The Small Satellite Market Is a Launch Manifest. Read the Manifests.

There is no authoritative total for the small satellite market, and the reason is not a shortage of analysts. It is that the thing being counted has never been agreed. Count spacecraft buses and you get one number, add launch and you double it, add ground segment, terminals and spectrum and you multiply it again.

What does exist, and is checkable, is a set of fleet counts and launch schedules published by the companies themselves. Read together, they turn an unmeasurable market into a set of dated, falsifiable claims. That is more useful than a forecast.

What is actually flying

Two ratios fall out of that table immediately, and both matter more than any market size.

Starlink’s 10,000-plus satellites against an initial authorisation of 12,000 is 83.3% of a constellation completed. That is a capex programme approaching the end of its deployment phase and transitioning into a maintenance and replacement phase — a completely different business from the one everyone else is running.

And roughly 650 direct-to-cell satellites inside a fleet of more than 10,000 is 6.5%. The entire satellite-to-phone story, the one that generates the headlines and the operator partnerships, currently rides on about one satellite in fifteen. That is not a criticism; it is a capacity signal. The most commercially discussed layer of this industry is the smallest physical layer of it.

The launch arithmetic, and where it breaks

Deployment targets are the only genuinely testable claims in this sector, because a satellite has to be launched to exist.

AST SpaceMobile is the cleanest case. It targets 45 to 60 satellites in orbit by the end of 2026. It had flown at least 13 by 5 August, when BlueBird 11, 12 and 13 went up together. On the observed three-spacecraft-per-launch pattern, closing that gap requires 32 to 47 more satellites — roughly 10.7 to 15.7 further launches. There were about five months left on the calendar.

That is 2.1 to 3.1 launches a month. The company’s stated cadence is approximately one mission every one to two months.

The arithmetic does not reconcile, and the resolution is not mysterious. Either cadence compresses sharply, or the target slips, or the per-launch payload rises. Every one of those outcomes is observable, and each has a different meaning for the suppliers feeding that line.

Amazon Leo’s numbers tell a different story, and arguably a more interesting one. The company has contracted 80-plus launches for a constellation of more than 3,000 broadband satellites that will not begin direct-to-device deployment until 2028. It bought launch capacity years before it needed to fill it — and paid for the option, not the kilogram. In a market where the manufacturer is normally the one carrying schedule risk, the launch manifest became the balance-sheet item.

Then the constraint binds hardest where there is least capacity. Jio’s plan for 1,600 to 1,650 satellites over two to three years means 550 to 800 spacecraft a year, into a domestic launch industry that does not yet match SpaceX’s cadence, its vertically integrated manufacturing or its ability to loft large batches on reusable vehicles. The precedent is already set: OneWeb had to buy Falcon 9 launches in 2024 after Russia denied it launch services, with satellites sitting on the pad. The constellation that exists in orbit today was finished on a competitor’s rocket.

India’s supply side is genuinely moving — Skyroot Aerospace’s Vikram-1 became the first privately developed Indian launch vehicle to reach orbit in July, per ISpA’s 2025–26 annual report, and made Skyroot the country’s first space-tech unicorn. But one launcher reaching orbit is a capability milestone, not a cadence. A constellation programme is a cadence problem.

Revenue is the tell, and it is not service revenue

The most instructive number in the sector is AST SpaceMobile’s revenue mix. The company reported USD 70.9 million of revenue for full-year 2025. Of that, USD 54 million came in the fourth quarter alone — 76.2% of the year in three months — leaving USD 16.9 million across the first nine months.

That is not a business ramping on subscribers. It is a business ramping on gateway equipment sales to carrier partners on five continents and on early government service milestones. The 2026 guide has been raised from at least USD 140 million, stated on the fourth-quarter call, to USD 150–200 million — a 2.12x to 2.82x multiple of 2025’s outturn, if it lands.

Now the structural figure. AST’s backlog stands at approximately USD 1.30 billion, including commercial contracts and government awards, of which more than USD 125 million is government. Against USD 70.9 million of 2025 revenue, that backlog is 18.3 times the current revenue base.

Read that ratio properly. An 18.3x backlog is not evidence of a business that has arrived. It is evidence of a business that has been prepaid by its future customers — carriers buying gateways, governments buying milestones — against a constellation that does not yet exist in sufficient numbers, while carrying operating losses of roughly USD 260 million a year. The backlog converts only if the launches happen. Which is why the launch schedule, not the contract book, is the number to watch.

Around fifty gateways were in various stages of completion, installation or planning in August. Gateways are the part of the system a carrier pays for before service exists.

The companies that do not need a launch

The contrast inside this sector is starkest against the operators that already finished deploying.

Iridium runs 66 satellites in polar orbit and is layering a standards-based narrowband service, Iridium NTN Direct, onto that sunk capital during 2026. No new constellation is required to earn the incremental revenue. The same logic drives Skylo’s narrowband non-terrestrial work with Verizon, Google, Samsung and Comcast, Lynk’s commercial SMS in select markets with partners such as Vodafone Ghana, and Apple’s satellite SOS on Globalstar’s proprietary system.

