The 'brain drain' (as you refer to it) stems from intelligent/motivated grads in the US for the last two decades (at least) pursuing more lucrative fields like finance and adtech (re: Google, Facebook). Or some pursue management route (attending big MBA schools and switching to management roles where they climb corporate ladder). In other words, there are not a lot of college/grad students who want to pursue traditional engineering routes in the US.
I myself was an electrical engineering (EE) major until I switched to computer science in my third (junior) year of college because like a friend of mine at the time told me, "<my name>, if you don't major in computer science, you will not be able to find a job easily after graduation". He was right. All of my former college friends in EE ended up pursuing programming jobs (a few of them now works for FAANG; I used to work for one but left a year ago due to RTO). That is why the US has no sufficient personnel to do traditional engineering jobs and we have shipped off a lot of those to foreign countries.
It's not even brain drain, America's dominance came from the fact that for nearly a century the brightest people in the world were willing to give up everything to come here. That is no longer the case. Today's Einstein probably isn't going to immigrate here.
Alternate explanation: electrical engineering is actually really hard and some parts of computer science look comparatively easier. Plus coding is startups is cool, EE is still nerd as in Nerd.
> I myself was an electrical engineering (EE) major until I switched to computer science in my third (junior) year of college because like a friend of mine at the time told me, "<my name>, if you don't major in computer science, you will not be able to find a job easily after graduation". He was right. All of my former college friends in EE ended up pursuing programming jobs (a few of them now works for FAANG; I used to work for one but left a year ago due to RTO).
Nothing against you looking out for your future, but this is exactly what I describe to people when I say the industry has changed. It used to be nerds who were very passionate. Now it’s full of people who are just doing a job.
Hit the nail on the head. I went to the College of Nanoscale Science and Engineering in Albany for a master's in "nanoscale engineering" which essentially boiled down to a master's in being a fab line manager. I finished the degree since it was only a 3 semester program and I was getting paid for research work, but almost immediately after chatting with alumns that went to go work at IBM/Intel/etc it was pretty clear that software engineering was a much more lucrative and less stressful career.
Definitely true, as there weren’t EE jobs here. Now that we’re moving chip manufacturing back, and with programming job market being saturated, perhaps it will shift and EE will pay more due to being more in demand
I think your explanation about large numbers of motivated students pursuing lucrative Non-STEM degrees is incomplete without mentioning the cost of an undergraduate and graduate STEM education in the USA.
The most critical shortages of STEM graduates are in roles requiring advanced degrees. Your median undergraduate education (~$40k) and median graduate education (~$60k) saddles students with approximately $100k in unforgivable student debt! Never mind the years lost that one could otherwise be working. So it’s no wonder students are motivated by the ROI of their degrees, it’s why I chose Computer Engineering over Electrical Engineering.
These are expensive STEM degrees which students on visas are all too willing to pay for a chance at a residency and a pathway to citizenship. So no wonder the majority of undergraduate and graduate STEM students are foreign born in the US. The ROI is not worth it for the debt. We don’t have enough need based scholarships available to finance the STEM graduates this country claims it needs.
Really appreciate this comment and perspective! In the larger context of immigration and brain drain in other countries, how the US also has one, but of a different kind. Ultimately, it's a loss of potential. I'd somewhat disagree with the directionality of the correlative/causal relation, though. But what can be said is that the US also experiences a knowledge drain towards plainly lucrative jobs. I'd wager that it was/is a cyclical effect that just worsened over the decades and that neither engineers moving to fintech nor low-paying engineering jobs were/are the sole reason.
My hot take as to the reason EE is a bit of a dead end in the US is that the options outside of the handful of primary employers are limited. It is very capital intensive to run a semiconductor fab, design chips or assemble electronics at scale. Therefore the employer has all of the leverage. The equipment and/or factory worker infrastructure comes first and the engineering teams are just a cog.
Compare that to having all the degrees of freedom as a computer science student to start up a niche mobile app or internet based niche service after working at FAANG for 5-6 years. Even AI infrastructure will eventually go down in price making niche AI first startups a possibility. In finance its the same, as a post i-banker you have the option to start a boutique fund, a niche fintech or just invest your own savings.
What you said seems contradictory. You open with the premise that intelligent youth go the finance / CS / MBA path instead of engineering and then say that those who do go into traditional engineering can’t find jobs. Couldn’t it be that people don’t go into engineering because there aren’t any jobs? Wouldn’t the lack of jobs explain the low salaries and thus the preference for more high paying alternatives?
I read the main problem with hiring chip factory workers in Arizona was the factory just didnt pay enough for the long hours demanded. I looked up the median salary and its only 50k so I'm assuming it's not crazy skilled labor (e.g. brain drain). Taiwanese workers just seem more willing to do it.
I spoke to a Taiwanese person and apparently the salaries there are actually quite good, even by western standards (normal ones; not SF). The downside is they have very very long hours (996, barely any holiday, etc.).
50k is just a step above McDonalds these days in a lot of areas. Sure minimum wage might be $15k, but realistically nobody pays that little except in very rural areas (if you need a small number of low skilled employees a small rural town is a perfect spot to build - but if you need more than a small number they can't provide more at any price - you will pay more in the city but there are a lot more people around if you need more)
That's why manufacturing offshored in the first place, companies feel they're receiving better value for money on wages elsewhere for this kind of work (and these days not to mention more & larger facilities, proximity to component sources, and a strong ecosystem of supporting and complimentary facilities).
The chips still need to fly back to Taiwan to be packaged as packaging partner Amkor's facility in Arizona won't be ready until 2027*. I'm not sure the cause of the delta but it could be in part because Fab 21 got back on schedule rather impressively following earlier delays.
* updated to reflect newer article that Amkor's facility is delayed beyond late-2025
I was about to say, surly at some point in the near future the USA will introduce this capability. Shame they did not match each other in completion time.
For a new factory with a new entry into the local market it makes perfect sense to bring in experienced workers for knowledge transfer. This is more an issue if a decade later this is still how things are done.
Back when American companies were offshoring, the initial start up teams were comprised of a lot of Americans who would do commissioning and initial ramp ups while training up the foreign workers. It's a lot easier to train people on a production line that is proven to work.
