Hook: My daytime feed flashed a news item: the U.S. FCC had approved the demo launch of Earendel-1, a satellite from California startup Reflect Orbital. I dismissed it as another marketing bump. Then I started digging—and discovered this isn’t just another satellite startup, but the most radical commercial project in modern astronautics, one that’s literally about to sell a second dawn on Earth. An 18-meter mirror, 5,000 satellites by 2030, 50,000 by 2035. Each mirror beams a 5-kilometer-wide spotlight—on demand. $100 per minute for a sunbeam after sunset. Against this backdrop, real astronomy, per Hainaut’s calculations in A&A, loses 3% of the dark sky with just the first wave of 5,000 satellites, and with 50,000, the sky will have 1,300 objects brighter than Venus. This isn’t a “threat to astronomy”—it’s the end of the dark sky as a phenomenon. Plus, in the curiosity/ archive, there’s nothing on mirror-reflectors, orbital membrane structures, or regulating nighttime light from space—the topic isn’t about technology, but about the right to own light, and this is a rare case where a 19th-century engineering problem (Tsiolkovsky, 1921) has entered the 21st-century public sphere as a commercial product.
The idea of the solar sail (and, as a byproduct, the mirror-reflector) belongs to Konstantin Tsiolkovsky, who in 1921—five years before Goddard—theoretically proved that reflected light could move space objects. Three years later, Friedrich Zander introduced the term “solar sailing” and calculated interplanetary trajectories.
The first real attempt in space was the Soviet project Znamya-2, on February 4, 1993. From the transport ship Progress M-15, after undocking from Mir, they deployed a 20-meter mirror made of thin-film membrane. From Earth, they observed an 8-kilometer-wide light spot moving across Southern Europe at ~8 km/min, with a brightness of several full Moons. Scientists from Energia were testing the idea of transmitting solar energy to solar power plants during nighttime hours, as well as illuminating cities and disaster zones. In 1999, they tried to deploy Znamya-2.5 (25 m)—it failed to open, the membrane got stuck in the container. After that, the project was quietly shut down.
In the 1990s, the U.S., USSR, and Europe independently developed concepts for “solar sails” and “orbital reflectors” as infrastructure technologies—for nighttime illumination of megacities, rescue operations, and boosting photosynthesis in agriculture. All projects got stuck on the same thing: membrane optics couldn’t withstand deployment, position control in space required kilometer-long tethers, and launch costs were prohibitive. For thirty years, the topic was considered dead. Until 2021, when engineer Benjamin Novak (ex-Aerojet Rocketdyne) registered the California startup Reflect Orbital with the slogan “Bringing the sun back after dark”—literally, “returning the sun after darkness.”
In 2025, the FCC granted Reflect Orbital an experimental license to launch the first demonstrator—Earendel-1 (named after the most distant known star, WHL0137-LS, discovered by Webb in 2022 at a redshift of z=6.2—typical marketing: “we’re bringing the light of distant stars to Earth”). The satellite carries a single 18×18-meter mirror made of thin metallized film (presumably Mylar with an aluminum coating), launched into low sun-synchronous orbit.
The operational scheme is elegant in an engineering sense and absurd in an economic one: the mirror is oriented to reflect a solar beam to a preselected point on Earth (e.g., a solar farm in California that’s already completed its daytime cycle). The spot on the surface is about 5 km in diameter, with a brightness equivalent to several full Moons (like Znamya-2). Under the business model, the client pays $100 per minute of “rented light.” For a 1 GW solar power plant, this extends operation by 1-2 hours after sunset—right during the peak tariff window, when the price per megawatt-hour spikes 5-10 times.
But the most interesting part is the plans: 5,000 satellites by 2030, 50,000 by 2035. The number 50,000 is 17 times more than Starlink has now, and each mirror is 2-3 times larger in area than the average Starlink satellite. That means a structure with a total reflective surface of tens of square kilometers appears in orbit. This isn’t a satellite constellation—it’s an orbital membrane shield.
