A kilogram to which orbit, on whose schedule, with what else aboard?
Low Earth orbit is not one destination. A 200-kilometre equatorial parking orbit, a 500-kilometre sun-synchronous orbit, and a precise rendezvous with a space station impose different energy, launch-site, timing, and assurance requirements.
| Metric | Numerator | Denominator | What it is good for |
|---|---|---|---|
| Advertised capacity | Public base price | Maximum payload to a reference orbit | Comparing vehicle envelopes |
| Booked slot | Contracted rideshare price | Mass entitlement purchased | Budgeting a standard small payload |
| Realised mission | Launch + integration + assurance | Useful mass actually deployed | What the customer experiences |
| Operator cost | Vehicle + labour + operations | Useful mass actually deployed | Understanding the economic frontier |
| Delivered capability | All mission expenditure | Working payload in its final orbit | Comparing complete architectures |
The empty kilograms do not receive a refund
Change the useful mass and customer-side integration bill. The advertised maximum-capacity number stays fixed while the realised cost of this mission moves.
100% of selected vehicle or slot capacity used
- Launch service
- $7,000/kg
- Integration and assurance
- $2,000/kg
- Sticker-price floor
- $7,000/kg
- Denominator penalty
- 1.3×
This is an editorial arithmetic model, not a quote. “Useful payload” means mass delivered to the contracted orbit; it excludes the launch vehicle and may exclude dispensers or propulsion needed after separation. Falcon 9's 22,800 kg figure is an expendable LEO maximum, not the capacity of every recoverable mission.
The small-launch premium buys control
A dedicated rocket can leave when one payload is ready, insert it near the orbit it wants, and protect its schedule from co-passengers. That service can be rationally more expensive per kilogram than filling spare space on a larger vehicle.
Orbit is an energy state. The energy itself is cheap.
At 200 kilometres altitude, a circular orbit requires about 7.78 kilometres per second. Its kinetic plus gravitational energy above Earth's surface is roughly 32 megajoules per kilogram, or nine kilowatt-hours.
Nine kilowatt-hours costs less than one dollar at ordinary industrial electricity prices. But a launch vehicle cannot apply energy only to the payload. The rocket equation forces it to lift propellant that lifts more propellant, fight gravity and drag, reserve margin, and discard dry structure. The physical difficulty sets the architecture; it does not set a high commodity-energy bill.
The hard floor is not the fuel's price. It is the hardware and operating system required to spend that fuel at orbital velocity without losing the payload.
What the headline prices actually are
| Vehicle | Advertised price | Implied $/kg | Basis |
|---|---|---|---|
| Falcon 9 dedicated | ~$27M | ~$1,500 | Public list price and reference orbit capacity |
| Falcon 9 standard mission | ~$67M | ~$2,720 | Public list price at ~22.8 t to low Earth orbit |
| Falcon Heavy | ~$97M | — | Public list price; capacity depends heavily on recovery profile |
| Starship | ~$100M cited; $10M + $67–100/kg targeted | Target | Cited pricing versus announced goal—neither is a contracted market rate |
The list prices are what operators advertise; the per-kilogram figures divide them by a reference capacity, not by what a specific customer's spacecraft actually masses after integration. Target rows are announced goals, not transacted prices.
Recovering the booster removes neither the upper stage nor the mission
The useful unit is a chain from qualified spacecraft to correct orbital state. Every interface in that chain can dominate a lightly loaded or unusual mission.
Make a payload flightworthy
A satellite must survive vibration, acoustics, vacuum, radiation, and the launch provider's interface rules before it approaches a pad.
- Measure
- Qualification level, envelope, centre of mass
- Failure boundary
- Late payload readiness can waste a launch slot even when the rocket is available.
Aggregate and integrate
A dedicated customer buys control. A rideshare customer trades that control for a shared denominator and a standard port.
- Measure
- Manifested mass, volume, port, schedule
- Failure boundary
- Mass capacity may remain empty because volume, geometry, orbit, or timing fills first.
Operate the range
Weather, airspace, maritime exclusion zones, tracking, licensing, and pad turnaround all have to align for a launch window.
- Measure
- Attempts per window, pad occupancy, annual cadence
- Failure boundary
- A reusable booster does not create another pad, range slot, or launch licence.
