Cheap Enough to Lose: Attritable Mass, Deep Magazines and the New Economics of Defence
Somewhere over Ukraine this month, a missile worth more than a suburban house will destroy a drone worth less than a used car. Multiply that exchange across thousands of engagements, and you arrive at the central problem of modern defence economics: the West has spent thirty years perfecting weapons it cannot afford to use against adversaries fielding weapons they can afford to lose.
That inversion is now reshaping procurement doctrine in Washington, London, Brussels and Canberra, and it is - we would argue - the most investable theme in defence technology today. It travels under an ugly piece of Pentagon vocabulary: attritable mass. Systems cheap enough to lose, numerous enough to matter and, the part investors consistently underrate, manufacturable enough to replace.
The theme has two halves. The first is about the weapons themselves: a shift from relying on small numbers of extraordinary platforms to fielding very large numbers of merely adequate ones alongside them. The second is about what stands behind them: factories. Because the binding constraint on Western military power is no longer design brilliance. It is production capacity.
The Ledger Doesn't Lie
Let’s start with what Ukraine exposed. When Russia invaded in February 2022, the US Army had not purchased a Stinger missile in eighteen years. The contracting moved fast - a US$624.6m undefinitised order was issued on 25 May 2022. The production did not. Seeker components were no longer commercially available, and the workforce that once hand-assembled the missile had largely retired. The May 2022 order carried an estimated completion date of 30 June 2026. Four years, for a weapon designed in the 1970s.
The pattern repeats across the inventory. The Center for Strategic and International Studies (CSIS) estimated that the US donated roughly a third of its Javelin stock in the war's opening months, yet production was just +14% over the following two years; by January 2025 transfers exceeded 10,000 missiles - four to five years of output at prevailing rates. Artillery is the starkest case: US 155mm production stood at 14,400 rounds a month at the invasion, while Ukrainian guns at their peaks reportedly fired 6,000 to 8,000 rounds a day - a year of pre-war American output consumed roughly every month. Washington nearly tripled output to 40,000 rounds a month by late 2024, but the 100,000-per-month goal slipped from October 2025 into 2026. The US Government Accountability Office put the problem in one line: at 2022 production rates, two million shells would have taken nearly twelve years to build.
Nor is it only Europe's problem. A series of CSIS wargames of a US-China conflict over Taiwan found that the United States would exhaust some of its key long-range precision munitions in less than a week. Allies, meanwhile, wait in the queue: Taiwan's 400 Harpoons, approved in October 2020, reached contract in April 2023 and finish delivering in March 2029 - more than eight years, for a missile designed in the 1970s. Behind all of this sits a supplier base that has consolidated from 51 major primes in the early 1990s to just five today - a decline of roughly 90%, and deepest in exactly the categories now in shortage: missiles and munitions.
This year supplied the freshest proof. US air and missile defences performed well against Iran's missile and drone salvos - so well that interceptor inventories began depleting again, just as CSIS’s Tom Karako had warned after 2025's 12-day war. As he put it in June: "now the complaint is that we aren't hitting bullets with bullets cheaply enough, and that we are running out of anti-bullet bullets". Weeks later the US Army placed its first domestic order for PAC-2 Patriot interceptors in more than 30 years - a $441.6m contract reported to be in support of operations against Iran. As this edition went to press, the Pentagon confirmed a memo - first reported by The Washington Post - giving industry leaders no more than 21 days to submit plans for faster delivery schedules or increased production, with US deputy defense secretary Steve Feinberg writing that "years-long development cycles are not acceptable". CSIS now estimates Patriot interceptor stocks have fallen by at least 65% since the war began. A three-decade-old design you can buy now beats a state-of-the-art one you cannot.
The Surge That Wasn't
The instinctive response - "money will fix it" - turns out to be wrong. A National Defense University study published in December 2025 examined which US munitions actually surged after 2022. The answer is uncomfortable: the production lines that accelerated were those already warm. Guided Multiple Launch Rocket System (GMLRS) output +40%, helped by more than 26,000 rockets contracted between 2017 and 2021; PAC-3 interceptor production doubled, helped by pre-war procurement and a plant expansion begun in 2019. Stinger's cold line and Javelin's minimally active one barely moved in the immediate aftermath, no matter how much money arrived. As the study concluded: "absent pre-crisis preparation, a swift post-crisis surge in U.S. munitions production cannot be guaranteed."
