> The loss of Taiwan and with it TSMC — the thinking goes — would result in a global tech meltdown.
The irony is that for most of the tech industry TSMC is unavailable. Nvidia, Apple and AMD have booked all the manufacturing slots for the foreseeable future. It's been this way since 2020.
Yes, their loss would have a huge effect, but many are having to cope with that situation today.
> Yes, their loss would have a huge effect, but many are having to cope with that situation today.
I suspect they'd be in much worse shape if Taiwan were invaded.
First, they'd find themselves competing with current TSMC customers for the non-TSMC foundries.
And second, an invasion of Taiwan could lead to a full out war with China, which might mean the non-Taiwanese fabs have their production priorities overridden by the warring governments.
>an invasion of Taiwan could lead to a full out war with China,
What would be the casus belli?
Quite literally no country of even remote significance on the world stage (aside from the Holy See) officially recognizes Taiwan.[1] By what pretext would a country declare war against China as concerns Taiwan?
Please. Casus belli doesn’t matter these days. You just do enough propaganda with a number of justifications and control the news coming from the other side until the poll numbers look right.
Casus Belli is now translated into "America could win". There is no clear Casus Belli with China. Last time the US faced China in combat was Chosin Reservoir...maybe the worst ass-kicking the US has ever received. No way voters are going to support a President who oversees delivery of 50k dead US soldiers just to save a fab.
So, China aggressively threatens everyone who recognizes Taiwan for decades, and the resultant lack of recognition for Taiwan is used to delegitimize attempts to defend them?
The US has a security relationship with Taiwan that may or may not be a defensive alliance guarantee. Regardless of how ironclad that guarantee is, there is substantial likelihood of significant military supply to Taiwan in the event of a China-Taiwan conflict.
Should China decide to strike that supply chain--whether by accident or on purpose--the result would be US military casualties, and that would have a strong likelihood of precipitating outright war.
Multiple US companies with trillion-dollar market caps that are completely or near-completely dependent on TSMC.
Remember that Japan attacked the US in WWII largely because we cut off their oil supply. Now, let's watch China cut off the US's supply of advanced semiconductors, and see how that goes for them. You bring the popcorn, I'll bring the beer.
Unless the date in the comment is wrong and I missed some important update in 2024, this statement is not correct. Apple Books a chunk of the latest tech for a period of many months every now and then but there is a long tail of companies using TSMC for shorter projects and for using earlier tech.
Is it even realistic to expect US fabs to be competitive? I think this article sheds insight into why the US based semiconductor companies like Intel are falling behind.
> The American engineers complained of rigid, counterproductive hierarchies at the company; Taiwanese TSMC veterans described their American counterparts as lacking the kind of dedication and obedience they believe to be the foundation of their company’s world-leading success.
I mean, some humility is in order. TSMC are succeeding because of these things, not in spite of them. Given the supposed need for expedience it's best to copy that as-is and then you can tweak it once it's up and running.
If the US starts shipping nukes to Taiwan, we will for sure have a repeat of the Cuban missile crisis, with the roles reversed. Unless hundreds of nukes can be installed in full secrecy, China will almost certainly try to sink ANY ship trying to deliver anything from the US to Taiwan, using nukes if necessary.
If the weapons are not placed in Taiwan, there's no guarantee that the US would use them to defend Taiwan.
If the US formally makes such a guarantee, that in itself might trigger a nuclear war.
This is a game of chicken, and games of chicken are rather dangerous. Do you want your city nuked to save Taiwan? Does Xi want to risk having Shanghai and Beijing obliterated to maybe take Taiwan?
If there is an actual war over Taiwan, it's hopefully going to be a conventional one. But that would also mean China may risk it. Which would make TSMC production out of service for months or years (or forever).
Just because the weapons are in Taiwan doesn't mean that the US would use them to defend Taiwan - unlike Cuba I doubt if the military leaders would be given delegated authority to use nukes defensively and I doubt any US political leader would use nukes first when faced with a conventional attack on Taiwan.
To be a proper deterrent, Taiwan would need to be able to control their use. I agree that the US would not be likely to use nukes first over Taiwan. Even losing Taiwan would be far preferable to a nuclear exchange for the US.
We don't know if China would, but I don't think they would start a conflict over Taiwan if they expected to only be able to "win" if they use nukes.
Rather, the only way I can see this become a nuclear war is if Xi starts an invasion due to overconfidence (like Putin with Ukraine), but is then humiliated by a combined Taiwanese/US defense, proceeds to see nukes as the only way to avoid losing face, and then caring more about this than the future of the world.
But let's imagine for instance that Russia ends up conquering Ukraine completely, with NATO letting it happen. In such a case, Japan, Korea and many others may suddenly realize that they're not safe until they have a large nuclear arsenal. Once they have build up such an arsenal, it's not that unlikely (I would say) that Taiwan follows, and creates their own secret arsenal of nukes. Maybe even purchasing the tech from Korea or Japan.
Non-proliferation is a very important principle which has so far been almost entirely upheld. If the US did that, who would China donate nukes to in response? Iran?
