How real is America’s chipmaking renaissance?
economist.com
economist.com
https://www.ti.com/about-ti/company/ti-at-a-glance/manufactu...
https://news.ti.com/blog/2022/09/29/tis-new-300-millimeter-w...
https://dallasinnovates.com/finisar-considers-3-billion-semi...
https://www.dallasnews.com/business/jobs/2022/12/07/sherman-...
https://www.samsung.com/us/sas/Taylor
> Texas has led the country in semiconductor exports for 11 straight years, and has shown a commitment to expanding the industry.
https://www.texastribune.org/2022/07/26/texas-chips-act-semi...
Definitely a lot of heavy construction.
Hello from Richardson.
I used to work in semiconductors and transitioned to software when I realized how precarious that was.
This was years ago and not TX. I'm sure it has gotten slightly better but what I see on glassdoor still makes me very glad I listened to the job hunting postdocs in nanofab class and steered clear of this career path.
I wasn't actually suggesting that I return to my former employer, although the thought has crossed my mind...
My reasoning for a move would be about all of the higher-order consequences of having well-compensated technology enthusiasts imported by the thousands.
No worries, I was picturing a young person setting out to 'make their fortune' by moving across the country to make it. I just wanted to share the perspective I left the field with.
That doesn't mean it isn't possible to make it work!
From a North Texas perspective, the communities were already built up. The first integrated circuit was made in North Texas. DFW has some of the highest number of fabs per square mile, tons of small producers around along with some of the big behemoths.
North Texas has been a pretty big semiconductor manufacturer since semiconductor manufacturing has been a thing.
For a historical context of the history of the community of semiconductors in North Texas, JFK was shot on his way going to give a speech at what would become UT Dallas, a school whose entire history is focused on semiconductors.
You’ll work 60 hours a week for shitty pay, and be the Yoda of some tool. In 10 years there will be a new magic tool. Then you start learning the Starbucks menu.
You'll see salaries of around $70-110k in very expensive metros like the Bay, Portland, Dallas, Austin, etc.
Salaries just haven't kept up in the EE space.
They'll probably have a laugh while they're out boating.
I think there might be a bimodal distribution within EE salaries where either you work at an Nvidia/Google/Apple/AMD/TI type company which pays competitively and others which pay subpar wages.
Maybe this is heavily dependent on the subset of EE as well (eg. I'd assume the profit margins for a fabless GPU company like Nvidia are much higher than a company building/desigining ASICs for Telcom usecases like KLA).
That said, you seem to be from North Texas/DFW and EE wages and cost of living seem much more managable compared to those offered in the Bay Area and Portland.
Median home price in Richardson: $478K
Median home price in Garland: $301k
Median home price in Sherman: $308k
Median home price in The Colony: $411k
Median home price in Plano: $520k
Which places are you more likely to be able to own your home with ~$100k/yr income?
I'm just pointing out that EE salaries within the US are depressed due to the (relatively) low wages within DFW (which by DFW levels are competitive).
At the end of the day, you guys are definetly being underpaid compared to SWEs as well as certain specialities of EE (take a look at Nvidia's mid-level TC in the Bay verus TI's in DFW).
DFW has always been a major EE hub in the US, but SV and Portland are similarly sized EE industry hubs and their wages have been depressed.
Since DFW represents around 30-40% of the EE industry within the US, it's a major portion but plenty of industry share is still stuck in higher CoL areas, which incentivizes talent to either switch to SWE or outsource jobs to SK, Taiwan, PRC, India, and Israel where such salaries are more competitive
The pecking order was established in the parking lot. The execs drove BMWs and Teslas, the tenants varied. We were in a special lot.
The worker bees in the fabs drove older Hyundais and Nissans. Except for the trades - fabs require a lot of plumbing and electrical contractors, and those guys were all union, all minting overtime, and all driving fancy pickup trucks.
Everyone in a tech field has to retain the ability to learn new things over time. AI might not take our jobs but it will likely change them. I guess it sucks if you have a fear of learning. Admittedly, at first learning a new thing can be kind of uncomfortable, like getting into a cold swimming pool. But once you get over the initial shock and start moving around, it's not so bad.
It's not the worker's fault they can't learn the new tool (they absolutely can learn it). It's capitalism's fault for valuing workers only for the cost of their labor and nothing else.
Sigh. The value you're talking about here is the cost of their labour. This is like a business owner saying that employees should value their jobs differently to their salary.
You're right in the sense that capitalism is fundamentally "people make agreements and decisions within what's available to them" but what's the actual issue you want capitalism to solve, that doesn't create one of the massive downsides that any of capitalism's former rivals has?
