1. Long shifts are normal for fab operators in the U.S., eg Intel.
2. Certain engineers are always on call if an emergency happens, also true at Intel.
3. Samsung, not exactly the “live slow and enjoy life” type of culture as an employer operates fabs just fine in the U.S.
A big part of why this didnt bother me was the scale/impact of what I was doing - I was responsible for the primary business interface used across the entire factory floor.
Once you see a modern fab with your own eyes, it will change you in a deep way. I felt a sense of compassion for this incredibly complex and valuable thing that humanity is just barely able to scrape together. Nothing you see in media can prepare you for the real thing.
Definitely not a job for everyone, but its really easy for me to see how a lot of us were able to otherwise deal with the stress.
That being said, being on call is not for everybody, and a well managed planning goes a long way. It's one thing to be called for a true emergency, another because a colleague doesn't know how to turn on a monitor or because someone did something stupid on a weekend (extreme example)
Working in the US for a Korean megacorp, who rightfully demanded heavy standardization of manufacturing processes, meant I had basically zero latitude on this sort of stuff. I did manage to get approval for 1 minor development project which was ported back to Korea.
I still consider going back to SAS. I departed on amicable terms. The amount of impact I could bring with my skills today would probably open some additional doors. Working with stuff like Blazor in 2022 sometimes has me thinking about ways I could revamp that factory UX and solve problems that were impossible to solve before.
I support financial applications that could be considered critical infrastructure for the country. To the point where the government has set up special regulatory procedures to ensure it’s managed properly. We make sure everyone we hire in the support roles understands that this job isn’t for everyone, there’s work that needs to be done at possibly any hour of any day, and while we have offshore teams there can still be gaps due to sickness, power cuts, etc. we have in depth on call rotas. We pay overtime and on call supplements.
Even outside IT, there are power stations and telecoms systems, water mains, TV and radio stations, who knows what else that operate round the clock, round the year. There are a lot of people in every nation that work these kinds of jobs.
A standard expectation of 11 or 12 hour days is extreme though. No company should plan their operations around that as a standard practice. I’ve worked a 70+ hour week before when there’s been a proper crisis going on though. I got paid for every hour, and I know my work is appreciated so there should be a nice bonus on the way in a few months.
While working for Samsung Austin Semiconductor in 2013, I needed to travel to Korea for a month as part of our work on a global supply chain software project. IIRC, we were working at the semiconductor complex at Giehung. The fabs there are unbelievably enormous, much bigger than American fabs. Twelve stories tall, ~200x700 meters. A half dozen of these behemoths parked next to each other.
I felt like I was walking through an industrial city in Star Wars.
Edit to add: I believe you can probably contact the public relations department at Samsung Austin Semiconductor and schedule a tour for your group.
IIRC they do regular tours for schools and other STEM groups. Last I badged out, there was an entire museum room dedicated to the history/tech/etc.
https://www.mit.edu/people/dpolicar/writing/prose/text/think...
Then I got it. Brilliant.
I interned at Micron as an industrial engineer looking at capacity and equipment purchases. I can very much relate to that. The whole summer was my mind being blown by the orders of magnitude across the board and it's somehow it's an economically viable process.
First, we take this giant silicon crystal taller than humans and cut it into wafers. Then we take these pure materials (like 99.99999% pure) and transfer a tiny big on to the wafers in a successive layering process. Oh no, the deposition process wasn't perfectly even across the whole wafer (because of annoying laws of physics), so we'll throw it into the chemical mechanical planarizarion process to skim off a tiny layer keep the internal mechanical stressed down. Add in other mechanical, etching, lithography, and measurement processes and it gets crazy.
Wafers go through hundreds of manufacturing steps depending entirely on purpose, and with each step there's a possibility of messing up part or all of the wafer.
99% yield on a per-machine basis is sub-par in most manufacturing environments, but achieving that would be devastating in semiconductors. For sake of demonstration:
0.99^100{manufacturing steps} = 36.6% total process yield.
0.995^200 = 36.6% total process yield.
