People: We want stuff cheaply, immediately and in immense volume!
Production: Moves to Asia
People: We want jobs!
Production: Opens a new factory in the USA
People: We don’t want to work that much!
Production: increases pay, which increases product cost
People: this is too expensive. Can we buy from overseas cheaply instead?
Production: Closes USA factory, moves back to Asia.
People: we want jobs…
That's cute, but not what actually happened. It started with:
Company: We want more profit
Production: Moved to Asia
Other Companies: We can't compete any more
Other Production: Moved to Asia
Companies: Oh crap, we need production back in the US (for various reasons)
Production: Oh shit that's right, we moved because it cost less with no labor or environmental protections and we could work the shit out of people.
Companies: Let's call it a "Culture Clash".
Don’t believe me? See what changed during the last two decades of globalization, product cost and availability or profit margins.
Honestly, I'm surprised so much hands-on work is needed when we're talking about a semiconductor company. One would think a lot of the fine-tuning work, which seems to be what requires a lot of the labor, could be easily automated with current technology.
Ask anyone who worked in retail for a long time. When they replaced the point of sale and catalog search systems in the early 2000's, everyone's productivity plummeted.
They replaced old, DOS/Norton-style keyboard-only systems with a legion of shortcuts and archaic codes with Web-based touch-screen or mouse-based UIs with modern UX etc.
The new technology was obviously more modern and allowed faster training, and it's hard to imagine any world where that upgrade doesn't happen. But the old system was faster, and allowed experienced users to do more.
Now, if you're talking about a Blockbuster Video, maybe it's the right choice.
But if you're talking about extremely specialized semiconductor manufacturing, maybe you stick with the highly optimized system that works. Some things changed a lot since 1987. Others haven't.
There's 168 hours in a week. If people work for 40 hours a week, to cover the 168 hours you need to employ 4.2 people for every employee you want on the floor at any given time. That means everyone works 5 8-hour shifts, or 3.3 12-hour shifts per week.
If you reduce headcount by 33%, that works out to 2.8 people per employee on the floor, which means everyone works 60 hours per week -- 7.5 8-hour shifts, or 5 12-hour shifts. That's a lot.
That's one of the pitfall of low labor costs: since time is virtually free it tends to create a culture that doesn't use it efficiently.
I've seen it with offshored devs way too often; why take the time to learn version control when you can pass around code in emails and on thumb drives?
TSMC was never in North America, it was always a Taiwanese company. It's literally an Asian company moving production to the US.
Taiwan isn't some rando poor Asian country; it's a rich democracy. The work culture is no doubt different (as it is, say, in Japan or South Korea), but to pretend that it has no labor or environmental laws is somewhere between racist and naive.
The story as told above on how production moved to cheap labor countries, jobs were lost as well as purchase power has a weird continuation where the decline of purchase power scales down production and makes life harder for the cheap laborer therefore more fragile. We shift from [the absurd] putting the consumer in charge of the economic decisions to having the "choices" made by (or more like making the results depend on-) people all the way at the bottom of the food chain. We are pretty much left collectively hoping that they can keep up with our unreasonably demanding demands.
I'm not even complaining that we are squeezing people to hard. I have issues with a system that can just abruptly stop - entirely! Sure, I want cheaper products but I have no idea what kind of risks we are taking. I should not be the one to decide. I would rather have a large group of highly trained professionals calculate the risks and present them in a way we might understand. Make me a map of the [proverbial] meteor strikes and the global economic consequences. What parts of the show can go on if ingredient X is taken out of the soup?
(edit: Ofc there is more than cheap labor pushing localized specializations which increase efficiency and decrease resilience)
Again, there's a pretty racist subtext here that because it's in Asia, you assume that must be cheap labor, not innovation. Would you have the same argument if the fabs were in France or Spain, both of which have lower GDPs per capita than Taiwan?
From the article is seems workers in Taiwan work 12 hour days as normal. Couple that with the fact that manufacturing used by primarily US being done outside US borders is generally some for cost saving reasons and it seems to make sense to assume that Taiwan IS doing this cheaper somehow. If it were a matter of innovation wouldn't the same innovation hold in their new plant in Arizona?
There's also a difference between what you're saying and what the great-grandparent said: being cheaper can be because they have better processes or are more innovative, or have shorter supply lines, or better access to mainland China or whatever.
But the great-grandparent both (a) assumed that tech fundamentally belongs to the US, and if it's being done somewhere else, it's in service to the US, and (b) that the primary reason someone would produce things outside of the US is labor costs.
Labor costs are no longer the primary reason things are produced outside of the US. At this point, it's primarily industrial capacity and know-how.
