Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/tsmc-ceo-flags-3-nanometre-chip-production-in-japan-investment-reported-at-17-billion.24495/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2030871
            [XFI] => 1060170
        )

    [wordpress] => /var/www/html
)

TSMC CEO flags 3-nanometre chip production in Japan, investment reported at $17 billion

Daniel Nenni

Admin
Staff member
56d1d7ccb626abd41b7ca182b83b98fc

Japan's Prime Minister Sanae Takaichi meets TSMC's CEO C.C. Wei in Tokyo

TOKYO/TAIPEI, Feb 5 (Reuters) - TSMC plans to mass produce advanced 3-nanometre chips in Kumamoto in southern Japan, TSMC CEO C.C. Wei said on Thursday, an investment local media reported was worth $17 billion as the Taiwanese firm seeks to meet soaring demand for AI chips.

TSMC, the world’s ‌largest contract chipmaker and a key supplier of AI chips to companies such as Nvidia, currently produces its most advanced chips in ‌Taiwan. Previous plans for Japan had focused on less advanced technologies.

The announcement puts Japan on the map as the newest location for the production of the high-end 3-nanometre chips, which are widely used in high-performance computing and AI servers. The company plans to begin producing these chips at its second fab in Arizona in 2027.

The ⁠boom in AI has helped TSMC, Asia's ‌most valuable listed company, outshine its rivals.

"We believe this fab will further contribute to the local economic growth and, most importantly, form a foundation for Japan's AI ‍business," said Wei in Tokyo at a meeting with Japan’s Prime Minister Sanae Takaichi.

Wei added that the company was also in discussions with Japanese customers and partners on further collaboration across multiple areas important to the AI industry.

Takaichi said 3-nanometre chips hold great significance for economic security.

TSMC told Reuters in an email that the company is now ‌planning to use 3nm process technology for production at its second fab in Japan to meet strong demand driven by AI.

Japanese newspaper Yomiuri reported earlier on Thursday that the investment will be $17 billion. TSMC declined to comment on the reported investment figure.

Japan's government, which has subsidized Taiwan's leading chipmaker to build out capacity in Kyushu, is considering additional support for the new investment plan, Yomiuri reported.

TSMC had planned to invest $12.2 billion ⁠on 6-12 nanometer chipmaking capacity at its second fab in Kyushu but will discuss changes to the plan with Japan's government, Yomiuri reported.

The company said on an earnings call in January that the construction of its second fab in Japan has started, and "the technologies and ramp schedule will be ‍based on our customers' need ⁠and market conditions."

Japan is also heavily subsidizing homegrown foundry venture Rapidus, which will produce cutting-edge chips on the northern island of Hokkaido.

The government has determined the two companies' chips will have different ⁠uses and will not be in competition, Yomiuri reported.

Securing access to chips, which are critical for electronics, automotive and defence industries, ‌has become a priority for governments around the world.

 
Japan is also heavily subsidizing homegrown foundry venture Rapidus, which will produce cutting-edge chips on the northern island of Hokkaido.

The government has determined the two companies' chips will have different ⁠uses and will not be in competition, Yomiuri reported.
That's interesting, how does it know that?
 
56d1d7ccb626abd41b7ca182b83b98fc

Japan's Prime Minister Sanae Takaichi meets TSMC's CEO C.C. Wei in Tokyo

TOKYO/TAIPEI, Feb 5 (Reuters) - TSMC plans to mass produce advanced 3-nanometre chips in Kumamoto in southern Japan, TSMC CEO C.C. Wei said on Thursday, an investment local media reported was worth $17 billion as the Taiwanese firm seeks to meet soaring demand for AI chips.

TSMC, the world’s ‌largest contract chipmaker and a key supplier of AI chips to companies such as Nvidia, currently produces its most advanced chips in ‌Taiwan. Previous plans for Japan had focused on less advanced technologies.

The announcement puts Japan on the map as the newest location for the production of the high-end 3-nanometre chips, which are widely used in high-performance computing and AI servers. The company plans to begin producing these chips at its second fab in Arizona in 2027.

The ⁠boom in AI has helped TSMC, Asia's ‌most valuable listed company, outshine its rivals.

"We believe this fab will further contribute to the local economic growth and, most importantly, form a foundation for Japan's AI ‍business," said Wei in Tokyo at a meeting with Japan’s Prime Minister Sanae Takaichi.

Wei added that the company was also in discussions with Japanese customers and partners on further collaboration across multiple areas important to the AI industry.

Takaichi said 3-nanometre chips hold great significance for economic security.

TSMC told Reuters in an email that the company is now ‌planning to use 3nm process technology for production at its second fab in Japan to meet strong demand driven by AI.

Japanese newspaper Yomiuri reported earlier on Thursday that the investment will be $17 billion. TSMC declined to comment on the reported investment figure.

