The magazine

A Pacific war does not wait for the next training run. It waits on cells, canisters, and interceptors that already exist. The opening is a magazine problem. The model stack does not refill a vertical-launch system.

What gets shot first

Open analyses of a Taiwan or first-island-chain fight keep returning to the same list: long-range land-attack and anti-ship cruise missiles, air-launched standoff weapons, and the interceptors that keep bases and ships alive while those shots fly. Tomahawk-class cruise missiles. JASSM-class air-launched strike. PAC-3. THAAD. SM-6. Maritime strike variants. Those are the rounds that convert a plan into a hole in a hull or a crater on a runway.

Recent operations outside the Pacific already cut into several of those piles. Think-tank reconstructions after the Iran fight put Tomahawk use in the high hundreds to around a thousand, JASSM use over a thousand, Patriot use around half the prewar stack, and THAAD use a large share of a small inventory. Those are not Pacific numbers. They are proof that a few weeks of intense firing empties the cupboard the Pacific would also need.

The rate problem

The United States is finally writing framework agreements as if production were the strategy. PAC-3 capacity is supposed to move from roughly 650 a year toward 2,000 over about seven years. THAAD is supposed to move from 96 toward 400 on a similar clock. Tomahawk targets have been set above 1,000 a year; recent actual output sat far below that because past orders were small. Lead times on the critical interceptors and Tomahawk still run two to four years from contract to usable round.

Eighteen munitions plants are now described as under construction. Lockheed has announced an $8–9 billion manufacturing push. That is the right direction. It is not inventory on a railhead in 2026. A line that reaches the advertised rate in 2030 does not stock a fight that starts in 2027.

Why AI does not close this gap

Models can pick better targets. They can sequence a strike package faster. They cannot machine a rocket motor, qualify a thermal battery, or grow a solid-propellant line. The bottlenecks sit in those shops, not in the token budget.

Cheap mass — one-way drones, containerized cruise, attritable effectors — is the only munitions category that can move on a two-year clock. That is Part 7. It is not a substitute for the interceptors that keep Kadena, Guam, and a carrier alive while the cheap mass flies.

China designed for surge

Chinese munitions industry is built as a regional wartime system: state conglomerates, robotic lines, and a planned surge above peacetime rates. The U.S. system was built for exquisite weapons and annual buys. One side planned to refill. The other planned not to miss.

Bottom line

If the first two weeks consume the magazine, the rest of the war is fought with whatever the factories can still throw. Software does not change that arithmetic. Plants do, and plants take years.