The narrowest point in the world economy

Modern economies have concentrated supply chains everywhere, and most of them have substitutes. Advanced semiconductor manufacturing does not. It is the narrowest genuinely irreplaceable point in the global economy, and it sits in a politically contested location.

The chain, with the choke points marked

Design is concentrated but competitive. Intellectual property for instruction sets comes largely from one British company. Design software comes from three firms, two of them American. Lithography at the leading edge comes from one Dutch company, and each extreme ultraviolet machine involves thousands of suppliers and years of production.

Manufacturing at the leading node is where it narrows to almost nothing. A very large share of the most advanced logic chips is made in Taiwan. Memory is concentrated in South Korea. Advanced packaging, the step that bonds memory to processors, is a bottleneck in its own right and has repeatedly been the actual limit on how many AI accelerators exist.

Every link has a plausible substitute except manufacturing at the leading edge and lithography. Those two have none on any timescale shorter than half a decade.

What diversification has achieved

Subsidy programmes in the United States, Europe and Japan have funded new fabs. Several are operating or close to it. This is genuine progress and it is slower and more expensive than the announcements suggested, for three reasons that were predictable.

A fab takes years to build and more years to reach yield. The workforce does not exist locally and takes a decade to train. And the supply ecosystem, the hundreds of specialist chemical, equipment and service firms that cluster around a fab, does not relocate with the building.

The cost premium for manufacturing outside the established clusters is real and is commonly put in the range of twenty to fifty percent. Someone pays it: the subsidy programme, the customer, or the margin.

The scenarios, priced honestly

Disruption in the Taiwan Strait. Not a technology stock question; a global economy question. Advanced chips go into cars, phones, medical devices, industrial controls and weapons. A prolonged interruption would be a supply shock without a modern precedent. It is also, for exactly that reason, something every party involved has enormous incentive to avoid.

Export controls tighten further. The most likely scenario and already partly happening. Effects are bifurcated markets, separate product lines for different regions, and accelerated domestic development in restricted markets.

A natural event. Taiwan is seismically active and periodically short of water, both of which have interrupted production before. These are the underpriced risks because they are not political and therefore get no headlines.

What an investor can actually do

Not much, honestly, and that is worth saying. There is no hedge for a chip supply interruption, because everything correlates in that scenario. What you can do is know your exposure, including the indirect kind: an industrial company whose products contain controllers, a car maker, a cloud provider with a capital expenditure plan.

The useful discipline is scenario arithmetic rather than prediction. If leading-edge supply fell by a third for a year, what happens to the revenue of each holding. Most portfolios have never had that question asked of them.

What to watch

Watch advanced packaging capacity, which constrains accelerator supply more tightly than wafer capacity. Watch yield reports from new fabs outside Asia, which tell you whether diversification is working. And watch the equipment order book of the lithography supplier, which is the closest thing to a leading indicator for the entire industry.