Defence Budgets Are Up. Why Are the Weapons Still Late?

Once the money is approved, what stands between it and a finished weapon?

A defence budget buys orders before it buys weapons. In the United States, the annual National Defense Authorization Act sets policy and spending ceilings, and appropriations bills provide the money; the military turns that money into contracts, and contractors slot the work into their production lines. The early stages are budgetary and contractual, and the numbers can change within a year. The last stage depends on what the defence industrial base can actually make, and it cannot be hurried.

When capacity lags, the extra money simply joins a queue. The United States bought two Virginia-class attack submarines a year from FY2011 to FY2024, yet construction has never reached two a year. In testimony to Congress in April 2025, the Navy said the build rate had come close to 1.9 a year for about three years before falling after the pandemic, to 1.13 a year by the end of 2024. The Congressional Research Service describes the result as a growing backlog of boats procured but not yet built.

Nor is the problem confined to one programme. The US Government Accountability Office reported in March 2025 that the Navy’s shipbuilding budget had nearly doubled over two decades without any increase in the number of ships. New ships keep arriving late and over budget, and the billions recently put into shipyard facilities and workforces have yet to bring the improvements expected.

Where exactly does capacity get stuck?

Mostly in unglamorous places: drydocks and plant, welders and other skilled trades, and foundational processes such as castings and forgings. Orders stayed low for decades after the Cold War, and these capabilities withered with them. Naval Sea Systems Command said in 2024 that the submarine manufacturing base had shrunk to a third of its capacity of 30 years earlier, while output had to nearly double, requiring an extra 3.5 to 4.5 million hours a year of submarine module production and outfitting.

People are the hardest part. Workers can be hired; experience cannot. The Navy told Congress that hiring at the shipbuilders rose 41% in 2023 and beat its targets in 2024, yet submarine output did not rise as expected; it named workforce challenges, first-time quality, and material and supplier delays among the main causes of the fall in production. The GAO likewise found shipyards struggling to replace experienced, skilled workers with new ones.

The Navy has also conceded that it, the shipbuilders and the wider supply chain underestimated how hard it would be to move from the low production rates of the post-Cold War years to one Columbia-class and two Virginia-class boats a year. Taiwan’s F-16V order has run into production-line trouble of its own, discussed in the last section.

How can one small part hold up a whole production line?

Because a weapon is built by thousands of suppliers, and many of them have no substitute. The US Navy says its submarine industrial base has about 16,000 suppliers, 11,000 of them sub-tier vendors, that is, the suppliers’ suppliers. According to the Congressional Research Service, about 70% of the critical suppliers for submarine construction are sole sources. However fast the prime contractor expands, one sole-source supplier that cannot deliver leaves the work downstream waiting.

Drones show the same pattern more visibly. A policy issued in March 2026 by Japan’s Ministry of Economy, Trade and Industry (METI) notes that more than 90% of the drones used for industrial purposes in Japan are foreign-made. Motors, electronic speed controllers and batteries must meet drone-specific requirements, so general-purpose parts are hard to substitute; a particular country holds a large share of their supply, and disruptions have already occurred abroad. A drone designed and assembled in Japan still stops production when those parts stop arriving.

Large numbers of commercial drones have been modified for military use, so defence production in effect draws on the same parts as the commercial market and inherits its concentration. Dual-use supply keeps volumes high and prices low, and brings the concentration risk in with it. A single point of failure of this kind usually sits further down the supply chain, the same problem as in the chip supply chain: a review that stops at first-tier suppliers will not find it.

Why don’t manufacturers expand first and wait for the orders?

Because new capacity takes years to pay back, while defence orders are decided by governments one year at a time. Plant, equipment and trained workers become idle costs as soon as orders fall, so whether a company invests in production lines and suppliers depends largely on whether it can see demand several years ahead. Hideki Tsuchimoto, a former commissioner of Japan’s Acquisition, Technology and Logistics Agency, has argued that the government should publish clear procurement plans and a medium- to long-term roadmap so that companies can work out when an investment will pay back (see how Japan is rebuilding its drone industry).

METI’s policy shows what happens when support stops short of production. Past government support for drones covered only airframe research and development, not investment in equipment to mass-produce key components, so a domestic production base never took shape; with limited domestic volumes and each maker designing its own parts, production stayed in small batches and prices stayed high. The new policy calls for industry-wide standardisation of key components and sets a target of a domestic mass-production base for about 80,000 drones by 2030.

