SpaceX's Starship SN9 Nails the Ascent, Still Can't Stick the Landing
SN9 flew to 10 km and executed its belly flop reentry, but crashed on landing just like SN8 did in December.
SpaceX flew another Starship prototype today, and it ended the way SN8 did back in December: a fireball on the landing pad.
SN9 launched from Boca Chica, climbing to a planned altitude of about 10 km (6.2 miles). From there it went through the maneuver that makes Starship testing so weird to watch — the “belly flop.” The rocket shuts down its engines mid-flight and falls horizontally, using its body as a giant air brake while four flaps steer it through the atmosphere like a skydiver. It’s a wild way to bring a rocket down, and by all appearances SN9 handled that part fine. The flaps did their job and the vehicle stayed controlled on the way in.
Where it fell apart was the final flip. To land, Starship has to reignite its engines, rotate itself from horizontal back to vertical, and settle down tail-first onto the pad. SN9 didn’t manage the reorientation in time. It came down still tilted, hit the pad hard, and exploded — visually almost a replay of what happened to SN8 six weeks ago.
Same failure, different vehicle
That repetition is the headline here. Two prototypes in a row have nailed the hard, unconventional part of the flight — the belly flop descent — and then failed at the part that sounds simpler on paper: flipping upright and landing softly. Whatever is causing the issue, it doesn’t seem to have been fully solved between SN8 and SN9.
To be fair, SpaceX’s whole approach with Starship has never been about avoiding failures. It’s about flying fast, blowing things up, and mining the wreckage for data. And there’s real data here: engine relight performance during the landing flip, flap responsiveness during reentry, and now a second data point on exactly how the vertical reorientation goes wrong. That’s not nothing, even if the footage looks like a disaster.
Elon Musk has talked about needing all three Raptor engines to relight properly and about fine-tuning the header tank pressure that feeds fuel during the flip — small margins that matter a lot in the last few seconds of flight. Whether today’s failure traces back to the same root cause as SN8 or something new is going to take engineers combing through telemetry to figure out.
SN10 is reportedly already being prepped in Boca Chica, and given SpaceX’s cadence over the past few months, another attempt in the coming weeks wouldn’t be a surprise. The ascent and reentry profile now looks fairly repeatable. The landing is the piece still eluding them, and until a Starship prototype rotates upright and touches down gently, that’s the number one thing to watch for next.