Revision #1

On January 30, 2026, SpaceX filed what may be the most audacious regulatory application in the history of telecommunications: a request for the Federal Communications Commission to authorize up to one million satellites — orbiting data centers designed to run artificial intelligence workloads in space [1]. Six weeks later, Amazon fired back with a formal 17-page petition urging the FCC to deny the application outright, calling it "speculative" and warning it would hand Elon Musk's company "an orbital monopoly that will make it the gatekeeper to space" [2].

The confrontation marks a dramatic new chapter in the intensifying rivalry between two of the world's wealthiest men and their sprawling technology empires. But beneath the corporate warfare lies a set of questions that will shape humanity's relationship with space for generations: Who gets to fill Earth's orbit? How many satellites are too many? And can the regulatory frameworks built for the broadcast era govern a future of orbital computing?

The Application That Stunned the Industry

SpaceX's filing, accepted by the FCC's Space Bureau on February 4, 2026, proposes a constellation of up to one million non-geostationary orbit (NGSO) satellites operating at altitudes between 500 and 2,000 kilometers [1]. The satellites would be arranged in two operational tiers: sun-synchronous orbits at higher altitudes, where spacecraft would remain in sunlight over 99% of the time for continuous computing, and lower-inclination orbits to handle peak demand periods [3].

The system would communicate primarily through intersatellite optical (laser) links, connecting to SpaceX's existing Starlink network, which would then relay data to ground stations. Ka-band radio operations would serve only as backup for telemetry and command functions, operating on a "non-interference, unprotected basis" [3].

The filing arrived just one day before SpaceX announced its $1.25 trillion merger with xAI, Musk's artificial intelligence company — the largest corporate merger in history [4]. Musk justified the union by pointing to the energy crisis facing AI: "Global electricity demand for AI simply cannot be met with terrestrial solutions without imposing hardship on the environment," he wrote [5]. Space, he argued, offers abundant solar power, natural cooling at near-absolute-zero temperatures, and virtually unlimited room to expand.

Source: GDELT Project

Data as of Mar 9, 2026CSV

Notably absent from the filing were several details one might expect in an application of this magnitude: no satellite design specifications, no mass or dimension data, no deployment timeline, and no cost estimates. SpaceX also requested a waiver of the FCC's standard milestone requirements, which typically mandate that half a constellation be deployed within six years and the full system within nine [3].

Amazon's Counterattack

Amazon's petition, filed on March 6 — the final day of the FCC's public comment period — did not mince words. Filed under the Amazon Leo brand (the company's rebranded satellite division, formerly Project Kuiper), the document argued that SpaceX's application "lacks basic details including satellite design, the radio frequency characteristics of the units," and critically, "any plan for managing conjunctions or interference at million-satellite scale" [2].

The petition characterized the submission as "a speculative placeholder rather than a complete application under the Commission's rules" [2]. Amazon's most pointed argument centered on logistics. At the record pace of 4,526 global satellite launches in 2025, deploying one million satellites would take more than 220 years [6]. And because satellites in low Earth orbit have a typical lifespan of roughly five years, sustaining a million-satellite constellation would require replacing 200,000 satellites annually — more than 44 times the entire world's satellite launch output in 2025 [6].

Amazon's core accusation went beyond feasibility: the company alleged SpaceX was attempting to "warehouse" orbital altitudes between 500 and 2,000 kilometers, effectively blocking competitors from accessing those orbital shells [5]. For a company that currently operates approximately 212 satellites through its Leo program and faces an FCC deadline to deploy half of its 3,236-satellite Gen1 constellation by July 2026, this framing is both strategic and existential [5].

The Frenemy Paradox

The competitive dynamics are made all the more remarkable by the fact that Amazon has been forced to rely on its rival to meet its own regulatory deadlines. After development delays with ULA's Vulcan rocket, Ariane 6, and Blue Origin's New Glenn — the latter owned by Amazon founder Jeff Bezos — Amazon contracted SpaceX for three Falcon 9 launches to deploy Kuiper satellites [5]. A lawsuit has even alleged that Amazon leadership allowed personal rivalry with Musk to interfere with the company's procurement decisions by avoiding SpaceX when it was the most cost-effective option [7].

AWS CEO Matt Garman added a technical dimension to the rivalry in early February, publicly dismissing orbital data centers as impractical at a Cisco AI Summit. "I don't know if you've seen a rack of servers lately: They're heavy," Garman quipped. "And last I checked, humanity has yet to build a permanent structure in space" [8]. He described the concept as "just not economical," noting that "there are not enough rockets to launch a million satellites yet" [9].

