Conor here: According to the Twitter-referenced article in American Affairs Journal and the following piece from OilPrice, it would appear that the Biden Administration/Blob has realized that its efforts to curtail China’s semiconductor industry and build up the US’ aren’t working and believe it’s time to keep doubling down.
This is an excellent read from @pstAsiatech, maybe even THE authoritative article on America’s attempts to curtail China’s semiconductors industry and China’s response to the challenge:https://t.co/cp57Q1p6IM
The high level summary is that the US have – correctly – identified…
— Arnaud Bertrand (@RnaudBertrand) March 4, 2024
By Felicity Bradstock, a freelance writer specializing in energy and finance. Originally published at OilPrice.
- The 2022 bipartisan CHIPS and Science Act provided significant funding for semiconductor research and manufacturing, aiming to strengthen US competitiveness in the global semiconductor market and reduce dependency on imports.
- Despite initial investments, some government representatives and industry experts advocate for a second CHIPS Act (“CHIPS Act 2.0”) to accelerate semiconductor development, attract private investments, and solidify the US position in semiconductor manufacturing.
- While the US government has allocated billions in funding from the first CHIPS Act, discussions continue on the necessity of additional subsidies and financial incentives to compete with major producers like TSMC and to ensure long-term semiconductor manufacturing independence.
As it accelerates its green transition, the U.S. is looking to become a major player in the international semiconductor market, to become more competitive with Asia, the world’s biggest producer and consumer. The 2022 bipartisan CHIPS and Science Act introduced a high level of public funding for research and manufacturing into semiconductors in the U.S. This supports the overarching aims of establishing the country as a major manufacturing hub and strengthening the supply chains that aid the green transition. Now, some government representatives are suggesting the need for a CHIPS Act 2.0 to spur investment further and solidify the position of the U.S. in the global semiconductor market.
The 2022 CHIPS and Science Act provides around $280 billion in funding for research and development into semiconductor technology and manufacturing operations. The funding includes $39 billion in subsidies for U.S.-based chip manufacturing, further supported by tax credits for operational equipment. It also contributes significant funds to the science and technology sector. The Act aims to revitalise domestic manufacturing, create well-paid jobs, strengthen domestic supply chains, and accelerate the industries of the future.
Currently, China is the biggest semiconductor market, purchasing more than 50 percent of the global supply every year. China continues to be highly dependent on semiconductor imports, sourcing many of its chips from the Dutch giant Advanced Semiconductor Materials Lithography (ASML) and the Taiwan Semiconductor Manufacturing Company (TSMC). TSMC produces around 80 to 90 percent of the world’s advanced semiconductors. China also hopes to develop its manufacturing capabilities. In 2023, Chinese companies bought U.S. chipmaking equipment to make advanced semiconductors, despite U.S. policies aimed at restricting China’s import of semiconductor-related technology, to slow its progress in chip production.
The U.S. is attempting to counter China’s dominance in the global semiconductor market by rapidly developing its production capabilities and striving for technological advancements. While the CHIPS Act has gone a long way in establishing the U.S. role in the global semiconductor market, some industry experts believe more is still needed. The United States Secretary of Commerce Gina Raimondo recently emphasised the need for federal subsidies in the industry to enhance the position of the U.S. in the international microchips market. She believes this can best be done through the launch of a second CHIPS Act to spur more funding.
Raimondo said, “I suspect there will have to be – whether you call it CHIPS Two or something else – continued investment if we want to lead the world.” She added, “We fell pretty far. We took our eye off the ball.” The development of a second CHIPS Act could support the construction of new chip foundries and the financing of semiconductor startups. It could also help the U.S. develop its technological capabilities in the field of specialised and advanced chip manufacturing.
However, billions in funding have still yet to be allocated from the first CHIPS Act, with the White House only recently announcing a $5 billion investment in a new chip research initiative (NSTC). In February, the U.S. government awarded $1.5 billion to New York-based chipmaker GlobalFoundries, which was its third and largest grant in the field of semiconductors under the CHIPS Act.
Nevertheless, some industry experts believe that federal funding, alongside private investments in the sector, will help the U.S. achieve chip manufacturing independence within the next two decades. The ambitious policy has also encouraged other world powers to develop similar investment schemes. In 2023, the EU introduced the $46.53-billion European Chips Act to boost competitiveness in the international semiconductor market.
