Starlink Near-Misses Prompt SpaceX Push for Orbital Data Sharing

A close-call problem that tracking alone can't fix

SpaceX has turned its size into an argument. On October 6 at the International Astronautical Congress in Antalya, Turkey, Starlink executive Michael Nicolls told the audience that low-Earth orbit can hold many more megaconstellations, but only if every operator tells the others where its satellites are and where they are going.11 The company had already posted the same message on its blog that week. It reported that Starlink has had close approaches with about 650 different third-party satellites that maneuvered in 2026, and that only about half of those operators share their trajectory data.1713

The main point is narrower than general worry about space crowding. Nicolls described debris as something like obstacles on a road. If you can measure it, you can predict where it will be and steer around it. What Starlink cannot plan for, he said, are spacecraft that change course without telling anyone.11 In those cases, he said, the other operators either were not publishing their ephemeris (a forecast of a satellite's position and velocity over time) or were ignoring other operators' published data when they maneuvered. The result was passes of tens to hundreds of meters, which he called way too close for comfort.1113

The incidents SpaceX chose to show

The company pointed to two cases. In August, a Starlink satellite passed within 134 meters of another spacecraft after the other vehicle unexpectedly changed direction. The Starlink satellite turned its solar arrays edge-on, which SpaceX calls a "knife-edge" profile, to present a smaller target.1712 The second case, from October 2025, was closer. Another satellite's unannounced burn cut the predicted miss distance from about 9 kilometers to 57 meters.1319 SpaceX also released images from a Starlink satellite's onboard camera showing another satellite passing nearby. That is unusual for an industry that rarely shows what a near-miss looks like.1113

The overall numbers are large. SpaceX says Starlink logged about 164,000 potential conjunction events over six months.1219 One outlet says the count covers events where the closest approach came within four hours after an unannounced maneuver, which leaves very little time to react.15 Starlink satellites make about 1,000 collision-avoidance maneuvers a day, according to Nicolls. Getting out of the way autonomously is now a "fairly common occurrence," and when there is too little time to move, a satellite can do a "ducking maneuver" and turn to reduce its exposure.11 SpaceX also says it contacted some operators to warn them about flaws in their own collision screening, and some had not known about the problems until then.19

Where the coverage agrees, and where it doesn't

The core facts are consistent across outlets: the 650 satellites, the roughly 50 percent sharing rate, the 134-meter and 57-meter encounters, and the 164,000-event count.11121317 Most of the differences are minor, but they are worth noting.

The first is Nicolls's title. One report calls him SpaceX's vice president for Starlink and president of xAI.11 Another calls him senior vice president of space systems.12 A tracking site calls him the Starlink VP.20 This doesn't change the substance, but it shows how closely Starlink's leadership is now tied to Musk's AI business.

The second difference matters more. One aggregator reported near-misses as close as 10 meters at altitudes between 500 and 600 kilometers, and said the data had gone to federal regulators. That outlet's own fact-check note says those figures still need to be verified against the original disclosure.14 Nothing from SpaceX or Nicolls backs up a 10-meter figure. Readers should treat the 57-meter encounter as the closest one SpaceX has documented.

The third is what SpaceX is asking for. Nicolls's request at the IAC was voluntary and global: start sharing data across all vehicles, whatever country you are from.11 One outlet goes further and reports that SpaceX wants the FCC and international spectrum regulators to make orbital licenses conditional on mandatory ephemeris sharing.15 No other report confirms that, so it should be read as a possible direction, not an official position.

The China question, and the limits of blaming others

Nicolls said sharing data with international operators has been especially hard. He also said there has been progress, noting that SpaceX shares data closely with China's space station and some Chinese satellite operators.11 A chart in his presentation suggested most of the gap involves Chinese and Russian satellites, though some operators in the US, Europe and Japan also don't share.11 Another report, based on SpaceX's own data, found that most of this year's close-approach satellites that withheld trajectory data were Chinese.19 Public tracking data shows how often these encounters happen. In the first two weeks of October, Starlink satellites passed within a few kilometers of Chinese Yaogan and Shijian spacecraft and Russian Rassvet-3 satellites.212223

SpaceX also identified reasons beyond geopolitics. Some operators won't publish planned maneuvers because they consider them proprietary. Others are willing but can't get government permission. SpaceX called both policies counterproductive and said they create preventable collision risk.13 That is an important point. It suggests the gaps come partly from industry practice and export rules, not only from rivalry between countries.

