Taiwan Semiconductor Manufacturing Co. confirmed it began volume production of 2-nanometer chips in the fourth quarter, marking the chipmaker’s transition to gate-all-around nanosheet transistors that promise substantial efficiency gains for artificial intelligence applications.
The world’s largest contract chipmaker disclosed the milestone on its website rather than through a formal announcement, noting that production commenced at its new Fab 22 facility near Kaohsiung. The N2 process delivers up to 15% better performance at identical power consumption, or 25-30% power reduction at the same speed compared with its 3nm predecessor.
TSMC is simultaneously accelerating its U.S. expansion. According to Nikkei, the company plans to begin installing equipment at its second Arizona plant around mid-2026, positioning the facility for 3nm production by 2027—roughly a year ahead of the original 2028 target. Construction of the building has already been completed.
The acceleration reflects sustained demand from American technology giants seeking domestically manufactured chips. TSMC’s first Arizona fab entered high-volume 4nm production in late 2024.
Chief Executive Officer C.C. Wei said the company is preparing to upgrade Arizona operations to 2nm and more advanced technologies given strong AI-related customer interest. A third Arizona plant, designed for 2nm and A16 processes, broke ground in April.
The chipmaker expects roughly 30% of its most advanced production to eventually come from U.S. facilities once its NT$5.2 trillion (US$166 billion) American investment is complete.
