Finally. I was reading about someone that said Americans can't compete in high tech chip manufacturing because of work culture and it just felt like a ridiculous way to justify poor worker rights. Because ASML tech is just as cutting edge and you sure as hell know that Dutch workers will not accept TMSC standards.
I would say ASML is even more cutting edge in terms of technology than TSMC. It's higher up in the value chain, and so far nobody comes even somewhat close to ASML. And that's even with China having the deepest high-tech manufacturing engineering bench of any nation stealing ASML's tech and trying to copy it fully and not succeeding.
It's possibly the highest-tech manufacturing thing out there in the world nowadays, and the work culture is extremely dutch (very chill). Having said that, most of the engineers and researchers are expats. Although at the engineering-research management level you find ex dutch university professors and all other functions of the company are also mostly all dutch people.
High NA EUV lithography machines are the single most complex piece of technology that the human race is capable of making. Nothing, whether jet engines, space rockets or even particles colliders require that level of precision. It took decades of research and billions of dollars plus assistance from the big American chip makers to make it possible.
China has lots of money and engineering tech but they have a lot more obstacles to pass. Instead of being able to rely on a complex network of global suppliers, they have to manufacture every piece of the supply chain while also looking for a method that doesn't infringe on ASML's patents. Plus they're decades late.
The Chinese are determined,.but they aren't magicians. If they get the tech before 2030 then it'll be truly impressive.
Also, ASML does rely on migrants a bit though. Remember when they threatened to leave earlier this year after the Dutch government tried to restrict the inflow of migrants. Of course, it's all very high skilled people.
The Dutch government is stopping immigration/refugees from MENA. There are already talks behind doors about high-skill immigrant exceptions for ASML and others.
I live in SE Asia and have a LOT of experience with (allegedly) Confucian style business management. Somehow these pieces never laud the management in Vietnam. Nor do they ever talk about how amazing Singaporean companies are. Or Malaysian. Or Thai or Filipino or Indonesian where the Chinese minority still makes a huge amount of the business culture.
I'm reminded of something Noah linked a while ago from Ha-Joon Chang's book "Bad Samaritans: Rich Nations, Poor Policies & the Threat to the Developing World" about how we think certain countries have innate characteristics but it is largely a reflection of the country's economy. Japanese are lazy, sloppy, and undisciplined. Germans were doomed to perpetual poverty because they lacked enterprise. Koreans were dirty, sullen, and lazy. Poles had, within living memory, an entire genre of jokes dedicated to how lazy and dumb they were.
The overwhelming consensus of everyone I've worked with in Vietnam -- both Vietnamese and non-Vietnamese -- is that the (allegedly) Confucian-style Vietnamese business management is garbage and everyone except the managers hate it. Most Vietnamese try to get jobs with Western managed companies if they can possibly manage it.
The entire piece looks like survivorship bias mixed with Stockholm Syndrome. Business books (crack open Good to Great, for instance) are full of brilliant companies that thought they had unique insights into how to run a business only to eventually become forgotten dinosaurs.
Your second paragraph is eye-opening. The difference between reputations of shiftlessness and industriousness changes due to trade. The last people I would've thought of to be poor and unentrepreneurial were the Germans. Raised by a Hungarian mother and grandmother, I absorbed a sense of envy they had toward Germans and Austrians who seemed to have everything going their way.
I also learned that part of the Mussolini braggadocio of "making the trains run on time" was rooted in European racism. In this case, it's the northern, cold-weather Europeans -- seeing themselves as rational, stoic and industrious -- dismissing Mediterranean Europeans like Italians as shiftless, carefree olive eaters. It was the fear of fascist violence that motivated a work ethic into the railroad workers. (Narrator: Italy's railways were in a state of decrepitude due to age, materials shortages and the economic tribulations between the world wars, hence why they didn't run on time.)
In fact Mussolini's austerity measures made Italian trains more late than they used to be before.
Perhaps the best evidence comes from 1950s TV host/proto-mythbuster Bergen Evans, who assailed the “myth of fascist efficiency” in his 1954 book, The Spoor of Spooks, and Other Nonsense, based in part on his own first-hand experience.
The author was employed as a courier by the Franco-Belgique Tours Company in the summer of 1930, the height of Mussolini's heyday, when a fascist guard rode on every train, and is willing to make an affidavit to the effect that most Italian trains on which he travelled were not on schedule—or near it. There must be thousands who can support this attestation. It's a trifle, but it's worth nailing down
Generally agreed here, although I have come across a reasonable amount of very positive press for especially Singaporean companies (e.g. ST Engineering and Chartered Semiconductor, the latter now a big chunk of Global Foundries) and also Malaysian companies, although the Malaysian examples were somewhat lower on the value add food chain.
