RAGEX home

Western Weapons in Ukraine: What Worked and What Didn't

RAGE Global · RAGE X Analysis · Updated 2026-09-02 · 12 min read

RAGE X ANALYSIS — Expert assessment authored and reviewed by Carlos Kfoury. This is analysis, not reportage. Factual claims carry source attribution; judgements carry confidence labels.

The most capable Western tank in theatre was never defeated frontally and 57% of the tranche was still lost. A 1947 Swedish autocannon is doing more useful work than several systems costing a hundred times more.

Executive Assessment#

Four years of Western equipment in combat against a peer adversary has produced the largest body of evidence on Western military hardware since 1991. Most of the public conclusions drawn from it are wrong in a specific and consistent way.

First, the headline failure was quantitative, not qualitative. Ukraine received approximately 21 Leopard 2A6 hulls — 18 from Germany, three from Portugal. That is barely two companies. The strategic error was expecting 21 vehicles to function as a war-winning instrument. (Confidence: Analysis, high)

Second, where systems failed, they usually failed at the seams rather than in the specification. No Leopard 2 has been confirmed destroyed by frontal penetration of its composite armour. Roughly 12 of ~21 2A6 hulls appeared in confirmed loss imagery by September 2024, six destroyed. Mines immobilised; drones and artillery finished. (Confidence: Confirmed as OSINT floor / Analysis)

Third, the systems that performed best were the cheap ones. The Bofors 40 mm L/70, designed in the late 1940s, is downing Shahed-type drones because 40 mm shells cost hundreds of dollars against a $20,000–$50,000 target. Its manufacturer has scaled production three- to fourfold. (Confidence: Confirmed)

Fourth, the most expensive systems hit a cost-exchange wall that no amount of quality resolves. Ukraine's PAC-3 MSE stocks ran out around 1 July 2026. In July, Ukraine intercepted 87% of drones and cruise missiles and only 15% of ballistic missiles, allowing over 126 Russian missiles to reach cities. (Confidence: Confirmed as reported)

Fifth, the transferable finding is about integration, not hardware. A fleet of 2A6, 2A4, 2A4NO, Strv 122, Leopard 1A5 and 1A5DK is a maintenance problem in which the word "Leopard" implies commonality that does not exist at part-number level. Donor generosity produced an orphan fleet. (Confidence: Confirmed)

The Scorecard#

System Verdict Why
Leopard 2A6 Worked, too few Frontal armour never defeated; ~21 hulls was never a capability; ~57% documented attrition
Leopard 2A4 / Strv 122 / Leopard 1A5 Mixed Orphan configurations; spares sourced worldwide; Germany no longer fields older variants
M1 Abrams Adapted Anti-drone structures, reactive armour, localised control panels to compress crew training
M2 Bradley Worked Performed alongside 2A6 at Robotyne; consistently well-regarded by crews
HIMARS / GMLRS Worked, then countered Decisive in 2022–23 against fixed rear targets; Russian EW and dispersal degraded effect over time
Patriot PAC-3 Worked, unaffordable Effective against ballistics; stocks exhausted July 2026; ~€3.5m per interceptor
NASAMS / IRIS-T Worked, unaffordable at scale Firing a €3m IRIS-T at a $20,000 Shahed is mathematically unsustainable
Gepard / Skynex Worked Gun-based short-range; favourable exchange
Bofors 40 mm L/70 Best value in theatre 1947 design; shells cost hundreds against $20k–$50k targets; production tripled
Javelin / NLAW Worked Decisive in the 2022 defensive phase; less central once the war became static and drone-dominated
F-16 / Mirage 2000 Contributing Part of the layer intercepting drones and cruise missiles
Armoured recovery vehicles The gap Germany ordered 23 Büffel in June 2026 to replace transfers; deliveries Dec 2027–Jun 2029

What "Failure" Actually Looked Like#

The Zaporizhzhia counteroffensive of June 2023 is the case everyone cites and few describe correctly.

What happened. The 47th Mechanised Brigade attacked with Leopard 2A6 and M2 Bradley. Mine density materially exceeded pre-offensive estimates. Vehicles were immobilised inside belts they could not clear at the pace the plan demanded, then struck by FPV drones and artillery while stationary. Aerial footage of a burning 2A6 in a minefield went global within hours and straight into Russian information operations.