>TSMC are succeeding because of these things, not in spite of them
This was the conventional wisdom about Japan in the 1980s as well, and they fell into the abyss with everyone else when China was capitalized and could offer labor for 1/4 the cost. The tactics that get you there aren't necessarily the ones that keep you there.
>Is it even realistic to expect US fabs to be competitive?
The United States invented and has historically dominated semiconductors until the last decade or so, so yes.
The problem here is two-fold: Financial misincentives, and the fact anyone who achieves #1 in anything will always fall from grace eventually and often disasterously.
It also doesn't help that the Chinese/Taiwanese are very shrewd people compared to Americans who are... lazy, to say the least. I say that as an American.
The US has a pretty good work ethic. It's mostly different from East Asia. It's not like Korean, Japanese or Chinese tech companies are THAT much more productive than US ones.
If anything, lower salaries in Asian countries indicates that their productivity is lower than in the US.
People in the US work relatively long hours compared to, say, Europe. They're also some of the most creative and agile out there.
I suspect that part of the reason TSMC is having such extreme success right now, is that they're by far the most successful company in Taiwan. When they're looking for people, it's like OpenAI looks for people in the US, they can get almost anyone.
Put another way, TSMC gets the cream of the crop in Taiwan while US based fabs get 2nd/3rd tier talent who couldn’t get work in other sectors. I can see that.
I think Tesla/SpaceX are similar comparisons. They're known for a tough work schedule but people go there anyway because of the nature of the work. I imagine TSMC is similar in Taiwan.
I've never really equated "hours worked" with "work ethic" - if anything I would say that of all the nationalities I have worked with the people who seemed to have the best work ethic in the sense of turning up, focusing on work and going home again were actually German and they have the some of the lowest annual hours worked.
Being expected to come into the office and sit there just because its expected of you seems the opposite of "work ethic" to me....
I agree, this was kind of my point. Work ethic can mean somewhat different things to different people. Some emphasize hours worked, some effort, some obedience, some loyalty and other honesty, bravery or creativity.
Some of these can even be contradictory. Some workplaces reward the bravery to provide constructive, honest criticism to coworkers and especially senior management. In other places, such behavior can be seen as disloyal.
But there are also cultures that actively repress people who seem overly eager to give their best. I personally know of places where the union has been going after young people who work at their full capacity as it makes the older workers look bad.
In other places, I've seen laziness being seen as completely acceptable. And particularly in groups that feel underpaid, either relative to management or to coworkers in other countries with higher salaries. ("We pretend to work, they pretend to pay us").
A lot of asian work ethic is centered on appearances more than raw productivity. There is a big difference between being at work for 12 hours and doing work for 12 hours. There is tons of stupid appearances fluff going on. They're not super humans.
TSMC has had brilliant success, but this kind of work culture tends to negatively affect creativity and agility. TSMC seems to have found some way to either avoid that or compensate for it, possibly by having the ability to hire only the top technical talent in the territory/country.
And moving forward, people are starting to realize the risk of having TSMC manufacture everything in Taiwan, given how little competition they have. If TSMC wouldn't start to move some of their production abroad, they would risk facing import restrictions to western countries or (as already happened) government subsidies to local fabs.
To retain their market share, they were probably forced to set up some fabs abroad.
In the end, I think it's really good for those new countries to have TSMC set up fabs in their countries, both since it guarantees that some production stays available in the case of a war, but also because the TSMC workers will gradually naturalize in the US or Europe and start to work for companies like Intel or ASML and contribute to crosspollination of ideas abilities.
Short term though, cultural conflicts is exactly what one would expect. It's probably very similar to what Western companies experienced when first setting up shops in China or India.
In the beginning Western managers would try to have the locals behave like Western employees when it came to initiative and willingness to "say it as it is" when managers were giving impossible orders.
Over time many would learn that finding really good local managers able to bridge the east/west culture gap could be really helpful.
I'm guessing TSMC will need to do the same thing. Find Western managers (for instance second generation Asians) able to understand both cultures, so they can deal with local employees the way they're used to while also able to work productively with upper management (who would presumably still be Taiwanese).
I think it's hard to overstate the issues that stem from a workforce who cannot all speak the same language. In my experience it's an absolute showstopper for creating an effective organisation.
Add language barriers to solving very hard technical problems and you're going to have slow progress.
To be clear I am talking about just about language and not the cultural differences other comments are talking about.
This can easily become an economy-wide problem. Canada is an example of this, and it has certainly harmed the country's overall productivity.
Language has long posed a challenge for organizations in Canada with operations and customers in both the English-speaking and the French-speaking regions of the country. These days, though, that has become a comparatively minor issue to deal with.
Severely flawed immigration policies have very quickly brought in a huge number of people who don't know English well, and who also don't know French well, too. In effect, these people don't really fit in linguistically anywhere across the country.
As this third group has gotten larger, especially over the last decade, unnecessary friction has been introduced into even the most basic of interactions, for pretty much all of the participants involved.
In the major cities, for example, it's not uncommon to watch one customer after another with limited English and/or French skills have trouble interacting with a cashier who also has limited English and/or French skills.
Native English and French speakers are left in a position of constantly being unsure if they were properly understood, too.
Miscommunication, misunderstanding, and mistakes have become far more common than they were even as recently as the 1990s and early 2000s. Lots of time, money, effort, and resources are frequently wasted thanks to this.