E.g. monarchy can reward loyalty, but then so can capitalism. Monarchy can just also tell you to go die on a field if you don't kill the other king's people.
What makes you think that capitalism won't do the same to you? Capitalist[1] societies engage in pointless, bloody wars all the time.
[1] And communist, and other -ist.
Unlike monarchy/communism/fuedalism etc where power and resource allocation are by necessity mostly centralised, capitalism can (and wants to) be divorced from power, and only do resource allocation. Constant realigning it with power (e.g. regulatory capture) is what corrupts it.
The sort of (limited) power capitalism creates is by creating value in some way: labour, capital, contacts, resources. It's not derived from strength, unlike the other methods.
Of course a government, that is fundamentally derived from strength, and must be, can still fight wars. It can do it much better in fact because its economy was organised by the most efficient manner anyone's ever discovered, and so there's a surplus to tax and an efficient economy to flow requirements through.
It has been shown time and time again that the majority of the powerful players in a Capitalist economy inherited their wealth or opportunities.
Capitalism can only function in an economy if a government allows it by setting up the necessary structures and allowing private property ownership.
Society can certainly be Capitalist by extension of the economy being Capitalist due to policies and strucutes set up by goverments.
Seeing as it's more or less the de facto world order it should ideally solve global alignment for the long term wellbeing of Earth-borne life
That doesn't align well with efficiency or top-down world political agreements. But it has lifted more people out of poverty faster than any other system in history.
The fact that this problem pops up all over the damn place should tell you something about our relationship to u**ons.
children used to work 10 hours a day and lose hands in machinery. hell, kids are dying in factories now, and the rules just keep getting rolled back further.
miners get black lung or asbestos exposure.
dudes boiled alive in machinery that didn't have fail safes.
company towns that only paid in scrip and then charged you 3x the price of regular goods.
people had to fight and die for an 8 hour workday.
the bourgeoisie programmer types who've never had to work actually dangerous jobs tend to forget just how incredibly brutal working conditions used to be.
Basically, a big, inflexible place is a good unionization candidate. A smaller, nimble org is a lousy place for a union.
"No, we'll hire them away from our competitors. Let one of them invest in training."
Of course this accelerates the business’s desire to adopt the new tool sooner, so they best hurry!
New tool makes a different wavelength of light by shooting high energy beams at molten tin and all your PhD research isn't relevant to that.
It's not like moving from Adobe to Figma, it's like moving from physics to chemistry.
They're either smart or they're not. Which one is it?
If you're in the 1% of top performers, you have the cognitive ability to learn new things.
> If you're in the 1% of top performers, you have the cognitive ability to learn new things.
I had to stop and admire the subtle self-deprecating humour on display between those two statements. Well played!
So, ok, but anyone thinking about entering the field can see that they’ll be left in a lurch when they hit mid career.
We’ve already run this experiment for a generation or two; anyone smart enough to finish an engineering degree can just go do a programming career afterwards, where the work and career advancement is easier. Now we have imported chips and a million JavaScript frameworks.
He was laid off and unemployable in the field. He survived as a Starbucks manager and sold Herbalife for a year while getting some additional education. I was able to help him get a contract as a tech writer to get out of the retail stuff. Now he’s a director level person in a cybersecurity firm.
The one true path is obvious on message boards, but the reality is people find themselves in bad places sometimes.
When the cops are complaining about union-busting...
I can see why a lot of people hate the idea of semiconductor manufacturing. It is the apex of "by the books" while simultaneously exhibiting the most incredible shiny technology on the face of the planet. Being forced to follow all of these rules and procedures to get at the shiny tech rubs a lot of people the wrong way. "What do you mean I can't take my smart phone into the office? Y'all make its memory here." The pay is also arguably pretty bad unless you lock in for life.
I would say you get what you want out of it, just like anything. If you walk in with an entitlement mindset and expect the knowledge to be forced into your brain, you are in for a really shitty ride just as you describe. If you walk in thinking "I'll bust my ass here for 3-5 years and learn what I can", you will have a much better experience.
I didn't even know I wanted to write software full time until I was sitting in a cubicle at Samsung's ATX office for a few years. In my opinion, if you aren't sure what you want to do but you really love all things tech, why the hell not? Try it out. It was a great stepping stone for me.
I didn’t particularly like the job, but I’m grateful for the opportunities that time and effort has given me.
I definitely don't pretend its all just roses though. It being over 100F every day for the past month or so is definitely a big negative, and a good bit of the politics here is getting ugly. We need to continue to expand the public transit in North Texas and make it more useful to reduce car dependency. More dense housing options need to be available to allow people to choose different ways to live and alleviate rising housing costs. There's a lot of stuff that needs to be addressed, but there's movement in positive directions around here.
