🌏 中文版
What this article does not do: name tickers, recommend positions, or discuss timing. It applies an existing framework to a sector, and the conclusion may well be "doesn't qualify right now" — which it is.
This is the last piece of the drone series. Earlier entries covered industry structure, Taiwan's position, regulation, careers, and industry cycles. This one feeds those facts into the framework.
The criteria are the same ones this site uses on the AI supply chain: supply chain chokepoint, structural demand, high switching cost, long-term institutional holding. Using one ruler across sectors is how you learn what differs.
Criterion one: supply chain chokepoint ❌
The question: if this link were removed, would downstream stop?
Taiwan's position is unambiguous. Thunder Tiger's published BOM split shows Taiwan supplying battery modules, controllers, motors and ESCs, power modules, propellers, and frames — all Layer 2. The three Layer 3 critical modules (comms/GPS, flight control, camera) are bought from US, European, and Japanese vendors by the only company in Asia to clear Blue UAS.
Layer 2's problem is substitutability. Motors, frames, and propellers have many suppliers; switching costs weeks of qualification, not years of redesign. The declining localization curve says the same thing: roughly 70% for small airframes, 60% medium, 30% large — the more complex the aircraft, the less Taiwan can supply, because complexity concentrates in Layer 3. (in Chinese)
Verdict: does not currently hold. Taiwan sits at the most competitive, lowest-bargaining-power layer. The policy bet on "three chips, two software" (flight control, comms, and satellite positioning chip modules plus flight and ground control software) targets the right cell, but that's a future state, not a present moat.
Criterion two: structural demand ⚠️
The question: is demand driven by end-user behavior, or by a mechanism that can be interrupted?
Drone demand itself is real — battlefield consumption, public-sector inspection, supply chain reshuffling all have substance. But look at the structure on Taiwan's side: 2025 output was NT$12.9 billion against NT$2.95 billion of airframe exports — 22.9%. The other ~80% is domestic public-sector and defense procurement.
That is policy demand, not structural demand. The difference isn't scale, it's the interruption mechanism:
- Structural demand is driven by end-user behavior; killing it requires a substitute or a behavior change
- Policy demand is driven by budget process, and a single legislative deadlock can defer a full year
The live example: the Executive Yuan's special statute for defense-autonomous unmanned vehicle procurement proposes an NT$210 billion special budget and is still under legislative review, with opposition versions arguing for annual appropriations instead. The same demand, routed through a special budget versus annual budgets, gives suppliers two completely different capacity-planning rhythms. (in Chinese)
Verdict: partially holds, but the nature must be labeled. Treating "NT$44.2 billion has been appropriated" as evidence of structural demand is this sector's most common analytical error.
Criterion three: high switching cost ✅
This is the one criterion that genuinely holds, and it holds solidly.
Drone switching cost doesn't come from technical lock-in. It comes from certification:
- The Blue UAS bar is written into statute. Its basis is Section 848 of the FY2020 NDAA, prohibiting DoD procurement of drones made in China or containing Chinese critical components — the named components cover flight controllers, radios, data transmission, cameras, gimbals, ground control systems, operating software, and data storage. Essentially everything on a drone that stores or transmits data. Applicants submit a full chip-to-software BOM for review, with every covered-country source replaced.
- Once obtained, the position is hard to displace. Thunder Tiger's Overkill FPV cleared Blue UAS in September 2025 — Taiwan's first and still Asia's only. Roughly 35–40 platforms are on that list globally. (in Chinese)
- Taiwan adds a time lock. Under the Remotely Piloted Drone Management Regulations, from 1 December 2027, drones with navigation equipment used by legal entities in activities requiring application must carry a cybersecurity assessment report from a Ministry of Digital Affairs–recognized body plus commodity or type certification. Past that line, uncertified hardware is out of the corporate market.
Certification fits this criterion's definition precisely: long cycle, high cost, and an exclusionary list once passed. Same moat logic as semiconductor process qualification.
Verdict: holds. But note what it protects — the holders, not the sector. Taiwan has one seat on Blue UAS, so this moat currently shelters very few.
Criterion four: long-term institutional holding — not assessed here
This requires stock-level ownership data (foreign holding percentage, consecutive net-buy weeks, target price moves), and this article deliberately avoids single-stock analysis.
