The US is trying to hold its high ground of dominance of the semiconductor industry via export restrictions and subsidies to increased domestic manufacturing, notably via the Chips Act. Yet experts are quietly warning that this plan to decouple from China may backfire, particularly if pursued too aggressively.
Semiconductors are fundamental to the operation of commerce and consumer communications, so the US believes it has found a key choke point by which it can impede China’s further rise as an economic superpower. But the wee problem with that view is that the US view thought it had an even more powerful choke point with Russia via its supposed dependence on dollar payment systems. We know how that movie is working out.
Admittedly, the US actions against China’s chips industry are not of the “kill the economy” ambitions of its sanctions against Russia. But there’s a weird myopia in not understanding that China has plenty of ways of retaliating if thing were to get ugly, given US dependence on China for many imports, starting with pharmaceutical ingredients and seemingly humble chemicals like ascorbic acid. And as we’ll address soon, a broad analysis of technology leadership by an Australian think tank shows the China to be number 1 in 37 of 44 categories.
We’ll provide a high level treatment today and plan to go deeper in future posts. Let’s return to the various semiconductor protection moves. As far as I can tell, the justifications were to prevent Chinese spying on Americans and impede China from using advanced chips in military applications (although truth be told, armed forces don’t make much use of the super-small chips that are the focus of the curbs). But even experts who are sympathetic with the idea that the US should do more to protect its interests in its dealings with China think that even with obviously over-broad Trump era measures having been rolled back, the new restrictions aren’t well targeted. From Jon Bateman, a senior fellow at the Carnegie Endowment for International Peace, in Politico in January:
America has embarked on one of its most difficult and dangerous international challenges since the Cold War. The task: reversing decades of economic and technological integration with its chief rival, China.
This technological decoupling, if done selectively, will help to preserve America’s military edge, protect key U.S. industries from unfair competition, and push back on Beijing’s human rights abuses. But if decoupling goes too far, it will drag down the U.S. economy, drive away allies, stymie efforts to address global crises like climate change, and increase the odds of a catastrophic war….
Restrictions on Chinese technology make sense when they match the scale of specific threats and buy time for America to bolster its own tech base. But Washington seems intent on a grander crusade — to hobble China at a fundamental level — with little regard for the risks to global stability, the U.S. economy and American alliances.
The Biden Administration first focused on increasing investment in the tech industry, but then started deploying restrictions. Again from Bateman:
Even so, there were growing hints of a more aggressive agenda. First came reports in May that America’s most severe sanctions list — populated with terrorists, drug lords and war criminals — might for the first time target a major Chinese tech firm. Then came the bombshell announcement in October of new export controls on semiconductors and chip-making equipment…
The new U.S. export controls block China from importing high-end foreign semiconductors it needs to train artificial intelligence algorithms. At the same time, Washington sought to stop China from making homegrown versions of such chips, or even the mid-range chips that power the Internet of Things and other lesser devices. It therefore barred Chinese chip-makers from importing advanced manufacturing equipment and from working with U.S. personnel…
Officials cited the fact that advanced processors can help Beijing model nuclear explosions and missile aerodynamics. But these military applications comprise a tiny fraction of the countless important uses for powerful semiconductors and AI. The vast majority are benign: business process automation, e-commerce, cybersecurity, disease diagnosis and much more. Some uses, like climate change research, would actually benefit the United States and the world….
Alan Estevez, a senior official who oversees export controls, captured the gung-ho mood in late October: “I meet with my staff once a week and say, ‘Okay, what’s next? What are we going to do next? Who’s being bad? Where is the technology area that we need to address?” He said that future controls on biotech, quantum technology, and AI software and algorithms are likely.
That triumphalism seems awfully familiar. Foreign Policy sets forth that view in more detail:
To retain its role as the world’s sole superpower, Washington believes that it has to stop Beijing in its tracks…
In this economic war, the United States is unsurprisingly keen to put all forms of economic coercion to good use. The Trump administration imposed tariffs on $360 billion of U.S. imports from China; President Joe Biden has made it clear he is not lifting these…In the financial sphere, U.S. lawmakers are pondering whether to delist more than $1 trillion worth of shares of Chinese companies on U.S. stock exchanges. Congress is also considering barring the Thrift Savings Plan, which manages the pensions of millions of federal government employees, from investing in Chinese companies.
The Chinese economy, however, has grown far too big for Washington to sanction Beijing with its usual toolkit….
Semiconductors are the Achilles’ heel of the Chinese economy. Beijing buys more than $300 billion of foreign-made semiconductors every year, making computer chips China’s largest import, far above oil. This reflects the fact that Chinese factories import 85 percent of the microchips they need to build electronic goods. Most of these semiconductors are manufactured using U.S. technology. For Washington, this makes export controls a seemingly ideal tool to deprive Beijing of U.S. innovation and know-how. Such restrictions function in a similar fashion to financial sanctions: They seek to curb adversaries’ access to U.S.-made staples—the greenback for financial sanctions or computer chip technology for export controls—that have become so crucial that few countries can do without them.