It's been a while so I have forgotten whether the Singaporean articles subtly favored the overseas Chinese strain in Singaporean business culture or not. Looking just now, four out of five of ST Engineering's executive council are ethnic Chinese.
From an on the ground perspective, I'm part of the engineering management of a US company in a mid-tier industrial (definitely not consumer!) tech sector. Our observation trying to move some of our somewhat specialized manufacturing from China (where we have a captive subsidiary) to Malaysia is that the major problem has been not any complaints about the Malaysian companies as such, but rather lack of depth in the supply chain: our manufacturing process involves a fair number of custom to us "robots" - automated manufacturing cells - and the supply chain for those devices, e.g. the robot controllers, is almost entirely in China, so when something goes wrong we have to fly someone from China.
As you can imagine, during Covid that was quite a problem - we had one of our south China employees stuck in Penang for something like six months because he could not get back with the border controls.
This has actually been a common complaint of workers in Japan, the shift towards management by numbers (=disconnect of decisions from the factory floor) in the last 25 years.
Hard! I do think consultancies have had some good success stories. Sometimes it's good to have an outside abstract view. But yes, the mission is short term profit for the shareholder, abstracted away from the productive entity and the people that make it work, and certainly disconnected from any long-term gains - there, I've put the "make it away in the night rich" in a less beautiful and more convoluted way.
God yes. A decade in Forbes 100 corporate America has dramatically reduced my respect for America's executive class, with the massive waste of money and often active harm to the business bringing on mgmt consultants being a significant contributor.
I experienced a weird reversed deja vu while reading this article - it reminds me of the clashes I know about, the ones between European work culture and the American one. In those, the US side perceives things like 5 weeks of holidays and unwillingness to take work calls after work hours as lazy while the European side perceives the same things that Americans are proud of as toxic and unproductive.
It is interesting to see Americans being on the work-life balance side for once. Just a personal observation - not saying that any culture is better. The lesson I take from this interesting article is that it is more complicated than just "my culture good, their culture bad".
As someone who worked at Intel's foundry for a few years, this is absolutely correct. I'm compiling a list of the best writing on semiconductor manufacturing, and this is definitely going on it.
Also, the idea of "learning" TSMC's best practices isn't necessarily about operational efficiency. In fact, Intel under Gelsinger has adopted many of these practices as it re-oriented itself. The thing it can't replicate, though, is trust. I wrote in detail about that here: https://roblh.substack.com/p/the-technoeconomic-pillars-of-foundry
The short version is that, as an IDM, Intel's product teams could compete with their foundry customers at any time (and may already). I was disappointed that the CHIPs Act had no appetite for trying to make a greenfield, US-based, pure play foundry that could compete on a truly equal footing. This probably isn't the last round, though, and the article demonstrated clearly that American talent wants to work hard, wants to be even more productive than what TSMC is doing, and would benefit from a much lower cost-of-living (where I'm increasingly persuaded by the housing theory of everything driving these issues).
If you talk to them again, the issue I saw was always that the US was willing to put in money but not enough to pay for it outright or entice private investment. I had one pitch but our key investors walked as soon as the Feds started talking total amounts. It was always going to the existing players.
Speaking as a customer of foundry customers (i.e. my company buys product from some of TSMC's customers; since we're sometimes on the early edge we see a fair number of details of the foundry selection and schedule management, not to mention the device company's analysis of the costs of trying to have a second foundry on a cutting edge product, where the process stack is never quite the same between foundries):
not at the bleeding edge of fabrication, I don't think, pending major structural changes in the foundry industry: it's not clear that there is room in the global market for more than TSMC, Samsung, and Intel. The capital costs involved are, to say the least, frightening.
What I don't know and people like us are unlikely to see (IMHO) are the more detailed financial with life cycle ROI on process nodes. E.g. it's easy to see that TSMC has an estimated 61% of the worldwide foundry market (4Q23), but how that revenue side (to say nothing of some sort of attributable gross margin) is split between nodes is a different question. I happen to know :) that TSMC is still running wafers on some truly ancient processes, even though TSMC is (appropriately) jacking prices on those processes, because it's still cheaper for TSMC's customers than migrating the devices.
Global Foundries (generally #3 or #4 in the foundry market, trading places with Taiwan-based UMC), for example, made a decision to stop chasing the bleeding edge of geometries and concentrate on more niche process stacks.
