Starship V3 Debuts, with Mixed Results
For the first time in 2026, SpaceX’s Starship Super Heavy launch vehicle roared into the skies of South Texas on the afternoon of May 22nd. Powerful shockwaves, clearly visible in launch photos, battered onlookers a few miles away, providing tangible evidence of the rocket’s upgraded engines. But despite a suite of improvements touted for V3, the system’s third generation, Friday’s flight test ended with mixed results, advancing the program in some ways and stumbling backwards in others. And in the seven months since Starship’s last flight, the pressure for SpaceX to deliver has only increased as NASA’s Artemis program and a looming IPO raise the stakes.
SpaceX’s previous Starship design, called V2, was introduced in January of 2025, and by most accounts was a failure. At the time, SpaceX had caught its first Super Heavy booster, and Starship had survived its first few reentries, albeit with major damage. The new design was meant to iterate on the spacecraft’s flaps and heat shield and deploy simulated Starlink satellites. However, the first three V2 Starships failed spectacularly; only the last two vehicles evaded problems long enough to reenter the atmosphere successfully. All the while, several boosters were caught and one of them reused, allowing the Super Heavy side of the program to continue pushing its envelope.
Flight 12 debuted new vehicle designs, the new Raptor 3 engine, and a redesigned launch pad. The mission began with a successful liftoff and first stage ascent, but problems began at stage separation. When Booster 19 attempted to fly back towards the coast, several of its engines failed to restart, and an explosion disabled many others. The booster aborted its boostback burn and coasted far downrange of its intended landing site. Another failure during its landing burn doomed the rocket to crash into the Gulf of Mexico. The booster was meant to make a controlled water landing before a future catch attempt, but that milestone will now likely be pushed to a later flight.
Meanwhile, Ship 39’s ascent was marred by the loss of one of its 3 vacuum-rated Raptor engines. But while similar failures caused Starships to lose control on Flights 7 and 8, Ship 39 held stable, completing its ascent burn later than planned to compensate for the issue. Although the planned engine relight attempt was cancelled, Ship 39 successfully deployed 20 Starlink simulators and two modified Starlink spacecraft, which used onboard cameras to photograph heat shield tiles on the ship’s leeward side. This marks the first time Starship has deployed any kind of functioning spacecraft, though not yet in a stable orbit.
Finally, Ship 39 reentered the atmosphere more successfully than any vehicle before it, demonstrating sweeping turns and dramatic maneuvers to stress-test its flaps and heat shield. Unlike most previous Starships, Ship 39 appeared fully intact during its spectacular terminal descent and landing burn, and its heat shield weathered the test with minimal obvious damage. As an entry vehicle, Starship V3 has demonstrated its ability to navigate from space to a precision landing largely unscathed, one of the final steps towards becoming a fully reusable launch system.
Where does this leave the program? At a glance, some boxes on Starship’s to-do list have now been unchecked, stepping backwards from the successes of Flight 11. The performance of Raptor 3 is particularly suspect; engine failures plagued both stages of Flight 12, destroying the booster and perhaps coming close to dooming the ship. Since Raptor 3 is a radical redesign, failures are perhaps not surprising, but they are disappointing to encounter three years into the program. And if the design proves fundamentally flawed in some way, V3 may continue to suffer failures and near-misses much like V2, prolonging Starship’s development even further.
On the other hand, Ship 39’s performance otherwise seems to reflect a much more robust design. Several V2 ships lost control after catastrophic propulsion system failures, whereas the new vehicle handled a contained engine shutdown in stride. Its heat shield especially shows maturation from prior flights; it’s not hard to imagine a Starship gliding in for a catch attempt at the launch site sometime soon. The ship also demonstrated better attitude control thanks to a new set of thrusters, which will be crucial for docking and in-space refueling in the future.
All told, Flight 12’s outcome failed to reach SpaceX’s stated goals for the mission, and that itself is a bad outcome for the program. NASA’s new plan for Artemis III expects a Starship that can reach a stable orbit, rendezvous and dock with Orion, and potentially support astronauts onboard the ship itself for the first time—all about one year from now. Starship is then expected to land on the Moon twice in the next two years, first uncrewed and then with human lives at stake. This, of course, requires dozens of operational flights in total, let alone the test flights SpaceX deems necessary to reach these capabilities.
SpaceX now also plans to become publicly traded as soon as next month. The company and its CEO, Elon Musk, have made grand promises about orbital datacenters, AI satellites launched from the Moon, and most recently crewed tourism flights to the Moon and Mars, and SpaceX is now preparing to place its funding in the hands of many new stakeholders who will be expecting positive results. SpaceX has a loyal fan culture which is not easily disappointed, but the Starship program is a long way from delivering on most of these promises, to say nothing of the stability of the AI bandwagon upon which SpaceX’s core products now rest.
SpaceX is betting its future on meeting dizzyingly high expectations very soon, and a misstep at any time is unwelcome. Starship V3 is intended, in SpaceX’s words, to “unlock the vehicle’s core functions.” If this flight is only a temporary hiccup, Starship could truly be just a few months from deploying operational satellites into Earth orbit. But if Flight 13 offers a repeat showing of these issues, the program may well be in jeopardy.
