Every drone, robot boat and planetary probe is a stack of the same 27 component layers. This is a guided tour of who actually makes them — 546 source-checked manufacturer entries, chokepoints, and gap analysis.
| BALLOON | ASSEMBLY GROUP | MAPPED ENTRIES |
|---|
The grounding taxonomy catalogues the uncrewed world across eight domains — from sub-250-gram FPV quads to work-class ROVs and deep-space probes. They look nothing alike. Underneath, they are assembled from the same shelf.
FIG. 1.0 — Domain plates, typed rows per sheet of Uncrewed_Systems_Master_Taxonomy_v0_6.xlsx
The DIY domain matters more than its size suggests: it shares its entire supply chain with the attritable-drone tier proven in Ukraine.
Pick a domain and watch the stack light up. A universal core — structures, batteries, inertial sensing, compute, connectors, autonomy — is essential nearly everywhere. The rest is domain signature: sonar defines the underwater world, radar the counter-drone world, star trackers the space world.
FIG. 2.0 — GENERAL ARRANGEMENT, GENERIC SURVEY-CLASS USV, NOT TO SCALE · layers rated Core / Common / Selective per domain (workbook, CrossRef Matrix)
We traced 492 manufacturers behind these 27 layers — every entry carrying a confidence tier and a live-checked source. The regional split below is the honest one: Chinese makers appear only where they hold a dependency position, and they still account for one entry in ten.
FIG. 3.0 — 546 entries by headquarters region
Region = headquarters country. Foreign-owned UK factories are flagged per entry — chapter 8 is about exactly that distinction. Confidence: 394 confirmed on the maker's own site or filings · 113 reported by credible trade press · 7 inferred · 32 unverified leads kept visibly separate.
The China story is not evenly spread. It is concentrated — almost surgically — in the high-volume electric small-drone stack.
Motors, speed controllers, propellers, flight-controller boards: trade coverage puts China at 80–90% of the global drone component chain, and even most Pentagon-cleared "Blue UAS" aircraft still fly Chinese motors. T-Motor sits on the US Entity List; the boards under most Western integrators come from Shenzhen and Guangzhou.
No UK, EU or US company makes drone-grade lithium cells at meaningful scale — the entire UK "battery industry" here is pack assembly on imported cells. Automotive lidar is Chinese-led (Hesai and RoboSense are both on the US DoD's 1260H list), and DJI's Dock dominates drone-in-a-box hardware even inside UK CAA-category operations.
The deepest chokepoint is upstream of every motor, servo and gimbal on this page — the ring of sintered NdFeB magnets inside each one. China mines about 60% of magnet rare earths, refines 91%, and manufactures 94% of the magnets themselves. The funnel narrows precisely at the stage the West stopped doing.
Engines, sonar, radar, connectors, spacecraft subsystems: here Chinese makers barely feature at the grades these parts are actually bought at. Defence and certified-industrial buyers can't substitute a hobby-grade part for a qualified one, so it is the qualified tier that decides dependency, and in these five layers it is held by UK, European, US and allied firms. The dependency is specific, mappable territory. It can be exited layer by layer rather than all at once.
FIG. 4.0 — Editorial severity per layer, from sourced per-category findings · magnet stages: IEA 2025
Since 2024 the component layer has become an instrument of policy on both sides. If you build, buy or invest in this sector, here's a smattering of the policy shenanigans.
Each event is sourced in the workbook (Federal Register, MOFCOM announcements, FY2026 NDAA, UK trade press). Dates verified live during the research pass, September 2026.
Six layers came back with no dominant Chinese maker at the quality tiers that matter most. In several of them the UK is not merely present but world-class.
138 UK entries cluster into a recognisable geography: acoustics on the south coast, radar around Cambridge, space in Glasgow and Harwell, carbon fibre in the Highlands. Set against that, the gaps are just as specific.
