SpaceX Asks the FCC to Approve One Million Orbital Data Center Satellites, Claiming AI Compute Will Be Cheapest in Space
SpaceX filed to launch up to one million solar-powered satellites for orbital AI computing, drawing skepticism from astronomers and feasibility questions from experts.
Overview
SpaceX has asked the U.S. Federal Communications Commission for permission to deploy up to one million satellites in low Earth orbit that would function as solar-powered data centers for artificial intelligence workloads. The filing, submitted on January 30 and accepted by the FCC’s Space Bureau on February 4, represents the largest satellite constellation ever proposed and would dwarf every object currently in orbit by a factor of roughly 68.
The proposal has drawn immediate scrutiny from astronomers, space-debris researchers, and regulatory observers who question whether the project is technically feasible, environmentally responsible, or anything more than an aggressive spectrum-reservation play.
What We Know
The proposed constellation would place satellites at altitudes between 500 and 2,000 kilometers in 30-degree and sun-synchronous inclinations, according to SpaceNews. Sun-synchronous orbits would keep the satellites in sunlight more than 99 percent of the time, allowing near-continuous solar power generation with what SpaceX describes as “little operating or maintenance cost.”
Each satellite would generate approximately 100 kilowatts of computing power per metric ton of mass, according to details in the application reviewed by The Register. The satellites would communicate with each other using high-bandwidth optical intersatellite links and route traffic to the ground through SpaceX’s existing Starlink network. SpaceX also said the satellites will operate in the Ka-band as a backup for telemetry, tracking, and command on “a non-interference, unprotected basis,” according to SpaceNews.
SpaceX framed the project in sweeping terms. The filing describes the constellation as “a first step towards becoming a Kardashev II-level civilization — one that can harness the Sun’s full power,” as reported by TechCrunch. The company claims that Starship could deliver “unprecedented tonnage to orbit,” enabling orbital processing capacity that would eventually exceed “the electricity consumption of the entire U.S. economy.”
The economic argument hinges on a straightforward assertion: “Freed from the constraints of terrestrial deployment, within a few years the lowest cost to generate AI compute will be in space,” SpaceX stated in the filing, according to SpaceNews.
FCC Chairman Brendan Carr highlighted the filing on social media and indicated the Commission welcomes public input, as noted by Teslarati. The FCC is soliciting public comments with a deadline of March 6, according to The Register.
The Scale Problem
The numbers involved are staggering. Harvard astrophysicist Jonathan McDowell, who maintains a widely used satellite tracking database, told The Register that as of January 2026 there are roughly 14,518 active orbital payloads, of which 9,555 — about 66 percent — are already Starlink satellites. One million additional satellites would increase the active population by approximately 6,800 percent.
For context, China’s recent satellite constellation filings — themselves considered ambitious — total roughly 200,000 spacecraft, according to SpaceNews. SpaceX’s proposal is five times larger.
SpaceX has also requested several regulatory waivers, including exemptions from the FCC’s standard milestone requirements that typically mandate 50 percent deployment within six years and full deployment within nine. No specific deployment schedule or cost estimates were included in the filing, SpaceNews reported.
Concerns From Astronomers and Debris Experts
McDowell raised three categories of concern in his comments to The Register.
First, space debris: “A constellation like this will absolutely be required to have a fleet of tow-truck satellites to remove failed ones to avoid Kessler,” McDowell said, referring to the Kessler Syndrome — a theoretical cascading collision scenario in which debris generates more debris until entire orbital bands become unusable.
Second, astronomical interference: “One million satellites are going to be a big challenge for astronomy, especially as they are in higher orbits which is worse for us.” Existing large constellations already leave streaks across ground-based telescope exposures and interfere with space-based observations.
Third, atmospheric effects: SpaceX currently vaporizes an estimated one to two Starlink satellites per day during controlled reentry. Scaling to a million-satellite fleet would dramatically increase the volume of metallic particles deposited in the upper atmosphere, with effects that remain poorly understood.
The Register editorially characterized the proposal as a “ridiculous idea” that the FCC may have accepted for filing primarily to gather public opposition, though the agency itself has not publicly signaled a position.
What We Don’t Know
Several fundamental questions remain unanswered. The filing does not explain how SpaceX would dissipate heat from computing hardware in the vacuum of space, where convective cooling is impossible. It does not address latency implications for AI workloads that typically require low-latency interconnects between computing nodes. And it does not describe the satellite hardware in any technical detail beyond power-generation estimates.
The entire plan depends on Starship, SpaceX’s next-generation super-heavy-lift rocket, which has not yet demonstrated the routine, high-cadence launch operations that deploying a million satellites would require. SpaceX did not respond to requests for comment from The Register.
It also remains unclear whether the filing represents a genuine near-term engineering plan or a strategic move to reserve orbital slots and radio spectrum ahead of competitors. McDowell told The Register: “I think it’s unclear at this stage whether it’s feasible or not — but SpaceX seem to think it is.”
Analysis
The filing arrives amid an escalating race among hyperscalers and sovereign governments to secure data center capacity for AI training and inference. Terrestrial facilities face mounting challenges: power-grid constraints, water-cooling demands, permitting delays, and community opposition. SpaceX’s proposal reframes the bottleneck entirely — unlimited solar energy, no grid dependencies, no land-use disputes.
But the gap between filing an FCC application and placing a million functioning computers in orbit is vast. The proposal reads more like a declaration of ambition than an engineering blueprint. Whether the FCC treats it as a serious application or an aspirational placeholder will become clearer after the March 6 comment deadline passes.