TSMC Said to Start Trial Run of 2nm Chip, Backed by Nvidia AI System

BEIJING, June 6 (TiPost)— Taiwan Semiconductor Manufacturing Co. (TSMC), the world’s leading chip maker for other companies, was said to closer to mass produce its most advanced offering as schedule, with the help of Nvidia, the artificial intelligence (AI) semiconductor king.

Source: TiPost

TSMC has started trial run of chips using 2-nanometer (nm) process on Monday, and Apple and Nvidia are set to be the first clients of the product, the local economic and business newspaper Economic Daily News reported on Tuesday. TSMC was reported to begin to deploy engineers to its facility located in the Hsinchu Science Park, a key tech industrial park in Taiwan, and its research and development (R&D) team was expected to have more than 1,000 people. The company would initially build a small production line in the facility for the pilot, and aimed to make about 1,000 2nm chips by the end of this year, backed by, according to the report.

While not commenting on details of the repot, TSMC said development of 2nm technology went very well, and reiterated its target to kick off mass production in 2025. That was in line with the timeline the chipmaker provided in late 2022, when it announced to start to mass produce 3nm chips and expected demand for the product to be “very strong”.

TSMC has seen recovery signs of demand at some of end markets and the company could have a slight negative growth this year, while is ready for the next wave of growth, the chairman Mark Liu said at the 2023 annual shareholding meeting on Tuesday. Liu forecasted sales to decline in single-digit this year, and sales in U.S. dollar for the first half of year to fall 10% year-over-year (YoY).  Liu highlighted orders of TSMC increased quite a lot due to the booming generative AI, and the company is preparing for capacity expansion to meet such new demand.

TSMC CEO C.C. Wei also warned at the shareholder meeting that sales could slide 10% YoY, but showed more optimistic on performance for the second half of the year and 2024. At an earnings call in April, Wei told analysts sizable 3nm revenue contribution was set to start in third quarter and would contribute mid-single-digit percentage of our total wafer revenue in 2023.  

The AI system co-developed by Nvidia and TSMC will put into use for the latter’s 2nm chip production in June, Nvidia founder and CEO Jensen Huang said at the annual GPU Technology Conference, or GTC, in March. Calculation time of the system is expected to be just a fortieth compared with the previous, and the system would further streamline the mass production timetable, reduce their carbon footprint and set the foundation for 2nm and beyond, according to the executive.

Nvidia is the semiconductor designer that dominates the market for artificial intelligence (AI) chips, which empower AI systems including the large language model behind ChatGPT. The behemoth said in March it has modified some of flagship products including A100 and H100 for exports to China, as the U.S. regulators last year banned it from selling its most advanced chips to China.

Many media outlets said last week that Huang would visit mainland China this month despite ongoing China and U.S. tensions. The state-run newspaper Shanghai Securities News learned that Huang could arrive in Shanghai on June 6, or Tuesday. However, Huang was reported earlier this week to return the United States, skipping the planned trip to mainland China after wrapping up his travel to Taiwan.

更多精彩内容，关注钛媒体微信号（ID：taimeiti），或者下载钛媒体App

​Read More