technology
SpaceX Starship Flight 12: Largest Rocket Ever Launches Successfully
SpaceX's 407-foot Starship Version 3 lifted off from Starbase Texas on May 22, 2026 — Booster 19 splashed down in the Gulf, Ship 39 reignited its engines mid-coast, and NASA's Artemis lunar lander timeline just got a tangible push forward.
May 22, 2026407 ft tallShip 39 + Booster 19$15B+ program
407 ft
Vehicle height
22
Mock + live payloads
~1 hr
Total flight time
$3B
R&D in 2025 alone
Key Takeaways
- SpaceX Starship Flight 12 launched May 22, 2026 at 5:30 PM CDT from Starbase, marking the debut flight of Version 3 — the tallest, most powerful rocket ever built at 407 feet (124 meters).
- Both stages performed planned controlled splashdowns: Super Heavy Booster 19 in the Gulf of Mexico and Ship 39 in the Indian Ocean — a clean recovery profile after May 21's hydraulic-pin scrub.
- Mission tested in-space engine reignition ~39 minutes after launch — the exact maneuver NASA needs for the Human Landing System (HLS) to descend to the lunar surface on Artemis III.
- Two of the 22 deployed payloads carried imaging sensors trained on Ship 39's heat shield tiles — early data on the thermal protection system upgrade that has historically been Starship's weakest link.
- SpaceX has burned more than $15 billion on Starship development across the program's lifetime, including roughly $3 billion in R&D during 2025 — a price tag now hinged on Version 3 maturing fast enough for the late-decade Artemis cadence.
What Happened on Flight 12
Starship Version 3 cleared the tower at Starbase on the second attempt of the week. The first try on May 21 was scrubbed inside the final hour after a hydraulic pin on the ground support hardware failed safe checks — a 24-hour stand-down that SpaceX teams converted into a clean recycle.
At T+0 Booster 19's 33 Raptor engines lit on the new V3 configuration. Stage separation came on schedule and the booster executed a flip-and-burn back to a controlled water splashdown in the Gulf of Mexico, ending its mission as a paid full-duration test rather than a catch attempt. Ship 39 continued to orbital-energy trajectory, deployed its 20 Starlink mass simulators plus 2 instrumented test satellites, and re-ignited its center Raptors roughly 39 minutes into flight — the deep-space restart NASA wants to see before Artemis III. About one hour after liftoff Ship 39 met its planned fiery splashdown in the Indian Ocean.
Elon Musk called the result 'an epic' launch, with a minor engine anomaly noted during booster return that did not affect mission objectives. Crucially nothing on the vehicle had to be ad-libbed: every major event hit within seconds of its predicted timeline.
▸ Two consecutive booster-and-ship recoveries on V3 hardware means SpaceX can now collect repeatable telemetry instead of one-off post-mortem data.
Starship Version 3 vs Version 2 — What Changed
| Spec | Version 2 (Flights 9-11) | Version 3 (Flight 12) | |
|---|---|---|---|
| Stack height | ~397 ft (~121 m) | 407 ft (124 m) | |
| Raptor engines (booster) | 33 (Raptor 2) | 33 (uprated Raptor 3 mix) | |
| Heat shield system | Tile + ablative backup | Re-tiled with instrumented imaging payloads | |
| In-space relight test | Partial — single Raptor | Full sequence at T+39 minutes | |
| Payload deployed | 0 (test only) | 20 Starlink simulators + 2 instrumented | |
| Booster recovery profile | Tower catch attempt | Controlled Gulf splashdown |
Flight 12 Timeline — From Scrub to Splashdown
May 21, 2026
T-0:43 — Hydraulic pin scrub
Inside the final hour the launch director called a scrub after a hydraulic pin on the orbital launch mount failed an integrity check. SpaceX recycled to T-24 hours rather than push through.
▸ Hardware-side scrub means the vehicle never even loaded its critical countdown — propellant load stayed nominal for a second attempt.
May 22, 5:30 PM CDT
T-0 — Liftoff
Booster 19's 33 Raptor engines light up. The stack clears the tower and rolls onto its azimuth toward the eastern range.
▸ First V3 ascent — every second past T+30 is brand-new flight data SpaceX has never had on this configuration.
T+2:30 (approx)
Stage separation + booster boostback
Hot-staging separates Ship 39 from Booster 19. The booster flips, relights, and begins its return profile toward a Gulf of Mexico splashdown target.
