In July 2026, a research team led by Alessio Terzi, an assistant professor at Cambridge’s Bennett School of Public Policy, and Francesco Nicoli of the Politecnico Institute of Turin, published a paper in PNAS Nexus that stitched together 4,400 orbital launches from 1960 to 2025 — the largest global dataset of rockets ever assembled — and drew a curve that steepens rather than flattens. The cost of putting a kilogram into low Earth orbit, they wrote, has already fallen roughly 96% since 1960, from around $87,000 to $3,868 last year, and their model projects it hits $1,569 by 2030 and $273 by 2040. That would be steeper than the steamship revolution that reshaped the 19th century, and steeper than the solar photovoltaic collapse that reshaped this one.
Faster than steamships, faster than solar
The Terzi and Nicoli paper applies Wright’s Law — the empirical finding that costs fall by a predictable percentage each time cumulative production doubles — to six decades of launch data. Every doubling of total payload delivered to orbit has cut the average cost per kilogram by 21.2%. Steamship freight after the SS Savannah’s pioneering 1819 Atlantic crossing fell by roughly 15.5% per doubling of cargo volume such as wheat and cotton, according to the authors’ historical comparison. Solar photovoltaic modules, the standard benchmark for a technology that got cheap fast, fell at a slightly slower rate than launch costs are now falling.
The comparison is meant to be provocative. Steamships underwrote the first wave of globalization. Solar underwrote the energy transition. Space launch, on the numbers, is now falling faster than either.
The paper also splits its timeline at 1989. Before the fall of the Berlin Wall, when the space race was still a state-led prestige contest, costs fell at one rate. After it, when commercial operators began entering the market, they fell more than two-and-a-half times faster. “State-led competition during the Cold War did not foster cost efficiency on the same scale as the ensuing era of international space cooperation and private-sector involvement,” Nicoli told the Cambridge press office.
Put differently: Apollo did not lower launch costs. Falcon 9 did.
How the curve bent
Two structural shifts explain the acceleration. The first is reusability, pioneered at commercial scale by SpaceX and now spreading to competitors. On the night of June 28, 2026, a Falcon 9 first stage designated B1085 lifted a 15,400-pound SiriusXM satellite from Cape Canaveral on its seventeenth flight, according to Spaceflight Now. The same booster had previously carried Crew-9 to the International Space Station, Firefly’s Blue Ghost lunar lander toward the Moon, the classified USSF RRT-1 mission, and nine separate Starlink batches. Thirty days elapsed between B1085’s previous flight and SXM-11 — roughly the turnaround cadence Falcon 9 has been holding to in 2026.
A sister booster, B1067, has now flown more than thirty-five missions, and SpaceX flew a flight-proven Falcon for the 600th time on July 14. The marginal cost of a reused Falcon 9 launch now sits well below the manufacturing cost of any expendable Western competitor’s first stage.
The second shift is competition. At 12:15 a.m. Eastern on July 10, 2026, the first stage of a Long March 10B rocket descended over the South China Sea, deployed four metal hooks, and let itself be caught in a cable net stretched across the deck of a recovery ship called the Linghangzhe. With that catch, China became only the second nation ever to recover an orbital-class booster, and the first anywhere to do it without landing legs. The China Aerospace Science and Technology Corporation says the same stage will fly again before the year is out. Rocket Lab, Blue Origin, Firefly, and a growing cohort of Asian and European entrants are chasing the same economics.
The SpaceX problem
The Terzi and Nicoli projection has a hidden assumption. It assumes the payload-doubling engine keeps turning. That engine, at the moment, is SpaceX. According to the launch tracker maintained by NextSpaceflight and NASASpaceflight, SpaceX delivered approximately 2,413 tonnes to orbit in 2025 — about 86.6% of all known mass launched globally, and roughly 34.8 times the tonnage of the second-place operator, United Launch Alliance. If the doublings stop, the curve stops.
The company’s financial trajectory dramatizes the stakes. On June 12, 2026, SpaceX completed the largest IPO in capital markets history, pricing at $135 a share and closing its first session at $161. Within days, the stock surged to an intraday market cap above $2.9 trillion, briefly leapfrogging both Amazon and Microsoft, according to CNBC. Shares then retreated. By late June, SPCX was trading around $153, roughly a third below its post-IPO peak.
