Elon Musk is considering taking SpaceX public as early as mid-2026 to raise capital for a new line of business: data centres in orbit designed to power artificial intelligence systems, according to people familiar with the discussions cited by The Wall Street Journal.
SpaceX has for years signalled that it would avoid an initial public offering until it had achieved a sustained presence on Mars, arguing that remaining private allowed it to pursue long-term projects without the pressures of quarterly reporting. The reported shift reflects the scale of investment now required for AI computing infrastructure, and growing competition among technology and space companies to control the next generation of compute and connectivity.
The WSJ report said Musk has told associates he wants to complete an IPO by July 2026 and that SpaceX is expected to select banks to lead the offering. Separate reporting last month described SpaceX interviewing investment banks in a “bake-off” process, a step typically associated with preparation for a public listing.
The rationale, according to the WSJ’s sources, is to finance “space-based data centres” that would operate as solar-powered computing platforms, using satellites or modular spacecraft to host AI chips and related hardware. Musk is said to view the concept as a route to large-scale compute without the power, land, and cooling constraints that increasingly shape terrestrial data-centre buildouts.
The notion has moved rapidly from speculative to strategic. In November 2025, Google Research outlined a programme to test elements of a space-based AI infrastructure design, including a prototype mission planned for early 2027 as part of what it calls Project Suncatcher. The company’s published material describes a modular approach using interconnected satellites as a building block for future space computation.
Jeff Bezos’s Blue Origin has also been linked to orbital computing ambitions. Reuters reported in December that Blue Origin has been developing technology for AI data centres in space, as the industry explores whether compute can be shifted off-planet as demand rises. Separately, Blue Origin this week announced plans for a large satellite network aimed at enterprise and data-centre customers, underlining the commercial focus on high-capacity space infrastructure even where the primary product is connectivity rather than computation.
For Musk, the SpaceX angle also intersects with xAI, his AI company, which competes with OpenAI, Google and Anthropic. The WSJ reported that SpaceX and xAI already have operational and financial links, and that a SpaceX IPO could strengthen xAI’s access to infrastructure if orbital compute capacity were developed within the SpaceX ecosystem.
The wider AI sector has shown interest in pairing launch capability with compute. The WSJ has previously reported that OpenAI chief executive Sam Altman explored a deal involving rocket company Stoke Space, in the context of deploying satellites with AI computing capabilities. The competitive logic is straightforward: if compute and the means to place it in orbit are both scarce, the firms that control launch and platforms could shape pricing and access.
The engineering obstacles remain substantial. Space-based data centres would have to manage heat dissipation in vacuum, protect sensitive electronics from radiation, and maintain reliable high-bandwidth links to ground networks while moving at orbital velocities. Latency for some applications could be a limiting factor, depending on orbit and the distribution of ground stations. The WSJ said SpaceX believes it made a technical breakthrough in 2025, though details were not disclosed.
There is also a market question. SpaceX is already one of the world’s most valuable private companies, supported in part by its Starlink satellite broadband business and its dominant launch cadence. Reuters has reported that Starlink has about 10,000 satellites in orbit and more than six million users, figures that have helped underpin investor interest in any eventual flotation.
An IPO would bring new scrutiny. Public investors would want clarity on how much capital would be directed to orbital compute rather than launch vehicles, satellite internet expansion, or Mars-related development, and how returns would be generated in a market where the customer base for in-space AI compute is still emerging. It would also expose SpaceX to more detailed disclosure requirements, potentially affecting how it manages proprietary technology and government relationships.
For now, the plan remains contingent: on the readiness of the technology, on the willingness of customers to buy compute delivered from orbit, and on market conditions for a listing that could rank among the largest offerings of the decade. What is clearer is that the AI infrastructure race is pushing space companies beyond rockets and broadband, towards ownership of the compute layer itself.