The consolidation running through that layer is the tell. Amazon agreed in April 2026 to acquire Globalstar — satellites, infrastructure and mobile satellite service spectrum — with the deal expected to close in 2027, and separately agreed to support Apple’s satellite services. Device layer, spectrum and space segment are converging onto one balance sheet. That is what a maturing access business looks like: the scarce assets get bought, not built.

And the partnership count is far ahead of the revenue. The Global mobile Suppliers Association counted 170 publicly announced operator-satellite partnerships across 80 countries and territories by August 2025 — with 34 operators in 25 markets having launched commercial services. That is 20.0% of announced partnerships actually billing. Twelve launches had occurred; 24 projects sat at trial or licensed stage.

Twenty percent conversion is the honest state of commercial direct-to-device in 2025–26. Deloitte’s estimate that spending on direct-to-device satellite capacity could reach USD 6–8 billion in 2026, with more than 1,000 D2D-capable satellites potentially operating by year-end, is a statement about what has to be launched between now and December, not a statement about what has been earned.

India shows exactly where the margin sits

The clearest read on value migration in this industry is not in orbit but in licensing.

As of June 2025 India had three Unified Licence holders for Global Mobile Personal Communication by Satellite: Jio Satellite Communications, OneWeb India Communications and Starlink Satellite Communication. Alongside them sit eleven Unified Licence holders for Commercial VSAT CUG service — the layer that can actually invoice an Indian enterprise. Pheonix Research set out that eleven-against-three structure in October 2026, and it remains the single most revealing ratio in the sector: OneWeb’s constellation already exceeds 600 satellites, the satellites are in orbit, the licence exists — and the commercial relationship with the customer sits one layer down, with companies whose economics are set by wholesale capacity prices they do not control.

Domestic capacity adds a second thread. GSAT-N2 added up to 48 Gbps of high-throughput capacity in 2025, and TRAI’s Consultation Paper 06/2026 on a Satellite Communication Network authorisation and spectrum framework is where the split between network operator and retail service provider is being settled. For the smallsat industry this matters more than any bus specification: it decides whether the space segment is a utility selling wholesale capacity or a business with customers.

The Indian ecosystem’s own numbers show how much capital is now chasing that question. ISpA reports more than 380 space startups, over USD 150 million raised in the year — up 94% year on year — and a space economy of roughly USD 8 billion targeting USD 44 billion by 2033. Those are the industry association’s own figures, and the target is a target. But 380 companies competing for a position inside a licensing framework is a very different competitive dynamic from 380 companies building satellites.

Four numbers to underwrite with

• Satellites in orbit over satellites authorised. Starlink is at 83.3% of a 12,000-spacecraft authorisation. Jio is at zero of 1,600–1,650. Amazon’s deployment begins in 2028. This ratio separates the programmes that are capex-complete from the ones still consuming capital.
• Launches flown over launches contracted. Amazon’s 80-plus contracted launches are capacity bought, not satellites delivered. A manifest is a schedule of future liabilities.
• Backlog over revenue. At 18.3x, a backlog is a promise about launches.
• Service revenue over total revenue. Near zero for the pre-commercial operators, and the only figure that survives the day the gateway orders stop.

What this data cannot do

There is no consolidated, authoritative worldwide launch count in what can be verified here, so every launch figure above is company-specific and dated. The three-satellites-per-launch assumption comes from a single observed mission on 5 August 2026 and may not hold across configurations. Amazon’s 80-plus contracted launches are not publicly tied to specific spacecraft. Jio’s 1,600–1,650 is a plan, not a manifest.

And there is no Pheonix Research series for satellite buses, launch services, ground segment or user terminals — so there is no defensible total addressable market number for small satellites, and this article will not produce one. What it will do is rank the operators by how much of their promise is already in orbit.

The point

The small satellite market is best measured not in dollars but in the gap between satellites promised and satellites flown — and that gap is currently widest exactly where the capital is newest. Starlink is finishing a constellation at 83.3% completion with direct-to-cell hardware on 6.5% of its fleet. AST is guiding to roughly two to three times its 2025 revenue on a backlog eighteen times it, while its own cadence arithmetic points to a shortfall against a year-end target. Amazon has bought 80-plus launches for spacecraft that arrive in 2028. OneWeb finished its constellation on a competitor’s rocket, and Jio now plans 550 to 800 satellites a year into a launch industry that has not yet demonstrated the cadence.

Launch is not scarce in aggregate any more. It is scarce at the level of the individual company, on the individual schedule, and it is the only input in this industry that cannot be prepaid into existence.

Method note: fleet counts, launch dates, revenue, backlog and guidance are company disclosures and dated trade reporting; licensing counts, the GMPCS/VSAT structure and India’s space-economy figures come from Indian regulatory and industry sources; no third-party market-research estimates are used. Every ratio, sum and multiple above was computed arithmetically from those published figures.