What a kilogram costs

Almost every interesting argument in the space industry — about asteroid mining, lunar bases, orbital factories, megaconstellations, settlements on Mars — turns out, when you press hard enough on it, to be an argument about a single number. How much does it cost to put one kilogram of mass into low Earth orbit? Get that number low enough and entire markets become possible. Keep it where it is and they remain expensive thought experiments. The number has done one thing for fifty years and is doing something else now, and the next decade of the industry runs on which way it goes.

1969: $11,400 per kilogram on Saturn V

Saturn V — the largest, most powerful operational rocket ever flown until Starship — could put about 140 tonnes into LEO and cost, depending on how you do the accounting and what you adjust for inflation, somewhere between $1.2 billion and $1.6 billion per launch in 2024 dollars. Splitting the difference and dividing through gives roughly $11,400 per kilogram to LEO. That number is honestly impressive. It was the lowest cost-to-orbit in the world for thirty years.

The reason it was so cheap is that Saturn V was, by the standards of any subsequent crewed-rated heavy launcher, brutally simple. Three stages, all expendable, all kerosene-LOX or hydrogen-LOX, no exotic materials, no reusable elements, no avionics complexity beyond what it took to fly the Apollo stack. The economy was built on industrial-scale production at Boeing, North American, McDonnell Douglas, and the smaller suppliers. Once the stage manufacturing pipelines were running, marginal cost per vehicle was reasonable. Then the program ended and the pipelines closed.

1981-2011: $24,000-$60,000 per kilogram on the Space Shuttle

Then came the Shuttle. The cost story here is famously complicated and famously bad. NASA's official figures vary widely depending on whether you include development costs, infrastructure costs, the Solid Rocket Booster recovery operations, or just the marginal cost of one mission. The most commonly cited figure is around $1.5 billion per launch in 2024 dollars, with a typical 27-tonne payload to LEO — which works out to around $55,000 per kilogram. Other accountings put it as low as $24,000 if you exclude development sunk cost. None of them put it below the Saturn V number.

The brutal lesson of the Shuttle program is that "reusable" is not a synonym for "cheap". Refurbishing a Shuttle orbiter took roughly four months and cost more, in inflation-adjusted dollars, than building a new Saturn V S-IC stage. Reusing the SRB casings was nearly a wash on cost. The thermal protection system inspections alone took thousands of person-hours per flight. The Shuttle was an extraordinary piece of engineering and a deeply uneconomic launch vehicle, and it kept the world's cost-per-kilogram number stuck at expendable-launcher levels for another thirty years.

2010-2025: $2,500 per kilogram on Falcon 9

Falcon 9 broke the line. Initially it didn't — the early flights, expendable Block 1, were priced at around $4,000 per kilogram and were already cheaper than any Western alternative. The breakthrough was reuse, but a different kind of reuse than the Shuttle had attempted: just the first stage, refurbished in days rather than months, with engines designed from the start to be used many times. The 2015 first booster landing was the inflection point. Today, after about a decade of operations, SpaceX has reflown individual Falcon 9 boosters more than twenty times. The published list price of a Falcon 9 dedicated mission is $69.75 million; the company's marginal cost on a reused booster is, by reasonable estimate, well under half of that.

The accepted figure for Falcon 9's effective cost-per-kilogram to LEO is roughly $2,500. The Transporter rideshare program — where a single Falcon 9 carries dozens of small payloads simultaneously — gets to roughly $5,500 per kilogram once you factor in the integration and dispenser overhead, which sounds higher but is in fact lower per actual user because you don't need to buy a whole rocket. Either way: Falcon 9 is approximately a quarter the price per kilogram of Saturn V was, and a tenth of what the Shuttle cost. It is, by some distance, the cheapest reasonable way to put a kilogram in LEO that has ever existed.

This is not because SpaceX did one clever thing. It's because they did about a hundred reasonably clever things in sequence. The choice to standardise on a kerosene-LOX engine cluster, the gridfin-and-cold-gas booster recovery system, the propellant-densification trick that gave the early Block 5 boosters their margin, the relentless iteration on stage manufacturing in Hawthorne, the decision to vertically integrate avionics and software in-house, and — most importantly — flying frequently enough that all the lessons compound. By 2024 SpaceX was flying every two and a half days on average. Each flight made the next flight cheaper, until the curve flattened around the marginal cost of the upper stage and the propellant.

2026 onwards: Starship's promise of $200 per kilogram

If Starship works at the cadence and reuse SpaceX claims, the cost per kilogram drops by another order of magnitude. The company has publicly suggested $200 per kilogram to LEO at full operational maturity. That number is genuinely controversial, and it's worth being honest about how unsupported it is by current operations.

Starship has not yet completed a fully successful orbital flight with payload deployment. It has not yet refurbished and reflown either the Super Heavy booster or the Ship upper stage in commercial service (the booster catch in October 2024 was a major step but the same booster has not yet been re-flown to orbit). The propellant-mass fraction necessary to hit the published payload numbers depends on a clean Ship return-from-orbit, which has been demonstrated only in part. The on-orbit refuelling architecture that Artemis depends on has not been demonstrated at all.

That said: even with conservative assumptions about Starship reuse, even allowing for years of refurbishment overheads, even taking SpaceX's published $200 figure as half-fantasy, the realistic Starship cost-per-kilogram once it reaches operational cadence is probably between $500 and $1,500. That's still a roughly 5x improvement on Falcon 9 and a 50x improvement on the Shuttle. It is also genuinely transformational for what kinds of missions can credibly be funded.

What changes when the number drops

Most space-economy thought experiments that fail today fail because they assume a launch cost that doesn't yet exist. Asteroid mining penciled out at $200 per kilogram and made no sense at $11,000. Orbital data centres need cheap launch to be cost-competitive with terrestrial fibre. Crewed Mars missions become budgetable rather than aspirational. The hundred-thousand-satellite mega-constellations become technically possible (whether they make commercial sense is a different question — see "the mega-constellation thirst trap" — but at least the launch math closes).

What's worth noticing is that for the first half-century of spaceflight, the cost-per-kilogram number barely moved. The 1969 Saturn V figure was, in real terms, slightly better than the 2010 Atlas V. Five decades of nominally improving technology produced essentially zero progress on the one metric that drives everything else. Reuse changed that. The next change — when Starship matures, or when one of the Chinese or European reusable competitors catches up — will be the second time in human history that the unit economics of spaceflight have meaningfully shifted.

And then we'll see which of the thought experiments were waiting on it.

Cost figures derived from published list prices, NASA accounting reports, and contemporaneous analyst estimates. Inflation adjustments to 2024 dollars. See also the launch-market consolidation analysis and the launch tracker.