Musk Outlines Vision for Solar-Powered Data Centres in Space

Elon Musk is pushing the boundaries of data centre design beyond Earth, outlining a vision in which AI infrastructure operates in orbit and on the Moon. 

The plan follows SpaceX’s acquisition of Musk’s AI startup xAI, creating a single private company valued at an estimated US$1.25tn.

The combined organisation aims to build AI satellites that function as solar powered orbital data centres, shifting compute intensive workloads away from Earth’s constrained power and water resources.

The proposal represents an extreme response to familiar industry challenges around energy availability, sustainability and long-term scalability.

AI workloads already place significant strain on terrestrial infrastructure, particularly as high density training clusters demand reliable power and cooling at scale. Musk’s argument is that space offers a fundamentally different operating environment, where solar energy is abundant and uninterrupted.

Currently, Earth intercepts around 173,000 terawatts of solar energy, far exceeding global consumption. Musk believes even a fraction of this could support future AI infrastructure if accessed directly in space.

“To harness even a millionth of our Sun's energy would require over a million times more energy than our civilisation currently uses,” said Musk as he announced the companies' combination.

“The only logical solution therefore is to transport these resource-intensive efforts to a location with vast power and space. By directly harnessing near-constant solar power with little operating or maintenance costs, these satellites will transform our ability to scale compute.

“My estimate is that, within two to three years, the lowest cost way to generate AI compute will be in space.”

Orbital data centres and AI satellites

Under the proposal, AI satellites would operate as self-contained data centres in orbit, powered almost entirely by solar energy. The concept removes the need for water-intensive cooling systems and grid connections that constrain terrestrial facilities, while also sidestepping land availability and permitting challenges.

For Musk, orbital compute is not a supplement to Earth based data centres but a potential replacement for the most power-hungry AI workloads. Training and inference could take place in space, with data transmitted back to Earth as required.

This approach reframes data centre design around power capture and transmission rather than proximity to users or fibre routes. It also introduces new questions around latency, resilience and security that differ from conventional colocation or hyperscale environments.

Manufacturing and infrastructure on the Moon

Beyond orbit, Musk is also targeting the Moon as a manufacturing and logistics hub for AI infrastructure. SpaceX’s Starship, designed to carry up to 200 tonnes per flight, would transport cargo including satellites and equipment to lunar sites.

Once established, these sites could support both scientific activity and industrial scale manufacturing.

After cargo has landed on the Moon, it will be possible to establish “a permanent presence for scientific and manufacturing pursuits”, Musk continued.

He added: “Factories on the Moon can take advantage of lunar resources to manufacture satellites and deploy them further into space.

“By using an electromagnetic mass driver and lunar manufacturing, it is possible to put 500 to 1000TW/year of AI satellites into deep space, meaningfully ascend the Kardashev scale and harness a non-trivial percentage of the Sun's power.”

The Kardashev scale measures a civilisation’s ability to harness energy. Musk argues that space based AI infrastructure could move humanity closer to using a meaningful share of available solar power.

Energy strategy and reusability

Musk has reinforced these ideas in public forums, including the World Economic Forum’s Annual Meeting, where he linked AI growth, robotics and solar energy to future economic expansion. 

Central to the strategy is Starship’s reusability, which is intended to dramatically reduce the cost of deploying infrastructure into space.

He said: “If you had to throw away an aircraft after every flight, that would be a very expensive flight. If you only have to refuel, then it's the cost of the fuel.”

Lower launch costs are critical to making space-based data centres economically viable, particularly when compared with increasingly expensive terrestrial power generation and grid upgrades.

Implications for the data centre sector

While the concept remains speculative, Musk’s proposal highlights the scale of the challenges facing the data centre industry on Earth. 

Power availability, sustainability targets and long term compute growth are already reshaping where and how facilities are built.

Space-based AI data centres represent a radical alternative, shifting infrastructure design away from land, water and grid constraints. 

Whether or not the vision is realised, it underscores how energy has become the defining factor in AI infrastructure planning.

As demand for AI accelerates, Musk’s plan positions compute not as a fixed terrestrial asset but as something that could eventually operate wherever energy is most abundant.