0.9975^400 = 36.7% total process yield.
Each additional 9 on yield is really expensive to add.
In school we talked about 1/1000th of an inch being kinda tight to hold on a CNC mill, with 1/10,000th needing a lot more specialized processing and time. Suddenly I was hearing about nanometer thick layers with tolerances measured in angstroms.
And the capital expenditure was nuts! 6-figures hardly gets you anything in a fab, it's really in the 7-8 figure range where you see most of your equipment landing. That will be old but still viable equipment in a few years.
Somehow depreciating all that equipment designed to make chips sold at pennies to dollars each is profitable.
As far as I'm concerned, it's black magic and truly an incredible achievement for humanity.
That’s a really good point. It’s crazy to think that as we get smaller processes that not adding additional 9’s gets you relegated out of the leading edge, and possibly out of relevancy.
Black magic indeed!
Fabs seem to be the far edge of tools-to-make-the-tools dependency stack, with some of the necessary sorcerous gadgets (like EUV lithography) themselves only having a single supplier with the necessary secret incantations on the other side of the planet.
Does the knowledge even exist to bootstrap things back up again should the present infrastructure and supply chains be sufficiently disrupted?
I've long admired David J. Gingery's 'Build Your Own Metal Working Shop from Scrap' and wondered whether anything like it is even possible for integrated circuits.
Man, I'd love to know. Hypothetical scenarios are:
1) Everything is available at your fingertips supply chain wise, but you have no equipment. How long until we're at our current technical capability?
2) No equipment and the supply chain need to be rebuilt, but the designs are still present.
3) We have sticks and stones and whatever is in our brains. Start fresh like a game of Factorio and see how well you can do on a speed run.
I don't recall exactly where (some large ouvre like The Culture, or The Polity), but some SF novel referred in passing to this becoming an extreme sport: Your team is dropped with no supplies on an uninhabited planet, and you try to bootstrap back to an interstellar capability.
IIRC, the aside mentioned that the current record was ~70 years, by a team genetically engineered for ultra-efficient digestion and no need for sleep.
Based in Europe, Germany, we had this when we had meetings with chipmanufacturers in Japan vs. USA. So when a company comes from one culture into another another it always has a learning curve. Even when you employ local people helping you to adopt to the local culture. Because you have still managers in the other cultur with their expectations.
I learned that also from Daimler/Mercedes Benz. When they startet to copy their manufacturing plant from Germany into the USA, they had many quality problems and all kind of problems. Over the years there was a learning curve to adopt to the US work culture. So they still don't have only about a 2/3 output compared to the very same manufcaturing plants based in Germany.
While Asian companies have much higher working culture than Germany, I assume the experience at first is horrible for the Asian managers.
When I worked at Samsung, there was virtually always friction between the Korean and American stakeholders. It was something we learned to tolerate though and eventually these differences did bring a lot of combined value.
The general theme was something like - Korea uses American talent to shake up processes and innovate in ways they would be difficult to approach on cultural grounds. For instance, directly challenging an incorrect idea from your manager's manager. This would be a standard Monday morning activity for any native Texan/American working at SAS. It may initially cause an uproar (why didn't you notify xyz first... etc.), but it always got the ball rolling on a quick resolution and left an implicit obligation to review why certain management chain links were skipped over in the first place.
On the other hand, Korea brought a degree of consistency/discipline in their process engineering that staggered many of us. Virtually everyone balked at this stuff initially, but we grew to understand the value over time.
The language barrier was the hardest thing from the American perspective. I feel like many of us would have had an easier time if all of the documentation wasn't originally developed in Korean. Automatic translation tech was not able to help back in the day, so your next best bet was to find a Korean who worked in your office area to help you out. Needless to say, ad-hoc human translation services were hard to come by, especially if one didn't want to feel abusive about it.
2. It is possible that being on call at Intel requires fewer hours of attention than being on call at TSMC. It is also possible that Intel is offering higher starting salaries than TSMC, given that Intel is in a position of relative weakness.
