The key judgment
On the night of 4–5 August 2026 Russia attacked Ukraine with 115 drones, 24 ballistic missiles and four missiles the Ukrainian Air Force described as the combined category “Zircon/Oniks”. The Air Force reported 98 of the drones destroyed or jammed. It did not report intercepting any of the missiles. Hits were recorded at 26 locations.
Read as one event, that night looks like an air defence that half worked. Read as four events, it looks like exactly what the system is built to do in one lane and structurally unable to do in another. Ukraine does not have an air-defence problem. It has four of them, and they are only loosely related.
One-way attack drones, cruise missiles, ballistic missiles and glide bombs are answered by different equipment, drawing on different production lines, supplied by different countries under different political conditions, and adapting on different timescales. The lane Ukraine has improved most is the one it builds for itself. The lane it has improved least is the one it cannot build at all. The contrast reflects more than engineering. It also exposes the shape of the supply map behind Ukraine’s air defence.
It does not follow that Ukraine’s air defence is failing, or that more interceptors would fix it. Both of those are claims this note does not make, and the second is the more common error.
The drone lane: adaptation, and then counter-adaptation
The drone lane is a genuine Ukrainian success, and the numbers move in the direction that makes it interesting. Independent monthly analysis by the Institute for Science and International Security, compiled from Ukrainian Air Force daily reporting, records Shahed-type launches rising from 4,128 in August 2025 to a peak of 8,161 in May 2026 — while the share of those drones recording a hit fell from 16.47 per cent to 6.65 per cent over the same period. More arrived, and proportionally fewer got through.
The mechanism is cost. RUSI’s Justin Bronk describes the answer as “deep belts of cheap short-ranged air defence” — anti-aircraft artillery, MANPADS, mobile fire groups and, increasingly, interceptor drones. A one-way attack drone is individually easy to destroy; its effect comes from arriving in numbers designed to exhaust whatever is used against it. Engaging that with expensive surface-to-air missiles is not sustainable, and Ukraine stopped trying to. The interceptor drone — a cheap propeller-driven airframe built to kill another cheap airframe — restores the exchange rate.
Critically, this is the one layer Ukraine owns. It is built domestically, across more than twenty producers, and its output is not set by a decision in another capital.
Then Russia adapted back. Jet-powered variants entered the picture: the Geran-3 from July 2025 at over 200 mph on a Chinese Telefly JT80 turbojet, and the Geran-4 cruising above 200 mph with a sprint over 300 mph on a more powerful engine. As David Hambling notes, mobile fire groups have far less chance against a drone at 300 mph than at 120 mph, and interceptor drones slower than about 220 mph “have little chance against the Geran-4” outside its cruise phase. Russian sources indicate an intention to make half of Geran production jet-powered. Ukraine has responded with jet interceptors of its own, reaching around 400 mph.
The trend duly bent. Ukrainian interception of drones is reported to have fallen from about 92 per cent to about 87 per cent in July 2026 — small, but in the opposite direction to the preceding year. The adaptation is not free for Russia either: a Geran-4 costs upwards of $100,000, roughly twice a Geran-2, and carries perhaps a quarter of its range. Speed is being bought with reach and cost.
This is the correct shape for the whole note. Not a solved problem, but a working adaptation cycle: Russian mass, Ukrainian cheap counter, Russian counter to the counter, and a next test already visible.
The cruise-missile lane: a different economy
Cruise missiles are neither cheap enough to swarm nor fast enough to require the narrowest tier of defence, and they are answered by their own layer: principally NASAMS and IRIS-T SLM, with fighter interception alongside. These are medium-range systems whose economics sit between the two extremes.
The distinguishing feature is not capability but supply. Both systems are donated, and their resupply is set abroad. NASAMS fires an air-to-air missile family from the ground, which ties its replenishment to a munition stream also wanted for combat aircraft — an advantage in production terms and an exposure in allocation terms. Germany is the principal IRIS-T supplier, and the only primary source that separates what has been delivered from what is planned is the German federal government’s own itemisation, which records six IRIS-T systems delivered against ten planned. That document is explicitly an overview as of 17 April 2025 and states that it will not be updated further. Higher totals circulate; no primary source was found for them, and this note does not repeat them.
That is worth pausing on. For the cruise-missile lane, the best public record of what actually arrived stopped being maintained sixteen months ago. The number of launchers in Ukraine is not the binding variable in any case — the rate at which interceptor missiles are replaced is, and no supplier publishes it.
The ballistic lane: the hardest problem, and an unresolved diagnosis
Ballistic and quasi-ballistic missiles — Iskander-M, Kinzhal, S-400 in surface mode — are answered by Patriot PAC-3 and SAMP/T, and by nothing else Ukraine fields. It is the narrowest lane in the system and the only one with no domestic component whatsoever.