I wonder also to what extent TSMC's perceived "management" advantage is more just that they were able to leapfrog Intel after their big bet foundry R&D stumbled, and now Intel is struggling to catch up because TSMC hasn't had a similar ball-drop moment yet.
any chance you would be willing to share this list of "best writings on semiconductor manufacturing"? I'm doing a similar thing by bookmarking them on Zotero.
In my experience having spent some of my earlier career working in mainland China and East Asia more broadly - I think East Asian Management culture can be really good in the context of manufacturing and industrial applications where consistency, safety, and small incremental optimizations matter the most.
When it gets applied to fields that require rapid innovation and iteration things seem to fall apart. There is a lot of performative productivity (extremely long hours in the office, many measurable tasks accomplished) but at the end of the day theres a reason that the Japanese still rely on fax machines. Oh and lots of mandatory drinking.
Have worked and lived in Taiwan and hired people who worked for TSMC. Their stories confirm for me it is a harder working institution than any US companies I’ve heard of (even ones that really pride themselves on working people to death). It’s not East Asian management, it’s TSMC. There are cultural distinctions that are true across the different companies I worked with in Taiwan and counterparts in the US, but those trends have more to do with “style” of business conduct and didn’t indicate a more productive and successful business from those features alone.
I believe those driven personalities exist in the US. During the 2000’s they went into finance and entertainment, and away from medicine and technology.
I remember encountering the Toyota Production System in the early 1980s working on automotive industry policy for Australia's new industry minister and thinking it was an extraordinary development. Economists denied it for a while and then misunderstood as stricter inventory control. It's much more than that. As Deming joked "The Americans come and note everything down. Then they return to American and try to copy. But they don't know what to copy!"
Anyway, have a look at this graph. It complements your graph beautifully covering the 25 years before your chart shows that the Americans were learning from Japan. Look at how long it took for denial to be replaced by the attempt to learn. And then the really dramatic thing is that the big American car companies started learning and their productivity started growing strongly. But their predatory purchasing practices left their suppliers with no incentive to invest more in their own productivity — because it would be stolen from them by their customers. One aspect of Toyota's practice was to share the dividends of supplier productivity gains with their suppliers!
I remember Noah getting into a back-and-forth with Lee Harris about whether TSMC should be allowed to import labor from Taiwan for the fab in Arizona (he was for, she was against). Granted that "Americans can learn", I hope he's not backing away from his original position, because he's still right.
If there's a national-security argument for moving as much chip production as possible away from a potential war zone, you'll want to accelerate that by moving actual people from Taiwan to the US, not just training Americans. It sounds as if engineers immigrating to the US might easily get snapped up by social-media companies paying higher salaries (you can see why China is suppressing this sector) but that could be addressed by putting conditions on people's work visas.
Another plus would be that these Taiwanese engineers that come to the USA will have kids here and whom are more likelier to be smart and productive compared to your average american.
Sounds like typical toxic management that wants you to sell your soul to the company for pennies. You work hard so that they get to line their pockets with the profit you generated.
These problems occurred with the Japanese and Koreans. It was solved by the development of an American workforce that was bilingual and culturally competent in both countries. This took like a decade to put together and was done by basically sheer bloody mindedness. The big corporations deciding they were going to do this and if that meant just grinding through years of failure then they would do it because it ultimately guaranteed access to the American market. But once it was done it was done. And it was available to other Japanese/Korean companies (since they could poach people). Which is why Samsung didn't have this issue. They had their own people and, if they didn't, they could hire Korean-familiar American engineer.
The issue is that China doesn't have this cross-cultural workforce. There are Chinese-Americans, of course, but Asian business experience is not genetic. There's no large workforce of Americans who've worked in Chinese companies, are familiar with Chinese culture, speak Mandarin incidentally to being an industrial engineer or something, etc. But for the same reason this happened in the 1980s it will happen in the US: rising trade barriers, economic nationalism, and problems at home mean that an American workforce is the best way to guarantee access to the American market.
So this is a nothing new under the sun story. The interesting difference is the geopolitical differences if mainland Chinese companies decide it's in their interests to jump in too. But a lot of that depends on if and how they do since those factors aren't too relevant to Taiwan.
Yes? I don't see how this would mean there's a significant workforce of Americans who have language and corporate cultural skills. The Koreans and Japanese have high five to low six figures of these people. Samsung alone has like 60k American employees and by the time you're in the mid to upper layers they're regularly reporting to Korean superiors and moving up to international operations that involve Koreans.
It's a commercial bus plant that has been there before the whole Chinese EV's being a threat was big. I meant am actual passenger car plant that'll produce 200k cars a year.
