xAI powers data centers with natural gas, eyes orbital solar

Elon Musk's new AI company, xAI, is powering its data centers with unregulated natural gas turbines, a surprising move given the broader ecosystem.

DK
David Katzman

May 23, 2026 · 3 min read

Futuristic xAI data center powered by natural gas turbines, with a SpaceX orbital solar array visible in the background against Earth's orbit.

Elon Musk's new AI company, xAI, is powering its data centers with unregulated natural gas turbines, a surprising move given the broader ecosystem. This occurs even as his other venture, SpaceX, champions a future where orbital solar arrays generate over five times more energy than terrestrial ones, effectively challenging the notion that Elon Musk abandons solar power on Earth for his AI ambitions in 2026. A critical strategic divide is highlighted by this stark contrast.

Musk's companies are making substantial investments in terrestrial energy infrastructure for AI, but his long-term vision focuses entirely on unproven space-based power. This tension reveals a calculated risk.

Companies pursuing AI at scale may increasingly face a dilemma between immediate, often carbon-intensive power solutions and highly speculative, futuristic energy sources, potentially delaying sustainable transitions while accelerating AI development.

Current Investments and Ground-Based Solutions

  • XAI has spent $697 million on Tesla Megapack battery systems over the last two years, according to Zamin Uz.

Despite these substantial investments in terrestrial battery storage, these ground-based solutions appear to be temporary measures in Musk's broader strategy. The financial commitment suggests a need for immediate, reliable power, even if the long-term vision lies elsewhere.

The Vision for Orbital Data Centers

SpaceX touts space-based solar arrays as the future of data center power, claiming they can generate more than five times the energy of terrestrial ones. This is due to 24/7 illumination, according to TechCrunch. This ambitious space-based energy vision promises unprecedented power generation, positioning it as the ultimate solution for future AI demands.

Such a system would theoretically overcome the intermittency of ground-based solar.

Ground Operations as a Stopgap

Musk considers xAI's current data centers as stopgaps until SpaceX can deploy orbital data centers, TechCrunch reports. This perspective reframes current ground-based operations as provisional, highlighting a strategic pivot towards a radical, long-term space solution.

The current reliance on natural gas and battery systems serves an immediate operational need.

Challenges and Future Demands

The economics and feasibility of space-based data centers face questions, according to TechCrunch. Higher power prices for Starlink satellites and challenges in protecting chips from space environments are notable hurdles. Significant economic and technical obstacles must be overcome for the ambitious space-based data center vision to become a viable reality.

These issues complicate the path to orbital AI infrastructure.

Addressing the Terawatt-Scale Need

Why is Elon Musk giving up on solar power?

Musk's xAI currently uses natural gas turbines, and his long-term focus is space-based solar, not terrestrial solar. This shift suggests a prioritization of immediate, high-power solutions for AI over ground-based renewable adoption.

What are Elon Musk's current energy projects?

XAI is investing in terrestrial battery systems like Tesla Megapacks and using natural gas turbines. Simultaneously, SpaceX is developing concepts for orbital data centers powered by space-based solar arrays.

What is the future of solar power without Musk?

Terrestrial solar power continues to expand globally, driven by other companies and governments. Musk's focus on space-based solutions for AI does not halt the broader growth and innovation in ground-based solar energy, which continues to see significant investment. SpaceX forecasts that the world will need additional terawatt-scale computing power annually, according to Zamin Uz.