United Launch Alliance’s Vulcan rocket lifted off from Cape Canaveral Space Force Station at 12:40 p.m. EST on Sunday, 15 November, carrying 40 satellites for Amazon’s Leo internet network. About 10 minutes later, the mission was over. The satellites never reached orbit.
It was the first outright failure in ULA’s 20-year history and the fifth flight of Vulcan, which debuted in January 2024 and had not flown in nine months. For a company whose Atlas and Delta rockets built a reputation for reliability, the loss is a serious blow. For Amazon, it is the latest delay in a satellite program that is already years behind its own ambitions.
What changed: On 18 November, ULA said its initial integrated review of flight data pointed to inadequate fuel supply from the vehicle to the BE-4 engines, not to the performance of the engines. The finding moves the investigation away from Blue Origin’s engines and toward the first stage’s propellant feed system. It is preliminary, not a root cause. Vulcan remains grounded and ULA has not set a return-to-flight date.
What ULA says happened
ULA’s first detailed account, published on 17 November, said early indications pointed to “low booster performance” just after liftoff. In the company’s vocabulary, the booster is the core first stage, the part powered by two BE-4 engines. The shortfall meant stage separation happened when the vehicle was “low in altitude and below minimal velocity to reach orbit.”
A day later, ULA refined that account. “While work remains to determine cause, preliminary indications are that the low booster thrust was the result of inadequate fuel supply from the vehicle to the BE-4 engines, not from the performance of the engines,” the company said.
The Centaur upper stage “appears to have started nominally,” ULA said, and its two RL10 engines operated as planned. It performed well but could not recover from the performance loss. The six Northrop Grumman solid rocket boosters also performed nominally, ULA said, following nozzle improvements made for this flight.
ULA has not identified a root cause. It has not said which part of the fuel system failed to deliver enough propellant, or why. The company said it is very early in the process and wants to avoid speculation until the investigation is complete. The Federal Aviation Administration will oversee the ULA-led investigation, be involved at every step and approve the final report, including any corrective actions.
What “inadequate fuel supply” means
A rocket engine is only as useful as the plumbing that feeds it. Vulcan’s BE-4 engines burn liquid methane and liquid oxygen. Both propellants have to arrive at the engine inlet at the right pressure, at the right flow rate and in the right mixture. If the vehicle cannot maintain that flow, the engines produce less thrust even when every component inside them is working correctly.
ULA’s wording draws a line between the engine and the vehicle. The BE-4 is Blue Origin’s hardware. The tanks, pressurization system, feed lines, valves and the rest of the first-stage propellant system belong to the Vulcan vehicle that ULA builds and integrates. Saying the problem was “fuel supply from the vehicle” points investigators at that boundary, not at the engine’s internal machinery.
That does not identify the culprit. A drop in flow could come from a stuck or misconfigured valve, a pressurization problem in a tank, a restriction or leak in a feed line, contamination, or a sensor and control issue that pushed the system outside its intended operating range. ULA has not said which of those possibilities, if any, it is pursuing. The finding is a direction, not a diagnosis.
The distinction matters for more than blame. If the fault sits in the vehicle-side feed system, the fix belongs to ULA and the BE-4’s design is not implicated. If the fault sits at the interface between engine and vehicle, both companies will be involved. And if the finding changes as more data arrives, the implications for Blue Origin’s New Glenn rocket change with it.
Why the upper stage could not save the mission
A two-stage rocket divides the work. The first stage provides most of the thrust needed to climb out of the atmosphere and build speed. The upper stage takes over near the edge of space, where its engines are optimized for vacuum, and finishes the job of reaching orbital velocity.
That handoff has to happen within a narrow range of altitude and speed. If the first stage separates too low and too slow, the upper stage has to make up a deficit it was never sized for. Centaur can adjust its trajectory and burn longer, but it cannot manufacture the missing energy.
Spaceflight Now’s William Harwood reported that the first stage appeared to burn longer than ULA’s pre-launch timeline and that separation came later than expected. Telemetry-driven animation showed Centaur making unusual orientation adjustments early in its burn. ULA’s broadcast commentators did not mention a problem at the time.
Mark Peller, ULA’s president and CEO, put the same point more plainly at a Washington Times event on 17 November. The first stage had “a significant shortfall in booster first stage performance,” he said, leaving Centaur in “such a low-energy state” that it “couldn’t make up for that performance deficit.”
