The Rideshare Cliff

01Summary

SpaceX has closed its rideshare programme to new customers beyond late 2028. The capacity it carried, about half of the world's non-megaconstellation smallsats, has no replacement of comparable price, cadence or reliability on any current timeline.

Since late June 2026, at least nine SpaceX customers and integrators have confirmed that SpaceX is no longer accepting new Transporter (sun-synchronous) or Bandwagon (mid-inclination) reservations for dates beyond late 2028. Bloomberg reported on 23 July that the same applies to new dedicated Falcon 9 bookings. On 22 August Elon Musk wrote that SpaceX will be "winding down Falcon" once Starship flies reliably several times a week, and Ars Technica has reported that SpaceX told NASA it plans to end commercial Falcon 9 launches in 2028. SpaceX itself has announced nothing: there is no press statement, no filing language and no earnings-call remark. Existing bookings are being honoured, Transporter-18 and Bandwagon-5 are scheduled for no earlier than October 2026, and integrators have bought five whole Falcon 9s for 2027 and 2028. What remains open is whether the freeze hardens into a retirement date and whether Starship can ever serve sun-synchronous orbit from Texas or Florida.

The replacement side is the larger problem. Every medium-lift vehicle nominated as a substitute has, as of 25 August 2026, either never flown or had a 2026 that went backwards. The exceptions are small: Electron, Vega-C, Ariane 6 and H3 have flown well this year, and none of them carries Transporter-class volume. No Neutron stage has reached the pad, and Rocket Lab's language on 10 August shifted from a fourth-quarter launch target to "delivery of Neutron to the launch pad in Q4 2026." New Glenn has one failure in three flights and lost its pad to a static-fire explosion in May. Vulcan has been grounded since February. Firefly's Alpha has succeeded on three of seven flights and its Block 2 debut slipped to the fourth quarter. Isar's Spectrum is 0 for 1 with a second attempt now targeted for September; RFA One slipped again on 10 August; PSLV is grounded after two consecutive failures with an undisclosed cause; Japan's Kairos is 0 for 3. Terran R, Nova, Eclipse, Miura 5 and Maia are unflown. Starship has not yet reached orbit (Flight 14, the first attempt, was targeted for 28 August), has no rideshare product or price, and has no documented route to sun-synchronous orbit from any site: no west-coast pad or environmental process, and no polar-corridor approval from Florida.

Six findings

1. The gap opens in 2029, and it is already shaping decisions. The remaining Transporter and Bandwagon manifest plus five aggregator-bought Falcon 9s give 2027 and 2028 roughly 21 and 38 tonnes of SpaceX-sourced third-party capacity. Much of it is already sold: SEOPS reported its 2028 sun-synchronous Waymaker mission about 90% contracted by mid-August (a July report had said the manifest filled within three weeks, with about 30 customers waitlisted). Anyone planning a 2029 launch today has no SpaceX rideshare option, no dedicated Falcon 9 option if Bloomberg's reading is right, and no proven alternative at anything near $7,000 per kilogram.
2. Demand is larger than the anecdotes suggest. SpaceX's dedicated rideshares carried 436 payloads in 2025, about 46% of the count of non-megaconstellation smallsats launched worldwide and, by our estimate, 55 to 65% of Western ones. We put the rideshare-class market at about 22 tonnes a year today, growing to roughly 30 tonnes by 2029 as average smallsat mass rises; the mass figure is the least certain number in this report. 259 organisations launched a smallsat in 2025 and 106 were doing so for the first time; the long tail of first-timers has no comparable alternative.
3. Nameplate replacement capacity looks adequate on paper by 2029 and is not. Our base case sums to about 53 tonnes of non-SpaceX third-party rideshare-class capacity in 2029 against 29 tonnes of demand. About 61% of that supply is on vehicles that have never flown, and weighting each vehicle by the probability that it is operating at the assumed cadence, and by its flight record, cuts credible supply to about 32 tonnes (23 to 39 depending on how fast the new vehicles arrive): parity with base demand, short of the high case. In the bear case, credible 2029 supply is 11 tonnes. Vehicles flying today with an established record (Electron, Vega-C, Ariane 6, H3) supply about 7 tonnes, a quarter of base demand; adding PSLV once it returns and the early-record vehicles takes it to about half, and none of that capacity carries a 140-satellite manifest.
4. Neutron is the right shape and the wrong date. A fully loaded Neutron to polar orbit would undercut Transporter's list price per kilogram. But Rocket Lab is selling dedicated missions at $50 to 55 million with no discounting, has no scheduled rideshare product (one rideshare contract was pending in its backlog as of late 2025), says the Kepler deal "pushes Neutron's manifest toward full capacity through the end of the decade," and its stated cadence is one flight, then three, then five. In our base case, in which Neutron flies but is sold as dedicated missions, it contributes nothing to rideshare in 2027 and about 3 tonnes in 2028.
5. The price of access rises roughly 1.2 to 6 times, and the price of reliability rises more. Every alternative costs more per kilogram than Transporter, from Neutron-aggregated rides at an estimated $8,000 to 12,000 to Electron at $28,000 to 42,000. Moving a $10 million, 150 kg satellite from a 99% reliable vehicle at $7,000 per kilogram to an 85% vehicle at $10,000 multiplies the expected mission loss per kilogram roughly sixteen-fold and takes the reliability-adjusted cost from about $7,700 to $21,500 per kilogram. Insurance can transfer that risk only where capacity exists, and for new vehicles it is thin and priced by spot.
6. The winners are whoever holds inventory. Integrators that bought Falcon 9s (Exolaunch, now being acquired by EQT; SEOPS; Maverick with Portal), operators with dedicated-launch anchors (BlackSky and iQPS on Electron, Kepler on Neutron), and Rocket Lab itself. The most exposed are high-cadence rideshare users without an announced alternative, led by ICEYE (38 satellites on SpaceX rideshares in 24 months, ramping to 100 a year), Spire, Muon, HawkEye 360 and Planet's replenishment line, plus every university and NASA CubeSat Launch Initiative payload.

Confidence: high on the freeze and its scope as reported; high on vehicle status and flight records; moderate on demand mass (SpaceX does not publish per-mission mass); moderate on the 2029 supply model; low on any Starship sun-synchronous timeline.

02What SpaceX did, and did not do

The story surfaced on 25 June 2026 when SpaceNews reported, citing at least nine SpaceX partners and customers, that SpaceX was no longer accepting Transporter or Bandwagon bookings for late 2028 or early 2029 and beyond, and that available slots in the intervening years could fall to roughly half of recent levels (the SpaceNews text is not accessible to us; its wording is relayed by TechTimes and an industry summary). SpaceX did not respond to requests for comment. SatNews repeated the finding on 20 July with the same sourcing, small satellite operators and mission integrators rather than SpaceX; AIAA carried a similar headline on 21 July that we could not retrieve. Bloomberg's 23 July report, sourced to people familiar with the matter and quoted here second-hand because the original is paywalled, widened the scope: SpaceX has begun turning away satellite operators seeking dedicated Falcon 9 launches beyond 2028, has stopped new rideshare reservations, and "has also halted production of some non-reusable Falcon hardware." Falcon 9 would continue to serve NASA and Defense Department missions, and Crew Dragon still needs a Falcon.

Two later statements bound the intent. On 22 August Musk posted: "Once Starship is flying reliably several times per week, it makes sense to shift super scarce SpaceX engineering and production resources to Starship to get launch rate to several times per day, which means winding down Falcon." No date was attached. Ars Technica, as relayed by Space.com on 24 August, reported that SpaceX has told NASA it plans to end commercial Falcon 9 launches in 2028, and described NASA's concern about the retirement; Space.com adds that ISS crew and cargo flights are expected to continue until at least 2030. The Ars report itself is not accessible to us.

Why SpaceX is doing it

No source has SpaceX's reasoning in its own words, so what follows is inference from its disclosures. Three facts line up. First, Falcon 9 is at capacity and most of it is internal: SpaceX reached its 100th launch of 2026 in the third week of August (18 August by one count, 22 August by another) and about 78% of its flights to mid-August carried Starlink; Musk told investors on 4 August that Falcon delivers "roughly 2,500 tons a year to orbit." Second, external launch is a shrinking part of the company. Launch Services revenue was $648 million in the second quarter and $978 million for the first half, down from $1,056 million in the first half of 2025 even as cadence rose; the Space segment as a whole was 12% of second-quarter revenue. Third, Transporter was never lucrative. Novaspace's Gabriel Deville estimated in 2024 that 2023's Transporter missions earned SpaceX $12.5 to 20 million each against an operating cost of up to about $15 million, leaving "anywhere between $5 million and nothing per Transporter mission in profits." A product that consumes a Falcon core and a Vandenberg slot to earn a thin margin from 80 customers is the obvious one to close when the company's own Starlink V3, Starshield, Starfall re-entry and orbital data-centre programmes need every core through 2028 and the engineering organisation is being pointed at Starship.

