Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/chipmaker-tsmc-in-talks-with-suppliers-over-first-european-plant.17254/
        )

    [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
)

Chipmaker TSMC in talks with suppliers over first European plant

Daniel Nenni

Admin
Staff member
TSMC HQ Taiwan.jpg


Company to send senior executives to Dresden early next year to discuss potential factory project in Germany

TSMC is in advanced talks with key suppliers about setting up its first potential European plant in the German city of Dresden, a move that would allow the world’s largest chipmaker to capitalise on booming demand from the region’s car industry. The Taiwanese company is sending a team of senior executives to Germany early next year to discuss the level of government support for the prospective plant as well as the capacity of the local supply chain to meet its needs, according to people familiar with the matter. The trip will be the second in six months by TSMC executives and a final decision on whether to invest billions of dollars in a plant, which could begin construction as early as 2024, is expected to follow soon after, the people said. Last year TSMC was asked by customers to consider building a plant in Europe, but halted an initial review following the invasion of Ukraine.

But growing demand from Europe’s carmakers for a locally-manufactured supply of chips has prompted TSMC to revisit the idea, the people said. A decision to build the plant would be a major boost for the EU, which is racing to cut its reliance on importing semiconductors — vital components in everything from smartphones to cars — from Asia. Brussels earlier this year approved €43bn in subsidies in a bid to attract chipmakers to Europe. TSMC’s talks with several materials and equipment suppliers are focused on whether they can also make the investments required to support the plant, people familiar with the matter said. Manufacturing chips is a complex process relying on more than 50 types of equipment, such as lithography and etching machines, and over 2,000 materials including chemicals and industrial gases.

“We would try to support our customers. We wouldn’t let [them] walk alone in the desert,” said one executive from a supplier that would provide key materials to the Dresden plant, adding that state support would be required. Surging energy costs and higher inflation have already prompted US chip group Intel to seek more support from the German government for its plan to build a €17bn for plant in the eastern city of Magdeburg.

Intel is still committed to investing in Europe but the Magdeburg plant had to be competitive, according to people familiar with the matter. If TSMC presses ahead with a Dresden plant, it would focus on 22-nanometre and 28-nanometre chip technologies, similar to those it plans to make in a factory it is developing with Sony in Japan. Nanometres refer to the size of each transistor on a chip — the smaller the nanometre, the more powerful and advanced the semiconductor.

TSMC will have to weigh up whether building a plant in Dresden will put too much of a strain on its workforce. The chipmaker is already sending several hundred engineers to support new plants it is building in the US and has said it would need to deploy 500 to 600 more to help set up the factory in Japan. Europe, the Middle East and Africa account for roughly 6 per cent of TSMC sales, a fraction of the 65 per cent the group generates from North America.

A TSMC spokesperson said that “no possibility” was being ruled out regarding a potential plant in Dresden. TSMC’s overseas expansion comes as global chipmakers such as Intel and Samsung race to expand capacity. The world’s three biggest chipmakers are committed to investing at least $380bn over the next decade to build new factories in Taiwan, South Korea, the US, Japan, Germany, Ireland and Israel.

In the US, The Chips Act, which was proposed in 2020 and passed by Congress last year, has triggered $200bn of private investment in the country’s chipmaking capacity, according to the Semiconductor Industry Association. The speed of the global expansion has raised questions about the risk of the industry facing a glut of chips if global economic growth slows sharply. But with the global semiconductor market forecast to reach $1tn in value by 2030, chipmakers must decide now on how they will meet that expected demand given that it takes years to build plants.

 
This seems to be a case of "Anything you can do I can do better" between TSMC and Intel and clearly TSMC is winning. The TSMC event in AZ was the best foundry press event I have seen in my entire career, absolutely.

I have repeatedly warned my friends at Intel to tread lightly but they have not.

"I fear all we have done is to awaken a sleeping giant and fill him with a terrible resolve." Japanese Admiral Isoroku Yamamoto regarding the 1941 attack on Pearl Harbor by forces of Imperial Japan.