Reported interception in this lane has been low since the second half of 2025. RUSI records roughly 37.4 per cent across attacks in summer 2025, about 17 per cent over 1–24 October 2025, and around 15 per cent across 2025 as a whole, with a majority of individual attacks producing no interceptions at all. The night of 4–5 August 2026 sits inside that pattern rather than outside it.
Why the rate fell is not established, and this note does not resolve it. Three explanations are live in the specialist literature and none has been demonstrated to dominate.
The first is Russian missile adaptation. RUSI’s analysis of the Iskander examines steeper trajectories permitting terminal manoeuvre, improved 9B999 decoys, and coordinated multi-directional salvos — while stating plainly that “it is not yet possible to arrive at a conclusive answer”. The second is interceptor scarcity, the explanation most often reached for in public commentary. The third is engagement choice: Fabian Hoffmann suggests Russia may increasingly strike frontline positions and undefended cities where Patriot is not deployed, so that fewer incoming missiles are engaged at all.
The chronology matters, and it cuts against the simplest reading. The decline is documented from the second half of 2025. The interceptor-exhaustion claims most frequently cited for it are dated to mid-2026, roughly nine months later. Hoffmann makes the point directly: scarcity “may not fully explain the recent drop since shortages predated the August baseline”. Something that began in 2025 cannot have been caused by something reported in 2026.
What an interception percentage cannot tell you
There is a further difficulty underneath all three explanations, and it is the most important methodological point in this note.
Ukraine does not publish how many incoming missiles it attempted to engage.
A reported interception rate is therefore a ratio of successful interceptions to missiles launched, not to missiles engaged. It has no denominator in the sense that would make it a measure of system performance. “One of 27 intercepted” is consistent with twenty-seven engagements and twenty-six failures; it is equally consistent with two engagements, one success, and twenty-five missiles that no battery was positioned to reach. Those are completely different findings about the state of Ukrainian air defence, and the public figure does not distinguish between them.
Hoffmann states the constraint plainly: published intercept rates have “limited utility without additional context”. This note treats the ballistic interception rate as an outcome measure and not as a performance measure, and readers should treat headline percentages the same way.
Glide bombs: interception is the wrong frame
The fourth lane is the one most often misdescribed, because the word “air defence” invites the assumption that the answer is to shoot the weapon down.
Glide bombs are cheap, heavy and arrive in volume. Russian Su-34s typically carry four per sortie and release them from 35–50 km standoff; UMPK kits reach 60–70 km and the newer UMPB reportedly further. In July 2026 the UN human rights monitoring mission recorded 105 civilians killed and 753 injured by aerial bombardment — a 143 per cent increase on June, making glide bombs the second leading cause of civilian casualties that month.
NATO’s Joint Air Power Competence Centre assesses the counter as a three-tier disruption of the kill chain, and the tier that shoots the bomb down is the least useful of the three. First, deny the target: dispersal, concealment and movement, on the reasoning that prevention is more cost-effective than interception. Second, disrupt the weapon: electronic warfare and satellite-navigation jamming are identified as “the most effective countermeasure”, exploiting the kit’s navigation dependency, with its backup inertial system described as “prone to significant inertial drift”. Third, destroy the shooter: fighter interception and strikes against airfields and dispersed launch platforms, described as “the most decisive way to defeat glide bombs” — though requiring robust suppression of enemy air defences to attempt.
On terminal interception, JAPCC is explicit: continuous use of NASAMS, Patriot or S-300 against glide bombs is “unsustainable owing to war-reserve material and cost limitations”.
So the glide-bomb lane depends on Ukrainian electronic warfare, Western fighter aircraft, and the ability to strike Russian airbases. It has almost nothing in common with the interceptor supply chain, and no decision about Patriot production would affect it.
The supply map behind the four lanes
Set the four lanes beside each other and the pattern is not about capability at all.
The drone lane is Ukrainian. It scales because Ukraine can decide to scale it. The cruise-missile lane is European, donated, and its delivery record is no longer being published. The ballistic lane is American and effectively single-source: actual PAC-3 MSE production was 620 interceptors in 2025 against a baseline rate near 600 a year, with a seven-year framework agreed in January 2026 targeting 2,000 a year by 2030 and interceptors funded under an April 2026 contract assessed as unlikely to arrive before mid-2028. Ninety-four per cent of that funding is Foreign Military Sales, against an allied order backlog reported at more than 4,300 rounds — roughly seven years of output at 2025 rates. Bronk adds that PAC-2 and PAC-3 scarcity is global, driven partly by expenditure by Gulf states and by US forces elsewhere. Ukraine is one customer in a queue it does not administer. The glide-bomb lane depends on none of this and on aircraft and electronic warfare instead.
