Elon Musk’s decision to merge his artificial intelligence startup xAI with SpaceX has reignited serious discussion around a long-theorised idea: building AI data centres in space.
Once confined to science fiction and academic speculation, AI data centres in space are now being viewed through the lens of Musk’s expanding technological ecosystem, which spans rockets, satellites, AI systems and digital platforms. The deal has sharpened investor interest in whether Musk can turn a decades-old concept into operational reality.
“For the long term, space-based AI is obviously the only way to scale,” Musk said on Monday, arguing that even a tiny fraction of the Sun’s energy would dwarf the power currently consumed by human civilisation.
Old Concept, New Momentum
The idea of relocating energy-intensive computing to orbit has existed for decades. NASA engineers and U.S. government researchers explored space-based solar power in the 1970s, but abandoned it due to prohibitive launch and materials costs.
In recent years, the concept has re-emerged as computing demand has surged. Big Tech companies, including Alphabet and Jeff Bezos’ Blue Origin, have explored orbital infrastructure as a potential solution to energy constraints on Earth. Constant solar exposure and natural heat dissipation make space theoretically attractive for data centres.
What distinguishes Musk’s approach is scale and integration. SpaceX has already sought regulatory approval to deploy up to one million solar-powered satellites designed to function as an “orbital data-centre system,” according to filings with the U.S. Federal Communications Commission.
“Compute in space isn’t sci-fi anymore,” said David Ariosto, founder of space intelligence firm The Space Agency, pointing to Musk’s track record of commercialising ambitious ideas.
Economic and Technical Hurdles
Despite growing interest, experts caution that AI data centres in space face formidable technical and financial challenges. Radiation exposure, orbital debris, cooling requirements and maintenance costs remain major obstacles.
High-performance AI chips are particularly vulnerable to cosmic radiation. While radiation-hardened chips exist, they traditionally lag behind the performance of cutting-edge processors used in AI workloads.
Cooling is another constraint. Although space is cold, its near-vacuum prevents heat from dissipating naturally. Instead, systems must rely on large radiators to emit infrared energy, adding mass and cost to orbital platforms.
“There are real challenges here, particularly around cost-effectiveness,” said Armand Musey, founder of Summit Ridge Group. “But never say never. A lot of this is ultimately a bet on Elon.”
Several analysts believe commercially viable space-based data centres are still at least a decade away.
Why Musk May Have an Edge
Supporters argue the xAI–SpaceX merger controls more of the supply chain than any rival. SpaceX operates the world’s most active launch fleet, has demonstrated mass satellite production through Starlink, and has access to deep private capital.
“SpaceX has structural advantages that few others can match,” said Kathleen Curlee, a research analyst at Georgetown University, citing launch capacity, manufacturing experience and financial backing.
Other firms are also testing the concept. Google has subjected AI chips to radiation testing as part of its Project Suncatcher, which aims to create an orbital AI network using solar-powered satellites. A prototype launch is planned for 2027.
As Musk prepares SpaceX for a potential public offering valued at over $1.5 trillion, the xAI merger has added a new dimension to his long-term vision — one that places AI data centres in space at the frontier of future computing infrastructure.