▸ Confirms the upgraded hot-staging ring survived V3 thrust loads — a known stress point on V2 flights.
T+~8 minutes
Booster 19 controlled splashdown
Super Heavy executes its landing burn over the Gulf and touches the water in a controlled vertical posture — no catch attempt, but a clean instrumented recovery target.
▸ Minor engine anomaly noted during this burn — exactly the kind of edge data SpaceX needs before authorizing the next tower catch.
T+~39 minutes
Ship 39 in-space engine reignition
Ship 39's center Raptor relights on schedule — the deep-space restart that NASA's HLS variant must perform to drop out of lunar orbit and land Artemis astronauts.
▸ This single 20-second burn is arguably the most consequential event of the mission for Artemis III scheduling.
T+~1 hour
Ship 39 planned splashdown
Ship 39 re-enters and meets its target splashdown in the Indian Ocean. The 22 deployed payloads complete their separate orbital test profiles.
▸ The Indian Ocean target gives SpaceX a no-debris-risk recovery zone while still demonstrating full re-entry survivability.
Why NASA's Moon Plan Was Watching Every Second
Artemis III — NASA's first crewed lunar landing since Apollo 17 — depends on a Starship variant known as the Human Landing System (HLS). The HLS has to do something no Starship has done yet: launch, dock with an Orion crew capsule in lunar orbit, transfer astronauts, and then perform a precision powered descent to the surface using main-engine relights in the deep-space vacuum.
That last part is the high-stakes maneuver. Vehicle systems that work for an 8-minute boost to orbit do not automatically work after hours of cold-soak. Propellant boil-off, hydraulic actuators, igniters, header tanks — every one has to survive and then perform on command. Flight 12's in-space relight at T+39 minutes is the first time SpaceX has demonstrated the full procedure on V3 hardware.
The schedule pressure is real. NASA has publicly held Artemis III for late 2027. Internally, every quarter HLS slips, the program pushes commercial lunar service contracts, suit-development milestones and Gateway assembly with it.
▸ ZestLab estimate — at $15B+ sunk and roughly $3B/year burn, every 3-month HLS slip costs the program ~$750M before opportunity costs.
The $15 Billion Question — Where the Money Went
Starbase build-out
Two orbital launch mounts, the Mechazilla tower, propellant farms, and a tile factory next door — Starbase is now a vertically-integrated launch city.
Raptor engine program
Raptor 2 to Raptor 3 took an internal rebuild — fewer parts, higher thrust, full-flow staged combustion at production scale never attempted before.
Iterate-and-explode test cadence
Eleven prior flights, multiple anomalies, plus pad and tile rework after each. The expensive part of move-fast-and-break-things is the rebuilding.
An epic launch — minor engine anomaly during booster return, otherwise on the money. Version 3 is the ship that takes us to the Moon and Mars.
Heads up for sky-watchers in Vietnam
Future Starship launches with Starlink payloads can be tracked over Vietnam via the Heavens-Above app — V3's bigger payload bay means the Starlink trains will be brighter and longer-lasting in the night sky.
What Comes Next — Flight 13 and the Real Tests
With Version 3 having survived its debut, SpaceX has publicly hinted that Flight 13 will push three new envelopes: a full Starlink mass deployment with real V3 satellites, a tower catch attempt for Booster 20, and the first orbital refueling rendezvous demonstration. Each one cleanly maps to an Artemis III prerequisite.
The refueling demo is the next high-stakes step. HLS needs to be tanked up in low Earth orbit before its translunar burn — that requires a tanker Starship to launch separately, rendezvous, dock, and transfer cryogenic propellant in microgravity. SpaceX has tested propellant transfers between internal tanks; an external ship-to-ship transfer has not flown yet.
For Vietnam-based space enthusiasts, the practical fallout is faster Starlink coverage upgrades. The bigger V3 payload bay means each launch can loft 60+ next-gen V3 satellites, which are designed to handle higher per-cell throughput. Vietnamese Starlink resellers and rural broadband projects feel that capacity arrive within the next 12-18 months.
▸ ZestLab take — the gating risk is no longer 'can the rocket fly,' it is 'can SpaceX dock and refuel.' That's a software-and-mechanism problem, not a propulsion problem.