US Government's new export controls are wreaking havoc on China's chip industry
twitter.com
twitter.com
So i think this may be a mistake in the very long run - the profits are denied to the western chip makers, and such profits are necessary for continued investment and innovation; and secondarily, sanctions like this will give a massive incentive for indigenous industry development. If successful, their competitiveness is going to be higher than that of the US & west's.
Sure some countries could leverage the recent flushout of corruption and hundreds of thousands of employees of the microchip industry looking for work. But I doubt China will be able to.
Not really, it's expected quirks of ambitious PRC industrial policy which have historically been successful. Spam it until you make it. Inefficient on paper, but ultimately extremely efficient because the alternative is to have no indigenous options. Losing few billions out of ~100B worth of big funds over ~10 years is just cost of business (fast, cheap, good etc). It's about as inefficient as Intel insisting on divident payments + layoffs even with CHIPs. But in PRC, folks actually get punished / incentives aligned via anti-corruption / CCDI. Meanwhile, actual PRC semi capacity has increased dramatically in last few years, after leadership decided to stop halfassing post Huawei sanctions.
> some countries
Thousands semi shops closed down because that's what it takes to speedrun the ENTIRE semi supply chain, an effort no other country can even attempt.
> looking for work
US export controls just destroyed 1/3 of western semi revenues, there will likely be mass layoffs in west coming. Meanwhile PRC domestic semi industry just became a gold rush for non-American companies / talent and the incentive to de-Americanize US origin tech to create parallel semi supply chain to service PRC's 400B semi industry. There's a reason US Gov had to unilaterally announce these new measures on the steps of CHIP4 when other countries said they're not interested in proposed level export controls.
I've read about this, and chip manufacturing is not like other manufacturing. It's like when other places try to duplicate Silicon Valley, or Hollywood. I've read about this for 40 years, yet no one has made any significant inroads.
Why? The infrastructure. In Hollywood, for example, it isn't just about the physical act of making the movie - get a camera, some lights, bank you are done. There's a whole infrastructure - accountants that only work in the film industry, caterers that specialize in it, mobile dressing room companies, prop companies, movie animal companies, lawyers specializing only in movies. Advertisors, product placement specialists, casting companies, photographers for headshots, agents - the list is truly endless but it all works together in a symphony because everyone knows everyone and how the machine works.
I've lived in Silicon Valley for a long time, and it is out of the world how the entire SF Bay area - and especially San Jose to Palo Alto is geared to tech. Two world-class universities - Stanford and UC Berkeley, and other schools - San Jose State, SF State, East Bay, University of Santa Clara, University of San Francisco - and more - all contribute talent to Silicon Valley. It's literally everywhere, it is almost inescapable.
Hollywood's been there for more than 100 years, Silicon Valley for a long time.
Or like Delaware and corporations - I've read in depth about them and it's the same - their infrastructure and ecosystem is the geared towards corporations. Too long to go into. .
The same is true for Taiwan's chip industry. It is impossible to duplicate, even in the USA. They have the same infrastructure that I've read about. Anyone can go down to xyz nanometers, but go farther than that, and no.
For example, the Dutch company ASML, a $350 billion 37-year-old company with 31,000 highly specialized staff, is the ONLY firm in the world that can make the most advanced chips and Taiwan semiconductor companies buy their machinery to make chips. USA could put incredible pressure on that company not to sell to China, and that would be yet another completely different technology that China would have to develop, but by then, if they could even do that, technology would have improved and China would always be behind the curve. Here's the thing - they make $350 billion by selling to VERY fewe clients - TSMC, Samsung, Intel. Each machine has over 100,000 specialized components. It takes 40 fright containers or four jumbo jets to ship one. In 2020, it sold 31 of them. Trump did stop sales of them to China, and Biden is continuing this policy. ASML has 4,000 suppliers, and those suppliers have suppliers.
Their mirrors, for example, are made by German firm Zeiss in partnership with ASML and they’re the flattest structures humans have ever made.
In the 1960s and 1970s, they used a regular lightbulb to create wafers. Now they have carbon dioxide lasaers that is 400,000 degrees F in temperature. The process to make it took 30 years.
It’s an insane process to ship them and there’s a big learning process to get them up and running because the machines are so complicated. They’re not like an off the shelf, plug it in, turn it on and go. You need to be able to train the staff that are operating them. ASML staff is at every site that there is a machine. ASML is planning to release a next generation machine called High-NA, which stands for high numerical aperture, around 2025.
