What a Rocket Launch Really Costs, and Whether Reuse Is Cheaper

Someone tells you a Falcon 9 launch costs $74 million, someone else says SpaceX only spends $15 million, and a news story says one NASA moon rocket launch costs more than $4 billion. All three can be true, because they measure different things. The $74 million is a price a customer sees on a published list, the $15 million is a reported estimate of what a reflown rocket costs SpaceX to fly, and the $4 billion is an inspector general’s projection that spreads a whole program’s production and operations over its first few flights. Reuse really does lower the cost of flying, but it does not automatically lower the price a customer pays, and the gap between the two is the interesting part.

This note sorts the figures out, says exactly what each one is and when it was published, and points out where I could not verify something. If you want the story of how boosters came back, that is in our note on Falcon 9 reuse; this one is about money. A caution on sources: I could not open every primary page directly, so several 2026 figures rest on reputable reports of those pages, and I say so where it matters.

Price, cost and marginal cost are not the same thing

Price is what a customer is quoted. Cost is what it takes the provider to deliver, and that has several layers. The marginal cost of a launch is the extra money spent to fly one more rocket when the factory, the pad, the staff and the development bill are already paid for. The average cost spreads everything, including past development, across all flights. Quoting a marginal cost as if it were a price, or comparing one company’s price with another’s average cost, is the most common error in launch-cost arguments.

A fair example: an airline’s fuel and crew bill for one more flight is a tiny part of what a ticket costs, because the aircraft, hangars and booking systems have to be paid for too. Launch is the same, just with fewer flights and much bigger numbers.

Falcon 9: the published price

SpaceX publishes a standard price on its Capabilities and Services material. According to Spaceflight Now’s reading of it in February 2026, the standard plan through 2026 is $74 million, up from $70 million in 2025 and $67 million in 2022. That price is for a Falcon 9 carrying up to 5.5 metric tons to geostationary transfer orbit, and SpaceX’s material notes it includes range services, standard integration and liability insurance. I could not load the SpaceX document myself, so treat $74 million as a well-reported figure, not one I read at the source.

Three limits apply. It is a list price for a standard commercial mission, and government customers with special requirements often pay more. Rides on shared missions can cost far less per customer. And SpaceX can and does negotiate, so the final invoice on any particular flight is not public.

Falcon 9: what it costs SpaceX

SpaceX does not publish its internal costs. The figure usually quoted is about $15 million for flying a booster that has already flown, which Spaceflight Now reported in February 2026 as coming from Ars Technica reporting. A similar marginal-cost claim of roughly $15 million has been attributed to Elon Musk in earlier years. Either way it is a company-side or journalist estimate, never an audited number, and it covers a reflown first stage and the expendable second stage, but not the money already spent on developing the vehicle.

Take it as a rough scale, not a fact. If both figures are close to right, the gap between $74 million and $15 million is large, and it is the margin that pays for development, the next generation of rockets and Starship. It does not mean customers are overcharged. A company that has spent billions on a program needs more than its marginal cost to survive.

What reuse actually saves

A Falcon 9 reuses its first stage and fairing, not its second stage, which is thrown away every flight. The booster is the biggest and most expensive piece, so reusing it removes the largest single block of cost from each launch. In 2017, SpaceX president Gwynne Shotwell said that refurbishing a flown booster cost ‘substantially less than half’ of a new one, according to reports at the time. That is a company statement from the early days, and the fleet has flown many more times since. SpaceX has not published refurbishment figures for today’s boosters.

It also costs something. Landing means the booster carries extra propellant, legs and grid fins, and gives up some payload. Third-party sources put Falcon 9’s low Earth orbit capacity at about 22,800 kilograms when the whole rocket is thrown away, which SpaceX itself quotes, and around 17,500 kilograms when the booster lands on a drone ship, though I found that second figure only in secondary sources. Reuse also needed over a billion dollars in development, a figure Musk gave in 2017, so the savings had to pay back an investment first.