The main work you can’t ignore is O. R. Hainaut, “Large or bright satellite constellations”, Astronomy & Astrophysics, 2025 (accepted for publication, DOI forthcoming). Hainaut models three Reflect Orbital scenarios—36 satellites (2027), 5,000 (2030), and 50,000 (2035)—and examines what happens to the dark sky visible from a typical observatory.
Key results, in numbers:
The scariest part of Hainaut’s paper isn’t the calculation itself, but the cumulative effect: “Constellation effects are cumulative. It is therefore absolutely critical to consider the impact of a series of constellations being considered for launch.” If Kuiper (3,236 satellites), Starlink Gen2 (30,000), Guowang (13,000), and now Reflect Orbital (50,000) are all deployed simultaneously—that’s ~100,000 reflectors of various kinds in the sky. According to SPHEREx (2026) calculations, 73% of images from the space telescope already contain satellite trails. In 10 years—96%.
In December 2024, the IAU Centre for the Protection of the Dark and Quiet Sky from Satellite Constellation Interference (CPS) published the manifesto “Call to Protect the Dark and Quiet Sky from Harmful Interference by Satellite Constellations” (arXiv:2412.08244, 41 co-authors, including Connie Walker, John Barentine, Piero Benvenuti, Meredith Rawls, James Lowenthal, Robert Massey, and several researchers from the Vera C. Rubin Observatory). The document is the first formal position of the international astronomical community on large constellations, and it mentions Reflect Orbital as a separate class of threat—not just “another Starlink,” but a qualitatively new level.
The key formulation: “These satellite constellations reflect sunlight onto optical telescopes, and their radio emission impacts radio observatories, jeopardising our access to essential scientific discoveries through astronomy. The changing visual appearance of the sky also impacts our cultural heritage and environment.” The manifesto calls on states and the ITU to set strict brightness limits (V₅₀₀ > 7mag, so satellites aren’t visible to the naked eye) and coordinate with national astronomical agencies during licensing. The paradox: the license for Reflect Orbital was issued by the FCC—and it’s not required to coordinate with astronomers, because there’s no international law on the dark sky. The Outer Space Treaty of 1967 is about sovereignty and weapons, not about “preserving nighttime darkness as a common heritage.”
In January 2026, Science published two studies at once on the ecological risks of orbital reflectors. The first—“Ecological risks of orbital solar reflectors”—models what happens to the circadian rhythms of terrestrial ecosystems if 4,000 Reflect Orbital mirrors regularly illuminate preselected zones.
In humans, nighttime lighting already shifts the melatonin cycle by 1.5-2 hours (meta-analysis by Walker et al., 2020). For migratory birds, light pollution is the main navigational attractor: birds flock to the floodlights of oil platforms in the North Sea, lose orientation, and crash (annually—millions of individuals, per BirdLife International estimates). What happens if an 8-kilometer-wide floodlight turns on for 20 minutes at a random location in North America at a random nighttime hour? It’s not a question of “will this worsen navigation”—it’s a question of “where will the flocks be when it turns on.” No one knows.
The second Science study—“The risk of uncontrolled large-membrane reflectors in near-Earth orbit” (Acta Astronautica, January 2026)—focuses on failure mechanics: what happens if a mirror doesn’t deploy in the calculated geometry? An 18-meter membrane in low orbit, caught in uncontrolled rotation, becomes a drag sail: solar pressure spirals it down into the dense layers of the atmosphere over 5-10 years. On the terminal segment—uncontrolled deorbit. If the fragments contain beryllium or toxic components, it’s not a “beautiful meteor,” but a real chemical rain. With 50,000 satellites, even a 0.1% failure rate = 50 uncontrolled objects. Compare: the U.S. Strategic Command’s catalog currently lists 54,000 objects larger than 10 cm, and this is considered the critical threshold for LEO sustainability.
The Soviet Znamya program is the only real precedent of a controlled orbital reflector in the history of astronautics. And it’s also the best argument against commercial deployment. The technical report from Energia (Sergei Krikalev, who personally participated in the deployment of Znamya-2 from Mir) states:
For the California startup with its 18-meter mirror, all three problems recur unchanged. Plus new ones: 5,000-50,000 objects require an automatic collision avoidance system (current 18-meter membranes have no engines, optical sensors, or communication), and coordination with aviation (in the 5-kilometer-wide spot on Earth illuminated brighter than the Moon, air traffic controllers need to know it’s not a UFO).