What also happened, and received a fraction of the attention. Once through the belt, the Leopard 2A6 and Bradley combination performed. The brigade contributed materially to the capture of Robotyne in August 2023.

The correct verdict. The platform was not the failure point. Breaching under persistent aerial surveillance was the failure point — and it remains unsolved by any army on earth in 2026. Western equipment was asked to execute an operation that Western doctrine had not solved, in quantities that could not absorb the losses that operation generates.

The Recovery Gap#

This is the least-discussed finding and the most transferable.

In April 2024, Russian media published footage of the first captured Ukrainian Leopard 2A6 — a 47th Brigade vehicle disabled near Avdiivka in October 2023, still fitted with its anti-drone grilles.

Roughly six months between disablement and extraction. Neither side could recover it, because recovery vehicles operating that close to the line are themselves priority drone targets. Russian units reportedly used explosives to detach the tracks before towing it clear.

On a transparent battlefield, armoured recovery is not logistics. It is the mechanism that converts a mobility kill back into combat power — and its absence converts every mobility kill into a total loss plus an enemy trophy.

Berlin has internalised this. Germany ordered 23 Rheinmetall Bergepanzer 3 A2 Büffel in June 2026, explicitly to restore capability degraded by transfers to Ukraine — having handed over 21 Bergepanzer 2 and two Bergepanzer 3. Deliveries run December 2027 to June 2029. That is a three-and-a-half-year lead time on a lesson learned in 2023.

The Sustainment Answer#

The most valuable Western contribution was not a weapon.

Date Development
June 2022 Rheinmetall and KNDS stand up a maintenance and logistics hub in Lithuania
October 2023 Rheinmetall Ukrainian Defence Industry (RUDI) founded with Ukroboronprom
December 2023 German-funded spare-parts stockpiling contract concluded with Ukraine's MoD
10 June 2024 RUDI maintenance hub opens in western Ukraine — Marder first, now Leopard, Fuchs, Skynex, Gepard
Through 2025 Rheinmetall's Ukrainian footprint expands to four plants: armoured vehicles, artillery ammunition, air defence, propellants
December 2025 Contract to deliver five Lynx KF41 IFVs

Shortening the repair loop from "ship it to Lithuania" to "drive it to western Ukraine" is a larger contribution to combat power than any single tranche of vehicles. Damaged modern Leopard 2s could previously take many months to return to service. Turn time, not inventory, is the currency of an attrition war.

RAGE INTEL judgement: Rheinmetall's onshore presence — the deepest of any non-Ukrainian defence prime in Ukraine — will outlast the specific vehicles it was built to support and constitutes a structural shift in European defence-industrial geography. (Confidence: Analysis, high)

The Cost-Exchange Wall#

The clearest quantitative finding in the entire dataset.

Engagement Interceptor cost Target cost Verdict
Patriot PAC-3 vs ballistic missile ~€3.5 million High-value Justified, and stocks ran out
IRIS-T vs Shahed ~€3 million $20,000–$50,000 Mathematically unsustainable
Interceptor drone vs Shahed Under $15,000 (some €1,000–€4,000) $20,000–$50,000 Favourable
40 mm Bofors round vs Shahed Hundreds of dollars $20,000–$50,000 Strongly favourable

What happened when the wall was hit. PAC-3 MSE stocks ran out around 1 July 2026. Lockheed Martin produced approximately 620 PAC-3 MSE interceptors in 2025; a $58.62 billion July 2026 contract aims to triple output by 2030, but a 24-to-30-month production lead time means new interceptors will not arrive before late 2027 at the earliest.

The Western system failure here is industrial, not technical. The Patriot works. There are not enough of them, they cannot be built fast enough, and global demand is rising as Japan and South Korea build inventory. Ukraine competes with the US itself for the same production line.

What the West Should Have Learned#

Six findings, ranked by how badly they are being acted on.

1. Magazine depth beats platform quality. Every failure in this dataset is a depth failure. Twenty-one tanks. Exhausted interceptor stocks. Recovery vehicles ordered three years late with deliveries three years out.

2. Recovery is combat power. A three-and-a-half-year lead time on the Büffel order is the clearest indictment of Western defence-industrial tempo in the entire war.

3. Cheap effectors are not a downgrade. The Bofors L/70 and interceptor drones are doing work that expensive systems cannot afford to do.

