SpaceX Starship V3 Flight 12 Faces May 21 Launch From Starbase

SpaceX has Booster 19 stacked under Ship 39 at Pad 2, the fueling rehearsal in the bag, and a launch window that opens Thursday, May 21 at 6:30 p.m. EDT. Flight 12 will be the first Starship liftoff of 2026 and the debut of the V3 vehicle, a 407-foot stack running on 33 Raptor 3 engines, designed to carry NASA’s Artemis 4 crew to the Moon in 2028.

Seven months have passed since the last Starship flew. The launch window opens against a heavy backdrop: a contractor died at the Starbase assembly site on May 15, the U.S. Occupational Safety and Health Administration (OSHA, the federal workplace safety regulator) opened an investigation, and SpaceX’s planned June 12 initial public offering at a $1.75 trillion to $2 trillion valuation is riding partly on a clean splashdown off Western Australia.

The Seven-Month Rebuild Behind Flight 12

Flight 11 closed Block 2 on October 13, 2025. Booster 15-2 and Ship 38 separated cleanly, the upper stage relit its Raptor in space, and both halves splashed down on target. The program then went silent for the longest inter-flight gap since test launches began in April 2023.

What rolls out from the Megabay this week is a fresh architecture from the engines up, not a refined predecessor. The booster carries 33 Raptor 3 motors in place of the Raptor 2 generation, each producing roughly 280 tons-force at sea level against 230 for the older engine. The ship moves from six engines to nine, three sea-level Raptors and six vacuum-optimized. SpaceX flew five Starship tests across 2025; Flight 12 is the only one currently on the books for the first half of 2026.

The launch date has already slipped twice this week. The original May 19 attempt moved to May 20 without public explanation, then to May 21 after the Gigabay fatality. Two engine generations, a new pad assignment, a taller stack, and refueling hardware are arriving on the same flight rather than incrementally, and the scope of that seven-month reset is the bet itself.

What Is New on Ship 39 and Booster 19

Side by side, the new vehicle and the prior generation differ on more than the marketing reads. Engine count, engine power, tank volume, stack height, fin layout, and on-orbit hardware all changed. The headline numbers below are SpaceX’s own, drawn from the company’s official Starship vehicle overview and from Raptor 3 specifications released over the past year.

Engines and Thrust

Raptor 3 is the headline. SpaceX rates each sea-level Raptor 3 at roughly 280 tons-force, about 22 percent above Raptor 2’s 230 tons-force and nearly 50 percent above the original Raptor. The full 33-engine first stage now produces a combined 9,240 tons of liftoff thrust, the highest of any rocket flown. The upper stage rises from six engines to nine, which roughly matches the booster increment and gives the ship more margin for the planned in-space relight and landing burn.

Tanks, Height, and Mass

The vehicle is taller. Stacked, the new ship reaches 407 feet (124 meters), several feet above the prior generation. The Super Heavy booster now holds 4,050 tons of liquid methane and liquid oxygen, up from 3,650 tons on the earlier design. The ship’s own tankage rises from 1,500 to 1,600 tons. Gross liftoff mass sits near 5,300 tons of vehicle and propellant combined. The booster also drops a grid fin (three on this generation, four on the last) and uses the remaining fins as catch points for the Mechazilla tower arms during eventual return-to-pad attempts.

The Refueling-First Design

V3 is the first Starship designed to dock. Ship 39 carries four orbital docking ports and a redesigned quick-disconnect coupling for ship-to-ship propellant transfer in low Earth orbit. None of that hardware matters on Flight 12, which never reaches orbit. It matters on the propellant-transfer demonstration SpaceX owes NASA before Artemis 4, a mission whose architecture only closes if a tanker Starship can fill a depot Starship in space. SpaceX signed a $53.2 million NASA contract for that demo in 2020.

The Artemis Math That Forces a Refueling Demo

NASA’s Human Landing System program (HLS, the moon-lander track inside Artemis) requires SpaceX to fly a Starship to lunar orbit, demonstrate an uncrewed lunar landing and return, then launch the actual crewed Artemis 4 lander. The architecture only closes if SpaceX can refuel a Starship in low Earth orbit, because the lander has to carry its own propellant load to the Moon. NASA’s working estimate sits at roughly ten tanker launches per HLS mission.

