> ## Content Index > Fetch the complete content index at: https://www.satelliteinsights.com/llms.txt > Use this file to discover other available public pages before exploring further. # Every Constellation Is Now a Starship Derivative - URL: https://www.satelliteinsights.com/every-constellation-is-now-a-starship-derivative/ - Published: 2026-07-28T12:00:16.000Z - Updated: 2026-09-09T12:40:25.000Z - Description: Launch capacity, not capital or spectrum, now gates every satellite constellation. Why the whole market's schedule has become a Starship derivative. - Author: Glenn Canales - Tags: LEO broadband, Analysis, Launch Capacity, Starship, SpaceX, LEO Analysis The Launch Shortage July 28, 2026 [01The derivative chain](#thesis) [02What's actually for sale](#supply) [03Rationing, in plain sight](#rationing) [04Who needs the flights](#demand) [05When deadlines meet physics](#milestones) [06Who wins, who loses](#winners) [07What would prove this wrong](#falsify) [—Methodology & caveats](#method) Satellite Insights # Every Constellation Is Now a Starship Derivative The satellite industry planned the rest of the decade on the assumption that launch was a solved problem. In the space of a year, launch quietly became the scarcest input in the business, and one test program in South Texas became the variable every operator’s schedule depends on. **Glenn Canales** — Principal, Satellite Insights LLC · 40+ years in satellite communications; former Sr. Director, Enterprise Broadband Services, Viasat (2016–2024); iDirect, PanAmSat/Intelsat; USAF Communications Supervisor (TS/SCI). Disclosure: The author is a former Viasat employee. Viasat is a co-founder of the Equatys joint venture discussed in this analysis. All information here is drawn from public sources. 123 of 165 2025 Falcon 9 flights consumed by Starlink 2025 flight log, all providers Late 2028 SpaceX rideshare bookings closed beyond this point Transporter & Bandwagon · SatNews / SpaceNews, Jul 2026 $7,000/kg Published rideshare list rate, up from $5,000 at 2019 launch published list rate · $5,000 at 2019 debut 36 months New booking lead time, up from roughly 12 industry booking guidance, up from \~12 Card 01 · The Thesis ## The most important dependency in the industry runs through a test stand in Texas Why does one company’s test schedule now govern every operator’s deployment plan? **The extractable answer:** On July 20, SpaceX confirmed it has stopped taking commercial rideshare bookings beyond late 2028\. Read that again. The company that made launch feel infinite is telling the market there is none left to sell for the rest of the decade. Almost nobody in the satellite business has repriced their plans for it. Here is the sentence I would put on the whiteboard of every satellite operator, investor, and program office: the external availability of Falcon 9 is a derivative of Starship's test schedule. Not influenced by it. Derived from it. Starlink's next-generation V3 satellites are too large to fly on anything except Starship. Until Starship flies them at operational cadence, Starlink keeps consuming Falcon 9 internally, and everyone else in the market splits what's left. In 2025, SpaceX flew a record 165 Falcon 9 missions, and 123 of them carried Starlink. The most capable operational rocket fleet in history was roughly three-quarters an internal delivery truck. The commercial market never had access to Falcon 9\. It had access to the residual. Starship slips V3 debut slid from March to late May 2026; boosters still failing recovery; suborbital only → Falcon 9 stays internal Starlink V3 needs Starship, so V2 Minis keep consuming Falcon 9 flights → Everyone else waits External slots shrink, prices rise, bookings close beyond 2028 The propagation path. An operator who never books a Starship still inherits its schedule. That is why the correlation matters more than the delay itself. Starship's slips are well covered. What is not covered is the transmission mechanism: every quarter Starship stays suborbital is a quarter Falcon 9 remains pinned to internal Starlink duty, and a quarter in which the only meaningful relief valve for the entire Western satellite industry stays shut. Operators who have never spoken to SpaceX sales are exposed. Their backup providers are exposed. The whole market's schedule now correlates to one company's test outcomes, and I have not seen that risk priced into a single constellation business plan. Where does Starship actually stand? Flight 12 flew on May 22 after sliding from March. The ship did its job, deploying twenty-two payloads and splashing down off Western Australia. The booster did not: an engine-timing error left the post-separation flip misaligned by roughly ninety degrees, five engines failed to relight, and it struck the Gulf at about 1,450 kilometres per hour, scattering a debris hazard zone that put five aircraft into holding patterns and grounding the program for fifty-two days. Flight 13 finally flew on July 24, after an ignition abort on the 16th and a weather scrub on the 