Musk Predicts AI Will Move to Space Within Five Years

AI's Next Frontier: Orbital Data Centres

Elon Musk has laid out an ambitious vision for the future of artificial intelligence infrastructure, predicting that within five years, SpaceX could be launching more AI computing capacity into orbit annually than the entire cumulative total currently operating on Earth.

In a podcast interview released on Thursday with Dwarkesh Patel and Stripe co-founder John Collison, Musk argued that space is the only viable path for scaling AI. "The only place you can really scale is space," he said. "Once you start thinking in terms of what percentage of the Sun's power you are harnessing, you realise you have to go to space."

The Solar Advantage

The economic case rests heavily on solar energy. According to Musk, solar panels in orbit operate approximately five times more effectively than ground-based installations, free from atmospheric interference, day-night cycles, and cloud cover. He claims space-based solar is effectively ten times cheaper when factoring in eliminated battery storage requirements.

SpaceX filed with the Federal Communications Commission on the thirty-first of January seeking authorisation to deploy a massive constellation of up to one million satellites, operating at altitudes between five hundred and two thousand kilometres using laser links for data transmission. The FCC accepted the application on the fourth of February and opened a public comment period with submissions due by the sixth of March.

A Trillion-Dollar Bet

The filing followed the formal merger of SpaceX with Musk's AI venture xAI, announced on the second of February, creating a combined entity valued at one point two five trillion dollars. The merger integrates SpaceX's expertise in rocket launches and satellite deployment with xAI's artificial intelligence capabilities.

Expert Pushback

However, scientists and industry leaders have raised serious concerns. Thermal management in the vacuum of space, where heat cannot dissipate through conduction or convection, presents a fundamental engineering challenge. Space debris risks, solar radiation damage to chips, and the impossibility of on-site hardware repairs add further complications.

AWS chief executive Matt Garman dismissed the concept as "just not economical," noting that there are not enough rockets to launch a million satellites and that current payload costs remain massive.

Despite the scepticism, Deutsche Bank Research noted that these appear to be engineering challenges rather than physical impossibilities, and several other companies including Google are pursuing similar concepts.