TSMC: There is risk in Taiwan semiconductor supply chain, we should build complete local supply chain within Taiwan

[tonyget]

台積電：台灣半導體有斷鏈風險，應建立完整供應鏈

台積電副總經理余振華今天表示，世界朝本土化發展，台灣半導體供應鏈很長，有斷鏈潛在風險，台灣應建立完整本地供應鏈。 余振華今天出席國際半導體展半導體研發大師論壇，針對人才與產業永續發展，他說，台灣半導體產業特色是專業分工，有設計及晶圓代工等，供應鏈很長，晶圓代工廠要引進設備、材料，設計廠則需要電子設計...

technews.tw

Yu Zhenhua, deputy general manager of TSMC, said today that the world is moving towards localization. Taiwan's semiconductor supply chain is very long and there is a potential risk of chain disruption. Taiwan should establish a complete local supply chain.

Yu Zhenhua attended the Semiconductor Research and Development Masters Forum at the International Semiconductor Exhibition today. Regarding talents and sustainable development of the industry, he said that Taiwan’s semiconductor industry is characterized by professional division of labor, including design and foundry. The supply chain is very long, and foundries need to introduce Equipment, materials, and design factories need electronic design automation (EDA) tools, and it is very dangerous if the supply chain is interrupted.

Yu Zhenhua said that the current world is moving toward localization and does not encourage globalization. He said that some foundries were previously banned from supplying three materials by competing countries, which was greatly affected. The semiconductor industry believes that what Yu Zhenhua mentioned should refer to Japan's control of the export of semiconductor materials to South Korea.

Yu Zhenhua said that there will be more and more new opportunities in the semiconductor industry in the future, and it will develop towards packaging and system integration, requiring new materials and equipment, and the entry threshold is relatively low. Taiwan Semiconductor has the risk of breaking the chain. It is suggested that a complete local supply chain should be established, and schools should start cultivating key talents.

Speaking of innovation, Yu Zhenhua said that encouraging innovation is absolutely necessary, which is one of the reasons for Taiwan's success. He pointed out that TSMC encourages innovation, and it is a disciplined innovation, and promises that the time cannot be delayed. This is an important factor for TSMC's success.

 

[tonyget]

The West doesn't trust Taiwan's resilience in semiconductor supply chain，so they demand TSMC to build fabs outside of Taiwan. I guess the reverse is also true，Taiwan doesn't trust the West either. Now everybody wants to have a complete semiconductor local supply chain within their border，Morris Chang was right，globalization is dead，even the most globalized industry like semiconductor isn't immune to this trend.

 

[blueone]

The West doesn't trust Taiwan's resilience in semiconductor supply chain，so they demand TSMC to build fabs outside of Taiwan. I guess the reverse is also true，Taiwan doesn't trust the West either. Now everybody wants to have a complete semiconductor local supply chain within their border，Morris Chang was right，globalization is dead，even the most globalized industry like semiconductor isn't immune to this trend.

I would conservatively guess it would take more than a decade to de-globalize the semi manufacturing supply chain, if it's possible at all. How are you going to de-globalize ASML, for example? Even ASML can't de-globalize ASML.

 

[delong.height]

I would conservatively guess it would take more than a decade to de-globalize the semi manufacturing supply chain, if it's possible at all. How are you going to de-globalize ASML, for example? Even ASML can't de-globalize ASML.

you can't deglobalize ASML, but there can be a polarized version of AMSL in the East.

 

[cliff]

I would conservatively guess it would take more than a decade to de-globalize the semi manufacturing supply chain, if it's possible at all. How are you going to de-globalize ASML, for example? Even ASML can't de-globalize ASML.

Can we entertain the non conservative guess? Let's assume fewer meeting and getting stuff done. Yes, this is a fantasy for entertainment value...

How difficult is it for the US companies to build DUV lithography machines, the lenses, etc? Is the chips act funding this?

Let's suppose 16/14nm + MRAM (and other exotic memories work) + HBM + 2.5D SIPs are good enough. What prevents North America from isolating? What raw materials are missing?

