its flagship Terran R rocket, is a logical fit for such a strategy, according to Ars Technica. The company is one of the few independent U.S. launch providers developing a heavy-lift rocket. Its planned Terran R vehicle is projected to carry up to 33.5 metric tons to low-Earth orbit in expendable mode, or 23.5 tons with a reusable first stage. That capacity surpasses even a fully upgraded Vulcan Centaur from United Launch Alliance and rivals some configurations of SpaceX’s Falcon Heavy.
Other options for private launch at this scale are limited. SpaceX and Blue Origin are tightly held by billionaires Elon Musk and Jeff Bezos, respectively. ULA’s Vulcan is costly and already committed to existing customers. Rocket Lab’s Neutron may not meet the necessary payload specifications. Relativity, then, offers a rare opportunity for Schmidt to control launch infrastructure at the scale his vision may require.
As Ars Technica has previously reported, there have been a few setbacks. Relativity’s approach has suffered from engineering issues and is the subject of strategic questions. However, such issues may be solvable with deeper investment — and Schmidt, while less wealthy than Musk or Bezos, is reportedly courting co-investors to inject fresh capital into the company, Ars Technica reports.
Beyond launch logistics, Schmidt’s vision still faces several hurdles. Constructing and operating data centers in orbit would require new solutions in thermal regulation, reliability, and satellite servicing. There are also legal and orbital traffic challenges, given the already congested nature of low-Earth orbit. Whether solar arrays in space can produce the constant power needed—or whether large-scale orbital computing even makes financial sense — is still open for debate.