The transit bus plant is there because of Buy America rules on components and assembly. The FTA by setting standards for transit bus design, is effectively a monopsony buyer. Transit agencies, in order to receive grants to buy new buses, must adhere to them.
What's unusual about the BYD approach is that the buses have a universal design throughout the world. Basically, buses are a closed market. By the end of this year, BYD will be only one of three full-size bus manufacturers in the U.S., its only other competitors are Canada's New Flyer and U.S.'s Gillig, who basically have the transit bus market to themselves. The U.S. doesn't import or export transit buses.
A union, frustrated by its efforts to organize workers at the plant, did the time-honored tradition of California NIMBYism and lodged a CEQA lawsuit to stymie the plant.
They certainly have the option. There will no doubt be political issues. And my point is that fear of the Japanese as a geopolitical threat was imaginary while fear of the Chinese has more basis in fact so it might play out differently. But tariffs structurally encourage such investments in the US.
Agreed. Completely. But back in the 1980s Japan really was Japan. I was at NEC's Central Research Lab doing AI for 2 years (86-87), and during that time, one of the Japanese business magazines had a special edition with a big-print cover title to the effect "There Will Be A Forreigner Working Next To You. Soon". And the joke in my group was "Next to you, yes. Working, no.". (But a few years later I ran into the guys and they reported that they were still using the pattern matcher I had written while I was there: I had killed time by implementing Carl Hewitt's pattern matcher he described in his PhD thesis.) They did meetings during the day and the actual work overtime*: I was out the door at 5:00 every day. But I largely didn't go to the meetings.
*: Which meant that promotion to a management position was a financial disaster, since they depended on the overtime pay, and management doesn't get overtime...
I wonder what other cultural distinctions exist besides "saving face." I think much has been made of how Asian cultures do this, without acknowledging how much this happens in the US as well. (Remember Pence's Christian complaints about Trump's vulgarity and criminality? Me neither.) Few people can climb corporate ladders pointing out managements' flaws. Mostly this happens to people who constantly cover those up. That's many employees' primary function.
Furthermore, while we have a free press that is incentivized to criticize leadership, there are very real consequences to doing so. In Hollywood, writers and talk show hosts who tear down celebrities are not given interviews, and without such interviews there's no audience and no success. Much of what happens instead ends up being sycophantic, and then we're all "shocked" years later to discover these celebrities were suffering mental health crises, drug addictions, sexual impropriety, workplace harassment, etc.
Same goes with whistleblowing within industries -- even the media. I will be curious to see what becomes of Uri Berliner. It may be that he will find a news organization that wants to present itself as a counterpoint to NPR, like Fox News, and make a splash out of hiring him. But many will also be cautious of hiring someone who could turn against them.
And this will undoubtedly impact his social circle. People will be afraid of seeming like they are siding with him against everyone else who works at NPR and garnering the same stink that anyone who criticizes their stated politics does.
While the blacklisting of the 50s in America was official and reinforced by laws, it was also a manifestation of social behavior that does this more discreetly. It is often this subtler form of banishment that drives people's anxiety about "Woke" politics today. Any analysis, challenge or discussion of the Left's ideals raises everyone's suspicions and they race to castigate the speaker and outcast them, as if it means the person speaking or writing is a double-agent harboring patriarchal, white-supremacist agendas. The behavior of congresspeople like Marjorie Taylor Green going after Speaker Mike Johnson suggests this happens on the Right as well.
This is not indicative of a society that celebrates discussion or the challenging of authority or social norms. This is rigidly upholding ideologies. It's just the US is big enough for there to be two, major ideological poles and voting is private (at this point), so extremists still tend to be minorities.
Finally. I was reading about someone that said Americans can't compete in high tech chip manufacturing because of work culture and it just felt like a ridiculous way to justify poor worker rights. Because ASML tech is just as cutting edge and you sure as hell know that Dutch workers will not accept TMSC standards.
I would say ASML is even more cutting edge in terms of technology than TSMC. It's higher up in the value chain, and so far nobody comes even somewhat close to ASML. And that's even with China having the deepest high-tech manufacturing engineering bench of any nation stealing ASML's tech and trying to copy it fully and not succeeding.
It's possibly the highest-tech manufacturing thing out there in the world nowadays, and the work culture is extremely dutch (very chill). Having said that, most of the engineers and researchers are expats. Although at the engineering-research management level you find ex dutch university professors and all other functions of the company are also mostly all dutch people.
