Taiwan's documented programme of conscription extension, budget growth, reserve reform and civil-defence expansion since 2022. Its central unresolved question is the balance between conventional platforms and asymmetric systems, and that argument is conducted in public.
Summary
Taiwan has undertaken a sustained programme of defence and resilience reform since 2022. Conscription was extended from four months to one year, effective from 2024, reversing two decades of reduction. Defence spending has risen substantially as a share of the budget, including a special defence budget passed in May 2026 well above the previous baseline. Reserve mobilisation was reorganised under a dedicated agency, and civil-defence and disaster preparedness — long built around typhoons and earthquakes — has been extended toward broader contingency planning. Running through all of it is an unsettled strategic argument: whether to invest in conventional platforms that also serve peacetime presence and deterrence signalling, or in large numbers of cheaper asymmetric systems. Both approaches have advocates in Taiwan and among its partners, and procurement has followed both. The external supply picture moved in two directions at once during 2026: the largest package ever notified was entered in December 2025 and Abrams deliveries completed in April, while a further $14bn package was publicly held back as a negotiating element and several cases stalled on an appropriation dispute in Taipei rather than on production. What sharpened was not scarcity but the gap between what has been notified and what has been fielded.
Why it matters
This is the variable Taiwan actually controls. The pressure campaign is not something Taipei can decide to stop; the composition, size and readiness of its own forces and the resilience of its civil systems are. The programme is also the clearest available evidence about how Taiwan's own leadership and public assess their situation — conscription extension in particular carried a real domestic political cost, and paying it is more informative than any statement. For the module's other views it is the link between the military picture and the civil-resilience material, because the same programme covers both.
Stated objectives
What the participants say. Each entry cites where the position was stated; Vigil records that it was made, not that it is true.
Taiwan states its objective as a credible defence sufficient to deny a rapid outcome and to make coercion unattractive, alongside civil resilience adequate to sustain the society through disruption.
Successive governments have committed to raising defence spending as a share of output and have legislated special budgets to do so.
Assessed objectives
Vigil's reading, graded. Kept structurally apart from stated positions so an assessment can never be published as someone's declared aim.
The conscription extension is the programme's most significant element because it was politically costly and is hard to reverse — a better indicator of Taipei's own threat assessment than spending figures, which are easier to announce than to execute.
Procurement continues to reflect an unresolved argument rather than a settled doctrine, and the resulting mixed portfolio is a political outcome as much as a military one.
Taiwan's defence problem is not one problem, and the headline percentage of GDP measures the part that is least in dispute. Spending LEVEL is rising and is politically settled in direction: 3.01 per cent in the 2027 proposal against a stated 5 per cent by 2030. COMPOSITION is contested and is where the 2026 argument actually happened — the legislature funded foreign procurement and cut indigenous uncrewed systems, which is a choice about what kind of force to build rather than how much to spend. EXECUTION is a separate constraint again, because appropriated is not disbursed. PRODUCTION binds outside Taiwan's control entirely, in American factories and order books. DELIVERY is measurable and improving. Any account that reduces these to one number will misdescribe at least three of them.
Public uncertainty about US transfer decisions has increased the weight placed on domestic production and on asymmetric systems Taiwan can build itself, though the shift is early and its durability is not established. The evidence cuts both ways and is stated that way deliberately: notifications reached a record in December 2025 and the delivery backlog fell through the first half of 2026, so any account resting on transfers simply slowing does not survive the published figures.
Primary forces
- Republic of China Armed Forces Primary force · Confidence: High · as of 2026-06
- All-out Defense Mobilization Agency Primary force · Confidence: Moderate · as of 2026-06
The reserve and mobilisation reorganisation is the programme's least visible and most structurally significant component.
Supporting & external
- ROC Army Supporting force · Confidence: High · as of 2026-06
- Taiwan Coast Guard Administration Supporting force · Confidence: Moderate · as of 2026-06
Coast guard expansion is part of the response to law-enforcement-framed pressure around the outlying islands.
- US Indo-Pacific Command External supporter · Confidence: Moderate · as of 2026-06
Training and materiel relationship under the Taiwan Relations Act framework. Recorded as a supplier relationship, not as a commitment to participate in any contingency.
Reassessment after 2022
The scale of the August 2022 exercises and the war in Ukraine together prompted a public reassessment of Taiwan's force structure, reserve system and civil preparedness.
Conscription extended
Compulsory service was extended from four months to one year — the programme's most politically costly decision and the clearest signal of how the threat is assessed domestically.
Budget growth, and a cut with a shape
Defence spending rose as a share of the budget, culminating in a special defence budget passed in May 2026 at NT$780bn against a NT$1.25tn request. The reduction was compositional rather than proportional: the Legislative Yuan removed approximately NT$470bn of domestically produced uncrewed-systems funding while leaving foreign procurement intact, at NT$300bn for a first tranche and NT$480bn for a second. Reading the outcome as general austerity misses what was actually chosen.
The supply question sharpens, in both directions
Notification and delivery moved opposite ways. In December 2025 the United States notified Congress of more than $11bn in new cases — the largest package ever notified for Taiwan — while in May 2026 President Trump said a further $14bn package was unapproved and described it as a negotiating element with Beijing, prompting President Lai to restate that arms purchases are the most important deterrent available. Meanwhile deliveries continued: the Abrams case completed in April 2026 and the notified-but- undelivered backlog fell from roughly $32.0bn to $29.72bn. Individual cases run far longer than the aggregate implies, and 500 Stinger missiles took about a decade from notification to delivery.