Not "10x less" or anything incredibly dramatic, but for a business, money saved is money saved.
It's silly to have a whole thread of back and forth responses on "is it racist to say cost of labor is cheaper somewhere else" without looking at the numbers to see if it actually is. GDP per capita is very much not the same as salary.
Personally I'm waiting for engineering salaries in the non-US to catch up to the US because it'll make it easier for me to leave the US if I choose to do so but it doesn't look like it's happened yet.
Engineers in virtually every country in the world make less money than in the US. But the reason people have jobs in the rest of the world isn't because everyone wants to make things in the US, but it's just too expensive, so they'll settle for somewhere else. That's pretty heavy-handed American exceptionalism.
TSCM has the best fabs in the world. This article literally mentions them having to build the factory in the US and send people over to teach the Americans how to run it. They're just better at it. To keep insisting that that's not the case, and that this is really all about their lower salaries is, in fact, racist.
I'm talking about the how of them becoming the best in the world, not arguing that they aren't the best now. As money for semiconductor manufacturing moved to overseas companies, initially driven by costs, those companies ended up with the money to invest in getting good at it.
But that money was directed there as a result of specific actions by people in the US. So yes, all the conversation about immediate-cost-focused decisions, and the difference in the labor costs, are relevant.
It's not "American exceptionalism" to be aware of this history, it's "American exceptionalism" that caused this and that drives the laissez faire attitude that there wouldn't be domestic impact like loss of expertise and reliance on overseas producer from a focus on immediate cost over all else. The idea that if we wanted to, we could simply build the industry back up. Because we're America! Same as all the rest of "it can't happen here" BS that people use to excuse ignoring various domestic issues.
--
To frame the labor cost issue a different way, let's look at what the reverse process, bringing that expertise back to the US, would look like: let's say tomorrow the US wanted to build up a domestic best-in-the-world alternative company. They'd have to hire a bunch of engineers! Even best-case, it would take those engineers quite a while to reproduce all the earned knowledge and expertise of TSMC's engineers. All the while you're gonna have to pay those engineers more than TSMC's engineers are making, because it's a more expensive labor market. And then even if they do produce equally high levels of output after a while, all those higher costs are going to put them at a competitive disadvantage. So if you want to be good at it domestically, and you live in a country with more competition for labor, you really have to commit to spending extra to do it. (Getting TSMC to build some plants domestically is a way of short-cutting the knowledge acquisition step, but it's just a small step towards really rebuilding the industry.)
A lot of manufacturing did move to Taiwan in the 70s and 80s because labor costs were lower there then, and because the US was investing a lot in Taiwan because of its strategic role vis-a-vis mainland China. But that wasn't especially high tech, and was decades before TSMC became relevant on the global market for anything but very low-end chips.
The way that TSMC ended up on top was by beating Intel at its own game. It did that at a time where Taiwan was already rich, and the US wasn't actively subsidizing the semi-conductor industry, whereas Taiwan's government was. It seems naive to assume that Intel wouldn't have had a decade of missed deadlines if it'd just thrown more money at the problem. In that this article started off about corporate culture, it's not hard to argue that much of what killed US chip fabrication dominance was Intel's corporate culture (as opposed to cheap Asian labor).
Basically there was a lot of consolidation in fabrication in the US, then with all of the eggs in a couple of baskets, the US-based fabs dropped the ball, and TSMC and Samsung smelled blood and stepped up their game.
I'm not particularly focused on Intel at the very top-end here. TSMC capitalizing on Intel dropping the ball in the 2010s doesn't happen if there wasn't a trend that continued through the 90s and 2000s that helps them get in the position to capitalize. Labor costs then were part of the equation - after all, they still haven't fully caught up.
The US being down to essentially one player that made their living focusing on just the top of the market is because of that process that hollowed out the US-based industry.
The alternative scenario I'm pitching isn't "Intel doesn't screw up cause they throw even more money at it" it's "the US semiconductor industry has multiple big players so that the are more domestic players to try to pounce on Intel's mistakes." But, because of cost pressure, including cost of labor, I believe that would take specific action and policy choices.
(This also isn't just a "US vs Asia" thing - Japanese companies have lost business to other more recent Asian entrants in the electronics and related manufacturing space as well.)
What difference would it make to your quality of life if you made $60k in Europe instead of $150k in America? Sure, your net worth wouldn't increase at the same rate, but should we define our lives around our net worth?
Having money in the bank is basically the same as having choices.
>as well as our long-term economic security and flexibility >Having money in the bank is basically the same as having choices.
Do people in Europe not have long-term economic security and flexibility? Do they not have choices?
