This framework almost always finds the same thing: nothing has really run out, and the shortage is just money wearing a costume. Wartime spending is the case that runs the other way, for a reason worth understanding. A government that prints its own currency has already proved money is not its problem. So whatever is left over is what was actually scarce all along.
A reading against the treatise, and the second time this framework has found a real shortage rather than a fake one. The first was the Strait of Hormuz. The claim: money aimed at a physically limited thing does not buy more of it. It buys a higher price, and the price escapes into everything else that needs the same ingredients. Conditional, falsifiable, not policy advice.
In the first ten months of fiscal 2026 the United States government spent $6.284 trillion, collected $4.485 trillion, and borrowed the $1.799 trillion difference. Total debt hit $39.93 trillion on 13 August. The Congressional Budget Office expects the publicly held portion to pass its 1946 record before the decade is out. National defense got roughly $1.045 trillion, the largest peacetime allocation in the country’s history and a jump of more than 17 percent in a single year.
Every one of those dollars was found. Not a single appropriation failed for lack of money.
That is the ordinary condition of a government that borrows in a currency it issues, and it usually gets discussed as a warning. Here it is useful for a different reason. It makes this period a clean experiment.
If money is never what stops the government, then whatever went wrong in the thing it was buying cannot be a money problem. The framework’s usual answer is off the table, which means it has to take whatever the evidence says instead.
What the evidence says is that the money arrived and the shells did not.
The money showed up. The shells did not.
The Army set itself a target: 100,000 rounds of 155 millimetre artillery a month, to be hit by October 2025. As of March 2026, according to the Pentagon’s own Inspector General, it was making about 36,000 a month.
That is real growth. Before 2022 the rate was near 14,500. It is also a little over a third of the target, eighteen months late. And a contractor-run plant in Mesquite, Texas, built for $469 million, had not produced a single shell that met specification.
The spending side is worse. Roughly 319 Tomahawk missiles were fired in the first six days of the campaign against Iran. The number scheduled for delivery across the whole of fiscal 2026 is 190. Six days of use consumed more than a year and a half of production.
Patriot deliveries for the year run to 172 missiles. THAAD interceptors have seen no new deliveries at all since August 2023, and none are due until April 2027.
Behind all of it sits a problem that predates every one of these decisions. The United States has not made its own TNT since 1986. It buys it from allies, mostly Australia and India. The first new domestic plant, in Graham, Kentucky, is due to finish in February 2029, and its output is spoken for by the military, with nothing set aside for anyone else.
Four cases in which an unconstrained bid met a bounded capacity. Sources: DoD Inspector General via Military Times and Stars and Stripes, July 2026; CSIS Iran war cost estimates, munitions expenditure and delivery schedules; US Army contract announcements on domestic TNT production, Graham, Kentucky.
This one runs backwards
The treatise describes a money shortage working its way down through an economy and pretending to be a shortage of things. The test that separates a fake shortage from a real one is simple, and this framework applies it everywhere. Does the thing show up when you offer money for it?
If money appears and the goods follow, the goods were never short. The shortage was money. If money appears and the goods do not follow, the shortage is real.
By that test this one is not close. The money was offered, at enormous scale, publicly, with support from both parties and no financing constraint of any kind. What came back was 36,000 shells against 100,000, zero interceptors for three years, and a $469 million factory producing nothing anyone could use.
That is a real shortage by this framework’s own rule. It would be dishonest to reach for the usual answer here, and the framework’s default reading would have been wrong.
Fake shortage: offer money → the goods arrive → money was the problem
Real shortage: offer money → nothing arrives → the thing was the problem
Money offered: unlimited, about $1.045 trillion.
Goods received: 36 percent of target, eighteen months late.
Verdict: real.
The interesting part is what happens next. A real shortage meeting unlimited money does not simply fail quietly. The money does not evaporate. It is still there, still bidding, and it goes looking for the same scarce ingredients through every door it can find.
The damage is where you are not looking
Ask where the money goes when the thing it wants cannot be made any faster. It goes into the price of whatever is upstream. And from there into every other person who needs that same upstream thing.
Explosives are the clearest example. They are not only a military product. Ammonium nitrate and TNT are what mining and quarrying run on. Quarrying is where crushed stone comes from. Crushed stone is what concrete is. And concrete is houses, roads, bridges and data centers.
When defense buyers compete for a limited supply, the civilian market gets priced out of the same supply. And the new American capacity being built to fix the military shortage is not being built to fix the civilian one.
So a war in the Gulf raises the cost of a quarry in Pennsylvania. The quarry raises the cost of a road. And none of it shows up in any ledger anywhere as a war expense.
This is the whole thing running backwards. Normally a shortage of money fakes a shortage of goods. Here a flood of money, aimed at something that cannot be made faster, manufactures a genuine shortage of goods and hands it to everyone who needed the same ingredients.
It even reached the buyer’s own house, which is the detail that makes it hard to wave away as an accounting quirk. Operation Epic Fury cost roughly $29 billion. In May the Chief of Naval Operations told Congress that from July he would have to start cutting training, operations and certification. In other words, the Navy was eating the thing that produces future readiness in order to pay for present operations.