Washington knows that it has a massive trump card to play in the semiconductor sector: Virtually every microchip around the world has some link to the United States, be it because it was designed with U.S.-made software, produced using U.S.-made equipment, or inspected with U.S.-made tools…
U.S. firms manufacture only around 10 percent of the computer chips sold across the world. The world’s leading microchip foundries (as semiconductor assembly lines are called) are located in Asia, mainly in Taiwan and South Korea. However, a handful of U.S. companies control all of the higher, upstream echelons of the supply chain. Given the United States’ dominance over the microchip sector, Washington knows that measures curbing China’s access to U.S. semiconductor technology have every chance to deal a blow to Beijing’s technological ambitions….
In October, the Biden administration dealt an even more severe blow to China’s technological sector: Instead of targeting only high-profile Chinese firms, Washington clamped down on all exports of advanced microchips and semiconductor-making tools to China. U.S. citizens were also warned that without explicit (and unlikely) U.S. government approval, they are breaking U.S. law if they choose to work for Chinese technology firms.
Other experts are warning that the loss of the Chinese market will hurt the profits and even more so the R&D spending of key players. From Anjani Trivedi at Bloomberg:
China accounts for over a quarter of sales for chip equipment manufacturer Tokyo Electron Ltd., where they’ve been growing sharply over the past five years. For Nikon Corp., a maker of lithography machines, it’s around 20%, while Advantest Corp., which produces testing machines, depends on China’s evolving computing market for its customers, too. The country accounted for over a quarter of global billings — a gauge for demand — at the end of last year. Along with Taiwan and South Korea, China has been the top destination for capital spending for the past two years for the largest semiconductor equipment companies….
Here’s the rub: These firms don’t just invest in China, they sell equipment across the world, including to the US and Europe. That keeps the virtuous cycle of technology transfer and development humming along. If they’re hamstrung because major sources of revenue get cut out, then ultimately industrial innovation will struggle. Even if the US manages to stay ahead in terms of technological advances in lab projects and patents, it won’t be able to scale them.
Scaling is a key point. Some commentators, including NC readers, have opined that a major focus of the China-hawkish measures is to reduce the dependence of US chip designers on Taiwanese fabs and build up capacity in the US. The problem is the big-sounding numbers on that front don’t go all that far. From Yu Zhou in Issues.org:
The CHIPS and Science Act authorized $52 billion for domestic semiconductor chip manufacturers with the aim of enhancing the global competitiveness of the US chip industry, improving the security of the supply chain, and countering China’s ambitions in the sector.
While increasing investment in semiconductor research and development is welcome, whether it can improve US global competitiveness and prevent the rise of China is uncertain. In 1990, US companies manufactured 37% of semiconductors produced globally, but by 2020 that share had shrunk to 12%…
In this notoriously capital-intensive industry, the CHIPS Act’s $52 billion investment is relatively small. For example, in 2022, just one company, TSMC, announced new capital investments of over $40 billion, building on $30 billion invested last year. Samsung plans to invest $355 billion in its semiconductor and biopharmaceutical technologies over the next five years. Since the semiconductor industry is the single most important global niche held by South Korea and Taiwan, government and commercial conglomerates in those countries are likely to do whatever is necessary to maintain their supremacy. The CHIPS Act thus signals the start of a high-stakes global race, leading to more public and private money in the semiconductor industry.
A race to invest in manufacturing will ultimately flood the market with chips, which is likely to drive down the price and profit margin for all players—as is already being seen with memory chips. Given that such slumps are almost inevitable, it is unclear how American chip makers, with their long-standing focus on quarterly earnings, will deliver on promises of expanding capacity. Asian corporate structures, by contrast, are far more tolerant of temporarily low profit margins.
Yours truly is old enough to remember when the US was a serious semiconductor producer, and earnings of its very capital intensive big players were cyclical, on the order of boom-and-bust-ish.
A wee problem with this picture is that it leads the US public to think the US can cut China down to size with its chips curbs, when tech-wise, the US and China live more in a world of mutually assured destruction. We first mentioned years ago that 80% of US pharmaceutical ingredients, including some finished drugs, come from China.
A more even-handed, and sobering, view comes via a new paper, ASPI’s Critical Technology Tracker, from the Australian Strategic Policy Institute. Philip Pilkington provided an overview on Twitter:
2/ This in spire of the fact the US is still poaching Chinese talent. Although many of these people are likely just learning skills to transfer back to China. pic.twitter.com/TXSHJKAIG3
— Philip Pilkington (@philippilk) March 3, 2023
We’ll stop here for a second. All but the very few truly bicultural Chinese would hit a glass ceiling in US companies and would have good odds of returning to China either dispatched by their US employer (and they might jump ship when back home) or on their own. The new open hostility towards China is sure to reduce how much Chinese “talent” comes to the US. And before you declare than means China is depriving itself of access to science-and-technology leading US schools, think again. The paper lists top academic institutions in the various technology categories, showing Chinese leadership in research generally corresponds to a strong real-world position.
4/ They are also way ahead on communications, AI and other stuff that’ll be important in the commercial economy in the coming future. pic.twitter.com/NFJAi4smzp
— Philip Pilkington (@philippilk) March 3, 2023
5/ If America thinks this can be overcome by sanctions they’re deluded. Sanctions will only make US companies even less competitive as they’re starved of export income and CapEx. Compete or die.