3. You don't actually know if Samsung has been doing "just fine" recently in the United States. I'm sure it did in 2018 and 2019 and the beginning in 2020. I'm very skeptical that Samsung has been humming along without incident for the past six months.
The type of work described in the article has a reputation of low pay per hour and being undesirable.
> TSMC will have to change to an eight-hour work day five days a week.”
Is this US society breaking down and American people being “ungovernable”? Or is it people being able to negotiate reasonable pay to working conditions ratios?
These are people who want that job to exist, but are unwilling to work for the pay and conditions necessary for that job to exist.
Does you definition of free market come from the "you'll work as much as I need you to if you want to eat" camp or the "if you want me to work bad enough, you'll pay me what I ask and negotiate the hours I am willing to work for you" camp?
Two things can be true though and may not be related.
I don't really see it as a hit piece, all sides here sound reasonable. It will be interesting to see what the end solution is. I think that the author would have shown a strong opinion on how to resolve it if it was a hit piece.
Their version of governable seems to mean compliant.
It's not inherently good or bad but it is a particular perspective that I can't see working well in some cultures.
Compliant does not mean accepting piss poor pay and poor work life balance.
Characterizing TSMC not being able to find sufficient people to accept their working conditions/pay as "US workers not being compliant" is a meaningless use of the word compliant in this context.
there is an underlying agreement...just because it isn't directly stated doesn't make them not real. Social contracts arise from cultural norms and very much are a thing.
The OP article, IMHO, smells a bit like misinformation. There are interests (especially in the US) which want to see a never-ending stream outsourcing to "low-cost centers". Stuff like the article provides fodder for that point of view by painting a false picture that americans can't measure up.
Interesting word choice.
Supposedly, the word "ungovernable" was used by anticolonialist/anti-imperialist factions in the '70s to describe a deliberate strategy of breaking down the "peace, order, and good government" that their imperialist counterparts offered as justification for their existence. Closely related is "no justice, no peace". However, this attribution of "ungovernable" to left-actors came to me via a right-source, so I'm not completely sure about it. If true, however, then I think you're implicitly reaching back to that era, to conjure images of South Africa and Uganda. I think the recent coinage of "Brazilianization" has a similar goal. Anyway, I would be very interested if anyone could help trace this sense of "ungovernable" to its source.
Some possible leads:
[1] https://www.cambridge.org/core/journals/british-journal-of-p... and its sources, particularly
[2] King, Anthony, ‘Overload: Problems of Governing in the 1970s’, Political Studies, XXIII (1975), 284–96. ;
and
[3] https://ssc.wisc.edu/~wright/Soc924-2011/Offe%20--%20Ungover...
and its sources, particularly
[4] Fluno, Robert Y. (1971), “The Floundering Leviathan: Pluralism in an Age of Ungover- nability.” In: Political Research Quarterly, 24/3: 560-566.
These are all "establishment" sources though; there are no real Left sources so far -- though [3] does say,
> Ungovernability is a peculiar concept. It has been used both by academic political sociologists and by journalists and politicians; it has played a role in New Left as well as neo-conservative discourses.
so I think it should be possible to see where it went (and came from) on the Left.
EDIT: I now also find these:
[5] https://knowyourmeme.com/memes/become-ungovernable
[6] https://dailypost.ng/2021/01/22/us-were-ungovernable-protest...
Who knows what the latter source is though.
[7] https://itsgoingdown.org/16-things-you-can-do-to-be-ungovern...
EDIT: Yeah, I think it's real.
> A global pandemic that has resulted in 4.5 million people quitting their jobs in the United States in the last month
4.5 million people quit their jobs and there was a global pandemic. 4.5 million would need to be compared to historical rates to say the pandemic resulted in an increase of X people quitting their jobs, not the total of 4.5 million. But even that X increase is a correlation and I highly doubt that the pandemic was 100% responsible for every single one of those individuals quitting their jobs.
