Three of the world’s most influential astronomy organizations filed formal objections on March 16, 2026, against two proposals they say would permanently damage the night sky: Reflect Orbital’s plan to place giant mirrors in orbit and SpaceX’s bid to launch up to a million satellites to power AI data centers. The joint protest, backed by peer-reviewed research showing that existing mitigation strategies break down at extreme satellite counts, sets up a direct clash between the accelerating demands of artificial intelligence infrastructure and the scientific community’s ability to observe the universe.
Major Science Bodies Unite Against Two Orbital Plans
The Royal Astronomical Society, European Southern Observatory, and International Astronomical Union submitted comments opposing both the SpaceX and Reflect Orbital proposals. Their language was blunt: the plans would “permanently scar” the night sky. That phrase carries weight because these are not activist groups or hobbyist clubs. They represent institutions that operate major observatories, set standards for astronomical nomenclature, and coordinate global observation campaigns that depend on dark skies.
The two proposals are distinct in design but linked in consequence. Reflect Orbital wants to place a large mirror in low Earth orbit capable of reflecting sunlight back to the ground, essentially creating an artificial light source visible from the surface. SpaceX, meanwhile, has filed plans for a constellation that could eventually number in the hundreds of thousands or more, intended to serve the booming demand for AI computing power. Astronomers argue that both would flood telescope sensors with unwanted light, degrading data quality for surveys that depend on faint, uncontaminated signals.
Why SpaceX Says AI Demands Justify the Scale
SpaceX explicitly cited the International Energy Agency’s analysis in its filing narrative to justify the enormous scale of the proposed constellation. That report projects that electricity demand from data centers could more than double by 2035, reaching roughly 1,200 to 1,700 TWh. The implication in SpaceX’s filing is that terrestrial power grids alone cannot keep pace with AI’s appetite, and that orbital infrastructure, including satellite-based connectivity and computing support, will be necessary to distribute the load and tap underused renewable resources.
The argument has internal logic. If AI workloads are growing as fast as the IEA forecasts suggest, traditional data center expansion will strain electrical grids in ways that create bottlenecks and local political pushback. Satellite networks could, in theory, help distribute processing or connectivity to regions with surplus clean power, smoothing demand. But the astronomy community sees a critical flaw in this reasoning: the environmental and scientific cost of filling low Earth orbit with reflective hardware is being treated as an externality rather than a constraint. No cost-benefit analysis in SpaceX’s filing appears to weigh the loss of sky access for science and culture against the commercial gains of more AI capacity.
Research Shows Mitigation Fails at Extreme Numbers
The scientific objections are not based on speculation. A peer-reviewed study published in The Astrophysical Journal Letters simulated the effects of satellite constellations numbering in the tens of thousands on the Vera C. Rubin Observatory’s Legacy Survey of Space and Time, known as LSST. That study tested avoidance strategies built into Rubin’s scheduling software and found that while some scheduling workarounds help at current constellation sizes, they hit hard limits as satellite numbers grow. The American Astronomical Society cited this research in its own petition against the SpaceX plan, arguing that trail avoidance cannot simply be scaled up indefinitely.
A related technical preprint by Rubin Observatory collaborators examined a different set of fixes: darkening satellite surfaces, adjusting sensor algorithms, and using avoidance pointing to steer telescopes away from satellite trails. The results showed that even after SpaceX applied experimental coatings to reduce Starlink satellite brightness, residual light contamination persisted. These are not theoretical complaints. They are measured outcomes from real hardware experiments, and they demonstrate that engineering fixes have ceilings that a million-satellite constellation would blow through, especially for wide-field surveys that cannot easily dodge thousands of moving objects.
An Existing Agreement That Cannot Scale
The National Science Foundation’s coordination agreement with SpaceX, negotiated during the initial rollout of Starlink, established a framework for dialogue and mitigation, including brightness-reduction targets and data-sharing protocols. At the time, it was seen as a reasonable compromise for a constellation of several thousand satellites, acknowledging both the commercial importance of broadband coverage and the need to protect key observatories.
