Global Defence Spending: The Rearmament Cycle and the Industrial Base Behind It

The answer is that the present rearmament cycle becomes durable industrial demand only where political commitments pass into legal authority, funded obligations, firm multi-year orders, qualified production, accepted deliveries and support. Those states move at different speeds. SIPRI's $2.887 trillion estimate proves the scale of the financial input; NATO's 2035 framework proves political direction; neither proves what has been ordered or fielded. The harder public evidence is further downstream: whether specifications remain stable, contracts endure, equipment passes acceptance and a support system exists. On that public-record test, the available evidence indicates that money is arriving faster than the published conversion record, although some programmes show that conversion can work.S01S06S07S14S15

That distinction is more than accounting hygiene. Defence expenditure includes people, operations, infrastructure, research and equipment, and national definitions are not identical. A budget can rise without a matching rise in industrial orders. A legal instrument can authorise finance without disbursing it. A contract can exist before serial output is qualified. A delivery announcement can precede ownership transfer and customer acceptance. Initial operational capability can follow years later, and sustainable availability can remain outside the public record. Treating these steps as synonyms produces false confidence precisely when companies and public buyers are making long-lived capacity decisions.S03S04S07S14

Three contemporary records expose the gap. The Hague agreement points towards 5% of GDP by 2035, but it explicitly combines at least 3.5% for core defence requirements with up to 1.5% for wider defence- and security-related investment. NATO's Ankara declaration then states more than $50 billion in new procurements, yet the public component record mixes investment plans, contracts, initiatives, a lease, quantities without values and conflicting delivery language. By contrast, the multinational tanker fleet offers smaller numbers but a clearer chain: formal launch, firm orders, ownership transfer, delivered aircraft, initial operational capability (IOC), full operational capability (FOC) and ongoing support. The documentary quality becomes stronger as the headline becomes more specific.S06S26S08S09S10S15S16S17

For a buyer, supplier or capital allocator, the decision is therefore not whether defence spending is rising. It is where the increase is durable and specified enough to justify a bid, capacity investment, partnership or deeper diligence—and which public conversion dependency could still break the case. The report does not rank forces or infer classified readiness. It follows public money and programme evidence only, keeps alliance, EU and national aggregates separate, and asks what each document actually proves.S03S04

The long arc begins with a missing point. SIPRI cannot calculate a 1991 world total because it lacks a Soviet Union figure, so the series starts here at $1.415 trillion in constant 2024 dollars in 1992 rather than filling the gap. Real global spending then moved down through much of the 1990s. By 2001 the comparable total was about $1.297 trillion. This was the peace-dividend reset: not the disappearance of military industry, but a lower aggregate flow and a different expectation about demand duration. Capacity, skills and supplier structures inherited from the Cold War did not instantly vanish, yet the economic signal facing much of the industrial base had changed.S02S03

After 2001, spending rose through an expeditionary era in which operational demand, personnel, sustainment and protected mobility mattered alongside major platforms. The financial series records the resource input; it does not tell us how efficiently those resources became capability or how inherited stocks changed the result. That caveat matters across decades. A country drawing on installed fleets, existing depots and mature supply arrangements may obtain a different output from the same annual expenditure as one creating a new sovereign industrial capability. Spending is a flow through a system, not an inventory of what that system already owns.S02S03S04

The next breakpoint came in 2014. World military expenditure stood near $1.936 trillion in constant 2024 dollars as NATO leaders at Wales reaffirmed a 2% of GDP guideline and a separate expectation that more than 20% of defence budgets should go to major equipment and associated research and development. The dual test was revealing: how much was spent and how it was spent were already different questions. The commitment created direction, but implementation remained national, and a percentage guideline could not specify common equipment, aggregate orders or guarantee that industry would retain demand for long enough to expand.S02S05