None of this should have surprised anyone. In 1982, a US Department of Defense mobilisation exercise called Proud Saber 83 found that "a six-month industrial surge would yield only a negligible increase in production", citing long-lead components and sole-source subcontractors. Decades later, an FY2016 study of 35 key munitions found 98% of critical components in the lower supply tiers were single- or sole-sourced. Crisis money can buy capacity, but it cannot buy back time. Surge capacity is a stock, not a flow. If it is to matter when the shooting starts, much of it must exist before the shooting starts.
Production is Deterrence
Doctrine has finally caught up with arithmetic. The Pentagon's first-ever National Defense Industrial Strategy (2024) and its implementation plan elevate four priorities - resilient supply chains, workforce readiness, flexible acquisition and, most tellingly, economic deterrence. The Replicator initiative promised thousands of attritable autonomous systems inside two years; reporting after its August 2025 deadline suggested it fell short - a reminder of the gap between doctrine and industrial execution. The shift has been distilled into three words: production is deterrence. An adversary weighing aggression now counts factories, not just fleets.
Europe is running the same play: a European Commission target of two million shells of annual production capacity by end-2025 and France's restarted propellant production at Bergerac - even as RAND warns that NATO allies have yet to choose between a "peacetime paradigm" driven by economic competition and a "wartime paradigm" of alliance-wide cooperation. Australia has moved from words to concrete with its A$26-36bn Guided Weapons and Explosive Ordnance (GWEO) enterprise.
Capital has noticed. According to A&O Shearman's Resilient Returns, US defence funding reached US$962bn in 2026, NATO allies have pledged 5% of GDP annually on defence and security-related spending by 2035; private equity deal value in aerospace and defence hit a record $55.6bn in 2025, while venture investment into defence and dual-use assets nearly doubled to $49bn. Most intriguingly, some governments are beginning to explore "production as a service" - supporting trusted companies with flexible manufacturing capability, with diligence focused less on guaranteed orders and more on production agility. If those models take hold, governments would pay for capacity itself, not merely units delivered. The factory becomes part of the contracted capability.
The Investable Theme
McKinsey's recent work on "affordable mass" describes where this ends up: in the future defence tech stack, the physical platform - the drone, the ground vehicle, the satellite bus - becomes the commodity layer, and its primary requirement is scalable producibility. Russian and Ukrainian forces combined are losing uncrewed systems by the thousands every month, a consumption rate McKinsey says "defies peacetime logic". Deterrence and endurance, it argues, "now require attritable mass at scale, plus the ability to reconstitute quickly".
None of this retires the exquisite. No low-cost interceptor does what a $4m PAC-3 does, Karako notes - "no captain of a ship will pull out his slide rule" mid-engagement - and the real demand is "not an either/or, it's a both/and". Exactly the investment point: both halves of the high-low mix must now be manufactured at scale.
So what does exposure look like? Three layers: the consumable systems themselves; the exquisite re-engineered for affordability; and the industrial inputs beneath both. Three companies we know well - one in each layer - show how quickly the thesis is turning commercial.
Layer One - Robotics as Munitions: GaardTech
Queensland-based GaardTech builds what it calls consumable robotics across three product pillars: the Jaeger family of robotic ground munitions, full-size robotic decoys, and robotic training systems. The decoys are replicas of a force's own vehicles, fielded for battlefield deception; the training systems are replicas of enemy tanks, fighting vehicles and air-defence systems that give forces something realistic to fight against. The systems are not paper concepts. GaardTech publicly lists the Ukrainian armed forces among its primary customers, alongside the UK Ministry of Defence (MoD), the US Department of Defense and the Australian Defence Force (ADF) and describes itself as operationally proven.
The economics are the point. GaardTech's public materials describe the Jaeger as a low-cost, consumable combat robot - cheap enough to guard ground in numbers, then be sent by an operator or autonomously to destroy a vehicle worth many times its price. The full-size decoys hand the adversary a losing trade: expend a real munition on a fake tank, or ignore what looks like a real one. They replicate the appearance, movement and signatures of real vehicles and are built to be fielded at scale (the flat-packed vehicles ship ten to a 20-foot container). The training targets let tank crews, HIMARS batteries and RAAF strike pilots practise the kill on high-fidelity replicas - with the visual, thermal and radar signatures to convince modern targeting systems - at a tiny fraction of the cost of destroying the real thing.