The scope question matters for the model in Section 6. If only rideshare is closed (SatNews, AIAA, SEOPS), then aggregator-bought dedicated Falcon 9s could in principle continue past 2028 and the cliff would be a price step rather than a wall. If Bloomberg is right that dedicated bookings are closed too, the five Falcon 9s already bought for 2027 and 2028 are the last of their kind. The Ars Technica report that SpaceX has told NASA it will end commercial Falcon 9 launches in 2028 points to the second reading. The buyers' rush for 2028 vehicles is consistent with either, since a rideshare-only freeze would also send them to the product still on sale. We model the second reading as the base case, quantify the first in Sections 6 and 10, and on the evidence put the odds at roughly 60:40 in favour of the harder reading. A Starship that reaches operational cadence would also free Falcon 9 cores from Starlink duty before any Starship rideshare product exists, which is a second route to a softer landing.

Confidence: high that the freeze exists and covers rideshare; moderate that it covers dedicated Falcon 9; moderate on the motive reconstruction.

03The product being withdrawn

Transporter is not a rocket. It is a scheduled, priced, reliable bus service to a specific orbit, and that combination is what the market has to rebuild.

SpaceX announced the Smallsat Rideshare Program in August 2019 and flew Transporter-1 on 24 January 2021 with a record 143 spacecraft. Seventeen Transporter missions to sun-synchronous orbit, four Bandwagon missions to mid-inclination orbits (from April 2024) and one Twilight dawn-dusk mission (January 2026) had flown by 25 August 2026. Adding up the published per-mission counts gives roughly 1,650 spacecraft on 22 dedicated rideshare missions; press reports after Transporter-17 put SpaceX's cumulative rideshare count past 1,800; the basis is not published and presumably includes rideshares flown on other missions. Cadence rose from two missions in 2021 to six in 2025 (Figure 1). Every Transporter since April 2023 has flown from Vandenberg SLC-4E; the early missions used Florida's southerly polar corridor.

What the service offered

Three things made Transporter the default choice for Western smallsat operators. It was scheduled: two to four sun-synchronous missions a year at predictable times, so a constellation could plan replenishment. It was priced by the kilogram on a public rate card, which let a first-time operator budget a 50 kg satellite at $350,000 without a negotiation. And it flew on Falcon 9, whose in-flight record stands at roughly 690 successes in 693 attempts, so a customer could self-insure a $5 million satellite and be right to. The rate card history below shows SpaceX exercising pricing power as demand outran supply: the per-kilogram rate has risen 40% since late 2019 and the minimum ticket has risen 27% since October 2022, while the queue lengthened to two years or more.

How much demand it carried

SpaceX does not publish per-mission mass. Novaspace's Deville counted 29 tonnes across the first nine Transporters, about 3.2 tonnes a mission. Recent manifests are heavier: K2 Space's Gravitas bus (2 tonnes) flew on Transporter-16, MBZ-SAT (750 kg) on Transporter-12, CAS500-4 (about 500 kg) on Transporter-17, and Exolaunch alone integrated more than 1.5 tonnes on Transporter-13. Our estimate is that the six dedicated rideshares carried 17 to 25 tonnes of customer mass in 2025, and we use 20 tonnes; the research streams behind this report put the figure anywhere between 10 and 35 tonnes, so it is the least certain number in the model. Adding about 2 tonnes a year that already flies as co-passengers on Vega-C, PSLV, H3 and Electron gives a 22-tonne 2025 base for the rideshare-class market. The implied average of about 45 kg per payload is far below Bryce's 177 to 213 kg average for non-broadband smallsats because the Transporter population is CubeSat-heavy: Planet's SuperDoves and Spire's Lemurs are 5 kg-class, and universities fly 1U to 6U.

The count is better documented. BryceTech's Smallsats by the Numbers 2026 records 4,466 smallsats launched in 2025, of which 79% were LEO broadband constellations (Starlink, OneWeb, Amazon Leo, Qianfan, Guowang), leaving about 938 non-megaconstellation smallsats. SpaceX's dedicated rideshares carried 436 payloads, or 46% of that count, though the two figures are not on the same basis: the 436 includes hosted payloads, transfer-vehicle sub-payloads and a few spacecraft above Bryce's 600 kg limit. Restricting the denominator to Western operators, who cannot use Chinese or Russian vehicles, the share is plausibly 55 to 65%, consistent with Deville's estimate that Transporter captured 57% of Western smallsat launch demand between 2019 and 2023. Bryce also records that 259 organisations launched a smallsat in 2025 and 106 of them were first-time operators, up from 59 in 2024. In our judgement most Western first-timers reach orbit through a Transporter port; Electron, PSLV and Vega-C have carried some.

Demand is growing and getting heavier. Novaspace forecasts 16,900 smallsats under 500 kg over 2026 to 2035, about 1,690 a year, with growth "driven by growing sovereign constellation demand." Bryce's ex-broadband average mass sits around 200 kg (its 2026 edition gives 213 kg for 2024 and 177 kg for 2025 and describes smallsat mass as increasing). The bus market is moving up: NanoAvionics added a 500 kg-class platform on 24 August because customers "want that greater degree of capability in a larger platform," Kepler's Aether relays are 300 kg class, and Exolaunch's new EXOtube stacks are sized for 20 to 1,400 kg satellites. We hold the market at 22 tonnes in 2026, a supply-constrained year with one fewer dedicated mission, then grow mass 10% a year: 24 tonnes in 2027 to 32 tonnes in 2030 in the base case, with a low case 25% below (18 to 24 tonnes) and a high case 35% above (32 to 43 tonnes). Because 2025 throughput was flown under a two-year queue, the base is a floor under demand rather than a measurement of it.

Who rides it

The manifests of Transporter-10 through 17, Bandwagon 1 to 4 and Twilight-1 identify the dependent population. The table ranks the largest commercial users by verified 2024 to 2026 appearances and records what each has said or done since the freeze became public.

The channel through which most of this demand flows is even more concentrated than the operators. Exolaunch says it has flown on every Transporter and Bandwagon since 2021, has deployed 839 satellites on 48 missions and integrated 49 of Transporter-17's 81 payloads; it has 68 more manifested on Transporter-18 and Bandwagon-5. SEOPS integrated 10 on T-17, 19 on T-16 and 23 on T-13. Maverick, ISISpace, D-Orbit, Impulse and Momentus carry the rest. Section 9 returns to what the freeze does to that channel.

Confidence: high on mission counts and pricing; moderate on mass and on the Western share; high on the exposure table's manifest facts, moderate on the exposure grades.

04Starship: the eventual answer and the interim problem

Starship is the reason SpaceX can afford to close Transporter. It is not yet a reason for anyone else to relax.

The vehicle's 2026 record is real progress and still short of orbit. Flight 12 on 22 May debuted the V3 ship from Starbase's second pad; the booster suffered engine failures and a hard water landing, the ship lost a vacuum Raptor but completed a suborbital flight and re-entry. Flight 13 on 24 July deployed 20 Starlink V3 satellites on a suborbital arc, survived re-entry with what Musk called a solved heat shield, and splashed down upright; the booster's landing burn failed. All thirteen Starship flights to date have been suborbital. Flight 14, targeted for no earlier than 28 August, is the first orbital-insertion attempt and the first operational Starlink V3 deployment; the ship tower catch floated on the 4 August earnings call as possible on the next flight was deferred on 20 August to "a few months." No ship-to-ship propellant transfer has been demonstrated. Musk's stated ambition is "at least 1 flight a day, possibly more" a year from now; the Falcon wind-down condition he set on 22 August (Section 2) is several reliable flights a week.

Pads are further along than flights. The FAA published a Final Environmental Impact Statement and Record of Decision for Starship at LC-39A on 3 February 2026 (up to 44 launches a year) but had not issued a licence as of 25 August. The Department of the Air Force signed its Record of Decision for SLC-37 on 20 November 2025 (up to 76 launches a year), conditional on further FAA airspace analysis. SpaceX's CFO told investors on 4 August the company is activating pads at Starbase, LC-39A and SLC-37. Bloomberg's sources expect the first Florida Starship launch late this year or early 2027.

The sun-synchronous problem

Transporter's orbit is the issue. Sun-synchronous orbit needs a near-polar, retrograde inclination of about 97.5 degrees. From Vandenberg that is a straight southward ascent. From Florida, Falcon 9 reaches it through the polar corridor reopened in 2020: a south-southeast ascent, a dogleg down the coast, and an overflight of Cuba that the Air Force permitted only for vehicles with autonomous flight termination and, in practice, booster flyback. Transporter-1 through 6 used it. From Starbase, a southbound polar trajectory overflies Mexico and lies outside any existing authorisation.

Nothing in the public record addresses Starship and polar orbits. Neither the LC-39A nor the SLC-37 environmental documents specify azimuths. There is no Starship pad, licence application or environmental process at Vandenberg: the Department of the Air Force's October 2025 Record of Decision raised SpaceX's Vandenberg ceiling to 100 launches a year and authorised SLC-6, but for Falcon 9 and Falcon Heavy only. The single super-heavy signal on the West Coast is a Department of the Air Force request for information of 30 December 2025 for "heavy/super-heavy launch capability" at SLC-14, which named no company or vehicle and drew a 12 January 2026 letter of concern from Representative Salud Carbajal. For scale, the SLC-37 environmental process took 21 months from notice of intent to Record of Decision, with FAA work still pending. A Vandenberg Starship pad before 2029 is not supported by any document we found, and a Florida polar corridor for a roughly 5,000-tonne vehicle over Cuba is an approval that does not yet exist. The counterargument deserves stating: the Falcon 9 corridor went from proposal to first flight in about a year once the Air Force agreed, Starship's performance margin makes a dogleg penalty irrelevant, and silence in the environmental documents is not a prohibition. The binding constraints are SpaceX's priorities and the absence of any product, not physics. We weight the counterargument low because SpaceX has said nothing about polar Starship flights from any site and its own polar demand is small next to Starlink's.