— Philip Pilkington (@philippilk) March 3, 2023
It looked as if the study attempted to throw some bones to the US. It dignifies our balloon panic:
Although balloons are conceptually low tech, their ability to (at least sometimes) slip through detection systems and carry heavy payloads is extremely valuable. The Financial Times reported that Chinese state television showed footage of high-altitude balloons carrying hypersonic glide vehicles in 2018, but that the video is no longer available. Video matching the description can be found on Twitter and Toutiao. Comments below the video state these were scale models of hypersonic glide vehicles used for testing, and suggest the wing design matches the ‘I-plane hypersonic concept’ from the Chinese Academy of Sciences. The 2018 research paper describing this design has been cited by, so far, 19 subsequent research papers. Thus, it’s likely that high-altitude balloon research has
directly contributed to the cost-effective testing and development of nuclear-capable hypersonic glide vehicles.
Douglas Macgregor pointed out that for the US touring Chinese balloon to have had meaningful surveillance capabilities, it would have had to carry a payload similar to the one of the Goodyear blimp.
Oddly the report did not spend much time on medicine or pharmaceuticals despite continuing development in areas like robot assisted surgeries and the use of AI in diagnostics. Instead we get:
Now admittedly medicine took a big step back under Mao’s efforts to push traditional Chinese medicine, and the fact that doctors are not highly esteemed or well paid (while by contrast, both Singapore and Thailand are medical tourism destinations). And this study focuses on major areas of technology advancement, not routine practice. Nevertheless, I found this bit to be surprising:
It’s also in front in the crucial areas of quantum computing and vaccines (and medical countermeasures). This is consistent with analysis showing that the US holds the most Covid-19 vaccine patents and sits at the centre of this global collaboration network. Medical countermeasures provide protection (and post-exposure management) for military and civilian people against chemical, biological, radiological and nuclear material by providing rapid field-based diagnostics and therapeutics (such as antiviral medications) in addition to vaccines.
America’s terrible performance in Covid infections and deaths, and in our “medical countermeasures” as witness the inaction after the East Palestine toxic explosion, raises questions about whether our supposed excellence actually benefits anyone other than the vendors.
As I said, this was intended to be a high-level introduction, so forgive me for being broad brush. No doubt we’ll be returning to this topic.
60 comments
Comments are closed.
Andrei Martyanov made fun of this ASPI study, as Russia appears nowhere.
That is actually a side-effect of how the study was done. For each “key technology” they track publications, graduates and where they find employment. Among other things.
I assume that the chosen methodology allows for predicting future trends to some degree, it also misses a lot of what is actually happening. So Russia being in the factual lead of hypersonic engines can not be captured by these metrics, since Russians don’t publish their research nor do they hire foreigners to implement the research.
There is another aspect to this. From the report:
Results weren’t filtered by language, but the overwhelming majority of reports (98.7%) were written in English. This means that research papers published in domestic journals in, say, Japan, China, South Korea, France or Indonesia, outside of the world’s major journals, aren’t captured in this data collection, and that’s of course a limitation.
Of course, most major/high-impact science/technology/engineering journals are published in English and “everybody” tries to get articles in them, but for countries or domains where publications in national languages are more usual or whose researchers are less “globalized” (perhaps Russia), this may introduce distortions.
This might help explain why the U.K. does better and possibly Germany / France less well in many categories than I might otherwise expect. Was trying to figure that out.
May also be that the U.K. (and US) are flattered by many foreign researchers working in their research institutions. Chinese research seems 100% indigenous so that mean that their true lead or potential might even be understated, given that these people are likely to stay in China. The summary states some of this too.
Agree with Polar Socialist that this is probably helpful for predicting future trends. It is obviously not fully definitive.
About 20 years ago, I had to assist foreign language exam of a Ph.D. student by selecting a scientific paper in German in the research area of this students. I had to go at least 10 years back. Seems that these days all Ph.D’s learn English and write in English, except for national language versions when required. Compare with the time of Newton and Leibniz when learned articles and books had to be in Latin, cf. facsimile of Newton working copy https://en.wikipedia.org/wiki/Writing_of_Principia_Mathematica#/media/File:Principia_Page_1726.jpg
Of course, industrial and military research can be different, and there is a big market for advanced educational materials in national languages.
Another issue: the edge or lack of it in knowledge and industrial know-how cannot be measured precisely. One fact is that many countries are leaders at something, and splitting the world into blocks produces a lot of hurt. Like an operation on Siamese twins, even if there is a potential success, it is not a simple job.
Know-how requires also to know how things do not work, track records of bad ideas and errors of the past, so loosing continuity in an industry may be hard to overcome. For example, the West lost continuity in building nuclear power plants, and the efforts of the last decade are pitiful compared to Russian and Chinese who build within budget, on time and pushing the technology forward. Chinese got superiority in solar power. So “global reduction of carbon emissions” will suffer if the world is butchered into blocks. And this is but one of many aspects.
