One launcher, one bottleneck
The month had fewer launches than usual but mattered more than its count suggests: Starship reached orbit for the first time, SpaceX published the first full set of numbers showing where its money is made, and every challenger's schedule slipped further.
249 attempts year to date, exactly the 2025 pace. The US is 10 launches behind (147 vs 157), China 9 ahead (73 vs 64). SpaceX is down 16 launches yet still carried 78% of payload mass.
Flight Atlas · 23 OctFlight 14 on 28 Sep made the first Starship orbit, deorbit burn and controlled splashdown, but an early Raptor vacuum shutdown cut it to about 3 of a planned 10 hours. Flight 15, a two-stage catch attempt, is announced for November.
Industry-changing eventNeutron moved to 2027, New Glenn has been grounded since 28 May, and China's two recovered stages have not flown again. Falcon 9's list price works out at about $3,246/kg, 19% above 2018.
Prices not fallingQ3 results show 13.6M subscribers at a $69 ARPU and Connectivity operating income of $2,367M — a 45.1% margin and about 85% of group operating income — while the Space segment lost $7M on Starship R&D. Amazon Leo has more than 1,000 satellites built but only 396 in orbit.
SpaceX Q3 2026 · reported 20 OctLaunch and in-space transport took $1.82B of $2.69B (67%) in the September cohort. The window caught the tail: Hanwha joined Impulse Space's $308M extension, Infinite Orbits raised €52M and booked four Impulse flights. The Space Force reset SCAR as a commercial-services buy.
~$521M disclosed in window“One month, one pattern: launch supply is concentrated in one company, and every other part of the industry is now organised around getting access to it.”
Divrite Research synthesis of launch databases, SpaceX's Q3 release and New Market Pitch funding data
Launch statistics: flat count, shifted mix
2026 YTD vs same date 2025
The decline is almost entirely SpaceX (122 vs 138), while CASC is flat at 48 — so China's gain of 9 comes from its commercial providers.
Source: Flight Atlas (Launch Library 2 checked against GCAT), by country of the rocket; country data as of 21 Oct, provider data as of 23 Oct 2026.
The window, 23 Sep – 22 Oct
Launches in the last 7 days (365-day average)
Launches 28 Sep – 23 Oct across two databases, about 11 American and 3 Chinese
Reach-orbit rate; 6 failures and 1 partial year to date, no new failure visible in the window
Payload to orbit YTD vs 2,248 t a year earlier
Below average. At 6.3 a week, a typical 30 days holds about 27 launches. Friction showed up as slips, not losses.
The West Coast overtakes Florida
Vandenberg launches by 15 Oct — one short of California's record of 71
Space Coast launches by 20 Oct, against the 2025 record of 109
Space Coast orbital launches by year
Straight lines put the Space Coast near 75 for the year and Vandenberg near 89; Space Force officials expect 80 or more from Vandenberg (estimates). The record-tying Starlink 15-15 was scrubbed three times and moved to 23 Oct.
Sources: Orlando Sentinel launch log (20 Oct); Vandenberg SFB
China's growth comes from outside CASC
Chinese launches YTD, up from 64 at this date in 2025
Share flown by providers other than CASC (25 of 73)
Chinese launches by provider, 2026 YTD
Galactic Energy (4), ExPace (3), Orienspace (2) and Space Pioneer (1) make up the rest. A straight-line projection of about 91 puts China level with, not clearly above, its 2025 record of 92 (estimate).
Source: Flight Atlas, as of 23 Oct 2026
Major missions: one flight outweighs the rest
Starship Flight 14 — the first orbital Starship
Starship can now relight in space, reach orbit, release a payload and bring itself back down under control. All 13 earlier flights were deliberately suborbital; SpaceX's Q3 results call this its first revenue-generating Starship mission.
~10x downlink capacity
of one Falcon 9 Starlink launch, by SpaceX's filed figure; a spokesperson put it at about 20 Falcon 9 flights.
~43% of the stated load
V3 flights could later carry up to 60 satellites; three on this flight filmed the ship's heat-shield tiles.
Engines are the open problem
Reliability matters for NASA's plan to land astronauts with a Starship variant in 2028.
Engine performance by phase
SpaceX summary| Phase | Engines ran | Result |
|---|---|---|
| Booster ascent | 33 of 33 lit; 1 shut down in climb | Ascent and hot-staging completed |
| Boostback burn | 31 of 33 | Designed to use up remaining LOX and test performance limits |
| Booster landing burn | 11, then 5, then 3 of 13 | Flight termination system fired as a demonstration |
| Ship ascent | 6 lit; 1 Raptor vacuum shut down early | Other 5 burned longer to compensate |
| Ship burns in space and landing | All ran (1, 1, then 3 sea-level Raptors) | First-ever Starship deorbit; on-target splashdown |
Engines were the weak point. The booster ran short of engines in every burn after liftoff, and the ship's early shutdown cut the mission short. All three V3 boosters so far have had engine-relight problems on descent.
