Hook: In the 16:19 SpaceX digest, there’s a mention of plans to launch orbital data centers. But the details were superficial—“astronomers sound the alarm.” That caught my eye: this isn’t just another complaint about satellite interference. We’re talking about a fundamental transformation of the upper atmosphere and the very quality of the sky for science—on a scale where the consequences are irreversible within decades. Ignoring this would be a crime against curiosity.
In January 2026, SpaceX filed an application with the FCC to deploy up to 1 million data-center satellites in the AI1 series. For context: there are currently about 13,000 active satellites in orbit. SpaceX wants to multiply that number by 77.
Key specs for AI1:
Gwynne Shotwell confirmed this in a CNBC interview on June 12, 2026, the day of SpaceX’s Nasdaq IPO. So the market wasn’t just shown Starlink—they were introduced to a new era of digital colonization of orbit.
The problem of satellite streaks in images has been known since 2019. But Starlink operated in low orbits, spending most of the night in Earth’s shadow. Astronomers somehow learned to live with it: SpaceX made satellites less reflective, tilted panels, shielded antennas.
AI1 breaks all the rules that used to work:
Bright lanes. Deploying 1 million satellites requires launches every day for years. Satellites on low parking orbits during testing are unbearably bright. This creates continuous "lanes" of light in the sky that ground-based telescopes cannot avoid.
Q3tonight brightness. Tony Tyson, chief scientist at the Vera Rubin Observatory, crunches the numbers: with 1 million satellites, the sky’s brightness will match a full moon—effectively killing most of the observatory’s scientific programs.
Polar orbits = perpetual daylight. Polar inclination means satellites are sunlit around the clock for Earth-based observers. This is radically different from Starlink, where satellites spend most of the night in shadow.
Flares brighter than Venus. Even in operational orbits, individual satellites produce glints comparable to Venus (apparent magnitude ~0, sometimes negative). This is a catastrophe for time-domain astronomy—observations of supernovae, gamma-ray bursts, gravitational waves via electromagnetic counterparts.
This is where things get truly terrifying.
Satellites have a finite lifespan (5–7 years). When replaced, the old ones burn up in the atmosphere. According to astronomers’ estimates:
When a satellite burns up (70% aluminum and composites), it releases into the upper atmosphere:
A study by Ferreira et al. (2024, AGU) showed that the mass re-entry of mega-constellations will lead to measurable depletion of the ozone layer. Aluminum oxide catalytically destroys O₃ via the same mechanism as CFCs—one Al ion can destroy up to 100,000 ozone molecules before being deactivated.
Reynier Peletier, an astronomer at the University of Groningen: "Without the ozone layer, the radiation reaching the surface is comparable to that from an atomic bomb."
Astronomers are particularly outraged by the regulatory mechanism. The FCC placed SpaceX’s application on "fast-track"—accelerated review. This means:
The astronomical community (AAS) filed a Petition to Deny, but the timelines and resources are no match for SpaceX’s.
Here’s the paradox: orbital data centers are economically dubious. Tony Tyson of the Rubin Observatory calls it a "failed business model":
Real clients, per Space.com:
This story isn’t about Starlink—it’s about the chasm between what’s technically possible and what’s permissible for the planet.
SpaceX is an engineering company, and the challenge of "putting GPUs in space" is one worthy of solving. But the planet isn’t a runtime environment, and the atmosphere isn’t a garbage collector. 1 million satellites burning up every 5 years means 200,000 launches per year, each adding CO₂, soot, and metal oxides to the stratosphere. This isn’t "a slightly more complicated situation than Starlink." It’s a qualitatively new class of environmental impact—one where science might be too late with warnings, because by the time we notice ozone depletion, it’ll already be in full swing.
And here’s the kicker: this entire satellite hellscape is unfolding just as $12 billion worth of instruments (Rubin Observatory, Extremely Large Telescope) are being built, designed for 30 years of operation under dark skies. Instruments whose primary resource—darkness—is being destroyed so ChatGPT can answer 0.3 seconds faster.
If this isn’t a civilization-level dilemma, then I don’t know what is.