A lot of the debate about Texas power grid make things sound way worse than it is. In reality Texas is about average when it comes to reliability in the US. Meanwhile we produce the most renewable energy in the country, we enable people to buy 100% renewable electric plans, and our cost of energy is some of the cheapest in the country. There's definitely challenges ahead for the grid and some of the market strategies we have need to be re-thought, but overall I'm not too worried about the grid long-term. Texas isn't alone with its grid challenges though, most states need to address massively aging infrastructure.
Do you have a source for this? Anecdotally, I felt the need to buy a generator when I lived north of Austin, because outages were frequent enough that it seemed warranted. Since moving to North Carolina (just north of Charlotte), I’ve had exactly one outage in a year, and that was on a mild day, just had some equipment failure.
Never once here have I received notice that I needed to reduce consumption to help the grid.
Here they rank it 30/50 in reliability and performance metrics.
If we're sharing anecdotes, I've never lost power for more than a few seconds in the past decade or so, and before that it was pretty much only because of hurricanes or tornadoes in the area.
A roommate of mine once quipped on something: 'that's like saying my crap smells better than yours'.
Feel free to correct me, I'm not an expert, but Texas runs its own state-only independent grid, it got completely pants in the ice storm a few years ago, and at the time a lot of articles said the Texas grid is fundamentally less resilient that other power companies. Is any of that wrong?
However, Texas is loaded with sun/desert and I believe a ton of wind resources, so it has that going for it for next-generation industry.
Surprisingly, you got to thank George Bush for that
He made renewable wind and solar energy a major part of his energy policy when he was Govenor of TX. [0].
In 1999, he passed a law mandating Texas to get 2,000 megawatts of electricity from renewable energy by 2009. This helped spurt the entire renewable industry in TX
[0] - https://www.technologyreview.com/2016/08/29/70295/george-w-b...
McGraw Hill was founded by the Bush family before they decamped for Dallas in the 1920s.
You can get around this with DC, and grids do interconnect with DC power, but that requires more expensive equipment.
I tried reading this IEEE article about the time the Eastern and Western Interconnections were connected into a synchronous system (1967-1975) to learn more about the difficulties with scaling interconnections. It mentioned frequency oscillations causing voltage variations close to the East-West connection (p11), which sounds like the issue you're describing. If that is the issue though, then Texas isn't the size limit for a grid because the Eastern and Western grids exist and are far larger.
https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8594689
Right, in 2021... and in 2011, and in 2003, and in 1989... :p
So, if there is a bubble in the chip supply pipeline, auto manufacture doesn't happen. If there is undersupply of precursor chemicals it takes a while to make vaccines at scale...
What you are talking about makes me think of the very long lead product that is skilled people. A significant proportion of fab employees are genuinely highly skilled workers who need a specialised physics degree and then additional training in order to optimize yields and things like that.
I think they can build the buildings and buy the machines but unless a lot of Taiwanese are willing to get on a plane and move to the US, it is going to take a long time to be able to replace their labour at the required scale.
Edit to add:
I've met a few people who are aware of US military supply chain rules, and assume it means "oh, chinese $doodad must be crap, that's why we don't give them to our troops", when the reality is just as you said. Supply chain bubbles can be massively disruptive, so its a real strategic blunder to outsource critical components to the country we have designated as our number one rival. I'm not saying we're set up for a trade war, but we are well past due to calculate the cost of increasing short-term profits. Is a quarterly decrease in operating costs from consolidating fabs worth risking a massive disruption when that single fab is knocked out? Given the volumes that even small fabs do, I argue that quarterly increase is not at all worth it.
For reference PRC after elevating semi to first-level dicipline in 2018 is pumping out about 30k IC graduates per year. They're still about 200k short, ~520k/720k out of what IC talent 2018 white paper estimated PRC needed for complete semi industry. This is PRC that graduates ~5M STEM with huge population to draw from that bias towards S&T in tertiary. Which leads to the other uncomfortable point that a complete semi industry is more than just fabs, there's a lot of critical semi supply chain stuck in TW / East Asia (SKR/JP) that's going to get disrupted anyway in event of conflict. I don't see any indication US is resolving to reshore the entire semi chain.
Or, gee, let's throw this out there: perhaps if the pay wasn't shit, people would go into the field.
You know, just like computer science over the past couple of decades.
There's a few good things that the university offers which is surprisingly strong. When i went back to prepare for a CS masters (military, self-taught) I was surprised at the quality of the undergraduate degree in preparing students for industry, and how quickly the security program was invested in (INL does a lot of investment here).