The political layer sits on top of the industrial one and moves faster. Licensed Ukrainian co-production of Patriot interceptors was pledged at the July 2026 NATO summit on 20 July, publicly disowned by the US president on 31 July, described by the US ambassador to NATO on 3 August as something that would not be concluded before this winter, and asserted as decided by President Zelensky in August without US confirmation. As of 14 August 2026 there was no licence, no contract and no delivery. Four public positions in four weeks produced no instrument.
This is why Vigil’s standing record for this file treats air-defence sustainment as a system rather than a stockpile, and why it records Ukrainian interceptor holdings and delivery volumes as unknown rather than estimating them. Neither is public.
Ukraine is trying to build a fifth answer
The obvious response to a lane you cannot supply is to build one. Ukraine is attempting exactly that, and the attempt is more interesting than it first appears.
The terminology needs care, because it is easy to get wrong. FP-7 is an offensive ballistic missile. FP-7.X is a separate interceptor reworked from that line and, per Hoffmann, derived from the Soviet 48N6. Freya — spelled Freyja by several outlets — is the wider system the interceptor sits inside, and it is the flagship project of an Integrated Anti-Ballistic Missile Coalition formed in Paris on 13 July 2026 comprising Ukraine and nine European states. The FP-7 is not an interceptor, and calling it one would invert the finding.
The status as of 26 August 2026 is development. Fire Point conducted a fully guided manoeuvring flight test on 3 June 2026 and published footage; Ukrainska Pravda’s account is explicit that no target was engaged and that this was a flight demonstration rather than an interception test. Company executives report at least five tests by the end of July 2026, and thirteen industrial partners are now engaged. Fire Point targets a first successful interception in mid-2027; Forces News reports the objective as July 2027, with first interceptors themselves due only at end-2026. Ukraine’s National Security and Defence Council describes the first-half-2027 goal as a minimum viable product “capable of demonstrating its first practical results” — and the coalition’s partner companies are financing development until that milestone, which by their own framing places it in the future. No interception has been attempted.
The genuinely interesting finding is what the design depends on. FP-7.X is being built around a German seeker from Diehl Defence, Norwegian command and control from Kongsberg, European radars, and NATO Link 16 — for which Ukraine licensed the alliance’s CRC System Interface in May 2025. A successful programme would not give Ukraine an independent anti-ballistic capability. It would move the dependency from the United States to Europe.
That is still a meaningful change, because European industrial politics are not American industrial politics and the queue is a different queue. But it is not independence, and it should not be described as such.
What would weaken this reading
- Ballistic interception improves materially without any identifiable increase in interceptor supply. That would strengthen the adaptation explanation and weaken the scarcity one.
- Interceptor deliveries demonstrably increase and reported outcomes do not improve. That points the same way, more strongly.
- Ukraine, or a partner, publishes attempted-engagement data. That would resolve the denominator problem and could invalidate the caution urged here about reading interception percentages.
- Jet-powered Geran variants substantially reverse the drone-lane trend, rather than bending it by a few percentage points. The cheap-counter model would then be a phase rather than a structure.
- Glide-bomb effectiveness falls sharply because launch aircraft are pushed back rather than because bombs are intercepted. That would confirm the JAPCC hierarchy; the reverse would challenge it.
- FP-7.X demonstrates a ballistic interception substantially ahead of its stated 2027 objective, which would compress the timeline this note treats as the realistic one.
Observable next steps
- Whether reported ballistic interception moves before or after any concluded supply instrument. The order is the test that separates the explanations.
- Whether a Patriot co-production licence or contract is actually signed, as distinct from announced, disowned or asserted.
- Whether actual interceptor deliveries change, rather than announced production capacity — and whether any supplier resumes publishing a delivery record.
- Whether the jet share of Russian one-way attack drone launches rises toward the reported fifty per cent intention, and what happens to drone leakage as it does.
- Whether FP-7.X achieves a first demonstrated ballistic interception by its stated mid-2027 objective.
- Whether European partners convert the anti-ballistic coalition into orders and production, or it remains an integration project.
What this does not mean
It does not mean Ukraine’s air defence is failing. Three of the four lanes are performing at or above what their supply position would predict, and the drone lane has improved against rising volumes.
It does not mean interceptor scarcity is irrelevant. It means scarcity has not been shown to be the cause of the ballistic decline, that the chronology sits awkwardly with that explanation, and that the public figures cannot settle it either way.
It does not mean more Patriot interceptors would not help. It means that even a large increase would leave three of the four lanes untouched, and that “air defence” is too coarse a term to carry a supply argument.
And it does not mean the four-lane structure is permanent. It is a reading of how the system is currently organised, offered because it explains why two apparently contradictory results occur on the same night. A Ukrainian interceptor that works against ballistic missiles, or a Russian drone that defeats cheap defences outright, would redraw it.