Problem is, those jobs in emerging markets were desirable compared to other jobs (for pay and opportunities), which helped with talent growth. These factory jobs, in comparison to other jobs, aren’t that desirable.
Edit: This is not news. This (combined with their higher EE education) is why Taiwan won IBM PC-clone-related manufacturing in the 80s. And why they now have TSMC.
Making chips isn't something you learn the details of at University. You can take all the classes you want in advanced semiconductor techniques but the simple fact is University level manufacturing is nowhere close to SOTA.
Basically, you need fab workers to spend time in Taiwan/China, and then return to USA. It's the same model that most foreign students use at schools in USA/Canada. Get USA/Canadian name brand school on resume, learn english, and go back to home country = profit.
Re the first point: Why do you think it is so difficult to transfer chip production off Taiwan?
I don’t think this is about salaries. Nor is this about facilities.
This is about process know-how. And it’s currently not available outside of Taiwan. I’m glad we’re finally starting to transfer knowledge. It will take a couple more years.
Does anyone know the general path to get involved in this? Perhaps its romantic, but this seems important, it seems hard, and it seems like something I can be proud of working on (as opposed to maximizing ad clicks). I'm just a SWE w/ a Comp sci degree, so what's the entry-point here?
Your entry point is a masters and probably Phd in Electrical Engineering, specializing in some aspect of semiconductor manufacturing. It’s definitely not CS.
I'm not too sure but I would assume there's going to be faster turn prototype chips in the USA now? Is packaging needed to prove a prototype? Can we start buying IP blocks and make our own ICs? I'd love a MCU with built in IMU and wide range LDO, not sure if that's possible all on the same node.
There's going to be some niches opening as a result of this IMO.
it's first step. you gotta do something to bootstrap, solve chicken-egg problem. From what I can see around me, the "made in america" is a no joke branding. a lot of pppl going tobuyjust because of that. and may even consider it as social status and their policial support.
Sure, but this is how a supply chain gets bootstrapped. All those factories in China didn't magically appear one day. Just like they didn't appear when Apple started moving operations to Vietnam. You start piecemeal and build out.
I have two kids in grade school and middle school and I see why we have a STEM gap. I have to constantly correct the learning at home in math. Also, I think it's fair to assume that in Korea, Japan, Taiwan, and China the school kids are actually put on an academic grindset unlike here where there is such little academic rigor or discipline being enforced by the school it makes sense why the k-12 education numbers are as bad as they are in the USA.
It might be worth getting up in front of the kids in middle school + and saying "Hey you're in competition at a global scale here. You're going to have to work your butts off to stay relevant."
Maybe that's how US is going to have enough STEM talents -- just like WWI and WWII, take as many talents as possible when the other parts of the world are in shit.
The scenario that we’re going to be able to fight a war with another first world power, where we will attack their infrastructure but ours will be left untouched, seems unlikely.
I have no specific info regarding this plant, but for anyone who never experienced this: flying in people from other plants at the starts (and all 3rd party vendors for a hypercare phase at launch) seems pretty normal.
If they have to keep staffing it that way, that's different.
you can fly a few hundred million dollars worth of chips in a single flight. You need not be concerned. The impact from temu shipments is several orders of magnitude higher.
made in america is also a federally defined standard that these chips categorically fail to meet. assembled in the united states is more appropriate, and even then if you didnt hire americans to do it, what was the point?
this is starting to feel like the best of intentions that has spiraled into a political theatricality where close-enough will be good-enough.
given the current state of declining US college enrollment, the affordability crisis of college, the growing wage gap, the failure of the minimum wage to keep up with the cost of living, and the failure to reform predatory US student lending practices I do not see how the US will in the next 25 years ever manage to curate the type of braintrust for which it was once renowned across the globe.
This is so disconnected from reality. They've gone from breaking ground to replicating one of the most advanced fabrication processes in the history of the world _at scale_ in about 4 years, but they'll be sending the dies off for packaging while their packaging partner comes online so its just political theatre?
Also, over half of the employees are local hires and the ratio will increase as more of the fab spins up. IMO it would be much worse political theatre to delay and balloon the cost of the project by forcing TSMC to exclusively use a workforce that has no experience with the companies tools and processes.
There's simply no real reason we can't have a deep and robust manufacturing base in America. Well except for the fact that some specific people made a whole lot of money while letting it fall apart, and have paid for decades of media relations trying to convince everyone otherwise.
If you're reading this statement I just made and want to instinctively disagree with me, start by interrogating your own opinion. Why do you think America can't compete with China, for example, over the long term? What "well everyone knows" facts are you using to create that opinion that you don't have any first hand relationship to.
> Why do you think America can't compete with China, for example, over the long term?
Not saying I necessarily disagree with you, but just to give an example, the US has considerably better labor practices and labor laws than China. It's not perfect but there are protections about making sure people are paid what they're owed, how much you are allowed to work someone, safety protocols, etc. All of those things could, in theory, cost more money and make labor more expensive.
Compare this to nations that don't have the same work protections, where they can pay people peanuts and have them work much longer shifts with effectively no protection (e.g. Foxconn in China [1]).
This might translate to decreased cost, and Americans have made it excruciatingly clear that we're apparently fine with slave labor as long as it doesn't happen within the US.
> Well except for the fact that some specific people made a whole lot of money while letting it fall apart, and have paid for decades of media relations trying to convince everyone otherwise.
I like the idea of made in America and bringing manufacturing self sufficiency to the US. But I don’t like the idea of reducing dependency on Taiwan, which makes it so that the world may ignore their plight in face of increasing aggression from China. The CCP is an authoritarian dictatorial government that seeks illegitimate control over Hong Kong, Taiwan, Tibet, Xinjiang, and other areas. They need to be stopped and the solution isn’t to remove incentives to defend those areas.
There's a need to be pragmatic here; In the event of any kinetic Chinese aggression, TSMC (and other co's) fabs are going to be rendered inoperable, regardless of how well executed a US response is.
The world will will not be able to help with their plight, just as it was not with Ukraine and more recently Palestine. Might as well secure the supply chain.
Protecting Taiwan would mean WWIII. I hope the West and the Taiwanese figure this out ASAP and start moving as many people as possible out of there, and destroy all the fabs.
