Satellite operators are running out of room to dodge debris in low Earth orbit

The 18th Space Defense Squadron now sends out 600,000 conjunction data messages a day — automated warnings that two tracked objects in orbit are on a path to pass uncomfortably close. That is three times the daily average of around 200,000 in 2020. The volume alone tells you something has changed in low Earth orbit: collision avoidance has gone from an occasional maneuver to a full-time automated operation running continuously across thousands of satellites.

SpaceX's Starlink constellation is the biggest driver of that shift. Starlink satellites performed more than 355,000 collision avoidance maneuvers in the past year alone — over three times the total performed in 2024, and an average of roughly one maneuver every two minutes across the fleet. Most of these are small, automated nudges executed without human sign-off, using onboard propulsion and conjunction data from the Space Force's tracking network.

How close is "close"?

In August 2026, five separate Starlink conjunctions occurred on five consecutive days, illustrating how routine near-misses have become. In one of those encounters, a Starlink satellite and the MONOLITH spacecraft passed within six meters of each other — a distance smaller than a city bus, at closing speeds measured in kilometers per second. At orbital velocity, six meters is not a comfortable margin; it is closer to a coin flip that autonomous systems are increasingly being asked to call correctly, thousands of times a day.

The band that matters most

Not all of low Earth orbit carries equal risk. Researchers tracking debris density have flagged the 600-900 km altitude band as having already crossed the threshold past which a Kessler cascade — a self-sustaining chain reaction of collisions generating more debris, which causes more collisions — becomes a modeled possibility rather than a hypothetical one. Of the roughly 16 active or planned mega-constellations occupying LEO, six now exceed the critical satellite density associated with runaway debris growth in their operating bands, according to recent orbital-mechanics research.

The raw numbers underline the scale of the problem. As of February 2026, there were 8,377 active satellites and 5,214 dead ones in orbit — 13,591 tracked objects in total — alongside another roughly 16,200 tracked rocket bodies, mission debris, and fragmentation pieces. Every one of those objects is a potential conjunction partner for every satellite still under active control.

Nobody owns traffic control

The uncomfortable part of this story is not the technology — autonomous maneuvering systems work, and operators like SpaceX have gotten good at executing them at scale. The uncomfortable part is that there is no binding, universal space traffic management protocol. Each operator runs its own collision-avoidance logic against Space Force conjunction data, but there is no shared right-of-way rule, no mandatory minimum separation standard, and no single body with authority to order two operators to coordinate a joint maneuver. When two mega-constellations both maneuver independently in response to the same conjunction warning, they can — and occasionally do — move toward each other instead of apart.

Active debris removal exists as a partial answer, but it remains expensive and slow. ESA's ClearSpace-1 mission, contracted at roughly €86 million (about $93 million) to deorbit a single rocket adapter, illustrates the economics: removing one dead object costs tens of millions of dollars and takes years to plan and execute. Astroscale's ELSA-M mission, backed by €13.95 million from ESA and the UK Space Agency, will attempt something more scalable — repeated dockings with multiple OneWeb satellites using a single magnetic-capture servicer — but it is a demonstration, not yet an operational debris-removal fleet.

What to watch

The trajectory here is not ambiguous: CDM volume, maneuver counts, and mega-constellation density have all been rising for several years straight, and every operator's collision-avoidance workload will keep growing as more satellites launch. Three things are worth tracking as leading indicators of whether the situation stabilizes or deteriorates: whether the FCC, ITU, or a similar body moves from voluntary debris-mitigation guidelines to enforceable traffic-coordination rules; whether active debris removal missions move from single-object demonstrations to routine multi-object servicing at lower cost per object; and whether any operator reports a maneuver failure serious enough to generate a debris field, which would be the first real-world test of how fast a cascade in the 600-900 km band could actually unfold.

None of this requires a catastrophic collision to become expensive. Every operator flying in the 600-900 km band is already paying an ongoing tax in propellant, engineering attention, and mission risk simply to keep dodging. That tax is the actual cost of an unregulated orbital commons, and it is being paid right now, whether or not the worst-case cascade ever happens.