Whether SpaceX hits Elon Musk’s projected 2030 revenue targets matters less, in the Terzi and Nicoli framing, than whether anyone else keeps flying enough to sustain the doubling cadence. A monopolist has little reason to keep cutting prices once the competition has been squeezed out. The steamship analogy the authors draw on is instructive here too. Costs kept falling through the 19th century largely because dozens of shipping lines fought each other on transatlantic routes.
What lower costs unlock
At the current $3,868 per kilogram, sending a payload to orbit still costs roughly what a high-end laptop retails for. At $273 per kilogram — the 2040 target — it would cost less than most airline luggage overweight fees. That price threshold changes which industries plausibly close a business case in orbit.
Terzi and Nicoli list several: zero-gravity pharmaceutical research, orbital tourism, high-purity fiber-optic manufacturing, and 3D-bioprinted human organs — the kinds of goods that benefit from microgravity and can bear the cost of return. Longer term, the authors point to orbital solar power, asteroid mining, and lunar propellant depots. The global space economy, per Novaspace’s tracking, was already worth roughly $613 billion in 2024 — comparable to Sweden’s GDP — and every major forecaster now expects it to cross $1 trillion within a decade.
None of the more exotic industries exists at scale today because the launch cost overwhelms the business case. That is what steep cost reductions are meant to fix.
The next disruption is software
Reusability may not be the final act. A paper by Haipeng Chen, Xiaowei Wang, and Feng Zhang, published in the October 2025 issue of the Chinese Journal of Aeronautics, argues that artificial intelligence applied to launch operations could shrink turnaround times to the hour level and improve flight reliability by one to two orders of magnitude. The authors describe four capability domains: automated ground testing and fault diagnosis, real-time in-flight anomaly response inside the seconds available during a burn, machine-learning health-monitoring for rapid re-launch decisions, and autonomous orbit management for a sky that is filling with satellites at roughly 2,000 new objects per year.
If reusability made rockets like airplanes in principle, AI-driven operations aim to make them like airplanes in practice — high-frequency, high-reliability, ground-crew-light. The Terzi and Nicoli curve implicitly assumes something like this happens. Without operational automation, launch cadence hits a ceiling regardless of how cheap each rocket is to build.
What could go wrong
The Cambridge authors are careful to flag that continued cost declines are not guaranteed. Two structural risks dominate their downside scenarios.
The first is market concentration. If SpaceX’s dominance calcifies into monopoly before Blue Origin, Rocket Lab, or the Chinese state ecosystem builds real scale, price competition ends. The second is geopolitics. Export controls, sanctions regimes, and the fragmentation of orbital debris governance could all raise costs by restricting who can launch what for whom. A bifurcated space economy — one Western, one Chinese, with limited exchange between them — would erase some of the scale benefits that drive cost reductions in the first place.
There is also demand risk. The 32,000-tons-per-year figure Terzi projects for 2040 assumes buyers actually exist for that much orbital capacity. Starlink and its Chinese equivalents account for most current growth. If megaconstellations saturate before other applications mature, the volume growth that underpins cost reductions could stall.
NASA’s 2014 dual-source decision — the $4.2 billion Boeing Starliner contract and the $2.6 billion SpaceX Crew Dragon contract, signed on the same day at Kennedy Space Center — was designed precisely to prevent single-supplier dependency in crewed spaceflight. A decade later, the divergence between the two programs is unambiguous. Whether the broader launch market avoids the same fate is the empirical question that determines whether Terzi and Nicoli’s 2040 number is a floor or a fantasy.
The dataset stops at 2025. B1085 is being inspected in a Florida hangar for its eighteenth flight. The Long March 10B stage caught over the South China Sea on July 10 is being trucked back to Hainan for refurbishment, targeting a reflight before the calendar turns. Somewhere in the next doubling — the one that would, if the curve holds, put a kilogram into orbit for roughly what a cross-country checked bag currently costs — is the payload that finally makes an orbital factory pencil out. It is already on someone’s manifest.
