TL;DR
xAI powers its data centres with unregulated gas turbines while SpaceX’s IPO pitches space-based solar. Tesla’s solar business is ignored.
The SpaceX IPO prospectus, filed on Wednesday, contains a vision for terawatt-scale space-based solar power. It also reveals, through what it does not say, that Elon Musk’s AI company xAI is running its data centres on unregulated natural gas turbines, with plans to buy $2.8 billion more. Tesla, the company Musk built on the promise of eliminating fossil fuels, barely features as a power supplier. The contradiction is now a matter of SEC record.
Tesla has released four Master Plans over the years. The through line has been consistent: electrification of the economy. In 2006, Musk described Tesla’s “overarching purpose” as helping “expedite the move from a mine-and-burn hydrocarbon economy towards a solar electric economy.” Just three years ago, Tesla’s Master Plan Part 3 outlined a detailed path to eliminate fossil fuels entirely. The document was rigorous, optimistic, and specific about the role of terrestrial solar, battery storage, and electrified transport in decarbonising the global economy.
Then xAI arrived. The AI company, which merged with SpaceX in February at a combined $1.25 trillion valuation, has embraced the mine-and-burn economy that Tesla was founded to replace. Dozens of unregulated natural gas turbines power xAI’s data centres in Memphis, Tennessee. The $2.8 billion in additional gas turbine purchases disclosed in the filing is not a temporary measure with an expiration date. It is a capital commitment that cements fossil fuel infrastructure into xAI’s operations for years.
Musk’s companies are not strangers to buying from each other. SpaceX spent $131 million on 1,279 Cybertrucks. xAI has spent $697 million over the past two years on Tesla Megapacks, the grid-scale battery storage systems used to manage peak loads at its data centres. But xAI has not purchased a materially significant number of solar panels from Tesla Energy, the division that exists specifically to deploy the technology Musk once described as the foundation of the future economy.
The SpaceX filing does mention solar, but only in the context of space. The company argues that space-based solar arrays can generate “more than five times the energy” of terrestrial ones thanks to continuous illumination. As AI data centres have encountered opposition on Earth, from neighbours, regulators, and grid operators, Musk and other executives have begun floating the idea of operating server racks in orbit, powered by 24/7 sunshine. SpaceX’s Starship programme, which has cost more than $15 billion to date, is positioned as the launch vehicle that could make this economically viable.
The economics, as TechCrunch’s Tim De Chant notes, are challenging at best. Power prices for Starlink satellites are already multiples higher than what a terrestrial data centre typically spends. Protecting AI chips from radiation, thermal cycling, and micrometeorites in orbit adds cost that does not exist on the ground. It is also unclear whether AI training workloads can be distributed across multiple satellites, which would leave a significant portion of the most compute-intensive AI work earthbound regardless of how cheap launches become. Shipping solar panels on a flatbed truck uses less energy than sending them into orbit.
The filing contains a more revealing claim. SpaceX argues that “third-party estimates on data centre demand are constrained by the practical supply limitations that exist in a terrestrial context and the power shortage may be far greater than what research estimates suggest.” The company references “terawatt-scale annual AI compute growth,” a figure that would represent a transformative increase in global energy demand. Humanity currently uses approximately 4 terawatts on a continuous basis. All the world’s data centres together consume roughly 40 gigawatts. Musk is projecting that AI alone will require additions measured in terawatts, every year.
The SpaceX IPO, expected to raise $75 billion next month, will be priced partly on the strength of this vision. Investors are being asked to buy into a future where terrestrial energy infrastructure is fundamentally insufficient for AI demand, and SpaceX is the company that can solve the problem from space. It is a compelling narrative. It is also a narrative that conveniently excuses the fact that, right now, Musk’s AI company is burning natural gas instead of deploying the solar technology his other company manufactures.
The energy problem for AI data centres is real. OpenAI paused its Stargate UK project over industrial electricity costs that run at more than four times US rates. Global data centre power consumption is projected to reach 150 GW by 2030. The question is not whether AI will need more energy, but whether the answer is to build more terrestrial solar, which has fallen in cost by 90% over the past decade and can be deployed at scale today, or to wait for a technology that requires launching hardware into orbit on rockets that, as of Friday, still cannot land their boosters reliably.
Tesla’s solar and energy storage business generated $2.8 billion in revenue in Q1 2026 alone. The Megapack factory in Lathrop, California ships grid-scale batteries to utilities and industrial customers worldwide. Tesla Energy is, by any measure, one of the most successful clean energy companies on the planet. And yet its founder’s newest company chose gas turbines instead.
Enterprise AI spending is accelerating at extraordinary rates. Salesforce projects $300 million in Anthropic token spending this year. The compute infrastructure behind that spending requires energy, and the companies building it are making choices right now about where that energy comes from. Musk’s choice, for xAI, was fossil fuel. His justification, via SpaceX, is that something better is coming from space. The gap between those two positions is filled with natural gas, and the Master Plan that was supposed to eliminate it.