23rd, and it was the best flight of the V3 campaign so far. The ship deployed twenty real Starlink V3 satellites, the first functional V3s Starship has carried, communicated with all of them over radio and laser links, completed the in-space Raptor relight that Flight 12 never attempted, and made the softest splashdown the program has managed. Read the fine print, though. The satellites rode a suborbital arc and burned up on reentry about twenty minutes later, so not one joined the constellation; the flight proved the deployment mechanism, not operational service. And the booster, the reusable half that is the entire economic point of Starship, again fell short, firing only five of thirteen planned engines on its landing burn and hitting the Gulf hard for the third straight flight. Those are real milestones, and Flight 13 was genuine progress on the ship. They are also, three and a half years into flight testing, still suborbital test flights that have yet to place a single working satellite in orbit or bring a booster back intact, which are the two things that have to become routine before Starship offloads anything from Falcon 9\. NASA's Office of Inspector General concluded in March that Starship will not be ready to land crew on the Moon in June 2027, with the critical propellant-transfer demonstration a year behind. Designed capability: full reusability and a hundred-plus tonnes to orbit. Demonstrated capability: suborbital flights and expended boosters. That distinction is the whole schedule risk, and this publication's standing rule is to price the demonstrated, not the designed. Analysis The Launch Shortage July 28, 2026 [01The derivative chain](#thesis) [02What's actually for sale](#supply) [03Rationing, in plain sight](#rationing) [04Who needs the flights](#demand) [05When deadlines meet physics](#milestones) [06Who wins, who loses](#winners) [07What would prove this wrong](#falsify) [—Methodology & caveats](#method) Card 02 · Supply ## How much launch capacity is actually for sale? Walk the alternatives, and check the date on each claim, because mid-2026 delivered a rare synchronized supply shock. Blue Origin's New Glenn exploded during a pre-launch hotfire on May 28, destroying the vehicle that was to carry 48 Amazon Leo satellites and heavily damaging the company's only operational New Glenn pad, with reporting pointing to a year or more of lost access. ULA's Vulcan flew roughly six times in 2025 against a plan of about twenty, and its backlog of more than eighty missions is about 90 percent spoken for by Amazon and the US Space Force; ULA itself says the manifest is at capacity through 2027 and most of 2028\. Ariane 6 is fully booked for 2026 and 2027 with few 2028 slots, which is why Arianespace is already selling 2029 and 2030\. Rocket Lab's Neutron has yet to fly, with a debut targeted late this year, and Rocket Lab's pending acquisition of Iridium makes it a direct competitor to any mobile-satellite operator shopping for a ride. Smaller vehicles from Firefly, Relativity, and Stoke are at the credibility stage, not the capacity stage. The one genuinely underrated relief valve is India. NewSpace India launched 72 OneWeb satellites on two LVM3 flights when OneWeb lost its Soyuz access, and flew AST SpaceMobile's BlueBird-6 last December, so the precedent for US-content constellations on Indian vehicles is established, not theoretical. The constraint is cadence: an LVM3 gets built roughly every six to eight months. India is a pressure-release, not a solution. And the large Chinese launch base building Guowang and Qianfan, somewhere near 190 and 200 satellites launched respectively by mid-year depending on whether you count launched or operational, is walled off from Western payloads by US export controls. It relieves nothing; it adds a parallel demand system to the global picture. Card 03 · Rationing ## What does launch rationing look like? It looks like this Rationing rarely announces itself. It shows up in behavior, and the behavior is now unambiguous. SpaceX stopped accepting Transporter and Bandwagon rideshare bookings beyond late 2028, confirmed independently by at least nine SpaceX partners and customers to SpaceNews in June, with available slots potentially cut to half recent levels. Launch integrator SEOPS bought an entire Falcon 9 for a 2028 rideshare and filled its manifest in three weeks, with roughly thirty customers waitlisted. Exolaunch signed multi-year deals to lock capacity through 2028\. Industry guidance on booking lead time has moved from twelve months to thirty-six. And prices are going the wrong direction for a market that was promised reusability-driven deflation: the dedicated Falcon 9 list price rose to $74 million this year, rideshare sits at $7,000 per kilogram against $5,000 at the program's 2019 debut, and SpaceX has adopted a structured annual escalation of roughly $500 per kilogram per year. RIDE! Space's chief executive put the operator's-eye view plainly: there is a clear asymmetry between when satellites need to launch and what