Starship Block 2: Three Flights, Three Failures — What Engineering Got Wrong

SpaceX's Starship ran 11 flight tests as of October 2025. The score: 6 successes, 5 failures. Block 1 had its own rough start — FT-1 and FT-2 both failed. Block 2 was supposed to be better. It wasn't — at least not at first.

FT-7, FT-8, and FT-9 were all Block 2 failures. That's a 3-for-3 failure streak on the newest version of the world's heaviest rocket. Understanding what went wrong matters beyond Starship specifically — these failures represent an unusually transparent view into what breaks when you scale up by a factor nobody has tried before.

## What Changed with Block 2

Block 2 wasn't a minor revision. SpaceX redesigned structural elements, updated avionics, and changed the propellant handling systems. Ship 33, the first Block 2 upper stage, was also heavier (85 tonnes empty vs ~100 tonnes for Block 1, actually lighter due to structural optimizations) and carried more propellant — 1,500 tonnes vs 1,200.

The booster also got a new catch mechanism design for mechazilla. FT-5 had successfully caught the Super Heavy booster by its grid fins — a genuinely spectacular achievement in November 2024. Block 2 was supposed to repeat that routinely.

## FT-7: The Propellant Leak That Ended It

Flight test 7 launched January 16, 2025. Super Heavy — Booster 14 — performed well and was caught successfully at Starbase. The upper stage is where it went wrong.

Ship 33 suffered an internal propellant leak in the engine bay. The leak created a fire, the fire caused loss of control, and telemetry was cut. Range safety terminated the vehicle 11 minutes after launch over the Atlantic. The debris scatter from a termination that high up generated concern — and a lot of cleanup of reports from aircraft and ships.

A propellant leak in a cryogenic system of that scale is partly a materials problem and partly a pressurization problem. LOX at –183°C and liquid methane at –161°C in adjacent plumbing, vibrating through a launch profile, will find any joint or seal that isn't quite right.

## FT-8 and FT-9: The Pattern Continues

FT-8 and FT-9 continued Block 2's poor upper stage performance. SpaceX was methodical about diagnosing and adjusting between flights — they don't fly on fixed schedules, they fly when they think they've fixed what broke — but Block 2's Ship design had systemic issues that took multiple iterations to isolate.

The Super Heavy booster performance in these flights was largely solid. The booster architecture matured faster than the ship. That asymmetry is interesting: the first stage, which operates in the thickest part of the atmosphere and faces the most violent aerodynamic loads, is more reliable than the second stage operating in near-vacuum.

## The Engineering Context

What's rarely appreciated in media coverage is how different Starship's scale is from everything that came before it. Saturn V burned RP-1/LOX. Falcon 9 burns RP-1/LOX. Starship burns liquid methane. Methane as a rocket propellant at scale is less understood — SpaceX is running what is effectively the world's largest methane propulsion research program in real time.

The combustion characteristics, the material compatibility issues, the boil-off behavior — all of these differ from RP-1, and the body of operational data for methane at Starship's propellant volumes was basically zero before 2023.

FT-7's internal propellant leak is the kind of failure that happens when you're learning a new propellant system at scale. It's not a design philosophy failure. It's an empirical iteration cycle that SpaceX has always been willing to pay in hardware.

Block 3, with a 142-meter height and 200-tonne LEO payload capacity, is a significant step beyond Block 2. Whether those three Block 2 failures ultimately produced the lessons that make Block 3 reliable is the real question — and it won't be answered until Block 3 flies.

Starship Block 2: Three Flights, Three Failures — What Engineering Got Wrong

// COMMENTS

ON THIS PAGE