
According to reports, following advancements by Intel and Samsung, TSMC is accelerating its U.S. production plans. The company is significantly ramping up 3nm process manufacturing at its Arizona facility, with mass production now expected to start in 2027—almost a year earlier than initially planned.
Considering the scale and investment, TSMC's Arizona plant ranks among its largest projects. Originally, the company intended to begin U.S. production in 2025 in support of the "Made in America" initiative, planning up to $30 billion in investment to build a resilient local supply chain. Current developments, however, suggest an even more ambitious approach.
TSMC's first Arizona factory has already begun mass production with its 4nm process. The second facility is set to handle 3nm production, targeting 2027 for ramp-up. A key driver for this acceleration is the strong demand from high-performance computing (HPC) clients, which occupy a substantial share of TSMC's chip capacity and require advanced nodes such as 4nm, 3nm, and even 2nm. The ongoing AI boom further reinforces the need to expand overall production capacity, motivating the company to fast-track the Arizona upgrade.
Another important factor is mounting regional competition. In addition to Intel's progress with the 18A (1.8nm) process, Samsung is emerging as a formidable contender. Samsung plans to advance its Taylor fab with the SF2 (2nm) process, bypassing the originally planned 4nm, and has secured key agreements with clients like Tesla, signaling that alternatives to TSMC are gaining traction.
Given rising capital expenditures and labor shortages, how TSMC manages such a massive wafer fab network—especially while exploring expansion in Japan—will be closely watched. Still, with surging market demand, expanding production capacity appears not just strategic but essential.