The gap between when the failure was detected and when it began is the most important detail in ULA’s update. Losing telemetry at 10 minutes is an effect. The cause was already in motion.
The BE-4 question, and a reprieve for Blue Origin
Vulcan’s first stage uses two BE-4 engines built by Blue Origin. They burn methane and liquid oxygen, a combination Blue Origin chose partly to avoid the soot and handling problems of older kerosene engines. A variant of the same engine powers Blue Origin’s New Glenn rocket, seven at a time.
Before 18 November, the engine was an obvious line of inquiry. New Glenn exploded during a static-fire test on 28 May at Cape Canaveral. Blue Origin CEO Dave Limp later said the failure originated at the main oxygen valve on one of the rocket’s seven BE-4s, according to Spaceflight Now. Blue Origin modified the valves and delivered retrofit kits for the BE-4s it had supplied to ULA. Peller said before Sunday’s launch that the engines on this Vulcan had received the retrofit.
ULA’s preliminary finding now points away from the engine and away from that retrofit. Blue Origin said it worked alongside ULA to review the flight data and reached the same early conclusion. “Together, we identified that the low booster thrust was the result of inadequate fuel supply from the vehicle to the BE-4 engines, not from the performance of the engines,” the company said. “ULA is a valued partner, and we look forward to Vulcan’s return to flight.”
The reprieve extends to New Glenn. Ars Technica and Aviation Week both reported that Blue Origin could return New Glenn to flight as soon as January 2027, with its damaged pad rebuilt. That is a reported target, not an announced date. It depends on Blue Origin’s own investigation and readiness, not just on Vulcan.
None of this clears the BE-4 permanently. Investigators will still examine engine data, and the interface between engine and vehicle is part of the failure. But the early evidence has shifted the centre of gravity of the investigation.
The solid boosters were not the problem this time
For most of Vulcan’s short career, the solid rocket boosters were the worry. On the rocket’s second flight in October 2024 and its fourth in February 2026, the nozzle of one strap-on booster failed about 30 seconds after liftoff. Both missions still reached orbit, but the second failure led the U.S. Space Force to pause national-security launches on Vulcan while ULA worked the issue.
ULA said it had identified three credible causes and made design and process improvements to the nozzle. Peller told Aviation Week that the company had taken positive steps to mitigate all three. All six boosters on Sunday performed nominally, according to ULA, and Peller described them as “six clean SRBs.”
If that finding holds, it clears the component that had already caused the most trouble and narrows the investigation to the core stage. It also means the months ULA spent on the nozzle problem were not wasted, even though the rocket still failed for a different reason.
What Amazon loses
Amazon Leo, formerly Project Kuiper, is Amazon’s plan to sell broadband from low Earth orbit. The constellation is designed to compete with SpaceX’s Starlink, which has more than 11,000 satellites and millions of customers. Amazon has booked more than 100 launches across several rockets to catch up.
Sunday’s flight was the first of 38 Vulcan missions Amazon has purchased. Vulcan is slated to carry more than 1,500 of Amazon’s first-generation satellites, more than any other rocket in the program. That makes the rocket’s grounding a direct threat to the deployment schedule, not a distant supplier problem.
Before Sunday, Amazon had launched 396 production satellites across 14 successful missions since April 2025, plus two prototypes. Its first-generation constellation is licensed for 3,232 satellites. Sunday’s launch would have pushed the production total to 436. Amazon says it has built more than 1,000 satellites and has hundreds of flight-ready units waiting for rockets.
Amazon still says it plans to begin initial commercial service before the end of 2026. On 18 November, it told Fierce Network that it has enough satellites for continuous coverage at some latitudes and will begin offering service “in the coming weeks” in select areas. The company had already missed a mid-2026 target. The Federal Communications Commission granted a conditional waiver after Amazon missed the 30 July 2026 interim milestone of 1,616 satellites in service; the final deadline is 30 July 2029 for the full first-generation constellation.
The next Leo launch is scheduled for no earlier than 23 November, when an Ariane 6 is set to carry 36 satellites from French Guiana. Amazon also has 24 launches reserved on New Glenn, 24 on Ariane 6 and 13 on Falcon 9, plus additional Atlas V missions.
Amazon projected confidence in ULA after the failure. “There is risk with any launch, and that’s especially true for inaugural missions like LV-01,” the company said. “ULA has been a trusted partner for Amazon, and we believe they’ll learn from this mission and deliver on their long-term plans for Vulcan.”