the market can provide, and that could be lethal at some point. One reallocation detail deserves a caution label. Reporting attributes part of SpaceX's capacity withdrawal to reserving payload volume for orbital artificial-intelligence data-center payloads alongside next-generation Starlink. Single source That claim traces to one outlet and I have not found independent confirmation; treat it as plausible and unverified. The booking halt itself needs no such caveat. It is multiply confirmed and dated. The launch market was never abundant. It was one company's surplus, sold at their convenience, and the surplus is gone. Card 04 · Demand ## Who is competing for the flights that remain? Four demand engines are pulling on the same shrinking supply at once. Starlink itself, above 10,800 operational satellites as of July 20 and needing Starship for its V3 generation. Amazon Leo, which stood at 396 satellites deployed after the final Atlas V flight on July 2 and must reach 3,232 by July 2029, just lost two of its three near-term launch providers to grounding and backlog in the same quarter. The US government, which tripled the ceiling on its National Security Space Launch Phase 3 Lane 1 contract to $17 billion on July 17, an increase a Space Systems Command spokesperson said covers an additional 110 launches for a total of 170 over the decade, while the Space Development Agency and the Golden Dome architecture add tranche after tranche of satellites that must fly on schedule. And the sovereign constellations, the demand engine I traced in [Scale vs. Sovereignty](https://www.satelliteinsights.com/scale-vs-sovereignty-weekly-deep-dive/), all of which now need vehicles: Europe's IRIS² has reserved Ariane capacity from 2029, Telesat's Lightspeed has slipped to early 2028 on chip delays with first launches still targeted for late this year, and Equatys, the Viasat and Space42 joint venture planning up to 2,800 satellites for initial service around 2029, had announced neither a satellite manufacturer nor a launch provider as of mid-year, with executives describing contracts as imminent. Single source That last characterization rests on one outlet's reporting of executive comments; what is verifiable is the absence of any announced contract, which in a market now booking thirty-six months out is itself the story. Do the arithmetic on that last point, because it generalizes. A constellation targeting 2029 service needs first launches in 2028, which under current booking reality means contracts signed now, into a manifest that is closed. Every constellation on paper today that has not already locked its rides is competing for slots that functionally do not exist, at prices that rise annually, against buyers that include the Pentagon and Amazon. The honest planning assumption for any unlaunched constellation is a twelve-to-twenty-four-month slip from its published schedule, and the trigger to update that assumption is not a press release. It is a signed launch contract, or the continued absence of one. Card 05 · Regulation ## What happens when regulatory deadlines meet a market that cannot launch? Spectrum rights come with deployment clocks, and the clocks did not get the memo about supply. Filings now describe something close to [1.7 million satellites](https://www.satelliteinsights.com/1-7-million-satellites-astronomers-vs-the-fcc-analysis/) across every operator on the books, a number the launch industry cannot service this decade at any price, which means the milestone regime was always going to collide with physics before it collided with ambition. The template for how this collides got written on June 5, when the FCC waived Amazon Leo's July 30 milestone of roughly 1,600 satellites but attached a real cost: satellites launched after July 30 temporarily lose interference-priority protection until Amazon demonstrates compatibility with other systems, recoverable in March 2028 or sooner on progress, with the full 2029 deadline intact. Bend the timing, tax the priority. Expect that template to be applied again, because the alternative is revoking authorizations across an industry that physically cannot comply, and no regulator wants to hand the spectrum map to the one operator that can. The International Telecommunication Union's bring-into-use and deployment milestones for non-geostationary systems carry the same collision course at the global level. The practical consequence is perverse and underappreciated: scarcity forces operators to commit launch capital years earlier than their business plans assumed, precisely to protect spectrum priority, which transfers even more pricing power to the launch providers. Card 06 · Winners & Losers ## Who wins a launch shortage, and who loses? Winners first, because they are less obvious. Anyone with assets already in orbit just got more valuable: the incumbent fleets, from Starlink to Iridium to the geostationary (GEO) operators the market spent five years writing off, hold inventory that cannot be quickly replicated at any price. That logic extends