The debris inventory
What Is in Orbit: The Tracked and Untracked Object Population
46,000 objects tracked. 130 million estimated. Every one of them moves at 7.5 km/s.
ESA’s statistics, updated July 31, 2026, list approximately 46,160 objects regularly tracked around Earth. Roughly 16,000 are functioning satellites. The remainder: dead spacecraft, spent rocket bodies, hardware released during missions, and fragments from more than 660 breakup, explosion, collision, and anomalous events. The US Space Surveillance Network tracks objects larger than ~10 cm; as of 2026 it catalogs approximately 29,159 objects in this size class.
Below the tracking threshold, the population explodes. An estimated 1 million objects between 1 cm and 10 cm, and roughly 130 million objects between 1 mm and 10 cm, orbit Earth. These cannot be reliably tracked but can disable or destroy a satellite on impact. At LEO velocities (~7.5 km/s), a 1 cm aluminum sphere carries the kinetic energy of a hand grenade. A 10 cm fragment can shatter a satellite into thousands of new fragments, each of which becomes a projectile.
The tracked object count has grown 68% since 2019, from 17,700 to nearly 30,000 in seven years. The acceleration is driven by two sources: mega-constellation deployment (Starlink alone accounts for ~10,900 objects including operational satellites and associated debris) and legacy fragmentation events whose debris continues to spread.
The crowded shells
Altitude Bands and Orbital Shells: Where the Congestion Concentrates
Not all of LEO is equally crowded. The 500–600 km and 800–900 km bands carry most of the risk.
Orbital congestion is not uniform. It concentrates in specific altitude bands where operational constellations, legacy debris, and sun-synchronous orbits overlap. The most congested shells:
480–560 km (Starlink operational shell). The densest active constellation in history. Over 10,400 working Starlink satellites orbit in this band. SpaceX performed over 355,000 collision avoidance maneuvers in the past year — more than three times the 2024 total — with each satellite now maneuvering almost weekly. The band is actively managed: SpaceX’s autonomous collision avoidance system handles the conjunction volume, but every maneuver consumes propellant and shortens satellite life.
550–600 km (mixed use). Starlink V2, Amazon Leo (590–630 km), and numerous smaller constellations overlap in this band. It is also the altitude of the Hubble Space Telescope and many Earth observation satellites.
800–900 km (legacy debris peak). This is the most dangerous band in LEO. Atmospheric drag is negligible above ~700 km, so debris persists for centuries. The 2007 Chinese ASAT test (Fengyun-1C at 865 km) created over 3,000 tracked fragments still in orbit, spread across 200–4,000 km. The 2009 Iridium-Cosmos collision (at ~790 km) added roughly 2,300 more. Together these two events account for ~16% of all cataloged debris. Sun-synchronous orbits (used by most Earth observation satellites) cluster near 800 km, compounding the density.
1,100–1,200 km (emerging constellation band). OneWeb (648 sats at 1,200 km), Guowang and Qianfan (~1,100 km) are populating this band. Debris at this altitude persists for millennia.
35,786 km (GEO belt). Geostationary orbit is not congested in the LEO sense but is slot-limited. Only 1,800 orbital positions exist around the equator (spaced by 0.1°). GEO slot allocation is managed by the ITU and is increasingly contested — see The Orbits.
Kessler syndrome
Kessler Syndrome: The Cascading Collision Threshold
The point where collisions create debris faster than drag removes it. Some bands have already crossed it.
Donald Kessler proposed the cascading collision hypothesis in 1978: at a critical object density, random collisions between objects in orbit generate fragments faster than atmospheric drag can remove them, creating a self-sustaining cascade that progressively fills the altitude band with debris. The threshold is not a single number — it depends on altitude (drag rate), object mass distribution, velocity distribution, and whether active debris removal is occurring.
ESA’s 2026 Space Environment Report documents a 20% jump in collision probability in LEO. The 600–900 km band has exceeded modeled Kessler thresholds in some analyses. NASA debris evolution models suggest that preventing a cascade in the sun-synchronous band (near 800 km) requires actively removing 5–10 large objects per year — defunct satellites and rocket bodies with the highest collision probability. No operational active debris removal system exists at that scale, though several are in development (Astroscale, ClearSpace).
The five most consequential debris-creating events — the Fengyun-1C ASAT test (2007, 3,400+ fragments), the Iridium-Cosmos collision (2009, 2,300+ fragments), the Cosmos 1408 ASAT test (2021, 1,500+ fragments), and two legacy rocket body explosions — account for roughly 7,000 tracked fragments, nearly 25% of the entire catalog. Two deliberate weapons tests and one accidental collision define the modern debris environment.
Regulation
Debris Governance: The FCC 5-Year Rule and the International Gap
One country has a binding rule. The rest have guidelines.
The FCC adopted a 5-year post-mission deorbit rule in September 2022, effective September 29, 2024. It applies to all LEO satellites (below 2,000 km) that require FCC licensing or US market access — including foreign-licensed operators selling services in the US. It replaced the previous 25-year guideline (NASA, 1995; IADC, 2002). The first enforcement action was a $150,000 fine against DISH Network (2023) for failing to deorbit EchoStar-7.
The FAA withdrew its proposed 25-year disposal rule for upper stages and spacecraft in March 2026, creating a governance gap: the FCC covers satellites under its jurisdiction, but rocket bodies, experimental payloads, and non-US-licensed systems face no binding US deorbit requirement. Internationally, the IADC guidelines and the UN COPUOS Long-term Sustainability guidelines remain voluntary. No international body has enforcement authority over orbital debris.
