Satellite mega-constellations are creating a collision-avoidance problem no one has solved

SpaceX's Starlink satellites performed roughly 300,000 automated collision-avoidance maneuvers in 2025 alone, according to FCC filings analyzed by Aerospace America. That figure covers one operator, in one year, and it is already large enough to make orbital traffic management one of the least-solved problems in an industry that otherwise moves fast. Researchers now project the number could reach 1 million maneuvers annually by 2027 as competing constellations expand alongside Starlink. The safety record so far is clean. The margin behind that record is thinner than it looks.

Why the maneuver count is so high

Part of the volume is simply scale: SpaceX has regulatory approval to grow Starlink to roughly 42,000 satellites, and the company is expected to begin launching a more capable satellite generation aboard Starship this year. But a large share of the maneuver count comes from a deliberate design choice — SpaceX screens for collision risk at a probability threshold of 3 in 10 million, far stricter than the industry-standard threshold of 1 in 10,000. That means Starlink satellites move out of the way for encounters that most other operators wouldn't even flag as a conjunction event, let alone act on.

That stricter threshold is good engineering practice for SpaceX's own fleet, but it exposes the structural gap in how orbital safety actually works: there is no shared, binding standard across operators. Each company sets its own risk tolerance, runs its own screening software, and decides independently whether and when to maneuver. Starlink's conservatism doesn't make nearby constellations from Amazon's Kuiper or China's state-backed mega-constellation programs any safer — it just means SpaceX is doing more of the defensive work in low Earth orbit than its market share alone would suggest.

The events that show the stakes

Tracking data gives a sense of how close "clean record" can get to not being clean. In a four-day window in March 2026, SOCRATES-based conjunction screening flagged nine separate threat events for Starlink satellites, including one classified as HIGH risk. In August 2026, a cluster of conjunctions included an encounter with a maximum calculated collision probability of 23.99% — roughly a one-in-four chance by the screening model's own estimate — before passing without incident.

Neither event became a collision, and that's the point astronomer Samantha Lawler has been making publicly: "Our ability to keep using orbit depends on Starlink continuing to operate perfectly." A system where the safety margin is "the operator doesn't make a mistake" rather than "the system has structural redundancy" is a system one bad week away from a different outcome — a software bug in a screening pipeline, a late maneuver, a satellite that loses thruster function at the wrong moment.

The Kessler problem nobody owns

The underlying risk is Kessler syndrome — a cascade where one collision generates debris that increases the odds of further collisions, eventually making entire orbital bands unusable. Low Earth orbit isn't there yet, but the trajectory matters more than the current state: multiple mega-constellations are expanding simultaneously, with no single body coordinating conjunction data or enforcing a shared maneuvering protocol across operators. The FCC regulates U.S. launches and licenses; it does not run traffic control for low Earth orbit the way the FAA runs it for commercial aviation.

That gap is structural, not incidental. Commercial aviation works because every aircraft operates under one shared, binding set of separation rules enforced by a neutral authority. Orbital traffic currently works because individual operators, mostly SpaceX, are choosing to be cautious. Those are not the same kind of safety guarantee, and the difference becomes more consequential as more operators with different risk tolerances occupy the same orbital shells.

What to watch for

The actionable signal isn't another clean year of Starlink maneuver statistics — that outcome is compatible with the system still being fragile. The signal to watch is whether a binding, cross-operator conjunction-data-sharing standard emerges, backed by something more than voluntary coordination between companies that compete with each other. Until an independent body can mandate maneuver thresholds and enforce them across every operator in a given orbital shell, the current safety record reflects operational discipline at a handful of companies, not a resilient system. For anyone tracking space policy, the real story in 2026 isn't the maneuver count — it's whether governance catches up to it before the count the industry can't absorb arrives.