Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/local-tsmc-suppliers-divided-over-us-move.17171/
        )

    [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
)

Local TSMC suppliers divided over US move

benb

Well-known member

Local suppliers of Taiwan Semiconductor Manufacturing Co (TSMC, 台積電) appear to be divided over whether they should follow the chipmaker and also set up production facilities in the US, after TSMC announced that it would increase its investment in Arizona.

While some TSMC suppliers, including clean-room design service provider United Integrated Services Co (漢唐集成), have set up plants in the US, others, such as IC testing and analysis provider Materials Analysis Technology Inc (閎康科技), have hesitated to make the move because of high production costs in the US.

TSMC on Tuesday announced that it would increase its planned US$12 billion investment in Arizona to US$40 billion to build a 3-nanometer fab, scheduled to begin operations in 2026, near its 4-nanometer fab, which is scheduled to start production in 2024.

P12-221208-348.jpg

An engineer walks inside a laboratory of the Taiwan Semiconductor Research Institute in Hsinchu County on Sept. 16.​

Photo: Ritchie B. Tongo, EPA-EFE
Many TSMC suppliers had committed to following their biggest customer to the US well before the chipmaker announced its additional investment.

United Integrated Services said it has set up a subsidiary — United Integrated Services (USA) — to serve TSMC, to which it has provided integrated engineering system services since 1994.

The company said it provides services to TSMC facilities in the Hsinchu Science Park (新竹科學園區), the Central Taiwan Science Park (中部科學園區) and the Southern Taiwan Science Park (南部科學園區), and has been awarded as one of the chipmaker’s best suppliers.

Facility engineering turnkey project provider Marketech International Corp (帆宣) said it opened Marketech International Corp USA long before TSMC in May 2020 announced its plan to build an advanced semiconductor fab in Arizona.

It has doubled its team in the US to more than 100 specialists because of the TSMC move, Marketech said, adding that it is already profiting from the US expansion.

Its US subsidiary posted a profit of about NT$250 million (US$8.15 million) in the first nine months of this year, after securing orders from TSMC for work on its first Arizona fab, sharply up from a profit of NT$3.4 million for the whole of last year, Marketech said.

Mass flow controller and purifier supplier Taiwan Puritic Corp (和淞科技) said that although it only has paid-in capital of NT$600 million and sales of less than NT$10 billion a year, it still decided to follow TSMC to the US.

Taiwan Puritic in March last year set up a US subsidiary — Propersys Corp — after TSMC announced its Arizona investment plan.

Other suppliers have been more cautious.

Materials Analysis said the higher cost structure in the US is a major factor in its reluctance to set up a facility there.

It is unsure whether it would find workers with the needed skills in the US, it added.

Taiwan Speciality Chemicals Corp (台特化) has also been hesitant to invest in the US.

The company said it needed to see the economies of scale in the US market grow to a certain level before it might start production there, adding that it opened a warehouse in the US to handle and distribute its products made in Taiwan.
 
Opening a local warehouse would make sense, but planning for new production facilities doesn't seem to make sense, considering the amount of talk regarding 'over-expansions' and 'over-heated markets'.

Perhaps these companies are very confident of continuing high demand?
 
Marketech profit in 2022 in the US subsidiary in 9 months: US$8.15 million
Marketech profit in 2021: NT$3.4million (US$110,942)
For the first time, some numbers on the rumored repressive effect of TSMC on the Taiwan semiconductor supporting business industry. Huh, pretty big.
 
Hard to tell because of translation. But I would assume the non chemical companies are talking about having support engineers for TSMC. I can’t imagine they would need more than a dozen each. The fact that some can’t even commit to doing this is shocking. My opinion not having engineering offices in the US is a bit of a no go. Unfortunately changing your chemical suppliers or software firms is an even bigger no go. Even changing non advanced equipment like a different abatement unit, pump, or TCR can cause troubleshooting nightmares as well as thousands of manhours to qualify the new equipment.

As a side note it seems weird for a chemical company to complain that operating in the US is too hard. The US has orders of magnitude more chemists/chemical engineers than the ROC (many of whom have many years of experience). Yeah the environmental regulations are stricter and labor is more expensive. My counter point would be that chemical plants are even more manpower light than fabs, and the environmental costs can be offset by a much stronger/larger chemical supply chain than in the ROC (boosting capital efficiency and allowing for lower input chemical pricing). However the argument that the Arizona fab wouldn’t demand enough chems to justify building out grassroots capacity is a much more sound.

Just my 2cents as a green around the gills chemical engineer though.
 
Last edited:
Hard to tell because of translation. But I would assume the non chemical companies are talking about having support engineers for TSMC. I can’t imagine they would need more than a dozen each. The fact that some can’t even commit to doing this is shocking. My opinion not having engineering offices in the US is a bit of a no go. Unfortunately changing your chemical suppliers or software firms is an even bigger no go. Even changing non advanced equipment like a different abatement unit, pump, or TCR can cause troubleshooting nightmares as well as thousands of manhours to quality the new equipment.

As a side note it seems weird for a chemical company to complain that operating in the US is too hard. The US has orders of magnitude more chemists/chemical engineers than the ROC (many of whom have many years of experience). Yeah the environmental regulations are stricter and labor is more expensive. My counter point would be that chemical plants are even more manpower light than fabs, and the environmental costs can be offset by a much stronger/larger chemical supply chain than in the ROC (boosting capital efficiency and allowing for lower input chemical pricing). However the argument that the Arizona fab wouldn’t demand enough chems to justify building out grassroots capacity is a much more sound.

