War became a manufacturing question.

Mass-attritable autonomous systems and the doctrine that rebuilt European defence economics in 36 months. A study on what happens when the unit cost of effect collapses by three orders of magnitude.

Prologue. The 36-month pivot.

In late 2022 the Ukrainian Armed Forces operated a few thousand commercial drones, repurposed for reconnaissance and ad-hoc strike. By the end of 2025, Ukraine's domestic UAV industry produced approximately 1.6 million unmanned aerial systems across the short, medium-range and long-range categories, of which roughly 30 000 long-range strike drones were planned and delivered, with deliveries continuing through 2026. The Ukrainian General Staff has reported on the order of 820 000 confirmed strikes against Russian targets across the calendar year 2025. The Center for Strategic and International Studies has assessed that approximately 65 percent of Russian armoured-vehicle losses were attributable to unmanned-aerial-system effect, with artillery and anti-tank guided weapons accounting for the remainder.¹²⁵

Those numbers describe a different war from the one most European general staffs were planning to fight. The doctrinal pivot has happened. It is happening at a tempo that procurement-grade defence acquisition cannot match. This study argues that the pivot is permanent, that the European response has been institutional rather than industrial, and that the institutional response will fail unless it is paired with an industrial one. The libertarian read is implicit in the argument; it surfaces in the closer.

The thesis.

Mass-attritable autonomous systems redefine the unit economics of war. A first-person-view drone produced inside Ukraine for around 400 USD destroys an T-90M tank whose replacement cost runs to roughly 4.5 million USD. The kill-ratio in dollars is on the order of 1:11 000. Even allowing generously for the cost of the operator, the airframe attrition rate, the comms infrastructure and the share of strikes that miss, the asymmetry remains overwhelming. The defender that buys 4.5 million-dollar units and the attacker that ships 400-dollar units are no longer in the same engagement. They are in different engagements, played on the same map.¹

The point is not that drones win wars. They do not, on their own. The point is that the procurement-grade weapon and the mass-attritable weapon now belong to different categories of capability, and a force that has only the first cannot operate in a battlespace in which the second is present at scale. Russia's electronic-warfare improvements through 2025 have raised the per-strike attrition rate of Ukrainian UAVs significantly; they have not closed the cost asymmetry, because the cost asymmetry sits in the production rate, not in the per-strike effectiveness.

The numbers, in detail.

The Ukrainian state's drone-production reporting is reasonably granular. Euro-SD's April 2025 review, citing Ministry of Strategic Industries figures, breaks the 1.6 million 2025 production volume into approximately 1.3 million short-range FPV/quadcopter systems, on the order of 200 000 medium-range fixed-wing strike systems, and 30 000 long-range strike drones with reach of 1 000 km or more. The long-range tier is the strategically novel one. It is the tier that puts Russian energy infrastructure deep behind the front line into permanent risk and that forces the Russian Ministry of Defence to disperse air-defence assets across the entire western half of the country at significant cost.¹

Russia's mirror-image is the Shahed-136 production line, scaled at the Alabuga Special Economic Zone in Tatarstan. OPFOR Journal's situation report on the 2025 Russian Shahed offensive places monthly production volumes in the multiple thousands, with the airframe variant evolved well beyond the original Iranian design through Russian on-shoring of the engine, the airframe and the navigation stack. The Russian doctrine differs in target selection rather than in attritable economics; the cost-per-effect logic applies symmetrically to the attacker side of the same battlespace.⁴

Cost asymmetry, in concrete.

Three orders of magnitude is not a metaphor. The asymmetry below is selected from cases the Ukrainian General Staff has confirmed and that the Russian Ministry of Defence has not formally disputed.

Apply the same logic to a 60 million-dollar Western infantry fighting vehicle, to a 90 million-dollar surface-to-air missile system, to a 350 million-dollar combat aircraft. The asymmetry compounds. None of these targets stop being valuable; they stop being affordable in the volumes a force at war actually loses them.

Tank losses, by attribution.

The CSIS analysis of the state of the Ukraine war in 2025 provides one of the cleanest available decompositions of Russian armoured-vehicle losses by causal vector. The picture confirms the doctrinal pivot in a way that prior wars did not generate evidence for.

For comparative scale, the equivalent share for the 2003 invasion of Iraq was effectively zero, and as recently as the 2014 phase of this same conflict, the figure was on the order of 5 percent. The decisive vector of armoured-vehicle attrition has migrated from artillery to autonomy in a single decade, with the bulk of the migration concentrated in the past 30 months.

How Ukraine actually built the production base.

The Ukrainian production base is not a single line. It is a federated industrial network of an estimated 250 to 400 small and medium-sized firms operating under a state coordination layer. The coordination layer, branded Army of Drones, sets specifications, qualifies production, ports designs across firms when one is destroyed by Russian strike, and finances volume through a mix of state procurement and crowd-sourced civic funding. The civic funding component, an institutional novelty Western defence has no easy peer for, has supplied a meaningful share of frontline UAV procurement since 2023.¹⁹

Three properties of this production base matter doctrinally. First, the iteration cycle from a frontline complaint to a re-engineered airframe averages two to four weeks, an order of magnitude faster than any procurement-grade defence vendor can sustain. Second, the network is hardened by distribution; Russian strikes have removed individual production sites without removing production capacity. Third, the production base feeds operational learning back into the design pipeline at a tempo Western contractor-customer relationships are structurally incapable of matching, because the operator and the engineer in the Ukrainian case often share a Telegram channel.

