How a Choke Point Becomes a Weapon

An export control is a government rule that restricts what can be sold across borders, to whom, and under what conditions. Nations have used them for centuries against weapons, but applying them to commercial computer chips represents something newer and sharper: the deliberate use of a commercial supply chain as an instrument of national strategy. To understand how this works, you have to understand the concept of a choke point.

A choke point is a place in a supply chain where the alternatives collapse to one or a very few options. If your rival needs a resource and can buy it from twenty suppliers in ten countries, you have no leverage. If your rival needs a resource that flows through a single company or a single technology that you or your allies control, you have enormous leverage. You can turn the tap. The art of modern technology statecraft is identifying these choke points and deciding when to squeeze.

Recall the anatomy of the compute stack. The design of the most advanced AI chips is concentrated in American firms. The extreme ultraviolet lithography machines needed to manufacture leading-edge chips come from a single Dutch company. The most advanced manufacturing itself is concentrated in Taiwan. This concentration is precisely what makes controls feasible. A rival nation cannot simply shop elsewhere, because elsewhere does not exist, at least not at the frontier and not quickly.

In October 2022, the United States government did something unprecedented in scope. It issued rules restricting the export to China of the most advanced AI chips, the equipment to make such chips, and even the ability of certain personnel to support Chinese chip development. In 2023 it tightened those rules to close gaps that chipmakers had used to keep selling slightly downgraded processors. The intent was explicit and openly stated by officials: to hold Chinese frontier AI and supercomputing capability at a fixed remove, denying access to the compute that advanced systems require.

What makes this a weapon rather than ordinary trade policy is the theory of advantage behind it. Return to the compounding logic from earlier. If capability compounds, then a gap maintained today is a gap that could widen tomorrow, because the leader can use its superior compute to accelerate its own progress. Deny a rival the chips, the theory goes, and you do not merely slow one training run. You slow the entire flywheel: the models that would have designed better chips, written better software, and trained better successors. You attack the compounding at its root.

The United States also reached beyond its own borders to make the controls bite, and this is a critical feature. Because the Dutch lithography machines and the Taiwanese foundries rely on American technology and components somewhere in their own supply chains, Washington could pressure allies to align their export policies. The Netherlands and Japan introduced their own restrictions on advanced chipmaking equipment. A control is far more powerful when the choke points are held by a coalition, because it closes the side doors through which a determined rival might otherwise slip.

There is a name for this dynamic when a single power sits at the center of many networks and can weaponize that position: it is sometimes called the leverage of interdependence. The country through which the wires run can listen, and can also cut. The United States discovered that its central position in the chip supply chain gave it a lever few had fully appreciated until it was pulled.

But a weapon is defined as much by its costs as its effects, and export controls carry real ones. They punish the controlling nation's own companies, which lose a large market overnight. They signal to every other nation, friend and rival alike, that dependence on the controlling nation is itself a risk to be engineered away. And they light a fire under the constrained nation, converting a commercial dependency into a national mission to build domestically whatever was denied. A choke point squeezed hard enough teaches your rival exactly where to dig.

My read is that export controls are best understood not as a knockout blow but as a clock. They freeze the board for a period, buying the leader time. What the leader does with that time, whether it extends its lead or merely enjoys it, determines whether the controls were a masterstroke or merely an expensive provocation. The tool is real and the leverage is real. Whether it delivers lasting advantage is a separate question, and the answer is not yet written.