I was remiss in not pointing out the peculiar absence of Russia, but that omission does not affect the focus of the post, the US v. China. The fact that the US is tops in only 7 of 44 tech categories is bad enough. It could not do better if Russia were included.
Fascinating! Are there measures to increase funding for expanded policing of new US sanctions? I imagine that Milo Minderbinder is celebrating either way,
Pip-Pip!
This is an intriguing post, and I tend to agree with the overall assessment–that the U.S. foreign-policy claque thinks that sanctions are going to work *this time*–and not cause self-harm or blowback.
Lots of crunchy data in the post, too, to contemplate.
Philip Pilkington’s tweets reposting the info-graphic are especially interesting and explain certain, eerrrrrr, tensions in the alliances and lack of alliances.
Note the number of Indian flags. India is squarely in the middle and will profit from a neutral foreign policy. Even this rather simple info-graphic makes that clear.
The data that intrigue me most, here in the Undisclosed Region and its capoluogo, Chocolate City, are the appearances of Italian flags: Seven Italian appearances–as opposed to Germany at six. And the population of Italy is some 20 million less than that of Germany.
So: It is already dawning on Italians how much they have at stake: Premier economy of the Mediterranean. Traditional crossroads / trading nation. Energy resources that amount to un bel niente. + The Italians, justifiably, are immensely proud of the economy.
[[Don’t get me going on the weaknesses. Lacking a minimum-wage law? The absurd system allowing high-school students to be placed out as “interns”? Sheesh.]]
Like India, Italy (which is also geographically vulnerable) is better off in the middle, as a trading nation.
Yet when, for very obvious reasons, as Pilkington’s tweets clarify, Italy signed an agreement with China to engage with the New Silk Road, the U.S. foreign-policy claque had a cow.
Someone’s afraid of the Economic Spaghetti Monster?
All but the very few truly bicultural Chinese would hit a glass ceiling in US companies
Some time ago, I read an article about Morris Chang. I seem to remember that, although he made a brilliant career at Texas Instruments and reached what was just one step below CEO, he was ultimately stymied. Judging that there was no chance his ideas would be fully accepted in the USA, and enticed by the government of Taiwan, he went there, where he founded TSMC a few years after his return.
Now-a-days it probably would be recorded as he:
https://theintercept.com/2020/02/02/fbi-chinese-scientists-surveillance/
https://www.justice.gov/opa/pr/harvard-university-professor-and-two-chinese-nationals-charged-three-separate-china-related
Agree with above comment that this is an intriguing post.
China clearly has plenty of ways to hit the US / West too so unclear what the Biden administration really will achieve here, other than entering an escalator that ratchets up conflict from name calling to sanctions then more sanctions then provocations then worse name calling until we get to some form of armed conflict. The latter will likely be by proxy given US precedents and the general disliking of body bags flowing home. This is as opposed to limited concern if proxy nationals get killed. Sorry to be so macabre but it is hard to see how else this ends unless the collective west comes to its senses.
The various quoted articles on chips do not seem to mention Eindhoven based Dutch ASML, which was originally a Philips spin out, and who make the high tech machines for producing them. Am no expert on that full supply chain but my understanding is that a large part of the core lithography technology needed is controlled by them and then indirectly by Germany’s Zeiss who make the optics needed in the machines. There are competitors such as Nikon mentioned above but the high end manufacture is still very much dependent on ASML, I believe. Some of the technology may have originally been US but clearly cooperation with the Netherlands and Germany seems crucial for any form of effective sanctions. Perhaps this is just a “given”!
Ultimately, in the longer term sanctions will anyway spur China with over a billion people to invest in technologies that it might not have bothered with otherwise. Nice job, collective west. If I were ASML I might start to get more worried now that this highly successful carve out from Philips could eventually go the same way as the rest of the corporation did and lose its global leadership.
clearly cooperation with the Netherlands and Germany seems crucial for any form of effective sanctions.
That is basically what happened — for some definition of “cooperation”.
Since there are USA-sourced patents and components (hardware and software) relied upon in the IC-manufacturing tools produced in the Netherlands, Japan, and Germany, these tools fall in the scope of sanctions: firms are not allowed to sell US technology to China, directly or indirectly, which implies that if ASML attempts to sell one of its machines integrating some US component in it, or some subsystem including a US component, or some component or subsystem designed with US software, and so on recursively, then it falls foul of the sanctions.
Thus, despite being wholly incapable of providing complete systems or just subsystems for chip manufacturing, the USA can still prevent other countries that do (Japan, the Netherlands, Germany, Korea) to sell them to China. This is exactly the same technique that was used to starve Chinese firms like ZTE and Huawei from components and patents at the core of wireless networking infrastructure, of mobile devices, and of tools to design such products.
That makes sense. Unless Europe chooses to get into a full blown trade / international law battle with the US then clearly that works and can, of course, be applied to pretty much any supply chain. It is hard to think of one that does not have some US IP, component or US owned company somewhere. And of course, Europe is “cooperating”. At least under the current set of elites.
Underlines the comment of NotTimothyGeithner. Ultimately, these sanctions will hurt Europe, or at least contribute to European dependency on the US. Achieving that has been one of the “successes” of US foreign policy. The EU was originally conceived of by some as in part a way for Europe to become a super power in its own right. Instead, it has become a useful vehicle for external rule.
