Environmental organizations have fired a direct warning shot at the Federal Communications Commission. They want the agency to freeze approvals for a wave of proposed orbital data centers until regulators complete a broad review of the projects’ collective effects on Earth.
The petition, filed July 8, comes as SpaceX and several startups race to put computing power in space. Their plans call for well over a million satellites. Proponents promise cheap AI training powered by constant sunlight. Critics see an unchecked experiment with the night sky, atmosphere and wildlife. The stakes have sharpened fast.
Earthjustice submitted the document on behalf of DarkSky International, Environment America and Public Employees for Environmental Responsibility. The filing asks the FCC to prepare a Programmatic Environmental Impact Statement under the National Environmental Policy Act before granting any more licenses. Such a review would examine risks, alternatives, costs and whether the entire concept serves the public interest.
“If ever a situation warranted a PEIS, it is this one,” the petition states, according to The Register. It adds that the FCC’s assumption these projects have no individual or cumulative environmental impact “is plainly inapplicable here.”
The numbers tell a story of sudden scale. In 2015 roughly 1,400 active satellites orbited Earth. By 2026 the total reached 15,000. Projections show another 58,000 added by 2030. SpaceX alone seeks permission for up to one million data-center satellites in low-Earth orbit. Other applicants including Starcloud, Orbital, Blue Origin and Cowboy Space bring the combined total far higher, reports SpaceNews.
But. The rush carries real consequences. Rocket launches pump emissions into the atmosphere. Satellites reentering burn up and release pollutants. Ozone degradation, orbital debris and light pollution all factor in. The petition highlights threats to the quality of the night sky, changes to the stratosphere’s chemistry and disruption of wildlife behavior tied to natural darkness.
“These projects could permanently alter the night sky as we know it,” said Ruskin Hartley, executive director of DarkSky International, in the Earthjustice press release. “The FCC needs to take seriously its obligation to ensure these projects do not cause unnecessary harm to naturally dark skies, or to our overall environment.”
Tim Whitehouse, executive director of Public Employees for Environmental Responsibility, put it sharper. “Allowing a million orbiting data centers with no environmental review isn’t just irresponsible — it’s reckless,” he said. “The potential for these projects to degrade the atmosphere with pollution and debris and harm wildlife needs to be carefully considered before licensing these projects.”
Jan Hasselman, senior attorney at Earthjustice, noted the petition marks a shift. Groups have pressed the FCC on individual megaconstellation applications for years, especially SpaceX’s Starlink. This marks the first call for one overarching review. “Drastically expanding satellites in space has a direct impact on people’s everyday lives as well as the future of our planet,” Hasselman said. “If we have to sue so that they comply, we will.”
The FCC has licensed satellites without environmental review in most cases. It relies on categorical exclusions under NEPA. A 2022 Government Accountability Office report urged the agency to reexamine that approach for large constellations, a point referenced in ongoing rulemakings. Satellite operators, including SpaceX, have pushed for even broader exemptions, arguing space lies outside U.S. environmental jurisdiction.
Yet the proposals arriving now differ in kind. These aren’t communications relays. They function as floating supercomputers. Companies envision racks of GPUs cooled by radiative panels, powered by solar arrays and linked by laser. The pitch sounds efficient. Reality bites harder.
Heat rejection in vacuum demands enormous surface area. One Nvidia H100 GPU draws 700 watts and needs about 1.4 square meters of radiator to run at 60 degrees Celsius, according to analysis in IEEE Spectrum. Scale that to a 40-kilowatt rack or a 100-megawatt facility and the hardware sprouts wings that could themselves become visible from the ground or contribute to debris risks. Starcloud has flown a single H100 to test the concept. Its radiator proved too small. The chip could not run at full power.
Manufacturing and launch cadence present equal barriers. Producing one million satellites at current Starlink rates would take decades. Even aggressive growth in Starship flights cannot close the gap quickly. Latency between orbiting nodes complicates synchronized AI training workloads. Inference might work. Large-scale model training faces physics that terrestrial data centers avoid.
Elon Musk has dismissed such doubts. “The lowest-cost place to put AI will be in space, and that will be true within two years, maybe three at the latest,” he told the World Economic Forum in Davos in January 2026. SpaceX filed its FCC application for the constellation days before a related corporate milestone. Jeff Bezos has echoed interest in orbital computing through Blue Origin, though with more measured timelines.
Industry backers argue space solves earthly problems. Power constraints on the ground grow tighter. Land for new facilities draws opposition. Constant solar exposure and radiative cooling could cut energy bills. Yet the petition counters that proponents tout “civilization-changing” benefits while refusing to detail downsides. A single programmatic review, it says, would force that conversation.
Technical skepticism runs deep among analysts. Michael Pierce of Technology Strategy Partners told IEEE Spectrum that Musk’s timelines appear overly ambitious. Cost parity might arrive in five to 10 years, he suggested, and then only for certain workloads. Matt Hasan, an AI strategist, pointed to lingering questions around launch costs, maintenance, thermal management and latency. The underlying demand for compute remains strong. The path through orbit stays uncertain.
Recent coverage adds texture. A June filing by startup Orbital sought approval for 100,000 data-center satellites aimed at delivering 10 gigawatts of compute, per SpaceNews reporting from late June. Starcloud’s earlier application for 88,000 satellites drew comments from the Secure World Foundation urging the FCC to treat it as precedent-setting and to weigh environmental questions explicitly.
The petition arrives as the FCC reconsiders its NEPA rules amid explosive growth in the space sector. A 2025 rulemaking sought input on expanding categorical exclusions. Satellite operators lobbied hard for exemptions. Environmental groups countered that the scale now proposed demands more study, not less.
Light pollution sits at the center of DarkSky’s concerns. Megaconstellations already complicate astronomical observations. Data centers with large radiator arrays could worsen the problem, reflecting sunlight in unpredictable patterns. Wildlife that relies on natural light cues for migration, foraging or reproduction stands to suffer. The stratosphere’s chemistry faces alteration from repeated reentries. Each piece compounds.
And the FCC has tools. It can pause individual applications. It can demand data on cumulative effects. A programmatic statement would set a framework for future licenses rather than litigate them one by one. That approach, petitioners argue, aligns with both law and prudence.
Whether the commission agrees remains open. Past pressure to accelerate licensing for commercial advantage has clashed with calls for caution. Debris mitigation rules have tightened. Five-year deorbit requirements replaced older 25-year guidelines. Yet environmental review itself has stayed limited.
The coalition’s move raises the temperature. A lawsuit threat hangs if licenses issue without study. Court precedent on NEPA’s application to space activities exists but remains unsettled at this scale. The petition frames the moment as a choice between haste and foresight.
Proponents see orbital data centers as the next logical step for AI infrastructure. Detractors view them as an uncontrolled industrial experiment in a shared global commons. The FCC now sits at the intersection. Its response will signal how seriously regulators take the environmental ledger of humanity’s expansion beyond Earth.
Additional recent reporting underscores the tension. E&E News described Earth’s orbit as the next environmental battleground, noting risks to human health and climate alongside the billionaire-driven proposals. Law360 highlighted the petition’s criticism of grandiose claims absent supporting environmental analysis. PCMag framed the effort as a direct brake on approvals amid SpaceX’s massive filing.
The debate will not fade quietly. Compute demand surges. Launch costs fall. Technical barriers persist. Regulatory gaps yawn. Somewhere between those forces lies the decision that could shape both the night sky and the future of artificial intelligence.
WebProNews is an iEntry Publication