The carrier Gerald R. Ford picks up about 6 percent more maintenance for every 30 days its deployment is extended, so a five-month extension carries something close to 30 percent. None of that is a budget decision. That is hulls, dry docks, and the number of days in a year.
At the bottom of all of it, people
Follow every one of these bottlenecks down far enough and the same thing is sitting at the bottom. It is the hours ceiling, at national scale.
The Navy’s submarine programs need something like 140,000 skilled workers. Roughly 100,000 to build and 40,000 to maintain. Welders, machinists, pipefitters, people who can test a weld without cutting it open.
Those people do not exist. No appropriation creates them. What limits their number is a training pipeline measured in years for the ordinary grades and decades for the master ones.
Electric Boat is hiring on the order of 8,000 people in 2026 and had passed 4,000 by late July. That is a serious effort. It is also arithmetic that does not close the gap this decade. Virginia class boats now run an average of four years behind the schedules in their own contracts, and the delays got worse from 2025 to 2026 despite the money.
This is exactly what the hours ceiling describes. Everything else answers a higher price by showing up in greater quantity. Labor answers with the same quantity, because the person selling the hour and the hour are the same thing, and a master welder is thirty years of accumulated practice that either happened or did not.
The submarines were paid for. They were not built. And the reason has nothing whatsoever to do with whether they were paid for.
Who ends up paying
The pattern here is the one the Toilet Paper Theory predicts. It is worth saying plainly, because war spending usually gets discussed as though everyone shares the cost equally.
On the way out, a defense build-out turns borrowed dollars into contracts. Contracts into revenue. Revenue into earnings held by a fairly small group of shareholders.
On the way back, the price effects land on things ordinary production shares. Explosives into stone into construction. Machinists and welders bid away from civilian factories. Government borrowing competing for the same savings that price a mortgage.
A family in the top fifth gets the first effect and is shielded from the second, because necessities are a small slice of what it spends. A family near the poverty line gets only the second. The transfer runs uphill, and it runs uphill through the price of concrete.
The line item that buys nothing at all
There is one line in the federal accounts that is bigger than the entire military and purchases nothing whatsoever.
Over ten months of fiscal 2026 the government paid $1,169.6 billion in gross interest on its debt. Over the same ten months it spent $764.7 billion on every Department of Defense military program combined. Even netting out the interest paid to its own trust funds leaves roughly $963 billion, which is still more than the military. The Congressional Budget Office puts net interest at about $1.0 trillion in 2026 and $2.1 trillion by 2036.
Interest builds nothing. It commissions no shell, trains no welder, pours no concrete. It is the accumulated price of things already bought, and it grows on its own terms whether or not any of those things ever worked.
A country spending more to service what it borrowed than on the thing it borrowed for has arrived at this framework’s central problem from the fiscal side. Claims compound without limit. Real things are bounded by matter and by hours. And the gap between them does not close by issuing more claims.
Interest came to more than half again what the entire military cost. It is the one line in the budget that can never be delivered late, because there is nothing to deliver.
What would prove this wrong
$1.80T
The federal deficit over ten months of 2026. $6.284 trillion spent, $4.485 trillion collected.
36,000
Artillery shells made per month as of March 2026, against a target of 100,000 set for October 2025.
319 vs 190
Tomahawks fired in six days, against the number due to be delivered across the whole year.
$1,169.6B
Interest paid on the debt through July, against $764.7 billion for the entire military.
Claim
Measured by
Falsified if
Horizon
Money was never what was stopping the shells
Units actually delivered against dollars appropriated, program by program.
Deliveries catch up to targets inside the same funding cycle, without any new factory opening.
by 2029
Money aimed at a limited industry buys price, not more stuff
What each shell and each ship costs, against how many actually got delivered.
Costs per unit hold flat or fall while deliveries rise in step with the spending.
by 2030
The shortage escapes through shared ingredients
Prices for commercial explosives, crushed stone and building materials, against military demand for the same things.
Those civilian prices simply track ordinary inflation while military demand climbs sharply.
by 2029
Skilled hands are the constraint, not money
How far behind the shipyards are, against how much has been invested and how many hired.
The delays shrink noticeably while the shortage of skilled workers stays exactly where it was.
by 2031
Interest eats the budget faster than capacity gets built
Interest payments against everything else the government chooses to spend on.
Interest shrinks as a share of spending without the debt or rates falling to explain it.
by 2036
Two warnings, because this piece reaches further outside the framework’s usual evidence than most. Wartime cost figures are estimates, produced under time pressure by analysts working from public schedules, and the munitions numbers especially should be read as the best available rather than as audited. And this framework has no view on whether any of the spending was warranted. That is a question about strategy and about ends, which it is not equipped to answer and does not try to.
What it does have a view on is how the thing works. The treatise keeps finding that people are told something real has run out when what has actually run out is money, and that the fix is therefore easier than it looks. This case is the mirror image, and it carries the mirror-image warning. When the money really is unlimited and the thing really is limited, spending more does not produce more. It produces a higher price for the same amount, a queue in front of the bottleneck, and a bill that arrives in the price of everything else that needed the same ingredients. The factories are not standing and the shelves are not full. That is a rarer condition than this framework usually finds, which is exactly why it is worth saying out loud when it happens.