Do space-based AI data centers make economic sense?

Following the astronomical success of the SpaceX IPO — raising $85.7 billion, valuing the newly public company in the trillions, and minting Elon Musk as the world’s first trillionaire — what many skeptics still view as a pie-in-the-sky idea, building AI data centers in space, is coming into view. There is good reason for the skepticism, but the concept has potentially moved onto at least a more plausible path as a result of the SpaceX windfall.

SpaceX has reliable, reusable Falcon rockets — and a more powerful one in the wings — while its xAI has an insatiable need for compute power and its space-based internet service, Starlink, has upgradeable satellites. Now the interconnected entity’s engineering and technology has billions in new capital necessary to bring those components together in space, not only to feed SpaceX’s massive internal AI operations but also to provide commercial services for an array of paying customers such as Anthropic.

Some investors contend the company has no choice but to make the idea work if it hopes to justify its public market valuations over time. “The company comes down to data centers in space,” Duncan Davidson, a partner at Bullpen Capital, said on CNBC’s “The Exchange” the week before the IPO. “That is the big, long-term play.”

The engineering and technical issues are being solved, said Davidson, whose firm is not a SpaceX investor but has an indirect interest in space startup Starcloud. Though he added, “economically, right now, it’s marginal.”

Considering, too, the ever-increasing constraints on terrestrial data centers — practical, political and public — the prospects of launching them into low-earth orbit, where the sun shines 24/7, is no longer the stuff of science-fiction.

If, as Musk has stated, SpaceX’s heavy-duty Starship rocket becomes operational next year — definitely an “if,” given his track record of under-delivering on previously promised schedules — it will greatly lower launch costs, which are a critical barrier to affordability. Meanwhile, the cost of building Earth-based data centers might go up, while “the space ones are going to start getting cheaper and cheaper,” Davidson said. “So I think the [business] case is really strong for these things,” he said.

In January, SpaceX filed an application with the Federal Communications Commission for a constellation of up to one million satellites that would be the foundation for an orbital AI data center. Two months later, at an event in Austin, Texas, Musk reiterated past claims that space-based, solar-powered data centers will be more cost-effective than terrestrial ones in as little as two to three years. “Increasing power on Earth becomes harder over time and more expensive over time,” he said, “but in space it becomes actually cheaper and easier over time.”

The so-called AI1 satellites will be upgraded versions of those used for the existing Starlink communications network and will require exponentially more semiconductors. The sheer scale needed is so massive that SpaceX, Tesla and Intel have partnered to create Terafab, a 10-million-square-foot facility being built in Austin and slated to open in 2029 and which could cost up to $119 billion to build.

SpaceX declined to elaborate on its plans, providing previously released information on its orbital data center concept and Terafab.

Jeff Bezos, Alphabet are also in the race

SpaceX is hardly alone in what has become a race to compute in space. Amazon CEO Jeff Bezos has voiced similar aspirations for his rocket and AI ventures, Blue Origin and Prometheus, respectively. Last month, in a CNBC interview, Bezos said that building data centers in space is “very realistic,” though questioned how long it might take. “Some of the timelines we hear are very short. People would talk about two or three years,” he said, likely referring to Musk’s bold prediction. “That’s probably a little ambitious.”

In March, Blue Origin submitted plans to the FCC to launch 51,600 data center satellites into low Earth orbit as part of its Project Sunrise initiative. Deployment of the proposed constellation of satellites, dubbed TeraWave, is slated to begin in the fourth quarter of 2027, the company said.

Alphabet’s search giant Google has entered the race through a collaboration with Earth observation satellite maker Planet Labs on Project Suncatcher, an orbital data center initiative, with SpaceX (of which it owns 6.1%) as its potential launch partner. The project, Google said, will explore how an interconnected network of solar-powered satellites, equipped with its Tensor Processing Unit AI chips, could harness the full power of the sun.

A paper explaining Suncatcher notes how historically high launch costs have hindered large-scale space-based systems, but suggests that prices may fall to less than $200/kilogram by the mid-2030s. At that price, operating orbital data centers could become roughly comparable to the reported energy costs of an equivalent terrestrial data center on a per-kilowatt/year basis.

