The Exchange Ratio

Kills are expected values. Intercept odds apply per shot, so an engagement of k shots kills with probability 1 - (1 - p)^k. Fired all at once, every engagement spends k shots; fired one after another, the next shot goes only after a miss, which saves interceptors. Leakers come from magazine exhaustion, failed intercepts and threats the system cannot engage at all. Damage uses compounding leakers: the chance the asset survives is the leaker miss rate raised to the number of leakers.

Dollars the defender spends per dollar the attacker spends, at the shots and firing doctrine you set above and each defender's default odds per shot. n/a marks a pairing the system cannot engage; hover it for the reason. Green is a defender winning the accounting. Everything else is a defender losing it.

The screenshot everyone shares stops at the first division: four million dollar missile, thirty five thousand dollar drone, therefore the West is losing. CSIS has pushed back on exactly this framing, and they are right. The interceptor is not bought to destroy the drone. It is bought to protect whatever the drone was flying toward.

So the ratio only becomes an argument once you name the target. Against open ground, any interceptor above the price of a burst of 35 mm is waste. Against a transformer that takes three to five years to replace, against an aircraft worth eighty two million on the ramp, against forty people asleep in a building, the same shot is cheap. The US Department of Transportation values a statistical life at 13.7 million dollars. Two of those and the PAC-3 has already paid for itself.

The real failure is not the price per shot. It is magazine depth. A fully loaded Patriot battery carries ninety six MSE rounds. At two shots per target that is forty eight engagements, and roughly 372 million dollars of inventory that took years to build. An attacker can drain it with a night of drones costing less than a rounding error on that figure, then send the fifty ninth one through. Cost is survivable. Running out is not.

Which is why the Army is now shopping for a Patriot interceptor under one million dollars, and why the cheapest working answers in Ukraine are a Cold War anti-aircraft gun from 1973, pickup trucks with heavy machine guns, and now 2,500 dollar interceptor drones that account for a third of all aerial kills.

The last caveat is that every row in this table is a snapshot of a moving fight. When the gun trucks were downing forty percent of the drones, Russia pushed the Shaheds up to two to five kilometres, above machine-gun range, and taught them to dive vertically onto the target. The guns got cheaper kills, so the threat left the guns' envelope, and the defence had to invent the interceptor drone. The same spiral runs in the radio spectrum: when Ukrainian spoofing sent 45 percent of a night's drones wandering off course, Russia fitted the Shaheds with 8-element jam-resistant antennas, then 12, then 16, and Ukraine answered with jammers built to beat those. An exchange ratio is not a property of a weapon. It is a property of one date in an arms race.

Read the results with these simplifications in mind. Real air defence is layered: a raid meets jamming first, then guns and interceptor drones, then missiles for what is left. This page models at most two layers, a cheap first layer firing one shot per target and a main system behind it; a real network has more. Within the main system, shots are re-fired only under the shoot-look-shoot doctrine, and decoys are spread evenly through the salvo because the defender cannot tell them apart. Radar, crews, fuel, logistics and launcher capital are excluded on both sides, which flatters everyone but flatters cheap systems most. The defended asset is a single point value, where real raids spread across many targets. And every figure is a snapshot: the arms race described above moves faster than any table of it.

est marks a figure that is a reasonable estimate rather than a published one. Default intercept odds are per shot and are estimates; the only published rate among them, the NASAMS 94%, is a per-target claim from the manufacturer's government, used here as a per-shot approximation. Every dollar figure above is sourced except the three per-engagement costs marked est (mobile fire group, Sky Sentinel, EW), which have no published figure. The headline pairing puts both sides on the same basis: the Shahed at Russian domestic production cost, the PAC-3 at US procurement unit cost. Unit costs in this domain are genuinely contested: domestic production, export contracts and analyst estimates can differ by an order of magnitude for the same weapon, which is why the Shahed appears twice at two real prices. EW is the softest entry of all: its per-target cost here is marginal operating cost only, Pokrova's capital price is not public, and a diverted drone is displaced rather than destroyed, so its intercept probability blends diversion with the chance the drone still lands on something.