TSMC is expected to reach a record $12.7 billion in revenue from its 3nm process in the second half of the year, surpassing the 5nm process. In the previous quarter, 5nm technology accounted for 33% of total revenue, while 3nm technology grew from 25% in the first quarter to 30%.
The company is scaling production for key clients including Nvidia, AMD, and Broadcom, with Nvidia being the primary driver. Wedbush analysts predict monthly 3nm wafer processing will increase from approximately 150,000 to 180,000 units by the next quarter.
TSMC is expanding its global infrastructure to support this growth. A new plant in Tainan, Taiwan, will start operations in the first half of next year, while a second facility in Arizona, USA, will begin 3nm production in the second half of next year. A second plant in Japan (JASM) is slated for 3nm output by 2028.
The company is also developing sub-2nm processes. The A16 process is being adapted for mass production this half-year. This technology will be used by Apple, Nvidia, and OpenAI. OpenAI's second-generation Jalapeno chips will combine A16 with CoWoS packaging, while Nvidia may use A16 for its Feynman GPUs and Rosa CPUs.
