CompetitorsWide moat

TSMC held roughly 73% of the pure-foundry market in the first quarter of 2026. Samsung, in second place, held about 6.5%; then SMIC at 5.1%, UMC at 3.9%, GlobalFoundries at 3.3% and HuaHong at 2.5%¹. Intel — the company that led semiconductor manufacturing for thirty years — does not appear in the top ten foundries at all. The share table is so lopsided that the usual competitive questions barely apply.

So these four pages ask a different one: what kind of relationship does TSMC have with each of them? Samsung is the only other company that can make a leading-edge chip, which makes it more valuable to TSMC's customers as negotiating leverage than as a supplier. Intel is stranger still — simultaneously building a rival foundry and buying TSMC capacity for its own processors, an arrangement that puts TSMC's founding promise never to compete with a customer under real strain. SMIC is not competing for TSMC's customers at all; it is serving a market TSMC is increasingly barred from. And UMC, GlobalFoundries and HuaHong occupy the mature nodes TSMC has largely declined to defend.

The existing root threat on competition and cyclicality works through what could go wrong. What these pages add is that TSMC's dominance is specific rather than general: it is close to absolute at the leading edge, real but unremarkable at mature nodes, and simply absent inside China. Three different markets, three different TSMCs.

Watch leading-edge share rather than foundry share. Foundry share includes mature nodes where competition is genuine and pricing is ordinary; the 90%-plus of advanced-node production is the number that supports the margins, and it is the one that would have to break for any of these rivals to matter.