Canada's productivity has certainly taken an unnecessary hit due to language issues.
I agree with you. Having said that, TSMC’s ramp in Japan has gone much more smoothly than in Arizona, which indicates that all the issues can’t be attributed to language barriers. There is a work ethic/culture delta as well.
In my experience, U.S. citizens are not likely to sacrifice their bodies and life to an entity that can cut them off at will. That sort of dedication comes at an extremely high premium in the us.
Honestly, it is just business. The companies need to adapt and stop complaining.
They are adapting by getting subsidies from the US government, which implies that US based manufacturing will never be cost effective. When the money flow turns off, these fabs might fade away.
It’s never going to be cost effective if you can outsource to a country that has less stringent laws and regulations.
The government has other levers it can pull, such as tariffs.
It also looks like the Taiwan employees are reluctant to transfer domain knowledge, based on the article. They don’t want to have to compete with U.S. workers.
The US cannot afford the risk of having virtually ALL advanced semi produced in a territory as exposed as Taiwan. At least until the threat against Taiwan is gone, they have to make sure a large percentage (>50%) of such strategic products happen outside Taiwan/China, meaning they have to continue subsidies (or switch to tariffs) until the threat is gone.
Anyway, just because TSMC puts Taiwan in a position similar to Japan in the 1980's, doesn't mean they retain that advantage into the future
At some point in the future, it's likely that invading Taiwan militarily will not longer be part of the Chinese agenda. There are multiple ways for this to happen:
1) China could have a regime change. This could either lead to voluntary reunification or at least to military action becoming irrelevant.
2) Xi could die, and the successor might have other priorities.
3) China could lose the ability or confidence to invade, for instance if their economy is destroyed by corruption or they stop investing in their armed forces.
4) Taiwan could develop their own nukes in secret, and be too scary to invade.
5) The threat could become reality (and thus no longer a threat). China may proceed to either win or lose such a war. In such a scenario, it's hard to say what remains of TSMC's capacity after.
1-4 are unrealistic. 5 is the likely outcome and Taiwanese production is no more.
In my view it’s not a China Taiwan conflict more a US China one.
Edit: China is trying to get hold of Taiwan politically not necessarily militarily and if I’m not mistaken the US said they will not tolerate this either.
I'm sure a lot of people inside Taiwan see this as a Taiwan China conflict, with the US being an ally that has shared interests with Taiwan.
From what I've been reading, the support for reunification within the Taiwanese population has been going down for a long time. The current talk of invasion could reflect Xi giving up on that possibility.
As for what would happen if some day the majority in Taiwan actually supported a peaceful reunification, the US wouldn't like it, but it wouldn't be easy for the US to defend attempts to stop it militarily.
Now, that wouldn't completely rule out some kind of covert support for a military coup or something. But if the support for reunification is large enough, the US would probably have to accept it.
Edit: Also, 1-4 are not that unrealistic, if you give it enough time. In fact, if enough time passes with no invasion (or peaceful reunification with the CCP still in power), one of them is likely to happen I would say.
It could take 30 years though, and we could be deep into some kind of singularity event before it happens. For instance, TSMC could end up building the largest AI on the planet using their own chips, and by using that become more powerful in all ways than China itself. (Possibly at the cost of human extinction).
Reunification is desired and could happen but unlikely on the basis of a regime change in China. It’s also very difficult because of the strategic position the island has for the US military in the broader US China conflict view. Looking at the map the US is surrounding China from the east like the Nato is surrounding Russia from the west.
As soon as the west is more independent from China again the conflict will enter into the next phase. Any military activity in and around Taiwan will brain drain the country very quickly and most of the talent will go to western companies.
The status quo can go on indefinitely. Taiwan gains almost nothing from actual independence. Chinese people don't seem to think of Taiwan unless it's brought up; they've already declared themselves the winner of the civil war and the whole world legally agrees. In dynastic times, China had many tributary states who acknowledged the power of China but otherwise went about their own business, and the current status quo is not far from that.
As long as the US government can afford to run with trillion dollar deficits. Once the budget cuts take place and the defense budget gets cut by half, all bets are off.
It has nothing to do with DEI. There's just a self serving political lie. Even the examples in the article make no sense, ok, some government agency is going to have a DEI officer, so what?
TSMC famously treats its workers like trash. I've seen it up close when my students would join and report back about how terrible the environment is. You should see what the leadership team has to say about how workers should behave. US workers don't want to put up with it so they're having a lot of difficulty finding people.
So the article is suggesting it comes down to this:
That Taiwainese workers will work 16 hour days and weekends.
That's true and probably won't happen in America with American workers.
But, before we just take that one factoid and call it settled, I'd love to understand what the ROI would be on a plant that just hired twice as many workers. My impression is that the margins on these chips are 100-1000x.
Did you read the article? Most of the article is about cultural differences among workers
> “They really are trying to push this narrative that Americans are slower because of lower technical ability, but I really don’t believe that’s the truth,” an American engineer who recently left TSMC told Rest of World. “The Taiwanese create this false sense of urgency with every single task, and they really push ‘you need to finish everything immediately.’ But it’s just not realistic for people that want to have some normal work-life balance.”