High NA EUV lithography machines are the single most complex piece of technology that the human race is capable of making. Nothing, whether jet engines, space rockets or even particles colliders require that level of precision. It took decades of research and billions of dollars plus assistance from the big American chip makers to make it possible.
China has lots of money and engineering tech but they have a lot more obstacles to pass. Instead of being able to rely on a complex network of global suppliers, they have to manufacture every piece of the supply chain while also looking for a method that doesn't infringe on ASML's patents. Plus they're decades late.
The Chinese are determined,.but they aren't magicians. If they get the tech before 2030 then it'll be truly impressive.
Also, ASML does rely on migrants a bit though. Remember when they threatened to leave earlier this year after the Dutch government tried to restrict the inflow of migrants. Of course, it's all very high skilled people.
The Dutch government is stopping immigration/refugees from MENA. There are already talks behind doors about high-skill immigrant exceptions for ASML and others.
I live in SE Asia and have a LOT of experience with (allegedly) Confucian style business management. Somehow these pieces never laud the management in Vietnam. Nor do they ever talk about how amazing Singaporean companies are. Or Malaysian. Or Thai or Filipino or Indonesian where the Chinese minority still makes a huge amount of the business culture.
I'm reminded of something Noah linked a while ago from Ha-Joon Chang's book "Bad Samaritans: Rich Nations, Poor Policies & the Threat to the Developing World" about how we think certain countries have innate characteristics but it is largely a reflection of the country's economy. Japanese are lazy, sloppy, and undisciplined. Germans were doomed to perpetual poverty because they lacked enterprise. Koreans were dirty, sullen, and lazy. Poles had, within living memory, an entire genre of jokes dedicated to how lazy and dumb they were.
https://www.dougsaunders.net/2007/10/lazy-germans-japanese-koreans-ha-joon-chang
The overwhelming consensus of everyone I've worked with in Vietnam -- both Vietnamese and non-Vietnamese -- is that the (allegedly) Confucian-style Vietnamese business management is garbage and everyone except the managers hate it. Most Vietnamese try to get jobs with Western managed companies if they can possibly manage it.
The entire piece looks like survivorship bias mixed with Stockholm Syndrome. Business books (crack open Good to Great, for instance) are full of brilliant companies that thought they had unique insights into how to run a business only to eventually become forgotten dinosaurs.
Your second paragraph is eye-opening. The difference between reputations of shiftlessness and industriousness changes due to trade. The last people I would've thought of to be poor and unentrepreneurial were the Germans. Raised by a Hungarian mother and grandmother, I absorbed a sense of envy they had toward Germans and Austrians who seemed to have everything going their way.
I also learned that part of the Mussolini braggadocio of "making the trains run on time" was rooted in European racism. In this case, it's the northern, cold-weather Europeans -- seeing themselves as rational, stoic and industrious -- dismissing Mediterranean Europeans like Italians as shiftless, carefree olive eaters. It was the fear of fascist violence that motivated a work ethic into the railroad workers. (Narrator: Italy's railways were in a state of decrepitude due to age, materials shortages and the economic tribulations between the world wars, hence why they didn't run on time.)
In fact Mussolini's austerity measures made Italian trains more late than they used to be before.
Perhaps the best evidence comes from 1950s TV host/proto-mythbuster Bergen Evans, who assailed the “myth of fascist efficiency” in his 1954 book, The Spoor of Spooks, and Other Nonsense, based in part on his own first-hand experience.
The author was employed as a courier by the Franco-Belgique Tours Company in the summer of 1930, the height of Mussolini's heyday, when a fascist guard rode on every train, and is willing to make an affidavit to the effect that most Italian trains on which he travelled were not on schedule—or near it. There must be thousands who can support this attestation. It's a trifle, but it's worth nailing down
https://www.bloomberg.com/news/articles/2016-11-15/stop-saying-mussolini-made-the-trains-run-on-time
Generally agreed here, although I have come across a reasonable amount of very positive press for especially Singaporean companies (e.g. ST Engineering and Chartered Semiconductor, the latter now a big chunk of Global Foundries) and also Malaysian companies, although the Malaysian examples were somewhat lower on the value add food chain.
It's been a while so I have forgotten whether the Singaporean articles subtly favored the overseas Chinese strain in Singaporean business culture or not. Looking just now, four out of five of ST Engineering's executive council are ethnic Chinese.
