How did TSMC’s Arizona operations become its most profitable overseas unit in the first half of 2026—earning NT$36.066 billion ($1.13 billion), up 662.8% year over year and more than double its full-year 2025 profit of NT$16.1 billion—what share did Arizona contribute to the NT$5

An important comparability caveat: The reported first-quarter Arizona result included roughly NT$16.9 billion of investment income. That means H1 subsidiary profit is not a clean proxy for fab-level operating profit or a recurring run rate.

Why Q2 fell despite high demand: Arizona reported NT$17.259 billion in Q2, down 8.2% sequentially from approximately NT$18.807 billion in Q1. The reported explanation is higher depreciation as production assets and capacity entered service; the decline therefore looks more like a cost-recognition/ramp effect than evidence of a demand collapse.

Near-term sustainability: Higher utilization, favorable advanced-node mix, and continuing AI demand can support positive Arizona earnings. Yet new U.S. fabs carry substantial fixed costs, construction and operating costs, and depreciation before each site reaches mature yields and utilization. TSMC has indicated that overseas expansion could dilute corporate gross margin by roughly 2–3 percentage points; separately, the initial 2nm ramp is expected to weigh on margins.

Technology roadmap and investment effect: Moving from current 4nm production to planned 3nm production in 2027 and around-2nm production around 2028 should improve revenue per wafer and preserve Arizona’s strategic relevance to advanced-chip customers. However, each transition entails a new yield-learning curve, equipment depreciation, and start-up expense, so profitability could be volatile rather than rise smoothly. TSMC’s additional $100 billion Arizona commitment brought planned U.S. investment to $265 billion, supporting long-run scale but raising the capital base that future profits must earn an adequate return on.