Elon Musk has forecast that orbital data centers will become the lowest-cost option for AI computing within two to three years. Speaking at the World Economic Forum in Davos this January, the SpaceX founder made the bold claim as his company prepared to go public. The timeline is characteristically ambitious for an executive known for missed predictions.
SpaceX filed an application with the Federal Communications Commission later that month for a constellation of up to 1 million satellites in low Earth orbit, positioned between 500 and 2,000 kilometers above the planet. Musk also discussed initial design specifications for a new AI-1 satellite data center in a video interview conducted just three days before the IPO. The filings suggest serious intent behind the vision.
Industry analysts note the staggering scale of the proposal. There are currently approximately 14,500 active satellites in orbit; SpaceX's Starlink constellation already accounts for roughly two-thirds of that total. Launch cadences and satellite-manufacturing capacity would need to increase dramatically to support the planned orbital infrastructure. Musk has a history of overpromising timelines.
The economics of orbital data centers face significant hurdles. Power requirements, radiation hardening, and latency concerns all pose technical challenges that terrestrial data centers don't face. If successful, however, space-based computing could revolutionize AI processing by bypassing terrestrial energy constraints and land costs.