Data centers are facing a pivotal year as leaders in the artificial intelligence industry ramp up spending to build out their infrastructure and develop the next generation of AI models. The issue is that these facilities consume vast amounts of energy, raising both affordability and environmental concerns. In response, some AI companies have proposed their boldest plan yet to continue pursuing their ambitions: Data centers in space.
Orbital Data Centers, Explained
Orbital data centers are networks of satellites operating within Earth’s low orbit. In theory, these networks could consist of thousands of satellites equipped with the necessary chips, software and IT components to store and process data involved in complex calculations. They would then beam these results down to Earth to power advanced AI applications, including large language models and agents.
Also referred to as “orbital data centers,” these structures would consist of thousands of satellites that can complete the computations needed to sustain AI’s growing energy demands. But the same problems that plague data centers on Earth could also follow them into space, raising questions around whether this idea will even get off the ground as more tech titans consider expanding their footprint to the heavens.
What Is an Orbital Data Center, Exactly?
Historically, data centers have been huge physical facilities that house companies’ networking components, computer servers, data storage systems and other essential IT infrastructure. Transporting these massive campuses into space is nowhere near plausible, so companies must think outside the box to get their infrastructure out of Earth’s atmosphere.
Enter orbital data centers, or networks of low-Earth-orbiting (LEO) satellites that collectively manage the complex computations typically handled by traditional data centers. These satellites are expected to be equipped with specialized computer chips, software components and other infrastructure needed to store and process data, sending the final results down to Earth.
With direct access to the sun, they’ll also be able to harness solar energy to fuel their operations. The hope is that this will be enough to power advanced AI tools, including the latest AI models and agentic AI applications. As opposition to traditional data centers intensifies worldwide, orbital data centers are being treated less as science fiction and more as an intriguing alternative to their terrestrial counterparts.
The Case for Data Centers in Space
It’s become undeniably clear that data centers come with a hefty price tag. According to a press release by the U.S. Energy Information Administration, power demand in the United States is expected to grow 1 percent this year and 3 percent in 2027 — the first time in two decades that demand will have risen four years in a row. The release explicitly singles out “large computing centers” as the “driving factor behind this surge.”
Of course, spikes in energy demand naturally lead to higher electricity bills for communities located near data centers, putting tech giants in the hot seat. In response, companies like Amazon, OpenAI and Microsoft have promised to cover data center-related costs, and President Trump has even called for companies to sign a ratepayer pledge. But these steps may not be enough to win over local legislatures, with a number of U.S. states considering moratoriums that would temporarily ban the construction of new data centers until they better understand their effects.
Such laws would hinder companies’ efforts to improve their AI models at a time when they need to justify increased investments or risk fears of an AI bubble all over again. That leaves space as a possible alternative to this AI backlash on Earth. In theory, space provides plenty of room to build and scale satellite networks without the constraint of limited resources. There are no variables like shade or bad weather blocking the sun either, giving satellites 24/7 access to solar power that could make rapid expansion even more feasible.
“To harness even a millionth of our Sun’s energy would require over a million times more energy than our civilization currently uses,” as tech mogul Elon Musk puts it in a SpaceX blog post. “The only logical solution therefore is to transport these resource-intensive efforts to a location with vast power and space. I mean, space is called ‘space’ for a reason.”
With AI becoming a divisive political issue here on Earth, more tech leaders are joining Musk in the quest to develop the infrastructure for orbital data centers. The question is who can gain an edge as the AI race kicks into overdrive.
Early Frontrunners to Watch in the Orbital Data Center Space
Established titans and rising startups alike are competing to be the first to store data closer to the stars. Here are a few key players to keep an eye on in this fast-growing niche.
SpaceX
SpaceX already has an advantage in the aerospace industry, thanks to Musk’s government contracts. And now, with xAI, his AI startup, under its umbrella, the company has even more resources to work toward Musk’s goal of launching orbital data centers. Starlink’s existing satellites could pivot to supporting computational needs as well, and SpaceX has applied to launch another million satellites to strengthen its network.
Blue Origin
Not to be outdone, fellow tech mogul Jeff Bezos and his company Blue Origin plan to send more than 5,400 LEO satellites into space that could deliver data at speeds up to 6 terabits per second. Known as TeraWave, this network aims to provide fast and secure service in both rural and suburban areas, as well as businesses and governments.
Blue Origin has also asked the U.S. government for permission to launch a network of more than 50,000 satellites to create an orbital data center. In a document filed with the Federal Communications Commission (FCC), the company described the so-called “Project Sunrise” as a network of spacecraft that will perform advanced computation in orbit to “ease mounting pressure on U.S. communities and natural resources by shifting energy — and water-intensive compute away from terrestrial data centers.” The exact timeline of this project is unclear, but experts reportedly told TechCrunch that it would be unlikely to come to fruition until the 2030s.