It sounds like much of the problem is not time spent working, but cultural friction. The American's expect a work environment similar to in American companies, while the Taiwanese (both management and others) expect Taiwanese standards, and neither side wants to back down. This will cause both sides to have something resembling racist feelings towards the other group.
And when there isn't even a shared language that all can use (and are willing to use), it doubles the problem and introduces a lot of inefficiency.
English is generally the current Lingua Franca, but TSMC employees appear to be either to proud of their language or improperly selected for language mastery to be willing or able to switch. And finding enough staff able to work using mostly Mandarin is probably even more unrealistic.
TSMC will probably overcome all of these issues, if the motivation is there. It's just that things will move a bit more slowly than expected until the culture issues settle down.
> The American's expect a work environment similar to in American companies
American culture resulted in chip-making moving from the US to Taiwan. That is why the foreigners are building this plant - they do things better than the locals. I hope the Arizona employees are expecting some level of change to happen. Chinese culture has resulted in substantially more prosperity being created than the US has managed over the last 50 years and it is obvious they are doing something better.
Although I'm not sure how much to read in to these sort of articles that are awash with confident anonymous sources. It is hard to get a grip on how much is disgruntled employee chatter and how much is serious.
> Chinese culture has resulted in substantially more prosperity being created than the US has managed over the last 50 years and it is obvious they are doing something better
Perhaps, although even with three times as many people the GDP of China is still less than that of the US. We should also ask in both cases "prosperity for whom" and "at what cost" (and to whom as that cost fallen). The median American is still wildly better off than the median Chinese resident, as can be seen from the migration flows.
Not to mention that Taiwan is not, despite all the polite fictions, China, and TSMC is anyway a unique project. You might as well ask "can we clone Morris Chang?"
The Taiwanese population is to a great degree Chinese, but not only have they enjoyed a much freer and capital friendly economy (and society in general), they also started out with much of the economic elite of China that was fleeing when Mao took over.
It seems like a larger percentage of the population in Taiwan (compared to mainland China) have the kind of productivity that has been gathering in a few mainland cities, such as Shanghai.
Younger Taiwanese people in particular also seem to be building a special non-Chinese Taiwanese identity. I'm not sure if that has any significant effect on productivity though.
> We should also ask in both cases "prosperity for whom" and "at what cost"
Prosperity for Chinese people? The stats and anecdotes are both fairly unambiguous. Chinese policy just catapulted ~800 million people out of extreme poverty. Some of my friends just got back from Ancestors' Day and were talking about how their grandfather's village had plumbing now. Shenzhen went from fishing nets to microchips in about a generation, and TSMC is only a little less than 40 years old. These people are wealth creators in the most respectable sense; the fruit of their work is spread extremely wide.
And I'm not sure exactly what we'd call the cost as being but I'm glad they paid it. I think extreme poverty is a bad thing and we'll be doing well to see the back of it. If they can do it more cheaply in Africa and India then that would be good, but the important thing is that it happens ASAP.
A lot of manufacturing started moving out of the US around 1970s due to how much higher salaries were for workers in the US at the time. Both Japan and Korea had surging economies for a while, with rapidly increasing salaries. But salaries are still far from US levels (if anything, the US have been pulling ahead).
Taiwan and China have been following a similar pattern, but even Taiwan has a long way to go to reach US productivity overall. What makes TSMC special, is that they don't have to compete for the top talent within the nation. They're kind of what Nokia used to be for Finland.
But the rest of Taiwan's companies are not doing as well. If more companies like TSMC emerge, the talent pool available will shrink. Also, most companies with little competition tend to experience entropic decay over time.
Now, it IS possible that Taiwan/China will eventually overtake the US eventually, even in terms of GDP/capita. But I would say it's equally possible that they will follow a path similar to Japan/Korea, where they reach a plateau.
And if there is something that holds these countries back, I suspect it's that the highly hierarchical social structure is not ideal for creativity in the long term, especially once people forget that their parents or grandparents were poor, and start to getting used to relative wealth.
And then there's the fact that the brightest minds from all over the world tend to be quite mobile, and tend to go to the place where they're likely to get the greatest reward for their talent. Currently that's the US, and I think it's likely to stay that way until the US has some kind of cultural collapse or a civil war.
"Gross margin for the quarter was 53.1%, operating margin was 42.0%, and net profit margin was 38.0%"
(and of course if the factory isn't operating, its margins are zero!)
No, this is an expertise problem, and I suspect that like a lot of software projects it would get worse with more people because the vast majority of the problem is addressing specific details that require the knowledge of particular individuals. It's not like you can materialize IC factory engineers from anywhere as if they were Uber drivers, either.
I think the question still holds of -- what % of operating expenses are headcount? Is it feasible to double that number.
Because frankly I don't think American engineers will (nor should) work 18 hour days unless they are getting paid I don't know, 200k->300k.
If we can't find a way to make the factory break-even meeting industrial regulations for US citizens then perhaps it needs to be staffed by H1B workers or something.
In my experience 40 hours plus 40 hours does not equal 80 hours in the office. Someone working twice as long is just able to accomplish a ton more than two people working 40 hours because theres less collaboration friction.
