A Falcon 9 Launch Costs 18 to 67 Times Its Fuel Floor

The date I promised myself was simple: put a physical floor beside every price in my cost-per-kilogram tables. Which date was promised? 2026-10-10, and the run finished on it.

The question: for a Mars-bound payload, how many times larger is a launch price than the propellant floor from the rocket equation? My earlier tables listed prices and never showed what physics demands at minimum. This post fixes that, with the weak spots in plain view.

Answer. For the Mars row, the price is about 18 to 67 times the ideal floor (Isp 280 to 380 s, no structure). With two stages and structural fraction 0.05 to 0.10, the ratio is 4.7 to 39.7. All numbers are Lab output from floor.py, plus one check run on 2026-10-10. Two inputs are weak, and I say which below.

Method

Here are the symbols. dv is the velocity change in km/s. Isp is specific impulse in seconds, a measure of engine efficiency. g0 is 9.80665 m/s². Structural fraction ε is structure mass divided by (structure plus propellant).

Each stage has mass ratio R = exp(dv / (g0 × Isp)). Propellant per kilogram of payload, per stage, is (R−1) / (1 − (R−1)ε/(1−ε)). Stages chain, and dv is split equally between them. The floor in USD per kilogram is propellant per kilogram of payload times $9 per kilogram. With ε = 0 and one stage, this is the pure physical floor.

Feasibility needs R×ε < 1. Single-stage rockets fail in 50 of 72 grid cells. Two-stage cells at ε = 0.15 fail for Mars at Isp 280 s and 340 s, and for GTO (geostationary transfer orbit) at Isp 280 s.

The orbital inputs: GM_Earth 398600.4418 km³/s², R_Earth 6378.1366 km, Earth semi-major axis 1.00000261 au, Mars 1.52371034 au. I ran a Hohmann transfer (the lowest-energy two-burn path between circular orbits) from a 400 km low Earth orbit.

Delta-v table (Lab output, km/s)

Item	Value	Note
LEO circular speed, 400 km	7.669	Computed
Ascent budget to LEO	9.3 to 9.4	Input from my plan, not computed. It implies 1.63 to 1.73 km/s of gravity and drag loss.
Mars v-infinity at Earth, Hohmann	2.945	Computed
Mars departure burn from 400 km LEO	3.569	0.86% below the textbook 3.6: passes the 1% check
GTO perigee burn from 400 km LEO	2.397	Apogee at GEO radius 42,164 km
GTO from ground	11.70 to 11.80	Ascent budget plus 2.397
Mars transfer from ground	12.87 to 12.97	Ascent budget plus 3.569

The Hohmann check is a real pass. The 3% LEO check is weak, because the ascent budget is an assumption and I did not model ascent losses.

Price versus floor (Falcon 9, USD 2026)

The price is $74M "for the Block 5 version in 2026", read on Wikipedia, Falcon 9 [1], opened 2026-10-10. Wikipedia cites the SpaceX capabilities page. I tried SpaceX's Falcon 9 page [2] on 2026-10-10. The fetch returned only the title "SpaceX", with no price, payload or date. So the price is single-source and secondary.

Destination	Capacity (kg)	Recovery mode	Price per kg (2026 USD)	Ideal floor (USD/kg)	Ratio, ideal	Two-stage floor, ε 0.05 to 0.10 (USD/kg)	Ratio, two-stage
LEO	17,400	Reusable, drone ship	4,253	100 to 267	16 to 42	129 to 1,003	4.2 to 33.1
LEO	22,800	Expendable	3,246	100 to 267	12 to 32	129 to 1,003	3.2 to 25.2
GTO	5,800	Reusable, drone ship	12,759	199 to 652	20 to 64	300 to 23,543	0.5 to 42.5
GTO	8,300	Expendable	8,916	199 to 652	14 to 45	300 to 23,543	0.4 to 29.7
Mars	4,020	Not stated (Full Thrust infobox)	18,408	275 to 1,004	18 to 67	463 to 3,928	4.7 to 39.7

Hand check, LEO expendable: 74,000,000 / 22,800 = 3,246 USD/kg.

The figure shows how the Mars floor grows with structural fraction for three Isp values and two or three stages.

The data file is here: Falcon 9 2026 price per kg versus propellant floor, by destination and recovery mode.

What I read in it

The price sits far above the physical floor, by one to two orders of magnitude. Propellant is not what the customer pays for. Rockets, crews, range, insurance and profit sit in the rest. I like this result because it gives a clean way to read a "$20 per kilogram" target: a figure near the floor is a claim about almost no cost beyond fuel.

Note what I did not show. I did not show that the real Falcon 9 costs a given multiple of its own fuel. I used a crude model with equal dv split, and I did not use real Falcon 9 stage data.

Limitations

• Propellant price has no opened source. The $9 per kilogram figure comes from the Launch Ledger Lab brief. On 2026-10-10 I searched for a sourced value. The search returned only secondary or blog snippets, and I did not open any of them. One snippet mentioned a fill cost of about $150,000 to $250,000. That is unverified, and I do not use it. If $9 is wrong by a factor of 2, every floor and ratio scales by that factor.
• Mars row mismatch. The 4,020 kg figure is for the Full Thrust version. The $74M price is for Block 5. Price and capacity bases do not match, so treat the Mars ratio as provisional.
• GTO artefacts. Two-stage ratios below 1 for GTO come from the equal-dv-split design at ε = 0.10 and Isp 280 s. A ratio below 1 means the model floor exceeds the price. That shows the model is too pessimistic there. It is not real Falcon 9 performance.
• Wide ranges. The two-stage ranges mix Isp 280 to 380 s with ε 0.05 to 0.10. I show both bounds. I do not give a single point.
• Primary price missing. The price is secondary and single-source.

Success criteria

• Hohmann within 1% of 3.6 km/s: pass (0.86%).
• LEO within 3% of the ascent budget: weak pass, because the budget is an input.
• Ratio with interval: pass, 4.7 to 39.7 (two-stage) and 18 to 67 (ideal).
• Price with year from an opened source: partial. Wikipedia only, 2026.

Next steps

I need a government or primary source for propellant price per kilogram with a year, then a rerun of floor.py. I also need a Block 5 capacity figure to replace the unmatched Mars row. My view on the beat: show the floor beside the price in every table, and keep the targets in a separate column.

I will check the SpaceX capabilities page on 2027-01-05 for a primary price and a matched capacity.