4. Configuration commonality matters more than generosity. Six Leopard variants is worse than half as many of one.

5. Onshore repair beats offshore repair by a wide margin. Turn time is the currency.

6. Six-week adaptation cycles defeat multi-year procurement. Ukraine's Brave1 ecosystem iterates in six weeks. No Western procurement system operates at that tempo; most cannot approve a requirement in six weeks.

The Lowy Institute's critique is the sharpest available: Western military institutions exhibit a systemic learning deficit, favouring exploitation of existing competencies over exploration of new solutions, and adapting dangerously slowly despite unprecedented open access to battlefield evidence.

The Counter-Argument#

Stated fairly, because the "Western weapons failed" narrative is as wrong as the "wonder weapon" narrative it replaced.

Ukraine is still fighting. That is the outcome the equipment was supplied to produce, and it has been produced.

The 2022 phase was decisive and is now under-credited. Javelin, NLAW and HIMARS materially shaped the defensive phase and the 2022 counteroffensives. Systems that stop working after two years because the adversary adapted did not fail — they were countered, which is what happens to every weapon eventually.

Quantity was a political constraint, not a technical one. Twenty-one Leopard 2A6s reflected donor decisions, not a limitation of the tank.

The professional literature does not support the strong failure thesis. The European Policy Centre's assessment is that despite mines, drones and battlefield transparency making armour easier to detect and destroy, tanks continued to enable advances, support infantry, deliver protected firepower and hold ground.

Key Judgements#

# Judgement Confidence
1 No Leopard 2 has been confirmed destroyed by frontal composite penetration Analysis — high
2 ~12 of ~21 Leopard 2A6 hulls appeared in confirmed loss imagery by September 2024 Confirmed as OSINT floor
3 The 2023 counteroffensive failed at breaching under surveillance, not at platform performance Analysis — high
4 Recovery capability, not armour, is the decisive survivability variable Analysis — high
5 Germany's Büffel order carries a 3.5-year lead time on a 2023 lesson Confirmed
6 Onshore repair in western Ukraine contributed more combat power than any single vehicle tranche Analysis — high
7 Ukraine's PAC-3 MSE stocks ran out around 1 July 2026 Confirmed as reported
8 July 2026: 87% interception of drones and cruise missiles, 15% of ballistics Confirmed as reported
9 The Patriot shortfall is an industrial failure, not a technical one Analysis — high
10 The Bofors 40 mm L/70 is the best cost-per-effect system in theatre Analysis — high
11 Six Leopard variants created an orphan fleet; commonality beats generosity Confirmed
12 The systemic Western learning deficit is the finding least acted upon Analysis — high

Indicators to Watch#

  1. PAC-3 MSE delivery against the tripling target, and whether Ukraine or US inventory is prioritised.
  2. Büffel delivery schedule adherence, Dec 2027 – Jun 2029.
  3. Any Western army adopting a Brave1-equivalent procurement mechanism.
  4. Interceptor drone adoption by NATO members for the Eastern Flank Drone Wall.
  5. Whether Leopard 2A8 and M1E3 specifications make overhead protection and counter-UAS baseline rather than retrofit.
  6. RUDI throughput — mean repair turn time is the truest sustainability metric.
  7. Any confirmed frontal composite penetration of a Leopard 2, which would revise Judgement 1.
  8. Ammunition and interceptor production rates, in units per year rather than contract value.

Sourcing and Methodology#

Tier 1 to Tier 3: Oryx-derived visually confirmed OSINT tallies as reported in defence trade press, official German and US procurement announcements, manufacturer disclosures, and named institutional analysis from the European Policy Centre, Modern War Institute and Lowy Institute.

Loss figures are visually confirmed minima with snapshot dates stated, not totals. Interception percentages are as reported by Ukrainian authorities and allied press and are not independently verified. Cost figures vary by configuration and source.

Performance verdicts in the scorecard are RAGE INTEL analytical judgements based on the cited evidence, not measured assessments, and are presented as such.

This assessment discusses platform performance at the level already debated publicly by serving officers, manufacturers and the defence press. It contains no vulnerability analysis, no engagement technique, and no dispositions.

Principal references

Corrections policy — errors are corrected promptly with a notice appended. No silent edits, ever. Corrections to: intel@ragex.co

RAGE INTEL — Intelligence that moves before the news does.

RAGE X Corp · Decode. Dominate. Deliver.

Continue the assessment