Two further gaps matter to the customers in Section 3. SpaceX has said nothing about a Starship rideshare product, price or first mission; the only booking with a rideshare character is ispace's 500 kg lunar slot no earlier than 2030, and SpaceX's prospectus commits only to "payload delivery to orbit in the second half of 2026" for Starlink V3. And SpaceX's own Transporter-class demand, Starlink V3, Starshield, Starfall and the orbital data-centre programme, is queued ahead of any third party. The realistic reading is that Starship will eventually offer sun-synchronous capacity at a price that resets the market, that "eventually" means no earlier than 2029 to 2030 with low confidence, and that the 2029 gap it leaves behind is the subject of this report.

Confidence: high on flight and pad status; high that no Starship SSO route or rideshare product exists in the record; low on any date for one.

05The replacement field

More than thirty vehicles could in principle carry a Western smallsat to sun-synchronous orbit between 2027 and 2030. Apart from Falcon 9, four are flying today with an established record, and none of the four sells a scheduled sun-synchronous rideshare at scale.

Neutron: timeline, economics and the rideshare question

Rocket Lab's Neutron is the vehicle the market most wants to be ready, and the one whose timeline has moved most. Announced in March 2021 for a 2024 debut, it slipped to mid-2025 (May 2024, engine-driven), then to the second half of 2025, then to 2026 after a November 2025 call that moved pad arrival to the first quarter. A first-stage tank ruptured in hydrostatic qualification on 21 January 2026, which Peter Beck attributed to a hand-laid section built by a contractor before Rocket Lab's automated fibre placement machine was running. The February 2026 call reset first launch to the fourth quarter of 2026; the May call held it. The 10 August results release and call changed the metric: production is "aligned with the target delivery of Neutron to the launch pad in Q4 2026," exact launch timing depends on the outcome of first-stage qualification and other tests later in 2026, and, in Beck's words on the call, "the window for an end-of-year launch is narrowing." Third-party trackers moved the debut to 2027 the same day; a prediction market put a 2026 launch at about 14%.

The hardware is close. Archimedes has passed 400 hot fires across both stage variants, Stage 2 was qualified in April 2025 and is having flight avionics installed, the fairing and thrust structure are qualified, interstage load tests are complete, and Launch Complex 3 at Wallops has been open since September 2025. What remains is the sequence that has undone other new vehicles this year: first-stage qualification, integration, static fire, licence, wet dress, launch. Beck said as much on the call, noting "for another space company that when that doesn't go well, it really doesn't go well," evidently a reference to New Glenn's 28 May static-fire explosion. No FAA Part 450 licence for Neutron is in the public record, and the "Return On Investment" landing barge, needed from Flight 2, was still under construction as of the last tracker update in May.

On paper Neutron is exactly what Transporter's customers need. The payload user's guide gives 13,000 kg to a 500 km orbit at 40 degrees with a barge landing and 10,100 kg to polar orbit; 8,500 and 6,200 kg with return to launch site; 15,000 and 11,800 kg expendable. At the $50 to 55 million price the CFO reaffirmed on 10 August, a full barge-landing polar mission works out at $4,950 to 5,450 per kilogram, below Transporter's $7,000 list, and a return-to-site polar mission at $8,070 to 8,870. The guide describes multi-manifest and rideshare missions using the restartable second stage and ESPA-class dispensers, and a 24-month standard integration schedule.

Three things stand between that arithmetic and a Transporter substitute. First, Rocket Lab is not selling per-kilogram slots; it is selling dedicated missions at full price and says it has "absolutely zero issues in selling full price Neutrons pre-test flight." Its backlog included one rideshare contract pending manifest confirmation as of November 2025, but a scheduled rideshare product would have to be built by an aggregator buying whole vehicles, which is what Exolaunch and SEOPS now do with Falcon 9. Second, the manifest is filling with dedicated customers. They include two launches for a confidential constellation operator (November 2024), the May 2026 five-launch deal, Kepler's 2028 mission, the $397 million Space Force SB-AMTI award to build and launch Flatellites, and the Air Force Research Laboratory point-to-point cargo experiment. They also include the NSSL Phase 3 Lane 1 on-ramp, which requires one successful flight before Neutron can compete for task orders, and "future Iridium capacity" once the $8 billion Iridium acquisition closes in mid-2027. The Kepler release says the deal "pushes Neutron's manifest toward full capacity through the end of the decade"; Kepler itself attributes the purchase to orbital control and schedule rather than to SpaceX's availability. Third, cadence. Beck's stated ramp is "1-3-5": one flight, then three in the following year, then five. If the first flight is in the first half of 2027, the five-flight year is 2029 at the earliest, and Rocket Lab said in November 2025 that the first flight would carry no customer payload.

Beck's reading of the market on 10 August is the clearest executive statement on launch scarcity in the record, though he did not refer to the SpaceX freeze. In one answer: "I have personally never seen launch so constrained pretty much ever. Other launch providers are backing off and focusing on their own needs as well." In another: "Even as new capacity comes on market from some competitors, a lot of that capacity is already spoken for their own internal programs. Whether it be internets or AI data centers or whatever... I see this constrained launch market persisting for quite some time." Rocket Lab has the balance sheet to finish the job: $2.13 billion of cash at 30 June, a $2.36 billion backlog, record quarterly revenue of $234 million, and a $1.94 billion at-the-market programme partly earmarked for the Iridium cash component. It also has 93 Electron missions of experience with a 1-3-6 ramp of its own.

Electron: the wrong size at the right cadence

Rocket Lab's small launcher is the only vehicle in this report whose 2026 record is unambiguously good: 21 launches in 2025, 14 by 21 August 2026, 100% mission success in both years, 52 consecutive successes since the September 2023 failure and 89 successes in 93 attempts overall. It is also structurally unable to absorb Transporter volume. Electron lifts about 300 kg to low Earth orbit and 200 kg to sun-synchronous orbit at an implied price near $8.5 million, which is $28,000 to 42,000 per kilogram, four to six times Transporter. Twenty-five launches a year at 200 kg is 5 tonnes of theoretical capacity, of which dedicated constellation customers (iQPS, Synspective, BlackSky) and a rising share of HASTE suborbital missions (at least 35 of the 90-plus launch backlog) leave perhaps 2 to 3 tonnes open, roughly one Transporter's worth. Electron is the answer for a 100 to 300 kg satellite that needs a specific orbit on a specific date, not for a 5 kg CubeSat or a 40-satellite batch.

The rest of the field

The table summarises every other candidate, ordered by realistic relevance to the sun-synchronous rideshare gap. "Record" is orbital attempts and successes as of 25 August 2026 with partial failures counted against; "SSO reach" notes whether the launch site can reach sun-synchronous orbit without a large performance penalty; "Rideshare product" records whether a scheduled, priced multi-customer service exists. OED SEPI scores are from the 25 August pipeline run on the one-year window, shown with the number of observed factor families so the reader can see how much evidence sits behind each score (see Appendix C).

Two observations follow from the table. The vehicles that could carry Transporter-class mass in 2027 and 2028 at a Transporter-class price are Falcon 9s that integrators have already bought and, at full fill, Neutron once it flies. Everything else is either too small to matter (under 1.5 tonnes a flight, often under 0.5), too expensive per kilogram (Electron, Alpha, Vega-C whole-vehicle), too committed to anchor tenants (Vulcan, New Glenn, Ariane 64 to Amazon Leo; Alpha to Lockheed Martin), unflown, or legally closed. The second observation is that the European Launcher Challenge, with EUR 902 million subscribed at the November 2025 ministerial for Isar, MaiaSpace, RFA and PLD (Orbex withdrew in February after entering administration), buys ESA missions from 2026 to 2030 and requires an orbital flight by 2027; it does not create commercial sun-synchronous rideshare slots. Combined nameplate of all four challengers is about 3 tonnes to sun-synchronous orbit per round of flights.

Orbital transfer vehicles and integrators

The companies that sit between operators and rockets are the ones adapting fastest, and the way they are adapting tells us what the post-Transporter market looks like. Impulse Space has flown three Mira missions on Transporters. It now markets its Helios kick stage as compatible with Falcon 9, Starship, Terran R, New Glenn, Vulcan, Neutron, Eclipse, Ariane 6 and H3, sells plate-based rideshare on it, raised a $500 million Series D on 2 June, and on 8 July became the first upper-stage prime on-ramped to NSSL Lane 1. Momentus (Vigoride-7 on Transporter-16 with ten payloads, Vigoride-8 sold out for 2027) remains Transporter-dependent. D-Orbit, Exotrail, Starfish and Turion likewise reached orbit on Transporters in 2025 and 2026. Orbital transfer vehicles solve altitude, phasing and plane spacing; they do not solve inclination, because a 45 to 70 degree plane change from a Cape launch is beyond any chemical or electric tug in service. Sun-synchronous rideshare therefore still has to originate from Vandenberg, Kourou, Andøya, Sriharikota or Tanegashima, or accept a dogleg penalty from Wallops or the Cape.

