SMIC has been forced to focus only on DUV technolog, and take that as far as it can go. That’s not bad for the short term. DUV is actually just as good if not better than EUV for the nodes they’re working on.
TSMC is putting a lot of their R&D into next generation technologies like EUV, backside power, advanced packaging/chiplet stuff, glass substrates etc
I’m not sure SMIC will be able to keep up when these efforts bear fruit. Eventually they’ll need EUV to unlock further progress.
SMIC has been able to poach talent by throwing mountains of money at them. This might not be possible going forward. Not being able to work on EUV could be career suicide, and I’m not so sure China can be so frivolous with their money anymore. They have to start raising taxes to compensate for the loss of land lease revenue.
To quote a random site that I googled while looking for the acronyms:
>Think of EUV as a precision scalpel, capable of carving out the finest details, while DUV is more like a skilled artist's brush – versatile and reliable, but with limitations when it comes to the tiniest features.
The current technology to generate extreme ultraviolet involves a container of molten/liquid tin that is released drop by drop; these are vaporized by a laser. The equipment for this is sold by one vendor, ASML (a company in the Netherlands). It is extremely expensive.
China has taken large strides in every advanced manufacturing process. Made in China used to indicate low quality but that's far far from the current situation.
I just don't see why they wouldn't eventually be able to dominate high end chip production, just like they've catched up in other areas where people said they won't be able to compete.
The sanctions might delay this for a couple of years, but it seems rather futile to think it can be delayed more than that.
In my opinion it's a bad look on the US part to try to deny a hard working people access to tools to develop technology and produce wealth.
There was some news a while back that they had mastered making ball point pen tips. But you gotta keep in mind that these efforts have to be sustained long term and have to be a self-reliant economy on their own.
Anyone can compete short term if you fuel your efforts with mountains of debt. But China's debt level (shadow debt included) has reached a breaking point. How many chip company bankruptcies can they take before the efforts break down?
China can use a lot of "tricks" with their economy, since they're so authoritarian. But they're not THAT special. Their economy is not magic. The local governments have an insane amount of debt now, and they're in a situation where they can't make as much money off land leases anymore, so they have to raise taxes to avoid becoming insolvent. Yet foreign companies are already moving factories out of the country, so how can they stay competitive while raising taxes? This was already a huge debt, it was just never on paper. Now that debt is coming due.
Keep in mind that the local governments are often involved in financing China's domestic industries..
Think about how the outlook would be if SMIC was in any other country. How would a chip company ever survive producing near cutting edge chips without a strong international market and high margins? If a western country bought hundreds of insanely expensive chip manufacturing machines that ended up gathering dust in a warehouse, it would be considered insane. This isn't sustainable.
China has a decent domestic market, but it's very hard to sustain good margins there. Margins they desperately need to sustain an industrial effort that is being run in a much less economically efficient way than the western chip industry.
If anything I think the sanctions encourage China to make domestic competitors for what they're cut off from.
And it's not like they're new to this field, chinese nationals almost certainly were among those that contributed to EUV development in the first place. There just isn't enough space in the market for more players like ASML without some backing geopolitical reason forcing the creation of new players.
> China has taken large strides in every advanced manufacturing process.
No they haven't, as in foreign companies in China have and invested in that. Not China itself. This is the big difference. So who is building these advanced manufacturing processes for them?
> I just don't see why they wouldn't eventually be able to dominate high end chip production
They've never bothered to. There's been huge corruption, lots of mismanagement, etc. Imagine a worse Intel.
> hard working people access to tools to develop technology and produce wealth.
Who says they're hard working? At least not in this sector. China has had plenty of time and resources before the sanctions and they've wasted it. They spent a staggering amount of $$$ (huge government investments), poached infinite amounts of talent and floundered it all already. Who says this will change?
China leads in 57 out of 64 technologies, up from just 3 twenty years ago. [1]
The US lost its research advantage: it was leading in 60 technologies 20 years ago, down to just 7 today.[1]
If this is in reply to China being hard working, then sorry, no. They're definitely hard working in obtaining technology. Maybe not by means you're thinking of (see e.g. Nortel, HTC or Via).
Also, compared to 20 years ago? They were less open about it. Plenty of US, Singapore, Taiwan, Hong Kong, etc based companies were all just fronts for actual Chinese companies. The "made in China" logo back then didn't have Western appeal. Nowadays people are much more accepting of it.