From an on the ground perspective, I'm part of the engineering management of a US company in a mid-tier industrial (definitely not consumer!) tech sector. Our observation trying to move some of our somewhat specialized manufacturing from China (where we have a captive subsidiary) to Malaysia is that the major problem has been not any complaints about the Malaysian companies as such, but rather lack of depth in the supply chain: our manufacturing process involves a fair number of custom to us "robots" - automated manufacturing cells - and the supply chain for those devices, e.g. the robot controllers, is almost entirely in China, so when something goes wrong we have to fly someone from China.
As you can imagine, during Covid that was quite a problem - we had one of our south China employees stuck in Penang for something like six months because he could not get back with the border controls.
It’s when East Asian management gets enamoured with McKinsey style MBA-drone management by numbers and offshoring that you get the truly toxic messes.
This has actually been a common complaint of workers in Japan, the shift towards management by numbers (=disconnect of decisions from the factory floor) in the last 25 years.
It destroys every field it touches, but its perpetrators make it away in the night rich.
Hard! I do think consultancies have had some good success stories. Sometimes it's good to have an outside abstract view. But yes, the mission is short term profit for the shareholder, abstracted away from the productive entity and the people that make it work, and certainly disconnected from any long-term gains - there, I've put the "make it away in the night rich" in a less beautiful and more convoluted way.
God yes. A decade in Forbes 100 corporate America has dramatically reduced my respect for America's executive class, with the massive waste of money and often active harm to the business bringing on mgmt consultants being a significant contributor.
I experienced a weird reversed deja vu while reading this article - it reminds me of the clashes I know about, the ones between European work culture and the American one. In those, the US side perceives things like 5 weeks of holidays and unwillingness to take work calls after work hours as lazy while the European side perceives the same things that Americans are proud of as toxic and unproductive.
It is interesting to see Americans being on the work-life balance side for once. Just a personal observation - not saying that any culture is better. The lesson I take from this interesting article is that it is more complicated than just "my culture good, their culture bad".
As someone who worked at Intel's foundry for a few years, this is absolutely correct. I'm compiling a list of the best writing on semiconductor manufacturing, and this is definitely going on it.
Also, the idea of "learning" TSMC's best practices isn't necessarily about operational efficiency. In fact, Intel under Gelsinger has adopted many of these practices as it re-oriented itself. The thing it can't replicate, though, is trust. I wrote in detail about that here: https://roblh.substack.com/p/the-technoeconomic-pillars-of-foundry
The short version is that, as an IDM, Intel's product teams could compete with their foundry customers at any time (and may already). I was disappointed that the CHIPs Act had no appetite for trying to make a greenfield, US-based, pure play foundry that could compete on a truly equal footing. This probably isn't the last round, though, and the article demonstrated clearly that American talent wants to work hard, wants to be even more productive than what TSMC is doing, and would benefit from a much lower cost-of-living (where I'm increasingly persuaded by the housing theory of everything driving these issues).
That's awesome, thanks.
As for the CHIPS people supporting a pure-play foundry, I know for a fact that they are thinking very hard about this.
If you talk to them again, the issue I saw was always that the US was willing to put in money but not enough to pay for it outright or entice private investment. I had one pitch but our key investors walked as soon as the Feds started talking total amounts. It was always going to the existing players.
Is there really any realistic option for this other than Intel?
Speaking as a customer of foundry customers (i.e. my company buys product from some of TSMC's customers; since we're sometimes on the early edge we see a fair number of details of the foundry selection and schedule management, not to mention the device company's analysis of the costs of trying to have a second foundry on a cutting edge product, where the process stack is never quite the same between foundries):
not at the bleeding edge of fabrication, I don't think, pending major structural changes in the foundry industry: it's not clear that there is room in the global market for more than TSMC, Samsung, and Intel. The capital costs involved are, to say the least, frightening.
What I don't know and people like us are unlikely to see (IMHO) are the more detailed financial with life cycle ROI on process nodes. E.g. it's easy to see that TSMC has an estimated 61% of the worldwide foundry market (4Q23), but how that revenue side (to say nothing of some sort of attributable gross margin) is split between nodes is a different question. I happen to know :) that TSMC is still running wafers on some truly ancient processes, even though TSMC is (appropriately) jacking prices on those processes, because it's still cheaper for TSMC's customers than migrating the devices.
Global Foundries (generally #3 or #4 in the foundry market, trading places with Taiwan-based UMC), for example, made a decision to stop chasing the bleeding edge of geometries and concentrate on more niche process stacks.
I wonder also to what extent TSMC's perceived "management" advantage is more just that they were able to leapfrog Intel after their big bet foundry R&D stumbled, and now Intel is struggling to catch up because TSMC hasn't had a similar ball-drop moment yet.
any chance you would be willing to share this list of "best writings on semiconductor manufacturing"? I'm doing a similar thing by bookmarking them on Zotero.
