Back to the archive

The encyclopedia · Engineering & Operations · Strategic decision · 2026

TSMC's Arizona mega-fab costs US$165 billion and still runs 30% dearer than Taiwan

America's biggest-ever industrial build ran over budget and schedule, showing how hard it is to export Taiwan's cost model.

TSMC · 2026-01-16

What happened

TSMC's Arizona complex has become one of the most expensive industrial builds ever attempted: NT$5.2 trillion (US$165 billion) across a 1,149-acre site north of Phoenix. One fab is already producing chips, two more are under construction, and when the site is fully built out TSMC expects nearly a third of its most advanced chips to be made in the US.

The decision to reshore was strategic — driven by Washington's push and by US customers like Apple, NVIDIA, AMD and Qualcomm that wanted domestic supply. But the cost of making that decision work on American soil has been far above a Taiwanese build. Operations are reported to cost roughly 30% more than comparable Taiwan sites, driven by higher tariffs and transport costs for importing materials from home.

The overrun is not only money. Permitting takes at least twice as long as in Taiwan, and TSMC was forced to help draft about 18,000 technical rules to meet local requirements at a cost of NT$1.1 billion. A shortage of skilled workers pushed the company to import hundreds of technicians from Taiwan, which triggered lawsuits from US workers and backlash from local unions.

The project secured NT$207.9 billion in federal CHIPS Act subsidies, but the episode is a case study in a decision made strategically and paid for operationally: the incentives cover building the plant, not running it at a permanently higher cost in a market with fewer of the skills and suppliers the model was built on.

Why it happened

  • TSMC bet that its Taiwan cost model could be transplanted via subsidies, without pricing in dramatically higher operating costs
  • Reshoring ignored that US permitting, water and environmental rules run at least twice as long as Taiwan's
  • A thin local skilled-labour pool forced importing hundreds of Taiwan technicians, eroding the cost and goodwill advantage
  • US$165B of capital is now committed to a site that runs ~30% dearer than home, with no exit path
What it costUS$165B project, 30% higher running costs, years of delaycostly

The lesson

A subsidy pays for the plant, not the operating model — when the running cost of a site structurally exceeds the home base, the capital is sunk before the first wafer pays for itself.

Sources

spotted an error? The club wants to know.

Comments · 0

    Sign in to join the comments.

    More like this

    Somewhere, someone solved the problem this company failed at. 2nd Opinion →