The integrators are becoming inventory holders. Exolaunch, which EQT agreed to buy on 18 June (about $339 million per Dealroom, with revenue expected to rise 250% in 2026), holds Transporter capacity through 2028 plus its two dedicated Falcon 9s, and has flown on Electron and integrated five satellites for Isar's second flight. SEOPS launched Waymaker, a 2028 sun-synchronous Falcon 9 rideshare that was about 90% contracted by mid-August, then added a mid-inclination vehicle for early 2028; SEOPS now numbers the mid-inclination mission Waymaker-1 and the sun-synchronous one Waymaker-2. Its president, Evan Hoyt, has been quoted urging customers to plan 36 months ahead instead of 12. RIDE! bought 1,000 kg on the sun-synchronous Waymaker mission. Maverick is retailing half of Portal's 2028 Falcon 9. No US integrator has publicly booked Alpha, Eclipse, Nova, Terran R, New Glenn or Vulcan rideshare capacity, and no aggregator has announced a dedicated non-US sun-synchronous mission on Vega-C, Ariane 6 or PSLV since the freeze. The first such booking will be the real signal that replacement supply is being productised.

Confidence: high on vehicle status and records; moderate on prices where marked as estimates; moderate on the Neutron base case.

06The capacity model, 2027 to 2030

Adding up published payload figures gives one answer. Weighting them by whether the rocket will exist gives a different one.

Method

We model the mass of rideshare-class capacity, sun-synchronous and mid-inclination, that could plausibly be sold to Western third-party smallsat customers each year, by provider, in bear, base and bull cases. "Third-party" excludes anchor-tenant mass (Amazon Leo, NSSL, Lockheed Martin's Alpha block buy, Iridium replenishment, Starlink) and counts only what an unaffiliated operator could buy. For vehicles launching from the Cape or Wallops we apply a 25% penalty as a blended allowance for the dogleg to sun-synchronous orbit, since most of the demand is sun-synchronous. Rideshare shares per vehicle are assumptions, not disclosures, and are listed in Section 11 with their sensitivity. SpaceX-sourced capacity for 2027 and 2028 has two parts. The first is the remaining Transporter and Bandwagon manifest, assumed at 14 to 15 tonnes a year, about three-quarters of the 2025 level: bookings are honoured, and the "roughly half" that SpaceNews's sources described refers to the unsold slots now closed to new customers. The second is the five aggregator-bought Falcon 9s at about 6 tonnes each (4.5 in the bear case, 8 in the bull), above the 3 to 3.5 tonne historical fill of Transporter and dedicated Falcon 9 LEO missions because the integrators now sell stacks for satellites up to 1,400 kg. It is shown because it meets demand, though much of it is already sold.

Nameplate capacity is then converted to credible capacity by multiplying by two probabilities: that the vehicle is operating at the assumed cadence in that year, and that a given flight succeeds. For vehicles flying today with an established record the first is one and the second is the Laplace success estimate at the current record, (successes + 1) / (attempts + 2), with partial failures counted as failures. For vehicles with an early record the first is 0.7 to 0.9; for grounded vehicles it is below one. For vehicles that have not yet reached orbit the first is a readiness probability set from each programme's status (35% for Neutron in 2027 rising to 85% in 2030, 15% for Eclipse and Terran R in 2027 rising to 70%, and so on) and the second is 85%, the typical early-life reliability of a new vehicle once it is operating. This is a judgement-based measure; Section 11 shows how it moves when readiness is scaled down or up by 40%, and how it compares with a flat prior. It answers the question a customer signing a 2029 contract has to answer, which is how much of the promised capacity is likely to exist. Demand is the Section 3 estimate. The full provider-level table is in Section 11.

What the base case says

In 2027 the market is adequate on paper and tight in practice (Figure 2): 33 tonnes of capacity against 24 of demand, but 21 of the 33 are SpaceX vehicles already sold. The 12 tonnes of non-SpaceX supply are spread across Electron (2.5), Vega-C, Ariane 6 co-passenger slots, Alpha, a first Isar year, a first Nova year, and PSLV and RFA One flights that depend on a return to flight and a first flight respectively. In 2028 the aggregator Falcon 9s arrive and total capacity reaches 68 tonnes nameplate and 55 credible against 27 tonnes of trend demand. That apparent cushion is being consumed as we write: operators whose satellites will be ready in 2028 or early 2029 are booking the 2028 vehicles now, which is why SEOPS's sun-synchronous mission is 90% contracted and why 2028 demand will run above trend. After 2028 the SpaceX-sourced capacity ends.

In 2029 the base case shows 53 tonnes of non-SpaceX nameplate capacity against 29 tonnes of demand, an apparent surplus of nearly two to one. Credible capacity is 32 tonnes, between 23 and 39 depending on how quickly the new vehicles arrive: parity with base demand and short of the 39-tonne high case. The composition is the problem (Figure 3). About 61% of the 53 tonnes is on vehicles that have never flown: Eclipse (9), Nova (7), Terran R (5), Neutron (5), RFA One (2.7) and the smaller European and American vehicles. A further 9% is on vehicles that have flown but not reached orbit (Spectrum, Kairos, Hanbit-Nano, Eris), 12% on vehicles with fewer than five flights or without a majority of successes (Alpha, New Glenn, Vulcan, Vikram-1), and 4% on PSLV, which is established but grounded. Vehicles flying today with an established record (Electron, Vega-C, Ariane 6, H3) supply 7 tonnes, a quarter of base demand; with PSLV back and the early-record vehicles counted, about half. None of that established capacity carries a Transporter-sized manifest: Electron takes one to ten satellites at a time, Vulcan and New Glenn offer ESPA-class secondaries, and Vega-C's figure assumes a dedicated SSMS mission that is not yet manifested. By 2030 the base case reaches 77 tonnes nameplate and 49 credible, above demand if the new vehicles fly.

What the bear case says

The bear case is not exotic; it is the pattern of the last twelve months continued for two more years. Neutron's first customer flight slips into 2028 after a ground-test or first-flight failure; Eclipse, Nova and Terran R each lose a year; Isar and RFA reach orbit but fly two or three times a year rather than five; PSLV returns but flies little foreign mass until the root cause is disclosed; New Glenn and Vulcan serve their anchor tenants only. Non-SpaceX supply in 2029 is then 18 tonnes nameplate and 11 tonnes credible against 22 to 29 tonnes of demand (Figure 4): a shortfall of about 50 to 60% on the credible measure, and of nearly three-quarters against the high-demand case, in a year when the Transporter option no longer exists at any price.

What the model does not capture

Tonnage understates the problem in three ways and overstates it in three. It understates because parcel size matters: a 2029 market made of 700 kg Spectrum flights, 260 kg Vikram flights and 2.3 tonne Vega-C flights cannot carry a 140-satellite manifest the way Transporter-15 did, and each small flight carries its own integration schedule, licence and range slot. It understates because price matters: Section 7 shows every alternative above $7,000 per kilogram. And it understates because product matters: no vehicle in the table offers a scheduled, priced multi-customer sun-synchronous service today; Vega-C's SSMS, Ariane 6's MLS and PSLV's co-passenger slots are sold flight by flight around institutional primaries. The model overstates the problem by ignoring demand destruction and migration: the CubeSat tail that a price doubling pushes below break-even is large in count and small in mass, perhaps 10 to 15% of the tonnage, and operators that hedge by going dedicated (Kepler, Planet through Isar, BlackSky and iQPS on Electron) take their mass out of the rideshare pool; both effects sit inside the low-demand case. It overstates by ignoring re-timing: some 2029 payloads will fly in 2028 on the aggregator Falcon 9s, and Bandwagon-class payloads can move to mid-inclination slots that New Glenn, Vulcan and Eris serve without a dogleg. And it overstates by ignoring the roughly 150 payloads by which press tallies exceed our per-mission sum, which we take to be rideshares on non-dedicated missions, a channel that persists as long as Falcon 9 flies Starlink from Vandenberg.

Three developments would change the answer. A Rocket Lab or aggregator decision to run Neutron as a scheduled sun-synchronous rideshare from 2028 would lift Neutron's contribution from the 5 tonnes assumed in the 2029 base case to 15 to 30 tonnes, enough on its own to cover the high-demand case; nothing Rocket Lab has said suggests it, and its manifest suggests the opposite. A SpaceX decision to keep selling dedicated Falcon 9s to integrators past 2028, which the SatNews and SEOPS reading of the freeze allows and the Bloomberg reading does not, would add 18 to 30 tonnes a year (three to five vehicles at 6 tonnes) at roughly twice Transporter's price and convert the cliff into a price step; we put the odds of that reading at roughly 40%. And a Starship that reaches operational cadence would free Falcon 9 cores from Starlink duty before any Starship rideshare product exists. None can be assumed by a customer contracting today.