Why bring DEI into this? Is there any factual information that DEI is the reason why the build out failed to meet expectations?
Or is it more likely that TSMC is building a highly specialized factory and is used to working with Taiwanese contractors who already have the skills, equipment, and training? Even more likely that their estimates and schedules are all built around prior experience in Taiwan and thus those estimates and planning all failed to account for the scale of challenges in dealing with an entirely different labor market.
I see no reason why DEI is even brought into this discussion.
Commentators have noted that CHIPS and Science Act money has been sluggish. What they haven’t noticed is that it’s because the CHIPS Act is so loaded with DEI pork that it can’t move.
The law contains 19 sections aimed at helping minority groups, including one creating a Chief Diversity Officer at the National Science Foundation, and several prioritizing scientific cooperation with what it calls “minority-serving institutions.” A section called “Opportunity and Inclusion” instructs the Department of Commerce to work with minority-owned businesses and make sure chipmakers “increase the participation of economically disadvantaged individuals in the semiconductor workforce.”
The department interprets that as license to diversify. Its factsheet asserts that diversity is “critical to strengthening the U.S. semiconductor ecosystem,” adding, “Critically, this must include significant investments to create opportunities for Americans from historically underserved communities.”
Lmao how did you turn this into a political issue? Intel loves Israel and Poland because of labor costs and regulatory framework. If DEI (the largely overstated and mysterious hiring policy that mostly doesn't exist in the way you think it does) is even a feature of Intels corporate culture it would still be implemented in Israel and Poland.
I'm sorry, but this is one of the most poorly written articles I've ever read. It feels like someone had a word limit to hit. I feel like I've read the same three sentences paraphrased a million times in the article.
Let me save you some time:
- TSMC's USA plant is delayed a year (this is apparently "woefully delayed" ?)
- There is apparently a language barrier between TSMC Taiwan and TSMC USA workers.
- Repeated references to a "strict hierarchy", but no details beyond that.
This shouldn't surprise anyone here because we've seen it happen time and time again with Silicon Valley. The urban infrastructure in the Bay Area is truly awful. The cost-of-living is out-of-control. There has been many attempts to recreate SV elsewhere [1] and while there has been some success, nothing has rivalled the original.
So people are in the Bay Area because people are in the Bay Area and that's hard to transplant or reproduce because a startup in Ohio will ultimately have better access to capital, people and an entire ecosystem by relocating to SV.
A lot of Chinese (and Taiwanese) companies now benefit from a culture they've created. Additionally there are supply issues. Factories in Shenzhen or Guanghzhou or Taiwan can now get all the supplies they need from all the factories that exist in that area. There's no cultural or language barrier with suppliers and the rest of the company.
It is going to be extremely difficult and require sustained dedication to recreate chip manufacturing somewhere other than Taiwan, China, Japan or South Korea. Those places have decades of experience at this point and (now) an ecosystem advantage.
"How did we get here?" is an interesting question because Taiwan's chip dominance was largely created by US manufacturers repeatedly seeking short-term cost-cutting. Dell (for example) would start assembling PCs in Taiwan to save on labor costs. And then they'd make PCB boards to save costs. And then they'd produce the RAM and storage there. Ultimately the entire PC was made there and suddenly they've created Asus and have funded every step of its development.
The difference is the massive strategic importance of not having such a reliance on a single location that your primary rival is claiming the right to take by force at any time.
Think of fabs as having similar advantages to being distributed as do data centers. Imagine the added cost and latency if every data center on Earth was located around SF?
90 comments
[ 3.0 ms ] story [ 157 ms ] threadThe irony is that for most of the tech industry TSMC is unavailable. Nvidia, Apple and AMD have booked all the manufacturing slots for the foreseeable future. It's been this way since 2020.
Yes, their loss would have a huge effect, but many are having to cope with that situation today.
I suspect they'd be in much worse shape if Taiwan were invaded.
First, they'd find themselves competing with current TSMC customers for the non-TSMC foundries.
And second, an invasion of Taiwan could lead to a full out war with China, which might mean the non-Taiwanese fabs have their production priorities overridden by the warring governments.
What would be the casus belli?
Quite literally no country of even remote significance on the world stage (aside from the Holy See) officially recognizes Taiwan.[1] By what pretext would a country declare war against China as concerns Taiwan?
[1]: https://en.wikipedia.org/wiki/Foreign_relations_of_Taiwan#Ty...
I'm not remotely qualified to predict that. Perhaps I shouldn't have mentioned it.
Should China decide to strike that supply chain--whether by accident or on purpose--the result would be US military casualties, and that would have a strong likelihood of precipitating outright war.
Multiple US companies with trillion-dollar market caps that are completely or near-completely dependent on TSMC.
Remember that Japan attacked the US in WWII largely because we cut off their oil supply. Now, let's watch China cut off the US's supply of advanced semiconductors, and see how that goes for them. You bring the popcorn, I'll bring the beer.
> The American engineers complained of rigid, counterproductive hierarchies at the company; Taiwanese TSMC veterans described their American counterparts as lacking the kind of dedication and obedience they believe to be the foundation of their company’s world-leading success.
