A SpaceX Starlink satellite broke apart in low Earth orbit on March 29, scattering debris at roughly 560 kilometers altitude. SpaceX confirmed the anomaly the next day. LeoLabs, which tracks objects in orbit, detected tens of fragments and said more may be present. SpaceX stated the debris posed no threat to the International Space Station, the upcoming Artemis 2 lunar mission, or other active satellites.
The more interesting question is what happened five months before it, and what SpaceX did differently each time.
On December 17, 2025, Starlink satellite 35956 experienced a failure that caused its propulsion system to vent and triggered a rapid altitude decay of about four kilometers, according to Ars Technica. LeoLabs identified the event as originating from an internal energetic source, most likely a propellant tank or battery, rather than a collision with orbital debris. After that failure, SpaceX said it was deploying software updates to increase protections against similar events. The company then paused Starlink launches: after a pair of missions on December 17, the next Starlink launch did not occur until January 4, SpaceNews reported.
The March 29 failure shares the same broad classification: an internal energetic source producing debris. SpaceX has not specified whether the two events share a root cause. The company did not pause launches this time. A Falcon 9 lifted off from Cape Canaveral Space Force Station roughly six hours after SpaceX confirmed the March anomaly, carrying 29 Starlink satellites on the Starlink 10-44 mission, SpaceNews reported.
That difference in response is the part worth examining. Either the software fix worked and this was a new, unanticipated failure mode, or it did not work and SpaceX determined the risk acceptable enough to continue flying. The company has not publicly explained the reasoning. Neither explanation is comfortable.
One possibility is that the December software update addressed the specific failure mode on 35956 but missed something broader. The December event involved propulsion system venting, which is a particular class of failure. The March breakup, at a different altitude with a different satellite, may involve a different component or mechanism that the previous fix did not cover. Propellant tanks and batteries are both internal energetic sources that LeoLabs has flagged in this class of anomaly. They are not the same subsystem.
Another possibility is that SpaceX has decided the statistical risk of a launch pause no longer justifies the operational cost. The Starlink constellation now numbers over 10,020 satellites in low Earth orbit as of March 2026, per Wikipedia. At that scale, individual satellite failures are not exceptional events. The calculus for when to stop flying changes as the fleet grows.
The fragments from the March 29 event are expected to deorbit within a few weeks given the low altitude, LeoLabs stated. That is the one piece of good news in this: unlike the December event, which produced debris with a longer orbital lifetime, the March fragments are not lingering.
What is worth watching is whether SpaceX provides a root cause analysis for the March failure, and whether it matches or diverges from what caused the December event. The company has not committed to a public explanation in either case. If the failure modes are different, it suggests either that the December fix was narrower than advertised or that there are multiple unaddressed failure pathways in the Starlink bus that have not yet fully revealed themselves. If they are the same, it means the fix did not work.
The Pentagon has contracted SpaceX for Starlink services to Ukraine through $40 million in extensions by mid-2025, according to AInvest citing DSCA and Bloomberg, separate from a $150 million Foreign Military Sales agreement. Ukraine's experience with Starlink as critical wartime infrastructure has already raised questions about reliance on privately operated satellite networks for national resilience, as documented by CircleID. Two failures in under four months, both internal in origin, adds a hardware reliability dimension to that debate that SpaceX has not yet addressed publicly.
For now, the debris is the least of the problems. The pattern is the story.