Confidence: moderate on the base case; moderate on demand; moderate on the composition finding, which moves between about 58% and 70% depending on how Neutron's rideshare share and the derivative vehicles (Vega-E, Alpha Block 2, Vikram-2) are classed. The finding that established, flying vehicles cover a quarter to a third of demand holds across those choices.

07Price and unit economics

The per-kilogram rate card is what made Transporter a market rather than a procurement. Its replacement will be dearer, and the increase lands hardest on the smallest satellites.

Three price bands emerge (Figure 5). At or below Transporter's $7,000 per kilogram sit only vehicles that a customer cannot buy by the kilogram today: a full Neutron to polar orbit with a barge landing, and a full Terran R. Between $7,000 and $16,000 sit a full H3-30, Ariane 62 and PSLV at whole-vehicle cost, Isar's target, Phantom's quote and, at the top of the band, the aggregator-bought Falcon 9s: a $70 million vehicle filled to 4.5 to 6 tonnes costs $12,000 to 16,000 per kilogram before the integrator's margin, which is why the SEOPS president has said Waymaker will cost more than SpaceX's rideshares did. This is also the band a Neutron rideshare would land in once an aggregator's margin, a partial fill and a return-to-site penalty are applied, and it is where we expect the 2029 market to clear if Neutron flies. Above $20,000 sit the vehicles that exist and fly today: Alpha, Vega-C as a whole vehicle, and Electron. The market's centre of gravity therefore moves from $7,000 to roughly $10,000 to 15,000 per kilogram in the base case, and to $20,000 or more if the new medium-lift vehicles are late.

What that does to a satellite business case

For a 100 kg Earth-observation satellite with an all-in mission cost of $5 to 10 million, a Transporter slot at $700,000 is 7 to 14% of the budget. At $12,000 to 15,000 per kilogram the launch is $1.2 to 1.5 million, or 12 to 30%. On a dedicated Electron at $8.5 million the launch is half to two-thirds of the mission and only works with a paying anchor, which is the BlackSky and iQPS model. For a 500 kg ESPA-Grande class satellite at $20 to 40 million all-in, Transporter's $3.5 million is 9 to 18% of the budget; the base-case band takes that to $6 to 7.5 million, or 15 to 37%. A doubling of launch cost is absorbable for government and defence programmes buying ESPA-class spacecraft; it pushes the marginal CubeSat, the university payload and the thin-margin IoT constellation below break-even. A tripling forces dedicated-launch economics that only anchored constellations can carry.

The quotes from the buyers are consistent with that arithmetic. RIDE!'s chief executive: "There is a clear asymmetry between the timeline for which satellites need to be launched and what the market can provide." Exolaunch's Jeanne Allarie: "We need a big European rideshare program and we need it now." Portal's chief executive, on buying a whole Falcon 9: "I needed to secure Portal's launch future for the next couple of years." None of them is waiting for Starship. The behaviour we observe is the opposite of an Osborne effect: operators are paying up for 2027 and 2028 certainty, and the only place a wait-for-Starship posture shows up is in bus design, where K2's Giga-class and Starcloud's 88,000-satellite filing presume Starship economics that do not yet exist.

One more consequence deserves a line because it is regulatory rather than commercial. FCC milestone schedules were set against Transporter-class pricing and availability. Amazon Leo, with 90-plus launches under contract at a reported cost approaching $10 billion, had 396 satellites in orbit against a 1,616 milestone on 30 July 2026 and received a waiver. Dozens of second-round NGSO licensees with far fewer resources were authorised on the assumption that cheap rideshare would be there through the decade. As SatNews put it in July, launch manifests are becoming spectrum policy.

Confidence: high on list prices; moderate on whole-vehicle and derived figures; the unit-economics ranges are illustrative.

08Reliability and insurance

Transporter's customers rarely bought launch insurance: on Falcon 9 the expected loss was too small to justify the premium and the administrative cost. The vehicles replacing it raise the expected loss sharply, and the capacity to insure it is, for now, thin.

The flight records

Figure 6 sets out the records as an underwriter reads them.

Falcon 9's record is the benchmark against which every replacement is priced: about 690 successes in 693 in-flight attempts, a current streak in the low 300s since the July 2024 Starlink upper-stage failure, and it is the reference point against which every other vehicle's rate is set. Four upper-stage anomalies in nineteen months (July 2024, September 2024, February 2025, February 2026) cost one payload and produced short groundings, which matter to schedule buyers but not to launch policies. Electron is the proven small-lift benchmark at 89 of 93. Vega-C and Ariane 6 are re-earning trust with five and seven consecutive successes respectively. Vulcan is certified but paused. PSLV, which had the third-longest record in the table, lost its proven status with two consecutive failures and an undisclosed cause. Everything else is new-vehicle risk: Alpha at three of seven, New Glenn at two of three, Spectrum, Kairos, Hanbit and Eris without an orbit between them, and seven vehicles with no attempts at all.

How the market prices reliability

Launch insurance is a spot market. Rates are finalised at binding, quoted per placement, and repriced across every open placement on a vehicle the moment it fails. The observable bands for launch-plus-one-year cover are 3 to 6% of insured value on Falcon 9 and Ariane in a soft market, 8% typical, and 12% or more for a hardened market or a new vehicle; Falcon 9 GEO rates rose from under 6% to about 10% during the 2023 loss year and have been falling since. As the SpaceNexus commentary put it in March, "a vehicle with 80 consecutive successes is rated very differently from one on its third flight," and Gallagher noted in the first quarter of 2025 that insurers were "doubling down on their technical assessments" of new vehicles "due to a perceived increase in risk for early adopters." We could not find a published maiden-flight rate or exclusion in any broker document. The empirical evidence is that Ariane 6's and New Glenn's first flights carried no insured commercial payload, and that an insured $30 million commercial satellite, AST SpaceMobile's BlueBird 7, flew on New Glenn's third flight. It became 2026's first insured launch-caused total loss when the upper stage under-performed. No broker states a rule; our reading of the proxies is that three to five consecutive successes is where capacity becomes broadly available, and ten or more where rates approach the proven band.

Gallagher's second-quarter 2025 update contains the sentence that explains the market's exposure to what is coming: "As with all lines of insurance, there is a compression of pricing such that the best (e.g. mature) risks pay more than their calculated technical failure rate would suggest, while newer, untested risks pay less than experience would indicate." Underwriters have been recovering on Falcon 9 and Ariane what they could not charge on new vehicles. A market in which Falcon 9 commercial business shrinks and new-vehicle business grows removes the subsidy and exposes the underpricing at the same time.

Why smallsat operators do not insure, and what changes

About 3% of LEO satellites carry insurance, up from 1% in 2021 (Gallagher, end 2023), and roughly 300 spacecraft in total carry comprehensive in-orbit cover. The economics explain it. A launch-only premium at 5 to 10% on a $2 to 5 million CubeSat is $100,000 to $500,000, comparable to or above the $350,000 Transporter ticket, and the administrative cost of underwriting a sub-$1 million unit exceeds the premium. Operators buy redundancy instead: spares, extra capacity on orbit and, until now, a launch vehicle whose 1% failure rate made the expected loss trivial. Where Transporter-class payloads have been insured they were ESPA-class: MethaneSAT (about $40 million claim in 2025; launched on Transporter-10), AstroForge's Odin ($3 million), Utilitysat ($30 million, 2024).

The arithmetic in Figure 7 is the core of the insurance argument. Moving from a 99% vehicle at $7,000 per kilogram to an 85% vehicle at $10,000 raises the sticker price 43% and the expected loss per kilogram sixteen-fold, from about $740 to $11,500. The reliability-adjusted cost of launch, price plus expected loss, goes from $7,700 to $21,500 per kilogram. No launch price recovers that gap at 85% reliability; a 95% vehicle would have to charge about $4,200 per kilogram (about $3,700 if a $2 million delay cost is included) to match Falcon 9's reliability-adjusted cost. The 85% case is kinder than the current record of every unflown vehicle in Appendix B, which is the point: it describes a new vehicle that is working. The arithmetic assumes the launch price is lost on failure; a reflight guarantee, which Rocket Lab, Firefly and Isar have offered, reduces the second term but not the first. Operators have three responses. Constellations will keep self-insuring, spread their manifests across two or three vehicles and hold spares, as AST SpaceMobile did when it flew BlueBird Block 2 satellites on Falcon 9 in June and August, within weeks of the New Glenn loss, while collecting its $30 million claim. Single-spacecraft ESPA-class operators, the EO, IoT and MicroGEO companies flying $20 to 80 million assets, are the most exposed and will seek launch-only cover on Neutron, Vega-C and Electron at 8 to 15% or more, and negotiate refund and reflight clauses. Lenders will re-underwrite: "if the insurance isn't available, then the finance isn't available," as Atrium's David Wade has put it, and a facility whose condition precedent is launch insurance at commercially reasonable rates can be breached by a change of vehicle.