ASML’s headquarters is in Veldhoven, Netherlands - home to around 45,000 people and thousands of ASML employees, for full assembly. And all those people there and in the surrounding area, who don't work directly at ASML, work for companies that work for ASML, and all part of the ecosystem that supports ASML.
And PRC is not like any other country.
At the end of the day semi is just a niche industry iterated by combined talent from US/TW/SKR/JP/NE (maybe DE). PRC now produce more STEM talent than all these countries combined. I think too many people drink the semi conductor is thinking sand indistinguishable from magic juice. Same was said about aerospace and other precision manufacturing sectors that many predicted PRC wouldn't surmount but now have indigenous solutions and rapidly closing high-end/value gap on. Yes, semi has very high walls / barriers, but PRC's advantage is being able pile enough bodies against a high wall until they climb over, freqently quicker than anticipated.
What group of countries need to collaborate on, PRC has talent and resources to do indigenously. After elevating semi to first-level dicipline in 2018, PRC is spitting out 30k IC graduates per year. They're still about 200k short, ~520k/720k out of what IC talent white paper estimates PRC needed for competitive semi industry. I wadger that's comparable to total _direct_ semi industry talent globally. Consider PRC is graduating 11m STEM talent a year, and the amount of resources they can spam into strategic industries is enormous. The reality is, PRC can duplicate what others do, and do so at larger scale. Or that industrially, PRC is TW on steroids, nearly every sector TW has outsource onto mainland, PRC has been able to replicate, scale and then inject extra does of performance, unlike TW semi expansion in west running into culture (read: work ethic) issues.
Ultimately it's about talent and time and these bans are about denying talent and buying (US) time. Consider ASML CEO last year thought PRC was 15 years from tech soverignty. IMO PRC could likely do it sooner, otherwise US would not have unilaterally place these export restrictions when last year US strategist thought it was merely enough for US to "sprint" and stay a few generations ahead. Now it's as big of a lead as possible, which either insinuates greater lead than before or more likely, the race is closer than we thought.
while i generally do agree with the premise of your argument, the large amount of STEM graduates is actually irrelevant or at best tangential. China has resources to spend - they can do so at the cost of profits (or people). it doesn't take 11 million people to run a semi conductor industry - it takes 10 thousand, but those 10 thousand must be nurtured and be well trained, and have personal profit motive and incentive.
I'm sure the chinese gov't can make this incentive - but whether their environment makes for innovation at an individual level, remains to be seen. After all, there's a huge amount of graduates who see graduation as a form of ritual to obtain a stable, easy and well paying job, rather than as a form of education for which you have to continue being enterprising.
I just think the US's strategic export controls won't end up as effective as they imagine it would be, and in fact, might be counter productive.
But yeah, PRC higher ed does have reputation for making coasters who thinks life is easy after getting into tier1 school after rigourous GaoKao prep. Which is why I think worth keeping eye on innovation, S&T indexes to see how PRC is doing in absolute terms. But ultimately, spamming chabuduo tier efforts that occasionally generates something high quality, but do so at PRC scale so as to generate enough incidental quality to be globally competitive is a numbers game PRC often plays to her favour.
This is a lot more info, not directly related to the conversation, but directly affects what is happening in tech and the rest of China's economy, and why their tech may still fail.
One thing I'm reading about, which is related, is that China's population is about to crash, and crash big time. It is all over the world, but particularly going to hit China hard because of the one child policy and they are still not a first world country.
In 2018, birth rates dropped 12% to 15.2 million, with some provinces droping by 35%. Yi Fuxian, a professor at the University of Wisconsin-Madison, has written that China’s government has obscured the actual fertility rate to disguise the disastrous ramifications of the “one child” policy. According to his calculations, the fertility rate averaged 1.18 between 2010 and 2018. Replacement is 2.1 child per couple.
Fewer children were born, and because of cultural preferences for male offspring, fewer of them were girls.
There are 122 boys for every 100 girls born, so this is an issue - not all men will find a mate and competition for women will be intense. 1 in 5 men won't find a mate, at all. 34 million frustrated men. I've read before that this is one of the main contributors that nations go to war - for single men to get women, if there is a great inbalance. It would certainly make the wifeless men very angry and ready for war, and motivated. The male/female birth imbalance has already fueled human trafficing and child kidnapping.