• ~10 tanker launches per single Moon landing, per NASA’s working figure
• $53.2 million NASA contract signed with SpaceX in 2020 for the propellant-transfer demo
• ~5,300 tons gross liftoff mass for the new vehicle, the heaviest rocket ever to fly
• Early 2028 NASA target for Artemis 4, contingent on the demo working

Ten tanker flights per Moon landing is a lift cadence roughly six times what SpaceX has achieved with Starship across all twelve test flights to date. Each tanker has to launch, dock, transfer cryogenic methane and oxygen, undock, and either return or deorbit. The propellant-transfer demo Ship 39 is built for is the bridge between a test program that splashes ships into the ocean and an operational program that flies them in pairs.

The Flight Profile and Starlink Test

The Thursday window opens at 6:30 p.m. EDT (2230 GMT, 5:30 p.m. local time at Starbase) and runs for 90 minutes. SpaceX livestream coverage begins about 45 minutes before liftoff. The mission profile is a suborbital arc partway around the world, with no second-stage burn to circularize. Both stages will splash down rather than return to the pad.

The full sequence is short for a flight that has taken seven months to schedule.

1. T+2 minutes: hot-stage separation, then Super Heavy ignites its boostback burn
2. T+7 minutes: Super Heavy splashes down in the Gulf of Mexico off the Starbase coast
3. Mid-coast: Ship 39 deploys 22 Starlink simulators sized to next-generation operational satellites
4. Late coast: two modified probes scan the ship’s heat shield and beam imagery back to Earth
5. T+65 minutes: Ship 39 reenters and splashes down in the Indian Ocean off Western Australia

Neither stage will be caught. Pad 2’s Mechazilla arms have not yet been used to grab a booster; Pad 1 caught Super Heavy on several earlier flights but is offline while crews retrofit it for the new generation. The two probe-mounted cameras are the part of the mission that most directly serves the next milestone, because a returnable Starship has to certify its own heat shield in flight before NASA buys the architecture for crew.

The Starbase Death and an OSHA File Already Open

A contractor died at Starbase’s Gigabay assembly building around 4 a.m. local time on May 15. The U.S. Occupational Safety and Health Administration opened an investigation the same day. The worker, employed by an outside firm, fell during pre-launch operations on Flight 12 hardware.

SpaceX has not publicly tied the launch delay to the fatality, but the original May 19 launch attempt slipped to May 20 within hours of the incident. The slip itself surfaced first through Starbase road-closure notices, and only later through company statements. Public visibility into the cause remains thin.

Starbase has a longer safety file than the corporate narrative usually reads. Federal injury data showed the site logging worker-injury rates roughly six times the average for comparable space-vehicle manufacturers, and nearly three times the broader aerospace industry rate, in 2024.

OSHA issued seven serious safety violations against SpaceX in January after a crane collapse at Starbase in June 2024, with proposed penalties totaling $115,850. The Federal Aviation Administration (FAA, the U.S. agency that licenses commercial launches) environmental review covering the new vehicle is still active, and any change to its flight or recovery profile can trigger fresh review.

The IPO Calendar Sitting Behind the Pad

SpaceX’s planned June 12 Nasdaq listing under the ticker SPCX is set at a valuation between $1.75 trillion to $2 trillion, with the offering targeting more than $75 billion in fresh capital. The road show leans on Starship as the asset that lowers launch cost per kilogram into orbit, supports the Starlink revenue line, and underwrites the company’s lunar and Martian programs. A clean Flight 12 lets the bookrunners go into pricing with a recent successful test of the launch vehicle in the deck.

A failure, especially one that destroys Ship 39 or the booster on ascent, would hand underwriters a price-discovery problem three weeks before the listing date. SpaceX has not adjusted its IPO calendar in response to the launch slip, according to public filings as of this week. That sequence implies management treats the suborbital arc as a high-confidence test rather than a coin flip, and the company is willing to put a trillion-dollar listing behind that confidence. Two pieces of evidence support it: Flight 11 in October closed the prior generation without an explosion, and the May 20 wet dress rehearsal completed without abort.

The Suborbital Cap Still Hanging Over the Program

Across twelve test flights, no Starship has reached orbit. Every flight has ended in a planned ocean splashdown. The program has demonstrated stage separation, in-space relight, booster catch on Pad 1, and a clean payload-bay door cycle, but none of those are the orbital insertion the propellant-transfer demo requires.

V3 is the architecture meant to break that ceiling. Flight 12 still does not break it. The suborbital trajectory was set before launch, has not been amended, and the manifest’s next item, the ship-to-ship transfer demo, sits behind a clean Thursday.

If Ship 39 splashes down on target off Western Australia about 65 minutes after liftoff, the propellant-transfer demo moves from paper to next on the launch manifest. If it does not, the schedule that ends with crew at the lunar south pole on Artemis IV in 2028 will be redrawn at NASA headquarters, not at Starbase.