How has the cost of reaching orbit changed since the 1960s, and what drove the changes?
The short answer. for forty years the price of a kilogram to low Earth orbit barely moved, at roughly $10,000–20,000 in today's money for most rockets and far more for the Space Shuttle. Then, between 2010 and 2020, it fell by about an order of magnitude, to around $2,700 per kilogram for a Falcon 9 and $1,400–1,500 for a Falcon Heavy at list price. The fall came from one company's decisions to build its own engines, standardise on one vehicle family and recover the first stage, not from a general improvement in technology. The next fall, if it comes, depends on a fully reusable vehicle that flew its first operational payload in September 2026.
Why price per kilogram is the right measure, and its limits
Price per kilogram divides the price of a launch by the mass a rocket can lift to a reference orbit, usually low Earth orbit. It is the simplest way to compare rockets of different sizes, and it is the number that determines whether a satellite constellation, a space station or a factory in orbit can pay for itself. It has three limits that this study keeps in view.
First, rockets rarely fly full. A satellite operator pays for the whole rocket whether its payload uses 40% or 100% of the capacity, so the price per kilogram actually paid is often double the headline figure. Rideshare missions exist to fill that gap.
Second, orbit matters. A kilogram to geostationary transfer orbit costs two to three times a kilogram to low orbit on the same rocket, because the rocket carries less mass to the higher energy orbit.
Third, list price is not cost. For most of history the two were close, because launch was sold by governments or by companies with thin margins. Since 2015 they have diverged: SpaceX's internal cost of a reused Falcon 9 launch is far below the price customers pay (Chapter 7), and that gap is now the most important fact in launch economics.
The long plateau, 1960–2010
| Era | Representative vehicles | Price per kg to LEO (approx., 2018 dollars) | Why |
|---|---|---|---|
| 1960s–1970s | Saturn V, Titan, Atlas, Delta, Soyuz, Proton | ~$5,000–20,000 | Expendable rockets built in small numbers for governments |
| 1981–2011 | Space Shuttle | ~$54,500 (nominal) to ~$61,700 (2018 dollars) | Partially reusable in principle, but each refurbishment was costly and slow; flew ~4–5 times a year |
| 1970–2000 average | All vehicles | ~$18,500 | NASA cost analysis |
| 1990s–2000s | Ariane 4 and 5, Proton, Atlas V, Delta IV, Long March | ~$5,000–15,000 | Commercial competition for geostationary satellites; Russian and Chinese prices undercut Western ones |
Source. NASA cost study (2018 dollars) for the averages and the Shuttle; the ranges are Setcoin summaries of published estimates.
Three things kept prices high. Rockets were thrown away after one flight, so every launch paid for a new vehicle. Production runs were short, a few to a dozen vehicles a year per type, so fixed costs were spread over few launches. And the main buyers were governments, which valued reliability and industrial policy more than price.
The Space Shuttle is the cautionary tale. It was sold as a reusable, low-cost truck to orbit. In practice the orbiters needed months of inspection and refurbishment between flights, the programme flew about 135 missions in 30 years, and its cost per kilogram was among the highest of any vehicle. The lesson that reuse only pays if refurbishment is cheap and fast shaped every reusable programme since.
The fall, 2010–2020
| Vehicle | First flight | Price per kg to LEO (approx.) | What changed |
|---|---|---|---|
| Falcon 9 (expendable era) | 2010 | ~$2,700 | In-house engines, one engine type on both stages, commercial development under NASA's cargo programme |
| Falcon 9 (reused boosters) | 2017 onward | List price stable; internal cost falling | Booster landing and reflight; fairing recovery |
| Falcon Heavy | 2018 | ~$1,400–1,500 | Three Falcon cores; high capacity for a modest price premium |
| SpaceX rideshare (Transporter) | 2021 | $5,000/kg at launch of the programme in 2019; ~$7,000/kg in 2026 | Selling spare capacity in small lots |
Sources. NASA cost study; CSIS Aerospace Security data as cited in market commentary; SpaceX rideshare price list (February 2026).
The striking feature of the table is that the price per kilogram paid by customers fell once, between 2010 and about 2018, and has since been roughly flat or rising in nominal terms. Rideshare prices rose from $5,000 to $7,000 per kilogram between 2019 and 2026. A standard Falcon 9 launch has been sold at about $60–70 million throughout. SpaceX's own filing implies that customer launch revenue averaged about $58–60 million per customer launch in 2023–2025 (Setcoin calculation, Chapter 7). What fell after 2018 was SpaceX's cost, not its price.
Why the price stopped falling
A competitive market would pass cost savings to customers. Launch is not a competitive market. In 2025 SpaceX conducted about half of all orbital launch attempts in the world and, by most estimates, most of the mass to orbit; for many Western commercial and government customers, it was the only rocket available with a strong record and an open manifest. A seller with that position prices against the next-best alternative, which for most of the period was an Ariane 5 or 6, an Atlas V or Vulcan, or a Russian or Chinese launch closed to Western customers. Those alternatives cost more, so Falcon 9 could hold its price and keep the savings.
SpaceX used the savings in two ways. It launched its own Starlink constellation at internal cost, which became the largest satellite system ever built (more than 11,000 active satellites in October 2026, CelesTrak). And it funded Starship. Both uses have shaped the industry more than a price cut would have.
The next step: full reuse
The economics of a fully reusable vehicle differ from those of a partly reusable one. On Falcon 9 the second stage, about a fifth of the vehicle's cost on most estimates, is thrown away every flight, along with the propellant and the work of integration. Starship aims to recover both stages and refly them with little refurbishment, like an aircraft. In that case the cost of a flight approaches the cost of propellant, operations and amortised hardware.
| Starship cost figure | Source and status |
|---|---|
| ~$2M operating cost for a mature, fully reusable mission | Company projection from 2019; not a price and not demonstrated |
| ~$10M per launch | Statements attributed to the company's chief executive; no primary source found |
| $90M for a dedicated launch | Reported from a customer's annual report (Voyager); a commercial price, not a cost |
| 100 tonnes to LEO (V3), double in later generations | SpaceX S-1, 2026 |
At 100 tonnes, a $90 million price would be $900 per kilogram; a $10 million cost would be $100 per kilogram (Setcoin calculation). The distance between those two figures is the central uncertainty of this study, and Chapter 10 builds the cost stack that sits between them.
What history suggests
Three patterns from sixty years of launch prices frame the rest of this study:
- Cost falls come from architecture decisions (engines, reuse, production rate), and they arrive in steps, not trends. There was one large step in 1960–2010 (the move to commercial expendable rockets competing for satellites) and one in 2010–2020 (Falcon 9). The next is Starship or nothing.
- Reuse pays only when refurbishment is cheap and fast. The Shuttle failed that test; Falcon 9's boosters passed it (Chapter 9).
- Price follows competition, not cost. Customers benefit from a cost fall only when a second supplier can match it. Who that second supplier will be is the subject of Chapters 14 to 23.
What would change this view
A competitor flying a reusable medium or heavy rocket at a high cadence (New Glenn, Neutron, Zhuque-3 or a successor to Ariane 6) would restore price competition and pass some of the cost fall to customers. Conversely, Starship reaching routine full reuse while no competitor follows would widen the gap between cost and price further.