Even as Google dramatically expands its portfolio of clean energy infrastructure, the hyperscaler anticipates that it will nonetheless need to find much more energy to fuel its operations.
One potential solution: space-based machine learning.
Last November, Google announced the launch of Project Suncatcher, an endeavor to someday scale machine learning in outer space. As an initial step, Project Suncatcher plans to send two solar-powered satellites equipped with Tensor Processing Units (TPUs) into orbit next year, where exponentially more solar energy is available to power the operations than on Earth.
“About [one ten billionth] of the energy that the sun emits hits the earth,” explained Blaise Agüera y Arcas, Google’s CTO of Technology and Society and a central member of the Project Suncatcher team, on a recent episode of Where the Internet Lives.
Agüera y Arcas brings a singular perspective to the conversation around extraterrestrial data centers. The author of the 2025 book “What is Intelligence?,” his approach to artificial intelligence centers around the provocative idea that it isn’t artificial at all, but instead that computers’ parallel processing mirrors the architecture of biological evolution. And as a result, he believes the technology will scale dramatically.
“ I think our thirst for intelligence is unlimited,” Agüera y Arcas explained on the podcast. “In nature, brains get bigger until they can’t get any bigger. Cities grow until they kind of can’t grow anymore. Even assuming that we gain another factor of a thousand in [energy] efficiency, we have to look at where the energy comes from. The answer is obvious: Space is where there is a huge amount of energy available.”
Through Google’s Paradigms of Intelligence team, which he founded, Agüera y Arcas helped design Project Suncatcher and its initial satellites. In order for those satellites to maximize solar absorption, they will need to be as flat as possible. “Any volume, any third dimension, is just more mass that you have to launch,” Agüera y Arcas explained. “So how two-dimensional can you make it is kind of the name of the game.”
The project comes with other significant engineering challenges. For instance, there’s the question of temperature control, and how to radiate excessive heat away from the TPUs. “In space, you can’t have liquid cooling. The only way to dissipate heat in space is radiative through infrared coming off the object,” Agüera y Arcas said. “And so you have to have large surfaces to radiate the heat.”
He cited communications as another hurdle. While data centers typically use fiber optics to transmit data, that physical infrastructure won’t be possible in space; as an alternative, Agüera’s y Arcas is pushing “free space optical links,” meaning communicating with satellites via laser. “In other words, you’re communicating by laser within a fleet of satellites. That’ll require the development of new kinds of optical communications that are still very experimental even on earth,” he explained.
While Agüera y Arcas expects hyperscalers to continue embracing terrestrial energy options like conventional solar, he maintains that the sheer scale of how much solar is available in outer space means that extraterrestrial options have real potential.
For the full conversation with Blaise Agüera y Arcas, listen to his interview on season 5 of Where the Internet Lives:
This is partner content, brought to you by Google. It borrows from an interview that appeared on Where the Internet Lives, a podcast produced in partnership by Latitude Studios and Google.
Where the Internet Lives is an award-winning podcast about the unseen world of data centers. Follow on Apple, Spotify, or wherever you get your podcasts.