Chinese women born during the years following the “one child” policy are now reaching or have already passed their peak fertility age. There are simply not enough of them to sustain the country’s population level. The declining population will create an even greater burden on China’s economy and its labor force. With fewer workers in the future, the government could struggle to pay for a population that is growing older and living longer.
Additionally, much of China's population have put their wealth into property based on the greater fool principle. This has resulted in entire ghost cities being built, with nobody living in them, and the apartment buildings are half-done (no one can live there even if they wanted to), and with the population collapsing, property value will precipitously drop, and soon. The value of chinese real estate is $52 trillion, twice the USA's, and greater than the USA's entire bond market. This reminds me when an acre in Tokyo in 1980 was worth more than the entire state of California's real estate - it's just crazy on principle. In China, 78% of wealth is in property, compared to 35% in the USA. 44% of homes are 2nd homes, 25 are 3rd homes. 30% of home sales are for people to actually live in. China's upcoming crash will make USA's 2008 crash look like a fender bender.
A study published in The Lancet forecasts after hitting a maximum population of 1.4 billion, China will lose nearly half its population by the end of the century, no matter what they do. It's too late.
The People’s Bank of China warned China had only about a decade left to enjoy the benefits of its large working-age population, which has helped propel growth over the past four decades,”
All retirement in China is unfunded. The children might support 2 sets of parents - his and hers, and 8 grandparents in their home, and can't split up between kids - 1 child policy. The working population is getting very resentful. The goverment wants this generation to work the 996 schedule [9 a.m. to 9 p.m., six days a week]. This is why the "Lie flat" movement is spreading in China, and leadership is upset by it.
Adding to this, desertion (not desertification) of rural areas are so severe that agriculture is not happening and Xi Jinping as admitted that China can't feed itself anymore. Scarier: the full effects of climate change are yet to be felt in full force.
With the war talk, and everything else, companies are pouring out of China, so China's income is going to take a hit. And rising employee wages means that China is pricing itself out of the market against new low cost suppliers.
While you say that they can do it inhouse, they still need to have buyers. And the buyers are the first world, who has all the money. With China being belligerent, is the West going to purchase from them, or will we comletely boycot them to starve their incipient industry? Or will they be good only with internal customers? I don't know.
Will China be able to create an advanced semiconductor industry? Who knows. Only the future will eventually tell the tale.
First some PRC numbers to contexualize actual cost of current PRC demographics, 600M undereducated stuck in informal economy, ~20% (200M+) farmers most of whom who could be replaced by machines, trillions of SEO inefficient waste poured into essentially make-work jobs. Height of PRC manufacturing employed like ~300M. All of which is to say PRC HAS too many people, only fraction of demographic divident was exploited to essentially address global demand, while huge % have always been demographic drains, and the sooner they clear off the demographic balance sheet, the "stronger" PRC's actual geostrategic/geoeconomic posture will become. PRC excessive population has (IMO) on balance been more demographic curse than divident.
There are two demographic shifts happening in PRC.
1. Net demographic decline which will only reduce PRC import dependancy and increase strategic flexibility (less mouths = more energy/food security).
2. Massive demographic boom in high skill workforce - PRC generating as much as all OECD countries combined in skilled talent. That's largest high skill demographic divident in human history, not the poverty farm to factory transition, but profitable factory to laboratory one, with decades left to milk.
What does this mean? Medium/long term PRC is shifting towards more strategic independance and advanced economic development. Yes, there will be some heavy social pressures weathering the transition, but PRC is culturally and structurally well calibrated to weather demographic decline. Unliked developed welfare states, PRC is a modestly developed society with high home ownership and high savings rate because, bluntly there is no expectation of comprehensive safety net. Look at civil protests in the past few years, it's mostly developed societies protesting losing perks, not developing countries protesting to get perks they are not fathoming in the first place. Don't get me wrong, PRC society will be way more stressful than developed economies, but it will also be able to maintain political serenity easier because governance in developed countries with huge welfare obligations has more brittle modes of failure than PRC's alternative of having their citizens largely own their own homes and retire on their own savings. There's also additional dynamic of wealth transfers that will stimulate currently stubborn consumption because all this money is locked away as nest egg. Or that the poor in PRC are so poor (huge income disparity) that it doesn't take huge wealth transfer to keeps "dregs" from uprising, not that they could because they're old, and whatever welfare they get from the state will likely be better than what they grew up with. PRC decline will be one of "stagnation", western decline will be one of "loss" especially drama with increased immigration, it's easier to govern/manage the former than latter.
