TSMC delays U.S. chip plant start to 2025 due to labor shortages | Hacker News Reader

TSMC delays U.S. chip plant start to 2025 due to labor shortages

asia.nikkei.com

110 pointsironyman155 comments

I think the push to fab SCs in the US is going to quickly run up against the fact that consumer technology pricing is incompatible with our labor standards. We've spent the last 50+ years exploiting global wage inequality to flood the country with cheap consumer goods, and we've reached a point where our sense of self-satisfaction is entirely fused to the acquisition of these cheap goods. Along the way we've ceded a lot of ground in the way of healthy communities and a rich public sphere. National pride is only going to make up for so much when it causes the iPhone 16 to jump $500 and people are still stuck in their isolating suburban hellscape. IMO the only way for this to work in the long-term is a national reset about what "quality of life" really means, and I don't think our political class is willing to engage with all of the questions that will entail.

edit: "American Factory" is a really interesting documentary that covers something like this dynamic in the context of an auto-glass manufacturing facility. US workers fought hard for their labor rights and capital responded by fleeing the country to more permissive political regimes. We accepted this loss of jobs because it kept the all-import car prices down. Now that other countries are starting to eat our lunch in the global economy we're suddenly not so stoked about globalization, oh no the consequences of my own actions!!

> "American Factory" is a really interesting documentary that covers something like this dynamic in the context of an auto-glass manufacturing facility.

probably fab employee adds more value, and can be paid much higher and work with higher labor standards.

If the product is being sold for less than its true value then the person making that product will be underpaid and/or overworked. Sure someone choosing between "glass factory" and "semiconductor fab" may choose the latter if they can, but that doesn't mean that they will get standards and pay commensurate with other jobs in the US requiring similar training or effort. Taiwanese labor standards are way lower than in the US, and the price-point of our consumer electronic goods requires that. Either the prices go up (and people get mad), the labor standards come down (and people get mad), or the tech companies accept lower profit margins (and capital owners get mad, and they tend to get their way). There's no magic fix to this trilemma, historically we've just moved the bad labor standards out of sight so US voters can get their labor standards and Tim Cook can get his ballooning valuation. One of these two things is going to have to give if we want to make chips on-shore.

> Taiwanese labor standards are way lower than in the US, and the price-point of our consumer electronic goods requires that.

not necessary. tsmc has 40% profit margin, so they could totally pay more to taiwanese employees, but decided not to and pocket profit instead.

Big-time agreement with you there

Where is the exhaustive list of jobs they are trying to fill with open reqs?

There is a very long list if you search linked in.

Personally, would you:

* move to Pheonix

* do shift work

* complete a special local college course at your own expense

* to work for a company with notoriously demanding standards for staff

and there are no salaries listed as far as I see...

What kind of garbage output can we expect from such labor abuse? Even with testing and all those processes this has to increase costs?

Looking forward to my nvidea GPU failing a week into use...

That’s what they currently do in their Tainan, Taiwan factory. I thought they were known for good quality chips?

Yes, it's not an issue with the management style, but the high expectation of working conditions of the American worker.

While the whole world is busting their ass, they're hooked on 9-5s.

Tesla is a great example of a manufacturer that pays top dollar and gets back top-quality work.

> top-quality work

Why are they rated so poorly for initial quality?

They started a car company from scratch, it's not an easy feat.

The build quality isn't at par with the rest of the manufacturers, but the drivetrain and batteries are by far the best.

(disclaimer: I own a Ford Mach-e and tried all the Teslas)

So they're a battery company that makes average cars

"high expectation of working conditions"

Working conditions in the Unites States are quite bad compared to some other places in the world. "At will" employment, wages so low that tips are required to offset them etc. I can go on and on. Just take a read in /r/antiwork/ at reddit. A lot of those things are even illegal but people don't even bother anymore to report them.

We are at a point in time where we have the luxury to improve working conditions and not go back to basically slavery. The issue is we are enriching the rich instead of those that are doing the hard labor.

We have the whole thing backwards, actual labor should be rewarded the most as it takes that one thing away from us that we have a limit resource of which is time.

If we don't strive to improve our lives then what is the point? Work and die to make the life for a few excessively luxurious?

Yes, and we should encourage people to go in this direction instead of becoming influencers, Youtubers, and models.

The West needs to start building again or it will become an extinct empire.

The article says they’re looking for people who can install their equipment:

“We are encountering certain challenges, as there is an insufficient amount of skilled workers with the specialized expertise required for equipment installation in a semiconductor-grade facility,"

As far as I can tell, there's something like 52 positions they are trying to fill. https://ro.careers.tsmc.com/search/?createNewAlert=false&q=&...

I suspect instead a wage shortage, not a lack of people to hire.

I'm not so sure - this is incredibly technical work so it doesn't ring as hollow as it does when McDonald's or whatever claim the same.

A lot of the work is actually just working with dangerous chemicals, and isn't a technical desk job with prestige.

It's factory work!

This - chip factory work is actually a factory job, with a obssesive compulsive quality control part. One stop, one power drop and the yield goes nowhere..

Yeah, my brother worked on stuff like this while working in a Physics lab, he got transferred to doing the software dev parts in the end as working with the acid was very stressful.

My family member works at TSMC. Last I talked to him (8mo ago), American engineers quit left and right due to working conditions and pay.

Can you expand a little bit on the issues with working conditions/pay at all? It would be good to have something written down publicly, if employees are being mistreated or taken advantage of.