That consumption matters because defence manufacturing creates a financing paradox that investors from other industries consistently miss: governments are reluctant to commit to large-scale operational programs before capacity is proven, while suppliers struggle to finance capacity without the visibility of backlog. Capacity is part of the qualification, not merely the consequence. GaardTech's model is designed to keep its lines running year-round on products bought, consumed and reordered in peacetime, while building and proving the capacity to surge production for operational-scale contracts. It plans to scale its footprint across Australia, the UK and the US accordingly - toward munitions-plant volumes rather than workshop ones, ahead of the contracts that require them.Its footprint already spans Australia, the UK and the US - subsidiaries and what it calls stress-tested production facilities delivering localised product today.
Layer Two - Making the Exquisite Affordable: Hypersonix
Attritability is also climbing the value chain, into the most exquisite category of all: hypersonics. On 27 February 2026, Brisbane's Hypersonix flew its DART AE vehicle on the US Defense Innovation Unit's (DIU) Cassowary Vex mission, aboard Rocket Lab's HASTE from NASA's Wallops facility. DIU described it as sustained, air-breathing cruise above Mach 5 by a manoeuvre-capable platform - a critical milestone in its Hypersonic and High-Cadence Airborne Testing Capabilities (HyCAT) program.
What matters for this theme is not that DART flew, but how it is built. DIU describes it as the world's first entirely 3D-printed airframe of a hypersonic launch platform in high-temperature alloys. Its SPARTAN hydrogen-fuelled scramjet has a fixed geometry with no moving parts and is printed, not machined - the company says it can go from design file to finished engine in a matter of weeks. The whole program went from kick-off to flight in under 36 months. That is attritable discipline applied to a Mach 7 design: the US runs some 70 hypersonic programs, many queuing for scarce flight-test windows, and a single-use test vehicle only works as a product if it is cheap, printable and repeatable. The market agrees - three DART flights contracted, a conditional Kratos agreement to procure up to 20 systems once the combined flight system is demonstrated, a DIU agreement opening a pathway to follow-on production contracts potentially worth up to $500m, and allied doors opening through the UK MoD's hypersonics framework and a strategic collaboration with European missile prime MBDA.
The direction of travel is clear enough: a vehicle designed from the outset to be printed, flown and replaced is a vehicle built for production at scale. The factory is as much the product as the vehicle.
Layer Three - The Energetics Beneath: Black Sky Industries
Every attritable system still has to be thrown, boosted or propelled. Strip any guided weapon back to its parts and the same thing sits at the bottom of the bill of materials: a solid rocket motor. Most of a motor, by mass, is propellant, and roughly 70% of that is a single oxidiser, ammonium perchlorate. No ammonium perchlorate, no propellant; no propellant, no motor; no motor, no missile. Hence the least glamorous layer of the stack may be the most strategic: rocket motors are among the West's tightest chokepoints - deep tiers of specialised suppliers, much of it sole-sourced, $211m of emergency US replenishment funding, and a dedicated CSIS study in June.
Australia's version is starker. The GWEO enterprise is committing A$26-36bn over the decade to sovereign guided weapons, yet the flagship response - motor production at Mulwala with Northrop Grumman Australia as preferred partner - delivers its first GMLRS motors in 2030. Logan-based Black Sky Industries is not waiting. It describes itself as Australia's only sovereign developer and supplier of solid rocket propellant and motors to the defence sector. It was the first company to produce ammonium perchlorate on Australian soil, and it has flown the output: guided weapons, sovereign practice missiles, sounding rockets, and Wagtail rocket-assisted take-off boosters that exist to throw drones into the air by the fleet. When Defence announced the Mulwala program in April, it named three companies as partners to explore novel ways of manufacturing rocket motors: DefendTex, Anduril Australia - and Black Sky.
