United Launch Alliance’s Vulcan rocket failed about 10 minutes after liftoff on November 15, destroying all 40 Amazon Leo broadband satellites aboard the mission. The loss is the first outright Vulcan mission failure and a material setback for Amazon’s plan to build a large low-Earth-orbit internet network.
What happened to the Vulcan mission
ULA said the vehicle “suffered an anomaly 10 minutes into the flight” and failed to deliver the Amazon mission to orbit. The company has not publicly identified the cause. The rocket launched from Space Launch Complex 41 at Cape Canaveral Space Force Station, Florida.
The payload was the first batch of Amazon Leo satellites assigned to a Vulcan rocket. Amazon had planned to place 40 spacecraft in low Earth orbit, where they would join a constellation designed to provide broadband connectivity beyond the reach of conventional terrestrial networks.
A major setback for Amazon Leo
Amazon had contracted ULA for 38 Vulcan launches. The agreement is intended to deliver more than 1,500 first-generation Leo satellites, making Vulcan a central part of the deployment plan rather than a one-off supplier.
The failed mission therefore affects more than one launch. Amazon loses the satellites on this flight, while ULA must investigate the vehicle, identify the failure boundary and determine what changes are needed before the next Vulcan mission carrying Leo hardware can fly.
Amazon’s broader launch strategy uses multiple providers, including Arianespace, Blue Origin and SpaceX. That diversification can reduce dependence on any one rocket, but it does not remove the schedule impact of losing a large payload at once. A constellation needs enough satellites in the right orbital planes to provide useful coverage, and missed launches can create gaps in that sequence.
Why the failure matters beyond one rocket
Large satellite networks are often discussed as software or connectivity projects. In practice, they are tightly coupled hardware, logistics and operations programmes. Guava Insights has examined a similar infrastructure lesson in its coverage of AI cloud expansion and the capital required to scale it: fast growth creates value only when the underlying capacity and operating model can keep up.
Service availability depends on manufacturing, launch cadence, orbital deployment, ground infrastructure, spectrum coordination and the ability to replace failed spacecraft.
The incident also shows why launch cadence is not the same as deployment resilience. Amazon and ULA had invested in dedicated processing infrastructure at Cape Canaveral so two Vulcan missions could be prepared in parallel and follow-on flights could occur within weeks of one another. That capacity is valuable only if the launch vehicle remains available and the investigation does not require a prolonged stand-down.
For businesses that may eventually depend on satellite connectivity, the practical question is not whether a provider has a large planned constellation. It is whether the provider has credible fallback paths when launches, satellites or ground systems fail. Redundant terrestrial links, multi-provider connectivity and clearly defined service limitations remain important even when low-Earth-orbit coverage expands.
What happens next
ULA’s immediate priority is to preserve telemetry and determine what caused the anomaly. Until that work is complete, it would be premature to predict how quickly Vulcan flights will resume or whether Amazon’s next Leo launch will use the same vehicle configuration.
Amazon has other launch providers and additional missions in its manifest. That gives the programme options, but it does not make the lost satellites recoverable or guarantee that launch slots can be rearranged without affecting the service rollout timetable.
The clearest near-term test is whether ULA can explain the failure with enough technical detail to support corrective action and whether the next Vulcan mission demonstrates that the fix works. For Amazon Leo, the episode is a reminder that building a global network requires resilience across every layer—not only the satellites and the software that coordinates them.