It depends on what's available. If you grew up in abject poverty on a rural farm then getting a job in doors, in air conditioning, without requiring brutal, repetitive physical labor is like heaven - requiring you to work 15 hour days and 7 days a week is a small price to pay. If you grew up in a wealthy suburb, then the baseline is 40-hour weeks in the same conditions, and you have aspirations to either a) work less for the same money or b) make a lot more money. Anything else is "abusive" to you. There's something like labor arbitrage going on here...

Those people are gone from the labor pool. They were a momentary artifact of class mobility that was present up to the early 90s but is no longer a thing

I totally feel that! :)

One of the things that I found great about working with computers was that it got me away from being a few inches away from a fryer for 8-16 hours a day. Not having to deal with screaming parents, sobbing children, and people literally (yes, literally) throwing hot food at me that "had too much (whatever) on it". Instead, getting to sit in a calm quiet environment for literally something like 5x the pay, for barely 3 to 4 hours of work, was a massive improvement. Over time, I managed to get into a role that has even less customer interaction, for higher pay, and less stress. It's also in something I actually care about, too! It's really great how freeing things can be, once you start to get specialized into specific areas.

Anyway, I was kind of tripped out on how the commenter had phrased their initial comment, "people quitting left and right". I had taken to mean something like "...because of a recent change", and was wondering what it was.

The commenter made a follow up and it sounds closer to something like, "People would join and work for TSMC for a short time, then get disgusted and quit." I had assumed that there was a long-term American work force that had been floating around, but left because of something that wasn't covered in the article, and I wanted to know what that was. It sounds like that's not really what they meant.

Just to add more, TSMC hires cohorts of US college grads to move to Taiwan for training for their first (2?) years. Then they are sent to Arizona for another 2 years of work.

There are monetary incentives to complete the 4 year contract, but many (50%?) terminate their contract early while they are in Taiwan or before transferring to Arizona.

These American workers are hired to work in Taiwan for training and then sent to Arizona to setup the factory. By local Taiwan standards, they are treated well (relatively high pay and don’t have to work on weekends).

But by US standards, working 12 hours per day 5 days per week with no overtime or free coffee isn’t competitive in our job market.

I don’t think anyone is mistreated or taken advantage of. But TSMC doesn’t provide a competitive work environment if you have access to the US job market.

Idk my math says that working 60 hrs should at minimum provide a 150% premium over working 40s. If you want 60 hrs and pay me for 40 or 32 I think I’m getting taken advantage of. What else do you call that?

Each additional hour of work has a greater opportunity cost than the previous hour of work.

I agree it’s a floor but a ceiling

From what he described, if you take a (local's salary + overtime) * 1.x, you would get his non-overtime US-salary pay. US workers also get more paid holidays than locals (doubly so depending on the month you join).

US workers making x% more than locals for the same job seems like the locals are taken advantage of, not the US workers.

It's weird because it sounds like what you're describing isn't purposefully malicious, but more so an issue cultural perspective. One would think that a major company like TSMC would have hired some kind of cultural liaison to assess what benefits, pay, etc. that Americans would be willing to accept before attempting to open a plant in America.

There's clearly a need TSMC has for highly skilled, niche work. Why bother posting the job if they aren't going to cover the associated costs?

I mean, TSMC is welcome to put out as many job adverts for as many positions as they like, offering the legal mandatory minimums, whether it be:

* Minimum wage

* No benefits

* No overtime

* No holidays

* No sick days

* No vacation days

* No workman's comp

* Swing shift

* 7-day a week schedule

I just can't fathom how TSMC would feign surprise when an already limited labor pool isn't tripping over themselves to apply to degrading work conditions. The culture of the higher ups might not call those things "taking advantage", but, if TSMC doesn't have perspective of the people they're trying to win over it is just a waste of everyone's time.

I find the whole thing to be queer because, more often than not, it's America that is being accused of poor workers rights and work conditions in comparison to European worker regulations. It's odd that, even when the roles are switched with some other nation, it's still somehow on Americans to capitulate. It'd be funny if it weren't incredibly tedious. Not enough benefits? America hates workers. Too many benefits? America hates working.

> One would think that a major company like TSMC would have hired some kind of cultural liaison to assess what benefits, pay, etc. that Americans would be willing to accept before attempting to open a plant in America.

They don't want to open a plant in the US, but are doing so because Apple, the US Gov, and maybe the threat of the CCP, are pushing them to.

There is also a reason much of the electronics manufacturing went to Asia, and long hours + low cost is why. Trying to reconcile that with US or EU working conditions may not be possible without entirely retooling their approach and management -- which is challenging to do even in a company that very much wants the change.

> They don't want to open a plant in the US, but are doing so because Apple, the US Gov, and maybe the threat of the CCP, are pushing them to.

Fair point. I guess I would just imagine that, if they are going to build the factory, whether it be by choice, or by burden, they would make a point of doing an adequate job at the feat. If the building is going up because of pressures from outside the company, it would seem to follow that they would want the money spent on that to be worthwhile, right? Or is the assumption that if they drag their feet long enough due to "no fault of their own" they can import people to do the work at a faction of the cost, under deplorable conditions?

Personally, I cringe at the idea of the idea of terrible working conditions essentially being imported into America, but I certainly wouldn't put it past a country trying to make a buck.

> Trying to reconcile that with US or EU working conditions may not be possible without entirely retooling their approach and management -- which is challenging to do even in a company that very much wants the change.

Very true! One of the hopes that I had had with this process was that, perhaps because TSMC would be under US working conditions that, perhaps new processes could be made/discovered, meeting both good conditions and whatever expectations they had in mind. Perhaps that was a bit (or very) naive on my part; I still want to believe that a better, more humane process can be found to meet the bottom line and make the work less grueling for the people on the factory floor.