I mean, some humility is in order. TSMC are succeeding because of these things, not in spite of them. Given the supposed need for expedience it's best to copy that as-is and then you can tweak it once it's up and running.
If the US formally makes such a guarantee, that in itself might trigger a nuclear war.
This is a game of chicken, and games of chicken are rather dangerous. Do you want your city nuked to save Taiwan? Does Xi want to risk having Shanghai and Beijing obliterated to maybe take Taiwan?
If there is an actual war over Taiwan, it's hopefully going to be a conventional one. But that would also mean China may risk it. Which would make TSMC production out of service for months or years (or forever).
We don't know if China would, but I don't think they would start a conflict over Taiwan if they expected to only be able to "win" if they use nukes.
Rather, the only way I can see this become a nuclear war is if Xi starts an invasion due to overconfidence (like Putin with Ukraine), but is then humiliated by a combined Taiwanese/US defense, proceeds to see nukes as the only way to avoid losing face, and then caring more about this than the future of the world.
But let's imagine for instance that Russia ends up conquering Ukraine completely, with NATO letting it happen. In such a case, Japan, Korea and many others may suddenly realize that they're not safe until they have a large nuclear arsenal. Once they have build up such an arsenal, it's not that unlikely (I would say) that Taiwan follows, and creates their own secret arsenal of nukes. Maybe even purchasing the tech from Korea or Japan.
This was the conventional wisdom about Japan in the 1980s as well, and they fell into the abyss with everyone else when China was capitalized and could offer labor for 1/4 the cost. The tactics that get you there aren't necessarily the ones that keep you there.
Gross manufacturing productivity in, say, Germany may still be higher than most places in Asia.
The United States invented and has historically dominated semiconductors until the last decade or so, so yes.
The problem here is two-fold: Financial misincentives, and the fact anyone who achieves #1 in anything will always fall from grace eventually and often disasterously.
It also doesn't help that the Chinese/Taiwanese are very shrewd people compared to Americans who are... lazy, to say the least. I say that as an American.
Americans can dominate in anything they feel motivated to, the problem is getting motivated.
If anything, lower salaries in Asian countries indicates that their productivity is lower than in the US.
People in the US work relatively long hours compared to, say, Europe. They're also some of the most creative and agile out there.
I suspect that part of the reason TSMC is having such extreme success right now, is that they're by far the most successful company in Taiwan. When they're looking for people, it's like OpenAI looks for people in the US, they can get almost anyone.
Being expected to come into the office and sit there just because its expected of you seems the opposite of "work ethic" to me....
Some of these can even be contradictory. Some workplaces reward the bravery to provide constructive, honest criticism to coworkers and especially senior management. In other places, such behavior can be seen as disloyal.
But there are also cultures that actively repress people who seem overly eager to give their best. I personally know of places where the union has been going after young people who work at their full capacity as it makes the older workers look bad.
In other places, I've seen laziness being seen as completely acceptable. And particularly in groups that feel underpaid, either relative to management or to coworkers in other countries with higher salaries. ("We pretend to work, they pretend to pay us").
And moving forward, people are starting to realize the risk of having TSMC manufacture everything in Taiwan, given how little competition they have. If TSMC wouldn't start to move some of their production abroad, they would risk facing import restrictions to western countries or (as already happened) government subsidies to local fabs.
To retain their market share, they were probably forced to set up some fabs abroad.
In the end, I think it's really good for those new countries to have TSMC set up fabs in their countries, both since it guarantees that some production stays available in the case of a war, but also because the TSMC workers will gradually naturalize in the US or Europe and start to work for companies like Intel or ASML and contribute to crosspollination of ideas abilities.
Short term though, cultural conflicts is exactly what one would expect. It's probably very similar to what Western companies experienced when first setting up shops in China or India.
In the beginning Western managers would try to have the locals behave like Western employees when it came to initiative and willingness to "say it as it is" when managers were giving impossible orders.
Over time many would learn that finding really good local managers able to bridge the east/west culture gap could be really helpful.
I'm guessing TSMC will need to do the same thing. Find Western managers (for instance second generation Asians) able to understand both cultures, so they can deal with local employees the way they're used to while also able to work productively with upper management (who would presumably still be Taiwanese).
Not that this is the same class of problem, but still.
Add language barriers to solving very hard technical problems and you're going to have slow progress.
To be clear I am talking about just about language and not the cultural differences other comments are talking about.
Language has long posed a challenge for organizations in Canada with operations and customers in both the English-speaking and the French-speaking regions of the country. These days, though, that has become a comparatively minor issue to deal with.
Severely flawed immigration policies have very quickly brought in a huge number of people who don't know English well, and who also don't know French well, too. In effect, these people don't really fit in linguistically anywhere across the country.
As this third group has gotten larger, especially over the last decade, unnecessary friction has been introduced into even the most basic of interactions, for pretty much all of the participants involved.
In the major cities, for example, it's not uncommon to watch one customer after another with limited English and/or French skills have trouble interacting with a cashier who also has limited English and/or French skills.
Native English and French speakers are left in a position of constantly being unsure if they were properly understood, too.
Miscommunication, misunderstanding, and mistakes have become far more common than they were even as recently as the 1990s and early 2000s. Lots of time, money, effort, and resources are frequently wasted thanks to this.