Amazon has petitioned the FCC to roll back the 5-year rule, arguing it disproportionately burdens operators still in the deployment phase. SpaceX, which operates below 600 km where atmospheric drag assists natural deorbit, is the rule’s primary beneficiary: its satellites deorbit naturally within 5 years even without propulsion. Operators at 1,100+ km (OneWeb, Guowang, Qianfan) face a much harder compliance problem — debris at those altitudes persists for centuries without active deorbit.
Spectrum
Radio Frequency Spectrum: The Other Congestion Problem
Orbital space and radio spectrum are both finite commons. Spectrum may exhaust first.
Every satellite that communicates — downlink to users, uplink from gateways, inter-satellite links — occupies radio spectrum coordinated through the International Telecommunication Union (ITU). Spectrum is allocated by band (Ku, Ka, V, Q), by orbital regime, and by geographic service area. The ITU coordination process requires operators to demonstrate non-interference with existing filings — a process that can take years and increasingly constrains new constellation designs.
The problem is compounding. Starlink, Amazon Leo, OneWeb, Guowang, Qianfan, and the filed orbital data center constellations (Starmind, Starcloud, Sunrise) will collectively demand orders of magnitude more spectrum capacity than the current allocation framework was designed to manage. V-band and Q-band (above 40 GHz) offer more bandwidth but are more susceptible to rain fade and require more ground stations. Optical inter-satellite links avoid the spectrum problem entirely for inter-satellite routing, but every ground link still requires RF spectrum.
GEO orbital slots — fixed positions around the equatorial belt — are managed separately through the ITU and are already fully allocated in the most commercially valuable longitudes. The distinction between orbital slot congestion (GEO) and orbital space congestion (LEO) is important: LEO has no assigned “slots,” which is why it can become crowded without any single authority having said yes.
The darkening sky
Satellite Constellations and the End of Dark Skies for Ground-Based Astronomy
If all proposed constellations are built, the night sky will never be the same.
When the first Starlink satellites launched in 2019, astronomers immediately observed that reflected sunlight from the spacecraft was visible to the unaided eye and left bright streaks across telescope exposures. The problem has only grown. With 10,900 Starlink satellites in orbit and second-generation V2 satellites that are physically larger and more reflective, satellite trails now contaminate a measurable fraction of ground-based astronomical images at every wavelength from optical through radio.
The numbers. A May 2026 paper in Nature modeled the impact of all proposed constellations (assuming ~100,000+ satellites by 2040). The findings: 20% of midnight images from wide-field telescopes like the Vera C. Rubin Observatory will contain satellite trails. 30–80% of twilight exposures will be affected. The contamination extends to space-based telescopes as well — Hubble, SPHEREx, Xuntian, and ARRAKIHS will all see satellite trails from LEO constellations below them. The current ~16,000 satellites are less than 3% of those filed for launch in the next decade.
Radio interference. The IAU reported in 2024 that SpaceX’s second-generation Starlink satellites emit radio signals up to 30 times stronger than the first generation, interfering with radio telescopes operating in adjacent bands. The Square Kilometer Array (SKA), under construction in Australia and South Africa, is designed to detect extremely faint radio signals from the early universe. A full Starlink constellation operating in adjacent frequencies may render some SKA observations impossible.
The million-satellite scenario. If SpaceX’s Starmind (1 million satellites), Blue Origin’s Sunrise (51,600), Starcloud (88,000), and the Chinese constellations are all built as filed, the naked-eye night sky will contain more satellites than visible stars at any given moment during twilight hours. The cultural, scientific, and philosophical implications of permanently industrializing the visual night sky have no precedent. The IAU Centre for the Protection of the Dark and Quiet Sky from Satellite Constellation Interference coordinates the scientific community’s response, but there is no legal mechanism to enforce brightness limits on satellites. Mitigation efforts (darkened coatings, sunshades) are voluntary.
The counter-argument. Constellation operators note that the services these satellites provide — global broadband, Earth observation, GPS alternatives — benefit billions of people, including the developing world. SpaceX has invested in darkening coatings (VisorSat, DarkSat) to reduce reflectivity. The question is not whether constellations have value, but whether an irreversible transformation of the night sky is an acceptable externality of that value, and who gets to decide.
The million-satellite question
What Happens When a Million Satellites Are Filed For
The FCC filings add up to 1.3 million satellites. The current total is 16,000.
In the first quarter of 2026, FCC filings for orbital data center constellations alone exceeded 1.1 million satellites — SpaceX Starmind (1 million), Starcloud (88,000), and Blue Origin Sunrise (51,600). Add the existing broadband constellations (Starlink 15,000, Amazon Leo 7,736, Guowang 13,000, Qianfan 15,000), the filed Orbital constellation (100,000), and dozens of smaller systems, and the total filed satellite count approaches 1.3 million.
Whether these constellations are actually built depends on launch economics (Starship), regulatory approval (FCC proceedings are contested; Amazon has petitioned to block Starmind), spectrum availability, and capital. But the filings represent stated intent from organizations with demonstrated capability to build and launch at scale. If even 10% of the filed total is deployed, the LEO satellite population rises from 16,000 to over 130,000 — an eightfold increase that would transform every aspect of orbital congestion, debris risk, spectrum management, and astronomical observing conditions.
The governance gap is the structural problem. No single authority manages orbital space as a commons. The FCC regulates US-licensed satellites. The ITU coordinates spectrum. The IADC recommends debris guidelines. National space agencies enforce their own rules on their own licensees. No body has jurisdiction over the aggregate — the total number of objects in a given altitude band, the cumulative collision risk, or the brightness of the night sky. The question is whether governance will scale before the commons is consumed.