I had a bunch of friends working in the fab industry in the US at Micron. They said TSMC's offer to poach them was 50k, no insurance, no 401k, no stock, 1 week PTO with no increases (that includes sick days). It was an absolutely insulting offer.

My friends all retrained into IT/software and jumped ship out of the industry. They all make 1.5-2.5x what they used to and no longer risk exposure to deadly chemicals every day.

> They said TSMC's offer to poach them was 50k, no insurance, no 401k, no stock, 1 week PTO with no increases (that includes sick days). It was an absolutely insulting offer.

I made more money bartending after college in the DC suburbs.

> American engineers quit left and right due to working conditions and pay

Which TSMC location has American engineers working at it?

TSMC hires US college grads and flies them to Tainan for training for a couple years. Then they go back to Arizona to setup the factory.

As any economist would tell you, "Labor Shortage" is a myth in any industry.

Not in new or rapidly expanding fields that require expertise.

Hardware (And even embedded programming) pays absolute ass since it's all outsourced to Asia. Zero reason not to do literally any other job which will use the same skills, and pay double. Bringing it to America, means upping the salaries to match what Americans expect, and alot of jobs just lowball still.

I agree. Labor shortage is mostly just lowball pay most the time. List the work and salary range and then let's see if it's a real shortage

It is also a shortage of workers willing to take slave driving conditions.

For something as exotic as chip manufacture, you genuinely can have a labor shortage. We are idiots not to immediately green light any visas for people with these skills.

It can also happen for skilled craft professions. It takes time to train people in that stuff, and if there's been a prolonged period of reduced investment in domestic manufacturing we might genuinely not have enough. Sometimes not having key people like welders, electricians, skilled plumbers, etc. can hold up a project since other things depend on those things.

You could argue that it still ultimately is a wage shortage, but even if the wages are high it takes time to train.

It will take a while for the US to rebuild domestic manufacturing capacity and talent base, but we can do it.

>but even if the wages are high it takes time to train

Exactly. There's also a brain surgeon shortage even though salaries are great, but not everyone has the will, skills, time, money and patience to study and become a credentialed brain surgeon despite the high pay.

Same with semi jobs, you need a hard science background that not everyone has or is willing to go through, including SW devs. Increasing wages might not fix the shortage because not everyone who can do other well paid jobs, can also do semi work.

>Sometimes not having key people like welders, electricians

I would add to this list the specialized construction workers and engineers who can build the buildings required to house a fab. This is similarly specialized, I imagine, as things like hospital construction. You can't just pull 100 crews used to building homes and say "Hey build this fab mkay?"

>We are idiots not to immediately green light any visas for people with these skills.

Are we, though? Isn't there some potential for industrial espionage? You know, the kind where the people that desperately want that tech for themselves are kind of the enemy of the main people that have that tech?

Yes, although paying people arbitrarily large amounts makes most shortages go away.

That said, there are a ton of things with a large labor component--and probably most things have a large labor component if you look far enough up the chain--that myself and others will just find a way of doing without if the price increases by 2x/3x/4x/10x.

> For something as exotic as chip manufacture, you genuinely can have a labor shortage. We are idiots not to immediately green light any visas for people with these skills.

All evidence suggests these roles pay between 50-70k USD, require insane hours, and a short course at the local community college is all it takes to be qualified (or qualified enough) to do the job.

For high-end computer-engineering grads? absolutely. For the fork-and-spoon operators in sector 7-G who just press the buttons? nah.

They ain't trying to recruit PhDs or brain surgeons, and may be trying to drag their feet as they do so.

Most definitely.

I suspect this is just the start of such stories being an extremely common theme for almost all industry other than white collar prestige careers moving forward.

The demographics are pretty much written in stone at this point for the foreseeable future.

The world is running out of children:

https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_pr...

And soon, the world will run out of people:

https://twitter.com/dylanmatt/status/1676590258857603076

Run out of people... for what? For buying Tesla model 3s? For logging the Amazon?

Pffffff the Amazon will be long gone by then.

they'll be fighting over arable land in now-tropical Upstate New York

what evidence do you have to suspect that? any info on current wages in that field vs what TSMC were offering etc?

Labor is a market, which means that the only two possibilities are 1) the offered pay is too low, 2) literally nobody in the entire country exists with these skills (in which case, the company needs to pony up the money to train them).

They have to be able to sell chips at a competitive price. In the extreme/absurd case they could pay each employee $1B/year and I’m sure they could find qualified people but they would immediately go out of business.

So thus, the right salary to hire the people they need is somewhere between what they're currently paying, and $1B/year. My guess—and this is only a guess, mind you—is that they'd still be able to staff up on something significantly lower than $1B. But I'm not an economist, so I might be wrong there.

TSMC has a long history of increasing double digit profit margins, in the range of $10B+ per year, and is amongst the most profitable businesses in the world. I would err on the side of TSMC deciding not to pay sufficiently to incentivize labor (which may still be the optimal move for TSMC).

https://www.macrotrends.net/stocks/charts/TSM/taiwan-semicon...

Or 3) we are talking about real markets instead of spherical frictionless cows. The market will eventually balance itself, but it doesn't happen immediately.

Consider yourself (and your family, if you have any). Assuming that the pay is right, how long would it take for you to uproot your life and move to a faraway place where you don't know anyone. And how long would it take for you to familiarize yourself with the new job and become reasonably productive?

My answer would be 1-2 years.