Technology decoupling, i.e. making sure China (and Russia, etc) does not depend directly or indirectly on US components and intellectual property, is extremely difficult.
It essentially implies autarchy at the level of a trade block, and may eventually entail branching-out (e.g. different and non-interoperable mobile networking standards, components and products; chips designed and manufactured following different approaches and that cannot be mixed and matched; propulsion systems designed for incompatible fuels or battery technologies; etc)
So then we go the old way of just copying as best as we can other’s tools, eh? And once you know it can be done, it will be quite motivational…
And when the US will be left behind what will happen? Try to wage wars of conquest to appropriate others’ achievements, as the UCS did previously with all the German technology they captured after WWII? I don’t think it will work this time…
So then we go the old way of just copying as best as we can other’s tools, eh? And once you know it can be done, it will be quite motivational…
And when the US will be left behind what will happen? Try to wage wars of conquest to appropriate others’ achievements, as the UCS did previously with all the German technology they captured after WWII? I don’t think it will work this time…
I was trying to look up good and brief history of US (and Allies’) seizure of German patents during World War 1 (the story of Aspirin is well known, but I couldn’t remember the details, and I knew there were lots of other stuff.). What I came across seems a lot more distressing and, worse still, dangerously familiar:
https://www.smithsonianmag.com/history/us-confiscated-half-billion-dollars-private-property-during-wwi-180952144/
IP is easier to just middel finger though, as Russia did a year back by declaring it would not uphold copyright on foreign software. Thus basically legalizing piracy of everything from IBM, Microsoft, Oracle, etc etc etc.
And IP claims are hard to enforce internationally, unless you want to get into a shooting war over a piece of paper.
Even USA bootstrapped its industry by ignoring English patent claims. Just look at the likes of Samuel Slater, aka Slater the Traitor in UK.
‘I cannot see how this will be China’s century’: Dutch PM Rutte praises ‘the great role’ of the U.S. after the country backs Biden’s chip controls
The Dutch government early on made show of seeming independent, but unfortunately that went out of the window pronto. Disappointing, I say, being Dutch.
Lets face it, not even France, for all its complaining, can really claim to be independent these days (though it is one of few, if not only, NATO nations without a US base).
I think Michael Hudson is right. These aren’t sanctions against Beijing but against Europe. China might take a hit, but between a strong government and domestic goodwill from a rally around the flag event as the US imposes sanctions, China will be fine. Remember the Inflation Reduction Act is out there too, putting obstacles to Euro exports to the US. Once is happenchance, but three times is Biden pantsing Scholz on CNN. Even a potential labor force in Ukraine is being destroyed. Borrell types who can’t see past their own skin are eager helpers in bringing Europe down.
Remember Bush and Obama vowed of have no competitors. Arguably the real competitor to the US is the EU with its “liberal democracy” and more recent multiculturalism. The Middle Kingdom and Russia are huge, but they’ve been huge for a while. They aren’t places people simply join. EU elites just bought the propaganda about Trump and decided Biden would be nice to them or something.
Rumsfeld’s “new Europe” Microstates are out making deranged demands of Berlin.
I think commenters are missing Stephen’s very important point – these sanctions are likely to spur China to compete with ASML. ASML currently has a near monopoly on the key high end chipmaking equipment – but if China puts her considerable resources into building a competitor she can well do that. This is made easier by the fact that China can attack from below (in a manner straight out of Clayton M Christensen’s ‘The Innovator’s Dilemma’) because the higher volume simpler integrated circuits don’t need 5nm processes.
All in all it looks to me like ASML is on its way to getting Clayton-M-Christensen-ed.
It already has.
China did make some chip fab announcements recently.
and there is also at least one Chinese company that announced a “home grown” (execs are former Nvidia employees) GPU.
For the moment that GPU is being made at TSMC fabs but the pieces are in place for China to pivot away from needing TSMC.
Never mind that you do not need the latest desktop hardware to make weapon systems etc.
Frankly USA is trying to close the barn door after having sold all the horses to China.
I recall the big elephant in the room regarding semiconductors was that Ukraine and Russia produced over 80% of the neon gas needed in photolithography. No neon, no wafers, no chips.
Apropos, on Thursday Russian Ministry of Industry and Trade will grant $14M (but in rubles) to Russian companies for developing domestic 248 nanometer eximer lasers for lithography. They also have two Russian chip companies lined up for development feedback before becoming “end users”. The news itself mentioned 14nm process, but I’m not sure they’ll get there by 2025 (which is the target).
So how important are all the latest whiz bang chips anyway? Cars don’t need them as 1990s era chips do just fine for engine control functions. Indeed it has been suggested that the shortage of those older style chips can partly be blamed on chip foundries switching to the latest and therefore more profitable chips even as the personal electronic sales that promote them are cratering. Personally I think I have all the consumer electronics I will ever need and planned obsolescence game is hampered by the fact that digital devices are far more durable than the partly or mostly analog versions that preceded them.