Raimondo was quick to say that she doesn’t expect all semiconductor manufacturing to be situated in the U.S., but she believes the country’s role in the chip industry can expand substantially. “To be clear, we can’t and do not want to make everything in America. We don’t want to make every chip in America. That isn’t a reasonable goal,” Raimondo added. “But we do need to diversify our semiconductor supply chains and have much more manufacturing in the United States, particularly leading-edge chips, which will be essential for AI,” she explained.
Raimondo is not the only one who thinks the government’s investment in the semiconductor industry is too low to ensure meaningful change. The cost of developing chip manufacturing facilities in the U.S. is far higher than in many other parts of the world, such as Taiwan. Meanwhile, TSMC, the world’s largest and most advanced semiconductor manufacturer, spends almost $40 on equipment and research and development every year to advance its capabilities. Therefore, for the U.S. to catch up with TSMC and other major producers, it will likely need to invest significantly more in the sector in the form of grants and financial incentives to encourage higher levels of private investment.
35 comments
Comments are closed.
> sourcing many of its chips from the Dutch giant Advanced Semiconductor Materials Lithography (ASML) and the Taiwan Semiconductor Manufacturing Company (TSMC)
ASML does not produce chips, they produce chipmaking machinery.
On a different important note, this article does not talk about the money in context, which is a bit of an oversight, to put it lightly. The Chinese are investing trillions. The $50B Chips Act, and some more billions here or there is peanuts compared to that. And that is even before we take into account that like any other US govt spending initiatives, the Chips Act money is mostly going to be a free cash grab for govt cronies, who will not bother to have anything to show for it.
“The Chinese are investing trillions”, do you have a source for that? The Chinese GDP is around 18 trillion USD, so trillions sound a bit much.
Taiwanese are more thrifty: “Meanwhile, TSMC, the world’s largest and most advanced semiconductor manufacturer, spends almost $40 on equipment and research and development every year to advance its capabilities. ” I spent more on my wind jacket… Perhaps we need to check about that $40.
I have no single source to refer to, and I have also lost my main source. But google a bit and you will see the numbers add up fast. Also do keep in mind that what they have going on is sustained investment over years, with proper industrial policy backing, as opposed to the single and now maybe second shot from the hip that the US has going for it. As to the relation with the size of the GDP, do also keep in mind that in general the Chinese have historically not been afraid to throw their weight behind something if it is needed, and in particular, after the US has designated them as the enemy no.1 in the recent years, with the associated attempts to cut them off of global semiconductor supply, China views achieving their own supply as existential.
In any case they started investing more seriously in 2014, to the tune of what seems to have started at around $50B a year. https://www.usitc.gov/publications/332/journals/chinese_semiconductor_industrial_policy_past_and_present_jice_july_2019.pdf
And are now up to at least $150B a year. https://www.digitimes.com/news/a20230627VL205/china-ic-manufacturing-semiconductor-chips+components.html . This will depend a lot on who is doing the counting though, and what is counted into it, and I am yet to find someone who cares to show they have kept track of all of the different funds that are going on. See also example this article which talks about a new smaller fund of $40B, but gives a great graph of global expenditure share by countries, and the growth of China in there https://www.fdiintelligence.com/content/data-trends/china-semiconductor-capex-adds-to-overcapacity-risks-83365 .
The most recent numbers seem to be more difficult to find, but what is known is that as the US has stepped up the trade war significantly in the last few years, so have the Chinese their self-sufficiency efforts.
https://futuresupplychains.org/chinas-semiconductor-supply-line/
What I would also be interested in is more info about the spending on local level. We in the West tend to disregard local governance as insignificant, yet China is extremely decentralised by our standards, with ~85% of the national budget spent on the local level. https://chinapower.csis.org/making-sense-of-chinas-government-budget/ There is some attempt here https://www.trendforce.com/news/2024/01/25/news-over-350-chinas-semiconductor-industry-projects-thriving-in-2023/ . This map definitely shows the industrial structure of the country, but I would suspect the money reflected there is not the whole picture.
As to what the money is buying them, https://www.asiafinancial.com/china-to-lead-2024-chip-expansion-with-18-new-fabs-semi . In the West a single new fab is a big deal. China builds 18 at the same time. The money China has to offer also goes a long way to poach talent, and institutional knowledge, from Taiwan https://www.aljazeera.com/economy/2022/5/4/taiwan-is-trying-to-thwart-chinas-efforts-to-poach-tech-talent , significantly lowering the barrier to entry. All of this combined is already bringing unexpectedly quick and advanced results https://arstechnica.com/information-technology/2023/11/how-huawei-made-a-cutting-edge-chip-in-china-and-surprised-the-us/ https://www.cnbc.com/2024/02/12/china-making-more-advanced-chips-but-beijing-still-faces-challenges.html https://www.ft.com/content/b5e0dba3-689f-4d0e-88f6-673ff4452977 . As if we here have needed someone to tell us that sanctions are counterproductive. I call in a few years the West will have to forget about banning exports to China to maintain a tech lead, and instead start banning imports from China to maintain a domestic tech manufacturing sector at all…
It’s all going to backfire on the West.