> how long would it take for you to uproot your life

If the price is right, I can be there tomorrow. Companies pay to relocate employees all the time; show me the money.

> how long would it take for you to familiarize yourself with the new job

According to this article, Intel and TSMC have waived any need for an advanced degree and have endorsed this program from Maricopa Community College, which takes ten days to complete: https://qz.com/you-can-now-become-a-semiconductor-technician...

'Traditionally, semiconductor technician jobs required an associate degree. Dropping this educational requirement helps widen the talent pool amid the ramping up of a domestic supply chain, Pearson said. Employers are realizing that “as long as they just get the skills, they really don’t need the degree,” he said.'

I didn't mean temporary assignments. I meant a permanent relocation and everything that it involves. Leaving your former life permanently and building a new life in another place. And dealing with this not only for yourself, but also for your spouse, children, other family members, friends, and pets.

And once you start in your new job, you still need to familiarize yourself with it until your productivity reaches what was expected from you. After all, the goal is not just employing someone but getting them to make substantial net contributions towards the organization's goals.

Not OP, but here's a link to their job listings for AZ. I checked a few random job descriptions and none of them mentioned hourly wage/salary, which is usually a bad sign.

https://ro.careers.tsmc.com/search/?createNewAlert=false&q=&...

On an interesting note, I did notice the wording of what was said in the article was odd

> [...] skilled workers and technicians needed to move equipment into the facility.

> [...] specialized expertise required for equipment installation in a semiconductor-grade facility

This doesn't sound so much like the day-to-day factory work that they struggling to hire for, but more so something like: delivery of equipment, warehouse receiving, and equipment install & maintenance personnel.

> I checked a few random job descriptions and none of them mentioned hourly wage/salary, which is usually a bad sign.

It's simply not a legal requirement in Arizona to disclose wage/salary range on job listings.

You need to look at TSMC listings in states like California[1] or Washington[2].

[1] https://ro.careers.tsmc.com/search/?locationsearch=CA

[2] https://ro.careers.tsmc.com/search/?locationsearch=WA

Which is an even worse look for TSMC. They could easily do it, since they already do for other states, but they want to pay bottom dollar so much that they would rather retain that card (a relatively minor one for an employer) in their hand than play it.

Is Intel voluntarily disclosing in Ohio job listings? Nope.

Is Samsung voluntarily disclosing in Texas job listings? Nope.

A "worse look" argument simply has no basis in reality when this corporate behavior is effectively status quo across all relevant competitors in the industry.

Relevant competitors would include higher paying software technology businesses because a smart enough person to work at TSMC doing advanced microchip work will instead choose to go into a different field.

I'm not saying such a thing isn't status quo, but simply that, if TSMC wishes to be perceived as "competitive", then an exceedingly easy way to that is to buck the status quo by listing their salary ranges - a competitive action. It seems disingenuous that they would prefer to have it both ways, to cry that they are struggling to attract/keep employees in those positions, yet choose not compete with other companies in that area, by listing information they aren't required to supply.

It just seems like that classic quote, "We've tried nothing and we're all out of ideas." That is what makes it a bad sign - they're excited to complain, but (apparently?) completely incompetent when it comes to fixing the problem they're in.

The evidence is an informal thought experiment: if the salary was $100 million for each position, they would have every role filled. Therefore, there must exist a clearing price for each role between $0 and $100 million such that each role would be filled at or above it (and below which it would not be filled). If there are unfilled roles, it is because the role is below the clearing price.

EDIT: all other effects can be ameliorated eventually by paying more. If there isn’t enough specialized labor for the role, the unspecialized labor will be incentivized to specialize themselves. If there isn’t enough labor in the local market, they will move there.

> The evidence is an informal thought experiment: if the salary was $100 million for each position, they would have every role filled.

If the salary was $100 million for each position they might fill every position--but could they keep them filled? If I were getting paid $100 million a year a single month's salary after taxes is enough to retire on at well above my current standard of living. I suspect it would be hard to retain people if they have the option to at any time quit and will never need to work again.

The other lever is quality of life at work.

The relevant measure is the ratio of pay to quality of life at work, not just pay.

What happens after the position is filled doesn’t matter for this argument. The point is that there is a price that will cause somebody to say “this is a good deal for me” and fill the position. If there’s a better deal somewhere else, they’ll go do that instead.

> EDIT: all other effects can be ameliorated eventually by paying more. If there isn’t enough specialized labor for the role, the unspecialized labor will be incentivized to specialize themselves. If there isn’t enough labor in the local market, they will move there.

Proof in the pudding -- working at oil / gas / mining camps in Canada.

All you need is the ability to climb a ladder and be a proud 8th grade graduate; pay is $100k+ CAD, and often much higher for skilled or specialist roles with degrees and certs; easily 200k+ in some cases.

No shortage in finding bodies. Working conditions are brutal, long times away from home, and no shortage of Dear John letters, but there wasn't a shortage of labor.

When there is good $$$ on offer, people will line up around the block. Esp. if all it takes is a 2-week course at the local community college.

I think they were offering like 140 for senior positions. There was a post 3-6 months back I can’t find it at the moment

Someone posted some jobs listings a while back. They pay on the low end and they want a lot of work in return. It’s not surprising at all.

I suspect not...

BLS national semiconductor industry-specific occupational employment data[1] circa 2022 suggests there are roughly as many 15-1252 title-specific "Software Developers"[2] (13,220) as there are all positions categorized under 49-0000 "Installation, Maintenance, and Repair Occupations"[3] (13,410)...never mind that what requires specialized handling is bleeding edge ASML EUV tech for which there is a single deeply backlogged global source of, substantially increasing risk to TSMC at what is clearly a critical phase.