The state of the market that has to absorb this

The space insurance market entered 2026 profitable and small. 2024 produced about $211 million of claims on roughly $580 million of premium. 2025 was tracking a net profit near $500 million and the lowest claims volume in almost a decade. Then SpainSat NG-2, insured for more than $400 million, was struck by a micrometeoroid in late December 2025; its operator disclosed the impact on 2 January 2026 and confirmed the satellite unrecoverable on 16 January. The notice of potential loss, received on 26 December, moved the 2025 loss ratio from about 15% to about 75%. Theoretical capacity per risk is $500 to 600 million; competitive capacity is nearer $300 million. Brit, Canopius, the Hiscox and Munich Re syndicates and Volante have left since 2023; Tata AIG, Phemis, Aesir, Whitecap, Hive and Relm have entered, mostly as small-line MGAs, and Howden opened a dedicated space practice in December 2025. Gallagher expects 2026 premium to "significantly contract" because few large GEO satellites are launching and constellation operators still do not buy, and it reports rate reductions available on prime business. In other words, the market has the balance sheet for a handful of $300 million towers and the appetite for softening, not for hundreds of $5 to 50 million placements on vehicles with two flights. The 2026 tower is 43% sourced from MGAs, and the number of senior space underwriters worldwide is well under a hundred according to Gallagher as cited by spacepolicies.org; bandwidth, not capital, limits a shift to many small policies.

Implications for underwriters and brokers

We found no underwriter or broker comment on the freeze as of 25 August; the absence is itself a finding. Four consequences follow from the analysis above. Demand for launch-only cover on ESPA-class spacecraft riding new vehicles will rise, and capacity will be offered selectively at spot rates (Gallagher's own statement is that small-satellite capacity "is heavily influenced by the risk profile and Sum Insured"), which favours brokers with vehicle-level technical assessments. The cross-subsidy problem becomes acute as Falcon 9 commercial volume falls, so rates on new vehicles must move toward actuarially fair levels or losses will follow. Product gaps open: multi-launch aggregate policies for constellation operators flying across three vehicles, launch-delay and manifest-slip cover for a market planning 36 months out, and slot-cancellation cover should SpaceX cancel any post-2028 commitment. And the single most useful data point for pricing an n-of-three vehicle, the rate paid on BlueBird 7's New Glenn placement, is not public; a broker with it holds an advantage. Vehicle-level reliability tracking, the reliability-adjusted cost of launch per kilogram, and manifest-level exposure by operator are inputs this market does not currently get from any single source; Off Earth Data intends to maintain all three alongside this report.

Confidence: high on market figures from broker reports; moderate on rate bands (no primary maiden-flight rate found); high on the expected-loss arithmetic given its stated inputs.

09Downstream effects on the satellite market

The freeze reprices launch, but its second-order effects run through manufacturers, integrators, investors and regulators, and they are already visible.

Operators: who loses and who is hedged

The exposure table in Section 3 ranks the operators. The pattern is that the companies with the fastest replenishment cycles and the least dedicated-launch history are the most exposed, and that the hedges announced so far are partial. ICEYE is the largest unhedged block: 38 satellites on SpaceX rideshares in two years, a production line doubling to 100 a year by 2027, a $520 million Series F and a EUR 1.7 billion German defence contract, and no announced launch contract with anyone but SpaceX. Spire's replenishment is entirely on Exolaunch-integrated SpaceX missions. Muon has thirteen satellites manifested to launch in the next twelve months and is building toward 500 satellites a year. Planet's Isar contract is a real hedge for Pelican but rests on a vehicle that has not reached orbit. Kepler's Neutron buy is the cleanest hedge in the group, though Kepler frames it as a choice for orbital control rather than a response to the freeze, and the mission is not before 2028. The government and academic tail, NASA's CubeSat Launch Initiative, the Space Test Program's SEOPS-integrated payloads, ESA's technology demonstrators and the university programmes on every Transporter, has no budget line for dedicated launch and no agency has said what it will do.

The operators that are structurally insulated are those that already fly dedicated: BlackSky and iQPS on Electron (iQPS has contracted eighteen Electron missions in total), Synspective on Electron, the SDA and NRO constellations on NSSL vehicles, and Amazon Leo on everything. Their advantage in 2029 is not price; it is schedule certainty and orbit choice, which Beck has long argued brings rideshare customers back to dedicated launch once they can afford it. Consolidation follows: Novaspace's 2026 outlook notes that "consolidation dynamics are also accelerating," Seraphim's index shows M&A running at four times the prior year, and the launch-plus-constellation integrators (SpaceX, Amazon, Blue Origin and now Rocket Lab with Iridium) are the ones that hold what Beck calls "the keys to space."

Manufacturers: inventory risk

The bus and satellite manufacturers sized their 2026 to 2028 production ramps against Transporter-class launch availability. Apex builds Nova buses ahead of orders and has raised three rounds in fourteen months; Muon is scaling from 50 to 500 spacecraft a year; NanoAvionics to hundreds of platforms by 2028; ICEYE to 100 a year; Loft has bought eleven buses for thirty satellites by end-2027; K2 plans constellation deployment in 2028. A 2029 launch gap converts finished spacecraft into working-capital inventory and pushes revenue recognition to the right. Novaspace's April forecast notes that "as vertical integration accelerates, independent suppliers face a narrowing addressable market," so the smaller-constellation channel is what remains, and that is the Transporter channel. None of the manufacturers surveyed had made a public statement on the freeze.

Integrators: from brokers to inventory holders

The most consequential structural change is happening in the channel. Until 2026 Exolaunch, SEOPS, Maverick and ISISpace were integrators and brokers: they bought Transporter ports, added dispensers and sold slots. Since May they have become owners of scarce inventory. Exolaunch's two Falcon 9s and its multi-year Transporter block are, on our reading, the asset EQT is paying for; SEOPS's sun-synchronous Waymaker mission was about 90% contracted within three months and the programme now has a second vehicle; Maverick is selling half of a customer's rocket. The integrator's margin rises with scarcity, the customer's planning horizon stretches from 12 to 36 months, and a secondary market for booked capacity is forming in the gap. Whether SpaceX rideshare contracts are transferable, whether deposits are refundable if SpaceX cancels a post-2028 mission, and whether integrators are marking up Transporter inventory are all questions we could not answer from public sources; they are the questions a buyer should be asking.

Orbital transfer vehicles: a lost ride and a new role

Every OTV operator reached orbit on a Transporter in 2025 and 2026, so the freeze removes their primary. It also creates their reason to exist. Impulse's Helios is explicitly designed to sit under New Glenn, Vulcan, Neutron, Terran R and Eclipse and to sell plate-based rideshare on primaries that fly to other orbits; Impulse has been building a GEO rideshare product with Exolaunch since 2024, when its president Eric Romo framed the question as "can we aggregate enough demand to make this work?" Bryce's forward view is that OTVs and dispensers "may further increase selection of medium to heavy launch vehicles." The catch, again, is inclination: an OTV can distribute forty satellites across altitudes and planes from a Neutron polar flight, but it cannot take a satellite launched from the Cape at 28 degrees into sun-synchronous orbit.

Investors

Capital was moving before the freeze became public. Seraphim's second-quarter 2026 index recorded $7.5 billion across 141 deals, the second-largest quarter on record, with the launch category up 162% year on year on Impulse's $500 million round (closed 2 June) and Isar's $311 million (9 June); Apex's $200 million (5 June) also fell inside the quarter, and the index, which ends 30 June, does not mention the freeze. Since then Muon ($250 million, 20 August) has closed and Stoke is raising about $1 billion at a $9 billion valuation. We found no company that cited the freeze in its own announcement; Payload framed Starcloud's $250 million extension on 24 August against "a looming launch shortage" as SpaceX "prepares to sunset the Falcon 9 rocket and pauses new bookings for rideshare missions." The venture view at the SpaceX IPO in June was that competing with SpaceX on launch was foolhardy; the freeze gives launch startups a demand narrative they did not have. The risk for investors is the composition finding in Section 6: the money is flowing to vehicles that have not flown, and the 2029 market they are being valued against depends on all of them flying.

Governments

The policy response has been fastest in Washington. The August presidential memorandum sets a target of 1,000 or more US launches a year by 2030 against a 2025 baseline of 193 launches, 85% of them SpaceX; as reported it does not mention rideshare, but the trade coverage of it placed the freeze and the integrators' Falcon 9 purchases in the same frame. NSSL Phase 3 Lane 1 now has seven providers after Impulse and Relativity were on-ramped on 8 July (the new entrants must each fly once before competing for task orders), and its ceiling has been raised to a reported $17 billion. The Space Force's $397 million SB-AMTI contract is a demand anchor for Neutron; its $266 million Kodiak award is for Electron-derived HASTE suborbital flights. NASA's launch-services and CubeSat programmes have said nothing. In Europe the Launcher Challenge predates the freeze and buys institutional missions rather than commercial slots; Arianespace's July rideshare announcement targets GEO from 2029. India transferred PSLV, LVM3 and SSLV production to an HAL-L&T consortium on 23 August with NSIL retaining commercial procurement, which is the right structure for a cheap sun-synchronous rideshare provider if PSLV's reliability returns. Japan's H3 has taken its first foreign commercial secondary. None of these is a Transporter replacement before 2029.

Confidence: high on the documented facts; moderate on the exposure ranking and consolidation reading.

