TSMC made an important decision, and Liu Deyin expressed his attitude
TSMC officially announced that the start of production of its U.S. factory has been postponed again.
After the change of chip manufacturing rules, Liu Deren announced that he would build a factory in the United States, investing more than 40 billion yuan, not only to build a 5-nanometer chip factory, but also to build a 3-nanometer chip factory, and directly produce 4-nanometer chips after the 5-nanometer factory is completed.
Liu Deyin also said that it is impossible to build a chip production line using non-American technology, and it is in the interests of TSMC to build a factory in the United States.
However, three years on, production in the U.S. has struggled to get up and running, with Dane Liu announcing last year that it would postpone mass production of 4nm chips in the U.S.
TSMC has now announced its decision to postpone the production of 3nm chips in the United States, which was originally scheduled for 2026, to 2027 or 2028.
In fact, the outside world's attitude towards TSMC's construction of a factory in the United States is not unanimously optimistic.
Zhang Zhongmou has publicly stated that TSMC has been building factories in the United States for 20 years, but has not expanded production.
The chairman of TSMC said that due to the high cost of building factories, lack of water and electricity, and shortage of skilled labor, TSMC relies on Taiwan's mature industrial chain, low-cost chip production, high international standards and profits, and building a factory in the United States will lose these advantages. He said.
TSMC has repeatedly delayed building factories in the United States, of course, because the United States requires too much data, otherwise it will not give subsidies; When TSMC invested in building the factory, the United States promised to give 15 billion yuan in subsidies, but it was too late.
Therefore, Liu Deying believes that foreign decisions are based on customer needs and the level of subsidies and support required locally.
According to people familiar with the matter, the construction of TSMC's U.S. factory may be postponed indefinitely.
This is a unilateral decision made by Liu Deying, who has already announced his resignation. He wanted Wei Zejia to succeed him and postponed the construction of the American plant indefinitely.
In contrast to Liu Deren, Wei Zejia focuses on Taiwan Province and the domestic market.
Wei Zejia publicly said that new energy vehicles will make the chip process mature and bring huge opportunities, which are greater than smartphones, and Japan is the world's largest new energy vehicle production and sales market.
Wei Zejia not only led a team to participate in the Shanghai forum, but also visited important customers in China.
Wei Tze-Chia revealed that TSMC plans to stay in Taiwan, build multiple 2nm chip factories, and open a new global R&D center.
Of course, the reason for TSMC's sudden change of hexagram is the rapid development, subsidies and high costs of domestic manufacturers such as Huawei.
According to Bloomberg, the Kirin 9000s chip is made in Japan and is built on a 7nm or near-7nm process, but the performance is comparable to Qualcomm's Snapdragon 888 chip with a 5nm process.
ASML said that Huawei and other companies are independently developing the semiconductor chip industry chain to subvert the global industrial chain; Nvidia's Huang Jianxun said that Huawei is one of the strongest competitors in artificial chip manufacturing.
Huawei also announced that its chip self-sufficiency rate has reached 70% and urged domestic manufacturers to launch more products, including domestic chips.
Most importantly, SMIC has completed the expansion of wafer production, with a monthly output of more than 100 million chips of various types, and 75% of chip orders come from domestic manufacturers.
The domestic industry also has the capacity to develop duvi lithography technology, which can reduce the chip process to 7nm.
ASML has again delivered all prepaid lithography orders, including models such as the 2000i and other DUV lithography machines that can scale chip processes down to 5nm.
SMIC has already completed the R&D of 7nm and started the R&D of 5nm; With a DUV lithography machine like the 2000i, the results are predictable. Otherwise, Zhang Zhongwu would not have suddenly advocated for the United States to launch a chip war against China.