 
Is European market too small for the TSMC to build a fab there?
I don't think it has much to do with local market size, because, as we all know, semiconductors are easily and relatively cheaply shipped (on a ship-cost / revenue basis). As expensive as building and staffing in the US is, Europe appears appears to be even more challenging. Tesla's challenges in Germany have been very telling.


The permitting process for the plant's construction in Germany was apparently even worse than the US, as difficult as that is to imagine.



I'm amazed at the quick progress TSMC has made in Arizona so far, considering US permitting processes. Perhaps TSMC is too.
 
The EU is regulation and bureaucracy hell and Germany is the the tenth level of it from what I've heard. I wonder what sweeteners and incentives they are going to have to give to make this effort at all worth it to TSMC. Europe and Germany can move fast when they want to, I.E the rapid construction and commissioning of their new LNG terminals which was done at lightspeed.
 
Good point, but the LNG terminal was a national emergency. A fab isn't that.
I would agree, but maybe the EU feels differently. They were very speady in funding their version of CHIPS and if memory serves Italy is begging on their hands and knees for an intel packaging facility. Rumor has it that they were willing to pay up to 40%. Who knows anymore though with the new government in Italy and the further deterioration of Intel’s financials. Back to the matter at hand if Intel is getting that kind of treatment I don’t see why TSMC might not get treatment on a similar magnitude (probably a smaller scale given the relatively small size of the foreign investments that TSMC is making).
 
I don't think it has much to do with local market size, because, as we all know, semiconductors are easily and relatively cheaply shipped (on a ship-cost / revenue basis). As expensive as building and staffing in the US is, Europe appears appears to be even more challenging. Tesla's challenges in Germany have been very telling.


The permitting process for the plant's construction in Germany was apparently even worse than the US, as difficult as that is to imagine.



I'm amazed at the quick progress TSMC has made in Arizona so far, considering US permitting processes. Perhaps TSMC is too.
"I don't think it has much to do with local market size, because, as we all know, semiconductors are easily and relatively cheaply shipped (on a ship-cost / revenue basis). "


Exactly that's the problem. If shipping is cheap, TSMC's European customers can order the chips from Japan or US if some of them feel it's better to avoid Taiwan.

To begin with the European fab cost and operation cost are higher and European buyers don't have too much leverage due to their smaller volume.
 
"I don't think it has much to do with local market size, because, as we all know, semiconductors are easily and relatively cheaply shipped (on a ship-cost / revenue basis). "


Exactly that's the problem. If shipping is cheap, TSMC's European customers can order the chips from Japan or US if some of them feel it's better to avoid Taiwan.

To begin with the European fab cost and operation cost are higher and European buyers don't have too much leverage due to their smaller volume.
In a post a while back Dan quoted AMD's founder Jerry Sanders' famous statement that "Real men have fabs." I responded that now the feeling is that "Real countries have fabs". I think that's a lot of what's going on here. It also seems especially true since competing with Taiwan or Korean gigafabs on cost looks like a mirage, due to the oft-explained cost advantage of far higher wafer throughput. I'm not smart enough about the fabrication industry to predict how this fab frenzy will turn out, but it doesn't smell like long-term success to me.
 
In a post a while back Dan quoted AMD's founder Jerry Sanders' famous statement that "Real men have fabs." I responded that now the feeling is that "Real countries have fabs". I think that's a lot of what's going on here. It also seems especially true since competing with Taiwan or Korean gigafabs on cost looks like a mirage, due to the oft-explained cost advantage of far higher wafer throughput. I'm not smart enough about the fabrication industry to predict how this fab frenzy will turn out, but it doesn't smell like long-term success to me.
The local taxpayer gonna be on the hook for plenty thats for sure!
 
