In the space roundup for September 26, there was a line easy to scroll past: at World Space Business Week, Firefly's VP of strategy stated that customers "got used to high reliability and low prices that, frankly, didn't support the launch market," and that the industry faces a "reset" lasting from six months to two years. Nearby lay Astra, promising to return to flight in early 2027, and news that SpaceX is winding down the Falcon 9 commercial program.
Reads like the usual supplier complaint about low prices. But at the Paris panel they weren't discussing price levels — they were discussing the unit of measurement itself: price per kilogram to orbit, which the industry had measured by for more than ten years, was declared obsolete. I got curious where this ruler came from and what it showed all those years it was considered objective.
At the Global Launch Providers session on September 15, Michael Creech from Firefly said cheap launch was "subsidized," not economically justified: "this subsidized price-per-kilo model didn't support launch companies, which led to today's supply shortage crisis." Nicole Jordan from Northrop Grumman agreed more softly: the metric was once useful, but their customers talk about something else — "total cost, risk, access to less congested launch sites." The next morning Eddie Kato from Interstellar Technologies called price per kilogram a confusing figure.
Essentially one person held onto it — Arianespace head David Cavaillolès, and he offered a narrower formulation: price needs to be weighed against injection accuracy, because accuracy buys service life for a ten-billion-dollar telescope. No one in the room argued the metric was outdated. This consensus is exactly what's worth unpacking.
First thing that jumps out: everyone burying price per kilogram sells launches. Firefly is sold out 24 months ahead, Blue Origin calls the market "completely seller's market," Exolaunch says customers are booking slots "one to three years ahead." With a full order book, the buyer's favorite metric becomes an obstacle, and the proposal to replace it with "access and schedule" should be read as a negotiating position. Most noticeable with Creech, who directly says "the race to the bottom is breaking right now, and that has to change": a supplier announcing that the customer's selection criteria urgently needs changing is telling you first and foremost about his cost table.
There's also an internal dispute in the Paris materials worth keeping in mind. D-Orbit calls 2029–2031 the critical window, Aerospacelab fears that by then "many satellites will remain sitting on the ground," while Exotrail calls the situation "absolutely predictable" and says they're confident in its resolution. Ars Technica journalist Stephen Clark, who runs a specialized newsletter, also rejected the thesis: "Not sure this philosophy will work long term." An editor who's been writing about launches for years publicly doubts his main sources' version.
Price per pound to orbit wasn't born as a market indicator. It was born as a promise in NASA press materials. In the fact sheet from January 6, 1972, the day after Nixon's shuttle development announcement, the agency wrote: the reusable system will reduce the cost of putting a pound of payload into orbit "from the current 600–700 dollars to 100 dollars," and one flight will cost "less than 10 million dollars — much less than any other spacecraft with comparable lift capacity."
The document's logic is clear. The shuttle was sold as business: budget relief, new users, commercial customers, "opening space for humanity's practical benefit." For conversation with Congress and taxpayers they needed one number that could be compared with disposable launcher prices, and that number became the pound.
In March 1972, NASA released a new fact sheet naming a different number: at full capacity of 65,000 pounds — about 160 dollars per pound versus 900–5,600 dollars for conventional launchers.
Fifteen months later, on June 1, 1973, GAO auditors in report B-173677 — the very one written at Senator Walter Mondale's request — carefully dissected NASA's index and showed why 160 dollars becomes 3,500. Two reasons, both structural. First: the figure excludes 8.1 billion dollars in development and production. Second: the calculation assumes a shuttle loaded to the brim, whereas in NASA's own model for 581 flights, average payload is about 12% of maximum. The agency itself responded to auditors that this index "would be misleading and has no place in cost-benefit analysis" — and this is exactly the index it published in an official statement with economic justification a year earlier.
Here's the mechanism that repeated for half a century. Two dials — whether to include development or not, whether to count by maximum or average load — in this specific case spread the result thirty-five times from the January promise and twenty-two from the March figure. Any launch comparison by kilogram price throughout subsequent history turns exactly these two screws, and almost never reports what position they ended up in.
Contrary to promises, the shuttle failed on cost, and this failure is described by two incompatible numbers. Counting only incremental flight costs, in 2010 it was 409 million dollars, meaning 14,186 dollars per kilogram to low Earth orbit. Counting the entire program, with development and maintenance, it's 1.5 billion per flight and 60,000 dollars per kilogram in 2008 dollars. Against the originally conceived 260 dollars per kilogram in 1972 dollars.
Both numbers are simultaneously correct and convenient for opposite conclusions. Then the same thing happened with Falcon 9. In a 2020 NASA ICES paper, Harry Jones calculates: the shuttle put up 27,500 kg for 1.5 billion, that's 54,500 dollars per kilogram, Falcon 9 puts up 22,800 kg for 62 million, that's 2,720 dollars per kilogram — roughly a twenty-fold improvement. The same text analyzes ISS delivery: the shuttle carried not 27,500 kg there but 16,050, and the kilogram came out to 93,400–105,800 dollars, while Falcon 9 with Dragon capsule — around 25,000. That is, four times better, not twenty.