> directly affects
IMO less than one thinks.
I'm bundling your (familiar) points under the customary PRC demographic collapsist narrative that tries to insinuate social pressures that will lead to inevitable stalled ambitions/developement or decline. But a country can decline in terms of QoL but still improve in PRC's nomenclature - "comprehensive national power". Consider all the Asian Tigers with shit tier TFR that move up value chain (especially semi) simply because more % of population became high skill over time. They've all experienced fair share of economic and social drama in past decades, and relative to boomers, current QoL are in continuous decline. Yet these mid to large sized (20-120m) societies managed to specialize in some strategic high value industries to become very competitive economies.
And this is again where PRC is not like any other country.
1.4B pop with even bad TFR by virtue of base effect is enough to generate surplus talent to necessarily develop every sector simply because there's too many excess bodies not to. While current 20% youth unemployment is nothing to brag about, nor is lying flat not upsetting, but these are indicators of excess talent and extremely competitive labour enviroments. Sure, not good for society, but great for cut throat development. It's high skilled hunger games. Big reason TSMC thinks US fabs might fail is "cultural" issues (read: American's don't slave hard enough).
Consider USA at peak of hegemony had ~250M pop to maintain technologic autarky in the 90s; IMO where PRC is trending towards is highly developed and competitive tier1/T2 coastal region with multiple times the skilled talent of US (maybe on par with OECDs), that will close parity and then lead in SciTech in the coming decades. Combine with cheaper labour from T3/T4+ regions (ability to arbituage regional labour costs great for PPP/maintaining all that infra) and PRC acquiring more automation / industrial robots than next 14 countries combined, and we have recipe for a PRC were social fabric will be increasingly miserable, but with almost foregone potential to leapfrog in comprehensive national power, with capacity to lead in many sectors while maintaining much larger modern military due to talent and PPP advantage. Like this is not how one would structure a society to maximize human flourishing or per capital wellness but it is well calibrated to maximize comprehensive national power for great powers competition.
>buyers
Covid export bump aside, PRC or US are not particularly export dependant countries (14% vs 20%)... with exports to west accounting for ~10% of PRC GDP. Ergo US/PRC dragging on trade war, because ultimately both countries run on dominant internal markets. And IMO, long term west will keep walls up anyway to potential competitive PRC products that encroach on their high value niche (look at what just happened to YMTC). PRC's long term game is to chip away at emerging market share and deny domestic market access. The new export controls could cost western semi 1/3 of revenue (PRC huge semi market) and have to scale back R&D / competitiveness. When PRC catches up, they'll start hammering industrial policy to make western products increasingly uncompetitive. It will not be about selling to west, but denying west sales, taking their sales from other markets, and reducing competitiveness of western industries to the point where they become loss leading strategic sectors sustained by billions in subsidies / industrial policy. My long term trend guess is convergence where PRC will grow slowly, but every unit of PRC growth will soon come at cost of more units of western growth.
It is too early to tell how it will play out.
Here is another perspective:
https://asiatimes.com/2022/10/china-chip-ban-a-us-exercise-i...
China chip ban a US exercise in extreme self-harm
CapEx and R&D implode in hard-pressed Western semicon industry while China pours massive funds into chip independence
Also it is not correct that the value the Chinese semiconductor industry has been "reduced to zero overnight". If anything, judging by the stock market price of the China's largest contract fab SMIC relative to TSMC, the Chinese have been hurt less.
All this does is push Russia and China into each others arms – not a strategy designed to maintain a globally dominant position.
Also note that Asia Times is out of Hong Kong and we know that press there has to tow the CCP party line or face ramifications if they don't.
China's morals hasn't changed since the west outsourced it's production to it.
What has changed is that China has gone from being a contractor to a competitor.
The Biden administration has simply acted on the reality that the PRC to all extent and purpose is an enemy of the USA as it strive to undermine our political system and influence around the world, and is preparing for more aggressive action to support that agenda.
Clear as day, expect more action on that front.
US numerical impact is not limited to their own population count; if they get Europe to do their bidding then add them to the mix as well as other NATO aligned countries.