Twenty seconds after liftoff on Feb. 12, one of the four solid boosters strapped to United Launch Alliance’s Vulcan rocket began burning through its own nozzle. A plume of flame showed where the exit cone should have been. The rocket kept climbing. Its methane-fueled core stage and hydrogen upper stage took on the load of the lost, uneven thrust, and the Space Force payload reached orbit anyway.
That flight, USSF-87, was Vulcan’s fourth. It was also its last. ULA said on Wednesday that the rocket will fly again on Nov. 13, nine months later, from Space Launch Complex 41 at Cape Canaveral Space Force Station in Florida. The date is pending approval from the Eastern Range.
The mission, called Leo 1, will carry 40 satellites for Amazon’s Leo broadband network. It is the fifth Vulcan flight and the first to use six of Northrop Grumman’s GEM 63XL boosters rather than two or four. Every one of those boosters uses the nozzle design the company has now modified twice.
The first failure came on Vulcan’s second flight, in October 2024. A nozzle on one of the two boosters disintegrated. Investigators traced that one to a manufacturing defect in an insulator meant to shield the nozzle’s metal structure from superheated exhaust. ULA changed the design and expected the problem to be behind it.
Sixteen months later the same part failed the same way. That is the fact that matters most about Nov. 13: the company is not returning to flight with a proven fix, but with a revised one that has never flown.
ULA has declined to describe the root cause in public. In a statement, the company said an integrated team of ULA, Northrop Grumman and Space Force engineers had “made great strides” in understanding the anomaly and had “identified several potential contributors which have been mitigated through design and process improvements to the nozzle.” Northrop test-fired a 63XL motor earlier this year as part of the work. The company said additional testing and analysis will continue alongside the mission, and that flying the improved design is the fastest way to gather data for customers.
The customers are watching closely. Vulcan is one of the rockets the Pentagon has certified to launch its most sensitive satellites, and it holds roughly half of the national security launch orders through the early 2030s. On Oct. 21, Space Systems Command confirmed it had moved six of those missions to SpaceX while Vulcan sat on the ground. Lt. Gen. Philip Garrant, who leads the command, disclosed the reassignment.
Brig. Gen. Eric Zarybnisky, the Space Force’s portfolio acquisition executive for space access, told reporters the same day that clearing Vulcan for military payloads would depend heavily on ground engineering and tests, and also on flight data from the commercial missions ULA flies. Leo 1 is therefore two tests at once: an Amazon deployment and an informal qualification flight for the Defense Department.
Amazon has its own deadline pressure. The company has nearly 400 Leo satellites in orbit and a license covering 3,236. Federal rules required half of that constellation, 1,616 satellites, in orbit by July 30. Amazon missed the date.
In June the Federal Communications Commission granted a conditional waiver, accepting the shortfall but stripping spectrum priority from satellites launched after the deadline until March 2028. Consumer service is planned to begin once 578 satellites are working.
That arithmetic explains why Amazon booked 38 Vulcan launches, enough to carry the bulk of the network, after starting with eight flights on ULA’s older Atlas V. It also explains why a nine-month gap in the launch schedule hurts more than a single lost mission.
Satellites are built in batches and stored. Rockets are the constraint. The 40 spacecraft on Leo 1 will be released from an adapter on the Centaur upper stage over about half an hour, a deployment sequence ULA has not run for this customer before.
The flight also tests a configuration, not just a component. Six boosters give Vulcan its highest thrust and its most asymmetric failure mode: if one nozzle lets go at low altitude, five others are pushing against a core stage that must correct for the difference. The previous two failures were survivable at two and four boosters. Whether the rocket can absorb a sixth has not been demonstrated in flight.
A broader question sits under the schedule. For two decades the U.S. government bought launches from a duopoly, and Vulcan was the replacement for the Atlas V, which depends on Russian-built engines that are no longer politically acceptable. ULA has spent more than a decade developing it.
Falcon 9 now flies most American payloads, and Blue Origin’s New Glenn is competing for the same national security work. A rocket that cannot fly on schedule is a rocket the Pentagon has to route around, as October showed.
ULA has not published a date for Vulcan’s next national security mission, and the Space Force has not said what data it needs before it restores certification. The range must still sign off on Nov. 13. Then the boosters have to behave for roughly two minutes.
If they do, ULA gets a launch cadence back and a case for its half of the military manifest. If a nozzle burns through a third time, the six missions already moved to SpaceX may look like the first of many.