In my experience having spent some of my earlier career working in mainland China and East Asia more broadly - I think East Asian Management culture can be really good in the context of manufacturing and industrial applications where consistency, safety, and small incremental optimizations matter the most.
When it gets applied to fields that require rapid innovation and iteration things seem to fall apart. There is a lot of performative productivity (extremely long hours in the office, many measurable tasks accomplished) but at the end of the day theres a reason that the Japanese still rely on fax machines. Oh and lots of mandatory drinking.
Just my take from somewhat limited experience.
Have worked and lived in Taiwan and hired people who worked for TSMC. Their stories confirm for me it is a harder working institution than any US companies I’ve heard of (even ones that really pride themselves on working people to death). It’s not East Asian management, it’s TSMC. There are cultural distinctions that are true across the different companies I worked with in Taiwan and counterparts in the US, but those trends have more to do with “style” of business conduct and didn’t indicate a more productive and successful business from those features alone.
Just re-subscribed after reading this. The clarity!
Thanks!!
I believe those driven personalities exist in the US. During the 2000’s they went into finance and entertainment, and away from medicine and technology.
A disaster
Thanks for this Noah
I remember encountering the Toyota Production System in the early 1980s working on automotive industry policy for Australia's new industry minister and thinking it was an extraordinary development. Economists denied it for a while and then misunderstood as stricter inventory control. It's much more than that. As Deming joked "The Americans come and note everything down. Then they return to American and try to copy. But they don't know what to copy!"
Anyway, have a look at this graph. It complements your graph beautifully covering the 25 years before your chart shows that the Americans were learning from Japan. Look at how long it took for denial to be replaced by the attempt to learn. And then the really dramatic thing is that the big American car companies started learning and their productivity started growing strongly. But their predatory purchasing practices left their suppliers with no incentive to invest more in their own productivity — because it would be stolen from them by their customers. One aspect of Toyota's practice was to share the dividends of supplier productivity gains with their suppliers!
The chart is here
https://clubtroppo.com.au/wp-content/uploads/2020/05/Screen-Shot-2022-06-30-at-4.52.46-pm.png
Thanks!
I remember Noah getting into a back-and-forth with Lee Harris about whether TSMC should be allowed to import labor from Taiwan for the fab in Arizona (he was for, she was against). Granted that "Americans can learn", I hope he's not backing away from his original position, because he's still right.
If there's a national-security argument for moving as much chip production as possible away from a potential war zone, you'll want to accelerate that by moving actual people from Taiwan to the US, not just training Americans. It sounds as if engineers immigrating to the US might easily get snapped up by social-media companies paying higher salaries (you can see why China is suppressing this sector) but that could be addressed by putting conditions on people's work visas.
Another plus would be that these Taiwanese engineers that come to the USA will have kids here and whom are more likelier to be smart and productive compared to your average american.
Sounds like typical toxic management that wants you to sell your soul to the company for pennies. You work hard so that they get to line their pockets with the profit you generated.
"And the man in the suit has just bought a new car/ From the profit he's made on your dreams"
Was thinking more 16 tons, but it amounts to the same. Another day older...
These problems occurred with the Japanese and Koreans. It was solved by the development of an American workforce that was bilingual and culturally competent in both countries. This took like a decade to put together and was done by basically sheer bloody mindedness. The big corporations deciding they were going to do this and if that meant just grinding through years of failure then they would do it because it ultimately guaranteed access to the American market. But once it was done it was done. And it was available to other Japanese/Korean companies (since they could poach people). Which is why Samsung didn't have this issue. They had their own people and, if they didn't, they could hire Korean-familiar American engineer.
The issue is that China doesn't have this cross-cultural workforce. There are Chinese-Americans, of course, but Asian business experience is not genetic. There's no large workforce of Americans who've worked in Chinese companies, are familiar with Chinese culture, speak Mandarin incidentally to being an industrial engineer or something, etc. But for the same reason this happened in the 1980s it will happen in the US: rising trade barriers, economic nationalism, and problems at home mean that an American workforce is the best way to guarantee access to the American market.
So this is a nothing new under the sun story. The interesting difference is the geopolitical differences if mainland Chinese companies decide it's in their interests to jump in too. But a lot of that depends on if and how they do since those factors aren't too relevant to Taiwan.
Taiwan is explicit about it being an English as a second language country.
Having lived in both China and Taiwan, Taiwanese people have a higher base level of English even if it is still low.
