What does a modern satellite contain, and which of its parts matter for supply?
The short answer. a satellite is a power plant, a radio station, a computer, a guidance system and a small rocket engine packed into a frame. A typical low-orbit communications satellite of a few hundred kilograms contains solar arrays and batteries, power electronics, one or more flight computers, attitude sensors and actuators (star trackers, reaction wheels, magnetorquers), a propulsion system, radios and antennas, often optical terminals to talk to other satellites, and the structure and harness that hold it together. By cost, the payload and power system dominate; by supply risk, a few small, specialised parts matter far more than their price suggests.
The subsystems
| Subsystem | What it does | Typical parts | Why it matters for supply |
|---|---|---|---|
| Power generation | Turns sunlight into electricity | Solar cells (III-V or silicon), cover glass, array substrate, deployment hinges | Cells are made by few qualified firms; III-V cells depend on germanium and gallium |
| Energy storage | Powers the satellite in eclipse | Lithium-ion cells, battery management | Several space cell formats are foreign-sourced |
| Power conditioning | Regulates and distributes power | Converters, switchgear, fuses, relays | Space-grade switchgear has the longest reported lead times |
| Command and data handling | Runs the satellite | Flight computer, memory, FPGAs | Radiation-tolerant parts come from very few foundries |
| Attitude determination and control | Points the satellite | Star trackers, sun sensors, gyros, reaction wheels, magnetorquers | High unit counts per satellite; mostly in-house at the largest builders |
| Propulsion | Raises orbit, keeps station, de-orbits | Hall or ion thrusters, propellant tanks, valves, xenon, krypton or argon; or hydrazine thrusters | Propellant gases have narrow supply; thrusters few suppliers outside integrators |
| Communications | Talks to the ground and users | RF front ends, GaN and GaAs amplifiers, phased arrays, antennas | Compound semiconductors depend on gallium |
| Crosslinks | Talks to other satellites | Optical inter-satellite terminals | Fewest merchant suppliers relative to demand |
| Payload | The mission | Transponders, imagers, infrared focal planes, radar | Defence sensors have the narrowest supply |
| Structure and thermal | Holds and cools | Aluminium, carbon-fibre composites, heat pipes, radiators | Composite fibre and resin combinations need long qualification |
| Harness | Connects everything | Cables, connectors | Connectors named as acutely short of capacity |
A reference satellite
To make the supply questions concrete, this study uses a reference low-orbit satellite. It is a Setcoin illustration built from public descriptions of current small and medium constellation satellites; it does not describe any one product.
| Item | Reference value (Setcoin illustration) |
|---|---|
| Mass at launch | ~300 kg |
| Solar array power (beginning of life) | ~3 kW |
| Battery | ~3–5 kWh lithium-ion |
| Flight computers | 2 (redundant) |
| Star trackers | 2 |
| Reaction wheels | 4 |
| Electric thrusters | 1–2 Hall thrusters |
| Propellant | ~10 kg xenon or krypton |
| Optical terminals | 0 (simple missions) to 3–4 (mesh networks and defence transport) |
| Expected life | 5–7 years |
The reference satellite feeds the constellation bill-of-materials model (Chapter 46). A defence tracking satellite or a geostationary satellite differs by an order of magnitude in power, mass and part quality; the chapters on each component note where the reference does not apply.
Where the money goes
Public cost breakdowns are rare. Industry descriptions and Setcoin's reading of programme budgets suggest the following split for a commercial low-orbit satellite built in volume:
Exhibit 2.1
The payload and power system take most of the cost; the parts that delay programmes take little of it
Share of satellite unit cost by subsystem (midpoints, Setcoin estimate)
Source table: Range (6 rows, shown as an exhibit above)
| Share of satellite unit cost (Setcoin estimate) | Range |
|---|---|
| Payload (communications, imaging or sensing) | 30–45% |
| Power (arrays, batteries, conditioning) | 15–25% |
| Avionics and attitude control | 10–20% |
| Propulsion | 5–10% |
| Structure, thermal and harness | 10–15% |
| Integration and test | 10–15% |
Setcoin estimate from public descriptions; individual programmes vary widely.
Cost is not risk
The parts that most often delay programmes are not the most expensive. A connector costs dollars, but a satellite cannot fly without hundreds of them, and the PwC/AIA study of March 2026 named connectors among the parts with acutely low production capacity. A reaction wheel costs tens of thousands of dollars, and one radar operator keeps three to five satellites' worth of wheels in stock because it cannot rely on delivery times. The supply question is therefore not "what costs most?" but "what is hardest to get on time, and what happens if it does not arrive?" That is what the scorecard in Chapter 4 measures.
Three kinds of satellite, three supply problems
| Class | Examples | Volume | Supply problem |
|---|---|---|---|
| Mass-produced constellation satellites | Starlink, Amazon Leo, Qianfan, Guowang | Hundreds to thousands a year per programme | Throughput of parts made in-house; launch |
| Medium-volume government and commercial satellites | SDA tranches, IRIS², Earth observation fleets | Tens to low hundreds a year per programme | Merchant parts with few suppliers: optical terminals, processors, solar cells |
| Bespoke high-value satellites | Geostationary communications, missile warning, science | A few a year | Qualification, radiation-hard parts, specialised sensors, test facilities |
The middle class is where the bottlenecks bite hardest, because it is too small to justify making parts in-house and too large to be served by craft production. Most of the defence and sovereign demand of the next five years sits there (Chapter 41).
What would change this view
Standardised, catalogue-bought satellite buses with interchangeable parts would reduce supply risk for the middle class. A wave of new merchant suppliers in optical terminals and processors (Chapter 49) would move the bottleneck from parts to integration and test.