China is building more warships than the rest of the world combined. And NATO can't even recruit enough people to man their existing fleet.
But it seems the West is a victim of its own propaganda so there's no political will.
"TSMC does not have an advanced packaging facility in the U.S., and its partner Amkor will only start packaging chips in Arizona in 2027. As a result, Blackwell AI silicon produced in Arizona will need to be shipped back to Taiwan for final assembly, as all of TSMC's CoWoS packaging capacity remains in Taiwan."
Given that there may be a 25% chance that China invades Taiwan by 2030, having the ability to package SOTA chips in the US by 2027 seems "soon enough".
"Packaging" in this context means taking the wafer of compute die (made in Arizona), dicing it up into individual die, mounting it onto a silicon interposer (an even bigger die, no idea where that's made, but probably taiwan) along with a bunch of HBM die, then mounting that Si interposer on a somewhat larger, very fine-pitched circuit board ('substrate') that is essentially a breakout for power and high-speed I/O from the compute die. That thing is the packaged 'CoWoS' system, where CoWoS==Chip-on-wafer-on-substrate, that eventually gets attached to a 'normal' PCB.
I'm making an educated guess but probably the cutting of chips from the wafers, placing them into the appropriate ceramic socket types (DIP, BFGA, SMD etc), soldering the line wires from chip to pin, encasing the chip, etc.
I believe packaging in this context means taking the raw silicon dies and assembling them into a package which can be soldered onto a PCB (or put in a socket, but Apple doesn't socket anything).
The machines and processes needed to package the individual integrated circuits are fantastically expensive but the margins are so low in that step that it's only profitable at massive scales.
So you put the fantastically expensive machines near where most of the customers are and most of the customers are in Asia.
Works the same way with fiber optic cables. Making the long skinny bits is hard and high-margin. Actually turning them into cables is easy and low-margin.
So Corning makes huge spools of fiber optic cable in Arizona, North Carolina, and New York (I think) and ships it off to Taiwan and China where it is made into the cables that you plug into stuff.
Marine shipping is just about the most fuel efficient way of moving things between any two places, by a lot. A 100,000 dwt ship can get 1050 miles per gallon per ton of cargo. It takes about a teaspoon full of fuel to move an iPhone sized device across the pacific when I ran the numbers last.
Plus, chips are small in size and cost a lot so you can fit a lot in a container. Per unit shipping costs probably come out to be pretty low. Especially when compared to the political costs and risks associated with not onshoring.
I'm suprised they can't ship (flat) packaging that could be used in Arizona with a simple assembly line.
If they had that packaging design then for this to make financial sense the two way shipping (and loading, unloading, custom clearance etc) would have to be less than shipping the packaging, the setup cost per unit cost of putting the chip in a box
I keep hearing about a skills gap in the US for fabs, what skills or jobs are actually suffering from this? people with masters in nanotech, compeng, EE?
Perhaps there is a skill gap because nobody actually knows there is a demand? I have no idea what to recommend to people who are trying to choose a college degree.
With my industry in infosec, at least there are certifications one can take, even proper masters degrees these days. In my experience, there is no skills gap in cybersec, despite what CEO's and linkedin-types' sentiment. They just don't want to pay market price for skilled talent. "skills gap" has meant "we need more talent so we can pay less", there is no actual shortage of people who can do the jobs adequately.
Is it different for chip fabrication? and if so, how can regular people work/study to obtain these skills? If I, having read HN for years and reading about the fab process have no clue, how can regular people who don't visit HN?
If you all can help me answer this, I'll try to recruit a few people into pursuing the right career to help meet this demand.
Off topic... Taiwan also machines and heat treats some of the best cutlery steels in the world. Taichung City is famous for this. This is not as delicate a process as producing CPU chips, but it is hard to get right consistently.
Most all major cutlery companies have product lines that are produced solely in Taiwan (Spyderco, Cold Steel, Demko, etc.)
It would be nice to see Taiwanese steel industy move some production to the US as well.
High-quality knives come from proper metallurgy, especially as it relates to proper hardening steps. If you don't get these things exactly right, the best machining on earth is not going to produce even mediocre knives.
The announcement of this plant coincided with the announcement of the Endless Frontier Act and CHIPS for America act, which is what eventually became the bill we call CHIPS and Science Act.
This plant was the foundation that the CHIPS act was built upon. The Secretary of State had to secure an agreement with TSMC to build this fab before the bills could be drafted, as a lot of the recipients of the funding are suppliers for this plant.
It is completely truthful to assert that this is the result of the CHIPS act. Congress agreed to introduce the bills as a result of TSMC's agreement to build the fab in Arizona. If you have to avoid giving Biden credit, then you can point out that it was Trump's SoS who negotiated this original agreement.
I agree that the CHIPS Act was likely contingent on someone showing semiconductor manufacturing could actually be onshored. I’m not sure I buy that TSMC’s investment was contingent on an Act that was contingent on them investing in the first place. It’s not like TSMC was ever going to get a check to just reimburse themselves. Even now, their subsidies are only for new plans.
Intel, on the other hand, is a great example of a how a company dependent on the government funding for semiconductor manufacturing behaves. Heck, look at the Foxconn debacle; companies prefer incentives up front.
If you remember, TSMC had the immediate fear of losing ~15% of their revenue with the Huawei export ban. I wouldn’t be surprised if that influenced their decision to cozy up to America.
What makes you think construction was completed in July 2022? The shell of Phase 1 may have been completed, but even now the construction continues in Phase 1B and Phase 2.
Half of the works are from Taiwan
All machines were imported to build the factory.
USA can't do anythings without immigrants.
China was able to develop its own chip factory without immigrants and without buying machines (because USA blocked the 'free market')
This is really exciting. It'd be awesome if the rebirth of American industrialism was tech hardware driven. It sounds like this being mass production ready is still a few years off, but kudos to Apple and TSMC for working to make this happen.
My guess is no, it won't. This is US taxpayer money being used to increase the manufacturing capacity available to the market so that the US has domestic manufacturing when stuff goes sideways. A similar thing regularly occurs with auto manufacturing and manufacturing in country A usually frees up capacity for other countries, resulting in slightly lower prices.