[1] https://www.bls.gov/oes/current/naics4_334400.htm

[2] https://www.bls.gov/oes/current/naics4_334400.htm#15-0000

[3] https://www.bls.gov/oes/current/naics4_334400.htm#49-0000

Sorta. If the internet is to believed:

> “The reality for people from Taiwan is that they are doing even more than 12-hour days often,” said the American engineer on Glassdoor. “There’s also the night shifts and weekend shifts on duty and/or on call.”

- https://www.eetimes.com/tsmcs-arizona-culture-clash/

>“Sure, TSMC might allow a reasonable expression of opinion [on work-related matters]—but only from an engineer or deputy manager to the department manager,” Joey, who has worked as a 5-nanometer chip engineer for TSMC in Taiwan for nearly six years, told Fortune. “It’s impossible for managers to express their opinions to upper-level management. This simply cannot be done,” Joey said. (He asked to be identified only by his nickname due to fear of reprisals.)

> [...]

> “Our salary is only [for] 10 hours [a day], [but] we don’t leave until we’re done. And we’ve never been willing to report it,” a member of a private 85,000-person Facebook group for current and former employees of TSMC in Taiwan wrote in February.

- https://fortune.com/2023/06/03/tsmc-arizona-plant-jobs-salar...

This doesn't sound like it's only a wage problem. This sound like TSMC trying to use a work culture that just doesn't mesh with what Americans are willing to tolerate.

Pay high enough and people will tolerate everything that you describe.

Just imagine if every chip assembly worker made 500K/year - would they care that upper management isn't listening? No, they wouldn't. Would they be okay with 12h/day? A lot of people absolutely would.

The problem is rock bottom pay for rock bottom work environment.

While I completely agree, there might be something else, i.e. the elephant in the room. While having TSMC factories in the USA is great for everyone, it's not ideal for Taiwan as a state. Therefore it's in the best interest of Taiwan to prolong the current situation where the whole world depends on TSMC factories located in Taiwan, and not anywhere else.

TSMC Taiwan is keeping their leading-edge nodes in Taiwan with a 3-4 year delay, which appeased the Taiwanese public/officials.

The bigger issue is that China might catch up to TSMC USA fabrication by the time that they actually start fabbing due to all these delays and China will have a superior chip capacity, making the defense aspect completely moot.

> Making the defense aspect completely moot

The biggest defense aspect (for the US) is preserving supply in the face of a Chinese invasion of Taiwan, not beating China in chip capacity?

I would not be opposed to the gov subsidizing 50% of salaries of chip manufacturer salaries to achieve this.

It's a crucial part of industry growth and military superiority. We have some of the top-tier minds for chip design, we need top-tier manufacturing.

Nvidia/TSMC has already proven that hardware superiority is a huge advantage. We absolutely need it.

Unfortunately, this will set America behind.

Big tech all had employees working 10-16h days in the beginning. Even now, software startups are working their employees hard. If Americans aren't willing to work extra hard, we're heading towards what happening in Europe (negative GDP growth, increasing wealth gap, high youth unemployment).

If anyone has reason to believe that TSMC is gaming the US visa system, or even suspect it, they should probably give the United States a heads up.

https://www.uscis.gov/scams-fraud-and-misconduct/report-frau...

* The H-1B worker is not or will not be paid the wage certified on the Labor Condition Application (LCA).

* There is a wage disparity between H-1B workers and other workers performing the same or similar duties, particularly to the detriment of U.S. workers.

* The H-1B worker is not performing the duties specified in the H-1B petition, including when the duties are at a higher level than the position description.

* The H-1B worker has less experience than U.S. workers in similar positions in the same company.

* The H-1B worker is not working in the intended location as certified on the LCA.

TSMC is Taiwan's single largest strategic national interest. It was started with government investment and it likely remains under a degree of national control. If you replicate the fabs of TSMC in the US, Taiwan will lose its strategic bargening chip. Why have the US defend the island if it offers no additional value?

With no 2nd source supplier the US is in a position to provide strong deterrence against invasion.

If that's the only reason the US is continuously poking the Chinese panda, we should keep that in mind the next time they invoke the humanitarian card.

It's possible to take this too far. I agree that TSMC is a major piece of the U.S.'s interest in Taiwan. but it's not the only thing. The U.S. supported Taiwan long before TSMC came around.

It's not and let's refrain from characterizing complex geopolitical issues as simply the "US is continuously poking the Chinese panda" because it does all of us a disservice and lowers the bar.

Anyway - the United States places value on Taiwan not just because of TSMC (otherwise why was there value before TSMC even existed) but because Taiwan serves as a key piece in the so-called first island chain of US-allied nations that sit on the eastern coast of China. This includes Taiwan, but also Japan and others.

The reason today that America is very interested in Taiwan is because in the event of a Chinese invasion of Taiwan - American commitment to the Pacific comes in to question. Will the US respond to, and go directly to war with China? If it does, can it win? If it does, what other nations will come to the aid of the US (or China), and who will stay out? The key piece here is that it puts the American alliance system to the test and some (not I) question whether the United States is committed to defending other countries in the Pacific theater if it means a "real war" - i.e. Japan, Korea.

The US already doesn't need TSMC for any strategic purposes - while the loss of the fabs in Taiwan would be devastating to the world economy, America (and other nations for that matter) still produce semiconductors, just not quite as good. You can still build cars with Intel chips, but you aren't building the iPhone 16 with them. It's not an existential threat, though certainly a very bad thing.

