The next data race may begin above Earth.
Boca Chica, March 2026.
Elon Musk has unveiled one of his most ambitious space technology concepts to date: orbital data centers designed to power artificial intelligence from space. The project, presented through SpaceX, envisions a future constellation of giant satellites equipped with enormous solar panels and radiators, built to deliver computing capacity on a scale that would dwarf current orbital infrastructure. The proposal is striking not only for its engineering ambition, but for what it suggests about the next phase of the AI race: the search for energy, processing power and physical scale beyond the limits of terrestrial grids.
According to the plan described by Musk, these satellites would be larger in length than both Starship V3 and the International Space Station. The first model is presented as only a smaller precursor to much larger systems still to come. Their purpose would be to capture vast amounts of solar energy directly in orbit and use that power to sustain high density AI computation. Early versions are projected to provide around 100 kilowatts of computing power, with later generations expected to move into the megawatt range.
The deeper logic behind the project is easy to identify. Artificial intelligence is no longer constrained mainly by algorithms. It is increasingly constrained by electricity, cooling and data center capacity. On Earth, the infrastructure needed to train and run frontier models is becoming more expensive, more politically sensitive and more energy intensive. By pushing part of that architecture into orbit, Musk is effectively proposing an extreme answer to a real problem: if power and heat management are becoming bottlenecks on the ground, then space itself could be reframed as a new industrial zone for computation.
That vision, however, comes with extraordinary scale. Musk’s projection reportedly includes the possibility of launching up to one million satellites for orbital data operations, eventually producing thousands of gigawatts of AI computing capacity. Such a scenario would mark a radical transformation of near Earth orbit, not as a communications layer alone, but as an industrial computing environment. In strategic terms, this would turn space into something closer to digital territory, where control over launch systems, orbital slots, satellite constellations and collision management becomes inseparable from control over the future of AI infrastructure.
The promise is matched by equally serious concern. Experts in astronautics and orbital debris have warned that a constellation of this size would generate extreme traffic management challenges. Collision avoidance, already a growing issue with today’s satellite populations, could become a continuous operational burden if orbital AI platforms are deployed at the scale Musk is suggesting. A system involving millions of spacecraft would require an unprecedented number of risk mitigation maneuvers every year, raising questions not just of feasibility, but of sustainability, regulation and space governance.
This is where the project becomes more than a technological curiosity. It exposes the convergence of three strategic domains: artificial intelligence, energy systems and space power. If Musk succeeds, he will not simply launch new satellites. He will push the idea that the infrastructure of intelligence itself can be externalized beyond the planet. If he fails, the concept will still leave behind a powerful signal about where the frontier mindset of the tech sector is heading.
What matters now is not whether orbital AI data centers will appear tomorrow. It is that the competition for computing power has become so intense that even space is being recast as a possible solution. The age of digital expansion is no longer thinking only in terms of faster chips or larger campuses. It is beginning to think in constellations.
Phoenix24: intelligence for free audiences. / Phoenix24: intelligence for free audiences.