What could happen is that once the US has manufacturing capacity it decides to tariff imported chips, causing your country to retroactively do the same. This is decades away, and the US has a problem sourcing chips it can trust right now, so it's not currently on the radar. It's not something I'm going to worry about.
Viewed through a pessimistic eye, the US finally is realizing that its arms production critically relies on chip production and it can't says its chips are US made when selling arms on the market. A change in mindset like this typically takes a generation and so even though this change in weapons really happened around the turn of the century, the people in power have mostly retired and the new generation now understands this reality.
Do we know why the US government did not promise to buy chips but to give tax breaks (or investment thereof)? Wouldn't promise to buy create a better incentive to the manufacturers?
There's just such a big shift between parties right now that when the current admin is done, you're not gonna know what to expect with the next. Especially with something that's more policy (purchase orders) than law (taxes). Better to just codify the benefits.
They'll be flown from the US to Taiwan for packaging, at least until packaging services exist here. Then they'll be flown to China, Southeast Asia, India, or possibly Brazil for final assembly into an iPhone or computer, at least until lower cost assembly plants are built here or someplace cheaper like Mexico.
I'm not interested what Apple says. What they do in FB is so many posts they posting what ever they like even posts of almost naked women and girls, looks like prostitution in FB. They are saying is their right to do that and their policy. It is disgusting thing in my point of view. Used to be postings of my friends now is totally disaster.
Maybe this is the discussion worth having. Taiwanese engineers competed to get into TSMC. Their management practically lived in the factory to solve production issues when needed. The local workers in the Arizona factory said the pay was pretty good per another comment. Yet somehow we thought that we were slaving the labors? What is the fundamental difference here? Personally, if I were a worker who could find just a service job that pays $30K a year or less, I'd kill to work for TSMC for $50K+/year and learn everything I can about chip manufacturing in my capacity. It would be proud to do it, and I wouldn't mind some overtime.
And I'm not sure why this got downvoted. Not that it matters, but I'm very curious about why people were not happy with the questions. My fundamental belief is that if someone chose to accept an offer and then work hard, it's not slavery but free will. But well, I guess American culture is interesting in the regard. If I study STEM hard in school, I'll be a "teacher's pet" or a nerd who knows only "how to cram". On the other hand, if I free throw under a hoop 4000 times a day, I'm DA man and it's worth the highest praise on the level of "have you seen the LA of 4:00am". Or if I'm a banker or a startup employee who worked 100hr+, I'm building the future of the US, yet if I worked in a fab 996 on my own will, I'll be a slave?
- Over 50% of the workers flew in from Taiwan to work on this plant and make these chips.
- The chips still need to fly back to Taiwan to be packaged as there are no facilities here with such a capability.
Made in america is a hard sell. But at least showing the glaring STEM field gap in the U.S. is a start to finally addressing the brain drain.
The 'brain drain' (as you refer to it) stems from intelligent/motivated grads in the US for the last two decades (at least) pursuing more lucrative fields like finance and adtech (re: Google, Facebook). Or some pursue management route (attending big MBA schools and switching to management roles where they climb corporate ladder). In other words, there are not a lot of college/grad students who want to pursue traditional engineering routes in the US.
I myself was an electrical engineering (EE) major until I switched to computer science in my third (junior) year of college because like a friend of mine at the time told me, "<my name>, if you don't major in computer science, you will not be able to find a job easily after graduation". He was right. All of my former college friends in EE ended up pursuing programming jobs (a few of them now works for FAANG; I used to work for one but left a year ago due to RTO). That is why the US has no sufficient personnel to do traditional engineering jobs and we have shipped off a lot of those to foreign countries.
Everyone I know that was in EE falls into two camps basically:
1. Became web developers
2. Work in Defense or some other regulated industry that has protections from being outsourced to China
22 replies →
It's not even brain drain, America's dominance came from the fact that for nearly a century the brightest people in the world were willing to give up everything to come here. That is no longer the case. Today's Einstein probably isn't going to immigrate here.
58 replies →
Alternate explanation: electrical engineering is actually really hard and some parts of computer science look comparatively easier. Plus coding is startups is cool, EE is still nerd as in Nerd.
13 replies →
> I myself was an electrical engineering (EE) major until I switched to computer science in my third (junior) year of college because like a friend of mine at the time told me, "<my name>, if you don't major in computer science, you will not be able to find a job easily after graduation". He was right. All of my former college friends in EE ended up pursuing programming jobs (a few of them now works for FAANG; I used to work for one but left a year ago due to RTO).
Nothing against you looking out for your future, but this is exactly what I describe to people when I say the industry has changed. It used to be nerds who were very passionate. Now it’s full of people who are just doing a job.
Hit the nail on the head. I went to the College of Nanoscale Science and Engineering in Albany for a master's in "nanoscale engineering" which essentially boiled down to a master's in being a fab line manager. I finished the degree since it was only a 3 semester program and I was getting paid for research work, but almost immediately after chatting with alumns that went to go work at IBM/Intel/etc it was pretty clear that software engineering was a much more lucrative and less stressful career.
Definitely true, as there weren’t EE jobs here. Now that we’re moving chip manufacturing back, and with programming job market being saturated, perhaps it will shift and EE will pay more due to being more in demand
7 replies →
I think your explanation about large numbers of motivated students pursuing lucrative Non-STEM degrees is incomplete without mentioning the cost of an undergraduate and graduate STEM education in the USA.
The most critical shortages of STEM graduates are in roles requiring advanced degrees. Your median undergraduate education (~$40k) and median graduate education (~$60k) saddles students with approximately $100k in unforgivable student debt! Never mind the years lost that one could otherwise be working. So it’s no wonder students are motivated by the ROI of their degrees, it’s why I chose Computer Engineering over Electrical Engineering.
These are expensive STEM degrees which students on visas are all too willing to pay for a chance at a residency and a pathway to citizenship. So no wonder the majority of undergraduate and graduate STEM students are foreign born in the US. The ROI is not worth it for the debt. We don’t have enough need based scholarships available to finance the STEM graduates this country claims it needs.
