In April 2026, Taiwan’s financial regulator did something unusual: it raised the cap on how much of a single stock a domestic fund could hold, from 10% to 25%. The change was not ideological. It was an admission. One company had grown so large that fund managers tracking Taiwan’s main index physically could not hold enough of it to keep pace with their own benchmark. That company is Taiwan Semiconductor Manufacturing Company, and it now accounts for over 40% of the Taiex and more than 60% of the MSCI Taiwan Index.
Single-stock concentration at that level is rare in any developed market, and it has unnerved investors used to diversified indices. TSMC’s shares are up more than 130% over the past year, and its market capitalisation has roughly doubled over the same period to around $2 trillion as of mid-2026. When TSMC moves, the entire Taiwanese market moves with it. The deeper problem is not the size of the position. It is what sits underneath it.
The company a national index now rests on
TSMC does not design chips. It manufactures them under contract, a model called pure-play foundry that its founder Morris Chang built in 1987, with backing from the Taiwanese government and Philips. The proposition was to separate the design of a chip from its manufacture, pool demand across hundreds of customers, and reinvest the proceeds faster than anyone else into the next process generation. It worked. TSMC currently commands close to 70% of the global foundry market across all nodes, and at the leading edge — the sub-5-nanometre processors inside frontier AI training clusters — its share is estimated at 90% or more.
The figure that matters for the concentration story is not the headline 90%. It is that there is no second source. No backup fab on another continent makes the same chips at the same node. The most strategically important manufacturing process in the technology economy has been consolidated into one supplier, operating from a string of fabs in Hsinchu, Tainan, and Taichung.
Why nobody can build a second one
The recipes are not secret. Advanced semiconductor manufacturing is documented across thousands of patents and academic papers. The barrier is everything around the recipe. Running a leading-edge fab requires extreme ultraviolet lithography machines built by a single Dutch company, ASML, each costing several hundred million dollars; a workforce of tens of thousands of engineers trained on those specific machines; supply contracts with hundreds of chemical and gas vendors; and capital expenditure in the tens of billions of dollars per fab before a single chip ships. TSMC has been compounding that knowledge for nearly four decades, and the gap between it and its rivals has widened with every node.
Samsung, the only other foundry running EUV at scale, has struggled with yields and lost major customers. Intel, after losing its process leadership, is attempting a comeback but remains behind on adoption. For the chip designers who depend on the leading edge, the practical substitutability is close to zero on a five-year horizon. Redesigning a chip for a different foundry’s process takes 12 to 18 months and tens of millions of dollars per design, and even then performance and cost would regress by a full generation. There is the leading edge, and there is waiting for it.
An economy bent around one product
The concentration does not stop at the stock market. It runs through the whole Taiwanese economy. GDP rose 8.63% in 2025 and expanded at an annualised 13.69% in the first quarter of 2026, almost entirely on the back of chip exports. Semiconductors alone account for more than 20% of the island’s GDP. But the prosperity is unevenly distributed across the island, concentrated in the engineer-heavy science parks while the wider economy lags.
Taiwan’s central bank governor has warned of an emerging “K-shaped economy,” where the sectors riding the AI wave pull away while everyone else stagnates. The semiconductor industry employs roughly 300,000 people in a workforce of 11 million, yet sets the salary expectations, the housing prices, and the index returns for the entire country. A single firm is now the GDP engine, the dominant employer of the high-wage cohort, the largest line in the national index, and the reason foreign capital flows into Taiwan at all.
The overhang nobody can price
This is where the financial story collides with the map. Taiwan is an island of roughly 23 million people that the People’s Republic of China claims as its own territory, sitting just over 100 kilometres off the mainland coast across one of the most militarised stretches of water on Earth. Beijing has not ruled out the use of force to achieve unification.
Analysts refer to the result as the geopolitical overhang on TSMC’s stock: the permanent, unpriceable question of what happens to the world’s chip supply, and to a national index, if anything disrupts operations on the island even temporarily. A blockade, a strike on the Hsinchu Science Park, a cyberattack on the supply chain, or a prolonged power crisis would interrupt the manufacture of the chips that train every major AI model. The recovery would be measured in years, not weeks, because no replacement capacity exists. The same event would take a 40%-weighted stock, and the index built on it, down with it.
