Starship Reuse Is a Manufacturing Problem, Not a Rocket One

Every argument about reusable rockets eventually turns into a picture of a booster landing upright, and that picture is the least part that matters of the economics. Reuse only pays if the vehicle flies again soon enough to spread its construction cost across many flights, which makes the whole business a question about inspection labour, engine life and how fast you can refill a tank farm. Flight rate is the number that decides whether a launch market exists. Payload is the number that wins arguments on the internet.

The Vehicle Stops Being the Expense Once It Flies Enough

The arithmetic is unglamorous. Cost per flight is the vehicle's construction cost divided by the number of flights it survives, plus refurbishment, plus propellant, plus range and operations. Everything interesting sits in that divisor. A vehicle that costs a great deal and flies once is an expendable rocket with extra steps and a heavier structure. The same vehicle flying twenty times turns its own manufacturing cost into a small line item, and the cost per flight collapses toward whatever the turnaround actually costs in labour and facilities.

This is why the factory matters more than the launch. Stainless rings welded in a shipyard and engines built on a production line are a different industrial proposition from hand-building a bespoke machine per flight. Starship's real break with the past is that it is designed to be manufactured at rate, in the open, with welding fixtures instead of clean rooms. Reuse is the second half of that idea, not the whole of it.

And reuse carries a permanent tax. A stage that comes back needs structure to survive reentry, propellant held back for landing, and a flight profile that costs payload. Every kilogram of heat shield and every kilogram of landing propellant is a kilogram that does not go to orbit, so a reusable vehicle is always less efficient per flight than an expendable one. It wins only by flying enough times to make efficiency irrelevant.

Inspection Is the Flight Rate

After a flight, a vehicle is not a vehicle, it is a pile of things that need to be looked at. The heat shield has to be walked tile by tile, because a single missing tile can put a hot spot on a stainless tank wall. The engines have to be borescoped, and the turbopumps on a full-flow staged combustion engine are the most highly stressed machinery on the vehicle — two preburners, high chamber pressure, and seals fresh off a thermal cycle nobody designed a consumer product around. Tanks get ultrasonic inspection for cracks at welds. Plumbing gets checked for leaks that only appear when the lines are cold.

None of that is easy to automate, because the failure modes are visual and tactile and the standards are written by people who have watched hardware fail. Commercial aviation solved this with maintenance intervals derived from decades of fleet data. Rocket reuse has a small sample, so intervals get set conservatively, which means more inspection per flight than the physics requires. Flight rate and inspection depth are coupled: fly more, learn faster, inspect less, fly more cheaply. Fly rarely and you never escape the conservative setting.

The Falcon 9 programme is the settled proof that the model works — boosters have been reflown repeatedly, and the turnaround between flights has been measured in weeks. It is also the honest warning: reuse did not make launches cheap by magic, it moved the marginal cost out of airframe construction and into labour and range operations — which is exactly why cadence is the lever.

The Pad and the Propellant Farm Are the Ceiling

A launch site is a chemical plant that happens to have a rocket on it. Methalox means liquid methane, which boils near 111 kelvin at atmospheric pressure, and liquid oxygen at 90. Both are loaded subcooled to increase density, which means chilling them below their boiling points, holding them there, and loading late enough that they do not warm up on the pad. That is a cold-chain problem with a schedule attached, and the schedule is the launch window.

The farm feeding the pad has to be refilled between flights, and refilling means deliveries, chilldown and quality checks. Add the water deluge, quick-disconnect hardware that has to mate reliably with a flown vehicle, and an airspace and maritime exclusion volume that has ed and released every time. A pad turnaround measured in hours requires all of that to be resized for hours, and most of it was sized for weeks.

This is where the honest answer to "how many launches a year" comes from: not the factory, not the fleet, but the number of pads times the number of cycles each pad can complete, bounded by propellant logistics and range availability. Additional pads buy cadence, and they buy it at the price of a second chemical plant.

Payload Is the Wrong Metric

Payload capacity is what a vehicle can do once. Cadence is what a business can sell. A customer buying a launch is buying a date, and a date is only worth something if it holds. Constellations are the clearest case: a megaconstellation needs a steady replacement rate for satellites that fail or deorbit, and that demand is a function of launches per month, not tonnes per launch. A vehicle that can lift a hundred and fifty tonnes twice a year is worth less to that customer than one that lifts fifty tonnes twenty times.

The same logic explains why high cadence is a prerequisite for markets that do not exist yet. Nobody plans a business around a capability that might be available in a three-month window. Propellant depots, orbital manufacturing, and any service that assumes frequent access all price in the schedule risk, and schedule risk is what a low flight rate is made of.

What I Think

I think the industry keeps measuring the wrong variable because the impressive one is easy to photograph. A landing is a miracle of control theory and it tells you almost nothing about whether the economics work; a shipyard that can weld a tank section a week tells you nearly everything. If Starship reaches a high flight rate, the reason will be boring: enough vehicles, enough pads, and inspection procedures that stopped being conservative because there was finally data to relax them.

My scepticism is reserved for the timeline, not the direction. The hard part was never catching a booster. It is running a chemical plant, an inspection shop and an airspace reservation system on a schedule that makes a customer's business plan possible — and that is a manufacturing and logistics problem, wearing a rocket costume.