This is not investment advice. The author has no position in any of the stocks mentioned. Wccftech.com has a disclosure and ethics policy.
The Taiwan Semiconductor Manufacturing Company (TSMC) has commenced pilot production of its next-generation 3-nanometer semiconductor manufacturing node. This report comes the courtesy of the Taiwanese publication Digitimes, which also reports that executives from U.S. chip giant Intel Corporation are set to visit Taiwan later this month to finalize orders for the 3nm process. TSMC's manufacturing technologies often have a limited capacity during the early stages of their lifecycle, which sees only a few companies being able to procure products manufactured through them.
TSMC's Initial 3nm Production Capacity Will Be Limited To 40,000 Wafers Per Month Believes Report
The report about 3nm pilot production comes as TSMC and Korean chaebol Samsung Group's semiconductor manufacturing arm Samsung Foundry are neck to neck when it comes to estimated mass production timelines for next generation semiconductor manufacturing technologies. While Samsung expects that its 3nm process will be ready during the first half of next year, TSMC's chief executive officer Dr. C.C. Wei has outlined that his company will enter mass production during H2 2022.
TSMC's 3nm pilot production kickoff meets Dr. Wei's estimates given earlier this year at an investor call. His company is marketing the manufacturing technology as its most advanced to date and believes that it will be able to reap the dividends from it for years to come.
With transistor sizes continuing to shrink, it takes chip fabricators years to develop new technologies, even as TSMC maintains that it is on track to double its process technology performance every two years.
Furthermore, as outlined by @chiakokhua on Twitter, Intel is eager to make sure that it is the first in line when it comes to procuring TSMC's 3nm products. Generally, the freshest silicon off of the Taiwanese company's manufacturing lines is thought to go to Cupertino tech giant Apple, Inc, which then uses them to maintain its technological advantages in smartphones and notebooks.
According to @chiakokhua, who also quotes Digitimes:
Intel execs to visit Taiwan mid-Dec to:
- Finalize scope of cooperation with TSMC.
- Solicit/ensure N3 capacity will not be affected by Apple.
- Initiate discussions about cooperation for N2.
- First wave of N3 capacity <60K, will only reach 40K+ wpm in 1H'23.
Rumors of companies soliciting TSMC for the top spots for its latest chip technologies have also surfaced before, with Santa Clara semiconductor designer Advanced Micro Devices, Inc (AMD) and San Diego firm Qualcomm Incorporated both rumored to be preferring Samsung's 3nm process over its Taiwanese counterpart.
More interestingly, Digitimes's belief that Intel will also discuss cooperating with TSMC for the 2nm process is telling, since the Santa Clara, California company also has plans for its own 2nm node. Dubbed as 20A (angstrom), this technology might make its way to the production line in 2024, according to Intel's roadmap revealed in July this year.
This might be a full year earlier than TSMC's 2nm node, which can enter production in 2025 if the back of the hand estimates given by Dr. Wei turn out to be accurate.
Intel's cooperation with TSMC for the 3nm process has been reported since January 2021, and the latest report suggests that the pair will work together on a graphics processing unit (GPU) and three datacenter central processing units (CPUs). Both are crucial products for Intel, which is yet to solidify its place in the rapidly growing GPU market and is competing with AMD in the data center segment, with the latter consistently demonstrating strong sequential and annual revenue growth in the segment.
