Read it again. That includes the employee's salary, in your link.Clarification: $37/hour worked, for all benefits including healthcare
The Average Cost of Employee Benefits in the US
The average cost of employee benefits varies based on location, industry, company size and type of worker. However, most employers spend about $21,726 per employee, per year, on benefits, bringing the average cost per employee to $71,334 in the United States.work.chron.com
Local TSMC suppliers divided over US move
- Thread starter benb
- Start date
Amazing. Compared to the hours I worked as an engineer and various levels of management, that would be like a part-time job.
Regular 12 hour shifts, that I believe.
ashwinbrao1
New member
The semiconductor chemicals market is a very tricky one.
Most chemical companies have a hard time dealing with the risks, investments and service levels associated with serving semicon customers.
Chemical plant investments are largely volume driven and with a few exceptions, semiconductor chemicals are a small volume market. Yes, there are plenty of chemists and chemical engineers in the US but a very small fraction of them have worked in the semiconductor chemicals market and this workforce is moving towards retirement.
For the next few years, I suspect that the chemical suppliers will be scrambling to fill their positions
Excellent points. On the point of retirement, I can't imagine that the ROC's chemical engineer population is all spring chickens either though. And I would have more trust in those positions being filled faster in the US than in Taiwan. As for companies not wanting to get into this sector that argument doesn't really work for the incumbents. TSMC's chemical suppliers have already chosen to play in this space, as well as the already strong ecosystem of suppliers for the large number of American semiconductor firms. Obviously US fab capacity (especially for foundry, 300mm, and memory) isn't as high as in the ROC PRC or ROK, but I can't imagine it is that far behind. The chems industry gets a further boost with all of those 200mm fabs that need far more chems per die than a 300mm wafer fab would.
Mooredaddy
Well-known member
Anecdotal but I’ve heard a lot of Taiwan based suppliers have either completed or are in the process of working through incentive programs with the Arizona commerce board. TSMC is just such a 800 pound gorilla in the room that you really do want to follow them around. I’m even seeing news reports from the Phoenix area of Taiwanese bakery’s, banks and service industries setting up shop to much fanfare. Apparently these places are chock full of expats and locals alike. Truly a great thing to see!! Exciting times
M. Y. Zuo
Active member
So does that mean local TSMC suppliers are no longer divided about the move?
Or is it just the pro-move faction that are going all in?
Mooredaddy
Well-known member
I think the “divided” narrative is likely overblown in the first place. I’ll be in Phoenix in the new year but my impression so far through the grapevine is everyone is excited about the opportunities from being set up stateside. The CHIPs ACT provided a lot of powerful incentives all down the supply/support chain. See Global wafers setting up in Texas etc.
M. Y. Zuo
Active member
Hmm it is possible the Taipei Times exaggerated the division.
It also would make sense that these things will be planned well ahead of time and that TSMC has enough competent folks in their procurement team to prevent undesirable scenarios from developing.
Why would they intentionally be using the same suppliers Intel uses? TSMC probably wants to keep the same suppliers they have back home. You are asking for trouble if one fab is using different suppliers from everyone else. Even if they charge you an extra 20% for a abatement unit as a tax for setting up an engineering office in AZ, it is still a no brainier to use them. This gives you one less thing that could potentially drag yields down and make troubleshooting with other ROC fabs more difficult since their setups wouldn't be the same. You would also need to spend many hours doing tests and writing internal whitepapers to confirm that "Yes: this part is compatible with out process.". TSMC and Intel might have their tools configured differently/have different upgrades. These differences might necessitate different components in the subfab (note many of the items in the subfab do not come from the tool vendor). If you don't think this is a concern then I would love to hear why.
Last edited:
There are only so many suppliers, should be the same coincidentally.
I don’t know. TSMC has a little bit more than 50% of foundry. By dollar amount US semis are like 15 or 20% (minus memory) of the global value. The US is making much lower asp wafers than TSMC’s avg spread (more wafers per dollar of the semi pie). The US also has more memory fabs than the ROC.
On the 200mm side, something that was interesting to me was how many wafers they can make per month. Skywater has a tiny cleanroom, but can spit out wafer starts at a faster than expected clip (probably due to simpler process flows).
Leading edge fabs are driving chemical use up. There was an inflection point with finfets.
Its also a specialty chemical story; semiconductor grade is 10-100x the base chemical price, with a lot of value-added that makes it worthwhile for businesses to pursue this. Specialty chemicals are differentiated, patent protected, and indispensible to the process. Much of the difference between Samsung and Intel and TSMC is not the equipment but the process, and much of that is the chemicals.
Its also a specialty chemical story; semiconductor grade is 10-100x the base chemical price, with a lot of value-added that makes it worthwhile for businesses to pursue this. Specialty chemicals are differentiated, patent protected, and indispensible to the process. Much of the difference between Samsung and Intel and TSMC is not the equipment but the process, and much of that is the chemicals.