Beyond that paper, “We have nothing new to share,” a Google spokesperson wrote in response to a request for comment.

Satellite, rocket and robotics startups are in testing

Outside of the trillion-dollar-plus tech stock universe, several startups are also eyeing the skies.

Starcloud has already sent an Nvidia H100 GPU into space on a test satellite aboard a SpaceX Falcon 9 rocket. “It will just simply be cheaper to put them in space,” CEO Will Marshall told CNBC in a recent interview. Another benefit, he said, is not having to compete for water and electricity in people’s communities. It’s a longer-term project, Marshall said, “but an exciting one, too.”

Starcloud is also teaming with Rendezvous Robotics, a builder of modular spacecraft systems that self-assemble in space, to generate power for its orbital data centers. The spacecraft comprise hundreds of interconnected, hexagonal tiles, each about five feet in diameter, that are stacked into a launch rocket.

“Our tiles have been tested three times,” said Rendezvous president Joe Landon, “once on a Blue Origin New Shepard flight and twice on the International Space Station.”

Another test on the ISS is scheduled for later this year. “In 2028, we’ll be able to deliver full-scale systems,” he said.

Rocket Lab‘s Electron rocket has launched nearly 90 of its proprietary satellites into space for NASA, U.S. Space Force and numerous global clients. Founded in 2006 by self-trained engineer Peter Beck, Rocket Lab is constructing a more powerful reusable rocket, the Neutron, which will give the vertically integrated company the capability to compete with SpaceX — if on a vastly smaller scale — in the orbital data center market.

“If this turns out to be a big market, we’ll be in a great position to attack it either as a merchant supplier or for our own application or a combination of those things,” said CFO Adam Spice. If push came to shove, though, “We would much rather turn customers into tenants of infrastructure that we own rather than help them build out their own,” he said.

Cowboy Space, established in 2024 by Robinhood co-founder Baiju Bhatt under the original name of Aetherflux, has an end-to-end strategy as well, but it is still developing its rockets and data center infrastructure in-house. Its novel approach “involves using the second stage of the rocket as the data center satellite itself,” Bhatt said. “We’ll have more to reveal and show on that in the not too distant future,” he said, adding that the company is targeting its first launch to space for later this year. In the meantime, it has applied with the FCC for a 20,000 satellite constellation.

Even with all of research, development, and ongoing investment into orbital AI data centers in space as the new commercial space economy quickly takes shape, the fundamental economic question has not changed, according to Mark Weinzierl, a Harvard economist focused on tax policy who became interested in space-based businesses about a decade ago and now writes and teaches about the topic. “One of the biggest questions is, are you sure that we can’t just do that cheaper on Earth?” he said.

“I haven’t seen any that say that right now [they’re] cost-competitive,” he said of his analysis of current business models in the space sector. Yet Weinzierl believes it is reasonable to predict a future in which “the costs of doing it on Earth are going to go up over time and the costs of doing it in space are going to come down over time. And at some point those two curves are going to meet.”

But that requires assumptions about future economics that are based on existing trends today.

“Technology can always change,” he said. “Maybe our next generation of chips won’t be as energy-hungry as the current generation,” which could help bring down the cost of terrestrial data centers.

There are the regulatory, environmental and political constraints on Earth, which Weinzierl said do currently make the space-based case work better, at least in theory. Indeed, the mounting backlash against the prolific build-out of data centers has prompted more than 100 proposed moratoriums at the local, county, state and national level. The public is pushing back, too. A Heatmap News poll conducted in May found that seven in 10 Americans would oppose a data center being built near where they live, up from four in 10 last August. What’s more, a newly released study from First Street, a climate risk analytics firm, finds that 79% of data center capacity faces elevated risk to acute climate hazards.

In Weinzierl’s view, it remains somewhat of a gamble rather than a certainty, but not necessarily an outlandish one.

“If you’re optimistic that the cost declines we’ve seen in launch, satellite technology and solar [power] keep happening, then those [curve] lines are going to cross sooner. It’s always going to be a bet, but that seems like a reasonable story to me,” he said.