All that to say: it's a pretty good school, especially for it's price. I only ever had one bad professor, and it was in the supply chain management department. And it was (is still?) the only raptor biology masters in the country.
I would say exactly the opposite, the vaccines were made quite quickly and there weren't major supply issues.
The Defense Production Act (er... central planning as you put it) was used to say what was going to happen when with money to back it up and... things happened quickly.
DPA wasn't invoked to build you new cars or supply you toilet paper so... those things had their supply issues.
The US actually can do logistics very well, especially when there's somebody willing to foot the bill for big asks. The lack of corruption and general merit-based placement here compared to so many other places combined with a large degree of vertical integration means that we can accomplish quite a lot and is one of the underappreciated cornerstones of our ability to make war (and why nobody wants to fight us).
We are decimating (and then some) Russia's military by giving away second tier equipment to Ukraine, and the amount we're giving away is basically an afterthought. A one-time superpower is scraping the bottom of the barrel for equipment fighting against largely NATO supplied spare equipment.
Unfortunately, US industrial vertical integration now includes foundations placed in global low-labor-cost countries.
The crash vaccine manufacturing scaling program was possible because we were rich enough to throw mountains of money (of which we have the most) at any international supply problems.
If the Atlantic or Pacific had been hostile to commercial shipping, or countries hadn't cooperated, it would have had a very different result.
> We are decimating (and then some) Russia's military by giving away second tier equipment to Ukraine, and the amount we're giving away is basically an afterthought. A one-time superpower is scraping the bottom of the barrel for equipment fighting against largely NATO supplied spare equipment.
"We" are not. Ukrainians are dying and holding their own against a much larger country, with the help of NATO surplus equipment.
20,000-50,000 Ukrainians now have at least one limb amputated as a result of the war. [0] [1]
And Russia's military is not being decimated. It's holding its own with an elastic defense, that's making progress difficult and costly for a Ukrainian army that's trained halfway between Soviet mass doctrine and Western maneuver warfare.
At some point the Ukrainians will hopefully manage to exhaust Russia's logistics sufficient to break through their defense, but that's in no way a sure thing.
[0] https://www.wsj.com/articles/in-ukraine-a-surge-in-amputatio...
[1] https://www.telegraph.co.uk/news/2022/05/19/ukraines-amputee...
1. Russian/Wagner (now admitted to be the same) neocolonialism in African states, most recently Niger. Within days Niger stopped trading Uranium to France. Conveniently, France is the least dependent on Russia for energy and will force the French to buy outside of Niger and then refocus on Russia.
2. The latest naval drone hit on a Russian oil tanker in the Black Sea took place well beyond the expected range of Ukrainian capability, which likely means someone else (Turkey?) helped at least get it into launch range. This coupled with the Ukrainian position that all russian ships, and all ships trading with russia are now targets in the Black Sea is also big for the area.
Currently Russians may not be dying in droves defending their positions. However, the knock on effects for russian military capacity, industrial capacity, and naval capacity are incredible. The longer they flounder and the more they commit to Ukraine, the more their traditional enemies are able to take advantage of their weakness.
If russian gas and oil cannot flow out of the black sea safely, their PMC backed interests in Africa and South American become less effective, and the corresponding governments will look elsewhere for influence/protection/trade.
The naval drone question is especially fascinating.
IMHO, they have to be transported closer to impact site by some other vessel.
Ukraine doesn't have submarines, and Russia has 6(?) Kilo's in the Black Sea, so traditional submarines would be insane in that environment anyway.
And whatever transport method would need to be stealthy with respect to Russia's sensing methods: acoustic, radar, to a lesser degree visual.
Given Ukraine's technical capabilities and traditional shipbuilding expertise, I'd hazard there's a somewhat-stealthy drone carrier boat (probably also unmanned).
They could also be hiding/launching the drones directly from commercial shipping, but that's an awfully big risk for the flag country/ship owner with respect to Russia boarding and searching, or even back tracing launch points.
> IMHO, they have to be transported closer to impact site by some other vessel.
The specifications I have seen is that they have somewhere between a 400km - 800km (I know it's a big spectrum) range.
If they were about mid that, they wouldn't need launching from outside the Ukrainian mainland to hit pretty much anything of military interest.
First of all, other countries managed to develop and produce large amounts of vaccine in very short timeframe during the pandemic. Second of all, the reason nobody wants to fight the US is that the US spends as much on its military as the next ten or so countries. The US makes up 40% of all global military spending.