Alternatively, if China were to invade Taiwan and the US failed to respond, that would be an existential threat to the United States because it would effectively kick the U.S. out of the Pacific and cause long standing allies such as Japan and Korea to play both sides instead of being firmly in America's camp.

Coming back to TSMC building a chip plant in the US - even if they were to (and they probably will) it won't really diminish or change the Taiwan Question and undoubtedly as part of increasing US defense aid to Taiwan the US is going to want to hedge its bets a bit and have an operational fab in the US so that if a war does break it the impacts are lessened. I don't see this as a change to US Pacific strategy because it's not a fundamental concern.

> if China were to invade Taiwan and the US failed to respond, that would be an existential threat to the United States because it would effectively kick the U.S. out of the Pacific

Hyperbolic? The US not having a wall of ships on China's east coast is not an existential threat for the US.

No it's not really hyperbolic. I wrote why the US is so committed to the region in my post above.

I'd also say characterizing things as "the US having a wall of ships on China's east coast" is itself a bit hyperbolic itself and taken without context quite meaningless.

One might wonder, well why does the US have aircraft carriers in the region?

Maybe it has something to do with things like China's continued threats of taking over Taiwan, or North Korea's launching of ICBMs into the Sea of Japan?

There are plenty of reasons but none of them include existential threat.

The comment belies basic geo political understanding. US for a change didn’t start the hostilities. It is China that has a stated policy of “de throning” the US and replacing the US led world order.

China’s actions have been equally concerning, the build up in South China Sea is a naked attempt at imperialistic expansion. Its activities on its land borders echo much of the same sentiment. It is a de facto revanchist power.

When an ex US colony Philippines invites the US to set up military bases it is not the US doing it, it’s because Philippines has concerns about China.

When S.Korea and Japan start putting differences aside it’s because they share a common threat perception of China.

Not to mention China’s abhorrent pillaging of technology, export of dystopian authoritarian control mechanism to government, and its human rights violations which you seem to ignore, yeah the US isn’t the problem here for a change. The CCP is an absolute challenge to liberal values it must absolutely be restrained. Taiwan is a small piece of that the game is much bigger.

China doesn't have the ambition to become the hegemon, this is a exclusively american neocon disease. The Chinese are instead interested in a multi-polar world, and let me tell you, there is a lot of countries out there that are also tired of American Hegemony and would welcome a change in a multi-polar direction.

No. A simple google search would reveal what China intends to achieve . Speech after speech given by Xi are available to refute your point entirely. While I agree China would want a multipolar world but that is just a stepping stone to its eventual goal of being the Middle Kingdom, this really isn’t in doubt, plenty of CCP literature exists to support this argument.

A simple internet search will give you the translated, summarized, and out-of-context Xi speech the neocon propaganda machine wants you to see. A classical case of manufacturing consent.

Like I alluded to before China’s actions speak louder than words. If speeches aren’t good enough an indicator for you, actions surely should be. China absolutely wants to be a hegemon and is an imperialist power as its actions since the early 1900s depict. Invasion and occupation of Tibet, SCS disputes, routine clashes with Vietnam, Philippines and India. US isn’t even the only aggrieved party here, hell you can even keep Taiwan out if it. It takes staggering amount of stupidity to ignore this reality and parrot tankie nonsense.

> Why have the US defend the island if it offers no additional value?

The US defended S. Korea for ideological reasons, and it will do the same again.

But even if you're cynical about these sort of things, the TSMC plant in the US will only provide enough chips for DoD usage. It definitely can't support the commercial sector.

>the TSMC plant in the US will only provide enough chips for DoD usage

Huh? DoD doesn't use any TSMC fabs, they have their own security accredited fabs for that on US shore that mostly cater to US government customers and not much else.

Similarly, TSMC and most other consumer fabs, have no desire to take any DoD contracts in peace time, as the legal and security related paperwork are not worth the huge hassle and low margin prices the government is willing to pay in peace time compared to what Nvidia, AMD, Apple, etc. fork up.

TSMC makes semiconductors used in F-35 fighters and a wide range of “military-grade” devices used by the U.S. Department of Defense (DoD).

https://www.csis.org/analysis/semiconductors-and-national-de...

TSMC makes commercial off the shelf chips fore everything, and the DoD happens to make use of some of those off the shelf parts as well, like the FPGAs in the article, but they aren't bespoke designs made to order from the DoD like the ones made on the fabs specific for that.

DoD primes integrate COTS systems, many of which heavily integrate COTS components, e.g. FPGAs designed by Xilinx and manufactured by TSMC.

What the parent is referring to is DMEA's Trusted Foundry Program[1], of which TSMC is not an accredited supplier of foundry services.

[1] https://www.dmea.osd.mil/otherdocs/accreditedsuppliers.pdf

Which ironically they are trying to move away from(custom ASICS), towards more fpga based systems, since they can be somewhat harder to reverse engineer (battery back your memory/processor and scramble the bitstream when the device is tampered with).

"Why have the US defend the island if it offers no additional value?"

They won't. That's why they are cloning it in the west.

Taiwan is a good bargain chip and has been one for many years, and also a good source of income for the US. You sound like US is doing Taiwan a favor but in reality all favors of Uncle Sam need...repay.

Like the Godfather.

The people with the expertise to design, setup, run, maintain, and improve the fabs will still be in Taiwan. The US will seek a secure source no matter what. By building fabs in the US, TSMC reduces the incentive for the US to bypass it and develop those expertise for themselves.

Conversely, if Taiwan relied on its fabs to incentivize protection, what happens after those fabs get destroyed in the opening minutes of the conflict?