And the latest AI craze seems to be turning into something of a joke. There are those military and science applications but somehow the Russians appear to be doing just fine in this area despite the heavy sanctions that were supposed to cripple them. Surely any techno war against China is going to be hampered by the fact that intellectual property is the most portable and stealable kind there is. If you make China an IP precrime outlaw then they will simply become the real thing as they have in the past.
Out here in the peanut gallery none of this makes any sense. Since the PMCs aren’t really interested in bringing back manufacturing and the working class then what are they up to? Do they even know themselves?
I think they are interested in theory, but they are doing it through rules and “classical liberalism.” Biden can’t conceive of industrial policy, but ceos make the same excuses to politicians about jobs in the US. It’s cheaper other places, even Europe because they don’t have to worry about Healthcare costs. Biden et al are too full of themselves to understand the Chinese industrialization wasn’t the result of being showered with gifts by the US. That helped, but it’s not the full story.
all of this was predicted in the 1990’s. its to late now to close the barn door, that time was in 1993, and bill clinton should have been roped in.
https://www.youtube.com/watch?v=wwmOkaKh3-s&t=180s
so what is more efficient, human and environmental degradation for cheap trinkets and more profits for wall street, or our standard of living, the environment, and our technology?
I cannot go into details, let’s just say way back in the day I reported to the ERDA, and had a clearance with a letter.
Let’s just say that the DOD was clearly told in the very early 80’s that the out sourcing/off-shoring of chip technology would result in today’s situation. And people need to understand that despite everything Silicon Valley later became, there was really only ONE customer that made Silicon Valley happen:
Silicon Valley rooted in backing from US military
https://www.ft.com/content/8c0152d2-d0f2-11e2-be7b-00144feab7de
Not disputing what you say, just that many powerful people in the DOD knew what was coming a DECADE before you point to, and it was a much more pointed warning. They were POWERLESS to stop it.
I’ll be blunt, America, and Americans were betrayed, betrayed by the American elites that REALLY run the country. (I tend to lump main stream politicians as for-hire employees of the elites that actually run the country – follow the money.)
but it was bill clinton that made it happen. america like almost all countries have a business section sick with greed.
but its politicians driven by idiotologies that make it happen.
According to director of Zelenograd Nanotechnology Centre Anatoly Kovalyov the most used chips in today are 65nm to 180 nm (so tech from 1999 to 2005). They cover all from cars to telecommunication.
And Russia can get all they need at the moment by “parallel import” but they still need to develop domestic technology to gain more independence.
Production of chips under 65nm is also much needed to retain the Russian expert teams that can design those, but it’s a much bigger problem to solve. I guess the same would apply to China, too.
You highlight the annoying gross simplications used when reporting on the semiconductor industry, such as referring to the entire sector as “computer chips”. Microprocessors and memory are a fraction of the product lines.
The real idiocy of this is that so many commodity parts are available cheaply in China which are unheard of by Western design teams, so they use Texas Instrument, Analog Device, Microchip, ST, Renesas, NXP.
Example: Can get a stick of 100 pieces of a full-featured microcontroller for 3.44c each off the shelf. Nothing from the suppliers mentioned above is available at 4x the price.
https://www.lcsc.com/product-detail/Microcontroller-Units-MCUs-MPUs-SOCs_Nyquest-Tech-NY8A051H_C5143390.html
So Western volume-made consumer-targetted equipment such as washing machine controllers, solar inverters, USB chargers heating controllers will end up being completely uncompetitive because access to the cheap parts will disappear, not just the cheap manufacturing.
What people who have never worked in manufacturing don’t always understand is that most “high tech” components are supported by lots of “low tech” components. With the exception of the defense industry(limited production with minimal surge capacity) almost all the low tech components are made in China. High tech components are useless without fasteners to hold them in place, glue to dampen vibration and impacts, and power supplies to give them power. We’ve outsourced ALL of those types of items.
I remember reading that the iPhone glass and the technology for the very thin and strong case moulding was only available in China….
Actually, the curved glass for Samsung phones currently is only made by a US company. My husband was the VP of Marketing and Sales back 20 yrs. ago. Don’t know about the I phones. They were also highly involved in supply material to almost all of the semiconductor manufacturers.
And all the AI libraries are highly parallelizable which means that if you don’t have the most advanced chips you simply train your model on 8000 1-core CPUs instead of 1000 8-core CPUs. Big deal.
The bozos at the think tanks are obsessed with semiconductor technology because they know the AI libraries that everyone uses are all open source – so good luck trying to restrict China’s access to the stuff that really matters (which is, as always, the software).
As i understand it, AI depend more on GPU than CPU. This because GPUs are set up for doing massive number of iterations on the same instructions really fast.
From the documentation for XGBoost – the library that the data scientists at my company like best:
“XGBoost supports fully distributed GPU training using Dask, Spark and PySpark.”
So whether you are using CPUs or GPUs – you can just use more of them if the ones you have are slower. All of the software is written from the ground up to be massively parallelizable whether running on CPUs or GPUs.
That is the basic thing, a upgrade in chip fabs can do multiple things.
The first is to allow for higher performance chips by allowing for more transistors etc on a die.