The full-scale Russian invasion of Ukraine in 2022 then changed the industrial problem. The comparable world total rose from about $2.299 trillion in 2022 to $2.771 trillion in 2025, a real increase of roughly one fifth in three years. Europe recorded the sharpest broad regional acceleration, but the rise was not European alone: spending outside the United States increased 9.2% in real terms in 2025. The shock exposed the difference between drawing from inherited stocks and sustaining replenishment, yet aggregate expenditure still cannot reveal classified stock conditions, production bottlenecks or operational availability. Public analysis can identify the conversion challenge without mapping vulnerabilities.S01S02

The period from 2025 to 2035 is therefore a proof period, not simply the next budget decade. Hague raises the political ceiling and requires annual national plans, with a formal review in 2029. NATO's production plan says what must improve around demand aggregation, capacity, standards and cooperation. The outcome will be visible only in later records: national authority, multi-year obligations, stable specifications, qualified output, equipment accepted by customers and support sustained after delivery. If those records accumulate, the current spending cycle becomes an industrial cycle. If they do not, part of the political ambition will remain stranded between annual budgets and usable capability.S06S07

What has changed since the first post-2022 surge is not only the volume of spending but the attempt to give demand duration and industrial form. EDA's 2024 record is one useful checkpoint: EU member states reported €343 billion of defence expenditure, including €88 billion of equipment procurement after a 39% annual rise and €13 billion of defence research and development. Those are recorded flows, not contract backlog or delivery value. EDA's 2025 figures—€381 billion total, equipment procurement above €100 billion and €17 billion of R&D—remain projections. Keeping the two years in their correct states avoids converting an institutional outlook into realised demand.S11

SAFE shows a financing mechanism moving through several more precise stages. The regulation creates legal loan capacity of up to €150 billion and sets eligibility and reporting conditions. By the retrieval cutoff, maximum amounts across 18 approved national plans summed to €130.16 billion. The Commission also recorded €7.95 billion of first pre-financing to four member states by 10 July 2026. This is genuine progression from legal capacity to approved allocation to cash movement. It still stops before the supplier chain: a loan tranche is not evidence of an equipment award, a firm production order, accepted delivery or readiness. The next public documents have to carry that burden.S12S13S25

NATO's production plan addresses the commercial reason this progression matters. Suppliers expand against credible duration, common requirements and orders, not an aggregate percentage alone. The plan calls for clearer demand signals, multinational and multi-year procurement, better capacity information, engagement across the lifecycle, attention to critical supply chains, and stronger standardisation and interoperability. Those mechanisms can reduce fragmentation and allow investment to be spread across a longer run. But the plan itself is policy. Whether it changes behaviour must be tested against contract duration, configuration stability, qualified throughput and accepted output.S07

Ankara is the sharpest warning against shortcut arithmetic. The summit declaration calls more than $50 billion a new-procurements aggregate, while the forum and speech records link it to production, cooperation, co-production and joint procurement. The public components include investment plans, mixed initiatives, disclosed contracts, a private lease, quantity-only items and value-undisclosed items. They do not provide a component table with common lifecycle states, inclusion rules or overlap tests. No ordered share or residual is therefore calculated. The safe result is more informative: a declared aggregate whose industrial meaning remains unresolved.S26S08S09S10

The multinational tanker fleet demonstrates both the value and the limit of milestone evidence. An initiative began in 2012 and the programme was formally launched in 2016. Firm orders grew as participants added requirements. NSPA records describe ownership transfer in the delivery process, providing evidence beyond a press announcement. The first aircraft arrived in 2020; the fleet reached IOC in 2023 and FOC in March 2026. NATO then published two statements that do not reconcile: on 7 July it announced delivery of the tenth aircraft; a programme page updated 9 July still recorded nine in service and one expected in summer. The report therefore leaves the tenth aircraft's ownership transfer, acceptance and service-entry status unresolved. Support continues through the multinational structure. The wider chain remains useful because each proven state is kept separate from the one that is disputed.S09S15S16S17