A more complete story can be found from the bureau of labor statistics.
https://www.bls.gov/news.release/jolts.nr0.htm
> Over the 12 months ending in November 2021, hires totaled 74.5 million and separations totaled 68.7 million, yielding a net employment gain of 5.9 million.
I still agree that the idea of bullet 1. is likely valid, that people are leaving jobs that are undesirable and getting better ones - but I'd be careful to add the reason why. Causation has been thrown around casually during the pandemic to influence policy without any evidence of the causation or the magnitude of that causal relationship. You could just delete "A global pandemic that has resulted in" and the sentence would stand alone.
> “If you are a process engineer, it will be more stable. Maybe you can start work at 8:30 a.m.and leave before 7:30 p.m. If there are some urgent matters, you may have to stay later.”
This doesn't sound like long shifts, but regular 60+ hour weeks. If a 13 hour shift (8 to 9) is "not normal" but happens 2 to 3 days a week, it doesn't sound like they are just working 3 days a week. A "more stable" position works 11 hours.
As an European these hours would be completely unacceptable to me unless it's 3 days on 4 days off, which it doesn't seem to be.
[0] Hiring as permanent employee vs fictitious "B2B" contracting can have different trade offs to employer, and while the later is cheaper in terms of wages, things like due diligence or contractual responsibility to customers/suppliers might be easier with permanent workers.
That seems reasonable, as long as workers have a stable schedule rather than the 'flex time' insanity that plagues many US workplaces.
That said, I imagine there are productivity tradeoffs between fresher workers (esp. at end of shift) vs. more handoffs between shifts and higher headcount of both workers and managers (so more communication overhead). How does that work out in practice?
Americans aren’t interested in 13 hour shifts, either, unless the pay is extraordinarily high.
I agree: Something is missing from this article. If they’re running 3-on, 4-off schedules with these hours, that’s one thing. But if they’re asking 13-hour days and 5-day weeks then it’s an open and shut case: Their jobs are a poor value proposition and people will (rightly) leave.
The factory: runs 24x7x365 - it has to, most processes have enough controlled time processes that it is a effectively week (or weeks) long process to shut down certain steps in the factory. Note for this whole comment I'm speaking of a typical 300mm high output FAB from a couple of years ago when I left the industry...for mental health/work life balance reasons. The reality is, if a single piece of critical equipment is down, it doesn't matter who you are, you will be on the phone, computer, or in the factory if your services are needed. During new process qualification it is not unheard of for equipment/process engineers to be expected to be at their equipment when certain special FOUPs of wafers are run - day or night.
The operators: typically 12ish hour shifts. Really 12 hours with an extra 15-30 min at the end to do an effective handover of the current situation of the fab. Typically in US fabs from my experience these are 3on 4off 4 on 3off schedules, often with wednesday or saturday/sunday as the swing day. In the olden days, these folks physically moved wafers and wafer boxes about the factory and it was a menial, fairly low skill, but highly precise job. Nowadays, most of them are sitting in some form of ops center monitoring looking at screens - because they are an unnecessary contaminant in the FAB. Due to automation, their number has gone down but their value and cognitive ability has definitely gone up because they are looking at the state of many pieces of equipment and many steps in the factory in an abstract way on a computer screen and controlling the movement of thousands of wafers and millions of dollars simultaneously. Good ones seem like they can see the future and are worth at least their weight in silver. I spent a lot of time trying to turn what the good ones saw/observed intuitively into systems, metrics, and control logic - that was cool and humbling.
The technicians: typically 12hour ish shifts 3 on 4 off unless doing cross training or something - with the same handover process that might last a little longer if they need to show something in person or show engineers something. These are the people physically working on the equipment by hand to perform preventative maintenance and repairs. These are the guys in the bunnysuits in the fab. Basically like the PhD version of a car mechanic. Good ones are paid more than a lot of the engineers - and are worth every cent. Beloved by ex navy folks (especially navy nukes and air force enlisted) - I knew of a fab in the North Eastern US where something like 16 techs 'hot bunked' in a nearby 2 bedroom apartment, 4 per shift, 4 shifts per week. Made local rent incredibly cheap and let them live somewhere else further away and nicer the remainder of the time.
