The Drone Gap Is a Manufacturing Gap

The lesson of recent conflicts is not that small uncrewed systems are decisive. It is that quantity is, and quantity is a manufacturing question.

Western defense industry is built to produce small numbers of extremely capable platforms, on long timelines, to exacting standards. That model is well suited to a carrier and poorly suited to an item that is expended by the thousand.

The constraint is components. Motors, batteries, optics, and control boards come from a supply base concentrated in a few places, and standing up alternatives takes years even with money committed.

Programs exist and are funded. What remains unresolved is whether a procurement system designed around certification and oversight can buy something disposable, which is a bureaucratic problem rather than an industrial one.

The word gap invites the wrong response, because a gap sounds like something you close with money. Money is not the binding constraint here, and the programmes that already have it are the evidence.

What is scarce is the industrial capability to make a cheap thing repeatedly, which is a different competence from making an expensive thing well. The second is what the defence base was built for and is genuinely excellent at.

That specialisation was a deliberate choice, reinforced by the authorities that govern buying and by decades of programmes structured around small numbers of exquisite platforms with long service lives.

The consequences show up in the rules rather than in the factories. The acquisition regulation is built to protect public money across a long procurement with a small number of units, which is exactly the wrong shape for something consumed by the thousand.

Consider what certification means for a disposable item. The cost of qualifying a component is fixed, and it is amortised across the production run. For a hundred aircraft that cost is trivial per unit. For an expendable airframe it can exceed the airframe.

So the procurement system does not forbid buying cheap things at volume. It simply prices the overhead in a way that makes cheap things stop being cheap, and it does so through rules each of which was a reasonable response to a real failure.

The supply base compounds it. Motors, cells, optics and control boards are commodity items produced at scale in a small number of places, and a defence buyer is a marginal customer for most of them rather than the anchor tenant.

Being a marginal customer is the position from which it is hardest to demand a second source, because the supplier's incentive to build one is proportional to the business it protects.

The money is visible and it is real. Programmes appear in the budget documents every year, and the award data shows where it goes. What the numbers cannot show is whether the capacity being funded is the capacity that would matter.

The distinction is between a line that buys units and a line that buys the ability to make units. The first produces inventory. The second produces a factory that can still be running in year three of something nobody planned for.

That is the actual question, and it is a procurement design question rather than a technology one. The engineering is largely solved and widely distributed. The part that is not solved is institutional.

Which suggests where to look for progress. Not at demonstrations of what a system can do, which are plentiful and encouraging, but at whether anyone has bought a production line rather than a production run.

There is a counter-argument worth taking seriously, which is that disposability is a false economy in a system that has to work. A cheap airframe that fails on the day is not cheap, and the certification overhead exists because somebody once died for the lack of it.

The honest answer is that both are true and the balance differs by item. The failure is not choosing volume over quality. It is having one procurement pathway, calibrated for the second, and running everything through it regardless of what is being bought.