Japan's government, which has subsidized Taiwan's leading chipmaker to build out capacity in Kyushu, is considering additional support for the new investment plan, Yomiuri reported.

TSMC had planned to invest $12.2 billion ⁠on 6-12 nanometer chipmaking capacity at its second fab in Kyushu but will discuss changes to the plan with Japan's government, Yomiuri reported.

The company said on an earnings call in January that the construction of its second fab in Japan has started, and "the technologies and ramp schedule will be ‍based on our customers' need ⁠and market conditions."

Japan is also heavily subsidizing homegrown foundry venture Rapidus, which will produce cutting-edge chips on the northern island of Hokkaido.

The government has determined the two companies' chips will have different ⁠uses and will not be in competition, Yomiuri reported.

Securing access to chips, which are critical for electronics, automotive and defence industries, ‌has become a priority for governments around the world.


This is an interesting development. There is a high possibility that Germany and EU/Europe will do the similar thing to encourage/subside TSMC to build second German fab for leading edge nodes.
 
This is an interesting development. There is a high possibility that Germany and EU/Europe will do the similar thing to encourage/subside TSMC to build second German fab for leading edge nodes.

I agree. TSMC N3 is the last of FinFETs and will most definitely be manufactured in Japan, USA, Japan, and maybe even India. You never know. It is Deja vu 28nm which was another dominant TSMC process.
 

"Japan is also heavily subsidizing homegrown foundry venture Rapidus, which will produce cutting-edge chips on the northern island of Hokkaido. The government has determined the two companies' chips will have different ⁠uses and will not be in competition, Yomiuri reported."

That's interesting, how does it know that?

Interesting. When Rapidus first launched they were positioned to beat TSMC N2. I met with them last year and suggested that going against TSMC is a very bad strategy. Many fabs have tried and many fabs have failed. Maybe they listened to me? I introduced them to the NOT TSMC market concept, hopefully they took it to heart because as capacity constrains the NOT TSMC market strengthens, absolutely.
 
I agree. TSMC N3 is the last of FinFETs and will most definitely be manufactured in Japan, USA, Japan, and maybe even India. You never know. It is Deja vu 28nm which was another dominant TSMC process.

We will see. I'm not sure if Europe / Germany has presently the funding to support a 20-30 B$ investment by TSMC for a 3nm plant. Remember, for the ESMC fab in Dresden this was in a joint venture with 3 other (European) companies. TSMC got 5 BEuro from the German goverment as real support.

The EU/Nato countries need to invest 5% of their GDP for defense, related to security and the war in Ukraine. This puts a lot of strain on national budgets.

https://www.prnewswire.com/news-rel...024 /PRNewswire/ --,a boost to the initiative.
 
We will see. I'm not sure if Europe / Germany has presently the funding to support a 20-30 B$ investment by TSMC for a 3nm plant. Remember, for the ESMC fab in Dresden this was in a joint venture with 3 other (European) companies. TSMC got 5 BEuro from the German goverment as real support.

The EU/Nato countries need to invest 5% of their GDP for defense, related to security and the war in Ukraine. This puts a lot of strain on national budgets.

https://www.prnewswire.com/news-releases/tsmc-breaks-ground-secures-subsidy-for-german-plant-302226476.html#:~:text=20, 2024 /PRNewswire/ --,a boost to the initiative.

Agreed, it would have to be a joint venture like last time. TSMC, along with European partners Robert Bosch GmbH, Infineon Technologies AG, and NXP Semiconductors, established a joint venture called European Semiconductor Manufacturing Company (ESMC) GmbH to build a semiconductor fabrication plant in Dresden, Germany. Something like that but maybe in France, Italy, or the UK? Ireland may have some skilled fab workers to spare?
 
"Japan is also heavily subsidizing homegrown foundry venture Rapidus, which will produce cutting-edge chips on the northern island of Hokkaido. The government has determined the two companies' chips will have different ⁠uses and will not be in competition, Yomiuri reported."


Interesting. When Rapidus first launched they were positioned to beat TSMC N2. I met with them last year and suggested that going against TSMC is a very bad strategy. Many fabs have tried and many fabs have failed. Maybe they listened to me? I introduced them to the NOT TSMC market concept, hopefully they took it to heart because as capacity constrains the NOT TSMC market strengthens, absolutely.
Dan puts this well. No one is really saying "I will look at TSMC and Rapidus in parallel to see who wins". If you are "not TSMC" person, then you look at Intel, Rapidus, Samsung etc.

you cannot have a sane comparison of TSMC and Rapidus.

even Not TSMC people are wildly different:

1) Rapidus: wants to do a foundry, has never shipped anything and is basing their process of IBM technology
2) Samsung: Huge company, lots of foundry experience, Lots of historical execution issues,
3) Intel: Huge company, wants to do foundry, minimal foundry experience, Leading technology, unknown execution

When these decisions are made: the simple question I would ask (like you do anytime you do not go with the leader). "why would you choose someone other than TSMC?"
 
