According to reports, TSMC's 2-nanometer chips are in increasing short supply, with order visibility extending to after 2028

According to Taiwan's Economic Daily, TSMC (TSM.US) is experiencing an expanding supply-demand imbalance for its 2-nanometer chip series. In addition to existing customers such as NVIDIA (NVDA.US), Broadcom (AVGO.US), and MediaTek, who have successively booked capacity for the next two years, Intel (INTC.US) has also joined in on the reservations, with new products expected to be introduced as early as next year. Recently, Meta (META.US) has also entered the competition for production capacity, but due to customers having already pre-booked capacity, the visibility of TSMC's 2-nanometer series orders has been extended to beyond 2028.

The report indicates that NVIDIA is unable to secure sufficient TSMC A16 capacity needed for the next-generation Feynman platform, leaving only the most critical bare chips to use A16, while some bare chip designs will switch to the N3P process. On the other hand, Broadcom has secured all high-bandwidth memory (HBM) and TSMC's advanced process capacity needed for AI chip production until 2028, clearly indicating the advance booking of TSMC's advanced process capacity for the next two years.

TSMC's 2-nanometer chips have been mass-produced as scheduled in the fourth quarter of last year at the Hsinchu and Kaohsiung facilities. The N2P process technology derived from the 2-nanometer chips is expected to begin mass production in the second half of this year, and the A16 process, also part of the 2-nanometer series, will enter mass production in the second half of the year