directly to satellite life-extension and servicing, and the market is confirming it in real time: Northrop Grumman launched the first privately owned operational robotic GEO servicing mission on July 21 with SES and Optus as named clients, each extension pod adding up to eight years of life, while Astroscale took investment from SKY Perfect JSAT in May explicitly for GEO life extension and refueling. When replacing a satellite means joining a three-year launch queue, keeping the old one alive is the best capacity trade in the industry. A commonly cited rule of thumb holds that ten or more GEO satellites run out of propellant each year with revenue-earning transponders still aboard. Single source Treat the figure as industry shorthand rather than audited data; the direction is what matters. The launch providers themselves win pricing power they have not held in a decade. And India's launch sector wins by being the only externally available capacity with proven Western-payload credentials. The losers are the businesses built on the abundance assumption. Smallsat startups that modeled multiple cheap rideshares a year now face closed manifests and compounding prices. New constellations without sovereign backing or a captive rocket face the roughest math of all, because the escape routes both require capital at a scale that reprices their plans: block-buy launches years ahead, as Amazon did with more than eighty contracted flights, or own the rocket, which is precisely the logic Rocket Lab just paid about $8 billion to execute in reverse by buying Iridium. I argued when that deal was announced that [a flywheel is built, not bought](https://www.satelliteinsights.com/bought-not-built-why-rocket-labs-iridium-deal-isnt-the-next-starlink/), and I still think the acquisition buys less than the comparison to Starlink implies. What it unambiguously does buy is immunity from the queue everyone else is standing in. The strategic lesson of this cycle is uncomfortable for everyone who spent the last decade preaching disaggregation: vertical integration is being validated in real time, and launch access has replaced spectrum as the barrier to entry that decides who gets to exist. Card 07 · Falsifiability ## What would prove this thesis wrong? An analysis that cannot be falsified is a mood, so here are the benchmarks. The shortage thesis weakens when at least two of these occur, and I will say so in print when they do: **The benchmarks:** 1. Starship deploys Starlink V3 satellites from orbit at a repeating monthly cadence, which begins offloading Falcon 9. 2. New Glenn returns to flight and sustains better than six launches a year. 3. Vulcan sustains a twice-monthly cadence for two consecutive quarters. 4. SpaceX reopens rideshare bookings beyond 2028, the single cleanest market signal available. Until then, the operating assumptions I would carry: published constellation schedules are twelve to twenty-four months optimistic unless launch contracts are signed; launch pricing escalates annually; regulators follow the FCC's bend-timing-tax-priority template; and the correlation risk to Starship test outcomes sits, unpriced, inside every satellite business plan in the Western market. The industry spent five years arguing about spectrum and capital. The scarce input was sitting on the pad the whole time. ## Methodology & caveats - **Live-figure discipline:** every launch count, price, backlog figure and date above was pulled from live July 2026 sources at compile time, not from memory or prior sessions. - **Designed vs. demonstrated:** cadence targets, payload capacities and market projections are announced or designed figures unless labeled demonstrated. Starship’s designed capability is full reusability and 100-plus tonnes to orbit; its demonstrated record is suborbital flights and expended boosters. - **Single-sourced items are flagged in print** at three points: the orbital AI data-center reallocation claim, the Equatys “imminent” contracts characterization, and the \~10-GEO-satellites-per-year propellant rule of thumb. Treat each with the caution the flag implies. - **Flight 13 status:** the July 24 outcome reflects reporting available at publication; check SpaceX’s post-flight recap before citing this piece for the flight record. - **Chinese constellation counts** differ by launched-versus-operational definitions and are given as ranges. - **Sources:** SpaceNews, Via Satellite, SatNews, Spaceflight Now, Payload, NASASpaceflight, GeekWire, The Register, NASA Office of Inspector General report IG-26-004, Federal Communications Commission (FCC) order DA-26-553, and company releases and filings. ## The intelligence behind the headline numbers Satellite Insights publishes independent weekly analysis on satellite operators, constellations and ground networks. No hype. A point of view. [Subscribe free](https://www.satelliteinsights.com/#/portal/signup) Satellite Insights · Analysis · July 28, 2026 © 2026 Satellite Insights LLC. Compiled from live sources; figures verified at time of publication.