The difference between twenty and four is explained not by the rocket but by where we're flying. Cargo to the station eats a quarter of lift capacity due to orbital inclination, and the capsule eats another half. The same pair of launchers gives both a triumphant and much more modest figure depending on the point on the map.
About the "era of cheap launches" there's a detailed chronicle, and it's one-directional. At the 2019 SmallSat Rideshare launch, SpaceX offered a million dollars for 200 kilograms plus 5,000 for each additional, in March 2022 — 1.1 million and 5,500 per kilogram, in March 2023 — 1.3 million and 6,500, in August 2024 — about 1.2 million and 6,000.
In February 2026, the price page looks like this: 350 thousand for the first 50 kg plus 7,000 dollars for each kilogram above. Dedicated Falcon 9 over the same period went from 70 to 74 million.
No secret here: in 2023 SpaceX itself announced it intends to raise rideshare price by about 500 dollars per kilogram annually to keep up with inflation. That is, the "era of cheap launches" had a built-in indexing clause announced in advance. And it ended administratively: according to Space Intel Report from September 15, 2026, the company began telling potential customers that new commercial Falcon 9 reservations are no longer accepted, and Transporter and Bandwagon, per SatNews, are closed beyond end of 2028. Even earlier, on September 10, Avio's head said on a call about the discount rideshare: "The party's over, and they're not doing this anymore."
Firefly's version that low prices bred the shortage can only be checked by numbers, and the numbers say something else. In 2025 there were 325 orbital launches worldwide, SpaceX performed 165 of them — almost 51% of global volume and 85% of all deployed satellites. Of those, 122 launches out of 165 were Starlink, meaning 74% of the largest carrier's launches serviced its own product. When the market's main supplier becomes its own largest customer, scarcity is explained faster than through subsidies.
Most vivid in this story is the example of a company that built its entire identity on the metric. Astra reached orbit in 2021 faster than anyone in industry history, and Chris Kemp never tired of repeating the rocket cost "just over a million dollars." Then — arithmetic the company couldn't rewrite: Rocket 3 was retired after two successes, and here sources diverge. SpaceNews writes two successes out of seven attempts, Reuters — two out of six. I couldn't resolve the difference, and it's significant: 29% and 33% success are different diagnoses of the same disease.
The shareholder story ended in July 2024. SPAC placement in 2021, valuation above 2 billion, stock drop over 99%, market cap around 12 million, founder buyout at 50 cents per share — and a final deal for 11.25 million dollars. Two years later, on August 14, 2026, the company seeks 250 million at a 1 billion valuation. That's roughly an 89-fold valuation increase over two years during which it performed zero orbital launches.
New launcher Rocket 4 — 18.9 meters, up to a ton to low orbit, launch in Q1 2027, no reusability, but "we're trying to build a soda can" and "hundreds of rockets per year." Here too there's a place where I can't reconcile. Reuters reports "target launch price of 5 million dollars," while Kemp himself in an interview calculates differently: "if rockets at modest production rates can cost a million dollars, and we're throwing a thousand kilograms, that's a thousand dollars per kilogram." Five million per ton comes out to 5,000 dollars per kilogram, not a thousand. Perhaps it's about rocket price and launch price with markup, but both figures in one article look like a contradiction, and I'm leaving it as is.
Kemp's comparison — "competitors now sell launch at 25–35 thousand per kilogram, so the difference is thirty times" — is correct in exactly one window: against dedicated small launcher flights. Against the rideshare marginal price of 7,000 dollars per kilogram, the advantage shrinks to seven times or less, and is achieved only in a world where Astra actually churns out hundreds of rockets a year, which it never has.
Listing the beneficiaries is straightforward, they were all in that room. Price per kilogram is disadvantageous to any launcher whose kilogram costs more than Falcon's, and that's the entire second tier of the market. With a full order book, it's logical to shift the conversation toward access, risk, schedule, and uncongested sites where there's nothing to compare you with. Telling that not a single player proposed a replacement — no clear new metric appeared at the panel, except "total cost of ownership" and injection accuracy as a clarification to price.
And the metric has one group of genuine allies left — buyers at the very bottom of the price curve. Google researchers estimate an orbital data center becomes competitive at launch prices below about 200 dollars per kilogram, and hope for this threshold by mid-2030s. In February 2026, the same provider's price card stated: plus 500 dollars per kilogram annually. The ruler didn't die, they just turned it the other way.
The ending here is shorter than the hook promised: price per kilogram wasn't a measurement, which is why there's no mourning it. It was born as an office number for Congress, depended its entire life on two assumptions, and died the moment it stopped flattering anyone. And the picture from the Paris panel remains charming: five executives from five different companies explaining that the industry's bottleneck lies in someone else's business.