They thought that they were going to kneecap China. Instead all they’ve one is applied a sense of urgency to the Chinese and increased their semiconductor investments. In some ways, they have made the same mistake with the sanctions against Russia. They don’t learn.
As far as banning Chinese imports, at some point, we are going to find that Chinese imports are much cheaper, rendering the West less competitive economically, and eventually, China will end up surpassing the Western world in terms of technology too. When that happens, it will not be viable to ban imports without kneecapping the West itself.
The ruling class just doesn’t learn from their own failures. Their greed has destroyed the West.
True, they make Semi equipment that companies like Intel and NXP buy. I worked in ASML circa 2016-17. The Dutch incidentally are even worse employers than our dear own PMC. The work is technically challenging and uses a lot of concepts from Thin Film technology, Lasers, Optics and Signal Processing. ASML uses something like the 80-20 model where 80% of the Engineers are Contractors and 20% are permanent employees. They pay well per hour though. A ritual there was the early morning standup where everyone would be crowded into a room and a so called “Scrum Master” would stride in and start yelling at each person. He would attack and humiliate any Engineer who did not make progress on the previous day task. There was a sea of H1-B engineers as well. One question about the Chips act is how many jobs would flee to Bangalore in a few years time? Intel for example has a huge presence there. Thankfully a lack of water prevents Semicon companies from moving wholesale in India. I have read that these companies require a lot of water for the fabrication process. Recently there was a controversy in Indiana when the State was competing with Ohio to get Intel’s Semicon operation flowing from the Chips act and the lawmakers were falling over each other trying to divert all of Indiana’s available water to make Intel happy. Thankfully Intel went to Ohio.
Came here to say this. ASML is only company producing EUV photolithography equipment for advanced process nodes. (At least on MP scales)
At present ASML is the big player. One of the problems with their machines is that they have to periodically disassembled and cleaned because they get a layer of tin on the innards. (Details will be too prolix for this post.) But new technologies are coming.
See for example Nakamura et al. (2023), Jpn. J. Appl. Phys. 62, SG0809
https://doi.org/10.35848/1347-4065/acc18c
Cute paper about using free electron laser to generate tunable EUV and beyond. I’m sure the Chinese have read it, if i did.
sidd
We’ll know these CHIPS acts are working and accomplishing their goals, when stock buybacks among the susidized companies equal or exceed their CHIPS funding.
The US initiative doesn’t have a prayer. You can’t just build fabs, you need a manufacturing ecosystem that we no longer have. And you can’t recreate that ecosystem within our current economic environment, especially with business leaders that can’t conceive of building anything without guaranteed profit margins that require theft to achieve. The workers, components, resource inputs and all other factors of production are too expensive to compete. Subsidies can push past a 10% gap, or maybe 30%, but not the order of magnitude required. And even if you did produce these chips, either nobody would buy them because they are too pricey, or the industry would bleed to death on losses at every stage after the political will to squander subsidies runs out. Ain’t gonna happen. This is California High Speed Rail at the federal level. After $100B they’ll promise a 16-bit processor in 7 years. In the meantime grifters will grift.
Perhaps a few tweaks will improve the program. For axample:
Raimondo said…“To be clear, we can’t and do not want to make everything in America. We don’t want to make every chip in America. That isn’t a reasonable goal,”..
If she tweaks her view to…
“To be clear, we can’t and do not want to make ANYthing in America. We don’t want to make ANY chip in America. That isn’t a reasonable goal,”..
…then Washington will approve of her views and policy.
Lots and lots of complex global supply chains to ensure cheap labor and zero jobs in USA, lots and lots of big stock buybacks is the way to make it all work for Washington.
Everyone in Washington has known that for decades.
Thank you. Wasn’t it Wall Street that sent all that manufacturing overseas? Maybe they are the problem rather than China.
Maybe the ecosystem exists in India. Texas Instruments went to India in the 1980s long before Outsourcing became fashionable. I worked in TI circa 2004-08 and they had more or less taken over chip design from Dallas and Nice (France). One of the Senior designers even quit and started his own company which then became very successful (IITIAM – I think therefore I am).