FIG. 7.0 — Stylised chart, not to scale · hover or tap a cluster to open its dossier
The most consistent UK pattern in this dataset is not lost factories — it is lost ownership. The plants stay; the decisions move. Individually each deal is defensible. Collectively they are a policy question.
Each transfer individually sourced (company announcements, RNS filings, press); dates re-verified live 25 September 2026.
No hedging. Across 546 sourced entries, these are the layers where Britain either has nobody at all, or is one sale or one site away from having nobody. Each line is drawn from the per-category findings in the workbook.
Every UK "battery maker" in this map assembles imported cells; Volklec in Coventry is the single exception, on Chinese-licensed chemistry. The cells themselves come from China, Korea, Japan and Taiwan, and US FEOC-style sourcing rules are already cascading toward allies.
The part that matters is the motor-drive stage itself: a bridge of fast-switching transistors, gate drivers and control firmware that turns battery DC into the precisely timed currents a motor needs, qualified for flight vibration, temperature and current spikes. UK firms already build the neighbouring boxes, TT Electronics in DC-DC conversion and Alexander in battery management, but stop one board short. Nothing about that board is exotic: the silicon is commodity, the control theory is published, and Australia's Currawong shows a small allied firm can own the niche. What is missing is a funded product programme and a defence launch order, which is why this reappears as Opening 01.
China makes 94% of the world's sintered NdFeB magnets, the ring inside every motor, servo and gimbal on this page, and its export-control regime is live. The UK's answer so far is two recycling ventures and no primary capacity.
FPV flight boards, video links and propellers are near-totally Chinese, and the British Army was printing FPV airframes on Chinese desktop printers until the Defence Secretary ordered a review in 2026. In Ukraine this class is expended at higher rates than any other airborne system and is treated as ammunition rather than aircraft. Sustaining that rate is a manufacturing problem, and the UK currently manufactures almost none of the electronics involved.
Not thin. Zero, across survey-grade, automotive-grade and robotics perception, in a category the US DoD considers sensitive enough to blacklist the Chinese leaders.
No UK maker of rugged GCS hardware at scale was found, and the two strongest UK autonomy-software houses were both acquired: BlueBear by Saab in 2023, Callen-Lenz by BAE in 2025. UK crews fly UK drones from foreign-made controllers running foreign-owned code.
Every claim above is source-linked in the workbook's Category Summaries and Master sheets. The constructive readings of these six sit in the next chapter.
Judged against demand signals found in the research — NDAA sourcing rules, the Blue UAS gap, post-Ukraine jamming demand, DJI security reviews — these are the credible UK and allied entry points.
If cheap drones are ammunition, then the factory is the magazine. Fix the component gaps above and the next bottleneck is how fast you can assemble the things, because this kind of war is won by whoever replaces losses fastest, not by whoever builds the cleverest machine. The rough shape of the answer already exists. In Ukraine, volunteers run distributed farms of 3D printers turning out drone parts. The British Army has printed FPV airframes in the field on consumer machines. The deliberate version of that idea is a factory built as a module: a containerised production line that prints the structural parts, assembles the boards, builds the wiring and tests the finished aircraft at the door, and can be copied ten times over or driven to where it is needed. None of this needs inventing. It needs engineering, and the UK has several of the pieces: it builds metal 3D-printing machines, it has deep test infrastructure, and its motorsport industry runs on exactly this kind of rapid engineering. Designing the factory unit before the demand spike is the cheap part. Whoever holds the proven, replicable production-line design when volume is suddenly needed owns the surge, as long as the supply chain holds up.
This is the dataset the whole report is built on: 546 manufacturer entries across the 27 layers, each with a confidence tier and a source link. The grid shows who makes what, and where; the red NIL cells mark where no maker was found. Pick a layer, or a single cell, and the makers appear alongside. The search box and chips narrow things down.
FIG. 11.0 — The capability console · left: makers per layer per region (hollow red = nobody found) · right: the selected layer's makers with tiers & sources