In a post a while back Dan quoted AMD's founder Jerry Sanders' famous statement that "Real men have fabs." I responded that now the feeling is that "Real countries have fabs". I think that's a lot of what's going on here. It also seems especially true since competing with Taiwan or Korean gigafabs on cost looks like a mirage, due to the oft-explained cost advantage of far higher wafer throughput. I'm not smart enough about the fabrication industry to predict how this fab frenzy will turn out, but it doesn't smell like long-term success to me.

Onshoring seems like a fad that will fade in time but I can assure you it is part of TSMC's master plan. TSMC gives a master class in collaboration where ever they go and the result is a fast growing ecosystem that is the base of TSMC's strength. Politics is now a part of the semiconductor industry and I can assure you TSMC will master that as well. The AZ event is a prime example. There were some very rich people at that event and those people get politicians elected:

Yesterday, it was our privilege and honor to welcome and host U.S. President Biden and Commerce Secretary Raimondo at #TSMC #Arizona along with many of our valued customers, partners and supporters. During this event, we celebrated the arrival of first batch state-of-art #semiconductor #innovation semiconductor tools which are designed to support production of leading-edge semiconductor process technology.

In addition to TSMC Arizona’s first fab, which will now manufacture our N4 process technology in 2024, we announced plans for a second fab which will start production of 3nm process technology in 2026. TSMC’s investment is $40 billion, representing the largest foreign direct investment in Arizona history, and one of the largest in the US. These two fabs are expected to create an additional 10,000 high-paying high-tech jobs, including 4,500 direct TSMC jobs. When complete, TSMC Arizona will produce the most advanced semiconductor process technology in the country.

“When complete, TSMC Arizona aims be the greenest semiconductor manufacturing facility in the United States producing the most advanced semiconductor process technology in the country, enabling next generation high-performance and low-power computing products for years to come,” said TSMC Chairman Dr. Mark Liu. “We are thankful for the continual collaboration that has brought us here and are pleased to work with our partners in the United States to serve as a base for semiconductor innovation.”

“I’m deeply honored that we have been able to gather here to celebrate such a historic milestone for TSMC, which represents our dedication to building a semiconductor ecosystem here in the USA in order to better serve our customers,” said TSMC CEO Dr. CC Wei during the ceremony. “This project is truly a testament to the Grand Alliance TSMC has formed with customers, suppliers and partners to further our collaboration and unleash innovation.”
 
Onshoring seems like a fad that will fade in time but I can assure you it is part of TSMC's master plan. TSMC gives a master class in collaboration where ever they go and the result is a fast growing ecosystem that is the base of TSMC's strength. Politics is now a part of the semiconductor industry and I can assure you TSMC will master that as well. The AZ event is a prime example.
FWIW, I do think TSMC's AZ facilities have the best chance of long-term financial success. I suspect Apple will be a big part of enabling that success. Samsung's TX facilities come in second place in my mind, but they seem like a distant second place. The others in the US and Europe, especially Intel's build-it-and-they-will-come strategy, look risky. And Intel's internal demand looks risky too. I hope Micron is planning a lot of automation to make memory fabs work financially at scale in the US.

(As an aside, I'm finding it fascinating how many of my friends who were Android phone users have switched to iPhones lately. Including some who were the biggest complainers about Apple's ecosystem of services. When I get texts from them, their text bubbles are now blue, and it provides a very interesting indicator of the changes of who's doing what with Apple and Android. I have not noticed even one going from blue to green. Of course this sort of anecdotal sample is not a valid indicator of reality, but in my boredom it does get a bit of my attention. The Apple/TSMC partnership would make an engrossing business school case study.)
 
FWIW, I do think TSMC's AZ facilities have the best chance of long-term financial success. I suspect Apple will be a big part of enabling that success. Samsung's TX facilities come in second place in my mind, but they seem like a distant second place. The others in the US and Europe, especially Intel's build-it-and-they-will-come strategy, look risky. And Intel's internal demand looks risky too. I hope Micron is planning a lot of automation to make memory fabs work financially at scale in the US.