At least government policy has shifted from this earlier view of technology:
“As Bush I [chair of the White House Council of Economic Advisors] Michael Boskin memorably quipped…, ‘Potato chips, computer chips, what’s the difference? A hundred dollars of one or a hundred dollars of the other is still a hundred dollars.'”
I’ve lived in China and watched it develop for over two decades and they do not do thing’s half ‘buttockeds’ to be polite. The US does about the same at 25% or less. Guess who wins? And well they should. What astounds me is that where the US could actually make some big money from China, they just ‘urinated’ that chance away. I guess that’s just a great example of how even a nation can grow senile with age, Mike Liston
They can be lazy but the point is that stuff like EUV is something one can throw brainpower at. It’s light, physics, nothing that needs to be interpreted right or wrong. Work it out.
The tech export ban sets them back but also makes them more inclined to develop homegrown faster. One of the recent chips in the new Huawei phones is domestically manufactured 7nm. Shocked the West thinking these communists are good for nothing and cannot think for themselves to make anything.
https://www.techinsights.com/blog/techinsights-finds-smic-7nm-n2-huawei-mate-60-pro
There’s also homegrown processors and other things already in second or third iterations. Does China need ASML in the future? Do they necessarily need smaller? If everyhing right now was on 7nm, that’d be good enough to work with.
In the Syracuse area Micron is supposed to build a dram fab facility on 1000+ acres county owned land (filling in 250 acres of wetlands). The state will provide $5.5B with the feds kicking in a similar amount if the planned 4 facilities are built over a 20 year period. Supposedly 9000 jobs will be created. The initial grants have not been released. The site will consume 5% of NY state’s electricity and will surely affect regional rates. There are 3 aging nuclear plants and multiple CCGT gas fired plants in the region. Massive amounts of water will come from the county system, drawn from Lake Ontario. The county IDA and political hacks are in Taiwan now drumming up the area for suppliers to locate here
Needless to say, this area is seriously deindustrialized rust belt. There are a lot of universities in the region that eventually could supply professional workers, but skilled manufacturing workers and trades are generally few and not many pipelines for that. The highway infrastructure on that end of town is poor and housing already a problem (but cronies are already getting sweetheart tax deals).
We’ll see if this actually happens; I’ll believe it when I see it. Given the issues with other such mega projects (eg, TSCM Phoenix), call me skeptical.
I think there are bigger issues. American corporations tend to be very short term oriented when it comes to their profit margins. Intel once had the best fabs in the industry and fell behind, in no small part due to doing things like shareholder buybacks. That played a major part of why TSMC overtook Intel. Subsidies won’t fix what is a moral failing among the elite. They are too “short term greedy” to make this work out.
Even with enormous public subsidies (which are essentially giveaways, especially considering there won’t be demands for concessions like equity in exchange for cash), it won’t change the fundamental nature of American capitalism. Normally, the ruling class claims to hate doing this, but the problem is that they desperately want to hold onto hegemony and they see China rapidly rising.
The whole ecosystem would have to be expanded. There are 3 basic categories:
1. Fabs like TSMC (Intel also owns its own fab and has been trying to win the business of other companies)
2. Fabless companies like Nvidia, AMD, which pays category 1 to make their chips (Intel has its own in house design team)
3. Suppliers like ASML, which supply equipment to 1 and have hundreds of suppliers as well
This is a very simplified view of the whole situation. For the US, it would mean expanding everything. Semiconductor engineering would have to be a more remunerative profession compared to careers like investment banking, management consulting, and software engineering. Then there’s the matter that the US economy is so financialized that living costs tend to be higher than the US, making the US less competitive.
One major barrier for China is ASML’s EUV, which has been barred due to US export restrictions, something China is trying to substitute through 193 nm lithography with multiple patterning (and it appears that they have succeeded in 5nm lithography). The Chinese are spending a lot of time to build their own EUV domestically. Certainly the decision to poach foreign talent by offering very attractive salaries has accelerated the pace of Chinese innovation, but we are also seeing the domestic talent being built up internally in China as well. Universities throughout China are increasing their semiconductor engineering majors.
We are already seeing some of these trends. In category 2, Nvidia has indicated that Huawei (notably not Intel or AMD) is the biggest competitor in AI technology, for example. Nvidia and ASML have both lobbied against these export bans, noting that they are self-defeating. I don’t usually agree with business lobbies, but in this case, they have a point.