Let me chime in to say that salary isn't everything in life, or the key to live a good, fulfilling life. I'd even argue that socialized healthcare and worker's rights, ensured in numerous European countries, do far more to achieve good life than, say, an absurdly high salary.
In the latter case, that point is different for different people, but if you're making FAANG seniorSWE+ level compensation (especially if it's at one of the easier FAANGs), I would think the options to optimize your quality of life is probably higher in the US than in the hypothetical more socialistic European countries you mention.
What chip foundries exist in France exactly? Strawman much?
46,000 employees.
> Americans workers at TSMC’s Arizona fab may chafe at becoming the humble students of their mentors in Taiwan, who nearly 50 years ago transferred semiconductor technology from RCA of the U.S.
It was more geostrategic rather than capitalist outsourcing that lead to the knowledge transfer, but after that transfer was complete, the economic factors mentioned by the person you're responding to would fully be in effect. A nontrivial part of Taiwan's economic success has been that knowledge transfer that enabled the creation of TSMC.
This isn't about a particular company, it's an industry-wide trend. Semiconductor manufacturing was predominantly in the US, then it wasn't.
> The work culture is no doubt different (as it is, say, in Japan or South Korea), but to pretend that it has no labor or environmental laws is somewhere between racist and naive.
This is actually the root of the problem. The problem isn't the existence of labor or environmental laws, it's that many labor and environmental laws exist in the US that can't survive a cost benefit analysis, so the US loses manufacturing to countries with a more efficient regulatory environment.
People act like regulation is some binary thing where the only choices are "on" or "off" when the reality is that the details are what matter.
There's more to it than that.
The ban on leaded gasoline makes gasoline somewhat more expensive. If you multiply that by all the unleaded gasoline everybody has bought since the 1970s, that's a lot of money! But lead is really, really, really bad. That ban is cost effective even if it costs billions of dollars. Which it does, but the alternative is much worse.
There are other regulations that cost just as much and provide nowhere near the same benefit. The deterrent effect on doing business in the US of both regulations is the same, but the benefit we receive isn't.
And the deterrent is cumulative. There are advantages to doing business in the US. An educated workforce and a stable government etc. Companies move out when the cumulative regulatory deterrent exceeds the advantages.
Which means you can have a ban on leaded gasoline without anyone leaving, because the cost of that doesn't exceed the other advantages of the US. Which means that anyone trying to repeal the ban on leaded gasoline is doing it wrong.
But if you have a ban on leaded gasoline and 250,000 other regulations that each cost just as much but are barely breakeven if that on the benefit side, you exceed the threshold and lose the jobs. To get the jobs back, you have to get rid of some of the regulations.
We need to get rid of the ones with a poor cost benefit ratio, not the ones with a great cost benefit ratio. Which is, to be honest, most of them -- but it's still really important that it be the right ones.
The advantages of doing business in the US that are balanced against "cumulative regulatory deterrent" are largely independent of the specific regulations involved, though not always (much harder to get an educated workforce when everyone is lead-poisoned in childhood). Deciding which costs should be considered baked in (you're saying "no leaded gasoline") and which are on the margin (you're saying whichever 250,000 you claim are as costly as no leaded gasoline) is not inherent to the problem based on "efficiency" alone.
"Benefit" is rarely neutral. Values and politics are its domain. ("It's your luggage that has put the plane over its weight threshold, not my luggage!")
Also,
> To get the jobs back, you have to get rid of some of the regulations.
presumes that other countries' regulatory environments are functionally fixed and not impacted by international cooperation. International effort imposed extra costs on those who would employ child labor, for instance. Other types of responses exist besides "well we have to let them give Kentuckian factory-workers spinal problems because otherwise they'll go off and give Vietnamese factory-workers spinal problems!"
TSMC was also founded in 1987, Taiwan has changed a lot over the decades and at a certain point inertia works for a lot even as costs may even out compared to in the past. What were those salaries in 2000?
No, but their biggest customers were making chips in NA - Apple (via Motorola or Intel) and AMD.
The reason?
"Korean Pride".
You were expected to go above and beyond for sake of the glory and well being of the Republic of Korea and the Korean People, via the proxy of The Company.
Most of us (myself included) internally thought this was total bullshit, and were on the lookout for the first opportunity to jump ship specifically for a non-Korean company. Of course we would never admit that to a manager, especially the expats from Korea.
It was also known that the best teams to work for in terms of work-life balance and just general pleasantness were ones where the manager and teammates were comprised mostly (or entirely) of non-Koreans, or at least heavily Americanized Korean-Americans. Having an older conservative expat Korean manager was the worst possible scenario.
One nice perk was having a big name like The Company on your resume wasn't bad at all for future job prospects.