But the second is that it can allow fabs to get a higher yield of working chips out of each production run (some chips will fail thanks to impurities in the wafer being used etc). This by simply packing more dies pr wafer. The big name home computer CPUs of the 80s, the Z80 and 6502, live on today as microcontrollers at happy meal prices.
You just know that as soon as Washington walks away from the Ukraine, that they are going to devote their entire attention to China and start them on the sanctions escalator. The sanctions on Russia should serve as a warning on the effects of blowback of sanctions but neocons never learn nor do they have a reverse gear. Chips is one of the first major steps but it will extend to everything that the US does not vitally need. But China is not Russia. Instead of the 800-pound gorilla in the room, they are more the 2,000 pound gorilla like it or not. Getting into a trade war with China is a magnitude of difficulty more harder and they are getting ready for the coming fight. And the worse of it is that they may make it an existential fight for China.
China has 3 millenia of history behind her, so they may do existential differently from The West.
For which it may turn existential much faster than we might hope for. Considering that 85% (or so) of the globe has decided to sit the NATO-Russia out, it may turn out that the percentage will be higher
ifwhen The West goes after China.Maybe to the extent that there will be no West after the dust settles. Or if there is, it’s an “outdoors museum” for the two centuries The West was lucky and punched way above it’s weight. Well, punched, kicked, raped, looted and tortured.
Much of it thanks to iterating on gunpowder way faster than anyone else.
It does seem that Ukraine was a Clinton-Biden project, as seen with Hunter partying it up there for a time.
China is a whole other ball game though, thanks to it growing big on US offshoring.
I’m starting to get nervous regarding anxiety meds. I know they are overprescribed, but for me they are a lifesaver to deal with anxiety. The last time I went off them I had lighting like electrical shocks for several months. Should supply chain issues arise, I can look forward to the double whammy of the return of my extreme anxiety coupled with about 3 months of physical pain I never want to experience again.
I wonder how many other Americans face this predicate. I am starting to think about talking to my doctor about a prolonged weaning off to try to lessen the physical effects.
The biggest flaw in the US ‘strategy’ is that it assumes that China is helpless to defeat this specific move. What will happen is that within ten / fifteen years China will have its own homegrown microchip fabricating technology and will lead the world in chip design and production. There is absolutely nothing that US techs can design or make that the Chinese can’t equal or indeed exceed. Chinese patent applications in 2021 were 1,538,558 – the US’s were 509,864 – only a third of China’s. Even if a percentage of Chinese applications were trivial or defective, they still are way ahead.
China is also turning out 8 STEM graduates for every one in the US. In my view the US is already behind and simply trying to hold onto second place.
The US has a strategy of skimming off the cream of other countries. When I interview to hire engineers for my software development team, 1 out of 20 of the candidates are born in North America.
Countermeasures can have offensive functions. NBC gear for combat troops would be an example.
I think the available evidence suggests this has been the US focus. The research proposal for an aerosolized, orthopox-vectored spike vaccine for bats (rejected by DARPA for, among other things, reasons of Chinese public safety) was for open-air animal testing of a means by which populations meant to be protected from a bioweapon could be covertly vaccinated en masse prior to its deployment. This is a big deal because it would, at long last, make bioweapons “useable” in a way that they never really have been.
Mass vaccination of wild animals, possibly using a replicating, transmissible vector, was a recurrent theme with the DARPA/Detrick adjacent virologists. This is so obviously an insanely bad idea that I have no compunctions about questioning their motives. That there is an obvious offensive application for the technology (which btw wouldn’t really have to work all that well to reach “usable” status) ought I think to put the research at WIV and a number of other places into the category of “Bioweapons Research Till Proven Otherwise”.
American history shows that only a severe crisis leads to an improvement in national leadership (e.g., Lincoln or FDR). The Washington neocons will not be driven out of our government until the U.S. suffers a significant military and/or economic calamity. Consider how the memory hole easily swallowed the U.S. Afghanistan debacle to understand the scale of disaster that could cause a change in course for American foreign policy. Things are going to get ugly.
That! I fear you are right. The scary thing is when reading user comments in, say, Der Spiegel, Zeit, NYT, NZZ, etc, it appears most readers seem to swallow the massive amount of propaganda portraying China as an aggressive menace and seem to welcome a possible conflict. It is insane.
Impeding China’s capacity to manufacture semiconductors might also be a strategy to deter them from invading Taiwan.
Apparently China only manufactures 6% of the semiconductors they use domestically, with Taiwan supplying most of the deficit. It’s been suggested that Taiwan has threatened to destroy all its semiconductor manufacturing capabilities if invaded, which would leave China in a very difficult position. My guess is this is why China is trying to build the capability to manufacture more of their own advanced chips. And why we are trying to block them.