Astronomers now argue that this earlier agreement actually strengthens their case. It proves that SpaceX and regulators acknowledged satellite interference with astronomy as a real problem worth addressing. But the agreement was designed for a constellation orders of magnitude smaller than what is now proposed. Scaling from thousands to potentially a million satellites is not a linear increase in interference; it is a qualitative shift that renders the old mitigation playbook inadequate. The NSF framework, in other words, becomes a benchmark that highlights how far the new proposal exceeds anything previously negotiated or tested.
Reflect Orbital and the Regulatory Vacuum
The orbiting mirror concept adds a different dimension to the controversy. Where SpaceX’s satellites would create streaks across telescope images, Reflect Orbital’s mirror could produce a single bright point of reflected sunlight visible to the naked eye, altering the visual character of the night sky for anyone looking up. “We just don’t have a regulatory process for these types of novel space activities yet,” said Roohi, a space policy expert quoted by The New York Times.
That regulatory gap is the thread connecting both proposals. The FCC reviews satellite applications primarily through a communications and spectrum lens, not a holistic assessment of cultural and scientific impacts on the night sky. There is no dedicated process to evaluate how a permanent artificial “star” or a sky crowded with moving lights would affect long-term survey projects, Indigenous sky traditions, or public access to natural darkness. Astronomers warn that without such a process, each approval sets a precedent that will be difficult to reverse.
Science Funding and the Asymmetry of Power
The clash also exposes an asymmetry between the resources available to commercial space operators and those available to the scientists trying to study their impacts. Major observatories and survey projects depend on competitive public funding mechanisms such as federal grant programs, which require detailed proposals, peer review, and long planning horizons. By contrast, launch providers can iterate hardware and deployment strategies quickly once they have a regulatory green light, reshaping the orbital environment faster than public agencies can respond.
Data on how much is invested in astronomy compared with commercial space ventures is scattered, but compilations like the National Center for Science and Engineering Statistics show that basic research funding, while substantial, is finite and often stretched across many disciplines. Astronomers argue that when orbital decisions compromise survey data, they effectively waste public investment in telescopes, instruments, and long-term observing campaigns. In their view, the opportunity cost of lost sky access should be counted alongside any projected gains from AI-optimized satellite networks.
Who Speaks for the Night Sky?
The current fight over orbital mirrors and AI-focused constellations is forcing a broader conversation about who, if anyone, has standing to speak for the night sky as a shared resource. Preprint servers such as arXiv’s member institutions have become crucial venues for rapidly disseminating technical analyses of satellite impacts, allowing observatories and researchers to coordinate responses before regulatory deadlines close. At the same time, the platform relies on community support, and appeals like arXiv’s donation campaigns underscore how fragile the infrastructure of open scientific communication can be compared with the capital backing commercial launch systems.
For now, the formal objections filed by the Royal Astronomical Society, European Southern Observatory, and International Astronomical Union do not guarantee that regulators will reject the SpaceX or Reflect Orbital proposals. But they crystallize a set of questions that can no longer be treated as niche concerns of professional stargazers. How many satellites are compatible with precision cosmology and planetary defense surveys? Should any company be allowed to place a permanent artificial light in the sky for marketing or demonstration purposes? And at what point does the cumulative impact of private projects amount to an irreversible transformation of a global commons that no single nation or firm has the moral authority to redesign?
As AI’s hunger for energy and connectivity accelerates, those questions will become harder to avoid. The outcome of this dispute will signal whether regulators see the night sky as an expendable backdrop for technological progress, or as a finite, shared environment that deserves the same level of scrutiny as land, air, and oceans. Astronomers hope that by putting hard numbers and peer-reviewed evidence on the table now, they can shift the debate from abstract enthusiasm about AI to a more grounded discussion of what humanity is willing to trade away in the name of computation.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
