Do space-based AI data centers make economic sense?

Following the astronomical success of SpaceX’s IPO (raising $85.7 billion, valuing the newly-listed company in the trillions, and launching Elon Musk as the world’s first trillionaire), the idea of ​​building artificial intelligence data centers in space emerges, which many skeptics still consider an idea in the air. There’s good reason to be skeptical, but the concept has potentially at least moved towards a more plausible path as a result of SpaceX’s unexpected crash.

SpaceX has reliable, reusable Falcon rockets and a more powerful rocket in the wings, while xAI has an insatiable need for computing power and space-based internet service Starlink has upgradeable satellites. Now the interconnected enterprise’s engineering and technology have the billions of dollars of new capital needed to assemble these components in space; This applies not only to fuel SpaceX’s massive internal AI operations, but also to provide commercial services to a range of paying customers like Anthropic.

Some investors argue that the company has no choice but to pursue the idea if it hopes to justify its public market valuations over time. “The company is coming to data centers in space,” Bullpen Capital partner Duncan Davidson said on CNBC’s “The Exchange” the week before the IPO. “This is a big, long-term game.”

Davidson, whose company is not an investor in SpaceX but has an indirect interest in the space initiative Starcloud, said that the engineering and technical problems have been resolved. But he added: “Economically it is marginal at the moment.”

Given the increasing practical, political and public constraints on terrestrial data centers, the prospect of launching them into low-Earth orbit where the sun shines 24/7 is no longer the stuff of science fiction.

As Musk noted, if SpaceX’s heavy-duty Starship rocket becomes operational next year, it’s definitely an “if” given its track record. poorly published over previously promised programs – will greatly reduce launch costs, a critical barrier to affordability. Meanwhile, the cost of building Earth-based data centers could rise, while “those in space will start to get cheaper,” Davidson said. “So I guess [business] “The 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 This will be the basis for an orbital AI data center. Two months later, at an event in Austin, Texas, Musk repeated past claims that space-based, solar-powered data centers would be more cost-effective than terrestrial ones in as little as two to three years. “Increasing power on Earth gets harder and more expensive over time,” he said, “but in space it actually gets cheaper and easier over time.”

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

SpaceX declined to provide details about its plans, citing previously released information about the orbital data center concept and Terafab.

Jeff Bezos and Alphabet are also in the race

SpaceX is not alone in what has become a computing race in space. Amazon CEO Jeff Bezos has expressed similar wishes for his rocket and artificial intelligence ventures Blue Origin and Prometheus, respectively. In an interview with CNBC last month, Bezos said that building data centers in space was “very realistic” but was questioned how long it would take. “Some of the timelines we’ve heard are very short. People talk about two or three years,” he said, possibly referring to Musk’s bold prediction. “That’s probably a bit 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 satellite constellation, called TeraWave, is planned to begin in the fourth quarter of 2027, the company said.

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

An article explaining Suncatcher He notes how historically high launch costs have hindered large-scale space-based systems, but suggests prices could fall below $200 per kilogram by the mid-2030s. At this price, operating orbital data centers could become roughly comparable to the reported energy costs per kilowatt/year of an equivalent terrestrial data center.

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

Satellite, rocket and robotic initiatives are in the testing phase

Outside of the multi-trillion-dollar universe of tech stocks, many startups are also looking to the skies.

Starcloud has already sent an Nvidia H100 GPU into space with a test satellite on a SpaceX Falcon 9 rocket. “It’s going to be cheaper to put them in space,” CEO Will Marshall told CNBC in a recent interview. Another benefit, he said, is that people don’t have to compete for water and electricity in their communities. Marshall said it’s a longer-term project, “but it’s also exciting.”

The star cloud is also Teamwork with Rendezvous RoboticsManufacturer of modular spacecraft systems that self-assemble in space to generate power for orbital data centers. The spacecraft consists of hundreds of interconnected hexagonal tiles, each about 1.5 meters in diameter, stacked in the shape of a launch rocket.

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

Another test on the ISS is planned for later this year. “We will be able to deliver full-scale systems in 2028,” he said.

Rocket LabThe Electron rocket has launched nearly 90 specialized satellites into space for NASA, the U.S. Space Force and numerous global customers. Founded in 2006 by self-taught engineer Peter Beck, Rocket Lab is building the Neutron, a more powerful, reusable rocket that would give the vertically integrated company the ability to compete with SpaceX in the orbital data center market — albeit on a much smaller scale.

“If this turns out to be a big market, we’ll be in a great position to attack it as a commercial supplier, or for our own application, or a combination thereof,” CFO Adam Spice said. But if the pressure continues, “we would rather turn our customers into tenants of the infrastructure we own, rather than helping them build their own infrastructure,” he said.

Cowboy Space, originally named Aetherflux by Robinhood co-founder Baiju Bhatt in 2024, also has an end-to-end strategy but still develops its rockets and data center infrastructure in-house. Bhatt said his new approach “involves using the rocket’s second stage as the data center satellite itself.” “We will have more to reveal and show in the not-too-distant future,” he said, adding that the company is aiming for the first launch into space later this year. Meanwhile, it has applied to the FCC for a constellation of 20,000 satellites.

While the new commercial space economy is rapidly taking shape, for all the research, development and ongoing investment in orbital AI data centers in space, the basic economic question remains unchanged, 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 lectures on the subject. “One of the big questions is, are you sure we can’t do this cheaper on Earth?” he said.

“I haven’t seen anyone say that right now [they’re] “cost competitive,” he said of his analysis of current business models in the space sector. But Weinzierl believes it is reasonable to envision a future in which “the costs of doing this on Earth will increase over time, and the costs of doing this in space will decrease over time.” And at some point these two curves will meet.”

But this requires assumptions about the future economy based on current trends today.

“Technology can always change,” he said. “Perhaps our next generation of chips will not be as energy hungry as the current generation.” This could help reduce the cost of terrestrial data centers.

Current regulatory, environmental and political constraints on Earth make the space-based case work better, at least in theory, Weinzierl said. Indeed, the growing backlash against the efficient creation of data centers has led to multiple movements. 100 proposed moratorium at the local, county, state and national level. The public is also taking a step back. A Heatmap News survey conducted in May found that: seven in 10 Americans More than four in 10 would oppose building a data center near where they live, up from four in 10 last August. What’s more, a newly published study from climate risk analytics firm First Street finds that 79% of data center capacity faces high risk of acute climate hazards.

This remains something of a gamble rather than a certainty, according to Weinzierl, but it’s not necessarily an odd one either.

“If you’re optimistic about the cost reductions we’re seeing in launch, satellite technology and solar energy [power] keep being, then these [curve] The lines will intersect more quickly. “There will always be a bet, but this seems like a plausible story to me,” he said.