One principle worth stating up front: policy-theme stocks tend to have a different ownership shape than the long-term institutional accumulation seen in AI supply chain names. The former shows short-term flows tracking news; the latter shows the tell of adding into bad news. Using this criterion means distinguishing accumulation from news-driven noise — which requires a long observation window, not a single week's net buying.
Three risks to understand first
Risk one: winning a bid is not revenue (acceptance risk)
This is the sector's most expensive lesson, and the case landed in July 2026.
The Army procured 26 fixed-position counter-drone systems with a budget of NT$989.63 million, awarded to Tron Future Tech at NT$987.81 million. The spec was explicit: detect a 10 cm² low-altitude target at 6 km and jam it at 4 km.
The result was failure at initial acceptance and both re-tests, with formal rejection at the second re-test review on 9 July 2026. Defense Minister Wellington Koo explained the disposition to the legislature: the military terminated under the contract, paid nothing, and returns the remaining funds to the treasury, forfeiting the roughly NT$98.78 million performance bond (10% of the award) with late-delivery penalties calculated separately at up to 20% of contract value. The vendor faces debarment and the project goes back out to tender. (both in Chinese)
A nearly NT$1 billion award recognized NT$0 of revenue and cost about NT$100 million. For this industry that isn't a freak event — it follows from the fact that specs are measured, not written, especially in counter-UAS, payloads, and long-range comms where Taiwan is structurally weak.
When you see "company X wins NT$Y drone contract," the second question is always: what are the acceptance conditions, how many delivery tranches, and what's the vendor's acceptance record?
Risk two: budget rhythm
Eighty percent of demand from the public sector means revenue visibility is tied to legislative process. Special statute versus annual appropriation, cross-party negotiation, line-item cuts and freezes all propagate into shipment schedules. In the case above, the legislature simultaneously cut NT$10 million and froze another NT$10 million of the 2026 budget pending a written report — that granularity of intervention doesn't exist in ordinary commercial orders.
Risk three: distributed awards compress single-vendor upside
Policy aims not just to buy equipment but to cultivate an industry. The National Police Agency's 50-system, NT$3.67 billion critical infrastructure counter-drone tender shows the design: systems are graded A, B, and C by verified capability, the top three in each grade deliver proportionally, and up to 9 vendors can win so that qualified suppliers all get a share. (in Chinese)
Good for the industry, compressive for any single vendor's revenue upside. "National team" reads as concentration in investment language; actual policy design often deliberately distributes.
The bottom line
Measured with one ruler:
| Criterion | Verdict | In one line |
|---|---|---|
| Supply chain chokepoint | ❌ | Taiwan sits at the most substitutable layer |
| Structural demand | ⚠️ | Demand is real, but ~80% is policy, not structural |
| High switching cost | ✅ | Certification is a real moat — sheltering very few |
| Long-term institutional holding | — | Requires stock-level data; not assessed here |
One of four holds. That doesn't mean the sector lacks opportunity — it has a clear geopolitical window, NT$44.2 billion of policy resource, and 21x export growth. But its current position is not 2023's AI supply chain; it's an earlier stage: the chokepoint isn't taken, demand still rests on budgets, and the moat is in very few hands.
Bringing semiconductor or AI-supply-chain expectations to it misreads "early" as "taking off."
If I tracked only one pair of numbers, it would be exports as a share of total output (rising from 23% means Taiwan is genuinely holding international ground) and Taiwan's seat count on the Blue UAS list (1 becoming 2 or 3 means the moat is spreading). Both are far more honest than output growth rates.
This is industry structure analysis, not investment advice. No securities are recommended; companies named are cited to illustrate industry mechanics. Evaluate and bear your own risk.
References
Framework and structure
- CNA — Legislature reviews drone statute; MOEA urges support for the Executive Yuan version (localization rates, competing statute versions) (in Chinese)
- CIO Taiwan — Thunder Tiger obtains Blue UAS certification (BOM split, Blue UAS bar) (in Chinese)
- Remotely Piloted Drone Management Regulations — Laws & Regulations Database (the 1 December 2027 cybersecurity and type certification requirement) (in Chinese)
Risk case
- LTN Defense — Army counter-drone system fails re-test; NT$980M contract to be terminated and retendered (spec requirements and test process) (in Chinese)
- LTN Defense — Army terminates counter-drone contract; Koo says remaining funds return to the treasury (termination, bond forfeiture, penalties) (in Chinese)
- UP Media — National Police Agency opens tender for 50 critical infrastructure counter-drone systems (graded award mechanism and budget) (in Chinese)
On this site
Loading...