China, as one of the other commentators indicated doesn’t do things in half measures. The amounts invested are staggering and if the other commentator is right, then the US is being out-invested by over 1 order of magnitude in investment in raw dollar figures. That doesn’t even factor in that China is likely to be far more efficient to be in their investments and they tend to take the long-term view (ex: they won’t engage in the kind of stock buybacks or that sort of rot that affects the US). I fully expect that the Chinese will build up a massive semiconductor manufacturing base,
At some point, I full expect that the Chinese will surpass the Western world and then even a ban would be counterproductive because by then, China would have the most advanced nodes. The West would be depriving itself of the best technology, rendering it even less competitive. It will also be made at a lower cost than the comparable Western chips. It’s going to end up like how high speed rail has ended up.
I recommend reading the article in “American Affairs Journal”, which thoroughly describes the various sectors involved in producing electronics, manufacturing tools, specialized materials, and software design tools.
It gives a very good survey of the players in China, how they compare to foreign players, how far they have already come, their inter-dependencies, and the challenges they face when setting up a completely domestic industrial sector free from interference from foreign powers.
One aspect is striking: the Chinese approach is one of husbanding a myriad of players, organizing and generously funding short-, medium-, and long-term projects, intervening in relationships between firms (as an example, the government made sure that Huawei would not poach engineers from another firm and collaborate with it instead), firms and academia (e.g. making sure the industry can access research results and IPR), setting up industry champions that will drive development, and supervising the whole effort at the very top, i.e. it is truly a matter for the boss (i.e. Xi Jinping).
In the USA, it looks as if the CHIPS act is essentially about throwing money to a variety of players, probably in the hope that they know what to do and thus basically need to be supported economically in their efforts.
China faces incredible challenges, but from what I have read, I rather think they will succeed: not only are they persistent and competent, but they will probably benefit from the deliquescence of Western industrial skills (firms focused on short-term advantages and stock buy-backs), the dislocation of academia (living off short-term contracts and low-paid researchers), and the loss of competence in government (no longer able to truly organize and coordinate, and ideologically reluctant to set industrial policies).
China having a lot of local players is nothing new. They have done this with EVs as well. Now the West is freaking out because they can’t compete with China on price.
https://fortune.com/2024/01/27/evs-byd-china-tesla-nio-xpeng-dan-ives/
That’s why there are trade wars going on.
The issue is about to repeat on semiconductors.
Already the West is worried about the older nodes.
https://financialpost.com/pmn/business-pmn/us-europe-are-growing-alarmed-by-chinas-rush-into-legacy-chips
The margins on these older and fully depreciated nodes is used to fund the capital and R&D costs for the latest nodes.
So we are in a situation where China is rising and cannibalizing sales already. It’s going to end with China surpassing the West.
They could be right about it being a national security issue for the US and wrong about the prescription (export controls on China). Turning it into a zero sum competitive game feels like a mistake when you’re starting this far behind, and when your system is financialized and likely to siphon off any investment for unproductive uses, as you stated.
I can’t think of an example of when one of these attempts to secure or preserve US technology supremacy through export controls actually succeeded. The obvious comparison is the Cold War era controls on strong encryption exports. Eventually it became so fundamental and pervasive that the export controls were doing nothing but hurt US exporters for no gain, and were dropped. I would expect something similar to happen eventually here.
I think you’ve hit on the key point here – there are bigger issues. America did not fall behind in chip fab or many other manufacturing sectors because of a shortage of money, and giving the exact same people that ran the American chip fab industry into second, third or forth place won’t fix the problem. It’s funny how Obama used to repeat that the “Republicans” had run the car into the ditch and he was going to take away their keys while watching him during the GFC bail out everybody, and nobody went to jail. A more accurate analogy for what Obama (and indeed almost all neoliberalcon cures) did is to go up to a meth addict and tell him that you’re going to give them $10M to stop being a meth addict. Now one in a hundred times that MIGHT work, but generally you’ll end up with a dead meth addict and local drug dealers $10M richer.
A better, more permanent fix is to have the SEC immediately stop share buy backs, raise tax rates on corporations (forcing corporations to pay taxes or invest in the corporation with capital expenditures and R&D). Bonus points for dragging all the industry CEOs to a Senate hearing and calling them unpatriotic SOBs that have wrecked their country. Along with all this, you have to fix higher education by turning universities back into very low cost for students, get rid of the PMC administrators, and ramp up R&D.
But I don’t expect any of that to be done so this whole CHIPS thingy is just one more giant government hand out to the exact same people that caused this problem.
