SpaceX has taken an extraordinary step toward redefining both space infrastructure and global computing, seeking approval from the U.S. Federal Communications Commission for a satellite constellation of unprecedented scale that would function as an orbital data center powered largely by the sun.
In a filing submitted late on January 30, the company proposed deploying up to one million satellites in low Earth orbit, a figure that dwarfs any constellation previously contemplated. The satellites would operate at altitudes ranging from 500 to 2,000 kilometers, spread across 30-degree and sun-synchronous inclinations designed to maximize exposure to sunlight and, by extension, solar power generation.
SpaceX framed the project as a response to the mounting constraints facing terrestrial data centers, where energy costs, grid limitations, and environmental concerns are increasingly colliding with explosive demand for computing power driven by artificial intelligence.
Register for Tekedia Mini-MBA edition 19 (Feb 9 – May 2, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
“By directly harnessing near-constant solar power with little operating or maintenance cost, these satellites will achieve transformative cost and energy efficiency while significantly reducing the environmental impact associated with terrestrial data centers,” the company said in the filing.
The language of the application goes well beyond near-term commercial logic. SpaceX described the constellation as “a first step toward becoming a Kardashev Type II civilization,” a reference to a theoretical stage of technological development in which a civilization can harness the full energy output of its star. In more practical terms, the company argues that space-based computing could support AI-driven applications for billions of people while advancing its long-term vision of humanity as a multiplanetary species.
Even by the standards of an industry accustomed to bold plans, the scale is enormous. For comparison, China filed plans with the International Telecommunication Union in December for two constellations totaling nearly 200,000 satellites. Rwanda, in 2021, submitted ITU filings for constellations exceeding 300,000 satellites, linked to proposals by startup E-Space, which no longer appears to be pursuing systems of that magnitude. SpaceX’s proposal would exceed both by a wide margin.
Despite the headline number, the filing offers relatively few technical specifics. Details such as satellite size, mass, and precise orbital configurations were largely absent. SpaceX said it intends to place the spacecraft in “largely unused orbital altitudes” within the proposed range, an assertion likely to draw scrutiny as congestion, debris risk, and space traffic management become central regulatory concerns.
The operational concept relies heavily on intersatellite optical links, allowing the satellites to communicate with one another and with SpaceX’s existing Starlink broadband constellation. Data would be routed through space before being relayed to the ground, reducing dependence on continuous ground station connectivity. Ka-band communications would serve primarily as a backup for telemetry, tracking, and command, operating on what SpaceX described as a “non-interference, unprotected basis,” which underpins its request for regulatory flexibility.
One notable omission is any clear deployment timeline or cost estimate. SpaceX asked the FCC to waive standard milestone requirements that typically mandate deployment of half a licensed constellation within six years and full deployment within nine. The company argued that those rules are meant to prevent spectrum warehousing and should not apply because the Ka-band spectrum would be used on a non-interference basis.
The proposal arrives as SpaceX and its founder, Elon Musk, increasingly emphasize space-based computing as a solution to the AI industry’s surging appetite for power and processing capacity. That narrative has also been tied to SpaceX’s long-anticipated initial public offering, which could come as early as this summer and potentially raise tens of billions of dollars. Analysts see AI infrastructure, alongside Starlink, as a key pillar of the company’s long-term valuation story.
Speculation has also intensified around Musk’s broader corporate ecosystem. Recent reports suggest he has explored merging SpaceX with xAI, his artificial intelligence and social media venture, or even combining SpaceX with Tesla, which has invested heavily in AI for autonomous driving and robotics. An orbital data center network would fit neatly into such a vertically integrated vision, spanning energy generation, launch, connectivity, and compute.
In its filing, SpaceX leaned heavily on the economics of falling launch costs and rising terrestrial constraints. The company argued that as demand for AI accelerates, Earth-based data centers are becoming more expensive and politically contentious, particularly where they strain power grids and water resources. In contrast, SpaceX claimed that space-based computing could soon undercut terrestrial alternatives.
“Freed from the constraints of terrestrial deployment, within a few years the lowest cost to generate AI compute will be in space,” the company said, predicting advances in AI models at “unprecedented speeds and scales.”
Central to that claim is Starship, SpaceX’s next-generation launch vehicle, which the company plans to use to deploy both future Starlink satellites and the proposed orbital data center constellation. SpaceX said Starship’s ability to deliver massive payloads to orbit could allow space-based computing capacity to exceed the electricity consumption of the entire U.S. economy, without the need to overhaul Earth’s already strained power infrastructure.
Still, the proposal raises profound questions. A constellation of one million satellites would intensify debates over orbital congestion, collision risk, astronomical interference, and regulatory oversight at a time when governments are already struggling to adapt rules to far smaller systems. It also challenges assumptions about where the future of computing should reside, shifting it from land-based facilities to a permanently orbiting infrastructure.