Really appreciate this comment and perspective! In the larger context of immigration and brain drain in other countries, how the US also has one, but of a different kind. Ultimately, it's a loss of potential. I'd somewhat disagree with the directionality of the correlative/causal relation, though. But what can be said is that the US also experiences a knowledge drain towards plainly lucrative jobs. I'd wager that it was/is a cyclical effect that just worsened over the decades and that neither engineers moving to fintech nor low-paying engineering jobs were/are the sole reason.
My hot take as to the reason EE is a bit of a dead end in the US is that the options outside of the handful of primary employers are limited. It is very capital intensive to run a semiconductor fab, design chips or assemble electronics at scale. Therefore the employer has all of the leverage. The equipment and/or factory worker infrastructure comes first and the engineering teams are just a cog.
Compare that to having all the degrees of freedom as a computer science student to start up a niche mobile app or internet based niche service after working at FAANG for 5-6 years. Even AI infrastructure will eventually go down in price making niche AI first startups a possibility. In finance its the same, as a post i-banker you have the option to start a boutique fund, a niche fintech or just invest your own savings.
What you said seems contradictory. You open with the premise that intelligent youth go the finance / CS / MBA path instead of engineering and then say that those who do go into traditional engineering can’t find jobs. Couldn’t it be that people don’t go into engineering because there aren’t any jobs? Wouldn’t the lack of jobs explain the low salaries and thus the preference for more high paying alternatives?
Your argument doesnt really make sense : there are no EE jobs in the use, therefore no one wants to pursue EE jobs, therefire there are no EE jobs.
I read the main problem with hiring chip factory workers in Arizona was the factory just didnt pay enough for the long hours demanded. I looked up the median salary and its only 50k so I'm assuming it's not crazy skilled labor (e.g. brain drain). Taiwanese workers just seem more willing to do it.
I spoke to a Taiwanese person and apparently the salaries there are actually quite good, even by western standards (normal ones; not SF). The downside is they have very very long hours (996, barely any holiday, etc.).
29 replies →
50k is just a step above McDonalds these days in a lot of areas. Sure minimum wage might be $15k, but realistically nobody pays that little except in very rural areas (if you need a small number of low skilled employees a small rural town is a perfect spot to build - but if you need more than a small number they can't provide more at any price - you will pay more in the city but there are a lot more people around if you need more)
10 replies →
...just seem more willing to do it
That's why manufacturing offshored in the first place, companies feel they're receiving better value for money on wages elsewhere for this kind of work (and these days not to mention more & larger facilities, proximity to component sources, and a strong ecosystem of supporting and complimentary facilities).
4 replies →
Cost of living can be a lot lower in Taiwan, if your property is already paid off.
Unfortunately housing is super overpriced, due to the Asian mentality resulting in high property ownership.
Real estate is always the monkey wrench in the gears of capitalism because of high necessity yet limited supply.
9 replies →
The chips still need to fly back to Taiwan to be packaged as packaging partner Amkor's facility in Arizona won't be ready until 2027*. I'm not sure the cause of the delta but it could be in part because Fab 21 got back on schedule rather impressively following earlier delays.
* updated to reflect newer article that Amkor's facility is delayed beyond late-2025
The hardest part is making the chips, no?
Packaging facilities cost ~20% of a fab, right?
Naively, I'm assuming packaging is also not as complicate and difficult as fabrication.
Surely if they can build a fab in the US, they can build packaging facilities, too.
Rome wasn't built in a day.
1 reply →
I was about to say, surly at some point in the near future the USA will introduce this capability. Shame they did not match each other in completion time.
5 replies →
[dead]
Also a lot of US STEM grads have their skills wasted in unproductive fields, like the ad business.
the internet ad industry is raking billions from all over the world into the USA, how can you call that unproductive.
57 replies →
If it's so unproductive why does it pay so well?
15 replies →
For a new factory with a new entry into the local market it makes perfect sense to bring in experienced workers for knowledge transfer. This is more an issue if a decade later this is still how things are done.
Back when American companies were offshoring, the initial start up teams were comprised of a lot of Americans who would do commissioning and initial ramp ups while training up the foreign workers. It's a lot easier to train people on a production line that is proven to work.
Problem is, those jobs in emerging markets were desirable compared to other jobs (for pay and opportunities), which helped with talent growth. These factory jobs, in comparison to other jobs, aren’t that desirable.
2 replies →
> STEM field gap
STEM salary gap
I suspect the Taiwan workers have on average much lower salaries.
Yes, roughly speaking 1:4 compared to California.
Edit: This is not news. This (combined with their higher EE education) is why Taiwan won IBM PC-clone-related manufacturing in the 80s. And why they now have TSMC.
28 replies →
That's really a training issue.
Making chips isn't something you learn the details of at University. You can take all the classes you want in advanced semiconductor techniques but the simple fact is University level manufacturing is nowhere close to SOTA.
Basically, you need fab workers to spend time in Taiwan/China, and then return to USA. It's the same model that most foreign students use at schools in USA/Canada. Get USA/Canadian name brand school on resume, learn english, and go back to home country = profit.
Re the first point: Why do you think it is so difficult to transfer chip production off Taiwan?
I don’t think this is about salaries. Nor is this about facilities.
This is about process know-how. And it’s currently not available outside of Taiwan. I’m glad we’re finally starting to transfer knowledge. It will take a couple more years.
How do we know there is knowledge transfer?
If I were Taiwan/TSMC, I would protect my trade secrets as if my life depended on it (which may actually be true).
2 replies →
> glaring STEM field gap in the U.S.
There is no such gap. The jobs do not pay Americans enough to tolerate the conditions.
And the few people who tolerate such conditions are already employed by game development companies.
Does anyone know the general path to get involved in this? Perhaps its romantic, but this seems important, it seems hard, and it seems like something I can be proud of working on (as opposed to maximizing ad clicks). I'm just a SWE w/ a Comp sci degree, so what's the entry-point here?
Your entry point is a masters and probably Phd in Electrical Engineering, specializing in some aspect of semiconductor manufacturing. It’s definitely not CS.
3 replies →
EDA software has some of the most amazing algorithms. I'm always surprised more CS people aren't into it.
You can find many great opensource projects here: https://theopenroadproject.org
But to get some context, and try out the flow and how everything works together, start here: https://tinytapeout.com
I'm not too sure but I would assume there's going to be faster turn prototype chips in the USA now? Is packaging needed to prove a prototype? Can we start buying IP blocks and make our own ICs? I'd love a MCU with built in IMU and wide range LDO, not sure if that's possible all on the same node.
