SpaceX space data centres AI 2026
Buried inside the biggest IPO in history is a plan that sounds like science fiction. SpaceX wants to launch up to one million satellites that work as a data centre network in orbit, built specifically to run AI. The company says deployment could start as early as 2028. Here is what that actually means, how serious it is, and why it matters for the UAE’s own AI ambitions.
The Plan Hiding Inside the World’s Biggest IPO
When SpaceX filed its IPO prospectus with the SEC, most headlines focused on the valuation. Buried in the filing was something far stranger. SpaceX is seeking regulatory approval from the Federal Communications Commission to launch up to one million satellites. Their purpose is not just internet connectivity. They are designed to function as a data centre network in space, built to support AI workloads.
The company says it expects to begin deploying these space based data centres as early as 2028. That is just two years from now. For a plan this ambitious, two years is an extremely short runway, even by SpaceX’s famously aggressive standards.
| One million satellites. A data centre network in orbit. A 2028 start date. This is not a thought experiment. It is written into the legal document that just helped raise 75 billion dollars. |
Why Would Anyone Put a Data Centre in Space?
The case for orbital data centres comes down to three things that are in short supply on Earth right now: power, cooling, and land.
AI data centres consume enormous amounts of electricity. The race to build new data centres on Earth has already triggered local backlash over energy prices and water usage for cooling, an issue Musk himself addressed when discussing Anthropic’s recent commitments to absorb energy cost increases caused by AI infrastructure.
In space, the calculation changes. Solar power is available essentially around the clock outside of brief eclipse periods, without weather interference. The vacuum of space provides a natural environment for radiative cooling, which could reduce or eliminate the need for water based cooling systems that have become a major point of local controversy around terrestrial data centres.
The challenge, of course, is getting the hardware up there, keeping it operational, and getting data to and from orbit fast enough to be useful. This is precisely where SpaceX’s core business, cheap and frequent launches through Starship, becomes the enabling technology for the whole plan.
The Moon Pivot Was Not Random
In the weeks before the IPO, Musk publicly shifted SpaceX’s stated priority from Mars to building what he called a self-growing city on the Moon. At the time, this looked like a strange moment to change a decades old mission statement. In context, it makes more sense.
Musk explained on the Dwarkesh podcast that the Moon could serve as a launch base for additional satellites and a manufacturing site for components needed for orbital data centres, specifically radiators for heat dissipation and solar cells for power. Lunar gravity is a fraction of Earth’s, which makes launching components from the Moon’s surface dramatically cheaper than launching them from Earth once any manufacturing capability exists there.
During an xAI all hands meeting reported by the New York Times, Musk told employees that xAI needs a lunar manufacturing facility, describing it as a factory on the Moon that builds AI satellites and flings them into space using a giant catapult. He told staff that if you are moving faster than anyone else in any technology arena, you become the leader, and said xAI intends to be that leader in computing scale.
| The Moon, in this plan, is not a destination. It is a factory floor for building the infrastructure that AI will eventually run on in orbit. |
The xAI Merger Is the Missing Piece
None of this fully makes sense without understanding the SpaceX and xAI merger that closed just before the IPO. xAI is Musk’s AI company, the one behind the Grok chatbot, and it also owns X, the social media platform formerly known as Twitter, which xAI acquired in 2025.
By merging xAI into SpaceX ahead of the public listing, Musk created a single company that controls rocket launch capability, the world’s largest satellite internet constellation through Starlink, an AI model development company, and a major social media platform with hundreds of millions of users generating the kind of data that trains large AI models.
Musk described the combined entity as the most ambitious, vertically integrated innovation engine on and off Earth, covering AI, rockets, space based internet, direct to mobile device communications, and what he called the world’s foremost real time information and free speech platform. Whatever you think of that framing, the structural logic is clear. SpaceX now has every piece needed to build, launch, and operate AI infrastructure in orbit under one corporate roof, funded by one of the largest capital raises in financial history.
How Realistic Is Any of This?
It is worth being honest about the gap between announcement and reality here. SpaceX has a strong track record of hitting ambitious targets eventually, even when the original timeline slips significantly. The company’s reusable rocket program was widely doubted before it became routine. Starlink went from concept to the dominant satellite internet provider on Earth within a few years.
At the same time, SpaceX’s own history includes plenty of timeline slippage. Musk previously said SpaceX would land humans on Mars by 2026. That did not happen, and the company has now pivoted away from Mars as the near term priority entirely. A 2028 start date for orbital data centres should be read as a direction of travel and a statement of intent backed by real capital, not a guaranteed delivery date.
What is different this time is the scale of capital behind the ambition. Raising 75 billion dollars in a single transaction, with a significant portion explicitly earmarked for satellite deployment and space based infrastructure, gives this plan a level of financial backing that most space based proposals never come close to.
What This Means for the UAE’s AI Infrastructure Race
The UAE has spent the last two years positioning itself as one of the most important hubs for AI infrastructure outside the United States and China. G42, Mubadala, and Khalifa University are all deeply embedded in partnerships with OpenAI, Microsoft, and other major AI labs. The OpenAI for Countries initiative includes a major computing facility being built in the UAE.
If SpaceX’s orbital data centre plan moves forward even partially on its stated timeline, it introduces an entirely new category of competition into the AI infrastructure race. Ground based data centres, the kind the UAE is currently building at enormous scale, would face competition from a fundamentally different architecture that does not depend on land, local power grids, or water resources at all.
This does not make the UAE’s current investments obsolete. Orbital data centres, even if they arrive on schedule, will likely serve as a complement to ground based infrastructure rather than a replacement, at least for the foreseeable future. Latency, bandwidth between orbit and ground stations, and the sheer complexity of maintaining hardware in space all remain significant constraints.
But for a country that has built its AI strategy around being an indispensable physical hub for AI compute, the emergence of a credible space based alternative, backed by the single largest capital raise in financial history, is a development worth watching closely. The UAE’s space agency and its growing satellite ambitions, including its own Mars and Moon missions launched via SpaceX, mean the country already has a foothold in the infrastructure layer this plan depends on.
| The UAE built its AI strategy on being the best place on Earth to put a data centre. SpaceX just raised 75 billion dollars partly betting that the best place for a data centre will not be on Earth at all. |
| Whether or not SpaceX hits its 2028 target, the direction is now impossible to ignore. The AI infrastructure race has a new frontier, and it is literally above everyone’s heads. |
Sources: SEC SpaceX IPO prospectus (June 2026), CNBC (June 12, 2026), Yahoo Finance and New York Times reporting on xAI all hands meeting, AOL Finance (Dwarkesh podcast coverage), Yahoo News SpaceX Moon pivot reporting, Microsoft AI Economy Institute Q1 2026, OpenAI for Countries UAE announcement.
Robius.news — Dubai, UAE — 2026 | Built to be first. Built to be trusted.