(As an aside, I'm finding it fascinating how many of my friends who were Android phone users have switched to iPhones lately. Including some who were the biggest complainers about Apple's ecosystem of services. When I get texts from them, their text bubbles are now blue, and it provides a very interesting indicator of the changes of who's doing what with Apple and Android. I have not noticed even one going from blue to green. Of course this sort of anecdotal sample is not a valid indicator of reality, but in my boredom it does get a bit of my attention. The Apple/TSMC partnership would make an engrossing business school case study.)
I would feel more secure with the Samsung Texas facility. They have a stronger presence in the US and China than TSMC. For this reason my gut says they should be more apt at distributed production. Part of TSMC’s business model is built on exploiting the synergies associated with concentrated production. This is not to say that the Az facility will fail. Rather it will likely be less profitable than TSMC’s other ventures (the smaller size doesn’t help either). Who knows though if the AZ sight grows big enough and a sizable EU sight is made, then maybe TSMC can have the kind of mastery over distributed manufacturing networks that intel and micron have.

As for Micron automating their plants, there isn’t much more to be done. 300mm fabs are about as barebones as you can get. Tool maintenance folks and engineers to run/improve the show. I’m sure there are further efficiencies to be gained (there always are), but all of the obvious things have already been automated.
 
I would feel more secure with the Samsung Texas facility. They have a stronger presence in the US and China than TSMC. For this reason my gut says they should be more apt at distributed production. Part of TSMC’s business model is built on exploiting the synergies associated with concentrated production. This is not to say that the Az facility will fail. Rather it will likely be less profitable than TSMC’s other ventures (the smaller size doesn’t help either). Who knows though if the AZ sight grows big enough and a sizable EU sight is made, then maybe TSMC can have the kind of mastery over distributed manufacturing networks that intel and micron have.
Interesting perspective. Perhaps I'm letting subjectivity get in the way of the evidence, but Samsung's US progress hasn't impressed me much.
As for Micron automating their plants, there isn’t much more to be done. 300mm fabs are about as barebones as you can get. Tool maintenance folks and engineers to run/improve the show. I’m sure there are further efficiencies to be gained (there always are), but all of the obvious things have already been automated.
You know more about fab operations than I do, but Micron is projecting 2000 employees for their new Idaho fab. That seems like a lot of people.

 
Interesting perspective. Perhaps I'm letting subjectivity get in the way of the evidence, but Samsung's US progress hasn't impressed me much.
What about it is underwhelming to you?
You know more about fab operations than I do, but Micron is projecting 2000 employees for their new Idaho fab. That seems like a lot of people.

AZ is 3000 people for one fab and will likely be lower wafer starts than Boise given the nature of memory fabs. If anything that says Micron is more efficient than TSMC on a person per wafer basis. Idaho is also a R&D site, so the number of engineers per tool should be much higher.
 
AZ is 3000 people for one fab and will likely be lower wafer starts than Boise given the nature of memory fabs. If anything that says Micron is more efficient than TSMC on a person per wafer basis. Idaho is also a R&D site, so the number of engineers per tool should be much higher.
Interesting that TSMC has more people, but I'm not a judge of how comparable the numbers are. Regardless, 2000 people is a lot for direct employees, and indicates to my admittedly not-an-expert perspective that there are probably more automation opportunities to be had.
 
Interesting that TSMC has more people, but I'm not a judge of how comparable the numbers are. Regardless, 2000 people is a lot for direct employees, and indicates to my admittedly not-an-expert perspective that there are probably more automation opportunities to be had.
Like I said there always are. Automating different spc and tool response systems require less engineers once these systems are properly setup. Better process engineers can setup and stage PMs to have fewer maintenance work spikes allowing for smaller technician pools, and the list goes on. But none of these projects can offer the same level of improvement you’d see with increasing wafer sizes, AMHS systems, and higher throughput/pm life tools.
 
Back
Top