There's going to be some niches opening as a result of this IMO.
EDA software?
1 reply →
[dead]
it's first step. you gotta do something to bootstrap, solve chicken-egg problem. From what I can see around me, the "made in america" is a no joke branding. a lot of pppl going tobuyjust because of that. and may even consider it as social status and their policial support.
The Purism Librem 5 phone is very expensive and unfortunately not that popular. Haven't met anyone who uses one yet
2 replies →
You have to walk before you can run.
You have to crawl before you can walk. Apparently this is where we are at.
Sure, but this is how a supply chain gets bootstrapped. All those factories in China didn't magically appear one day. Just like they didn't appear when Apple started moving operations to Vietnam. You start piecemeal and build out.
I have two kids in grade school and middle school and I see why we have a STEM gap. I have to constantly correct the learning at home in math. Also, I think it's fair to assume that in Korea, Japan, Taiwan, and China the school kids are actually put on an academic grindset unlike here where there is such little academic rigor or discipline being enforced by the school it makes sense why the k-12 education numbers are as bad as they are in the USA.
It might be worth getting up in front of the kids in middle school + and saying "Hey you're in competition at a global scale here. You're going to have to work your butts off to stay relevant."
isn't packaging tech mostly from american companies like applied mat/lam research? or am I missing something?
Maybe that's how US is going to have enough STEM talents -- just like WWI and WWII, take as many talents as possible when the other parts of the world are in shit.
The scenario that we’re going to be able to fight a war with another first world power, where we will attack their infrastructure but ours will be left untouched, seems unlikely.
3 replies →
I have no specific info regarding this plant, but for anyone who never experienced this: flying in people from other plants at the starts (and all 3rd party vendors for a hypercare phase at launch) seems pretty normal.
If they have to keep staffing it that way, that's different.
this is how chick-fil-a does it
Having a STEM degree isn’t a substitute for real world experience in a production facility.
Clustering is a feedback loop where production creates people with experience in production, something needs to kickstart that process.
> - The chips still need to fly back to Taiwan to be packaged as there are no facilities here with such a capability.
This seems to be a much more achievable barrier to work around than not having a fab.
I think people are missing something, training.
It's a new fab, and people need to be trained on current processes and work roles. If you have a skilled work force, you use them to train.
can you really say the chip was made in America when it is only the die wafer which was made there and the rest was made and assembled in Taiwan?
> The chips still need to fly back to Taiwan
The planet burned, but at least we made a few chips in America.
you can fly a few hundred million dollars worth of chips in a single flight. You need not be concerned. The impact from temu shipments is several orders of magnitude higher.
1 reply →
>50% of the workers flew in from Taiwan to work on this plant
I wonder what % of work they did.
It's a start
brain drain from where? thought a problem is influx of workers into us although more for software not sure of chip tech
made in america is also a federally defined standard that these chips categorically fail to meet. assembled in the united states is more appropriate, and even then if you didnt hire americans to do it, what was the point?
this is starting to feel like the best of intentions that has spiraled into a political theatricality where close-enough will be good-enough.
given the current state of declining US college enrollment, the affordability crisis of college, the growing wage gap, the failure of the minimum wage to keep up with the cost of living, and the failure to reform predatory US student lending practices I do not see how the US will in the next 25 years ever manage to curate the type of braintrust for which it was once renowned across the globe.
This is so disconnected from reality. They've gone from breaking ground to replicating one of the most advanced fabrication processes in the history of the world _at scale_ in about 4 years, but they'll be sending the dies off for packaging while their packaging partner comes online so its just political theatre?
Also, over half of the employees are local hires and the ratio will increase as more of the fab spins up. IMO it would be much worse political theatre to delay and balloon the cost of the project by forcing TSMC to exclusively use a workforce that has no experience with the companies tools and processes.
Off topic but currently relevant:
Over 50% of the workers flew in from Taiwan to work on this plant and make these chips.
Those are the 50% we’re willing to bring in no questions asked via any visa program.
Not the elusive Java developer.
Seems like this is actually happening.
I saw so many predictions of how this couldn't happen and "yeah but" ... but it seems to be happening for the most part.
Indeed. It's just bullshit, propaganda.
There's simply no real reason we can't have a deep and robust manufacturing base in America. Well except for the fact that some specific people made a whole lot of money while letting it fall apart, and have paid for decades of media relations trying to convince everyone otherwise.
If you're reading this statement I just made and want to instinctively disagree with me, start by interrogating your own opinion. Why do you think America can't compete with China, for example, over the long term? What "well everyone knows" facts are you using to create that opinion that you don't have any first hand relationship to.
> Why do you think America can't compete with China, for example, over the long term?
Not saying I necessarily disagree with you, but just to give an example, the US has considerably better labor practices and labor laws than China. It's not perfect but there are protections about making sure people are paid what they're owed, how much you are allowed to work someone, safety protocols, etc. All of those things could, in theory, cost more money and make labor more expensive.
Compare this to nations that don't have the same work protections, where they can pay people peanuts and have them work much longer shifts with effectively no protection (e.g. Foxconn in China [1]).
This might translate to decreased cost, and Americans have made it excruciatingly clear that we're apparently fine with slave labor as long as it doesn't happen within the US.
[1] https://www.dissentmagazine.org/article/chinese-workers-foxc...
7 replies →
> Well except for the fact that some specific people made a whole lot of money while letting it fall apart, and have paid for decades of media relations trying to convince everyone otherwise.
Who are some of these people?
1 reply →
I like the idea of made in America and bringing manufacturing self sufficiency to the US. But I don’t like the idea of reducing dependency on Taiwan, which makes it so that the world may ignore their plight in face of increasing aggression from China. The CCP is an authoritarian dictatorial government that seeks illegitimate control over Hong Kong, Taiwan, Tibet, Xinjiang, and other areas. They need to be stopped and the solution isn’t to remove incentives to defend those areas.
There's a need to be pragmatic here; In the event of any kinetic Chinese aggression, TSMC (and other co's) fabs are going to be rendered inoperable, regardless of how well executed a US response is.
