Why the First Shell From a $533 Million Factory Still Has No Unit Cost

Divide $533 million by one shell and the answer is $533 million per shell. Nobody writes that number down. Yet buyers accept its close cousin every time a ribbon gets cut: a completed factory treated as capacity, a first accepted article treated as a unit cost.

The $533 million is a hypothetical figure, an invented illustration rather than a reported one. I am not going to attach it to any particular plant, caliber, or schedule, because the argument does not depend on any of that. It depends on a habit that shows up across the whole munitions enterprise: the moment a facility exists and one round passes acceptance, people start talking as if the production economics are known. They are not. After the first usable round, the credible cost per shell is still undefined, and pretending otherwise distorts every buying decision downstream.

What One Accepted Round Actually Proves

A first article proves that the tooling can produce the geometry, that the process specification can be met at least once, and that the acceptance criteria are testable. That is real engineering progress, and I do not want to diminish it. But it is a statement about feasibility, not economics. It tells you the denominator of exactly nothing.

The same distinction is playing out in directed energy right now. The Pentagon task force working the counter-UAS problem is launching a directed-energy pilot program using lasers and microwave weapons. U.S. Army Brigadier General Matthew Ross put it plainly: the science has been proven for several systems, but successful use in an operational context remains unresolved. That is the honest formulation. Working once is a different claim from working repeatedly under real conditions, and the distance between those two claims is where most of the money lives.

Artillery is the same claim structure at industrial scale. One accepted shell demonstrates feasibility. A cost per shell is a statement about repetition, and repetition has not happened yet.

The Five Variables That Set The Real Number

Dividing sunk capital by a first article hides every variable that determines what a shell actually costs. There are at least five, and none is knowable from a ribbon-cutting. Each has to be measured, and none can be assumed.

Qualified yield: the fraction of starts that become accepted rounds. If yield sits at 60 percent, every accepted shell carries the material, energy, and labor of 1.67 starts. Yield on a new line is not designed. It is discovered, slowly, in production.

Cycle time: the interval between accepted units at steady state, not the time the first one took with engineers standing over it. Illustrative arithmetic only: amortize $533 million of capital over ten years. At 100,000 accepted rounds a year, capital adds roughly $533 to each shell; at 40,000 a year, it adds more than $1,300. Same building, same ribbon, wildly different economics, and the first article cannot tell you which world you are living in.

Rework: the shadow factory inside the factory. It consumes labor, floor space, and schedule while appearing in no brochure. Its size is unknowable until the line has run long enough to reveal it.

Labor: energetics is not a hire-and-train-in-a-quarter trade. Reporting on the munitions base this year made the point directly: building missile propulsion requires specialized facilities, chemists who spend careers learning the craft, and areas engineered to handle energetic materials without catastrophe. That workforce learning curve is a cost driver no amount of capital can purchase outright.

Repeatable accepted throughput: rate held across consecutive lots, all passing acceptance. This is the only denominator that produces a defensible cost per shell, and by definition it cannot exist on day one. Everything else is a proxy.

The Industrial Base Keeps Saying This Out Loud

The people closest to munitions production do not talk in ribbon-cuttings. Reporting on the stockpile problem after the Iran conflict stated that the United States depleted key missile inventories and that rebuilding some to prewar levels may take three to five years. The same reporting was blunt about why: you cannot simply run a munitions line faster, and there is no second shift that doubles output overnight. Three to five years is a statement about yield ramps, workforce qualification, and lot acceptance, not about square footage.

Notice, too, what the analysts identify as the real constraint. Justin Rodriguez, a partner and managing director at Boston Consulting Group, argues that much of the manufacturing capacity needed to increase production already exists. The Assembly report identified the larger obstacle as convincing suppliers that demand will remain stable enough to justify hiring, inventory purchases, and production investment. Installed capital is not the binding constraint, because throughput is a function of sustained conditions over time, which is exactly the thing a first article cannot demonstrate.

If you want the price of confusing an asset with an output, the submarine fleet already paid it. Over the last decade, attack-submarine maintenance delays produced an estimated $3.4 billion in costs for crews and submarines that provided no operational capability. Billions spent, hardware present, output zero. The factory version of that failure mode is a finished plant with one accepted round and no qualified rate behind it.

What A Maturity Label Would Actually Look Like

The fix is unglamorous. Every cost and output figure carries a label stating what kind of number it is, and the label is backed by evidence. Three labels cover most of what buyers see.

A target is a number someone wants. The current interceptor-cost effort is a clean example: the Department of Defense awarded X-Bow Systems a contract worth almost $11 million. The associated program aims at an anti-ballistic weapon for less than $750,000 a shot, roughly a fifth the cost of a Patriot missile. Stated that way, the $750,000 is honest. The failure mode arrives later, when targets get repeated without the label until they read like achievements.

Then there is the division, a number derived from a contract, and it needs the label too. Taiwan's Ministry of National Defense finalized an NT$26.9 billion, or US$847 million, contract with Anduril Industries for 2,032 ALTIUS unmanned systems, and dividing those figures gives about $417,000 per airframe. But the contract covers 1,554 ALTIUS-700M loitering munitions and 478 ALTIUS-600ISR systems. The purchase includes the drones, associated systems and launch equipment. So the raw division blends two different products with their ground infrastructure; it is a real number and it is not a unit cost.

A demonstrated number requires consecutive accepted runs: total cost across a stated span of consecutive lots, all passing acceptance at rate, divided by accepted rounds. Until that exists, the honest label is modeled or calculated, and a buyer should read it as such. The evidence is the run history, not the spreadsheet.

We hold ourselves to this at Kibernan. We have produced six complete, costed engineering programs, and every published figure in them carries its maturity: modeled, specified, calculated, or objective. They are proposals. No Kibernan hardware has been built, flown, or fielded, and we say so in writing, because a figure that hides its own maturity is not information.

Pay For Accepted Runs, Not Ribbons

For buyers, the translation into contract language is short. Stop treating factory completion or first-article acceptance as proof of capacity. Tie milestone payments to consecutive accepted lots at a stated rate. Require every cost and output figure in a proposal or a press release to carry a maturity label, and require the demonstrated ones to cite the runs behind them. When a program briefs a cost per round before a qualified rate exists, ask the only question that matters: how many consecutive accepted lots stand behind that number?

None of this slows anyone down. It changes what gets celebrated, from the ribbon to the third consecutive accepted lot at rate, which is the only milestone that ever moved a shell toward a magazine. Celebrations are cheap and magazines are not.

The first article is the entry ticket to qualification, not the finish line. Our hypothetical $533 million factory with one usable round has proven that it can begin. The cost per shell is still ahead of it, out in the yield curve, the rework bay, and the second year of a workforce learning to handle energetics. That is where the number lives, and no press release can move it.

Ribbons do not fill magazines.