Fly and recover the first stage
The first stage supplies most liftoff thrust and contains nine of Falcon 9's ten engines. Recovery preserves the most expensive reusable assembly.
- Measure
- Recovery margin, flight count, inspection work
- Failure boundary
- Every kilogram reserved for entry and landing is a kilogram not sold to the hardest orbit.
Spend the upper stage
The second stage completes orbital insertion at nearly orbital velocity. On Falcon 9 it is still discarded after each mission.
- Measure
- Insertion accuracy, restart count, disposal
- Failure boundary
- It is smaller than the booster but newly manufactured for every flight.
Deliver the last mile
A rideshare reaches the carrier's orbit, not necessarily the customer's final plane, altitude, or local time.
- Measure
- Delta-v after separation, transfer time, propellant
- Failure boundary
- A cheap ride can become an expensive mission if the spacecraft must carry its own transfer system.
The operating-system curve
The decisive change was not recovery alone. It was recovery combined with a manifest dense enough to keep vehicles, factories, pads, and crews in motion.
- 1981–2011
Shuttle proves reuse without low cost
The orbiter and solid boosters returned, but intensive inspection, a large standing workforce, and low flight rate produced an estimated $1.5bn mission cost in the NASA benchmark used here.
- 2010
Falcon 9 enters service
A vertically integrated commercial vehicle begins flying at a price far below the government-operated Shuttle programme.
- 2015
First orbital-class booster landing
The launch vehicle's largest hardware element returns to land after placing its second stage on the way to orbit.
- 2017
A flown booster flies again
Reuse moves from recovery demonstration to an operating fleet practice.
- 2020
Scheduled smallsat rideshare
Standard ports and recurring SSO missions turn spare capacity into a product with a public per-kilogram increment.
- 2025
Cadence becomes the advantage
SpaceX reports 165 Falcon 9 launches in one year, 157 using flight-proven boosters.
- 2026
Thirty-seven flights on one first stage
In August 2026 a single first stage flew for the 37th time as SpaceX logged its 100th orbital mission of the year. Booster life is no longer the immediate limit. Pad throughput, upper-stage production, payloads, and demand move to the foreground.
The first reflight matters more than the thirty-seventh
Amortising a booster over ten flights can remove ninety percent of its build cost per flight. Moving from ten to twenty removes only half of what remains. Meanwhile every mission still needs an upper stage, propellant, pad work, range coordination, recovery, inspection, and payload integration.
Reuse bends one line item, then cadence bends the rest
This deliberately transparent model does not estimate Falcon 9's confidential cost. It shows why booster life helps only the capital assigned to the booster; the upper stage, operations, recovery, and integration recur.
54% below the same assumptions on flight one
- Booster amortisation
- $3.0m
- Refurbishment / recovery
- $2.0m
- Other recurring cost
- $18.0m
- Share reuse can still attack
- 13%
SpaceX reported in June 2026 that a Falcon 9 first stage had reflown 34 times, that boosters were qualified for up to 40 flights, and that the second stage remained expendable. The dollar inputs above are scenarios, not SpaceX disclosures.
Cadence is now part of the vehicle
SpaceX reported approximately 620 Falcon 9 launches by March 31, 2026 and more than 540 launches on flight-proven boosters. In 2025 alone, 157 of 165 Falcon 9 flights used previously flown hardware. A design that is technically reusable but flies twice a year carries a different cost structure from one supported by factories, pads, recovery ships, and demand operating every few days.
Reliability is the other half of the record. Of the 330 orbital launches that finished 2025 globally, 165 were Falcon 9 flights, and 157 of those used previously flown hardware. Through September 1, 2026 the world had attempted 208 launches with 201 successes and seven failures—a 97% success rate—with the United States attempting 117, China 59, and the Falcon family logging 104 flights year to date.
Failures still punctuate the curve. An in-flight loss on July 11, 2024 ended Falcon 9's long streak of consecutive successes, and an off-nominal upper-stage event on February 2, 2026 triggered a fleet-wide pause the following day. Cadence that survives a pause and resumes is the asset; raw launch counts without that qualifier overstate the frontier.
Fixed infrastructure
Engineering, pad, factory, range, software, and fleet-support capacity are spread over flights. Higher cadence shrinks this amount per mission until another pad or shift must be added.