Even a functioning chain takes time. GAO's 2026 assessment found that 43 major US programmes which had reached IOC took 132 months on average from programme start; 17 programmes still awaiting IOC with published estimates averaged 149 months. Fifteen further programmes had no public IOC date or a classified date. These are not global production averages, and IOC is not a delivery date. They are nevertheless a disciplined reminder that fiscal calendars and capability calendars are structurally different. A political system can add money in one year; a complex programme may need a decade to pass through design, qualification, production, integration, acceptance and service entry.S14

The 2025 map is broad but not uniform. Europe reached an estimated $864 billion and rose 14% in real terms. Asia and Oceania reached $681 billion and rose 8.1%. Africa reached $58.2 billion after an 8.5% rise. The Americas remained the largest broad region at $1.065 trillion but fell 6.6% after the United States' exceptional 2024 increase. The Middle East was estimated at $218 billion and broadly flat, with SIPRI marking the level as uncertain. These are current-dollar levels paired with real annual changes; they answer different questions and should not be placed on one additive scale.S01

Economic burden tells a different story again. The Middle East had the highest regional average at 3.8% of GDP, followed by Europe at 3.2%, Africa at 1.8%, Asia and Oceania at 1.6% and the Americas at 1.2%. A high burden can coexist with a smaller nominal market, while a very large economy can carry the greatest nominal spend with a lower average share. SIPRI also excludes countries where data are insufficient and uses its own regional taxonomy. The sensible comparison is therefore within each panel and definition—not a league table of military effectiveness.S01S02S03

North America supplies unmatched fiscal scale and the strongest public oversight trail, but it also shows the conversion problem in plain view. GAO's programme cohorts put elapsed time around a decade, while the US industrial strategy implementation plan links resilient supply chains, workforce readiness, flexible acquisition and allied cooperation. That framing matters because appropriations alone cannot qualify a supplier, stabilise a design or create an experienced production workforce. The region offers abundant financial and audit data; the buyer still has to trace the specific programme from authority through obligation, order, acceptance and support.S14S18

Europe is the central test of aggregation. NATO supplies requirements, standards and a political framework; EU institutions add expenditure data, regulation and financing; member states retain core budget and procurement authority. That creates both leverage and the risk of double counting. EDA's equipment-procurement expenditure, SAFE's loan stages and NATO's percentage commitment overlap with national systems but do not represent one pot. Progress should be judged by whether common requirements become stable multinational orders and accepted equipment, not by adding alliance, EU and national headlines.S06S07S11S12S13S25

Asia presents separate accounting and industrial models. Japan's FY2026 document distinguishes annual expenditure from contract-basis commitments. India's official review separates approvals above ₹3.84 lakh crore, signed contracts of ₹1.82 lakh crore and about ₹1.2 lakh crore of capital expenditure; a later administrative release reports ₹1.78 lakh crore of defence production in FY2025–26, with public and private contributions split. South Korea's budget separately identifies total defence and force-improvement programmes. These native categories are valuable precisely because they are not forced into a converted global order total. They show how much meaning can be lost when a budget headline is detached from its jurisdiction's accounting state.S19S20S21S24

The Gulf links acquisition to domestic industrial policy. Saudi Arabia's official report placed its military-industry localisation indicator at 24.89% in 2024 against a target above 50% by 2030. That is a programme measure, not proof of sovereign design authority, qualified local content or accepted output, but it establishes what the public owner intends to monitor. Australia offers a different model: a transparent long investment plan that connects strategy and capability priorities while leaving individual contracts and delivery milestones to later records. Africa belongs in the global picture because its real spending rose, yet sparse and excluded observations demand restraint. None of these regions should be used as filler for a universal narrative.S01S22S23

The industrial bottleneck is best understood as a chain rather than a list of scarce items. It begins with durable specified demand. A supplier deciding whether to add equipment, people or a second source needs more than a national percentage pledge: it needs a credible requirement, funding path, order duration and configuration. NATO's production plan addresses aggregation and multi-year procurement for this reason. The US strategy reaches a similar conclusion from another direction by tying resilient supply chains to workforce, acquisition practice and allied industrial cooperation. Demand quality and production quality are joined.S07S18