The equipment/shift engineers: Half step between technicians and the process/yield engineers. Responsible for making sure individual pieces of equipment and fleets of equipment 'work' - meaning on target, clean, reliable etc. These are most frequently the people who's life get eaten by the job. They are attached to a physical asset that must work 95-99%+ of the 24x7x365 day year. This is where a lot of young engineers start and there is serious churn in some fabs. Compared to a lot of engineering roles, these folks really need good communication skills but it isn't necessarily obvious up front because it looks like a highly technical job on the surface. You have to build good relationships with your techs or you will fail, and often that means putting your ass in the FAB, listening, wiping, cleaning up tools to establish respect and trust.
The process engineers/yield team: Baseline is 9ish-5ish M-F, in my personal experience more like 7:15 to 5:30ish with coverage on weekends and nights available by a smaller group - not dissimilar to SREs in cloud companies. Its not that they don't work long hours, but they are not (generally) not on call as much or attached to a single physical asset.
Managers: work schedule unknown, except they are always there when you don't want them to be and never there when you wish they were :). Many of these roles become 'always available' roles, or historically they have been...some places have made moves to put two-in-a-box management models which seems to help burnout. I never did it officially but did a lot of coverage of these roles. It's kind of like air traffic control on + engineering + strategic planning mixed together.
Bigger picture - not everyone operates on the same schedule, because of the work they do and a variety of other factors. All of this is also dependent on when the 12/24 hour shifts start and end - some might run 7-7, others 8-8 or even 9-9. The exact shifts seem to shift FAB to FAB/company to company along with the boundaries of these roles. Production is different from 'ramp' is different from process development is different from R&D. What has seemed pretty universal is long hours for 'engineering types' who are doing the knowledge work and first level management of the factories physical assets. But long hours along with expectations for in person/on the phone/offline vary with a number of factors.
Would the FABS be better off with a larger workforce and a better work-life balance? My US/Euro work background perspective says yes - but I would also acknowledge there becomes, over time a huge amount of innate and tacit knowledge in these jobs that are highly experientially developed and just training more people is really hard. Coordinating amongst a larger group is even harder. That being said, I know a number of FABS that are really trying to push towards 'teams' of engineers that can slide and cross cover because the work life balance really can suck.
When I say suck I mean like (1) 'therapists near the plant who's entire patient list works for the FAB because its easier on the therapist and they still have a waiting list' sucks. And (2) 'people married to other FAB employees or single career households have longer lasting marriages' sucks.
An accident at a chip factory isn't going to be front page news the way an airplane crash is, but it could very easy cost millions in repairs and lost productivity. When I see talk about 12 hour shifts I always wonder if the management isn't costing themselves more in the long run, even if they aren't allowed to admit it for macho culture reasons.
Some may consider it worthwhile in their stage of life.
Correction: 14-month worth plus additional performance bonus.
Edit: it’s customary in Taiwan to measure the tear end bonus in 1 month pay equivalents (8.3% of base pay).
I consider my self a fairly hard working person, in my first job I've seen situations where people came in on a Monday and would go to their homes next Monday. I've done a full time start up with a day job, quite literally coding every waking hour of my life(like 2 - 4 hours sleep a day) for years. And then I relocated from Bangalore to Bay Area for a while, and met a few of our Chinese colleagues- Even by my standards the Chinese were in a different productivity league altogether.
One person I knew would have three monitors, would eat at desk all the three meals, not sure when he even took breaks, but he kind of worked all the time, I mean literally. Would watch tickets in one screen, production issues in one, and work in another one. He'd read every ticket, feature request, pull request, and comment every where, knew everything and would just outperform anybody by some factors(Mind you people in our team I knew were kind of legends themselves). It ain't just the hours, they through and through dominated the initiative. You stood no chance.
I knew several Chinese colleagues and friends like these. They are all good people wanting to make a good life by working hard, nothing to blame them. But it was kind of an unwritten rule that if you had two Chinese working in your team, you had no chance beating them.