Taiwan is a vital piece in the US’s first island chain policy of corralling China into the South China Sea. The whole plan falls apart without Taiwans naval backing.

Considering TSMC corporate culture and staffing practices, and the scale of raw material and supply chain complexity needed for advanced integrated circuits manufacturing - most of which China dominates - I think we're in a bind here. It was the overlap of American capital and engineering and East Asian resources and labor that made the current scale of compute possible.

As China absorbs Taiwan when their area denial capabilities become credible enough to keep the American navy away, we can expect an end to cheap compute. Either nationalists in Taiwan degrade assets enough or China simply shuts us out. But the second half of the 21st century won't be full of enterprise saas and a supercomputer on every pocket.

>we can expect an end to cheap compute

Hot take: If that means on-shored computing HW jobs back in the west, I'm all for it. I'm only upgrading my phone/PC every 6 years or so anyway, I can live with them being more expensive if they're made in the US/EU and provide local jobs.

Them being more expensive could also mean more focus on SW improvements and optimizations instead of throwing more HW resources at bloated and poorly written SW, more focus on longevity and repairability, and less e-waste for the environment, at the cost of corporate profits and the consoomerist attitude focused on hyping up this year's must-have "new-shiny" that's just 2% better than last year's "old-shiny". All I see is a win-win.

Fair points. Although if computers become more expensive the overall result is everyone becomes poorer.

Also, the energy transition towards decarbonization entails smart grids and intense use of computers for design, deployment and operation of a more diverse fleet of energy generating and storage assets.

People with the right skills might be better off, that's true.

>if computers become more expensive the overall result is everyone becomes poorer

How so? This might have been the case in the '90's and '00's when computers were still in adoption for home users, but nowadays, almost everyone already has a usable PC and a smartphone, even in developing nations in Africa.

Them being more expensive will mean only businesses and rich kids will get to play with the latest shiny, while Average Joes will have to stick to older tech and focus on sustainability and repairability instead of upgrading every 2-3 years.

I could be wrong, but I see this generating more HW repair jobs for technicians in every community like TV repair shops in the old days, and even more SW jobs needed to improve SW to run on older HW in order to target more consumers.

I don't see how this makes everyone poorer, other than execs at companies making a living selling a new HW widget with 2% improvement every year while abandoning SW support for last year's widget to force everyone to throw their old widget away, cough, Quallcomm, cough.

Or people have to get by with old hardware running no longer supported or patched software.

If things are more expensive, people can have less things and are thus worse off. This is circular, poor = less able to acquire = things are relatively more expensive.

> As China absorbs Taiwan when their area denial capabilities become credible enough to keep the American navy away

I don't think it's going to happen anytime soon

American generals have stated it could be as early as 2027 or 2030. When a country says they'll take land as soon as they can, I'm inclined to believe them. China missile deployment on South China sea is quickly becoming overwhelming.

The US has bases both north and south of Taiwan, in Japan and the Philippines respectively.

I’m against China seizing territory in the South China Sea and believe that the US needs to focus on rebuilding its MIC (as shown in the Ukraine conflict). We should address both of those serious issues promptly.

But let’s not exaggerate the case, either.

China says it's coming. US military says it's coming. Policymakers in US and EU are adapting to this shift. How am I getting downvoted and told off for trying to discuss the consequences to tech markets in a tech oriented forum on a thread about the consequences of it?

I’m not telling you off: I think you raised a common sentiment and I was responding.

I’m disagreeing that the saber rattling from two countries both experiencing civil unrest and economic hardship will translate into an armed conflict for Taiwan (generally) and that the missile buildup in the SCS is “overwhelming” (specifically). My point about US bases adjacent to Taiwan is a specific response: we already have forces with both access to the Pacific and clear lanes to Taiwan, positioned and ready.

The job of the DOD is to let the nation know our military status — and they’re letting us know that the building of US fleets versus Chinese fleets implies a weakness in the 2027-2030 range, which could be utilized by China. But that doesn’t automatically translate into an armed conflict during that period.

More broadly we’re shifting policy to compete (eg, by on-shoring) — but again, that’s different than an armed conflict.

I think navigating this conflict requires we not be hyperbolic in either direction: dismissal nor concern.

I agree with the content of your comment. I hardly think I was hyperbolic when stating "compute might become more expensive, thoughts?". A knee-jerk "whoa calm down" response to people raising such topics will inhibit healthy discussion. Someone said this was a non concern and I defended the value of thinking about it and discussing it. So at that point to say things are getting blown out of proportion is not warranted at all.

This isn't unforeseen. The bind is understood and being anticipated, hence the recent incentives to build factories in the US.

What we are seeing is the 'uncomfortableness' of shifting gears. US workers adapting and learning, populations moving, etc...

Everyone complained when jobs left. Kind of funny we complain when jobs come back.

Foreseen doesn't mean fixable. The scale of production achieved in East Asia is unique and doesn't look reproducible elsewhere at the same price points.

The same price point is the bit which they already surrendered upon by subsidizing production locally.

Which would indicates my analysis is aligned with that of American and EU policymakers.

I will never understand how they managed to convince TSMC to build a fab in the US, it's literally not competitive at all

What is the rational behind this choice?

Israel would have been a better choice

Israel is expensive to fab because of a labor crunch with American companies (eg. Intel, Micron) offering competitive salaries and land prices being Silicon Valley level insane.

This was why Tower Semiconductors (Israel's version of TSMC) negotiated sweetheart deals to open a fab in the US, China, and India before canceling them after an acquisition offer from Intel as well as COVID related supply chain issues in China's case.