A countdown with its own anomalies
The launch had already slipped twice. A Friday attempt was scrubbed after a problem with a gaseous nitrogen plant that serves multiple launch providers. A Saturday attempt was delayed by network problems between Vulcan and ground systems.
On Sunday, controllers held at T-minus 20 seconds because of an unexplained technical issue with the rocket. ULA resolved the problem and launched at the end of the window. It is not known whether the hold contributed to the failure. ULA has not connected the two, and investigators will examine the countdown data alongside the flight data.
The May “second failure” that wasn’t
One detail that has circulated in coverage of Sunday’s loss needs correcting. The May New Glenn explosion is sometimes described as a second Amazon Leo launch failure that destroyed 48 satellites. The satellites were not on board.
The rocket exploded during a static-fire test on its pad at Launch Complex 36. No one was injured. The blast badly damaged the pad, the only one configured for New Glenn, and delayed a launch that would have carried the 48 satellites. Amazon lost no satellites that night.
The distinction matters because it separates two different kinds of failure: a launch that fails with payload aboard, and a ground test that fails before launch. Both are serious. Only one destroyed satellites.
What happens next
Vulcan is grounded until the investigation produces answers. ULA has not given a return-to-flight date, and the FAA’s oversight means the company will need an approved final report before it can resume. The history of launch investigations suggests months, not weeks, especially when the fix touches propellant plumbing rather than a single replaceable part.
The stakes go beyond one rocket. Vulcan is ULA’s replacement for the Atlas V, which is due to retire after six more flights, and it is central to the company’s national-security business. The Space Force had already paused national-security launches on Vulcan while the booster-nozzle issue was resolved. It was waiting to see how Vulcan performed on this commercial mission before certifying the rocket again. The 18 November finding does not shorten that wait.
For Amazon, the cushion is diversification. It has contracts with ULA, Blue Origin, Arianespace and SpaceX, so one grounded rocket does not stop the program. But every delay tightens a schedule that was already behind. New Glenn is still recovering from the May explosion, and Reuters reported that SpaceX is no longer accepting new Falcon 9 bookings, with the rocket eventually due to give way to Starship. The launch market is tight, and Amazon needs capacity now.
The bottom line
Sunday’s failure is a reminder that Vulcan is still a young rocket. ULA’s reputation was built on Atlas and Delta, vehicles with long flight histories and deep engineering knowledge behind them. Vulcan has five flights, one failure and a long investigation ahead.
The 18 November update narrows the question. The engines probably did their job; the vehicle probably did not deliver the propellant they needed. That is a better place for investigators to start than a generic “booster performance” problem, but it is still a preliminary finding. The next update will matter more than this one.
What we added: We updated this article on 21 November with ULA’s 18 November preliminary finding, explained the difference between engine performance and vehicle-side fuel supply, added Blue Origin’s statement and the New Glenn return-to-flight reporting, and revised the Amazon service and launch-schedule details. The original reporting on the launch, the Centaur stage, the solid boosters and Amazon’s constellation is unchanged.
Last reviewed: 21 November 2026.
Sources: United Launch Alliance, Vulcan Leo 1 updates (15–18 November 2026); Spaceflight Now, “Vulcan failure investigation focuses on inadequate fuel supply” (18 November 2026); Ars Technica, “Vulcan failure caused by inadequate fuel supply to engines, ULA says” (18 November 2026); Aviation Week, “Vulcan Probe Shifts To Fuel Supply, A Reprieve For BE-4” (18 November 2026); Reuters (15 November 2026); Spaceflight Now launch report (15 November 2026); GeekWire (16 November 2026); Space.com (15 November 2026); The Verge (15 November 2026); GeekWire on the New Glenn explosion (2026); Fierce Network, “Amazon Leo on track for year-end service launch” (18 November 2026); Amazon Leo mission updates (15 November 2026); Light Reading on Amazon Leo’s satellite count (12 October 2026).
Have a tip? Email [email protected] or use the contact form and choose Editorial. We do not promise anonymity.
Read next: Starship Flight 15: historic catch, lost booster, Vulcan failure; The FAA nationwide ground stop is over; GTA 6 hacker Arion Kurtaj pleads guilty days before launch. Browse all Big Tech coverage.