Wonderful comments. Educational or refresher/updates for all. I would suggest NC readers imbibe these, then still read the American Affairs piece, especially for Chinese policy development past and present, and the distortions of the U.S.’s current/recent moves. Also good on scale, for those of us nonspecialists.
The newest I-phone, I-phone 15 Pro, has 20 billion transistors in its main chip. Moore’s Law has held up now for almost 50 years.
EUV chips, under 5nm, will be hard for China to duplicate in the next few years. But I understand that China may soon dominate the 7nm to 28nm market, which provides the bulk of chips in products like autos, household appliances, etc.
For a good history on this topic I recommend the book “Chip War” by Chris Miller:
https://en.wikipedia.org/wiki/Chip_War:_The_Fight_for_the_World%27s_Most_Critical_Technology
dday,
Do you really need a phone with a CPU that has 16 cores, or will 8 cores suffice? The fact is most of those cores are turned off all the time because they drain the battery – apple has put more cores on the die to convince people to upgrade to the new iphone but they are, by and large, not even being used.
7nm dies will more than suffice for the vast majority of applications. People who think that the US will “stay ahead” by having more transistors on their dies are people who don’t have engineering degrees.
“Do you really need a phone with a CPU that has 16 cores, or will 8 cores suffice?”
[ Really important question, the answer to which I think is now “no” and will be generally “no” for quite a while because of the battery drain problem. What then of particular problems to be solved? ]
“Do you really need a phone with a CPU that has 16 cores, or will 8 cores suffice?”
An answer to be question should affirm what questions will not be readily answered with 8 cores. I do not know the answer and need to ask about this. Ideas will be welcome for me to focus on.
Remember the old U.S.-Japan semiconductor dispute: In the 1980s, Japanese companies sold semiconductors more cheaply than U.S. companies and overtook the U.S. for the largest share of the world market. What did U.S. semiconductor companies do to counter this? Make their semiconductors cheaper than the Japanese? No, they cried to the government and begged it to declare Japan a dumping country and put pressure on it. And Tokyo bowed to Washington’s pressure. Japan agreed to let the U.S. set the selling price of its semiconductors and accepted restrictions on exports of semiconductors to the U.S. and on the purchase of a certain amount of U.S.-made semiconductors. This measure led to the collapse of the Japanese semiconductor industry. It is important to note, however, that the U.S. capitulated to Japan in the semiconductor market not because of disloyalty on the part of Japanese firms, but because of improper investment practices and lack of competitiveness on the part of U.S. firms. Will the U.S. make the same mistake again?
ciroc
The US did not “capitulate” to the Japanese – they thought that design was the high margin business and fabing was grunt work best outsourced to Asian vassals. This strategy worked very well for the US for many years – “fabless chip companies” like NVIDIA were made possible by this strategy and would not have happened otherwise.
But moore’s law eventually increased the barriers to entry in fabing and ended up making TSMC their like more powerful than any vassals should be allowed to become. Hindsight is 20/20.
Terrific exchange. I think I am close to understanding now.
I’m curious about the water issue – what happens to the water after it is used in the fabrication process? Is it polluted, can it be cleaned etc?
I’m curious about the water issue – what happens to the water after…?
[ Really important question. ]
The water, to not go into the detail too much, is basically used to wash out all of the weird stuff of the different manufacturing stages, and also just to keep even the smallest of dust and such away. When you are manufacturing stuff on the level of adding single layers of atoms of material, everything has to be extremely clean.
The weird stuff they use consists of basically everything you don’t want in your drinking water, from heavy metals and all sorts of toxic stuff, to good old halogens. High enough concentration to not send it to the water supply, probably low enough concentrations to just send it downstream with the approval of a friendly governing authority, and small enough particle sizes to make cleaning it as difficult as it can be.
What is actually been been done to the water so far seems not to be discussed publicly, but since the need for recycling that water has recently started to be discussed, I guess it is safe to say that so far, it’s indeed just been sent downstream. Maybe someone with industry background can chime in further.
Industry thinks water usage is a manageable problem https://spectrum.ieee.org/fabs-cut-back-water-use
Intel and TSMC are expanding chip manufacturing in Arizona. Not a place known for abundant water. The commenters on that Arizona article have more details on how water is used.
Both China and Russia have been under the mistaken idea that the US and the EU wanted to do business.
The US and its vassal , the EU are only in the business of Hegemony and the violence required to maintain it. China and Russia are moving on and its too late for the collective West to redeem itself.