The world will will not be able to help with their plight, just as it was not with Ukraine and more recently Palestine. Might as well secure the supply chain.
We can do lots of things to help. But we need to get our military operational in that case.
Protecting Taiwan would mean WWIII. I hope the West and the Taiwanese figure this out ASAP and start moving as many people as possible out of there, and destroy all the fabs.
China is building more warships than the rest of the world combined. And NATO can't even recruit enough people to man their existing fleet.
But it seems the West is a victim of its own propaganda so there's no political will.
These chips are still sent to Taiwan for packing, so it's a good step but not a complete step.
Until 2027, yes.
https://www.tomshardware.com/pc-components/gpus/tsmc-is-repo...
"TSMC does not have an advanced packaging facility in the U.S., and its partner Amkor will only start packaging chips in Arizona in 2027. As a result, Blackwell AI silicon produced in Arizona will need to be shipped back to Taiwan for final assembly, as all of TSMC's CoWoS packaging capacity remains in Taiwan."
Given that there may be a 25% chance that China invades Taiwan by 2030, having the ability to package SOTA chips in the US by 2027 seems "soon enough".
14 replies →
what is involved in the packaging process ? I believe they don't ship fully assembled chips to Taiwan only to be put in a pretty box ?
"Packaging" in this context means taking the wafer of compute die (made in Arizona), dicing it up into individual die, mounting it onto a silicon interposer (an even bigger die, no idea where that's made, but probably taiwan) along with a bunch of HBM die, then mounting that Si interposer on a somewhat larger, very fine-pitched circuit board ('substrate') that is essentially a breakout for power and high-speed I/O from the compute die. That thing is the packaged 'CoWoS' system, where CoWoS==Chip-on-wafer-on-substrate, that eventually gets attached to a 'normal' PCB.
2 replies →
I'm making an educated guess but probably the cutting of chips from the wafers, placing them into the appropriate ceramic socket types (DIP, BFGA, SMD etc), soldering the line wires from chip to pin, encasing the chip, etc.
1 reply →
I believe packaging in this context means taking the raw silicon dies and assembling them into a package which can be soldered onto a PCB (or put in a socket, but Apple doesn't socket anything).
Believe it or not, sending them overseas just to be put in a box actually can be cost-effective. Like with those pears: "grown in Argentina, packaged in Thailand, sold in UK" https://www.birminghamfoodcouncil.org/2022/01/16/part-i-pear...
I think "packaging" here refers to the process of putting the silicon die in its plastic casing and connecting the die's pad to the case's pins, see https://en.wikipedia.org/wiki/Integrated_circuit_packaging
https://www.youtube.com/watch?v=-egYoxajTz0
What does "packaging" mean in this context? I'm a total n00b when it comes to chips.
What comes to my mind is wrapping a piece of electronics in some bubble wrap and cardboard, which doesn't sound that hard...
How does this make any financial sense?
The machines and processes needed to package the individual integrated circuits are fantastically expensive but the margins are so low in that step that it's only profitable at massive scales.
So you put the fantastically expensive machines near where most of the customers are and most of the customers are in Asia.
Works the same way with fiber optic cables. Making the long skinny bits is hard and high-margin. Actually turning them into cables is easy and low-margin.
So Corning makes huge spools of fiber optic cable in Arizona, North Carolina, and New York (I think) and ships it off to Taiwan and China where it is made into the cables that you plug into stuff.
Marine shipping is just about the most fuel efficient way of moving things between any two places, by a lot. A 100,000 dwt ship can get 1050 miles per gallon per ton of cargo. It takes about a teaspoon full of fuel to move an iPhone sized device across the pacific when I ran the numbers last.
8 replies →
This is how most modern supply chains look like.
Plus, chips are small in size and cost a lot so you can fit a lot in a container. Per unit shipping costs probably come out to be pretty low. Especially when compared to the political costs and risks associated with not onshoring.
1 reply →
These are literally microchips. Tens of thousands of dollars of value in each gram.
Shipping cost is fundamentally irrelevant, you can put $100MM worth on a direct flight and have room left over for your family and friends.
7 replies →
I'm suprised they can't ship (flat) packaging that could be used in Arizona with a simple assembly line.
If they had that packaging design then for this to make financial sense the two way shipping (and loading, unloading, custom clearance etc) would have to be less than shipping the packaging, the setup cost per unit cost of putting the chip in a box
6 replies →
Made using which process? The article doesn't mention this.
https://www.tsmc.com/english/dedicatedFoundry/technology/log...
The smallest process they've got up and running right now is 4nm, last I checked
And for the record the A17 Pro chip is 3nm. Used in the iPhone 15 pro and the iPad mini.
But they could make iPhone 14’s and the smaller 15’s.
So which device will these be for then? I thought Apple stuff are always on the cutting edge node.
5 replies →
As an outsider that means somewhere in 2nm-10nm as everyone measures different things or have awfully off-standard rulers.
1 reply →
4nm
I thought Taiwan prohibited export of this kind of know-how? What did I miss?
24 replies →
I keep hearing about a skills gap in the US for fabs, what skills or jobs are actually suffering from this? people with masters in nanotech, compeng, EE?
Perhaps there is a skill gap because nobody actually knows there is a demand? I have no idea what to recommend to people who are trying to choose a college degree.
With my industry in infosec, at least there are certifications one can take, even proper masters degrees these days. In my experience, there is no skills gap in cybersec, despite what CEO's and linkedin-types' sentiment. They just don't want to pay market price for skilled talent. "skills gap" has meant "we need more talent so we can pay less", there is no actual shortage of people who can do the jobs adequately.
Is it different for chip fabrication? and if so, how can regular people work/study to obtain these skills? If I, having read HN for years and reading about the fab process have no clue, how can regular people who don't visit HN?
If you all can help me answer this, I'll try to recruit a few people into pursuing the right career to help meet this demand.
Off topic... Taiwan also machines and heat treats some of the best cutlery steels in the world. Taichung City is famous for this. This is not as delicate a process as producing CPU chips, but it is hard to get right consistently.
Most all major cutlery companies have product lines that are produced solely in Taiwan (Spyderco, Cold Steel, Demko, etc.)
It would be nice to see Taiwanese steel industy move some production to the US as well.