10Scenarios and signposts

Four futures are consistent with the evidence, and they are not mutually exclusive. The next twelve months will tell us which ones we are in.

The scenarios are not exclusive: C can combine with A or B, and D with any of them. We do not assign numerical weights to A, B and D, because their joint probabilities cannot be derived from the evidence; C carries the weight given to the rideshare-only reading in Section 2. The "OED view" column states the reasoning. What is not a matter of judgement is that Scenario D does not rescue 2028 or 2029 for anyone whose satellite is ready then, and that Scenario A depends on a commercial decision Rocket Lab has not made.

Signposts

Confidence: moderate that B is the most likely path; low on the probability assigned to C; high that the signposts listed will resolve most of the uncertainty within twelve months.

11Method, confidence and data notes

Sources and method

This report was built from primary and trade sources retrieved between 20 and 25 August 2026. They comprise company releases and investor materials (Rocket Lab, Firefly, SpaceX's prospectus and 10-Q, Exolaunch, SEOPS, Isar, Avio, Arianespace, Stoke, Blue Origin, ULA, Impulse, ICEYE, Spire, Planet, Kepler, Muon, Satellogic, HawkEye 360); regulatory documents (FAA and Department of the Air Force environmental decisions, 14 CFR Part 440, the ESA Launcher Challenge); broker market reports (Gallagher's quarterly space updates 2023 to 2026, Seradata via SatNews, Insurance Insider); industry datasets (BryceTech Smallsats by the Numbers 2025 and 2026, Novaspace's 2026 smallsat forecast and outlook, Seraphim Space Index Q2 2026); and launch coverage (Spaceflight Now, NASASpaceflight, European Spaceflight, Via Satellite, Payload, SatNews, Space.com, TechTimes). SpaceNews, Bloomberg, AIAA and Ars Technica could not be read directly and are cited as relayed by other outlets, marked as such. Every sourced numerical claim was traced to its source by a research analyst and re-checked by an independent verification pass; estimates and assumptions are identified as such in this section and in the tables. Where two sources disagree we say so in the text or below.

The OED graph (pipeline run 25 August 2026; 3,254 tracked entities, 3,185 scored) supplied SEPI scores, activity records and the spaceport registry. SEPI is a seven-factor composite (government contracts 22.5%, regulatory 18%, technical milestones 18%, funding 13.5%, talent 10%, partnerships 9%, sentiment 9%) blended with a Bayesian prior; scores shown here are on the one-year window and are accompanied by the count of observed factor families so that prior-dominated scores can be recognised. Methodology at www.offearthdata.com.

Model assumptions

Demand: the Western third-party rideshare-class market (sun-synchronous and mid-inclination) is put at 22 tonnes in 2025 and 2026: about 20 tonnes on SpaceX's dedicated rideshares (436 payloads on six 2025 missions; Novaspace measured 3.2 tonnes a mission for 2021 to 2023 and recent manifests are heavier; our range is 17 to 25 tonnes) plus about 2 tonnes already flying as co-passengers on Vega-C, PSLV, H3 and Electron. From 2027 mass grows 10% a year: 24, 27, 29 and 32 tonnes. The low case is 25% below (18, 20, 22, 24) and absorbs demand destruction and migration to dedicated launch; the high case is 35% above (32, 36, 39, 43) and reflects the two-year queue under which the 2025 throughput was flown. Supply: per-vehicle flights per year in bear, base and bull cases from company guidance less a typical twelve-month slip, times payload to sun-synchronous orbit, times an assumed third-party share (Alpha 35%, Eclipse 25%, Terran R 10% of 17.6 tonnes, or 1.8 tonnes a flight, Nova 40%, New Glenn and Vulcan secondary slots of 0.5 to 2 tonnes per host flight, Ariane 6 co-passenger slots of 0.5 to 1 tonne, Vega-C co-passengers on two institutional flights a year plus one dedicated SSMS every two years, PSLV and SSLV about 1 tonne in 2027 rising to 2 tonnes by 2029 as co-passengers of 400 to 500 kg return on three to four flights plus SSLV, Neutron sold as dedicated missions with rideshare-class sales of a fraction of a flight's mass a year from 2028, rising to about one flight in 2030). The share assumptions have no disclosure behind them; halving or doubling the Eclipse, Nova and Terran R shares moves 2029 base nameplate by about minus 11 or plus 21 tonnes. Cape and Wallops launches take a 25% sun-synchronous penalty. Credible capacity multiplies nameplate by a readiness probability (one for established vehicles flying today; 0.7 to 0.9 for vehicles with an early record; 0.6 to 0.9 for grounded PSLV; 0.1 to 0.85 for vehicles without an orbital success, set from programme status) and by per-flight reliability (the Laplace estimate for vehicles with at least one success; 85% for the rest; Astra and Epsilon S at 75% for partial heritage). Scaling the readiness probabilities of vehicles without an orbital success by 0.6 and 1.4 gives the 23 to 39 tonne range around the 32-tonne 2029 base; the older flat-prior method gives 22, 30 and 37 tonnes at priors of 30%, 50% and 70%. China and Russia are excluded as legally inaccessible.

Supply figures are tonnes of rideshare-class mass plausibly sold to Western third parties. Readiness is the assumed probability that the vehicle is operating at the modelled cadence in each year; reliability is the per-flight success probability applied to it. Pair bear supply with high demand for the stress case and bull supply with low demand for the benign case. The Section 5 table pools the smallest launchers into one row; the model carries Astra, Phantom, Epsilon S and Kairos separately and pools the rest.

Known disagreements between sources

• Cut-off wording: "beyond late 2028" (SatNews, Bloomberg) versus late 2028 or early 2029 and beyond (SpaceNews's sources, as relayed). SpaceX has confirmed neither.
• Scope of the freeze: rideshare only (SatNews, SEOPS) versus all new commercial Falcon 9 bookings beyond 2028 (Bloomberg). Modelled as Scenario C in Section 10.
• Cumulative rideshare count: about 1,650 by per-mission sum versus press reports of "over 1,800" after Transporter-17, whose basis is not published.
• 2023 rideshare rate: Payload reported $6,500/kg; the listed rate for 2023 to early 2025 was $6,000/kg. SatNews's "sub-$300,000" ticket in July 2026 is stale against the $350,000 reported by four other sources. The August 2019 launch price is given as $2.25 million (Teslarati) and $2.5 million (illdefinedspace).
• Bloomberg's "halted production of some non-reusable Falcon hardware" cannot mean all expendable upper stages while Falcon 9 flies three times a week; we read it as some hardware lines.
• Waymaker numbering: the May 2026 release called the sun-synchronous 2028 mission Waymaker-1; SEOPS's site and Payload's August coverage now number the mid-inclination mission Waymaker-1 and the sun-synchronous mission Waymaker-2. This report refers to the missions by orbit. Fill: Payload reports about 90% of the sun-synchronous mission's mass contracted by mid-August; TechTimes reported in July that the manifest filled within three weeks with about 30 customers waitlisted. The two reports are not fully consistent and we use Payload's.
• Portal's Starburst-1 is manifested on Bandwagon-5 (Payload, August 2026) after earlier reporting placed it on Transporter-18.
• SpaceX's 100th launch of 2026: 18 August (SpaceDaily) versus 22 August UTC for the 100th orbital mission (Spaceflight Now); the counts may treat Starship flights differently.
• Starship Flight 12 and 13 dates: 22 May and 24 July (Space.com live coverage, TechTimes, the OED launch feed) versus 26 May and 26 July in one summary; we use the former.
• Ariane 6 flight count: eight flights to 17 June 2026 (Arianespace and European Spaceflight, including Galileo L14 on 17 December 2025); one insurance-stream tally omitted that flight and gave seven. Appendix B uses eight.
• H3 return to flight: 11 June 2026 UTC (NASASpaceflight) versus 12 June in the OED feed. Innospace's suborbital test termination: 19 August versus 20 August; one tally counts Hanbit-Nano's December 2025 attempt, another does not. Starship Flight 14: NET 28 August per a 20 August report, September per the launch trackers.
• Firefly Alpha capacity: 1,030 kg LEO / 630 kg SSO (company site) versus 1,170 / 745 kg (likely Block 2). Alpha's ~$15 million price is widely reported but not on a primary source. Firefly's IPO proceeds: $868 million versus $998.6 million including the overallotment.
• PSLV-C62's lost co-passengers: about 18 (NASASpaceflight) versus 16 satellites or 12 commercial payloads in other tallies. Skyroot's Vikram-2 is given as 900 kg (company) and 1,000 kg (press).
• New Glenn's explosion is dated 28 May by Blue Origin, Spaceflight Now and NASASpaceflight; one Space.com article says 8 May. Blue Origin's reported $10 billion round at a $130 billion valuation (July 2026) rests on a single mention.
• Space insurance 2024 result: Seradata's profitable year ($211 million claims) versus a consultancy's "155% loss ratio"; we use Seradata. 2025 premium: $550 to 600 million (Seradata) versus $675 million (Space Intel Report). SpainSat NG-2: the notice of loss (26 December 2025) predates the operator's disclosure of the strike (2 January 2026); the strike itself occurred in late December.
• Falcon 9's consecutive-success streak is in the low 300s since July 2024 (our estimate); the "400 consecutive" figure sometimes quoted conflates total flights with the streak.
• HawkEye 360's May 2026 IPO: $435.9 million net (company) versus $416 million in one funding tracker.