The United States has also worked on the capital side. The revised Opportunity Zone rules put the main tax benefit after a ten-year holding period, rewarding money that stays in physical assets; one qualified opportunity fund used them to buy a 355-acre shipyard in Alabama, earmarked for submarine module production and training (see how the US is using tax incentives to rebuild shipbuilding capacity). A tax break can attract capital, but it cannot cure delays or cost overruns.

Besides spending more, what can governments do?

One option is to make orders bigger and steadier. In May 2025 the EU adopted SAFE (Security Action for Europe), an instrument offering up to €150 billion in long-maturity loans. To qualify, purchases must in principle be made jointly by at least two countries, a rule meant to build economies of scale and reduce the fragmentation of Europe’s defence industry; and components from outside the European Economic Area and Ukraine may make up no more than 35% of the cost of an end product’s components. Pooled orders give manufacturers more confidence to expand, and the origin rule in effect pulls capacity into the region.

Another is to add sources of production. Section 222 of the National Defense Authorization Act as reported by the Senate Armed Services Committee in June 2026 directs the Pentagon to enlarge its stock of air defence interceptors along several paths: cutting the cost of existing missiles and making them easier to manufacture; handing technical data packages to additional manufacturers, including commercial firms; sourcing interceptors from allies and partners while exploring co-production and licensed manufacturing to set up or expand US production lines; and identifying critical components. Britain and Ukraine have tried a different division of labour: an interceptor drone designed in Ukraine is mass-produced in Britain and supplied to Ukrainian forces (see who owns Ukraine’s battlefield knowledge).

A third is to turn purchases into domestic industry. Saudi Arabia is building up its own defence industry through procurement, technology transfer and co-production (see the Saudi–Turkish–Pakistani alignment); its General Authority for Military Industries reports that the localised share of military spending reached 24.89% at the end of 2024, against a target of more than half by 2030. Turkey’s drone, air defence, missile and naval industries have grown quickly, yet President Erdoğan still presses allies to lift their restrictions on defence trade with Turkey (see NATO’s cracks before the Ankara summit), which suggests that even an established domestic industry can remain subject to other countries’ export licences.

All three take time. Loans have to become contracts, new sources have to be qualified, and Saudi Arabia’s localisation target is set for 2030. They change capacity several years out, not next year’s deliveries.

What does this mean for Taiwan?

Equipment bought abroad arrives on the seller’s production schedule. In August 2019 the US State Department approved the sale of 66 F-16V Block 70 fighters to Taiwan, valued at an estimated US$8 billion; the government spent about NT$247.2 billion on them between 2020 and 2026, and they were due to be delivered in batches by the end of 2026. According to Taiwan’s Central News Agency (CNA), the programme was delayed by factors including the relocation and restart of the production line and disruptions to the supply chain, and the first two jets did not land at Chihhang Air Base in Taitung until 2 October 2026.

Washington has noticed. In its review of the FY2027 National Defense Authorization Act, the Senate Armed Services Committee directed the Government Accountability Office to examine delays in US arms sales to Japan, Taiwan, South Korea and the Philippines, and how they affect denial defences along the First Island Chain (see the Senate version of the 2027 NDAA). Arms that are approved but not delivered weaken deterrence all the same, so delivery belongs among the tests of what a defence commitment is worth.

When production cannot catch up, stocks positioned in advance offer another route. The Senate version of the same bill would let the president transfer to Taiwan munitions and supplies from Pentagon inventory that are obsolete or surplus, intended as reserve stocks for Taiwan and already stored there; the final text depends on negotiations between the two chambers. The logic is that of asymmetric defence: shipping is the first thing a conflict disrupts around an island, and resupply depends on shipping. Asymmetric warfare needs large numbers of replaceable systems, which makes pre-war stockpiles and local production and repair all the more important.

CNA also notes that domestic capabilities such as the F-16 maintenance centre run by Aerospace Industrial Development Corp. (AIDC) in Taichung can cut upkeep costs and strengthen Taiwan’s ability to repair battle damage quickly in wartime. There is opportunity for Taiwanese industry as well: semiconductors, servers, sensors, unmanned vehicles, motors and batteries could all enter allied defence supply chains, though the bar is set not only by price and performance but by transparent component sourcing, cybersecurity, export controls and reliable delivery. When reading a defence budget, look beyond the headline figure to delivery dates, where the production lines are and where the critical parts come from.