Cut short by caution
orbits planned, flight of just under 10 h
orbit reached with two burns
no stage recovery attempted
NASA crewed lunar landing plan
“Out of an abundance of caution”, controllers cut the time in orbit; the ship came down north of Hawaii, one of two pre-agreed zones, rather than off Chile as planned. CNN reported a harder touchdown than Flight 13's.
Planned timeline, brought forward
The pre-flight plan put the deorbit burn at about T+8:52 and entry at T+9:29; milestones after payload deploy were brought forward when the flight was shortened.
Other missions that mattered
Oct 1 · Falcon 9 and Falcon Heavy
Crew-13 and NROL-97
Crew-13 slipped from 12 Sep after a propulsion leak on Dragon Grace; Crew-12 splashed down on Oct 8.
Oct 9 · Falcon 9
21 Space Development Agency data-transport satellites
Launched after an Oct 4 scrub; half of 126 planned Transport Layer satellites were already in orbit.
Oct 9, 19:27 UTC · Long March 12
27th group of low-orbit internet satellites
Preset orbit from Hainan's commercial site; Xinhua named neither operator nor satellite count.
Oct 13, 10:33 UTC · Falcon 9 B1096 + Dragon C211
CRS-35 to the ISS
Booster's 8th flight, landed at LZ-40; capsule on its 4th flight.
Oct 14, 11:40 UTC · Zhuque-2E Y4
10 satellites to sun-synchronous orbit
8 ADAspace “Star Compute” satellites plus Guangchuan-03 and -04 laser-communication testbeds.
Oct 15 · Falcon 9
Classified U.S. Space Force payload
Vandenberg's 70th launch of 2026.
Oct 20, 11:04 UTC · Falcon 9 B1085
Bandwagon-5: Starfall plus 31 payloads
Booster's 19th flight; Exolaunch arranged 20 of the 31, including HawkEye 360, Iceye and Xona satellites.
Why they matter
More than 6,300 lb of cargo and the fourth and final pair of roll-out solar arrays, each above 20 kW for over 10 years. Installation still needs spacewalks; Dragon leaves in mid-November.
The 2,100 kg uncrewed capsule carries up to 1,000 kg. Reentry was not publicly confirmed, yet SpaceX has already signed three commercial Starfall missions.
Jiuquan settled a conflict with a Long March 4 in LandSpace's favour, a first. ADAspace targets 2,800 computing satellites and now has 20 in orbit.
Reusable rockets: recovered is not reflown
Reuse status as of 23 October 2026
Recovering a stage is no longer rare; flying it again at rate still is. Price per kilogram falls only when a second vehicle offers reflown capacity at rate — and without that competition, SpaceX keeps the savings.
Design life
Falcon 9 booster up to 40 flights; New Glenn booster up to 25; Zhuque-3 stage up to 20 uses.
The window keeps widening
Neutron's slip and New Glenn's grounding extend the period in which SpaceX has no reflown rival.
Flight 15 (Booster 22 / Ship 42) will try to catch both stages — potentially the first fully recovered orbital launch vehicle. All three V3 boosters had engine-relight problems on descent; Musk calls success “more likely than not”.
SpaceX also seeks FAA approval for up to 456 launches and 912 landings a year at Starbase.
Grounded since the 28 May pad explosion, traced to a BE-4 main oxygen valve. Bezos aims to fly before year-end; Reuters reports early 2027. More than 1,000 workers are rebuilding LC-36A.
A $330M factory expansion is to make up to 150 upper stages a year from summer 2027.
Inspectors found voids in the carbon-fibre layers of a newly cured forward propellant dome. A process fix produced Rocket Lab's “highest quality” aft dome; stage tests run into Q1 2027.
Fourth public slip: 2024 → 2025 → early 2026 → late 2026 → 2027.
LandSpace aims to refly the Zhuque-3 booster within six months of its 19 Aug land landing. Zhuque-3 burns methane and lifts up to 18,300 kg to LEO; Long March 10B carries up to 16 t.
At least six other Chinese firms are pursuing recoverable stages.
Reuse lowered cost, not price
$/kg to LEO from list prices (SpaceNexus, as of 11 Oct 2026)
List prices, not negotiated contract values. Sources disagree on reused Falcon 9: $1,500/kg (orbital-intel) vs about $2,700/kg (Orbital Radar). SpaceX does not publish its internal cost.