According to a 2021 U.S. Army paper about Deterring China from Invading Taiwan (I found this link in a recent NY Times editorial explaining the consequences of war with China):
“the United States and Taiwan should lay plans for a targeted scorched-earth strategy that would render Taiwan not just unattractive if ever seized by force, but positively costly to maintain. This could be done most effectively by threatening to destroy facilities belonging to the Taiwan Semiconductor Manufacturing Company, the most important chipmaker in the world and China’s most important supplier. Samsung based in South Korea (a US ally) is the only alternative for cutting-edge designs. Despite a huge Chinese effort for a “Made in China” chip industry, only 6 percent of semiconductors used in China were produced domestically in 2020. If Taiwan Semiconductor Manufacturing Company’s facilities went offline, companies around the globe would find it difficult to continue operations. This development would mean China’s high-tech industries would be immobilized at precisely the same time the nation was embroiled in a massive war effort. Even when the formal war ended, the economic costs would persist for years. This problem would be a dangerous cocktail from the perspective of the Chinese Communist Party, the legitimacy of which is predicated on promises of domestic tranquility, national resilience, and sustained economic growth. The challenge, of course, is to make such a threat credible to Chinese decisionmakers. They must absolutely believe Taiwan’s semiconductor industry would be destroyed in the event of an invasion.”
The West probably sources most of its chips from Taiwan (and South Korea). Any war in Taiwan would be the end of that.
And now that advanced chips from Taiwan are no longer going to be available to China, there goes the incentive for China to refrain from destroying TSMC by kinetic means. China would have little to lose in doing so. The country that has made itself China’s implacable enemy — the United States — would be the big loser in that (admittedly) far fetched scenario. China has little to gain in continuing to appease the United States in anything.
TSMC is being assisted under the CHIPS Act to build a manufacturing plant in Arizona, supposedly with first production starting in 2024. This would appear to be an obvious Plan B for US semiconductor supply in the event of a CCP invasion of Taiwan.
Of course this also raises the possibility of the likely invasion occurring sooner rather than the 5 year timeframe touted as likely by US Defence analysts so as to deter a scorched earth strategy by the Taiwanese.
This one fab will be completely unable to cover the US demand. Also it will only cover some more advanced technologies (not the most advanced, though) and will therefore be completely useless to supply chips for appliances and cars which use “older” technologies. A war with China over Taiwan will make the Covid supply chain problems look like a picnic.
Industries tend to cluster:
https://hbr.org/1998/11/clusters-and-the-new-economics-of-competition
… and what has been emerging in recent years is a semiconductor manufacturing cluster centered in Taiwan but increasingly including Chinese cities from Shenzhen to Shanghai. This cluster shares common suppliers and a common pool of Mandarin speaking experts who move back and forth between Chinese and Taiwanese firms.
I believe that it is this cluster that the US is trying desperately to disrupt. But it is very difficult for firms outside of a cluster to complete with firms inside of a cluster. Very difficult.
I think that is the fundamental paradox that a Taiwan crisis, if it does arise, will be coming across.
Majority of Taiwanese may not want to be ruled by Beijing, but…
Even more Taiwanese–even those who like the idea of independence–are quite happy being an integral part of the Sinocentric economic system, in which they happen to enjoy a significant and privileged position because they too are “Chinese” (I don’t know about the legal aspects–I think there are a lot there, too–but certainly sociocultural dimensions must be significant).
So the desire for “independence” is abstract and theoretical, even if the idea might be “popular.” It can stay that way as long as the reality doesn’t interfere with the fantasizing. But if the Taiwanese are forced to face the price they would have to pay, to have their “independence” enforced at the very distant gunpoint by the United States, can anyone expect how the Taiwanese will react? I don’t think any more than handful of Taiwanese will actually pay anything for “independence.” It’s too expensive and the cultural Chinese are practical people. They are not South Carolinians in 1860.
Shenzhen is an interesting one, as the place basically got started by being a sub-contractor to Hong Kong.
Do 99% of the production in Shenzhen, load it onto trucks, drive it over the border to then British Hong Kong, add the last few screws, put it in a box, stamp it all Made in Hong Kong.
And the effect is astonishing. If you look up historic images of the area, it was sleepy fishing villages and river delta. Now it is part of perhaps the world’s first mega city, stretching around the Pearl River estuary from Hong Kong to Macao.
Entry barriers are high in the semiconductor manufacturing industry, and especially in the industry that makes the tools to make the chips. Until Trump’s and Biden’s sanctions came along, there was little commercial reason for Chinese chip manufacturers to buy Chinese made lithography or etching machines — American, Japanese and Dutch companies dominated the industry — and so there was little incentive for Chinese companies to get into that line of business. Biden’s recent sanctions have changed that.
Now that the Chinese market for this equipment has been handed to Chinese companies, there will be commercial incentive to scale up all aspects of chip design and manufacturing using Chinese tools.
Dan Wang published a great article in Foreign Affairs this month on the strengths China will be bringing to developing its chip design and manufacturing industries, and the industry to manufacture chip making tools. According to Wang, China’s strength is in figuring out how to take scientific and technological advances and turn them into high quality products manufactured at scale. Chinese companies can do this partly because of the breadth and depth of China’s manufacturing ecosystem. Here is a link to Wang’s article, which is highly recommended: https://www.foreignaffairs.com/china/chinas-hidden-tech-revolution-how-beijing-threatens-us-dominance-dan-wang
William Verick,
I agree with you completely and I appreciate the link.
Reading this article, I have the impression that the writer sees Europe as an American sub-state. Obviously he sees all European tech as US’s. Not very friendly.