> We are decimating (and then some) Russia's military by giving away second tier equipment to Ukraine
The total value of NATO aid to Ukraine is on a similar scale as Russia's entire military budget. NATO is not just giving away second-tier equipment. It's giving Ukraine what it thinks Ukraine can use most effectively right away (such as upgraded T-72 tanks, man-portable air-defense and anti-tank systems early on), and some of that is cutting-edge gear (such as the Storm Shadow missile).
> and some of that is cutting-edge gear (such as the Storm Shadow missile).
Storm shadow is 20 years old from first deployment and 29 years since development started. Its replacement is being actively developed since 2017, it’s hardly “cutting edge”.
It’s just new enough that the Russians have trouble countering it.
The dominant ideology in the US, at least until recently, is that there should be no central planning. The invisible hand of the market was all-knowing and would allocate resources far more efficiently than any human planner.
I wouldn't blame central planning, I'd blame the lack of it.
It's an extremely efficient allocator... on a certain timescale, itself determined by the market.
It's an incompetent allocation on different timescales, typically when you get into the 5+ year range.
The ideal blend is when capability is centrally planned, but implementation and tactical scale is market-demanded.
- treating anyone who isn't "white enough" with dignity and without extreme suspicion of being in the country illegally. Driving in texas and not white enough? You might end up in jail if you don't have sufficient proof on you that you're not "an illegal"
- a reliable power grid, because they petulantly do not want to have to follow federal regulations in how they run said grid (also, electrical pricing that doesn't swing by 4 orders of magnitude or more)
- a well-funded public school system with science-based curriculum to develop their children into a skilled workforce capable of critical thinking, being retrained, etc.
- human rights, healthcare, mental health services, etc which highly skilled technical workers tend to strongly prefer/value (and in case nobody told Texas: there is a long, long history of highly skilled electronics engineers, programmers, etc being LGBTQ, autistic, having mental health needs, etc.)
Anyone who is highly skilled in semiconductors could write their own ticket. Why would they go to Texas?
There are only four states that require you have an ID during a police stop and Texas is not one of them. Obviously if you're driving, a driver's license is required. But for a random stop walking down the road? Nope, Texas does not require that you have ID.
To the rest of your post: I honestly don't know where to start. You're making a lot of assumptions that don't hold. Suffice to say that it's a lot harder to prevail in a case against the police than you realize.
People get arrested/thrown into jail for no reason far more often than you apparently think. I know someone (a software developer actually) who was attacked by an undercover officer in a case of mistaken identity. He fought back not knowing the guy was a cop. To prevent him from being able to sue, they charged him with assaulting a police officer. After months of trying to fight it in court, he finally gave up. City dropped the charges as long as he agreed not to sue them.
This is why people are so pissed and why there have been so many protests in the last few years. Everyone is tired of this shit.
That sounds like a massive conspiracy that all media (left, liberal, conservative) are part of.
Texas ranks about middle of the pack by most statistics in electrical grid reliability while also being one of the cheapest places in the US. Its also expanding renewable energy faster than any other state in the US, by a massive margin.
> electrical pricing that doesn't swing by 4 orders of magnitude or more
Clearly shows you don't actually know what you're talking about. The vast majority of households and businesses in Texas are on fixed rate contracts. The price of my electricity swings every three years or so. The headlines you're reading are spot prices; the vast majority of people don't directly pay those.
> a well-funded public school system with science-based curriculum to develop their children into a skilled workforce capable of critical thinking, being retrained, etc
Eh, PreK-12 35th overall according to US News, not great. 25th for higher education, so exactly middle of the pack. There's a lot worse of places to be education-wise. It really varies from school district to school district though. US News ranks the high school my kids will eventually attend in the top 16% of schools nationally. The smaller college I attended was ranked in the top third of all national universities. Not too bad.
> Why would they go to Texas?
As mentioned, loads of opportunity and comparatively low costs of living compared to a lot of other places.
all sorts of roles but middling pay and long hours. not quite the Chinese 996 but taking hints from it; the Taiwanese version, essentially
while China invading Taiwan is not a guarantee, it is certainly highly probable. this risk should be taken into account by the us government
Chinas activities in the South China Sea, the fact they openly say they will integrate Taiwan, by force if necessary and unemployment being 20-30% for young people are just a few red flags.
For me personally—in general, I read the story that the author/publisher intentionally posted for public discussion. While it is possible to circumvent paywalls, I wouldn’t want to go against the will of the author and their publisher.
I think it is weird that we allow discussion of unauthorized copying of text here, but we don’t seem to allow (at least as far as I’ve seen) links for unauthorized copying of, say, movies or music.
