Multiple media reports indicate that TSMC will cease contract manufacturing of 7nm and smaller node processes for Chinese AI chips from today. Whether it will continue to manufacture mobile phone chips has yet to be determined, but prospects are not optimistic under US pressure. Notably, not only TSMC but also Samsung have been asked to stop contract manufacturing 7nm and smaller node processes for Chinese chips.

Currently, most domestic AI chips are manufactured by TSMC using 7nm processes, while some mobile phone chips are produced using 6nm processes. Chinese chips contribute about 10% of TSMC's revenue.

Amidst the pressure on TSMC, its original intention was to stop contract manufacturing 7nm and smaller node processes for Chinese chips. However, there are reports that the US has demanded that TSMC cease all contract manufacturing services for Chinese chips using 7nm and smaller node processes, a significantly stricter requirement. Nevertheless, as the US president is still in transition, it is uncertain whether this stricter restriction will be immediately implemented.

Globally, Samsung is another company capable of providing 7nm and smaller node processes. However, Samsung's contract manufacturing services for Chinese chips have been quite limited, and Samsung tends to follow US directives closely. Naturally, Samsung will swiftly comply with US requests. In fact, Samsung is struggling in chip contract manufacturing, having lost almost all clients for 5nm and smaller node processes, causing its market share to plummet from around 20% to 13%.

While this US move will undoubtedly deal a significant blow to Chinese chips in the short term, China is not without alternatives. Moreover, it may accelerate the development of Chinese chip manufacturing processes and strengthen the domestic chip industry chain.

In terms of chip manufacturing, China's largest chip manufacturer can already mass-produce 14nm and larger node processes, and the N+1 process has achieved mass production. Industry insiders speculate that the N+1 process is close to TSMC's first-generation 7nm process, which does not use EUV lithography.

China's N+1 chip process uses DUV lithography with multiple exposure techniques, resulting in lower production yields and higher costs. In terms of performance, it is slightly inferior to TSMC's 7nm EUV process, which explains why most Chinese AI chips are manufactured by TSMC using 7nm processes.

With TSMC discontinuing 7nm and smaller node processes, Chinese AI chips will now be produced by domestic foundries, generating substantial revenue for these companies. This will accelerate the improvement of N+1 process yields and provide more funds for research and development of 5nm processes.

Additionally, other methods to enhance chip performance include chip stacking and chiplet technology. However, these technologies are less suitable for mobile phone chips due to their high power consumption requirements. They are more feasible for AI chips, PC processors, server chips, and other less power-sensitive applications. Combining domestic chip manufacturing processes with these technologies can meet most needs.

The US demand has already dealt a significant blow to the global chip industry, causing a significant drop in the stock price of lithography company ASML. This highlights global concerns that US actions may further reduce revenues for chip equipment and material companies, especially considering China's status as the world's largest market.

More revenue from the Chinese chip market will flow into the domestic chip industry. In fact, previous US actions have prompted China to accelerate the development of its domestic chip industry chain. In recent years, revenues of domestic chip equipment and material companies have grown rapidly. With increased revenue, China's chip industry chain has been rapidly improved, and domestic chip manufacturers are increasingly adopting domestic equipment. China's largest chip equipment company has become the world's sixth-largest, causing concern among US, Japanese, and Dutch counterparts.