Galileo Didn't Have to Put Up With This
SpaceX has asked for permission to put up to 1 million satellites in front of the most powerful survey telescope ever built. The agency deciding is the one that happens to hold the radio spectrum.
On January 30, 2026, SpaceX asked the Federal Communications Commission for authority to launch and operate up to 1 million satellites. The system is called the SpaceX Orbital Data Center, and it would occupy altitudes between 500 and 2,000 kilometers, some of it in sun-synchronous orbits chosen to keep the spacecraft in sunlight almost continuously. The Space Bureau accepted the application for filing on February 4 and opened it for comment. It has not approved the constellation.
The comment record closed on March 23. On June 29, with the application still pending, the Vera C. Rubin Observatory in Chile began the most ambitious survey of the sky ever attempted.
Those two facts may be on a collision course, and astronomers should be heard before the Commission decides.
The reason became visible in a simulation featured recently by Scientific American. Shane Ross of Virginia Tech modeled what 500,000 of the proposed satellites would look like from Blacksburg, Virginia, 80 minutes after sunset, once assuming little effort to reduce brightness and once applying measurements from current Starlink satellites, whose reflectivity SpaceX has worked to mitigate. The gap is substantial, and the optimistic result still shifts with latitude and season. Under some conditions the satellites become a conspicuous new population in the night sky.
For most of us that changes what we see when we look up. For astronomers it changes what can be measured.
Rubin's 8.4 meter telescope and 3.2 gigapixel camera will photograph the southern sky over and over for 10 years, returning to each patch roughly 800 times, hunting hazardous asteroids and supernovae while feeding investigations of dark matter, dark energy, and everything in the sky that changes. Exposures run roughly 30 seconds across an unusually wide field. A satellite crossing that field leaves a streak through the observation.
Milky Way Glows Over Rubin. NSF-DOE Vera C. Rubin Observatory on Cerro Pachon, Chile. Photograph by Hernan Stockebrand, NOIRLab Audiovisual Ambassador. Credit: NSF-DOE Vera C. Rubin Observatory/NOIRLab/SLAC/AURA/H. Stockebrand. NOIRLab Image of the Week iotw2616a, licensed under CC BY 4.0.
The problem exists already, before anyone launches a data center. Rubin reports that simulations involving 40,000 low Earth orbit satellites found about 10% of its images could carry at least one satellite trail, and that most twilight observations could contain streaks. Twilight is not a marginal loss. It is when astronomers look for the asteroids approaching Earth from the direction of the Sun, the ones that arrive out of the glare. Brighter satellites contaminate larger portions of an exposure and push systematic errors into precision measurements. The stakes are not hypothetical. In 6 weeks of early operations before the survey formally began, Rubin catalogued more than 11,000 previously unknown asteroids, including 33 near-Earth objects.
Now replace 40,000 with a proposal whose ceiling is 1 million.
The American Astronomical Society has gone past expressing concern and formally asked the FCC to deny the application. Its filing argues that data center satellites could be larger and brighter than present Starlink craft, because computing at that scale demands substantial solar arrays and radiators to shed the heat. At the proposed numbers, scheduling around satellites stops working as a mitigation, and some spacecraft could saturate an entire Rubin image. In reply comments the Society also disputes SpaceX's claim to have reached the visual magnitude 7 brightness target with Starlink, noting that the target was stated rather than met.
Optical astronomy is only part of it. Computing produces heat, and heat radiated into space is infrared interference. Communications among that many spacecraft raise separate questions for radio astronomy. Launches, failures, collisions, debris, and reentries all scale differently when a system moves from thousands of objects to hundreds of thousands.
SpaceX deserves credit for taking brightness seriously with Starlink. Its engineers have demonstrated that design choices reduce the problem, which is precisely why the company should be asked to demonstrate it again against a constellation nearly 100 times the size of Starlink, which crossed 11,000 satellites in orbit in August and already accounts for about two-thirds of everything active overhead. Satellite communications deliver genuine benefits, and orbital computing may eventually achieve efficiencies that terrestrial infrastructure cannot. None of that settles the question, because scale changes the question. A constellation of this size is not an extension of a communications network. It is a decision about the physical character of near Earth space.
Why the FCC?
Nothing about an orbital data center is fundamentally a communications problem. The satellites are computers. They would talk to each other over optical links and hand data to Starlink for relay to the ground, carrying Ka-band equipment mainly as a backup for telemetry, tracking, and command. That thin sliver of radio is the entire hook on which federal permission hangs.
Congress created the Commission under the Communications Act of 1934 to regulate interstate and international communication by wire and radio. Because every spacecraft eventually transmits something, spectrum became the one authorization no operator can route around, and the FCC has served as the country's de facto satellite regulator for decades on that basis. Part 25 of its rules governs commercial space station licensing, and in April 2023 it consolidated that work into a dedicated Space Bureau.
Other agencies hold adjacent pieces. The FAA licenses launch and reentry under the Commercial Space Launch Act of 1984. NOAA licenses commercial remote sensing. No agency licenses the underlying decision about how many objects belong in low Earth orbit.
The question arrives at the FCC because there is nowhere else for it to arrive. The Society's petition to deny, DarkSky International's opposition, and nearly 1,500 public comments all landed in one docket for that reason.
