There is a calculation most boards quietly avoid. Not because the numbers are hard to find, but because the answer is uncomfortable. If you build, sell, or depend on anything with an advanced logic chip inside it — and in 2026 that is very nearly everything — then a meaningful slice of your enterprise value rests on a single island, served by a single company, sitting roughly 130 kilometres off the coast of a strategic rival. I call working through what that actually means “strait math,” and it is the first piece of homework I set any client who tells me their supply chain is “diversified.”
The concentration is not a talking point — it is the base case
Let us start with what is no longer in dispute. Taiwan produces well over 90% of the world’s most advanced logic chips — the sub-10nm devices that run AI accelerators, flagship handsets and high-performance computing (Mordor Intelligence). Some datasets put the advanced-logic figure as high as 92% (Economy Insights). Across all nodes, the island accounts for roughly 60% of global contract chip output and, by some institutional projections, around 83% of total manufacturing capacity in 2026 (Storm Media).
Inside that island, the concentration tightens again. TSMC’s share of the global foundry market reached 72.3% in the first quarter of 2026, a record high, up from 70.4% the previous quarter (Semiecosystem / TrendForce). Its nearest competitor, Samsung, sits at 6.5% — a gap of nearly 66 percentage points (Facebook / Chen-yuan Tung). The top three foundries together hold close to 84% of the market. At the leading edge the picture is starker still: 2nm capacity is reported sold out through 2027, with Apple alone holding more than half of the early 2nm allocation, and TSMC’s advanced manufacturing and CoWoS packaging fully booked through 2028 (Byteiota).
This is not a diversified market with a clear leader. It is, as more than one analyst has now put it, a natural monopoly at the leading edge (Dataintelo). For risk purposes, you should treat it as a single point of failure and model it accordingly.
The cost of the disruption — the number that should be on the slide
Here is where the strait math gets serious. Bloomberg Economics has estimated that a genuine blockade cutting Taiwan off from the world would cost the global economy in the order of US$5 trillion; a full conflict scenario rises towards US$10 trillion (Insurance Journal). Those are headline figures, and headline figures rarely survive contact with a board’s procurement reality. What does survive is the production math underneath them.
Under a blockade scenario, modelling based on 2024 capacity suggests global supply of cutting-edge logic chips would fall by around 62%, and lagging-edge chips by roughly 31% (Insurance Journal). Translated into finished goods: output of advanced electronics such as smartphones drops by about 60% in a blockade and as much as 85% in a war scenario where South Korean production is also curtailed. Apple, as the single most exposed handset maker, could lose more than 90% of iPhone sales, with global smartphone volumes washing out by up to 80% once channel inventory is gone.
You do not need to believe any particular probability of conflict for these numbers to matter. The point of a stress test is to size the loss conditional on the event, and then let leadership decide what that conditional loss is worth insuring against. A 60% to 85% reduction in your ability to build product is not a supply wobble — it is an existential interruption.
How to actually run the calculation
I take clients through four steps, and none of them require a crystal ball.
First, map exposure by node and by part. The question is not “do we use Taiwanese chips” but “which of our products cannot ship if sub-7nm logic stops, and what share of revenue do those products represent?” For many technology firms the honest answer is “most of it.”
Second, identify substitution time, not substitution existence. It is comforting to say Samsung and Intel also make leading-edge silicon. It is more useful to know that no other company can replicate TSMC’s most advanced capacity inside three to five years (LongYield). Your substitution window is measured in years, not quarters. Plan to that horizon.
Third, price the disruption against the cost of mitigation. Strategic inventory, dual-qualified designs, and pre-negotiated allocation all cost real money today. The discipline is to weigh that certain, modest cost against the probability-weighted catastrophic loss. When the downside is a 60%-plus collapse in output, even a low annual probability justifies a serious mitigation budget.
Fourth, track the leading indicators rather than the headlines. Foundry concentration is rising, not easing — TSMC gained roughly five percentage points of share year on year (Facebook / Chen-yuan Tung). Reshoring helps at the margin: the US share of advanced capacity is projected to climb, but even then TSMC and Samsung are expected to own over half of that new American capacity, and leading-edge R&D and pilot lines remain anchored on the island (Mordor Intelligence). The “silicon shield” is real, but it protects Taiwan rather more reliably than it protects your order book.
The uncomfortable conclusion
Most resilience plans I review treat Taiwan as a tail risk — something to acknowledge in the appendix and move past. The strait math says otherwise. When a single supplier on a single contested island accounts for over 90% of the silicon your most valuable products depend on, that is not a tail. That is the centre of your distribution, and the tail is the small chance that nothing goes wrong for another year.
The firms that will navigate the next decade well are not the ones predicting whether or when something happens in the strait. They are the ones who have already priced the disruption, sized the loss, funded a proportionate hedge, and can therefore make calm decisions while competitors are still arguing about probabilities. Do the maths before the market does it for you.