Just asking...

 

[blueone]

you can't deglobalize ASML, but there can be a polarized version of AMSL in the East.

Very doubtful, even if someone had access to a state of the art machine to disassemble and analyze. In ASML's 2022 Annual Report they talk extensively about partner dependencies, especially with Carl Zeiss SMT.

 

[cliff]

Very doubtful, even if someone had access to a state of the art machine to disassemble and analyze. In ASML's 2022 Annual Report they talk extensively about partner dependencies, especially with Carl Zeiss SMT.

That is why I mentioned lenses. There will be plenty of people put out of work with our automation to work on lenses, no? I would prefer to stick to EDA, but I am considering making a lens company. Is it possible for a non-German to make a lens?

 

[blueone]

Can we entertain the non conservative guess? Let's assume fewer meeting and getting stuff done. Yes, this is a fantasy for entertainment value...

Are you trying to get a rise out of me for entertainment value?

How difficult is it for the US companies to build DUV lithography machines, the lenses, etc? Is the chips act funding this?

Go to any equipment company's annual report, like ASML, AMT, KLA, Lam, etc., and read about their partnership dependencies and risks. These companies took decades to get where they are, and that was with the freedoms of international partnerships and acquisitions. Where would ASML DUV be without Cymer, for example?

Let's suppose 16/14nm + MRAM (and other exotic memories work) + HBM + 2.5D SIPs are good enough. What prevents North America from isolating? What raw materials are missing?

16/14nm isn't good enough if you're a chip designer with competition who has access to 5nm. I realize 16/14 + MRAM is your current plan (you've posted about it), but is that a niche or something big enough to be a national objective for China?

Just asking...

Uh huh.

 

[blueone]

That is why I mentioned lenses. There will be plenty of people put out of work with our automation to work on lenses, no? I would prefer to stick to EDA, but I am considering making a lens company. Is it possible for a non-German to make a lens?

Are you off your meds again? Haven't you heard about Japanese lenses?

 

[cliff]

Yes, I have heard of Canon and Nikon. There was Paul Simon song about one of them.

This thread was a nationalistic one, but now that you brought the Japanese into it, the US and Japan are obvious partners. So let's expand my question to include Japan, at least in the short term. I think Tonygetsit. 10 years... really? I don't think we have 10 years.

 

[blueone]

Yes, I have heard of Canon and Nikon. There was Paul Simon song about one of them.

That was Eastman Kodak. "Kodachrome". Nikon just made the camera.

This thread was a nationalistic one, but now that you brought the Japanese into it, the US and Japan are obvious partners. So let's expand my question to include Japan, at least in the short term. I think Tonygetsit. 10 years... really? I don't think we have 10 years.

We don't have 10 years until what exactly?

 

[cliff]

Thanks for the nitpick.

10 years... I don't want to push this thread too far in that direction. Time is collapsing in several ways. I don't think we need to expand that. This thread is about certain counties being isolated / nationalistic. It is an important topic. I am looking for a western solution. This could include Japan and Australia if we are incompetent.

We are both aware, 5nm is being built in AZ.. at massive costs. Hopefully it will be completed, practical, and protected. IMO, having a self contained solution is more important than exponentially fighting physics, but this is the correct forum to ask these questions. I am looking (and creating) at/for a practical solution in a world that is quickly transitioning from global to nationalistic.

Note: I don't think 180-12nm is a niche, but in this forum it seems to be. Dan has the numbers. Perhaps he agrees with you.

 

[kingmouf]

I think that what our industry needs is a more pluralistic world rather than nationalistic. Despite the fact that I dont really know whether it is even possible to envision the semiconductor design and manufacturing sector in a localised manner, the choices that are available are detrimentally few (for the importance of the industry to the functioning of the whole world). Only a couple of bleeding edge foundries and no more than a handful of fab companies. Only a handful of lithography and equipment makers. Only a trio of EDA suppliers. Only a couple of choices for high performance solutions. Things need to be more pluralistic. Not nationalistic.