Amazon has challenged SpaceX's audacious plan to build a million-satellite orbital datacenter constellation, filing formal objections with US regulators and calling the proposal vague and incomplete.
In a March 6 petition to the Federal Communications Commission, Amazon Leo, the company's satellite broadband division, requested that the FCC reject SpaceX's January 30 application for permission to launch what would be an unprecedented orbital infrastructure project. The petition, filed just days after the FCC opened its public comment period, signals serious concern from a direct rival about the technical feasibility and regulatory implications of SpaceX's ambition.
SpaceX frames its proposed system as "a first step towards becoming a Kardashev II-level civilization," invoking a theoretical measure of technological advancement based on energy harnessing capacity. The company envisions constellations operating at altitudes ranging from 500 to 2,000 kilometres, with satellites serving as distributed computing nodes optimised for artificial intelligence processing. This would represent roughly 67 times the number of active satellites currently in orbit.
Amazon's objection, however, cuts to what it sees as a core problem: vagueness masquerading as engineering. The filing argues that SpaceX provides only "the barest outline of how SpaceX will deliver on its grand claims." Critically missing, Amazon contends, are satellite design specifications, detailed radio frequency characteristics, and any meaningful plan for managing conjunctions and interference at the scale of one million operational objects in low Earth orbit.
This last point carries particular weight. Current orbital altitude already hosts roughly 15,000 active satellites and tracked debris objects. A million-satellite constellation would introduce complexity that space operators have never managed. Amazon argues the application amounts to "a speculative placeholder rather than a complete application under the Commission's rules."
The environmental dimension cuts both ways. Amazon raises concerns about satellite interference with astronomical observations and the environmental cost of frequent deorbiting and reentry. Yet the irony is unavoidable: Amazon founder Jeff Bezos predicted last year that gigawatt-scale solar-powered datacenters will soon "fill Earth's orbit within two decades," suggesting Amazon's environmental argument may itself be selective.
What Amazon Leo really objects to, however, emerges in a single phrase buried in its filing: concerns that SpaceX is seeking authorization for "an orbital monopoly that will make it the gatekeeper to space." This cuts to the genuine commercial stakes. SpaceX's application poses a potential challenge to AI titans including Microsoft, Amazon, Google and OpenAI, all racing to secure orbital compute infrastructure for training and running advanced AI models.
Amazon Leo itself operates more than 200 satellites, with over 3,200 in its licensed constellation plan. The company must deploy half that fleet by July 2026 to meet FCC licensing requirements. SpaceX, meanwhile, already operates roughly 9,900 Starlink satellites. The orbital datacenter question is therefore not academic; it concerns which company will control crucial computing infrastructure for the next decade of AI development.
The question of feasibility remains open. Analyst firm Gartner recently described orbital datacenters as "peak insanity" because running spaceborne facilities would be uneconomical and could never satisfy terrestrial demand for compute. Starship, the launch vehicle SpaceX plans to use for deployment, is still undergoing test flights and has yet to achieve full operational capability.
The FCC's comment period has closed, and the regulatory review process will now determine whether SpaceX must provide far greater technical detail or whether the Commission accepts the application as sufficiently substantive for further review. Amazon's intervention ensures the decision will be closely scrutinised by an interested party with both technical expertise and commercial motivation.