The company's plan to launch orbital AI data centers could reshape compute economics for everyone, including decentralized networks
SpaceX wants to put up to 1 million satellites in orbit. Not for internet access this time, but to build the world’s largest AI data center, floating in space and powered by the sun. There’s just one catch: none of it works without Starship.
The company filed its FCC application for the constellation in late January 2026, and public comments on the proposal began in early February. The vision is staggering in scope.
What SpaceX is actually proposing #
The plan calls for orbital AI data centers equipped with massive solar arrays and passive cooling systems, each satellite sporting radiators with surface areas around 100 square meters for heat management. The initial spacecraft design stretches more than 170 meters in length, which is longer than the Starship V3 rocket that would carry them up.
The company envisions hourly Starship launches beginning as early as 2028, each hauling roughly 200 tons of payload. Initial satellite deployments are targeted for late 2027, manufactured at a new Gigasat Factory in Bastrop, Texas. Musk has claimed the constellation could eventually support AI applications for billions of users, targeting an annual compute capacity of approximately 100 GW.
SpaceX itself has acknowledged in regulatory filings that the orbital data center business might not reach commercial viability.
Why Starship is the bottleneck #
Everything about this project is contingent on Starship becoming what SpaceX promises it will be: the most powerful and cost-effective launch vehicle ever built, capable of frequent reuse. Without cheap, high-volume launches, the economics of putting a million satellites into orbit simply don’t pencil out.
Musk has suggested that within two to three years, these orbital data centers could represent the most cost-effective solution for AI compute needs.
SpaceX is approaching a potentially massive IPO that could value the company above $1.75 trillion. The orbital AI constellation is part of the narrative driving that valuation, alongside Starlink and the Mars colonization mission.
What investors should actually watch #
The challenges here are genuinely formidable: regulatory approval for a million-satellite constellation, manufacturing satellites at a pace never attempted, achieving launch frequencies that don’t exist yet, and managing orbital debris risks that could trigger international pushback.
A constellation of this size would significantly impact astronomical observations and add to an already crowded orbital environment.
For the crypto market, the key variable to monitor is Starship’s launch cadence over the next 18 months. If SpaceX demonstrates reliable, frequent launches by mid-2027, the orbital compute thesis gains credibility. If Starship continues facing delays, the million-satellite constellation remains what it is today: an ambitious FCC filing. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our