SpaceX's Starship V3 Flies: A Bumpy, Mostly Successful Debut

At 6:35 p.m. Central time on Friday, the South Texas coastline shook as the tallest rocket ever built lifted off from a brand-new launch pad. Thirty-three sea-level Raptor 3 engines pushed a 124-meter stainless-steel column upward through the humid Boca Chica sky, the shockwave rolling over the dunes of Starbase and out across the Gulf. Onlookers had waited days for this moment — Thursday's attempt had been scrubbed at the last minute when a hydraulic pin on a tower arm refused to retract — but on the second try, Starship V3 finally flew.

What followed over the next 75 minutes was, as SpaceX itself described it, a flight test in the most literal sense: a debut that delivered the first hard data on the company's most heavily redesigned vehicle yet, and one that ended with the upper stage executing a controlled splashdown in the Indian Ocean while its booster came apart in pieces over the Gulf of Mexico. Elon Musk called it "an epic first Starship V3 launch and landing," telling his team they had "scored a goal for humanity." Ars Technica's Eric Berger, who has tracked the program closely, called the flight "really, really important" — not because everything worked, but because enough did.

What's actually new in Starship V3

V3 is not an iteration. It is, by SpaceX's own count, a vehicle reworked across thousands of components, and Friday's flight was the first chance to see whether those changes hold up in the punishing environment of a real launch.

The headline numbers:

• Height: 124.4 meters (408 feet) fully stacked, roughly four feet taller than V2.
• Thrust: About 18 million pounds at liftoff, with Raptor 3 engines rated at roughly 280 tonnes-force each at sea level — close to 9 percent more than V2's Raptor 2.
• Payload: A target of more than 100 metric tons to low Earth orbit in fully reusable configuration, which would make it the highest-capacity rocket in history by a wide margin.
• Aerodynamics: Super Heavy's grid fins reduced from four to three, with each fin roughly 50 percent larger and stronger.
• Pad: First flight from the newly completed Pad 2 at Starbase, which uses electromechanical actuators in place of hydraulics on the launch mount.

Friday's vehicles — Booster 19 and Ship 39 — also carried 22 payloads: 20 dummy V3-class Starlink satellite simulators and two modified Starlink spacecraft outfitted with imaging sensors, two of which SpaceX engineers nicknamed the "Dodger Dogs." Some of those sensors were trained back on the ship itself, transmitting live imagery of the V3 heat shield as it heated through reentry.

Why the Raptor 3 upgrades matter

The Raptor 3 is the engine SpaceX has spent years trying to build: lighter, simpler, and — in theory — far more reusable than its predecessor. By internalizing what used to be a tangle of external plumbing and adding regenerative cooling directly into the engine structure, SpaceX has eliminated the need for the dedicated heat shield and fire suppression hardware that Raptor 2 required. The result is an engine that weighs roughly a tonne less per unit and exposes far fewer parts to the post-flight inspection process.

That matters because Starship's entire economic case depends on engines that can be flown, inspected quickly, and flown again. Without rapid reuse of the Raptors, the dream of "airline-style" Starship operations is dead on arrival.

Friday gave that case a mixed first verdict. The sea-level Raptors performed well enough to push the stack through max-Q and stage separation. But during Super Heavy's boost-back burn over the Gulf, multiple Raptor 3 engines failed to relight or shut down prematurely, and Booster 19 broke apart before reaching its target landing zone. On the upper stage, one of Ship 39's three vacuum-optimized Raptors (RVacs) shut down less than a minute after ignition, scrubbing a planned in-space restart later in the flight.

What worked, what didn't

For a maiden voyage of a near-total redesign, the win column was unusually long.

What worked:

• Clean liftoff from Pad 2, including the debut of the electromechanical launch mount.
• Stage separation and a powered ascent that reached second-stage engine cutoff on five of six Raptors.
• Deployment of all 22 payload simulators on a trajectory SpaceX spokesperson Dan Huot described as "within bounds" of the pre-flight analysis.
• A controlled, banking reentry over the Indian Ocean simulating a future return-to-tower approach.
• A soft splashdown of Ship 39 at roughly 7:36 p.m. Central time — followed, as expected on these test flights, by a post-touchdown explosion as residual propellant ignited.

What didn't:

• Booster 19's loss during boost-back due to Raptor 3 relight failures.
• The early RVac shutdown that aborted an in-space engine restart test.
• An initial trajectory that, in Huot's words, was "not exactly nominal orbital insertion" but still within the planned envelope.

Notably, SpaceX did not attempt a Mechazilla tower catch of Super Heavy on this flight — a deliberate decision given how much of the booster's hardware was flying for the first time.

The Artemis pressure

If the technical story of Friday was about Raptors and reentry, the political story was about the Moon. NASA Administrator Jared Isaacman watched the flight from Starbase and afterward declared the mission "one step closer to the Moon, one step closer to Mars."

That phrasing is not incidental. Starship is the foundation of NASA's Human Landing System contract — the vehicle expected to ferry American astronauts from lunar orbit to the surface of the Moon. Under the agency's revised Artemis plan, announced earlier this month, Artemis III will now stay in Earth orbit to rehearse docking between Orion, Starship HLS, and Blue Origin's Blue Moon lander. The first crewed lunar landing has slipped to Artemis IV, currently targeted for 2028, a deadline driven in part by Isaacman's public commitment to put Americans back on the surface before Chinese taikonauts arrive.

Between Friday's flight and any Moon landing sits a problem no rocket has solved at scale: cryogenic propellant transfer in orbit. Starship HLS will need roughly ten tanker flights to fill a propellant depot before a single lunar lander has the fuel it needs to leave Earth orbit. Friday demonstrated that V3 can reach space and return controllably. It did not, and was never intended to, demonstrate the refueling architecture the lunar mission rests on.

What's next at Starbase

SpaceX President Gwynne Shotwell, speaking after the flight, framed the result in characteristically optimistic terms: "Our collective next flying amongst the stars has become so much closer." The company says it expects to complete roughly 10 more Ship 39-class vehicles and around five additional Super Heavy boosters by the end of 2026 — a manufacturing tempo that, if sustained, would let SpaceX rapidly iterate on the issues that surfaced Friday.

The Federal Aviation Administration's commercial space office last year cleared Starbase for up to 25 launches per year, removing what had been one of the program's tightest external constraints. With Pad 2 now operational and Pad 1 still active, SpaceX has, for the first time, the regulatory and infrastructure headroom to chase a flight cadence measured in weeks rather than months.

Musk, who two days before the launch announced SpaceX's intent to take the company public in what is expected to be one of the largest IPOs in history, has reiterated that an uncrewed Starship flight to Mars remains on the books for the late-2026 transfer window. That target has slipped before, and the Raptor 3 reliability questions raised by Friday's booster loss will not make it easier to hit.

But for one evening in South Texas, the bigger story was simpler. The largest rocket ever built flew. It dropped its payload where it was supposed to. It came home, however imperfectly, and it gave SpaceX engineers their first real telemetry on a vehicle that will, if Friday is any guide, define the next decade of human spaceflight.