Authority and finance form the next gate. Different jurisdictions use requests, authorisations, appropriations, obligations, loan agreements and special funds in different ways. SAFE makes the distinction unusually visible because the legal ceiling, approved maximum allocations and first cash tranches can be observed separately. None is yet a supplier order. The same discipline applies to national plans and the Hague framework: a political commitment shapes expectations, while a binding obligation creates a different commercial signal. A capacity case should specify which document carries the demand and how long that authority lasts.S06S12S13S25

Specification and qualification then decide whether aggregated demand is truly repeatable. Common standards can widen a production run; diverging configurations can fracture it into small batches. Materials, components, software and sub-tier suppliers must satisfy programme requirements before assembly volume becomes useful output. Workforce, tooling, test equipment and process control have to mature together. Public strategy documents can name these functions without identifying sensitive facilities, stock levels or exploitable shortfalls. The relevant commercial question is whether the programme has a qualified system and an evidence-backed ramp, not whether a press release names a large nominal capacity.S07S18

Test and acceptance are where industrial output becomes customer evidence. The tanker programme is instructive because NSPA describes ownership transfer in connection with delivery, while NATO separately records service-entry milestones and the fleet in service. This prevents a forthcoming delivery from being mistaken for accepted capability. It also shows why sustainment belongs inside the conversion chain: a NATO-owned multinational fleet depends on management and support after the aircraft arrive. Training, spares, maintenance, configuration control and upgrades may not dominate a summit headline, but they shape whether accepted equipment remains usable.S15S16S17

The status chain also explains why one metric cannot govern every market. Equipment-procurement expenditure is closer to industrial demand than total military expenditure, but it is still a period flow rather than backlog. Signed contract value is stronger, but may cover services, options or long delivery profiles. Reported production value establishes output under the owner's definition, not acceptance or readiness. Delivered quantity is stronger again, but it may not reveal sustainment performance. The decision-maker needs a small set of programme-native proofs and a visible statement of what remains unknown.S03S04S11S15S20S21

That leads to a practical capacity test. First, identify the buyer and the exact capability requirement. Second, locate legal and fiscal authority in the relevant jurisdiction. Third, confirm whether a firm, funded, sufficiently long order exists and whether its specification is stable. Fourth, trace qualification, test and customer acceptance. Fifth, look for a public support arrangement that survives the delivery event. A positive answer at every stage does not guarantee returns or operational performance, but it turns a spending theme into a decision-ready industrial case. A gap tells the buyer where deeper diligence belongs.S07S13S15S17

The rearmament thesis should now be judged by progression, not by another round of bigger announcements. SIPRI's next annual releases will show whether real growth remains broad and whether the current cycle survives a slower fiscal year. NATO's annual plans and 2029 review will show how the 5% framework is composed. EDA can test whether projected equipment-procurement and R&D spending becomes reported outturn. The Commission's SAFE timeline can show whether financing advances into supplier obligations. Programme owners can show accepted deliveries and support milestones. Oversight bodies can show whether elapsed times begin to fall.S01S02S06S11S14S15S25

Each signal has an owner, a unit and a monitoring window. That is deliberate. Where no repeatable public measure exists, the report asks for a document rather than inventing a score. Ankara's unresolved aggregate, for example, should be revisited only if NATO or national buyers publish a component table with lifecycle states. Readiness remains outside the framework unless a competent public authority supplies a comparable definition and series. The aim is not to predict a classified operating picture; it is to identify the next public event that would strengthen or weaken the industrial case.S26S08S09S10S14

The most important confirmation would be simultaneous improvement on both clocks: funded multi-year orders and stable standards on the budget side, then qualified output, acceptance, delivery and sustainable support on the capability side. The thesis weakens if real spending narrows to a few states, new money remains outside industrial categories, financing stalls before contract, or delivery and elapsed-time data improve broadly enough to show that industry is catching up. That test keeps the argument falsifiable through 2030 rather than turning 2035 ambition into a foregone conclusion.S01S06S07S11S14S15S25