A similar labor crunch has started to occur in the cybersecurity industry in Israel as well as MNC salaries have begun to reach Seattle levels

Israel is an expensive country. It's literally the same size and population of the Bay Area, and has similarly high land prices. This is also a big reason a lot of Israelis started buying houses and apartments in settlements in the West Bank - for the same price you could get a crummy 2 bedroom apartment in TLV or Jerusalem, you could get a house in Ariel or Efrat instead

It’s quite competitive.

Some of the largest Intel fabs are 52 miles away. But TSMC screwed up when selecting a site in Phoenix. They built on the other side of town. Any engineer currently working at Intel would need to sell their house and move to the other side of town, or be subjected at a 90min commute each way. If their plan was to lure those engineers, they’re going to need to pay more or provide a better quality of life. They appear to plan neither.

https://archive.is/HdLIo

Skilled hardware technicians for such tools (some from ASML) have a sciences/maths background (no, javascript/c++ coder without such background does not cut it and where did you see "highly skilled" here??).

They need a lot of "training", so, if there is no long term outcome... that's going to be tough.

Are the skills even transferable or are there so few companies that do this is a one company job

If I recall well a documentary about ASML, each photolitography tool has such trained individual attributed during its whole life time. And that was for "simple" DUV tools (immersion photolitography).

So for the EUV-HighNA tools...

Why live in a state that is running out of water? Isn't water necessary for chip-making, too?

Water is also so essential that it's safe to assume there will be a solution. So it makes sense to look at other factors, which are less likely to be fixed as swiftly.

Because water can just be manufactured? I don't understand your making that assumption when the states in the region, plus Mexico and Native tribes, are fighting over what's available.

Yes, fresh water can just be manufactured from salt water. It's not especially cheap, but it is surprisingly affordable. Arizona is not worried about Mexico or the tribes, push comes to shove they're getting nothing.

uh oh, isn't China's taiwan invasion scheduled for 2024?

>labor shortages

There are more than 300 million people in the US. This is definitely a shortage, but this isn't a "labor" shortage. This is a "we're not willing to pay you enough why does nobody want to work" shortage.

Counterpoint - sending people from TSMC that already have hands on experience will make things much easier for them.

If they only want people like that, then yes, they can't just divide the facility in two, like a cell. They'll need to train more.

There are not 300 million ASML experienced engineers in the US.

If you are experienced with ASML machinery and are willing to more to Arizona, then you should contact TSMC, because they are paying over the market rate right now for those roles.

These Reddit-esque takes are not worthy of HN.

Is market rate just really low for the role? Do you have any sources for over market rates?

No, the salaries are very high. The challenge is finding sufficient number of people outside Taiwan with ASML experience.

What’s very high? Working 60 hrs a week in a factory I’d expect 200k starting

They’re not looking for minimum wage workers to assemble chips by hand - they’re looking for hundreds if not thousands of engineers and technicians with experience with advanced semiconductor manufacturing equipment. I’m sure the salaries are perfectly acceptable.

Sure there’s 300 million bodies in America, many of them appropriately skilled - how many engineers are there with the requisite experience that are open to work in Arizona?

I’m sure they could send hundreds of their own experts - if those workers weren’t needed for existing operations in Taiwan.

At the salary provided and the work conditions described it’s probably mostly industry rejects and new hires

I'd say it's more than wage. It's more about outsourcing manufacturing and other labors to Non US countries.

It's wage + education + a shift in mindset, which basically means you have to put a lot of elites into prison to achieve that.

You are going to touch the cake of a whole generation of aristocrats.

They are clever enough not to give away their shield...clever girl...

This sounds like a Lenin quote.

I don't care, if they didn't have that factory in Taiwan, there is no way the USA would be defending Taiwan like it is...why do you think the US want to build a factory in Phoenix?

The US has openly stated that it will scuttle the Taiwan factory if China invades.

Lack of skilled labor seems unlikely. Perhaps they believe that the US would be more willing to defend Taiwan if there was no alternative foundry in Arizona...

The fact they chose Arizona to build this thing is mind boggling.

There was a really good episode of the Odd Lots podcast where they discussed this. Arizona is a very favorable location for semiconductor fabs because there's a very low risk for natural disaster (no hurricanes or earthquakes), the temperature changes are relatively predictable, land and electricity are relatively cheap, and Intel already has a large fab there. Water is a little bit of a struggle, but they can buy water rights from farmers, who use most of Arizona's water.

Solar power is extremely cost effective in Arizona. It more than compensates for the additional heat. It also allows for cheaper data center hosting.

>> "but they can buy water rights from farmers, who use most of Arizona's water."

They probably waste less water, so this might be a net positive for the environment. Abuse of water rights given out in a very different time is a major point of contention in the desert region of the country.

Arizona has a low chance of natural disasters, land is relatively cheap, and they're known for being tax competitive.

As others have mentioned there is a large Intel fab (and other smaller fabs) in the area. I suspect this gives them insights into the feasibility of success there and also means there is an existing talent pool to draw from. It basically de-risks the investment. There are probably only a handful of likely places to do this - OR, CA, AZ, TX, maybe NY. Intel building in OH is more difficult from a bootstraping perspective.

It's dry, weather is consistent(though hot), has a robust power grid, has an engineering school who produces semiconductor grads, has a large pre-existing industrial base and has ready access to water (the plant is literally next to a 340 mile water canal that runs from the Colorado river and sits atop a water basin that has had water levels rising since the 80s).

I'm not sure there are many better places outside Oregon, California or Texas...

Reply on news.ycombinator.com