Elon Musk is moving the idea of orbital data centers from speculation toward a more concrete plan. SpaceX has filed plans with the Federal Communications Commission for a million-satellite data center network, while the formal merger between SpaceX and xAI has linked Musk’s space and AI ambitions more directly.
The result is a proposal with major technical and business implications: putting networks of computers in orbit to support AI workloads. The central claim is simple, but far from settled: space may become a better place to scale AI infrastructure than the ground.
A space plan tied to AI infrastructure
The source article says SpaceX filed plans with the FCC on Friday for a million-satellite data center network. A week later, the idea looked less like a passing remark and more like part of a larger strategy.
The most visible move was the formal merger between SpaceX and xAI, which went forward on Monday. That merger brings together Musk’s launch business and his AI company in a way that fits the concept of a shared infrastructure project.
In plain terms, orbital data centers would be networks of computers operating in space. Instead of building all AI compute capacity on Earth, some of that work would move into orbit, where power generation and launch economics become part of the infrastructure equation.
The FCC also moved the filing forward. On Wednesday, the agency accepted the filing and set a schedule for public comment. The source describes that as usually procedural, but notes that FCC chairman Brendan Carr took the unusual step of sharing the filing on X.
Why Musk says space could be cheaper
Musk’s public argument centers on power. On a new episode of John Collison’s podcast Cheeky Pint, which also featured Dwarkesh Patel, Musk said solar panels can produce more power in space than they do on Earth.
“It’s harder to scale on the ground than it is to scale in space,”
According to Musk, any given solar panel can deliver about five times more power in space than on the ground. From that point, he argues that one of the major operating expenses for data centers could be reduced by shifting AI compute into orbit.
The logic is important because AI data centers are power-hungry infrastructure. If energy is a major cost, then a place with more productive solar panels can look attractive. That is the case Musk is trying to make.
But the source also highlights a gap in the argument. Power is not the only cost in running a data center, and solar panels are not the only way to power one. Patel raised that point in the podcast, along with concerns about servicing GPUs that fail during AI model training.
The timeline Musk is putting on the table
Musk did not frame orbital data centers as a distant concept. He pointed to 2028 as a tipping point year for the idea.
“You can mark my words, in 36 months but probably closer to 30 months, the most economically compelling place to put AI will be space,”
That prediction sets a short clock. It suggests Musk believes the economics of AI compute, launch capability, and space-based power could converge quickly enough to shift the center of gravity for future data center growth.
He went further with a broader prediction about the scale of AI in orbit. Musk said that five years from now, his prediction is that more AI will be launched and operated in space every year than the cumulative total on Earth.
The source provides one benchmark for how large the terrestrial market is expected to become. As of 2030, global data center capacity will be an estimated 200 GW. The article describes that as roughly a trillion dollars’ worth of infrastructure when built on the ground.
The business case behind the technology case
The proposal also fits SpaceX’s business model. SpaceX makes money by launching things into orbit, so a future in which AI data centers need orbital deployment would naturally benefit the company.
The SpaceX-xAI merger makes that alignment more direct. If one side of the combined company needs enormous AI infrastructure and the other side can put hardware into orbit, the idea becomes more than a theoretical engineering discussion.
The source also notes that the new SpaceX-xAI conglomerate is headed for an IPO in just a few months. That timing matters because orbital data centers could become part of the story presented to investors, especially while tech companies are still pouring hundreds of billions of dollars into data center spending each year.
None of that proves the plan will work. The article makes clear that questions remain around cost, maintenance, and whether greater solar output in space is enough to outweigh the complexity of operating compute hardware in orbit.
What to watch next
The next phase is likely to involve both regulatory and public persuasion. The FCC has accepted the filing and opened a public comment process, giving the proposal a formal path forward.
At the same time, Musk has begun explaining the case in public: more power from solar panels in space, harder scaling on the ground, and a belief that AI compute will move toward orbit faster than many might expect.
For now, orbital data centers remain an ambitious infrastructure proposal rather than a proven replacement for ground-based facilities. But the pieces described in the source article are now connected: SpaceX, xAI, an FCC filing, a million-satellite data center network, and Musk’s prediction that space could soon become the most economically compelling place to put AI.