Profit has all sorts of accounting aspects such as depreciation (which, for a massive chip plant must be enormous).
Like, this is the primary goal; any other uses of profit are valid only with the expectation that they will allow the owners to extract more money later.
Sure, gross profit would not include either of those expenses. But operating profit would subtract research and development (e.g. the salaries and bonuses of the engineers working on the next process node) and depreciation on new factory (i.e. the cost of the new factory spread out over its useful lifetime - though it doesn’t include a factory under construction). Then you subtract interest and taxes to get to net profit. And TSMC has net profit margins of 39% in 2020!
> In November 2020 officials in Phoenix, Arizona in the United States approved TSMC's plan to build a $12 billion chip plant in the city. [...] In 2021, news reports claimed that the facility may be tripled to roughly a $35 billion investment with six factories.
> In November 2021, TSMC and Sony announced that TSMC would be establishing a new subsidiary named Japan Advanced Semiconductor Manufacturing (JASM) in Kumamoto, Japan. [...]. The initial investment will be approximately $7 billion with Sony investing approximately $500 million for a less than 20% stake.
When you're committing to $42B of investments, $28.4B doesn't seem like all that much to have on hand.
https://www.bloomberg.com/news/articles/2021-04-01/tsmc-to-i...
Crazy! Although, like most companies, they will work hard not to spend their own money on it :)
An idealized company in a static market that has been making the same real (i.e. inflation adjusted) profits for a while and is expected to for quite some time will have its price track inflation, so any dividends paid out will be on top of inflation.
Real life is much messier than the above, due to things like speculation (how much some people think other people will think the company will be worth in a year), changes in the broader market (if the "zero risk" rate of returns goes up or down, then equities, which have higher risk will tend to move in the opposite direction), and such can have drastic effects on a company's market value.
"we want reasonable jobs that allow us to have a life and do not functionally make us the property of a corporate entity that does not share sufficient amounts of the gains of our labor with us and labels requests to be treated with respect as unintelligible"
States/those with power claiming things are 'unintelligible' when in fact they just don't like them or want to treat them as valid is a classic form of control.
That ship has sailed.
Now they have to be more in line with the rest of the world.
What the "corporation" can offer and survive or stay ahead of the pack is a function of choices available to it. Which are different now than decades ago.
This myth that Americans can demand better things that the rest of the world can't needs to die. If they want to work less, great! They can. They need to become more efficient and produce more goods with less labor. Otherwise people in other countries will not want to trade with them. That is reflected to the American worker through the set of jobs and work conditions available to them.
And this article makes that point very salient, as does antiunion drives at many companies. Say it costs me X to run a PR campaign about lazy workers or hire an antiunion consulting firm. It might cost me Y to raise pay, hire more workers, and/or improve working conditions. If X is less than Y and achieves the same effect - I do X, if Y is less, I do Y.
Employees are allowed to make employers compete as well. This is HN, like the home of the switch companies regularly in silicon valley & silicon valley was created because no non-competes.
In a flush labor market, employees compete for jobs from companies. In a tight labor market companies might have to compete (and treating workers better/paying more is a form of competing). If you don't want to compete by paying more, you can compete by accepting lower quality - just like if I go buy a product from a company or its competitors.
The strident defenses of the poor companies having to compete for labor being unfair is just such a strange narrative to me but one that has weaseled its way into a lot of the American subconscious.
If TSMC did know enough about American semiconductor employee work culture before investing $12 Billion in a new factory that they are surprised by this that's on them. If we accept your claim about American workers demanding higher wages than international workers for the same effort, then TSMC certainly knew that going in. TMSC isn't some innocent and naïve child...but they did build near Intel's castle hoping (it seems) to extract intellectual capital. A lot of those ex Intel-ers leave because of burnout and may have a shorter tolerance for BS than TSMC. TSMC is one of the top 10 employers in Taiwan, maybe that was something they didn't consider the influence of that in their work culture enough?
My suspicion is that this is an effectively planted article as part of a drive for better incentives by TSMC. This stinks of Foxconn and Wisconsin.
TSMC is upset because of the lack of enough slave labor in Arizona.
But there's also the question of whether it is truly necessary. Will hiring more workers to cover more shifts really increase production costs that much?
If you're competing with cheap labor, possibly slave labor, then you have to look to other ways to push back, like fast tracking immigration of talent from places like Taiwan, imposing tariffs on imports from countries with cheap labor, investing in automation to reduce production time and costs, etc.
Sometimes there isn't much you can do, but generally speaking, while private business is a good thing, it is also true that private business cannot contradict the common good. The idea that you can simply do whatever you want because you're private is simply nonsense.