Learning by repetition
Frequent launches produce inspection data, standard work, supplier demand, and operational confidence. Reliability and cost can improve together.
Demand density
SpaceX is also its own anchor customer through Starlink. That internal manifest gives the fleet a utilisation advantage that a launch-only provider cannot assume.
Range and pad throughput
The FAA has evaluated up to 120 Falcon 9 launches a year at SLC-40 and up to 100 Falcon-family launches a year across two Vandenberg complexes. Permission and infrastructure are capacity.
The booster stopped being disposable. Much of the launch service did not.
There is no single next bottleneck. The constraint moves with mission class: upper-stage manufacture for a dense constellation, schedule for a rideshare passenger, assurance for a national-security payload, or propulsion for a spacecraft dropped short of its final orbit.
The upper stage
Falcon 9's second stage is not recovered. It contains a Merlin vacuum engine, tanks, avionics, and the precision insertion job, and must be built again for every mission.
Payload readiness
High rocket cadence does not make a delayed satellite ready. Qualification failures and late integration can leave capacity empty or move a mission to a later window.
Volume before mass
Fairing geometry, port location, centre of mass, and deployment clearance can fill before the published mass limit. Kilograms are only one scarce dimension.
Orbit and last mile
Plane change is particularly expensive in orbit. Rideshare passengers may need propulsion, months of drift, or a tug to reach the state where their business starts.
Assurance and insurance
A launch failure destroys the payload and can delay an entire programme. Mission assurance rises with consequence, even when the rocket's marginal flight cost falls.
Concentrated supply
More than half of 2025 global orbital launches and over 80% of mass to orbit were on Falcon 9 according to SpaceX. Low price comes with provider concentration.
The verdict
Below $1,000/kg is a logistics achievement, not a fuel achievement
A fully reusable architecture could attack the expendable upper stage, while autonomy and standard interfaces could attack labour and integration. But sub-$1,000/kg realised cost also requires enough compatible payload, often enough, to fill the machine.
Deep first-stage reuse
Thirty-seven flights on one booster and 157 reused-booster missions in one year move the debate from whether orbital hardware can be reflown to how a fleet is operated.
Standard, frequent access
Scheduled rideshare, online booking, standard plates, and flexible rebooking make launch resemble freight—within a narrow envelope of orbit and payload requirements.
Recover the second stage too
Full and rapid reuse would remove the largest remaining disposable vehicle element, but only after demonstrating payload delivery, recovery, inspection, and repeat flight at useful cadence. Starship is the test case and remains unfinished: IFT-8 on March 6, 2025 lost its upper stage—the second consecutive flight failure—and the first Starship V3 flew on May 22, 2026 after a ground-equipment scrub the previous day. Its cited ~$100M expendable cost is industry-reported; the $10M / $67–100/kg ambition is a Target, never a Measured price.
There is more than one finish line
- Recover hardwareLand an orbital-class first stage intact.
- Refly reliablyUse flight-proven hardware without a reliability penalty.
- Operate a fleetRepeat recovery and reflight hundreds of times at high cadence.
- Fill the denominatorAggregate enough compatible payload to use the capacity offered.
- Standardise the missionReduce integration work without excluding useful payloads.
- Reuse the whole vehicleRecover the orbital stage and fly it again with minimal inspection.
Sources, method, and boundaries
Every cost number on this page carries its evidence type. Historical programme cost, advertised price, booked rideshare price, recognised revenue, and operator cost are not interchangeable. Per-kilogram values are simple divisions by the stated mass boundary and are not inflation-adjusted unless a source did so. The orbital-energy estimate uses a 200 km circular orbit and ignores atmospheric rotation, drag, steering, and vehicle losses; it is a physical lower-bound illustration, not a launch-energy estimate.
- Price / revenue
- What the customer is quoted or what an operator recognises from launch contracts. It includes margin and service scope.
- Operator cost
- The provider's cost of vehicle and operations. Public audited figures are rare; Rocket Lab's quarterly metric is used where available.
- Maximum capacity
- A vehicle performance boundary for a stated reference orbit and configuration, not a promise that a mission uses it.
- Useful payload
- The mass that reaches the orbit the customer contracted for, excluding launch vehicle and any separately accounted transfer hardware.
- Target
- A future claim such as full rapid reuse or a projected cost; never treated here as demonstrated performance.