Just my 2cents as a green around the gills chemical engineer though.
This has been my thinking too. Taiwan for sure has the better ecosystem but people acting like the manpower issue can’t be addressed in the U.S sound crazy to me. The U.S has a population on the scale completely beyond Taiwan by many degrees. If there is anything the U.S has is tons of people, the best universities in the world and a technological pedigree to make any nation envious. The idea that a workforce can’t be trained in large enough numbers for this is far fetched to say the least.
 
This has been my thinking too. Taiwan for sure has the better ecosystem but people acting like the manpower issue can’t be addressed in the U.S sound crazy to me. The U.S has a population on the scale completely beyond Taiwan by many degrees. If there is anything the U.S has is tons of people, the best universities in the world and a technological pedigree to make any nation envious. The idea that a workforce can’t be trained in large enough numbers for this is far fetched to say the least.
My sense is that TSMC will tell their suppliers the support level they need, and the suppliers will just do what TSMC asks and deliver. I can't imagine any of them being recalcitrant.
 
There are many factors that can affect the supplier's decision. Cost, price, profit, revenue projection, commitment (both ways), scale of the economy, staffing, risk management, and business model. TSMC is well-known in managing their cost. No matter how big or how small a supplier is, they have to have a honest review of all those factors.

For example, Air Products (headquartered in Pennsylvania US) is the specialty gas supplier for the new TSMC's Kaohsiung Taiwan fab. But Linde Gas (headquartered in Ireland with Germany and American history) is the premiere speciality gas supplier for the TSMC Arizona fabs.
 
The big question is whether Chips Act will subsidise TSMC only, or those fellows in the supply chain as well.

From my impression, the further up you go along the semi supply chain, the thinner are the profit margins.
 
Why would they need subsidies? Their subsidy is fabs ordering more equipment than they would otherwise. I also assume that the US government doesn’t really care about getting people like TSMC’s mfc or chems suppliers to manufacture their products in the USA how many local alternatives there already are. However it seems like some vendors will get some money, (ie globalwafers or tool vendors indirectly getting some extra r/d spend).
 
Why would they need subsidies? Their subsidy is fabs ordering more equipment than they would otherwise. I also assume that the US government doesn’t really care about getting people like TSMC’s mfc or chems suppliers to manufacture their products in the USA how many local alternatives there already are. However it seems like some vendors will get some money, (ie globalwafers or tool vendors indirectly getting some extra r/d spend).
TSMC has to price in any increases passed on from their suppliers, or accept much lower margins.

If the markup gets too high then the fab is not viable, from what I understand at least.

For example,
A 50% markup for specialty chemicals in Arizona due to more stringent environmental standards would translate to an X% increases in final pricing. A 25% increase in labor costs would translate to a Y% increase. etc...
 
TSMC has to price in any increases passed on from their suppliers, or accept much lower margins.

If the markup gets too high then the fab is not viable, from what I understand at least.

For example,
A 50% markup for specialty chemicals in Arizona due to more stringent environmental standards would translate to an X% increases in final pricing. A 25% increase in labor costs would translate to a Y% increase. etc...
They were talking about the suppliers needing subsidies. We all know that TSMC is getting subsidies for the above reasons. But the suppliers just need to provide engineering support/a basic workshop. A canister of He or an mfc controller made in the ROC is just as useable as one made anywhere else.
 
Many parts are about 30% more expensive in the USA than Korea, Taiwan or China. Same part, same vendor, different pricing for different geography. So an Asian vendor locating in the US can expect to earn about 30% more right away, in terms of sales (and COGS). It remains to be seen if that hits the bottom line or not, but interestingly, looks like at Marketech it is.
On the other hand, chemicals coming from Japan have been cut 30% in the last 1-2 years as the JPY falls against USD, which makes the US much more cost effective against Asia in this one limited sense.
Environmental standards in the US are local, so they vary. It isn't as stringent as you would think in the US. Semiconductors remain under the radar as far as pollution goes.
Salaries are lower in Asia and the US salary doesn't include about $40/hour of healthcare cost. So the $10-20/hr difference balloons enormously due to the US treacherous healthcare "system".
So, there are meaningful, sizable differences between Asia and USA which if Asian vendors can navigate should spell a bright future. Most of all TSMC, but all of TSMCs vendors are in for a very sweet treat in the US.
 
Salaries are lower in Asia and the US salary doesn't include about $40/hour of healthcare cost. So the $10-20/hr difference balloons enormously due to the US treacherous healthcare "system".
How do you figure that healthcare benefits in the US cost $40/hour? Average healthcare benefits in the US for 2023 for family coverage cost about $22,500/year. (This means a PPO plan.) Dividing by only a 40 hour workweek (I'm assuming all TSMC employees will be salaried, so the number of hours they work is irrelevant) and 52 weeks per year, equating to 2080 paid hours per year, that's about $11/hour.


Edit: additionally. most US companies make the employee cover part of the premium cost. Not all, but most.
 
How do you figure that healthcare benefits in the US cost $40/hour? Average healthcare benefits in the US for 2023 for family coverage cost about $22,500/year. (This means a PPO plan.) Dividing by only a 40 hour workweek (I'm assuming all TSMC employees will be salaried, so the number of hours they work is irrelevant) and 52 weeks per year, equating to 2080 paid hours per year, that's about $11/hour.


Edit: additionally. most US companies make the employee cover part of the premium cost. Not all, but most.
Techs work hourly.
 
That’s at Intel. I wonder if TSMC will have a different work schedule, or salaried techs.
I can confirm these hours are the case at Samsung Austin, Micron, and ASML. If memory serves TSMC Arizona was also in this camp. TSMC ROC was more vague on their hours (12 hour shifts, but no day amount was given).
 
Last edited:
Clarification: $37/hour worked, for all benefits including healthcare

 
Back
Top