The sequence was supposed to go like this. Two minutes and thirty-two seconds after liftoff from Starbase, stage separation. Super Heavy Booster 19, the first Version 3 booster ever built, flips its orientation above the Gulf of America and starts relighting engines for the boostback burn. The burn fires. The booster arcs back toward Texas or descends toward the planned splashdown zone. The mission continues.
Instead, twelve of thirteen center engines ignited at boostback initiation. Then something went wrong with the outer ring. The burn started and stopped early. Booster 19 didn't arc anywhere. It continued on its ballistic trajectory and hit the Gulf at a speed that left nothing to recover.
The FAA declared it a mishap. Starship is grounded.
Chapter One: What Raptor 3 Was Supposed to Be
Flight 12 was the debut of the Raptor 3 engine. The Raptor has gone through three major design generations since it first flew on Starship test vehicles in 2023. Each generation pushed more thrust out of a simpler design with fewer components. Raptor 3 was supposed to represent the version SpaceX could actually manufacture at scale. The engine that would power a fleet of Starships flying dozens of times a year rather than a handful of test vehicles flying occasionally.
Raptor 3 operates at chamber pressures that no other production rocket engine reaches. It burns liquid methane and liquid oxygen in a full-flow staged combustion cycle, which means nearly all the propellant passes through the turbomachinery before combustion. The efficiency that architecture produces is why SpaceX chose it. Higher specific impulse means more payload for a given propellant load, and payload capacity is the number that determines whether Starship's economics actually work.
None of that mattered on May 22 when the engines didn't light.
SpaceX's post-mission assessment pointed to the Raptor 3 engines as the center of the failure investigation. Both the booster's boostback burn and the upper stage's vacuum engine count were affected on the same flight. Two separate Raptor 3 configurations failed on the same mission. That's not a coincidence you explain away.
Chapter Two: What the FAA Actually Does
A mishap determination under federal commercial spaceflight regulations means SpaceX can't fly again until the investigation is complete and the FAA approves a return to flight. The process involves SpaceX leading its own investigation with FAA oversight at every step. The agency reviews the findings, the corrective actions, and the reasoning before it signs off.
In practice, these investigations have taken anywhere from a few weeks to several months. The FAA cleared Starship for Flight 8 relatively quickly after the Flight 7 anomaly in January 2025, issuing an early return-to-flight determination before the full investigation closed. That's only possible when the agency concludes the anomaly didn't jeopardize public safety. For Flight 12, the FAA activated a debris response area during the booster's uncontrolled descent, putting five aircraft in holding patterns and delaying six departures. That's a harder fact to set aside quickly.
SpaceX has been through this before. The pattern since Starship's first flight has been anomaly, investigation, corrective action, next flight. What's different now is the timing. SpaceX filed its S-1 prospectus with the SEC two days before Flight 12, targeting a Nasdaq listing. The grounding arrives at a moment when the company needs to demonstrate it can execute on Starship's promise.
Chapter Three: What the Booster Felt
Stand close enough to a Raptor engine during testing and you don't hear it so much as absorb it. The combustion pressure is high enough that the sound waves carry a physical weight at short range. A pressure against the chest. A vibration in the jaw. People who've stood near Raptor static fires describe the experience as something their body registered before their ears did.
Booster 19 carried 33 of those engines. At full thrust during ascent, the combined output is measured in meganewtons. A force that pushes 5,000 metric tons of vehicle off a launch mount and through the atmosphere at accelerations the human body can't tolerate for long. The staggered shutdown sequence that prepares the booster for stage separation is a carefully choreographed reduction from 33 engines burning to 3, timed to manage the structural loads on both vehicles as they separate.
After separation, Booster 19 was supposed to relight 13 of those engines for the boostback burn. Twelve lit. Then the outer ring failed to follow. Whatever happened inside those engines happened fast enough that the flight computer couldn't compensate. The burn ended before it began in any meaningful sense.
The booster that had just pushed the most powerful rocket ever flown to separation altitude then fell, without thrust, into the Gulf.
Chapter Four: What Version 3 Was Supposed to Change
Starship Version 3 is physically larger than its predecessors. The propellant tanks carry more. The thrust is greater. SpaceX built a new launch pad at Starbase to accommodate the V3 configuration and reduce turnaround time. The plan was to use V3 to begin deploying Starlink Version 3 satellites in the second half of 2026. Satellites that require Starship's payload volume because they're too large for Falcon 9.
That deployment schedule is now uncertain. The Starlink constellation currently operates more than ten thousand satellites in low Earth orbit, and the V3 generation represents the next step in its capability. Starship is the only vehicle that can launch them at the required size and cost. If the investigation and corrective action push the Flight 13 window into late summer, the Starlink V3 deployment timeline moves with it.
SpaceX has FAA approval for up to 25 Starship launches per year from Starbase. It has completed one flight in 2026 and the year is nearly half over. The gap between Flight 11 in October 2025 and Flight 12 was more than seven months. If Flight 13 doesn't fly until July or August, the cadence SpaceX needs to make Starship's economics work remains theoretical.
Chapter Five: The Investigation That Matters
The engineering question at the center of the Flight 12 investigation is specific. Why did Raptor 3 engines fail to reignite reliably after stage separation, and why did the failure affect both the booster configuration and the upper stage vacuum configuration on the same flight?
SpaceX's engineers have a significant advantage in answering this question. The company has flown Raptor engines hundreds of times across static fire tests and eleven previous Starship flights. The telemetry data from Flight 12 is dense. They know what the engine was doing at every millisecond of the flight, which sensors recorded anomalous readings, and which components deviated from expected performance.
The history of Raptor development suggests they'll find the answer. It also suggests the fix will require physical changes to hardware. That takes time. The July to August window that analysts are citing for Flight 13 may be optimistic depending on what the data shows.
Starship has always been a program that tests faster than it fails. The question Flight 12 raises is whether Raptor 3 needs more testing before it's ready to do what the mission requires of it.
The answer is coming. The data already exists. Somewhere in SpaceX's telemetry archive is the exact moment things went wrong.
Someone is reading it right now.