Space · Tuesday, 18 August 2026
01 · Briefing · what happened
Rockets have never flown more often. Satellite firms still cannot get a ride.
An average of 270 orbital launches a year, prices at record lows, and an industry that says it is in panic. Almost all the growth came from one company, and that company is leaving the commercial market.
270
orbital launches a year
three-year average, triple the rate of a decade ago
165
Falcon 9 flights last year
about 20 fewer expected this year
$4.3bn
of $7.8bn quarterly revenue
came from SpaceX's own Starlink satellites
37,000
satellites needing launch
forecast for 2023 to 2033 by Analysys Mason
At a glance
- An average of 270 orbital rockets a year have flown over the past three years, more than triple the rate of a decade ago.
- Satellite operators still say they cannot get a ride, and one analyst calls it an industry in panic.
- Almost all the growth came from SpaceX, which flew Falcon 9 31 times in 2021 and about 91 times in 2023.
- SpaceX is now winding the rocket down for Starship and has signalled an end to commercial launches after 2028.
- Its own broadband satellites brought in $4.3bn of $7.8bn in second-quarter revenue; launching other people's cargo is a small side business.
- The replacements are late: Vulcan has flown two military missions, New Glenn's only pad exploded in May, Ariane 6 aims for ten flights a year.
- Four new American rockets are chasing 2027 debuts, and none of their makers has run a high-rate launch business before.
Forces in play
Amazon has slowed satellite production waiting for launches it bought five years ago, and rivals are asking Telesat to share its booked Falcon 9 flights.
SpaceX flew Falcon 9 165 times last year and plans about 20 fewer this year, with commercial flights signalled to end after 2028.
Blue Origin is rebuilding its wrecked pad and adding a second one, and four new American rockets target 2027 debuts.
HawkEye 360 and Spire have both locked launch capacity through 2028, and say the open questions start after that.
How it unfolded
- Feb 2022 Russia's invasion pulls Soyuz and Proton out of the Western market
- 2023 SpaceX flies Falcon 9 about 91 times, backstopping nearly everyone
- May 2026 A New Glenn explodes on the pad during a ground test, wrecking Blue Origin's only pad
- This week Operators use earnings calls to prove they have launches booked
- 2028 SpaceX has signalled commercial Falcon 9 launches end; new rockets still scaling
Where this points
Watch whether any of the four new American medium-lift rockets actually flies in 2027 rather than slipping again. Every slip pushes more traffic back onto a rocket its owner is trying to retire.
Full briefing
The paradox
An average of 270 orbital rockets have lifted off each year for the past three years, more than triple the rate of a decade ago. Prices have never been more competitive, flights never more frequent
“What we have is an industry in panic,” said Caleb Henry, director of research at the analysis firm Quilty Space
The abundance was never spare capacity
Nearly all of the growth came from one company. SpaceX flew its Falcon 9 rocket 31 times in 2021 and about 91 times in 2023
That surge arrived exactly when everything else fell away. Russia’s invasion of Ukraine in February 2022 removed the Soyuz and Proton rockets from the Western market. At the same moment Europe was retiring the Ariane 5 for the not-yet-flying Ariane 6, and United Launch Alliance was retiring the Atlas V for the not-yet-flying Vulcan. Japan’s H3 rocket debuted late and has failed twice. India’s LVM3 flies once or twice a year
Only one company had capacity to absorb the loss. Falcon 9 ended up carrying Europe’s Galileo navigation satellites, NASA astronauts and Amazon’s own constellation
Why the ride is going away
SpaceX is winding Falcon 9 down in favour of the much larger Starship. It flew the rocket 165 times last year and will fly roughly 20 fewer missions this year
The company’s own accounts show why. Of $7.8 billion in second-quarter revenue, $4.3 billion came from Starlink, its own satellite broadband service. Space services, of which commercial launch is only a fraction, made up less than 12 percent
The gap between the two is visible on the pad. A Globalstar replenishment mission was set to lift off from Cape Canaveral on August 15, less than 40 minutes before another Falcon 9 from California
The alternatives are not ready
Vulcan debuted in January 2024 and has flown just two military missions, held up by trouble with its Northrop Grumman solid rocket boosters. The Pentagon has moved several payloads off it onto Falcon 9
Ariane 6 works, but its production goal is ten vehicles a year
Analysts at Analysys Mason expect more than 37,000 satellites will need launching between 2023 and 2033
What operators are doing about it
Booking years ahead, and saying so on earnings calls. AST SpaceMobile has ten launches booked with two providers for the roughly 45 satellites of its first constellation, after the New Glenn explosion pushed its commercial service into 2027
The squeeze is reshaping spacecraft too. One industry engineer argues that as launch got cheap, mass stopped being the binding constraint and deployable surface area took its place. Power and heat both need area, and area has to fold into a fairing
Elsewhere this week
China returned to flight fast. A Long March 7A failed on August 10, its second failure in 18 flights since 2020, losing the Zhongxing-4B communications satellite. Six days later a Long March 12 carried 24 internet satellites from Hainan, and a Long March 2C followed within 24 hours from Taiyuan
NASA administrator Jared Isaacman said he is “extremely confident” Artemis 3 will fly in 2027. The mission now needs three launches. A New Glenn carries a test version of Blue Origin’s Blue Moon lander; then the SLS rocket and Orion follow, rendezvous with it and dock in Earth orbit
Intuitive Machines disclosed a $600 million order for three geostationary communications satellites from an undisclosed customer
In science, Hubble and Gaia data pinned down a collision between the young Milky Way and a dwarf galaxy called LKH. It happened about 11.8 billion years ago, extending our galaxy’s known history 1.8 billion years further back
02 · Lesson · why it matters
Why the thing everyone depends on was never chosen
The dependencies that hurt most are rarely decided. They accumulate out of other people's separate delays, until one option is the only one left.
