Elon Musk's SpaceX is eyeing a revolutionary leap in computing infrastructure: launching data centers into orbit. This ambitious vision seeks to address the growing energy demands of AI and next-generation computing by harnessing the near-constant solar power available in space. The company has filed a request with the U.S. Federal Communications Commission (FCC) to approve a new satellite system that would function as a large, space-based computing network.
The proposal details a constellation potentially comprising up to one million satellites in low Earth orbit (LEO), interconnected via laser links. These satellites would house computing hardware, effectively creating orbital data centers powered primarily by sunlight. The goal is to offload computationally intensive tasks, including AI workloads, from Earth-based data centers, reducing strain on terrestrial power grids.
SpaceX argues that this next-generation satellite network will surpass ground-based data centers in both cost and energy efficiency. Traditional data centers consume vast amounts of electricity and rely on water for cooling. Space-based data centers, conversely, can use the vacuum of space as a natural heat sink and tap into abundant solar energy. Starcloud, a startup focused on space-based data centers, projects energy costs in space to be ten times cheaper than those on land, even when factoring in launch expenses.
However, the sheer scale of SpaceX's proposal has raised eyebrows. As of January 2026, there are fewer than 15,000 active satellites in Earth's orbit. Deploying a million more could significantly increase the risk of Kessler Syndrome, a scenario where collisions between space objects create a cascade of debris, rendering certain orbits unusable. Harvard astrophysicist Jonathan McDowell notes that such a large constellation would also pose challenges for astronomical observations.
Experts also question the feasibility of cooling GPUs in the vacuum of space. While the lack of air presents a challenge, industry analyst Carlos Placido suggests that with a million satellites, data centers could be made relatively small and easier to cool.
Despite the skepticism, Musk remains confident in the long-term viability of space-based AI. He has stated that it is the "only way to scale" AI capabilities and has announced SpaceX's acquisition of xAI to further this vision. SpaceX is reportedly preparing an IPO expected to raise as much as $50 billion to fund its orbital data center ambitions.
Meanwhile, China is also making strides in space-based AI infrastructure. The China Aerospace Science and Technology Corporation (CASC) has announced plans to construct "gigawatt-class space digital-intelligence infrastructure" within the next five years, posing a potential challenge to SpaceX's dominance in this emerging field.
While challenges remain, the concept of space-based data centers is gaining traction as a potential solution to the growing demands of AI and high-performance computing. Companies like Starcloud are already working towards deploying data centers in orbit, citing benefits such as reduced costs, increased scalability, and faster deployment. Google is also exploring space-based machine learning compute, with plans to launch two prototype satellites by early 2027 to test the technology. These satellites will test models and TPU hardware in space and validate optical inter-satellite links for distributed machine learning tasks.
SpaceX's application for orbital data centers has been accepted for filing by the FCC, with public comments due by March 6, 2026. As the world grapples with the increasing energy consumption of AI, the prospect of data centers in space represents a bold, albeit challenging, vision for the future of computing.