OED data notes

The OED activity graph carried the Transporter-15, 16 and 17 launches, the Transporter-18 and Bandwagon-5 placeholders, and the Kepler, Portal and Rocket Lab news items used here, and its spaceport pipeline feed supplied the Starship environmental record. Its 2025 launch coverage and its FAA licence feed are incomplete for the vehicles in this report, no 2027 manifest is yet loaded, and the July 2026 freeze itself has no event row; those gaps are logged for the ingestion queue and did not affect the figures above, which were taken from primary sources. SEPI scores in Appendix C should be read with their observed-factor counts: several European and Asian launch startups are scored on the prior alone.

Appendix ASpaceX dedicated rideshare mission log

Per-mission counts from launch-day coverage. Sum of flown missions: 1,546 on Transporter (using 90 for T-9), 62 on Bandwagon, 40 on Twilight. Cadence by year: 2021, two; 2022, three; 2023, four; 2024, four; 2025, six; 2026 to date, three.

Appendix BFlight records

Appendix COED SEPI snapshot, launch providers and rideshare-dependent operators

One-year window, pipeline run 25 August 2026. "Factors" is the number of the seven factor families with observed evidence in the window; a score with zero or one observed factor is prior-dominated and should not be compared with a fully evidenced score. Tier 1 is Established, 2 Growth, 3 Early.

Appendix DPrincipal sources

1. SatNews, "SpaceX limits future commercial rideshare bookings past 2028 amid severe manifest saturation," 20 Jul 2026. satnews.com
2. AIAA, "SpaceX halts new commercial rideshare reservations beyond 2028 as launch schedule fills," 21 Jul 2026 (headline only; page not retrievable). aiaa.org
3. Bloomberg, "SpaceX is turning away Falcon customers in major bet on Starship," 23 Jul 2026 (paywalled; quoted via Talk of Titusville, 24 Jul 2026, and social-media excerpts). bloomberg.com
4. SpaceNews, "Small satellite operators confront a bottleneck to space access," 25 Jun 2026, and "SpaceX launches Transporter-17 amid concerns about rideshare program's future," 7 Jul 2026 (not retrievable; cited via TechTimes, SpaceQ and a Japanese industry summary at note.com). spacenews.com
5. SpaceQ, "SpaceX's rideshare freeze: what it means for the smallsat market and Canada," 6 Aug 2026. spaceq.ca
6. Via Satellite, "Launchers cite busy manifests and scarcity as commercial demand grows," 24 Mar 2026; "Rideshare providers SEOPS and Exolaunch buy dedicated Falcon 9 missions," 26 May 2026; "Will the European launch market finally ignite?" 27 May 2026; "SpaceX's Transporter-17 mission launches 81 payloads," 7 Jul 2026; "Skyroot Aerospace makes history for India with first launch success," 20 Jul 2026; "White House targets support for 1,000 launches per year by 2030," 21 Aug 2026; "NanoAvionics upgrades its flagship MP42 bus line," 24 Aug 2026; "New Glenn grounded and AST SpaceMobile satellite lost after launch anomaly," 20 Apr 2026; Deville, "5 years of SpaceX rideshare missions: the spoils of monopoly," 21 Aug 2024. satellitetoday.com
7. Payload, "SEOPS doubles down on Waymaker dedicated rideshare program," 17 Aug 2026; "Portal books dedicated Falcon 9 amid constrained launch market," 20 Aug 2026; "Starcloud announces $250M Series A extension," 24 Aug 2026; "Muon Space raises $250M Series C," 20 Aug 2026; "Transporter-16 sends 119 payloads to orbit," 30 Mar 2026; "Rocket Lab enters Europe's crowded launch market," Aug 2026; "Impulse is hunting for GEO rideshare customers," 6 Aug 2024. payloadspace.com
8. Space.com, "Goodbye Falcon 9? SpaceX planning to retire workhorse launcher when Starship is up and running," 24 Aug 2026. space.com
9. Teslarati, "Elon Musk just set a condition that could end Falcon 9 as we know it," 24 Aug 2026 (Musk post of 22 Aug). teslarati.com
10. TechTimes, "SpaceX Transporter-17 delivers first commercial nuclear CubeSat, 1,800-payload milestone," 7 Jul 2026 (relays SpaceNews; RIDE! quote; Waymaker waitlist), and "Vietnam signs first private satellite deal before SpaceX closes manifest," 11 Aug 2026. techtimes.com
11. SpaceX, Form 10-Q for the quarter ended 30 Jun 2026 (filed 4 Aug 2026) and Form 424B4 prospectus (12 Jun 2026); Q2 2026 earnings call transcript, 4 Aug 2026 (Motley Fool). sec.gov
12. Rocket Lab, Q2 2026 results and call (transcript via Investing.com), 10 Aug 2026; Q4 2025 results, 26 Feb 2026; Neutron test update, 21 Jan 2026; Neutron Payload User's Guide v1.0, Jan 2025; Kepler Neutron agreement, 10 Aug 2026; SB-AMTI award, 4 Aug 2026; Iridium acquisition, 29 Jun 2026; 93rd Electron mission, 20 Aug 2026. rocketlabcorp.com
13. Spaceflight Now, "Window for 2026 launch debut of Rocket Lab's Neutron rocket is narrowing," 10 Aug 2026; "Firefly Aerospace pushes debut of Alpha Block 2 to Q4 2026," 11 Aug 2026; "Blue Origin's New Glenn explodes during prelaunch testing," 29 May 2026; "ULA confirms successful SRB test as Vulcan anomaly investigation continues," 14 May 2026; Transporter launch coverage 2021 to 2026. spaceflightnow.com
14. NASASpaceflight, Terran R engine integration (Aug 2026), Stoke update (19 Aug 2026), Blue Origin update (Jul 2026), Firefly Block 2 and Eclipse progress (16 Aug 2026), H3 return to flight (11 Jun 2026), Zhuque-3 (18 and 23 Aug 2026), PSLV third-stage failure (13 Jan 2026), Isar "Onward and Upward" (15 Jun 2026). nasaspaceflight.com
15. European Spaceflight, European launch schedule 2026 (19 Aug 2026); ELC funding (30 Nov 2025); Ariane 64 Amazon Leo flights (Feb, Apr, Jun 2026); MaiaSpace 2027 (2 Mar 2026); RFA One NET 10 Aug (7 Jul 2026); ESA Sentinel-1C Vega-C cost (27 Apr 2026). europeanspaceflight.com
16. ESA, European Launcher Challenge; Arianespace Ariane 6 MLS user's manual; Avio FY2025 and Q1 2026 results; Isar Aerospace Series D, 9 Jun 2026; Isar-Planet contract (Via Satellite, 2 Jul 2026). esa.int
17. Exolaunch releases: capacity through 2028 (20 May 2025); Exo-1 and Exo-2 (26 May 2026); Transporter-17 (7 Jul 2026); Transporter-18 and Bandwagon-5 manifest (25 Aug 2026). EQT, "EQT to acquire Exolaunch," 18 Jun 2026. SEOPS Waymaker release, 26 May 2026, and seops.space/waymaker. exolaunch.com
18. BryceTech, Smallsats by the Numbers 2026 (7 Aug 2026) and 2025; Novaspace, "Global smallsat deployment accelerates with 16,900 satellites projected through 2035," 23 Apr 2026, and 2026 Outlook; Seraphim Space Index Q2 2026. brycetech.com
19. Gallagher Space Market Updates Q4 2023 through Q1 2026 and "Regulatory developments in the space race," Aug 2026; SatNews/Seradata, "Space insurance: promising start to the year marred only by MethaneSAT claim," 17 Aug 2025; Insurance Insider, "SpainSat loss jolts space insurance market after brief return to profit," Jan 2026; SatNews, "Launch manifests are becoming spectrum policy," 29 Jul 2026; SpaceNexus, "Space insurance: how underwriters assess launch and in-orbit risk," 21 Mar 2026; SDxCentral, "Paying the premium: why 2023 was a bad year for space insurance," 2024; OrbitalRadar and Eudaimon Consulting rate-band commentaries, 2026; spacepolicies.org, "Space insurance markets in the mega-constellation era"; 14 CFR Part 440. specialty.ajg.com
20. FAA, Starship at LC-39A Final EIS and Record of Decision (Federal Register, 3 Feb 2026); Department of the Air Force, SLC-37 Record of Decision (Federal Register, 18 Dec 2025) and Vandenberg Falcon Record of Decision (Space.com, 17 Oct 2025); Rep. Carbajal letter on SLC-14, 12 Jan 2026. faa.gov
21. ICEYE, Spire Global, Planet Labs, HawkEye 360, Satellogic, Muon Space, Kepler, Apex, K2 Space and Impulse Space releases and SEC filings, June to August 2026, as cited in the text. Advanced Television, "Arianespace to offer rideshare launches," 9 Jul 2026. Wikipedia launch lists for vehicle flight histories, cross-checked against launch coverage. Off Earth Data graph, pipeline run 25 Aug 2026.