TSMC plans to increase 5nm chip shipments in 3Q22
In the early days of TSMC’s N5 process node, Apple took most of the production capacity. However, as more and more companies place orders at this process node and the manufacturing process technology gradually matures, TSMC has correspondingly increased the output of 4nm and 5nm chips.
According to DigiTimes, a semiconductor equipment supplier revealed that TSMC is expected to increase the wafer output of the N5 process node from the current 120,000 wafers per month to 150,000 wafers per month in the third quarter of 2022, an increase of 25% to meet future needs. Order demand from companies like AMD, Nvidia, and MediaTek
TSMC’s N5 process node includes ordinary N5, performance-enhanced N5P, derivative N4, N4P, N4X, and NVIDIA’s customized N4 processes. Customers can choose from a lot of processes. At present, Apple uses N5 and N5P processes for chips such as the M1 series and A14/A15 Bionic. It is expected that A16 Bionic will use N4 or N4P processes. MediaTek’s Dimensity 9000 uses the N4 process, while the Dimensity 8100/8000 uses the N5 process. NVIDIA’s latest Hopper architecture GPU uses a custom N4 process, followed by the consumer-grade Ada Lovelace architecture GPU. This year, AMD’s Zen 4-based CPU and RDNA 3-based GPU will also use TSMC’s N5 process node.
Both Nvidia and AMD have spent billions of dollars to make sure they can get the chips they need next at TSMC. In addition, TSMC plans to mass-produce the N3 process node in the second half of 2022, with Apple as the first customer, followed by Intel. The first 3nm chips are expected to ship in 2023, not seen in 2022.