The appropriation deadlock, and where it reaches
Having lost the uncrewed-systems funding in May, the Cabinet proposed a standalone NT$210bn drone budget on 18 June 2026 covering reported procurement of 1,446 reconnaissance drones, 208,200 coastal attack drones and 1,320 uncrewed surface vessels. Opposition parties voted to delay review on 26 June and committee review closed on 27 July without agreement. The dispute is not confined to domestic programmes: the June 2026 review of the US arms backlog names NASAMS and the Tactical Mission Network as cases held up by the same special procurement deadlock, which makes a legislative argument in Taipei a documented cause of delay in an external deterrence pipeline. Reserve training was separately standardised at fourteen days in June, with a reported 93 per cent rate of matching reservists to their original posts.
The first trillion-NT proposal, and the chamber that decides it
The Executive Yuan approved a 2027 defence budget of NT$1.1225tn on 20 August 2026 — the first above NT$1tn, about 3.01 per cent of projected GDP and roughly 18 per cent above the 2026 figure — and sent it to the Legislative Yuan. It is a proposal, and the same chamber cut the May 2026 special budget request by more than a third and has not passed a proposed NT$210bn drone procurement bill. Taiwan ran its first live-fire FPV drone exercises on 17 August, and an official audit has reported technical shortcomings in counter-drone development. The reform's binding constraint remains appropriation and delivery rather than declared intent.
Outcome and present status
Ongoing and genuinely unresolved. The programme has delivered real, dated changes — longer conscription, larger budgets, a reorganised reserve system, expanded civil defence — and has not settled the strategic argument underneath them. Whether the resulting force is the one Taiwan needs is contested among Taiwan's own analysts and among its partners, and this module does not adjudicate it. What can be stated precisely is where things stop. A capability reaches Taiwanese forces through seven distinct states, and the module's standing discipline is that none of them implies the next. US NOTIFICATION: Congress is notified of a Foreign Military Sales case — more than $11bn in December 2025, the largest ever. TAIWANESE APPROPRIATION: the Legislative Yuan must fund the purchase, and in May 2026 it funded the foreign tranches at NT$300bn and NT$480bn while removing roughly NT$470bn of domestic uncrewed systems; the NT$210bn drone bill proposed in June to replace that funding had not passed by late August. CONTRACT: cases require signed contracts and first payments, and in March 2026 the HIMARS schedule was renegotiated against Lockheed Martin contract terms rather than cancelled. PRODUCTION: American capacity binds independently of either government — Paladin howitzers notified in December 2025 are reported delivering in 2030. DELIVERY: the notified-but-undelivered backlog stood at roughly $32.0bn at the end of 2025 and $29.72bn by mid-2026, and the Abrams case closed in April. FIELDING: delivered equipment must be absorbed, crewed and integrated. READINESS: demonstrated only in exercises, and Han Kuang 42 is the most recent test Vigil records without assessing. The instructive cases are the ones that stall at different rungs. NASAMS and the Tactical Mission Network are reported held at APPROPRIATION by the Legislative Yuan procurement deadlock — an American system delayed by a Taiwanese political dispute. Paladin is held at PRODUCTION, which no decision in Taipei affects. The drone programme is held at APPROPRIATION for a domestic system that needs no American release decision at all. Three delays, three different causes, and treating them as one story about slow transfers is the error this ladder exists to prevent.
Observed adaptations
Extension of civil-defence planning from its long-standing natural-disaster basis toward broader contingency preparedness, using structures that already existed and are exercised annually.
Increased emphasis on domestically produced systems following the 2026 transfer slowdown, reducing exposure to external release decisions.
Coast guard capability growth as a response to law-enforcement-framed pressure, matching the framing rather than escalating past it.
Political implications
Conscription and budget decisions are taken in a divided legislature, so the programme's pace is a domestic political variable as much as a strategic one.
Military implications
The conventional-versus-asymmetric argument determines what Taiwan's forces would actually be able to do, and it is unresolved.
Economic implications
Defence spending growth and domestic production emphasis both have industrial-policy consequences beyond the military.
Linked events
Locations
Places this module already publishes as map markers, listed because the campaign is about them. Not an operational area.
What remains unknown
- Readiness, stockpile levels and unit manning are not published by this module and are not the basis of any claim here.
- Budget execution data lags appropriation substantially, so announced spending overstates delivered capability at any given moment.
- Reserve system effectiveness after the conscription change has not been independently assessed and is contested among analysts.
- The unfilled 2026 arms package is reported rather than officially detailed, and its scope is not established.
Readings the evidence also supports
- The conscription extension can be read as a considered response to a deteriorating assessment, or as a measure taken substantially to reassure partners who had publicly questioned Taiwan's commitment to its own defence. Both were argued at the time and the decision is consistent with either.
- Budget growth can be read as capability building or as a political signal whose delivery will lag well behind its announcement, as execution data has repeatedly shown. The two are distinguishable only in retrospect.
Listed because the record's own assessment is not the only one its sources permit — not as a hedge on the assessment above.
Does appropriation become expenditure, and then capability?
Appropriation and expenditure are different figures, and both differ from fielded capability. Execution rates are published and are the meaningful measure. The NT$210bn drone bill proposed in June 2026 had not passed as of late August, and NASAMS and the Tactical Mission Network are reported held by the same procurement deadlock.
The conventional-versus-asymmetric procurement balance
Visible in what is actually contracted rather than in what is argued.
Conscription intake and retention
The extension's real effect depends on throughput and on whether trained reservists are usable, neither of which follows automatically from the policy.
Is the notified-but-undelivered backlog falling?
The backlog is the figure that reconciles notification with delivery, and it is published monthly: roughly $32.0bn at the end of 2025, $29.72bn by mid-2026 after the Abrams case completed. Watch the aggregate and named case milestones rather than announcement values, and remember that individual cases run far longer than the aggregate implies.
See the force running this in Explore Interactive: select this record and follow what connects to it.
Assessed as of2026-08Last reviewed2026-08-28