TSMC announced a second $100 billion investment in U.S. chipmaking capacity on July 16, bringing total U.S. commitments to $265 billion. The additional capex will fund four new fabrication plants in Arizona beyond the six previously planned, focused on 2-nanometer and below process nodes. The move follows an existing $165 billion commitment already underway. TSMC CEO C.C. Wei cited strong multiyear demand from leading U.S. customers (primarily Nvidia and Apple) and said AI-related demand globally remains 'extremely robust,' predicting strong demand through 2029–2030.
In its Q2 2026 earnings, TSMC reported record net profit of $22 billion (706.6 billion Taiwan dollars), up 77% year-over-year, with revenue of $39 billion (up 36% YoY). The company raised its full-year 2026 revenue growth forecast to 40%+ from a prior forecast of 30%+, reflecting sustained AI inference and training capacity demand. TSMC also increased its annual capital expenditure budget to $60–64 billion from an earlier $52–56 billion estimate, signaling aggressive infrastructure expansion across Taiwan, Japan, and the U.S.
For architects: TSMC's $265B total U.S. footprint effectively locks in foundry capacity for the 2026–2030 period. With Q2 profitability and utilization at record highs, TSMC is now the primary supply constraint-and-release lever for advanced AI chips (NVIDIA GPUs for Blackwell/Vera Rubin, Apple Silicon, custom AI accelerators). The capex expansion and confidence signal suggest no shortage of physical fab capacity for training or inference clusters planning 2027–2029 deployments.