Navigate
HomeStart here
MusingsResearch & long-form
BuildingProjects & learnings
WorkProfessional practice
RunningTraining & races
AboutValues & identity
Life & PlacesCulture, food, travel, cities
Notes & ArchiveJournals, essays, portfolio
Acquired / Case file

TSMC

One-line business model — only make chips for other companies — sustained for forty years.

Latest filing on file · 2024-12-31

Revenue
NT$2.89T
FY2024 · TWD
20-F · period ended 2024-12-31Source
Operating cash flow
NT$1.83T
FY2024 · TWD
20-F · period ended 2024-12-31Source
Capital expenditures
NT$956.0B
FY2024 · TWD
20-F · period ended 2024-12-31Source

Overview

Business model and origins

Pure-play foundry — manufacture chips for other companies only, never design or compete with them. Customers, not products, are the relationship the company defends.

2 episodes1 hook2026-05 case file as of

Origin

Morris Chang spent 25 years at Texas Instruments rising to Group VP before being passed over for CEO in 1983. He left and was recruited by the Taiwanese government's Industrial Technology Research Institute (ITRI) in 1985 with the brief: build Taiwan a semiconductor industry. Chang, then 54, proposed a pure-play foundry — a structurally new business model. Backed by ~$220M of seed capital (half from the Taiwanese government, half from Philips), he founded TSMC in 1987. The model was met with skepticism — no major customer would commit at launch. The compounding came from one principle Chang refused to bend: never compete with our customers. 39 years later TSMC manufactures ~92% of the world's most-advanced (≤7nm) chips.

Hooks from these episodes

A modern leading-edge fab costs ~$20 billion to build and depreciates over the better part of a decade. The whole semiconductor industry depends on a few of these existing in places they could plausibly be defended.

First time I understood why "fabs" appear in geopolitics articles.