Agreed, it would have to be a joint venture like last time. TSMC, along with European partners Robert Bosch GmbH, Infineon Technologies AG, and NXP Semiconductors, established a joint venture called European Semiconductor Manufacturing Company (ESMC) GmbH to build a semiconductor fabrication plant in Dresden, Germany. Something like that but maybe in France, Italy, or the UK? Ireland may have some skilled fab workers to spare?

I cannot imagine CC Wei or one of his successors will ever build another fab outside the Dresden area, this area is clearly becoming the oil producer for EU-based chip production:
https://business-saxony.com/en/a-business-location-at-its-best/strong-industries/silicon-saxony-more-than-just-chips#:~:text=Industry Diversity,is "Made in Saxony".

I'm sure TSMC will first want to see how the ESMC fab is ramping in 2027? Enough other fabs to build for TSMC these days in Taiwan and Arizona.....
 
Dan puts this well. No one is really saying "I will look at TSMC and Rapidus in parallel to see who wins". If you are "not TSMC" person, then you look at Intel, Rapidus, Samsung etc.

you cannot have a sane comparison of TSMC and Rapidus.

even Not TSMC people are wildly different:

1) Rapidus: wants to do a foundry, has never shipped anything and is basing their process of IBM technology
2) Samsung: Huge company, lots of foundry experience, Lots of historical execution issues,
3) Intel: Huge company, wants to do foundry, minimal foundry experience, Leading technology, unknown execution

When these decisions are made: the simple question I would ask (like you do anytime you do not go with the leader). "why would you choose someone other than TSMC?"

Why pursue the NOT TSMC market? If I was a semiconductor CEO I would be concerned with:

Wafer Pricing. The foundry business thrived on second and third source manufacturing to keep wafer prices down. Samsung is famous for selling on price. We do not have that ability at 3nm and 2nm as of yet.

US Based Manufacturing. This is a hot political issue with tariffs etc...

Capacity which we now know is a problem and may continue to be. Either way it is a lesson learned for the future.

Bottom line: Intel 18A needs to ramp up external customers ASAP followed by Samsung and Rapidus. Otherwise TSMC will run away with 2nm plus 14A and that may be the death of the competitive foundry business, absolutely.



 
We will see. I'm not sure if Europe / Germany has presently the funding to support a 20-30 B$ investment by TSMC for a 3nm plant. Remember, for the ESMC fab in Dresden this was in a joint venture with 3 other (European) companies. TSMC got 5 BEuro from the German goverment as real support.

The EU/Nato countries need to invest 5% of their GDP for defense, related to security and the war in Ukraine. This puts a lot of strain on national budgets.

https://www.prnewswire.com/news-releases/tsmc-breaks-ground-secures-subsidy-for-german-plant-302226476.html#:~:text=20, 2024 /PRNewswire/ --,a boost to the initiative.

Around the world, anything related to national security and sovereignty is not limited to cost considerations alone. Trump’s political decisions and actions, including those related to Greenland, have forced Europe to change. The discussion has shifted from how to defend Europe from Russian aggression, with or without US support, to how to defend Europe (including Canada) from US aggression.

There are typical political and bureaucratic debates and delays over the price tag of a leading-edge TSMC Germany fab, but the direction is clear.
 
Last edited:
Around the world, anything related to national security and sovereignty is not limited to cost considerations alone. Trump’s political decisions and actions, including those related to Greenland, have forced Europe to change. The discussion has shifted from how to defend Europe from Russian aggression, with or without US support, to how to defend Europe (including Canada) from US aggression.

There are typical political and bureaucratic debates and delays over the price tag of a leading-edge TSMC Germany fab, but the direction is clear.

Let's first see what the EU commission publishes, perhaps at the end of March 2026, regarding EU Chips Act 2:

https://www.europarl.europa.eu/legi...sperity-and-competitiveness/file-chips-act-ii
CONTENT In its 2026 Work Programme, the European Commission announced plans to publish a proposal for a Chips Act II in the first quarter of 2026. The Commission’s latest timeline indicates that the Chips Act II proposal could be released on 25 March 2026.
 
Agreed, it would have to be a joint venture like last time. TSMC, along with European partners Robert Bosch GmbH, Infineon Technologies AG, and NXP Semiconductors, established a joint venture called European Semiconductor Manufacturing Company (ESMC) GmbH to build a semiconductor fabrication plant in Dresden, Germany. Something like that but maybe in France, Italy, or the UK? Ireland may have some skilled fab workers to spare?

https://www.cna.com.tw/news/afe/202602050149.aspx
....................................................................
Experts predict that TSMC's advanced process capacity in Taiwan, the United States, and Japan will still be unable to meet the strong demand from AI customers in the future. It is expected that the Dresden plant in Germany may follow suit and adjust its process technology planning, which is worth tracking and observing.
 
Back
Top