Buck Knives at least, mostly manufacture in the U.S., and their 110 model at least still arrives shaving sharp and keeps a decent edge.
High-quality knives come from proper metallurgy, especially as it relates to proper hardening steps. If you don't get these things exactly right, the best machining on earth is not going to produce even mediocre knives.
Sorry, don't think that's a national security priority.
Funny enough, Fab 21 was announced in May 2020 and completed construction in July 2022, a month before the Chips Act was signed.
The announcement of this plant coincided with the announcement of the Endless Frontier Act and CHIPS for America act, which is what eventually became the bill we call CHIPS and Science Act.
This plant was the foundation that the CHIPS act was built upon. The Secretary of State had to secure an agreement with TSMC to build this fab before the bills could be drafted, as a lot of the recipients of the funding are suppliers for this plant.
It is completely truthful to assert that this is the result of the CHIPS act. Congress agreed to introduce the bills as a result of TSMC's agreement to build the fab in Arizona. If you have to avoid giving Biden credit, then you can point out that it was Trump's SoS who negotiated this original agreement.
I agree that the CHIPS Act was likely contingent on someone showing semiconductor manufacturing could actually be onshored. I’m not sure I buy that TSMC’s investment was contingent on an Act that was contingent on them investing in the first place. It’s not like TSMC was ever going to get a check to just reimburse themselves. Even now, their subsidies are only for new plans.
Intel, on the other hand, is a great example of a how a company dependent on the government funding for semiconductor manufacturing behaves. Heck, look at the Foxconn debacle; companies prefer incentives up front.
If you remember, TSMC had the immediate fear of losing ~15% of their revenue with the Huawei export ban. I wouldn’t be surprised if that influenced their decision to cozy up to America.
What makes you think construction was completed in July 2022? The shell of Phase 1 may have been completed, but even now the construction continues in Phase 1B and Phase 2.
I’m going off the purely structural construction of the first fab. There’s a timeline on TSMC’s site.
Half of the works are from Taiwan All machines were imported to build the factory. USA can't do anythings without immigrants. China was able to develop its own chip factory without immigrants and without buying machines (because USA blocked the 'free market')
USA lost.
[dead]
This is really exciting. It'd be awesome if the rebirth of American industrialism was tech hardware driven. It sounds like this being mass production ready is still a few years off, but kudos to Apple and TSMC for working to make this happen.
As an european, all I wonder is if this will make Apple devices even more expensive.
My guess is no, it won't. This is US taxpayer money being used to increase the manufacturing capacity available to the market so that the US has domestic manufacturing when stuff goes sideways. A similar thing regularly occurs with auto manufacturing and manufacturing in country A usually frees up capacity for other countries, resulting in slightly lower prices.
What could happen is that once the US has manufacturing capacity it decides to tariff imported chips, causing your country to retroactively do the same. This is decades away, and the US has a problem sourcing chips it can trust right now, so it's not currently on the radar. It's not something I'm going to worry about.
Viewed through a pessimistic eye, the US finally is realizing that its arms production critically relies on chip production and it can't says its chips are US made when selling arms on the market. A change in mindset like this typically takes a generation and so even though this change in weapons really happened around the turn of the century, the people in power have mostly retired and the new generation now understands this reality.
Is this the first “Made in USA” chip in Apple devices since the Fishkill PPC 970?
Weren’t the Intel CPUs made in the US?
Oops, forgot about Arizona.
How hard will this be to scale to up 50% of Taiwan production into the US?
Hard, given how many Taiwanese workers they had to bring in, and how all the dies has to go back to Taiwan for packaging.
Do we know why the US government did not promise to buy chips but to give tax breaks (or investment thereof)? Wouldn't promise to buy create a better incentive to the manufacturers?
There's just such a big shift between parties right now that when the current admin is done, you're not gonna know what to expect with the next. Especially with something that's more policy (purchase orders) than law (taxes). Better to just codify the benefits.
3nm? 5nm? What chips are being made? A chip isn’t a chip
If memory serves me right, it's the Apple S8 chip used in their watches, built on a 7nm process.
Surprised they're going to apple rather than say military purposes
They'll be flown from the US to Taiwan for packaging, at least until packaging services exist here. Then they'll be flown to China, Southeast Asia, India, or possibly Brazil for final assembly into an iPhone or computer, at least until lower cost assembly plants are built here or someplace cheaper like Mexico.
Your comment is correct but this was already stated in earlier comments. This is probably why you are being downvoted.
I'm not interested what Apple says. What they do in FB is so many posts they posting what ever they like even posts of almost naked women and girls, looks like prostitution in FB. They are saying is their right to do that and their policy. It is disgusting thing in my point of view. Used to be postings of my friends now is totally disaster.
Imagine TSMC not getting US funds to bring over a Taiwanese workforce large enough to result in "Little Taiwan" being constructed in the desert.
Taiwan exodus in 3..2..1
[dead]
[dead]
[dead]
[dead]
[dead]
[dead]
with required NSA backdoor of course.
Make America good at slave labor again basically
Maybe this is the discussion worth having. Taiwanese engineers competed to get into TSMC. Their management practically lived in the factory to solve production issues when needed. The local workers in the Arizona factory said the pay was pretty good per another comment. Yet somehow we thought that we were slaving the labors? What is the fundamental difference here? Personally, if I were a worker who could find just a service job that pays $30K a year or less, I'd kill to work for TSMC for $50K+/year and learn everything I can about chip manufacturing in my capacity. It would be proud to do it, and I wouldn't mind some overtime.
And I'm not sure why this got downvoted. Not that it matters, but I'm very curious about why people were not happy with the questions. My fundamental belief is that if someone chose to accept an offer and then work hard, it's not slavery but free will. But well, I guess American culture is interesting in the regard. If I study STEM hard in school, I'll be a "teacher's pet" or a nerd who knows only "how to cram". On the other hand, if I free throw under a hoop 4000 times a day, I'm DA man and it's worth the highest praise on the level of "have you seen the LA of 4:00am". Or if I'm a banker or a startup employee who worked 100hr+, I'm building the future of the US, yet if I worked in a fab 996 on my own will, I'll be a slave?
1 reply →
This is only the first (significant) step for the american continent to be able to build cutting edge chips (again).