I would say it's a little more ethically questionable than ad blocking. Ad blocking is like buying a paper from a newsstand and duct taping over the ads - totally fine, it's my paper and I'll do whatever I want with it. This extension is like hanging around the newsstand reading all the front pages until the owner shoos you away, and then coming back with a fake mustache to keep reading.
I’m very much ideological about being in the camp: you can send me whatever you want, and I’ll render it however I want, and anyone who has a problem that is the problem.
I’m not sure how I feel about impersonating a scraper or switching around my cookies so that the site sends me something else.
Actually, this is legitimately the first question in this area that has gotten me confused in a while; if the tool was impersonating somebody, a specific person with account, that’s just be straightforwardly unauthorized access. Impersonating a category of user that has no actual specific identity, but that they default to allowing access? That seems… I mean, doesn’t really seem like lying exactly, but it also doesn’t seem totally honest. Tricky!
So, IMO, it is more like: if most restaurants gave reviewers preferential treatment and tended to comp their meals, and this was very well known, and you went around telling everybody you were a reviewer. Which also seems dishonest, but it isn’t like you are exploiting some nice little cute good-faith thing like free dessert on your birthday, you are exploiting the restaurant’s attempt to get a better review. Seems, like, dishonest but fair play.
And, I’d love if search engines would stop giving me paywalled sites, haha. So maybe it would be nice if everyone would exploit this loophole, they could close it, and the whole system would get a little more honest. (I’m not doing it unilaterally though).
The trouble is that capturing government subsidies can become a business by itself.
As long as the subsidies are there to restart involvement I think you get the benefits without profitability. If real profitability isn't a medium-to-long-term goal, things could get messed up.
The relevant point is that while the manufacturing portion is real, successful reshoring also requires skilled labor. Skilled labor for fab production is not something US has now and it takes many years to grow; and it is not clear if the US has the desire and the will to foster this growth.
I switched away from electronics early in my career, and it seems like a strict win.
As much as I believe in on-shoring, I'm not sure those issues can, will, or should be fixed.
If I take TSMC, they have:
* $73b revenue
* $32b income
* 73000 employees
(2022 numbers from wikipedia)
That's a really high margin, and a really high revenue per employee. This means there is probably room to accommodate for higher wages.
The concern I have revolves more around the US work culture. Employees tend to switch company frequently. Given how critical institutional knowledge is for running these fabs, this is a handicap.
I also cannot shake the feeling that the lack of regulations regarding working conditions might also be detrimental. Better regulations and compensation rules regarding night shifts or on-call for example could improve things greatly and attract far more workers in this field.
It’s not just cultural and it’s hinted at in the second part of your comment. Lack of regulation in employment means you’re penalised financially for staying in one job. Most employees have little say in their pay and conditions.
Employees wages are rarely commensurate with the value they provide to a firm.
When I've seen people switch firms, the institutional knowledge they brought over was gold.
More than Asia, but same order of magnitude, maybe even the same ballpark for some areas.
I expect most chip design will still be done in Asia and those are the high paying jobs.
/s
On the other hand, EEs either can SWE or can learn quickly, and switching seems like a win in all respects, unless you have a bunny suit kink or something.
I'd be happy if even half of that money actually went into a fab, and only half was wasted on lawyers, fees, studies, and wherever else free money manages to go rather than to actually building things. The timeline doesn't matter, nor does the fraction of on-shore chips made. You created those measurements out of thin air.
The CHIPS Act was marketed as an "AI cold war", and needing to bring high-tech research and production back to the US. Is is actually doing that by incentivizing companies to build leading-edge fabs in the US, or are companies just using it as free money to expand the production capacity of their trailing-edge processes?
Sure, having locally-sourced trailing-edge chips is still better than getting them from overseas, but it doesn't exactly do anything to reach the intended goals.
Getting the enough ASML machines to supply a fab for an advanced process (and they have a backlog that is almost a decade long), starting up the whole complex logistical process to feed the fab (with lots of single source suppliers that are already committed to the existing fab capacity), getting all the necessary personal and skillsets started and getting the fab working, creating all the little feeder companies that are necessary around the fab, all that is hard work. It is is something that the US has largely lost most knowledge of how to do with all the outsourcing of industrial capacity.
And all this has to be done agains the upwind pressures of our service economy. Our politicians are largely lawyers, our financial sector focuses on short term returns, and our companies are rife with lawyers and MBA layers taking their cut out of the necessary investments in manufacturing and engineering. None of this is encouraging for the necessary long term slog to bring back industrial capacity, particularly something with as complex and deep logistical tail as semiconductor manufacturing.