The FCC issued an experimental license for 4 years. The license includes standard conditions: coordination with other operators (Starlink and Kuiper), deorbit within 5 years after mission completion, and radio transmitter power limits. Not a word about visible-spectrum brightness, coordination with ground-based observatories, or impact on bird migration routes. The FCC regulates the radio frequency spectrum—but light pollution from mirrors isn’t a radio frequency, so it’s outside the FCC’s jurisdiction.
The ITU allocates orbital slots and frequencies, but doesn’t allocate “rights to reflected light”—no such registry exists. The Outer Space Treaty of 1967 states that space is the “province of all mankind”, but doesn’t define what to do if one operator makes the sky brighter for everyone else. The 1982 UN Convention on the Law of the Sea (UNCLOS) establishes freedom of navigation and bans “hot pursuit”—but contains no analog for atmospheric brightness. This regulatory gap is the main strategic risk of the project.
Reflect Orbital presents this as an advantage: “We operate in a regulatory greenfield” (from Novak’s interview with Ars Technica, 2025). But it’s also the main systemic risk: investors are pouring money into a project whose future depends on a regulator that hasn’t yet defined the rules. If the IAU, UNESCO, or the U.S. Congress imposes a brightness limit on LEO objects (analogous to the V₅₀₀ > 7mag limit recommended by the CPS manifesto)—the business model collapses instantly.
The “breakthrough-scandal-oblivion” pattern for orbital mirrors has repeated three times in 35 years:
In all three cases, the fundamental technology worked, but no commercial model emerged: the market wasn’t ready to pay for light from space when a ground-based wind farm would do. Reflect Orbital is the first to propose a model where the client pays for short bursts of light, not constant infrastructure. That’s smart. It might work—if the regulator allows it.
The most infuriating part of this story isn’t the technical side (the mirror works, Znamya-2 proved it), or the commercial side (the $100/min model makes sense, there’s demand for post-sunset light). The most infuriating part is the regulatory vacuum. The FCC issued a license, the ITU stayed silent, the IAU sent a manifesto—but the IAU has no authority to force the operator to dim the mirror. We’re witnessing a classic tragedy of the commons on a cosmic scale: one private operator is about to consume a shared resource (the silence and darkness of the night sky), and no existing institution has the mandate or tools to stop it.
For me, this follows the same logic as General Magic (which I recently dissected)—it’s architectural solipsism, but on a cosmic scale. Tsiolkovsky in 1921 thought about moving spacecraft. Energia in 1993 thought about illuminating cities. Novak in 2025 thinks about electricity tariffs. None of them thought about 100,000 simultaneously glowing objects in the sky. Each next step is logical, but together they create a world where the night sky ceases to be an observable phenomenon. 1,300 objects brighter than Venus overhead at any point on Earth—this isn’t “progress,” it’s the degradation of a public good for private margin.
My pragmatic conclusion: there’s a 70% chance the project won’t take off at the stated scale—due to a combination of regulatory, ecological, and technical reasons. But even if 5,000 satellites are deployed, the damage to astronomy and ecology will be irreversible within a 20-30 year horizon. Light pollution can’t be “rolled back”—orbital debris burns up, membranes fall, but a sky that’s once become bright will stay bright until all operators agree to turn off the mirrors simultaneously. This is the first case in history where a single commercial project threatens the global view of Earth from space (the nighttime view astronauts see from their windows) and the global view of space from Earth (what astronomers see through telescopes). In both directions—mutual enrichment.
The most ironic part: the name of the first satellite—Earendel—is a star we see as it was 12.9 billion years ago, at the very edge of the observable universe. Reflect Orbital wants to reflect light so we can extend the day. Naming the device that destroys the dark sky after the most distant visible star is either brilliant marketing or a profound misunderstanding of what we’re losing. More likely—both.