Competitors are mostly supportive. Rajeev Badyal of Amazon Leo said Amazon coordinates every launch and orbit-raising phase with SpaceX and meets regularly with the company to exchange data, so for Amazon, sharing is a "non-issue."1118 Not everyone accepts SpaceX's terms, though. Space-traffic firm Neuraspace argued that the long-term solution can't depend on every operator joining one company's platform. It called for a federated, interoperable system and said it had approached SpaceX about connecting with it.13 In my view, this is the most important pushback in the coverage. Data sharing is a good idea, but having it run through the infrastructure of the biggest operator raises reasonable questions about governance.

SpaceX's own footprint

SpaceX can credibly claim to be a model for responsible operation, but that claim has limits. Starlink has more than 11,000 satellites in orbit, about two-thirds of all active spacecraft in low-Earth orbit.11 Tracking data shows 12,988 launched, 11,137 still in orbit and 11,123 working.20 The company has moved its whole active constellation below 500 kilometers. That keeps it away from other busy altitudes, makes failed satellites reenter sooner, and improves service.11 Earlier this year SpaceX said it would move about 4,400 satellites to reduce collision risk.19

Still, Musk has talked about up to 100,000 Starlink satellites and as many as a million satellites for orbital AI data centers.17 SpaceX has applied for permission to operate 100,000 Starlink satellites and up to a million Starmind AI satellites, which one critic called a significant contribution to crowding in its own right.13 SpaceX's own record isn't spotless either. In 2019, a bug in its on-call paging system meant a Starlink satellite didn't move out of the way of ESA's Aeolus spacecraft.13 In 2021, China complained to the United Nations about two Starlinks that came close to Tiangong.12 The best reading is that SpaceX's argument is right on the merits, and that the company also has the most to gain from a system it has largely built itself.

Why Starship raises the stakes

The timing is tied to launch capacity. On September 28, Starship Flight 14 became the rocket's first orbital mission and its first with an operational payload, deploying 26 Starlink V3 satellites.6 The mission was cut from about ten hours to about three after a Raptor Vacuum engine shut down early in the ascent.6 SpaceX says each V3 satellite adds 1 Tbps of capacity, about ten times what a Falcon 9 load of V2 Mini satellites adds.6 The V3 satellites will fly at about 350 kilometers, even lower than the current fleet.11

Flight 15 is next. It will use Ship 42 and Booster 22.1 Planning documents moved its no-earlier-than date from October 19 to November 8.10 On October 10, Musk said he hoped Flight 15 would catch the ship with the tower "in 4-6 weeks." No Starship upper stage has ever been caught.9 In Florida, SpaceX says the work at Pad 39A has moved from construction into acceptance and activation testing. At least one more Starbase flight is planned before the first Florida launch.7 A December 10 date has been reported but remains unconfirmed.4 SpaceX has also asked the FAA to allow as many as 456 Starship launches a year.4

The lunar connection

The pace of these launches matters beyond Starlink. A Starship variant is NASA's human lunar lander. The first Artemis flight involving the Ship is Artemis III, now planned as a 2027 docking test in Earth orbit with Orion, launching from Pad 39A.7 Starship still has to show that it can transfer propellant in orbit, which both lunar and Mars missions depend on. Without that demonstration, a Mars attempt during this year's launch window is off the table, and late 2028 looks like the next realistic opportunity.5

The high-cadence launch plans, the lunar lander program and a 100,000-satellite network all lead back to the same point. SpaceX's long-term plans need a low-Earth orbit where operators can predict each other's movements. On that, Nicolls is right. What's still open is who sets the rules for that coordination and whether operators that currently stay quiet will go along with them.