A Coalition of Concern

Amazon is far from alone in opposing the application. The FCC received over 1,200 public comments on the proposal — a volume that far exceeds typical satellite filings [6]. The American Astronomical Society and satellite operator Viasat filed separate comments also urging denial [10]. DarkSky International issued a public action alert, warning that the proposal would increase active satellites by roughly 70-fold and threatening to "end the night sky" as humanity has known it [11].

The scientific concerns are substantial. Aaron Boley, director of the Outer Space Institute at the University of British Columbia, warned that the plan "blows right past" safe orbital carrying capacity limits: "By almost all metrics...this is a bad idea in terms of our long-term use and access to space" [12].

The specter of Kessler Syndrome — a cascading chain reaction of orbital collisions that could render entire altitude bands unusable — looms large. Currently about 16,000 satellites orbit Earth, with approximately 14,000 active. SpaceX alone operates roughly 9,900 through Starlink [13]. Adding one million objects would fundamentally transform the orbital environment.

The atmospheric impact could be equally severe. Each satellite that reenters the atmosphere deposits aluminum and lithium residues. Dark-sky advocate John Barentine estimated that if all one million satellites were deorbited, it would mean one satellite reentry every three minutes [12]. Eloise Marais, an atmospheric chemist at University College London, warned: "We're doing this sort of experiment with the atmosphere when we don't really know what the result will be" [12].

The Regulatory Void

The FCC finds itself in an uncomfortable position. Its existing regulatory framework was not designed for proposals of this scale. The commission's ongoing "Space Modernization for the 21st Century" rulemaking, proposed in late 2025, aims to overhaul Part 25 of its rules governing satellite communications — replacing it entirely with a new Part 100 framework — but the process is still in the comment phase [14].

The proposed modernization includes a "modular licensing assembly line" for faster processing, new categories for non-traditional spacecraft, and expanded spectrum access in the upper C-band and millimeter-wave frequencies [14]. But none of these reforms were designed with million-satellite computing constellations in mind.

The comment period for SpaceX's orbital data center application closed on March 6, with responses to comments due March 16 and replies due March 23, 2026 [1]. No timeline for a ruling has been announced, and satellite constellation approvals typically take months to years.

Source: SpaceNews / SatelliteMap.space / FCC Filings

Data as of Mar 9, 2026CSV

The Bigger Picture: A Race to Space-Based Computing

SpaceX is not the only company eyeing orbital computing. Reports indicate that Google and Blue Origin are also exploring the concept [8]. Jeff Bezos himself has predicted that gigawatt-scale orbital data centers would "fill Earth's orbit within two decades" [2] — a statement that complicates Amazon's current opposition to SpaceX's plans.

The economic logic, if the engineering challenges can be solved, is compelling. Terrestrial data centers consume enormous amounts of electricity and water. The International Energy Agency has warned that global data center energy consumption could double by 2030, driven largely by AI workloads. Solar power in space is continuous and abundant, and the near-vacuum environment provides natural cooling that terrestrial facilities must achieve through expensive infrastructure.

But the gap between vision and execution remains vast. SpaceX's Starlink, the world's most advanced satellite constellation, has taken over six years to reach approximately 9,900 operational satellites [13]. Scaling to one million would require a manufacturing and launch revolution — even with SpaceX's Starship, the company's next-generation heavy-lift rocket that is still completing its test program.

What Happens Next

The FCC's decision will set a profound precedent. Approving the application, even in modified form, would signal that orbital computing is a legitimate use of near-Earth space and that the regulatory system is prepared to accommodate it. Denial would force SpaceX to return with a more detailed and demonstrably feasible proposal.

Several middle-ground outcomes are possible. The FCC could approve a drastically reduced constellation size, impose strict milestone requirements that SpaceX must meet to retain its license, or require detailed orbital debris mitigation plans before any launches begin. It could also defer the decision pending completion of its Space Modernization rulemaking, establishing clearer rules before adjudicating an application of this unprecedented scale.

What is certain is that the question of orbital computing is not going away. The convergence of AI's insatiable energy demands, the declining cost of space access, and the strategic ambitions of the world's wealthiest technologists has placed this debate squarely at the intersection of technology, regulation, and the future of the commons that is Earth's orbital environment.

As Boley put it: "It's really a global effect. So these launches affect everyone" [12]. The FCC's ruling — whenever it comes — will be watched not just by Amazon and SpaceX, but by every nation, scientist, and company with a stake in what happens above our heads.