Google is exploring the possibility of launching a network of solar-powered satellites under its initiative called “Project Suncatcher.” The plan involves equipping each satellite with tensor processing units — Google’s computer chips designed specifically to accelerate machine learning workloads. The project is still in its early stages, but Google hopes it can be as successful as its early endeavors in quantum computing and autonomous vehicles.
Starcloud
Backed by chipmaker Nvidia, Starcloud has outfitted its Starcloud-1 satellite with an Nvidia H100 graphics processing unit, allowing the satellite to train an AI model in space — the first time that’s ever been done. According to an Nvidia blog post, Starcloud is using space’s freezing temperatures to cool its satellite without water and foregoing backup power thanks to a steady supply of solar energy. This promise of sustainability has earned the startup a partnership with Amazon Web Services, with the goal of launching 88,000 satellites.
Axiom Space
Axiom Space has been on a win streak lately, partnering with NASA for its fifth private mission and securing $350 million in funding from various investors, including Donald Trump Jr.’s 1789 Capital. Besides building a successor to the International Space Station, the company is launching orbital data center nodes, laying the groundwork for reliable, secure space-based cloud computing.
Sophia Space
Another startup to follow is Sophia Space, which raised $10 million in February 2026. The company is creating solar-powered tiles to process computations in space and send the results to Earth. While the technology is designed to be compatible with any existing satellite and cloud infrastructure, Sophia Space’s partnership with Nvidia means its tiles work particularly well with Nvidia’s Jetson platform and Blackwell architecture.
Why Orbital Data Centers Could Be Problematic
Orbital data centers may be gaining traction among tech players, but the concept comes with several potential pitfalls. For one, constellations of satellites are costly to operate, considering the investments needed to launch, repair and replace them over time. Advances made by Starlink and Tesla may have brought these overall expenses down, but it’s estimated that orbital data centers remain three times more expensive than terrestrial ones.
And space-based data centers come with their own environmental impacts. According to a 2024 study, the process of launching satellites generates aluminum oxides — a possible pollutant that could affect the ozone layer. There’s also the issue of space debris resulting from launches and older satellites going out of commission, as well as the effects of overcrowding Earth’s lower orbit with thousands of satellites.
Perhaps most concerning, though, is how orbital data centers could widen the gap between the world’s superpowers and everyone else. With no regulations in this area, the U.S. and China can double down on orbital data center initiatives and solidify themselves as the only two real options in this field. Such a scenario would force other countries to rely on American or Chinese satellites for additional AI compute, leaving them completely vulnerable if the U.S. or China leverages this dependence to pressure them in the geopolitical arena.
If the race to build space-based data centers does indeed unfold as another U.S.-China showdown, there could be international efforts to halt these projects until there are guarantees that they would benefit the greater good, rather than just the tech elites. That said, national security concerns may be the very factor that compels companies to move forward anyway with their plans for orbital data centers, given the political undercurrents of the AI race.
Could Space Be the Next Frontier in the AI Race?
Ambitions to dominate the AI industry have spurred conversations around tech sovereignty, or a country’s ability to access and control technologies essential to its economy and national security. These discussions have only grown more urgent as companies turn their attention toward the stars, with countries beginning to view access to space as crucial to their autonomy and security.
At the same time, American and Chinese companies are vying to cement their countries’ superiority in the space industry, and neither side is backing down. With China aiming to put astronauts on the moon by 2030, Musk has shifted SpaceX’s focus to the moon and wants to develop satellite infrastructure there. This announcement has encouraged other U.S. companies to accelerate their own lunar missions, including two startups currently working on a fleet of robots that will harvest isotopes from the moon.
All this to say that space is becoming an attractive investment in the eyes of tech titans and government leaders alike, and orbital data centers are key to controlling this rapidly evolving sector. Presented with the reality of falling behind once again, countries around the globe may funnel resources into satellite networks and other space initiatives, hinting at a new era in which the AI race is no longer confined to this world.
Frequently Asked Questions
What is an orbital data center?
Orbital data centers are actually networks of satellites — sometimes numbering in the thousands — that are outfitted with specialized chips, software and other components necessary to process and store large volumes of data. Harnessing solar energy from the sun, these satellites are expected to handle the complicated calculations as part of AI workloads, powering applications like AI models and agents.
Why do AI companies want to move data centers into space?
The rising energy demands of data centers are beginning to strain resources and drive up electricity bills, sparking anti-AI backlash. Meanwhile, space theoretically provides more room to scale satellite networks without draining Earth’s resources, and satellites have constant access to solar energy. As a result, tech companies are starting to view orbital data centers as a convenient alternative to terrestrial data centers.
What are the potential downsides of orbital data centers?
Considering the costs of launching, maintaining and replacing satellites, orbital data centers are about three times more expensive than terrestrial data centers. A 2024 study also found that the launch process produces pollutants and space debris, introducing new environmental issues. Most consequential, though, is how space-based data centers could threaten the national security of countries already dependent on American and Chinese tech stacks, potentially drawing international resistance to these centers if they fail to serve the greater good.