Falcon 9 per kg vs about $2,720/kg in 2018 ($62M list)
Rideshare, from $5,000/kg (2019–21) to $7,000/kg (Feb 2026)
LandSpace long-term target per kg, about 14% below Falcon 9; ~$51M a full Zhuque-3 flight (estimate)
Hypothetical Starship at $74M for 100 t; it has no published price
Satellite internet: Starlink turns launch into margin
Starlink: volume beats price
The base nearly doubled (+89% year on year) while the average bill fell by about a sixth (−17%).
Source: SpaceX Q3 2026 results release, quarter ended 30 Sep 2026, published 20 Oct 2026; margins computed.
Q3 2026 operating detail
Net consumer additions in one quarter, mainly from developed markets
Starlink Mobile (direct-to-cell) monthly users, +37% quarter on quarter
Aircraft equipped, of about 12,000 under contract
Connectivity revenue (+69%), above the $4.84B consensus
Not all repeatable. The 55% gross margin includes a one-time benefit of about 3 points, mainly from tariff refunds.
SpaceX Q3 2026: launch breaks even, Starlink earns
Revenue $14,575M · +161% · net income $4.4Brevenue, +6%
operating income
A 75.4% gross margin is used up by R&D of $1,225M (+44%) as Starship spending rose. 35 launches vs 44; about $94M per customer launch, up from $67M (estimate).
revenue, +69%
operating income, 45.1%
About 85% of group operating income on 36% of revenue. Enterprise and government revenue rose 91% on Starshield and aviation demand; capex was $1,543M.
revenue, +830%
capex, of $19,593M
The largest revenue line and the main user of capital (about 84% of capex). Includes $608M from Cursor, consolidated from 14 Aug; operating income $431M.
China is short of rockets
China's two networks hold about 480 satellites against roughly 10,074 Starlinks.
more Guowang needed to hit its 310 target in ~10 weeks
more Qianfan needed to hit its 324 target
Qianfan added only 10 satellites in the last 90 days by Orbital Radar's estimate; each batch carries 18 on expendable Long March rockets, against up to 60 per Falcon 9.
Qianfan satellites launched per year (KeepTrack)
At its recent rate, the 1,296-satellite Phase 1 finishes beyond 2035 (Orbital Radar).
Amazon Leo: built, not launched
More than 1,000 built in Kirkland, at up to 27 a week, but 396 in orbit.
396 of 1,618 (24.5%) by the 30 July deadline; Amazon is seeking regulatory relief.
Prime Video, eero, Fire TV and Ring; beta in five countries expected late 2026 or early 2027.
AST SpaceMobile: carriers, not cadence
The ~45-satellite target moved to early 2027; BlueBirds 14–16 shipped on 30 Sep with no launch date.
Deutsche Telekom, Orange, Telefónica and Vodafone signed on 19 Oct to explore a joint bid.
Musk reportedly ordered Starlink capability over Ukraine limited on 14 Oct (WSJ).
Funding and contracts: paying for the bottleneck
September 2026 financing by segment
Launch and in-space transport took $1.82B of $2.69B across 12 financings; pure Earth observation raised $117M (4%).
Source: New Market Pitch, last updated 5 Oct 2026; its aggregates are a reconstruction (estimate).
Government: buying services, not systems
Space Force and MDA reopened the programme as a Commercial Solutions Opening, after cancelling a $1.4B contract for 12 phased-array antennas. White papers were due 15 Oct; awards are planned for early 2027.
SSC contract to design and demonstrate ultra-compact RF terminals for a space-based data layer; no value disclosed.
Commercial: launch slots as assets
Four firm slots through 2029 on Impulse vehicles flying on Falcon 9 — “the company's most precious asset”, per its chief executive. First servicer deployment in late 2027.
99.6% of votes cast in favour; closing expected by mid-2027, pending regulatory approvals.
Round sizes reset upward
Median September financing; a $100M round “no longer automatically” marks a company as exceptional.
2026 surge (tracker estimate)
Space-tech funding to date per Tracxn, against $9.28B for all of 2025 (about 2.5x).
Softer start to Q4?
Share of September's capital that fell inside this window; in-window rounds total about $521M.
Next month's launch calendar
Almost nothing outside SpaceX's Vandenberg cadence carries a hard date. November's two marquee events are “no earlier than” targets, and two heavy-lift returns have already slipped out of the year.
23 Oct – 30 Nov, with December flags
Falcon 9 · Starlink 15-15, 27 satellites
Vandenberg; B1063 on its 36th flight; scrubbed 20, 21 and 22 Oct.
Falcon 9 · Starlink 15-16, 27 satellites
Vandenberg; with 15-15, California's 71st and 72nd launches — one sets a state record.