Canada's productivity has certainly taken an unnecessary hit due to language issues.
Honestly, it is just business. The companies need to adapt and stop complaining.
The government has other levers it can pull, such as tariffs.
It also looks like the Taiwan employees are reluctant to transfer domain knowledge, based on the article. They don’t want to have to compete with U.S. workers.
Anyway, just because TSMC puts Taiwan in a position similar to Japan in the 1980's, doesn't mean they retain that advantage into the future
1) China could have a regime change. This could either lead to voluntary reunification or at least to military action becoming irrelevant.
2) Xi could die, and the successor might have other priorities.
3) China could lose the ability or confidence to invade, for instance if their economy is destroyed by corruption or they stop investing in their armed forces.
4) Taiwan could develop their own nukes in secret, and be too scary to invade.
5) The threat could become reality (and thus no longer a threat). China may proceed to either win or lose such a war. In such a scenario, it's hard to say what remains of TSMC's capacity after.
In my view it’s not a China Taiwan conflict more a US China one.
Edit: China is trying to get hold of Taiwan politically not necessarily militarily and if I’m not mistaken the US said they will not tolerate this either.
From what I've been reading, the support for reunification within the Taiwanese population has been going down for a long time. The current talk of invasion could reflect Xi giving up on that possibility.
As for what would happen if some day the majority in Taiwan actually supported a peaceful reunification, the US wouldn't like it, but it wouldn't be easy for the US to defend attempts to stop it militarily.
Now, that wouldn't completely rule out some kind of covert support for a military coup or something. But if the support for reunification is large enough, the US would probably have to accept it.
Edit: Also, 1-4 are not that unrealistic, if you give it enough time. In fact, if enough time passes with no invasion (or peaceful reunification with the CCP still in power), one of them is likely to happen I would say.
It could take 30 years though, and we could be deep into some kind of singularity event before it happens. For instance, TSMC could end up building the largest AI on the planet using their own chips, and by using that become more powerful in all ways than China itself. (Possibly at the cost of human extinction).
As soon as the west is more independent from China again the conflict will enter into the next phase. Any military activity in and around Taiwan will brain drain the country very quickly and most of the talent will go to western companies.
30 years is a hell lot of time these days.
That’s why CHIPS was passed, the brass told Congress to get them some fancy new fabs, and like all requests the Pentagon makes, they got it.
Long term it doesn’t matter if these fabs are hugely profitable relative to other locales, we’ll find a way to keep them producing.
There’s only one thing the agree on in DC: defense spending is the most important line item, and everything else is secondary.
Also, turns out when you have the ability to take virtually anything by force the concept of “deficits” is a bit fungible.
they are even queuing themselves up in distinct routes based on race and national security concerns.
this is what real politik looks like in modernity.
TSMC famously treats its workers like trash. I've seen it up close when my students would join and report back about how terrible the environment is. You should see what the leadership team has to say about how workers should behave. US workers don't want to put up with it so they're having a lot of difficulty finding people.
That Taiwainese workers will work 16 hour days and weekends.
That's true and probably won't happen in America with American workers.
But, before we just take that one factoid and call it settled, I'd love to understand what the ROI would be on a plant that just hired twice as many workers. My impression is that the margins on these chips are 100-1000x.
> “They really are trying to push this narrative that Americans are slower because of lower technical ability, but I really don’t believe that’s the truth,” an American engineer who recently left TSMC told Rest of World. “The Taiwanese create this false sense of urgency with every single task, and they really push ‘you need to finish everything immediately.’ But it’s just not realistic for people that want to have some normal work-life balance.”
And when there isn't even a shared language that all can use (and are willing to use), it doubles the problem and introduces a lot of inefficiency.
English is generally the current Lingua Franca, but TSMC employees appear to be either to proud of their language or improperly selected for language mastery to be willing or able to switch. And finding enough staff able to work using mostly Mandarin is probably even more unrealistic.
TSMC will probably overcome all of these issues, if the motivation is there. It's just that things will move a bit more slowly than expected until the culture issues settle down.
American culture resulted in chip-making moving from the US to Taiwan. That is why the foreigners are building this plant - they do things better than the locals. I hope the Arizona employees are expecting some level of change to happen. Chinese culture has resulted in substantially more prosperity being created than the US has managed over the last 50 years and it is obvious they are doing something better.
Although I'm not sure how much to read in to these sort of articles that are awash with confident anonymous sources. It is hard to get a grip on how much is disgruntled employee chatter and how much is serious.
Perhaps, although even with three times as many people the GDP of China is still less than that of the US. We should also ask in both cases "prosperity for whom" and "at what cost" (and to whom as that cost fallen). The median American is still wildly better off than the median Chinese resident, as can be seen from the migration flows.
Not to mention that Taiwan is not, despite all the polite fictions, China, and TSMC is anyway a unique project. You might as well ask "can we clone Morris Chang?"
It seems like a larger percentage of the population in Taiwan (compared to mainland China) have the kind of productivity that has been gathering in a few mainland cities, such as Shanghai.
Younger Taiwanese people in particular also seem to be building a special non-Chinese Taiwanese identity. I'm not sure if that has any significant effect on productivity though.