The under-appreciated asset is the test estate. Black Sky operates what it calls the largest private launch range in the world - three million acres at Thargomindah in western Queensland - and in July the Queensland Government put A$7.5m toward MaxQ, the company's planned guided-weapons test and evaluation range, billed as Australia's first in private hands. Owning the range collapses the iteration loop. In an industry where test slots are the bottleneck, that is a moat.
Nor is the opportunity purely domestic. As a sovereign producer outside the US International Traffic in Arms Regulations (ITAR) export regime, Black Sky can supply allies that incumbent supply chains keep waiting. Behind it all sits a founding team of aerospace and defence veterans including Blake Nikolic and Karl Hemphill, and Dr Vu Tran, co-founder of the multi-billion-dollar edtech Go1, that talksGo1 - a team that talks about "product velocity" in an industry that measures change in decades.
What We'd Underwrite
We would apply four tests.
First, reorder dynamics. Attritable systems are consumed in combat, but also in training, testing, and exercises. That means revenue should look like ammunition, not platform sales with 30-year tails. Ask what gets used up in peacetime.
Second, unit economics at volume. Affordable mass only works if unit costs fall as volumes rise; the business must hold margin down the cost curve. Vertical integration, additive manufacturing and design-for-production can help protect it.
Third, surge economics. Who pays for capacity that stands idle between crises? Peacetime consumables, licensing models, government-funded infrastructure and, prospectively, dual-use side-production are all answers to the same exam question. A company with no answer is a company whose factory is a liability.
Fourth, policy durability. Multiyear procurement, co-production and sovereign-capability mandates are powerful tailwinds, but programs of record slip, elections happen, and battlefield innovations can go from viable to obsolete almost overnight. Balance sheets and product roadmaps must survive both the queue and the counter-measure.
Magazines Win Wars
For thirty years the West designed for short wars and bought accordingly. It built museum pieces of extraordinary quality and kept magazines - the depth of ammunition stockpiles - shallow enough to empty in weeks. That era is over. The next major conflict - or, preferably, its successful deterrence - will be decided as much on factory floors as in laboratories.
The companies that matter in this cycle will be the ones that treat manufacturing as the product: systems designed to be built, priced to be lost and contracted to be replaced. The exquisite will still draw the crowds at airshows. But history keeps siding with the deeper magazine - and with the investors who helped fill it early.
Sources & further reading: US DoD, National Defense Industrial Strategy Implementation Plan (Oct 2024); GAO, Ukraine: Status and Challenges of DoD Weapon Replacement Efforts, GAO-24-106649 (Apr 2024); CRS, The U.S. Defense Industrial Base, R47751 (Sep 2024); CRS, Implementing the National Defense Industrial Strategy, IN12459 (Nov 2024); CRS, Defense Primer: U.S. Defense Industrial Base, IF10548 (Dec 2025); B. Loidolt, "Ukraine, the U.S. Defense Industrial Base, and the Elusive Crisis-Era Munitions Production Surge", NDU Press/JFQ (Dec 2025); CNAS, From Production Lines to Front Lines (Apr 2025); K. Marcinek, NATO and Its Defense Industrial Base, RAND commentary (Oct 2024); CSIS, Empty Bins in a Wartime Environment (Jan 2023); W. Rumbaugh and T. Karako, Solid Rocket Motors for Missile Defense, CSIS (Jun 2026); Breaking Defense, interview with Tom Karako (18 Jun 2026); The War Zone, US Army Buying Older PAC-2 Patriot Missiles for the First Time in Decades (23 Jul 2026); CRS, DOD Replicator Initiative, IF12611; McKinsey, Future Defense Tech: Multidomain Stacks to Build Affordable Mass (Feb 2026); A&O Shearman, Resilient Returns: Investing in Defense; Australian Defence Ministers, rocket motor manufacturing announcement (30 Apr 2026); Queensland Government ministerial statement on MaxQ (Jul 2026); DIU, Cassowary Vex Mission Success (Mar 2026); AP, Pentagon pushes defense companies to boost weapons production (Aug 2026); The Washington Post, Pentagon presses defense firms as Iran war depletes stocks (Aug 2026); Cosmos Magazine; AMGC; Australian defence trade press (ADM, Defence Connect, APDR, EX2); Hypersonix, GaardTech and Black Sky Industries public company materials and announcements. Institute for National Strategic Studies. Defenseiq.com. Hypersonix.com