TSMC Arizona profit/loss split
- TSMC’s 2025 filings show a sharp split overseas: its Arizona chip fab turned profitable while Japan Advanced Semiconductor Manufacturing in Kumamoto stayed loss-making. - The standout number is the gap itself — Arizona earned about NT$16.1 billion, while JASM lost roughly NT$9.8 billion in 2025. - That matters because both fabs got heavy public backing, but only Arizona lined up with hotter AI and HPC demand.
Semiconductor fabs are supposed to be the hard part. Pour the concrete, install the tools, hire the engineers, and the economics should eventually follow. But TSMC’s 2025 numbers show that the harder part might be choosing the right chips to make. Its Arizona operation swung to about NT$16.1 billion in profit in 2025, while its Kumamoto joint venture in Japan, JASM, posted a roughly NT$9.8 billion loss. Same company. Two subsidized overseas bets. Very different outcomes. ### What actually split the two results? The simple version is product mix. Arizona’s first fab is making N4 chips — advanced 4-nanometer-class parts — and TSMC had already said that fab entered high-volume production in the fourth quarter of 2024 with yields comparable toy. ### What is Japan making instead? JASM’s first fab in Kumamoto began volume production at the end of 2024, but it is aimed at older nodes: 22/28nm and 12/16nm, with 40nm also in the broader site roadmap. Those chips are still useful — especially for autos, industrial gear, and image sensors — but they do not sit in the fattessolid middle seats in a softer section of the stadium. ### So was Arizona just “better run”? Not necessarily. That is the tempting story, but it is too neat. Arizona had a messy buildout and a lot of skepticism around labor, cost, and timelines. Yet once the fab started high-volume production, it was attached to stronger demand. TSMC also said major U.S. customers were pushing haader plan that now covers six fabs, two advanced packaging facilities, and an R&D center. ### Why does demand matter this much? Because a fab is not a generic factory. It is more like a factory built for a narrow band of products, customers, and downstream packaging needs. Governments can help fund the building, but they cannot force a market to want a given nmoto’s. ### Did subsidies fail in Japan then? Not really. The lesson is narrower. Subsidies can get capacity built, de-risk investment, and anchor supply chains. But they do not erase the difference between a fab serving a hot market and a fab serving a slower one. TSMC received government subsidies across several regions, including the U.S. and Japan, in support of plis necessary for these projects more often than it is sufficient. ### Could Japan