FAA comments close on LC-36A cadence (12 → 50)
The virtual public hearing took place on 22 Oct.
Falcon 9 · NROL-195
Vandenberg, per SpaceNexus as of 21 Oct.
Starship Flight 15 · Booster 22 / Ship 42
Starbase Pad 2; first ship catch and first V3 booster catch. A reported 8 Nov date is unofficial.
Falcon 9 · Starlink 15-21 and Cygnus NG-25
Vandenberg Starlink; ISS cargo from Cape Canaveral SLC-40.
Ariane 64 · VA271, 36 Amazon Leo satellites
Guiana Space Centre ELA-4; about 2 h 4 min from liftoff to final deployment.
Ariane 62 MetOp-SG-B1 · Vega C VV31 (IRIDE)
Both planned from Kourou.
Falcon Heavy · Astrobotic Griffin lunar lander
CLPS mission from the Space Coast; date TBD.
New Glenn 7×2 · return to flight with Blue Ring
Cape Canaveral LC-36; the only route to a third heavy launcher for Amazon this year.
Atlas V Starliner-1 · Long March 10A Mengzhou-1
Both uncrewed; Mengzhou-1 is reported slipping to early 2027. Vulcan Leo 1 (44 Amazon Leo satellites) remains undated.
When can Flight 15 fly?
17 Nov – 1 Dec: Musk's “four to six weeks” from 20 Oct.
30 Nov: Pad 2's fastest turnaround is 63 days; 8 Nov would need 41.
Working view: NET November, more likely the second half (estimate). The FAA licence had not been reported issued as of 20 Oct.
Ariane 6 and Amazon Leo
Cumulative Amazon Leo satellites on Ariane 6 in 2026; VA271 would be the 4th of 24 contracted launches.
Why the calendar looks thin
Vandenberg launches on the November manifest (21 Oct)
Falcon 9 missions from Vandenberg in September
Starlink missions are added a week or two out, Starship dates firm up about a week before flight, and Chinese dates surface only days ahead through closure notices.
Slip log: what has already moved
Ariane 6 VA271
NET 7 Oct → NET 23 Nov (47 days)
Reported hydrogen turbopump anomaly in acceptance testing; Ariane 6 now likely flies six times in 2026, not seven or eight.
New Glenn return
Dec 2026 → early 2027
BE-4 main oxygen valve behind the 28 May explosion; CEO Dave Limp still saw “a path to launch this year” on 28 Sep.
Starship Flight 15
Late Oct, early Nov → NET Nov
FAA licence pending and hardware readiness; dates usually firm up a week before flight.
Starliner-1
NET Apr 2026 → Dec 2026 or Jan 2027
Contract changed to an uncrewed flight.
Mengzhou-1 (Long March 10A)
Late 2026 → reported early 2027
Cause not stated; Chang'e-7 has also moved to next year.
Starlink 15-15
20 Oct → NET 23 Oct
A technical issue, then weather twice.
Planning rule. Treat 23 November (VA271) and the Vandenberg Falcon 9 slots as the only dates likely to hold within a few days. Treat Flight 15 as a November-to-early-December window, and every December flag as at risk of moving into 2027.
What to watch and implications
What the month means, by stakeholder
What changed this month
Implication
Starship Flight 15 outcome
A caught ship opens the path to the weekly cadence targeted for H1 2027. A launch before 1 Dec stays inside Musk's window; later is slower than Pad 2's record.
SpaceX's year-end launch count
16 behind 2025. Watch whether Vandenberg ends near the 80+ officials expect and whether the Space Coast recovers from its pace toward about 75.
Ariane 6 VA271 on 23 November
If it holds, Arianespace's Amazon total reaches 136; another slip leaves Amazon Leo with no dated heavy launch before year-end.
Amazon's FCC milestone relief
With 24.5% deployed, the FCC's response sets how much schedule slack challengers can expect.
New Glenn's licence and date
Comments on LC-36A close on 30 Oct; a December flight needs a dated announcement soon, while Reuters points to early 2027.
The first Chinese booster reflight
A Zhuque-3 reflight before about mid-February 2027 would be the first non-SpaceX evidence that reuse can lower prices.
Q4 funding pace
About $521M in the window against September's $2.69B; a November with no $200M+ round would confirm a softer quarter.
Starlink subscriber adds
At Q3's 1.6M pace, the Q4 report would show about 15.2M subscribers (estimate).
Our judgement
The month strengthened SpaceX on every axis it measures — mass to orbit, broadband economics, Starship progress. The clearest risks to it are financial, with implied 2026 capex of $83–88B, and regulatory, not competitive.