The result so far is easy to state. Reuse made SpaceX much cheaper to run and much more profitable at the same price, and it has let the company fly a hundred or more Falcon 9 missions in a year. Whether the savings reach customers is a matter of competition. Prices have gone up slightly in recent years, not down, which tells you that a rocket with little competition is priced to the market, not to its cost.

Cost per kilogram, and why to be careful

Dollars per kilogram is the number that gets quoted most and explained least. It depends on which price, which payload, and which orbit. As an illustration only, my own arithmetic: $74 million divided by 17,500 kilograms is about $4,200 per kilogram, while the reported $15 million internal cost over the same mass is about $860 per kilogram. Divide the price by the expendable maximum of 22,800 kilograms instead and you get about $3,200. None of these is a figure SpaceX publishes. The price is for a geostationary mission that carries less than a low-orbit one, so mixing the two orbits flatters the answer, and a rocket seldom flies at its limit.

Use per-kilogram figures to compare rockets of the same class, quoted the same way, and be skeptical when a story compares a price per kilogram for one rocket with a cost per kilogram for another.

The other side: SLS and Artemis

NASA’s Space Launch System is thrown away after each flight, and its numbers come from a different kind of source, the NASA Office of Inspector General. In a report in November 2021 the OIG estimated that each of the first four Artemis launches would cost $4.1 billion, covering production of the SLS, the Orion capsule and the ground systems needed to fly them. In testimony in January 2024 it updated that to at least $4.2 billion per launch for the first four missions, and said that excluded about $42 billion of earlier formulation and development spending.

Three notes on this. It is an inspector general’s projection that divides program spending across a few flights, not a price or a marginal cost, and a flight rate that stays low makes each launch look more expensive. NASA disagreed with the OIG’s recommendation to keep per-mission cost accounting, so there is no matching official NASA figure to compare with. And I found no 2026 report that restates a per-launch number, although a June 2026 OIG memo reported on cancelled Artemis system contracts and I did not read it in full, so check the OIG site before quoting anything newer. Our note on Artemis delays covers the schedule side.

A tempting comparison is $4.2 billion against $74 million. It is not fair, because the rockets carry different loads. SLS is a super-heavy rocket sized for sending Orion to the Moon, and Falcon 9 is a medium-lift rocket. But the contrast still tells you that a government program built for capability and jobs, with little reuse, sits in a different cost world than a commercial one built to fly often.

Other price points, with caveats

Several other numbers circulate, and most come from secondary sources, not company price lists. Rocket Lab’s small Electron is commonly given at about $7.5 million, and one analyst put its average near $8.4 million. ULA’s Vulcan has been quoted around $110 million for a base version, and Blue Origin does not publish a New Glenn price, so figures between $60 and $100 million are estimates. I would not rely on any of these without checking the company’s own statement for the date you care about, because launch prices change and most are not published.

Starship: targets, not facts

Starship is designed to reuse both stages, and that is where the claims get large. The targets commonly cited are an internal flight cost under $10 million at maturity and a cost under $100 per kilogram to low orbit, with Musk once suggesting even lower. These are goals, they refer to SpaceX’s own cost and not to a customer price, and they depend on very high flight rates and many reflights per vehicle, which have not yet been shown. A June 2026 analysis argued that the claims are about internal cost and remain unproven. One report of a first commercial price of $90 million, tied to a company filing, is something I could not confirm.

Starship’s progress is real. Our notes on Flight 15 and the ship catch cover what was demonstrated. But a catch is not a price list, and until a stainless-steel ship flies the same hull many times in succession, the per-kilogram figures are forecasts.

So is reuse actually cheaper?

For the company that does it well, yes, and clearly: a reflown Falcon 9 costs a fraction of building a new one, and that is the reason it has so many launches. For a customer the answer is less direct. They get reliability and availability that the market never had, and a price that has stayed flat or risen. If you want one test for any figure you meet, ask three things: is it a price, a cost or a projection, which year is it from, and what payload and orbit does it assume. If the answer to any is missing, the number is not ready to quote.

Key sources: Spaceflight Now, February 2026 and NASA OIG report IG-22-003.