Helsing as a tell.

Western Europe has noticed. Helsing, the Munich-based defence-AI firm, has expanded from software for fighter aircraft into a vertical strike-drone product line under the HX-2 designation, with reported production targets in the low hundreds of thousands of units per year. Sacra's 2025 financial summary places Helsing's annualised revenue in the high single-digit-billions of euros after the Saab/Eurofighter strategic deal and a follow-on Swedish JAS-39E AI-cockpit programme. Helsing is the first European defence-AI firm to be valued in the bracket previously reserved for the legacy primes (Rheinmetall and BAE foremost, with Thales adjacent).³

The Helsing trajectory is a tell because it is the European response that resembles, in a small way, the Ukrainian model. A software-led firm with a designer-operator relationship that bypasses the prime-contractor stack is, structurally, the species of organisation that can iterate at war pace. That such a firm now sits inside a Eurofighter cockpit is institutional progress. Whether the rate at which Helsing-class capability proliferates exceeds the rate at which a future adversary's drone industry would, in a comparable contingency, is the unanswered question. As of April 2026 the answer is no, and the gap has not narrowed materially in the past 12 months.³⁷

European procurement at war pace.

The European Defence Industry Reinforcement through Common Procurement Act (EDIRPA) and the European Defence Industrial Programme (EDIP) constitute the EU's institutional response to the procurement-tempo problem. The 30 March 2026 EDIP financing decision allocated approximately 300 million euros in initial common-procurement funding for joint Member-State acquisition of priority capability categories, including loitering munitions and counter-UAS systems. The number is real and the procedural innovation is real; the volume is not at war pace.⁶

The bottom row is the rate at which the Eurofighter, the F-35 and a long string of multinational European defence programmes have actually delivered. It is also the rate against which the Ukrainian production-iteration cycle should be benchmarked. Three weeks against seven years is not a ratio defence procurement can close with policy reform. It can be closed only by structurally changing what the procurement is for: not capabilities, but production lines.

The libertarian read.

The clearest doctrinal lesson of the Ukrainian case is that the state does not have to manufacture. It has to qualify and to consume; the coordination layer follows from the first two. The actual manufacturing, when not done by a captive prime contractor, is faster and more iterable when distributed across hundreds of small firms operating under a thin coordination layer than when concentrated in a handful of state-anointed industrial champions. It is also cheaper per unit of effect by a margin that competitive analysis can no longer deny.

This is awkward for European defence-industrial policy of the past three decades, which has been built around national champions, controlled-share defence groups and consortium-owned megaprojects. The political logic for the model is intact. The military logic for the model is increasingly broken. A continent that wants to be defended at the rate at which the threat now manufactures itself has to lower the floor of who is allowed to build, and raise the floor of what the state is willing to buy from non-incumbents.

The libertarian variant of this point is uncomfortable: the European drone industry that materially matters in 2026 is the one that did not exist in 2022. It was built by founders, by reservists running garage-grade fab lines, by software engineers who found themselves working on guidance code without the credentialling apparatus a Western prime contractor would have required. The same people, in a Western jurisdiction, would have spent two years acquiring an export licence. The institutional permission required to build at war pace, in other words, is the kind of permission only a state at existential risk has been willing to grant. The Chronicle's view is that the rest of Europe should grant it before the existential pressure makes it cheap.

What it means for builders.

If you are building software for autonomous systems, defence or otherwise, the operational lessons of Ukraine's drone doctrine are as follows.

Iterability dominates capability. A product that ships a small improvement weekly outcompetes a product that ships a large improvement annually, given equal access to the operator. Build the feedback loop before you build the feature set.

Cost-per-effect is the only relevant unit. Marketing a single high-end capability at high price is a strategy that fails the moment the adversary fields a low-end capability at scale. Optimise your cost denominator, not your capability numerator.

Distributed production is a defence property. A production base concentrated in one factory is a strike target. The Ukrainian case has demonstrated that distributed manufacturing degrades gracefully under attack in a way single-site manufacturing does not.

The operator is your engineer. Every defence-software product that has succeeded in Ukraine has had an unmediated channel from frontline operator to product engineer. Every product that has failed has had a procurement officer in between. The engineering implication is the obvious one.

Software guidance beats hardware sophistication. The cheapest, most-attritable Ukrainian drones routinely outperform high-cost Western counterparts because the guidance software is updated weekly. The hardware-sophistication-first defence model is being out-iterated by a software-first attritable model that nobody bothered to build until they had to.

Closer.

The procurement-grade weapon is not dead. The procurement-grade weapon as the dominant unit of military force is. The conditional that follows is uncomfortable for every European defence ministry, and it is the right conditional to focus on. The Chronicle's purpose, in this study and the next, is to make it as hard as possible to keep ignoring.

Tobias Hager

Founder · CEO · Sole engineer · mi media intelligence SL

Tobias designs and operates every system inside the house. He builds AI-native products and the offensive/defensive cyber capability stack the company runs alongside its research thread toward agentic generative reasoning. He writes the Chronicle from inside the operational stack rather than from outside of it.

tobias-hager.com LinkedIn contact@mi.network

mi media intelligence SL · Palma de Mallorca · April 2026 · SIGIL CHR-0003