Google has spent a decade turning “moonshot” into a brand. Now, the company is drawing a line toward the actual Moon.
Alphabet CEO Sundar Pichai is now openly backing data centers in space, telling Fox News in a Sunday interview that the company will start launching AI hardware in 2027 and that within about 10 years, extraterrestrial infrastructure will feel like a normal way to power the AI boom.
“There’s no doubt to me that a decade or so away we’ll be viewing it as a more normal way to build data centers,” he said. So for once, Pichai’s moonshot rhetoric comes with rockets attached. And in the not-too-distant future, Google’s cloud could live far beyond the atmosphere and off the planet for good.
“At Google, we’re always proud of taking moonshots,” Pichai said. “One of our moonshots is: How do we one day have data centers in space so that we can better harness the energy from the sun, which is one hundred trillion times more energy than we produce in all of Earth today?”
Internally, the effort lives under the very on-the-nose name Project Suncatcher. In a November blog post, Google pitched it as a “research moonshot” to push AI compute into orbit, with solar-soaked satellite swarms running TPUs and talking to one another over laser links instead of fiber. Essentially, the company is looking to “one day scale machine learning in space.”
The opening act is small — two test satellites in a partnership with Planet by early 2027 — but the sketches jump straight to tight satellite clusters the size of a city neighborhood, an early draft of what tomorrow’s off-world data centers might look like in the 2030s.
“We are taking our first step in ’27,” Pichai told Fox News. “We’ll send tiny, tiny racks of machines, and have them in satellites, test them out, and then start scaling from there.”
The space pitch arrives when Earth is starting to look like a bad long-term landlord for the AI build-out. A 2024 Lawrence Berkeley National Laboratory report found that U.S. data centers already chew through about 4.4% of the country’s electricity, and that share could climb to as much as 12% by 2028 as GPU farms multiply. McKinsey puts a price tag on the race to scale data centers: roughly $6.7 trillion in global data center capex by 2030, about $5 trillion of that aimed at AI-ready infrastructure. At some point, “just build another region” stops being a strategy and starts being an electrical engineering problem.
Google is already living inside that math. The company’s disclosures and outside tallies show its data center electricity use more than doubling in four years, even as it tries to keep emissions flat with carbon-free power purchases. At the same time, the company just committed $40 billion through 2027 to campuses in Texas’s Armstrong and Haskell counties.
Those are the “normal” data centers. Suncatcher is the contingency plan for the moment when even friendly grids and fast-tracked permits can’t keep up.
The company’s pitch is that space solves a lot of the headaches its Texas builds create. Above the atmosphere, satellites can sit in near-constant sunlight, harvest far more solar power per panel, and dump their waste heat into deep space instead of warm rivers and furious ratepayers. Launch costs keep drifting lower, and with Project Suncatcher, Google’s research team has already sketched out constellations of tightly clustered satellites laced together with laser links and packed with TPUs, an orbital TPU pod that looks more and more like a data hall with a better view. Pichai’s “within a decade or so” line is essentially a wager that the cost curves on rockets and chips will intersect fast enough to make that architecture competitive with another concrete box in West Texas.
Pichai isn’t alone in thinking the next phase of the cloud might live off-world. Tech CEOs have become obsessed with the idea. Salesforce CEO Marc Benioff called space “the lowest cost place for data centers” thanks to continuous solar and no batteries — in reference to a video clip of Tesla CEO Elon Musk discussing orbital AI and space-based data centers. Musk is throwing around 300-gigawatt Starship numbers, Amazon founder Jeff Bezos is talking about 10-to-20-year timelines for data centers in space, and OpenAI CEO Sam Altman has mused about “maybe” putting data centers in space — “maybe we build a big Dyson sphere,” he said.
Meanwhile, Y Combinator- and Nvidia-backed startup Starcloud has been busy sending its own AI-equipped satellite into orbit and promising that extraterrestrial data centers could cut emissions by a factor of 10 compared with their earthbound cousins, even after you count the rocket fuel. In this version of the space race, Mars is just scenery. The real prize is the cheapest, cleanest marginal megawatt you can point at a language model.
There’s a risk that Google’s orbital annex becomes an extremely expensive science fair project with gorgeous launch photos. But if the AI boom is going to demand Texas-sized power plants anyway, Google would rather that some of them float. Most companies worry about running out of runway. Google is gaming out what happens when it runs out of planet.
📬 Sign up for the Daily Brief