The difficulty is that the agency with jurisdiction has spent 4 decades explaining why the night sky is not its concern. The Commission adopted its categorical exclusion from environmental review in 1986, written around effects on the Earth's surface. The Government Accountability Office reported in November 2022 that the FCC had not documented why that exclusion should cover large constellations, and recommended it publish the reasoning. The Commission agreed. As of mid-2024 the GAO still recorded the recommendations as open. In July 2024 the D.C. Circuit upheld the exclusion against the International Dark-Sky Association, holding that the Gen2 Starlink license fell within it.
Jurisdiction and appetite are not the same thing.
Its reach beyond radio is contested from the other direction as well. TechFreedom told the docket that astronomers are in the wrong forum, since Congress gave launch authority to the FAA. Brian Babin and Zoe Lofgren of the House Science Committee wrote in February that no clear congressional authorization exists for the FCC to regulate space safety or space traffic. The Commission adopted new satellite tracking rules in July anyway, unanimously.
So a decision about the observable sky will be made by the agency that happens to hold the radio spectrum, under a public interest standard written for broadcast licensing, constrained by an environmental exclusion adopted when a few hundred operational satellites orbited the Earth, while Congress disputes whether the agency may weigh space effects at all. The venue is an accident of statutory drafting from 1934.
Regulation has not caught up in the narrower sense either. Rubin notes that no limits exist on satellite optical emissions or reflectivity outside the radio spectrum, which leaves mitigation to voluntary cooperation between operators and astronomers. That works when dozens or thousands of objects need accommodating. It is much harder to defend when the number is six figures.
The commons above
Scientists have never been powerful lobbyists. American science built influence through universities, professional societies, federal research agencies, peer review, and the credibility of evidence. Those institutions can move public policy. They were not designed to compete with firms for which regulatory strategy is part of doing business.
Astronomers are not silent, and it would be lazy to pretend otherwise. The Society represents more than 8,500 astronomers, educators, and students, runs a public policy program, and files regulatory comments. Advocacy is real, but it remains adjunct to the scientific mission.
SpaceX operates differently. The company maintains an in-house Washington operation and retains outside firms. Federal disclosures compiled by OpenSecrets show $2.85 million in SpaceX lobbying across 2024, and $750,000 in the first quarter of 2026 alone. Lobbying dollars do not predict FCC outcomes, and they are not directly comparable to the work of a scientific society. Treat them as a measure of institutional design rather than influence. One organization was built to persuade regulators. The other was built to produce knowledge and was later asked to defend it.
The incentives point the same way. A company captures most of the value an orbital data center creates. The costs of interference with the night sky fall on everyone.
No company created that sky. No university owns it. No country contains it. Civilizations have used it for navigation, calendars, religion, storytelling, and science for thousands of years, and modern astronomy turned that inheritance into an instrument.
That instrument was expensive. Rubin cost roughly $810 million to build, and operating it runs about $72 million a year, split between the National Science Foundation and the Department of Energy. It is a decade of public spending aimed at a sky everyone assumed would stay observable.
It was also not built by one country. The telescope stands on Cerro Pachón in Chile. France supports construction and operations through CNRS and IN2P3. More than 40 international organizations and teams contribute, most of them in kind. Rubin is a coordination achievement across borders, and what passes in front of it will be settled in a single American docket, under a public interest standard written for American broadcasters, by an agency accountable to no one outside the United States. Astronomers in Chile and France can file comments. That is the whole of their standing.
AI now arrives as another claimant. The debate over AI infrastructure has concentrated on electricity, water, land, and the towns next to terrestrial data centers. Orbital computing looks at first like an escape from all of it, since sunlight is abundant, land is unnecessary, and cooling works by radiation. Moving infrastructure off the planet does not delete its externalities. It relocates them, and in this case the destination is above everyone.
The remedy is not prohibition. SpaceX should get the chance to show that its proposal can coexist with scientific observation, and the burden of proof should sit with the party proposing to place an unprecedented number of artificial objects into a shared environment. That means independent modeling of optical, infrared, and radio effects before deployment at meaningful scale, enforceable brightness requirements instead of voluntary promises, incremental deployment in which measurements from real spacecraft govern whether more launches are approved, and astronomers seated where the standards get written. A million satellites should be the end point of a long evidentiary process, if it is ever reached, not the number authorized before anyone understands the consequences.
SpaceX's filing describes the system as a first step toward a Kardashev Type II civilization, one capable of harnessing the full output of its star. The ambition is breathtaking. Civilizations should also know when to listen to their astronomers.
For thousands of years they have looked up and told the rest of us what is there. Galileo was told to stop, by an institution that said so plainly and wrote it down. His successors face nothing so direct. They face a categorical exclusion adopted in 1986, a comment period that closed in March, and an agency whose authority to weigh the question at all is disputed in Congress. Obstruction is easier to argue with when it announces itself.
Before we change what astronomers can see, we should hear what they are saying now.
Further Reading
- Space Bureau Accepts For Filing SpaceX's Application for Orbital Data Centers. DA 26-113, the public notice that opened the docket, Kardashev quotation and all.
- SpaceX files plans for million-satellite orbital data center constellation. SpaceNews on the architecture, including the optical links that make the Ka-band filing hook so thin.
- American Astronomical Society reply comments. Where the Society contests SpaceX's visual magnitude 7 claim directly.
- Satellite Licensing: FCC Should Reexamine Its Environmental Review Process. GAO-23-105005, the 2022 finding that the 1986 categorical exclusion was never reexamined for megaconstellations.
- International Dark-Sky Association v. FCC. The D.C. Circuit closing the NEPA route in July 2024.
- Who gets to make rules about space?. CNN on the turf war over whether the FCC may regulate space at all.