How it works
- One supplier is cheapest and most reliable
- Every buyer picks it, separately and sensibly
- Rivals get no orders, so they stay late
- The choice narrows with nobody choosing
- The supplier leaves, and there is no second option
The twist
The dependencies that hurt most are rarely decided. They accumulate out of other people's separate delays, and because nobody chose them, nobody ever has to defend them.
Where you've seen this
Software
one small free library ends up under half the internet, maintained by one unpaid person
A workplace
one colleague quietly becomes the only person who understands the old system
A town
one road, one bridge, one hospital, because the alternatives were each shelved for good reasons
The catch
Redundancy is not free. The alternative you keep alive costs real money every year for something you hope never to need, which is exactly why it keeps losing the argument.
Full lesson
Nobody signed anything
There is no meeting where the Western world agreed that one company would carry its satellites. No vote, no contract, no moment anyone had to justify. Yet by 2023, Europe’s navigation satellites, NASA’s astronauts and Amazon’s own broadband fleet were all riding the same rocket.
Now that rocket’s owner is moving on, and an industry that has never launched more is calling itself panicked.
Each delay was separate, and each was reasonable
Look at how it happened and you find no plan at all, just a run of unconnected slips.
Russia invaded Ukraine, and two Russian rockets left the Western market for reasons that had nothing to do with launch. Europe was mid-way through retiring one rocket for another that had not flown yet. An American firm was doing exactly the same, for its own unrelated reason: getting off Russian engines. Japan’s new rocket debuted late and then failed. India’s flies a couple of times a year because that is the pace it was built for.
Not one of those decisions was about market concentration. Each was defensible on its own terms. Nobody was asked to weigh them together, because there was nobody whose job that was.
The choice narrows without anyone choosing
Here is the part worth carrying. Every buyer, every year, faced the same easy call: pick the ride that is cheapest, most available and most likely to fly on time. That was the right answer each time. It is still the right answer.
But the answer everyone reaches separately is also the answer that starves the alternatives. A rocket becomes ready by getting orders. Orders go to the rocket that is ready. So the gap widens by itself, and each year the sensible choice is a little more obviously the same one.
The word “market” hides this. A market implies several sellers you can move between. What the industry actually had was one firm’s flight schedule that outsiders were allowed to buy into, while that firm’s real business was becoming something else. When your supplier’s own products earn four billion dollars a quarter and carrying your cargo earns a fraction of that, your access was never a right. It was a courtesy that made sense while it made sense.
The arrangement had no author, so it had no reviewer
An arrangement someone builds can be argued with. You can name who chose it, ask what it costs, propose an alternative.
An arrangement that accumulates has no such handle. There is no author to question and no date on which anyone approved it. It simply becomes the way things are, which is exactly how it comes to look natural rather than contingent. The industry could count its launches, price its rides and see records everywhere. None of those numbers measured the thing that mattered: how many different places a ride could come from.
That is the general shape. A number that goes up can be a real measure of health and still be blind to the only question that decides whether the system survives a shock.
Who is standing on it
This is not a distant industrial story. That rocket has been putting up the satellites behind Europe’s navigation signals, weather forecasting, broadband to places with no cable, and the imagery that watches wildfires. The people relying on those things never made a launch decision in their lives.
And nobody in the picture is above it either. The buyers who booked the cheap ride were being careful with money that was not theirs to gamble. The company that ended up holding everything did not conquer the market; it kept flying while others could not. Even it is now constrained by its own success, retiring a rocket the world would rather it kept.
What it looks like from any single seat
From inside any one of those seats, everything was fine. The launch analyst saw record cadence. The satellite operator saw a booking confirmed. The rocket maker saw a schedule that would surely hold this time. Each of them was looking at a true thing.
The dependence was only visible from a vantage point nobody occupied. By the time it showed up in every seat at once, it had been true for four years. That is the ordinary way these structures form: not by anyone’s choice, but by the slow subtraction of everyone else’s, while all the available instruments read normal.
03 · Lab · your turn
Six Years of Booking Rides
Rehearse the yearly choice that quietly removes your alternatives, and see what the cheapest, highest-output run leaves you holding.
04 · Hope · carry this
Four teams are building rockets right now that nobody strictly had to build. A dependence that formed by accident can be undone the same way, one deliberate choice at a time.
More from Space