Such as slashing the corporate tax rate (and the offset is to raise the individual tax rate on higher incomes). Which is the Ireland model, and it works extraordinarily well (even more so if you already have a desirable foundation). Cut it far enough and it'll lure more companies and jobs than you can absorb. And meanwhile the corporate income tax is an increasingly trivial component of the US Government's funding, it's no longer very important. If anything we should cut it further (~15%) and very aggressively pull companies, investment, R&D, manufacturing out of foreign nations and into the US. The corporate tax rate is a magnet, the lower you go the stronger it is (and there is no doubt a significant diminishing of returns as you go below a certain % figure).
Increase the standing incentives (not only one-off programs like the CHIPS Act).
It's been a long time since the US did industrial policy.
That's definitely the feeling I get when I see the grants that get given out for Broadband to underserved areas. So much money gets spent on infrastructure for not much actual benefit.
Hopefully these fabs don't follow down that road. If nothing else, it'll be a lot easier to check and see if the objective were achieved, compared to broadband which is inherently much more decentralized.
I've heard from multiple people that they've looked at the map and forced companies to make amendments when they claim to offer service in an area but don't actually.
I hope this at least means there’s no excuse to cut bonuses in upcoming years though.
The renaissance is real, or is about to be.
I just assumed they had processes for eliminating any and all contaminants.
So I'm wondering how much the initial ore purity really matters.
Here's a great video on it:
1) Employers are responsible for employee health coverage, inflating employee costs by a significant degree. The rationale for this model among the investor class is that it makes employees dependent on their employer and limits the power of unions, but for example in Germany, there's free state-supplied health care (paid for by higher taxes on individual and corporate profits).
2) Four decades of neoliberal trade policy and corporate profit maximization have created a situation of fragile global supply chain dependence and a lack of skilled experienced labor in many manufacturing sectors. High-skill manufacturing jobs have been replaced with low-skill service jobs and the brain drain into the financial sector has left the industrial sector in second place.
3) Construction and operation of new manufacturing facilities relies on a robust domestic infrastructure system, but American infrastructure (railroads, electricity grids, ports, etc.) is woefully decrepit in many areas and lags behind China and Europe. Again the reason is nobody wants to pay for it, and the issue is not 'overregulation' it's that it would require higher taxes and corrupt government contractors divert government funds into their own pockets instead of into the projects, resulting in constant cost overruns and low-quality outputs (reminscent with what went on with Iraq and Afghanistan reconstruction contracts).
It turns out worshipping at the feet of Milton Friedman & Co. wasn't such a good idea after all.
A whole lot of people want to pay for it. The problem is that a minority doesn't, and they've been able to ruin things for the rest of us.
> Employers are responsible for employee health coverage, inflating employee costs by a significant degree
> for example in Germany, there's free state-supplied health care (paid for by higher taxes on individual and corporate profits).
TANSTAAFL, just choose your favorite accounting structure. A company with a younger contract workforce (and likely lower spending/risk profile) might find the funding model advantageous.
$1.7 billion expansion in the US towards the end of 2022:
https://www.bmwgroup-werke.com/spartanburg/en/news/2022/BMW-...
Expanding on their already expansive presence in Spartanburg. The X3-X7 and their newest high end vehicle the XM are all built in the US:
https://www.bmwgroup-werke.com/spartanburg/en.html
Mercedes, another company that knows about manufacturing in Europe and the US, is producing some of their newest vehicles in the US, including new electronic vehicles that debuted in 2022, the EQE and EQS. The 2024 Maybach EQS680, with an estimated price of $200k, is also built in the US.
Toyota is spending $2.5 billion to expand their US facilities:
https://pressroom.toyota.com/toyota-announces-2-5-billion-ex...
I had a little luck with archive.org/web, but it's not quite as stream-lined.
For days now Firefox gets stuck in an infinite loop of captchas for all of the archive links posted to HN. Nothing changed at my end.
Why? Ease up a bit on immigration for skilled workers, subsidize however many CHIPs Acts it takes to get the motor running, export a higher standard of living for skilled workers thereby, and margins on the most elite manufacturing sector on earth stay…really friggin high?
How productive does the global economy have to get before paying people well stops being a non-starter?
We didn’t used to fuck around on this: Bell Labs, Western, DARPA, the Apollo Program. Shareholders got a solid, low-beta return, workers got benefits and an education for their kids, and technology improved at an explosive rate.
Anecdotal, of course, but there's some real things being built for sure.
I don't think any of this onshoring, if it ever actually happens, is going to be a net positive for the areas that host the fabs. TSMC is already sending hundreds of their own workers over to America to work on stuff because they aren't happy with the local workforce, I don't see any reason why this will change any time soon. The US is just going to be a place they send their workers to fab chips in plants that got sweet cost/tax breaks, and that they don't care at all about polluting the local environment.