Prosperity for Chinese people? The stats and anecdotes are both fairly unambiguous. Chinese policy just catapulted ~800 million people out of extreme poverty. Some of my friends just got back from Ancestors' Day and were talking about how their grandfather's village had plumbing now. Shenzhen went from fishing nets to microchips in about a generation, and TSMC is only a little less than 40 years old. These people are wealth creators in the most respectable sense; the fruit of their work is spread extremely wide.
And I'm not sure exactly what we'd call the cost as being but I'm glad they paid it. I think extreme poverty is a bad thing and we'll be doing well to see the back of it. If they can do it more cheaply in Africa and India then that would be good, but the important thing is that it happens ASAP.
Taiwan and China have been following a similar pattern, but even Taiwan has a long way to go to reach US productivity overall. What makes TSMC special, is that they don't have to compete for the top talent within the nation. They're kind of what Nokia used to be for Finland.
But the rest of Taiwan's companies are not doing as well. If more companies like TSMC emerge, the talent pool available will shrink. Also, most companies with little competition tend to experience entropic decay over time.
Now, it IS possible that Taiwan/China will eventually overtake the US eventually, even in terms of GDP/capita. But I would say it's equally possible that they will follow a path similar to Japan/Korea, where they reach a plateau.
And if there is something that holds these countries back, I suspect it's that the highly hierarchical social structure is not ideal for creativity in the long term, especially once people forget that their parents or grandparents were poor, and start to getting used to relative wealth.
And then there's the fact that the brightest minds from all over the world tend to be quite mobile, and tend to go to the place where they're likely to get the greatest reward for their talent. Currently that's the US, and I think it's likely to stay that way until the US has some kind of cultural collapse or a civil war.
That would be false; you can read the margins off the report: https://pr.tsmc.com/english/news/3131
"Gross margin for the quarter was 53.1%, operating margin was 42.0%, and net profit margin was 38.0%"
(and of course if the factory isn't operating, its margins are zero!)
No, this is an expertise problem, and I suspect that like a lot of software projects it would get worse with more people because the vast majority of the problem is addressing specific details that require the knowledge of particular individuals. It's not like you can materialize IC factory engineers from anywhere as if they were Uber drivers, either.
Because frankly I don't think American engineers will (nor should) work 18 hour days unless they are getting paid I don't know, 200k->300k.
If we can't find a way to make the factory break-even meeting industrial regulations for US citizens then perhaps it needs to be staffed by H1B workers or something.
Or is it more likely that TSMC is building a highly specialized factory and is used to working with Taiwanese contractors who already have the skills, equipment, and training? Even more likely that their estimates and schedules are all built around prior experience in Taiwan and thus those estimates and planning all failed to account for the scale of challenges in dealing with an entirely different labor market.
I see no reason why DEI is even brought into this discussion.
The law contains 19 sections aimed at helping minority groups, including one creating a Chief Diversity Officer at the National Science Foundation, and several prioritizing scientific cooperation with what it calls “minority-serving institutions.” A section called “Opportunity and Inclusion” instructs the Department of Commerce to work with minority-owned businesses and make sure chipmakers “increase the participation of economically disadvantaged individuals in the semiconductor workforce.”
The department interprets that as license to diversify. Its factsheet asserts that diversity is “critical to strengthening the U.S. semiconductor ecosystem,” adding, “Critically, this must include significant investments to create opportunities for Americans from historically underserved communities.”
Let me save you some time:
- TSMC's USA plant is delayed a year (this is apparently "woefully delayed" ?)
- There is apparently a language barrier between TSMC Taiwan and TSMC USA workers.
- Repeated references to a "strict hierarchy", but no details beyond that.
What a stinker of an article.
So people are in the Bay Area because people are in the Bay Area and that's hard to transplant or reproduce because a startup in Ohio will ultimately have better access to capital, people and an entire ecosystem by relocating to SV.
A lot of Chinese (and Taiwanese) companies now benefit from a culture they've created. Additionally there are supply issues. Factories in Shenzhen or Guanghzhou or Taiwan can now get all the supplies they need from all the factories that exist in that area. There's no cultural or language barrier with suppliers and the rest of the company.
It is going to be extremely difficult and require sustained dedication to recreate chip manufacturing somewhere other than Taiwan, China, Japan or South Korea. Those places have decades of experience at this point and (now) an ecosystem advantage.
"How did we get here?" is an interesting question because Taiwan's chip dominance was largely created by US manufacturers repeatedly seeking short-term cost-cutting. Dell (for example) would start assembling PCs in Taiwan to save on labor costs. And then they'd make PCB boards to save costs. And then they'd produce the RAM and storage there. Ultimately the entire PC was made there and suddenly they've created Asus and have funded every step of its development.
[1]: https://en.wikipedia.org/wiki/List_of_technology_centers#Pla...
Think of fabs as having similar advantages to being distributed as do data centers. Imagine the added cost and latency if every data center on Earth was located around SF?
Another miscalculation: undervaluing the ability to speak the same language.
Any organization except the U.N. should have a single working language and require everyone to use it exclusively.
Hell, I even think that English proficiency should be a prerequisite for immigration to the U.S., or at least for citizenship.