In March 2026, researchers at the Payne Institute for Public Policy at the Colorado School of Mines published a number in Foreign Policy that reframed how to think about modern conflict: the first 36 hours of the U.S. and Israeli campaign against Iran consumed more than 3,000 precision-guided munitions and interceptors. The team built the estimate with an open-source ledger and data-scraper tool, counting Iranian missile and drone attacks and the coalition responses they triggered, then converting the expended munitions into the minerals and materials required to replace them.

The ledger kept running as the war did. By the 96-hour mark the count exceeded 5,000 munitions. By day 16 it stood at 11,294, at an estimated cost of about 26 billion dollars, with Iranian attacks settling into a sustained rhythm of roughly 33 missile and 94 drone strikes per day. And by mid-2026, the Center for Strategic and International Studies estimated the cumulative toll on the systems that matter most: a prewar inventory of roughly 2,200 of the most modern Patriot interceptor variants reduced to between 759 and 827, and THAAD interceptors down from 452 to between 234 and 278.

It is tempting to read these numbers as a war story. They are more useful read as an industrial engineering story, because every one of them is the output of a system with measurable rates, capacities, and lead times. That reading is the point of this essay.

Two rates, one problem

Strip away the geopolitics and the situation reduces to a comparison every operations student learns in their first inventory course: a depletion rate against a replenishment rate.

The depletion rate announced itself in 36 hours. The replenishment rate is set by the production system, and its numbers are public. According to CSIS, the United States produces 183 of its highest-end Patriot interceptors per year, and a newly signed contract runs three and a half years from signature to first delivery. Reporting on the THAAD line puts current production at roughly 96 interceptors per year, with a framework agreement aiming to raise capacity to 400. Set those rates against a drawdown of roughly 1,400 Patriot interceptors and 200 THAAD interceptors and the arithmetic speaks plainly: at current rates, rebuilding what 39 days of combat consumed is a project measured in years, which is exactly what both the Secretary of Defense and CSIS have said. The Army's multiyear Patriot production contract, worth up to 58.6 billion dollars through fiscal 2032, is sized to that reality.

The Payne team put the strategic implication in a single phrase in a later analysis for RUSI: endurance in this kind of conflict is governed by command of the reload. Not the depth of the magazine on day one, but the rate at which the magazine refills.

Replenishment is a scheduling problem

Command of the reload is, at bottom, a family of operations research problems, and naming them precisely matters because each has known structure and known methods.

It is an inventory problem: how large should stockpiles be, given uncertain demand, multi-year replenishment lead times, and the enormous carrying cost of interceptors that sit unused? The 36-hour burn rate is a demand observation that any inventory model must now digest, and it moves every answer.

It is a capacity expansion problem: raising a production line from 96 units a year toward 400 is not a purchase order, it is a multi-year program of facility construction, supplier qualification, tooling, and hiring, with decisions today that bind outcomes half a decade out.

It is a scheduling problem under cascading constraints: production rates are bound by component availability, component availability by processed materials, and processed materials, as the Payne analysis quantified by converting munitions into their mineral replacement burden, by critical mineral supply chains concentrated in a strategic competitor. A replenishment plan that ignores the minerals layer is scheduling against capacity that does not exist. Readers of the previous essay in this series will recognize the shape of that constraint: the binding stage of the value chain is processing, and processing is an operations problem.

And beneath all three, it is a workforce problem. Facilities, suppliers, and processing capacity are operated by people: production planners, process engineers, quality technicians, supply chain analysts. Expanding defense industrial capacity, mineral processing, and munitions production simultaneously means competing for the same thin national pool of practitioners who can do this work. That pool, not capital, is the constraint the United States can least quickly expand, and it is the one Project FORGE exists to widen.

The stress test we should learn from

The Payne researchers called the opening of the war a stress test of Western industrial endurance. Stress tests are valuable when their results change behavior. The measurable result of this one is that depletion happens in hours and replenishment happens in years, and the gap between those two clocks is set by industrial systems that respond to exactly the disciplines this program teaches: inventory theory, production scheduling, capacity planning, and supply chain optimization, practiced by people trained to apply them.

A forthcoming entry in this site's Models section will make that concrete: a transparent, reproducible replenishment model built entirely on the public figures cited here, showing how stockpile recovery timelines respond to production rates, lead times, and capacity expansion decisions. The numbers in this essay are inputs anyone can check. The methods are teachable. That combination is the reason to be neither panicked nor complacent about industrial endurance, and instead to do what operations researchers do: model it, and then train enough people to manage it.

Previous in this series: The Optimization Gap.

Sources