TSMC held its annual technology forum on the 24th, US time, and announced that the most advanced N2 process will be mass-produced in 2025, followed by the A16 process to continue the development of advanced processes. TSMC executives said that AI chip manufacturers may become the first adopters of A16, leading smartphone manufacturers.
TSMC A16 will combine Super PowerRail and nanosheet transistors and will be mass produced in 2026. Super Power Rail moves the power supply network to the back of the wafer, freeing up more signal network space on the front of the wafer, improving logic density and performance, and is suitable for high-performance computing (HPC) products with complex signal wiring and dense power supply networks. Compared with TSMC’s N2P process, A16’s speed is increased by 8% to 10% at the same Vdd (operating voltage), power consumption is reduced by 15% to 20% at the same speed, and chip density is increased by up to 1.10 times, supporting data center products.
Market players pointed out that A16 may shake up Intel’s declaration in February that it would regain the chip throne with Intel 14A. Zhang Xiaoqiang, senior vice president of business development at TSMC, said that demand from AI chip companies has made A16 research and development faster than expected, but did not disclose customer information.
AI chip companies are eager to optimize the design to take advantage of the full performance of TSMC’s process. TSMC does not believe that it is necessary to use ASML’s latest high-numerical aperture (High-NA) EUV to produce A16 chips. Intel said last week that it would first use High-NA EUV to produce Intel 14A chips.
TSMC also demonstrated the super rail power supply that will be launched in 2026, supplying power from the back of the chip to help speed up the operation of AI chips. Intel also has similar technology, which is also regarded as a core competitive advantage. However, market participants have expressed doubts about Intel’s declaration. Analysts from TIRIAS Research pointed out that both Intel and TSMC are still several years away from practical applications. They must prove that the final product can reach the current claimed level in order to win the favor of customers.
(Source of the first picture: Photo by Science and Technology News)
Follow TechNews via Google News here
New scientific and technological knowledge, updated from time to time
Tags: TSMCs nanoscale A16 HighNA EUV shattering Intels #14A performance dominance claim