Im in serious awe of the Chinese culture. They also seemed to have very little gender discrimination, and in general little less hyper about things like religion etc. I guess all this things give them a natural edge.
So yeah, not all long shifts are equal.
I've had the opposite experience. Most of the ones I knew were work evasive, and spent substantial time craming for leet code. I think there might be a generation gap at work.
My interactions have been very positive, and being an Indian I have never experienced any racism from them. There is a language gap for sure, and due to that times, more than a round of clarification is needed to be on the same page, but that was nothing I couldn't work out.
I agree your interactions would have been different.
It is also possible that I interacted with subset of Chinese immigrants in the Bay Area, which generally happens to be a place for people who in general value diversity and happens to be a meritorious place.
>>You probably should re-evaluate your own objectives
I have, I don't believe in linear returns any more and my days of attempting to scale personal growth as a function of hours worked are behind me. You grow up with age and your perspectives change too.
At the same time I have mad respect for immigrants(as a former failed immigrant myself) and other communities that are working day and night for a good life. More power to them.
I am much more productive coding than 99% of the people around me. One of my secrets is I sleep well. Probably your definition of coding is different from mine because of your working area. People make stupid mistakes when they are sleepy, including me, and it is way more expensive to fix(finding bugs in a big codebase) than not making then in the first place.
I have met lots of Chinese people working and had the opposite experience. China is so big that outliers that represent one in a thousand are more common but most people are that, average at most, way worse than European trained workers.
The fact that they sleep on their shops and workshops does not make them super human. Quite the opposite, they are human and working more hours make them work less intense.
I don't believe the Story that Chinese can run at Marathon pace all day, because they are similar mental limits for humans as they are physical ones. And Chinese people get burnout like everybody else or worse. Apathy on workers is the normal state.
Based on what you just described?! I'm sure there are many great reasons to be in awe of Chinese culture, but what you just described sounds like an absolute nightmare. How is this stuff a positive thing? How is this person anything more than a work machine? What's the point of their life at that point?
Studies on working have consistently shown diminishing returns on longer and longer hours, generally past 30-40, depending on job type, i.e. physical jobs like assembly line work drop of slowly (mistakes rise but only gradually) vs intellectual jobs where mistakes either rise rapidly or people adjust by taking more slack time throughout the shift.
That might be the crux of it. I gather it's similar on e.g. offshore oil rigs, one week on, one week off, but long, grueling days.
That said, this article (and the other threads) describe being available 24/7; you probably weren't expected to be on call during your days off.
Anyway, I think being on call is fine, as long as you get compensated for it - because you have to change your lifestyle to be available, plan things like going out or having a drink.
Some plants run the DuPont shift pattern or they run 3 8 hour shifts per day.
A friend of mine works at a fab at another company and he's not working crazy hours. I think he currently works the three days on three days off shift.
It's one thing to have on-call hours, it's another to have on-call hours while also being expected to work 60 hours a week. It sounds like TSMC is expecting the latter.
It might also just simply be a problem with management.
There's other industrial manufacturing that happens 24/7 in the US, so it's possible
If they are having issues with this it sounds like a management problem.
Its a confirmation bias thing common with low level guys. It happens in the military too. For example it was a major takeway for me when watching Generation Kill. Like this interview the reporter has with the point he has to make at 1:50 int he video
https://www.youtube.com/watch?v=gguxM8eABP8
Fab plants like this usually run 3 shifts, thats 24/7 production right there. So you can still do 5x8's and run a plant 24/7. My father in law literally ran ops for a few places doing printed circuit board fab before it was all sent offshore. Most of what made him valuable was logistics and managing shifts/workers/on call and ops process to ensure it ran smoothly. Not everyone can do it and many that THINK they can would flounder when put in the hotseat. Similar when people ascend to Director or Executive roles.
I worked in system like this and I actually loved that, especially that those were night shifts
I'd really take 3(+1)x 12h nights over 5x 8h