Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/tsmc-to-build-fab-in-arizona-and-they-are-hiring.13209/page-2
        )

    [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] => 2021370
            [XFI] => 1050270
        )

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

TSMC to Build Fab in Arizona and They are Hiring!

@hist78 - USA IC manufacturing did not make a good transition to 300 mm (only a handful went there in the end) and did terrible at adapting to evolution (dominance?) of the foundry model.
We are still deriving many dollars out of the IC business in design, software and equipment, just not in physical factories. There is no real interest in long term changes that might revive manufacturing (healthcare cost, longer term outlook, worker training). Focus is on M&A. Additionally, we have a severe collective case of Nimbo that makes it impossible to quickly and simply put up a factory near large population centers. In the last 20 odd years, we had precious few green field factories, mostly expansion of existing ones.
Taichung, Tainan, Hsinchu Science parks and their counterparts in Korea are all located near major towns where they can attract all the talent they need (and a lot it needed). In the bitter end, fabs overseas did many things right (things that were congruent with their culture). That made all the difference in the end.
We will have several successful factories here, nearly all in 200 mm and many in niched areas (SiC, SOI etc.) that overcome issues mentioned above.
@Portland - We had an election, it is over, let's collectively move on. We'll be waiting for you on the other side.
 
@hist78 One last point that has a major impact on manufacturing: the value of US dollar is disconnected from "real" economy. The US dollar is both our domestic currency and a global instrument to settle payments between all major countries. Until a significant amount of 3rd party transactions are settled in Renminbi or Euros, manufacturing will not come back. The new Nafta was aiming to effectively change the peso to dollar exchange rate (via hourly wage rate) but the $ value would rise if there was an earthquake in Indonesia, a civil war in Iran etc, irrespective of hourly productivity in US economy vis a vis Mexico. Make no mistake about it, Taiwan CAREFULLY monitors (and effectively controls) the value of the NT$. That effect is a big part of the cost advantage of doing manufacturing outside of the USA.
 
Digitimes (https://www.digitimes.com/news/a20201113VL200.html?chid=9) indicates that TSMC "has already placed orders for at least 13 EUV sets with AMSL for 2021 in line with its 5nm capacity expansion and its advance to 3nm production". Considering that Anandtech (https://www.anandtech.com/show/16042/tsmc-we-have-50-of-all-euv-installations-60-wafer-capacity) estimated that TSMC currently "has around 30-35 [ASML machines], this could represent at least a 37% expansion of its 3nm/5nm manufacturing capabilities.
 
@hist78 - USA IC manufacturing did not make a good transition to 300 mm (only a handful went there in the end) and did terrible at adapting to evolution (dominance?) of the foundry model.
We are still deriving many dollars out of the IC business in design, software and equipment, just not in physical factories. There is no real interest in long term changes that might revive manufacturing (healthcare cost, longer term outlook, worker training). Focus is on M&A. Additionally, we have a severe collective case of Nimbo that makes it impossible to quickly and simply put up a factory near large population centers. In the last 20 odd years, we had precious few green field factories, mostly expansion of existing ones.
Taichung, Tainan, Hsinchu Science parks and their counterparts in Korea are all located near major towns where they can attract all the talent they need (and a lot it needed). In the bitter end, fabs overseas did many things right (things that were congruent with their culture). That made all the difference in the end.
We will have several successful factories here, nearly all in 200 mm and many in niched areas (SiC, SOI etc.) that overcome issues mentioned above.
@Portland - We had an election, it is over, let's collectively move on. We'll be waiting for you on the other side.

To revive US semiconductor industry is definitely not an easy task. Actually many the same challenge that many US industries and business are facing too. To name a few of them:

1. Lacking a long-term nationwide industrial policy. We have many large or small projects and initiates here and there, either by federal or states. But few people are willing to address those fundamental issues.

2. Unfair public policy and tax treatment across individuals, industries and regions. For example, a large business can threaten to move to another state in order to get special tax rebate or public subsidy from local and federal government. At the same time small to medium size business don't have the political power to get something meaningful. They are left behind to fight their own survival.

3. Mistakenly worship many merge and acquisition deals that they are merely playing the number or just the work of scam artists. To make it even worst, too often those scammers can get away by out of court settlements and suffer little or no consequence.

4. Combine above issues and many other factors, frequently we are witnessing a business or a factory was established for a wrong product at a wrong place at a wrong time by a wrong group of people. There is no sufficient debate and discussion beforehand.

Using the proposed TSMC Phoenix Arizona fab as an example. I can understand and agree some of the reason behind it, such as the abundant tech workforce, good education system, and good weather, etc. But is there ever any thorough discussion on it? As a nation we are going to subside it but as a nation we should ask where and what is the best approach to build such important fab too.

Unfortunately, Arizona is an important swing state to both Republicans and Democrats. Any necessary and healthy discussion disappeared. Many similar kind of dramas have been repeating across US again and again. What kinds of wonderful result can we expect under this kind of practice?
 
Last edited:
Digitimes (https://www.digitimes.com/news/a20201113VL200.html?chid=9) indicates that TSMC "has already placed orders for at least 13 EUV sets with AMSL for 2021 in line with its 5nm capacity expansion and its advance to 3nm production". Considering that Anandtech (https://www.anandtech.com/show/16042/tsmc-we-have-50-of-all-euv-installations-60-wafer-capacity) estimated that TSMC currently "has around 30-35 [ASML machines], this could represent at least a 37% expansion of its 3nm/5nm manufacturing capabilities.
It would be too rough(13/35=37%) to estimate 3nm/5nm manufacturing capabilities expansion, because as reported tsmc would implement >20 EUV masking layers in 3nm but 5nm using only ~14 layers. Based upon ASML capacity model, I made the following simple calculation:
1. 35 EUV tools for 5nm( 14 layers) could support 35/14*45K=112.5KWSPM.
2. 13 new EUV tools for 3nm could support 13/20*45K=29.3KWSPM.
3. 29.3k/112.5k= 26%
4. Above we assume there are 35 tools existed and all used to support 5nm. If there are some tools used in 7nm( 4 EUV layers), then the expansion % would be reduced to even lower value.
For your reference.
 
Last edited:
Using the proposed TSMC Phoenix Arizona fab as an example. I can understand and agree some of the reason behind it, such as the abundant tech workforce, good education system, and good weather, etc. But is there ever any thorough discussion on it? As a nation we are going to subside it but as a nation we should ask where and what is the best approach to build such important fab too.

Unfortunately, Arizona is an important swing state to both Republicans and Democrats. Any necessary and healthy discussion disappeared. Many similar kind of dramas have been repeating across US again and again. What kinds of wonderful result can we expect under this kind of practice?

I can understand that one. The semiconductor talent in America is mostly concentrated in California, but CA is both pretty darn short on greenfield land and extremely litigious. Propose some plant a couple hours outside the Bay that's going to be using a bunch of toxic chemicals, and it will get tied up in lawsuits and environmental impact studies for the next decade and probably never get built. Just how that state works.

AZ's about as close as you can get to CA while staying outside the borders of that regulatory nightmare, and they have plenty of desert to build on. That and west NV would be my top two picks if I was dictator of the world, and AZ's got more of an electrical engineering nucleus in place.
 
Back
Top