Of course, if you’re just getting started now, the knowledge from 1987 is of little use (not entirely useless though) and indeed drastically different from 2023 SotA.
How about all the other stuff that's required besides the dies themselves... chip substrates? packaging? Passive components? Onshore PCB manufacturers (that are actually affordable)?
We build 10 to 12 layer boards with BGA x86 CPUs. There's a MASSIVE premium on trying to have complex PCBs made in the US. The last time I looked I found a lot of PCB mfgs in Mexico, but they're seemingly owned by the Chinese.
For both Digikey and Mouser, there's no way to even see or filter for Country of origin.
That was a sad moment last year, I am extremely concerned we Zombified our tech sector.
We've got another $50 billion. And another after that. And another after that. As necessary.
See: Ukraine v Russia.
$50 billion ain't what it used to be (and Russia is an economic joke in scale compared to the US, we can shovel billions of dollars into that war endlessly). Its just that the US and China have gotten so massive compared to their former peers, that we can fund at a relative gigantic level quite easily. The US economy is now 50% larger than the EU economy, and six times larger than Japan. US GDP per capita is a shocking $80,000 - well over double that of both the EU and Japan (soon to be double that of France and Britain; or closing in on France and Britain combined to put it another way). $80,000 is equal to the combined GDP per capita of France + Japan.
Things that are difficult for the EU to fund, like $50 billion for chips, are now comparatively easy for the US (or China) to fund. We're only funding the CHIPS Act because that sum is no longer a huge deal to the US. That's a little toy program that Biden was allowed to play with precisely because it's not that big of a deal financially. A medium deal would be $200-$300 billion (eg helping Ukraine defeat Russia across several years); a big deal is $500b-$1 trillion (infrastructure level).
Germany's economy hasn't grown on an inflation adjusted basis in decades. The same is largely true of all of Europe's biggest economies.
US GDP in 2008: $14.7 trillion. Now: $26.8 trillion.
Germany GDP in 2008: $3.7 trillion. Now: $4.3 trillion. And that is a mirror of what has been going on in Europe broadly (with a few exceptions eg the Baltics and Ireland). Since 2008 the US has added economy equal to Germany + France + Britain + Italy.
$50 billion would have been a big deal for the US, two decades ago.
Producing more ammo isn't a serious problem, it's a trivial problem that the Pentagon & Co. are in the process of resolving.
Why Arizona and Texas...?
https://www.nytimes.com/2023/02/22/technology/arizona-chips-...
Basically just told them "Don't worry about it, we'll foot the bill for any infrastructure needed to get you your water"
That said, we don't pull sea water to supply water for industry - desalination is too costly.
We aren't talking about building fabs way out in Abilene are we?
Some other commenter was talking about building fabs in the desert. I think we’re certainly in hypothetical territory.
My point was that the “coastal state” comment doesn’t explain how that magically make clean water plentiful.
Have the power generators applied any of the recommendations in the report commissioned by ERCOT after the 2021 blackouts? When the 2021 blackouts happened, they had not implemented any of the winterization updates recommended by the another report a decade earlier (2011). The outcome of the 2021 blackouts was the legislature allowing utility companies to claw back their "losses" from customers' bills over 10+ years - with no obligation to make changes or investing in equipment to ensure there same cascading failures won't happen again.
Too many theorists decided to claim that "no one would ever risk their economy!" and since it made lines go up globally everyone bought in. Then Russia invaded Ukraine and suddenly the hawks who'd been screaming about supply lines and logistics for years were taken more seriously.
You can't simultaneously threaten military/economic action and also rely on them for major resources. Not to mention how countries like China have essentially warped the market for goods around their artificially cheap labor, and the threat of losing that plus one of the largest markets in the world is severe.
Now granted, we could all still wind up in world war, but to me it looks like the standard stuff.
What's changed is that China continues to rise in power and exert its influence, combined with Russia's invasion of Ukraine making the West antsy about dependencies on China as well as Taiwan (if China invades or blockades Taiwan).
It's ridiculous to criticise China on one hand, and then emulate every move they've made ten/twenty years later.
Critical industries are, well, critical. Securing redundancy for critical things is risk management 101, even on much smaller scale than geopolitics.
But yes; Taiwan will end up losing on two ends: Massive tech transfer to the US. And a loss of access to its biggest market - mainland China.
Regardless outsourcing all manufacture of one of the most critical resources we depend on to a tiny island nation halfway around the world is bad news.