Yes? I don't see how this would mean there's a significant workforce of Americans who have language and corporate cultural skills. The Koreans and Japanese have high five to low six figures of these people. Samsung alone has like 60k American employees and by the time you're in the mid to upper layers they're regularly reporting to Korean superiors and moving up to international operations that involve Koreans.
Mainland Chinese companies don't really have the option of jumping in though. American political environment would not tolerate it at all.
Imagine the outrage if BYD announces a plant in Texas.
BYD already has a plant in Lancaster, California.
There was zero outrage, as evidenced by the fact that you didn't even know about it.
It's a commercial bus plant that has been there before the whole Chinese EV's being a threat was big. I meant am actual passenger car plant that'll produce 200k cars a year.
Look what happened to Ford's plant with CATL.
The transit bus plant is there because of Buy America rules on components and assembly. The FTA by setting standards for transit bus design, is effectively a monopsony buyer. Transit agencies, in order to receive grants to buy new buses, must adhere to them.
What's unusual about the BYD approach is that the buses have a universal design throughout the world. Basically, buses are a closed market. By the end of this year, BYD will be only one of three full-size bus manufacturers in the U.S., its only other competitors are Canada's New Flyer and U.S.'s Gillig, who basically have the transit bus market to themselves. The U.S. doesn't import or export transit buses.
Well, there was one.
A union, frustrated by its efforts to organize workers at the plant, did the time-honored tradition of California NIMBYism and lodged a CEQA lawsuit to stymie the plant.
They certainly have the option. There will no doubt be political issues. And my point is that fear of the Japanese as a geopolitical threat was imaginary while fear of the Chinese has more basis in fact so it might play out differently. But tariffs structurally encourage such investments in the US.
"Humans are humans, and they can learn."
Agreed. Completely. But back in the 1980s Japan really was Japan. I was at NEC's Central Research Lab doing AI for 2 years (86-87), and during that time, one of the Japanese business magazines had a special edition with a big-print cover title to the effect "There Will Be A Forreigner Working Next To You. Soon". And the joke in my group was "Next to you, yes. Working, no.". (But a few years later I ran into the guys and they reported that they were still using the pattern matcher I had written while I was there: I had killed time by implementing Carl Hewitt's pattern matcher he described in his PhD thesis.) They did meetings during the day and the actual work overtime*: I was out the door at 5:00 every day. But I largely didn't go to the meetings.
*: Which meant that promotion to a management position was a financial disaster, since they depended on the overtime pay, and management doesn't get overtime...
This is a great piece, as usual, Noah. Thank you.
I wonder what other cultural distinctions exist besides "saving face." I think much has been made of how Asian cultures do this, without acknowledging how much this happens in the US as well. (Remember Pence's Christian complaints about Trump's vulgarity and criminality? Me neither.) Few people can climb corporate ladders pointing out managements' flaws. Mostly this happens to people who constantly cover those up. That's many employees' primary function.
Furthermore, while we have a free press that is incentivized to criticize leadership, there are very real consequences to doing so. In Hollywood, writers and talk show hosts who tear down celebrities are not given interviews, and without such interviews there's no audience and no success. Much of what happens instead ends up being sycophantic, and then we're all "shocked" years later to discover these celebrities were suffering mental health crises, drug addictions, sexual impropriety, workplace harassment, etc.
Same goes with whistleblowing within industries -- even the media. I will be curious to see what becomes of Uri Berliner. It may be that he will find a news organization that wants to present itself as a counterpoint to NPR, like Fox News, and make a splash out of hiring him. But many will also be cautious of hiring someone who could turn against them.
And this will undoubtedly impact his social circle. People will be afraid of seeming like they are siding with him against everyone else who works at NPR and garnering the same stink that anyone who criticizes their stated politics does.
While the blacklisting of the 50s in America was official and reinforced by laws, it was also a manifestation of social behavior that does this more discreetly. It is often this subtler form of banishment that drives people's anxiety about "Woke" politics today. Any analysis, challenge or discussion of the Left's ideals raises everyone's suspicions and they race to castigate the speaker and outcast them, as if it means the person speaking or writing is a double-agent harboring patriarchal, white-supremacist agendas. The behavior of congresspeople like Marjorie Taylor Green going after Speaker Mike Johnson suggests this happens on the Right as well.
This is not indicative of a society that celebrates discussion or the challenging of authority or social norms. This is rigidly upholding ideologies. It's just the US is big enough for there to be two, major ideological poles and voting is private (at this point), so extremists still tend to be minorities.
Outstanding article. I worked in Semi in the late 90s so a lot of this is familiar