This seems to be a joke. China has 70% of the world global research share on advanced aircraft engines Seriously? Maybe in writing papers . But that does not translate to actual product.
China has 50% of electronic warfare and 45% of radar? In what planet?
Any test to see if a lot of what is going on in other countries is not published?
Is that really an honest question? Because it looks like a setup for some trollish "hurr Intel doesn't count because they're stinky" nonsense. Yes, America makes chips. Always has and never stopped. Even of you don't like Intel, America's chip production is obviously sufficient for national security intents and purposes.
Intel still has fabs in the US, TSMC is building more fabs in the US. TI also makes chips in the us. The capability is there, but it's also interlinked with a global supply chain for silicon and other things.
So can the west make chips without TSMC? Yes. Very much so. Can the US make chips on their own, in a more limited way.
Intel is struggling, TSMC found it extremely difficult to set operations there, that doesn't smell good. For >20nm chips sure, for bleeding edge that's yet to see.
China is really really broke, from central government, to local government, to enterprise, to individuals. from 11 trillion off the book debts for local governments, to record number of Chinese blacklisted for debt defaults. from 7 trillion lost in stock market value in 3 years, to close to 20 trillion wealth wiped out in real estate.
I am not sure if being broke is important or not. They have created a dependency on themselves with the belt and road countries to where they have to buy Chinese products. China is doing currency swaps and lending in their own currency that countries then use to buy from China. Normally if you have a massive housing crises, jobs crises, debt crises, etc you would completely fail. But yet China keeps going. I believe they have captured a large group of forced consumers to prop up the Chinese government. Africans are not looking for the latest tech they just need things that work good enough. The US can not serve all of these markets (at a reasonable price point) so the Chinese companies dominate which feeds back up to the CCP staying in power.
I'm wondering how to think about debt load in China.
If asset prices are too high, they will go down. When a debt defaults, someone doesn't have as much money as they thought they had. This is deflationary.
Meanwhile, the government can print money or raise taxes if they need to.
What are their real (rather than financial) constraints?
If you borrow in your own currency there's no real constraints. The trade-off is fiscal deficit = printing money and may cause inflation. Then higher interest rates will put pressure on economic growth.
Entities in the government are broke so there will need to be a clearing event. What better way to restructure debt than a war? China is extremely rich in mid grade mfg capabilities, and would be able to create a large quantity of good enough weapons to successfully execute such an endeavor.
China's debt is mostly internal, not external. It doesn't need a war to restructure debt. However, it does need to pay out an increasing amount of social pensions to its aging population, maintain its increasingly expensive and decreasingly utilized rail infrastructure, fund its growing state enterprises, and maintain its increasing military endeavors, etc. In other words, China's debt is owed to the citizens, but it now lacks the ability to pay them due to collapsing revenue from real estate.
It could try to print massively, but then its cash will be toilet paper, and for a country that mainly imports its resources, that would be devastating. So right now it's just printing quietly and in small doses.
> China’s relevance as a manufacturing hub has risen from about 5% of global manufacturing production in 1995 to about 35% in 2020. Its production share was higher in 2022 than that of the next nine largest manufacturing countries in the world combined, including the US, Japan and Germany
> Jean et al. (2023) look at global dominance from another perspective. The authors identify product-level dominant positions in world trade, defined as a share of more than 50% of worldwide exports. Based on this measure, they find that out of a total of about 5,000 products, China held a dominant position in almost 600 products in 2019. This is outstanding and atypical when comparing this result to other countries and over time since 1970. For example, China holds a dominant position in at least six times as many products as the United States, Japan or any other country and twice as many products as the European Union considered as a whole
Every (sane) company now has a China+1 strategy, mainly manufacturing for Chinese audience in China, and elsewhere for other audiences. It's easier to decouple from China now, much easier than 3 years ago.
WTO, IMF have yearly reports and IIRC one of U.S. Central bank branch have published report about this. TL;DR version is there's a trade flow reshuffle but China's trade percentage barely changed. As most traditional economic theories suggested and predicted, China sells more to ASEAN, South America and they sell more to U.S.
And the +1 could never scale to the same capacity as China. Especially not at the same time. You can’t decouple from China, the global economy barely decoupled from Russia and even then it’s questionable how much decoupling actually occurred
That’s all measured as “value-add” manufacturing - that is, the value of the manufactured good minus the value of the raw materials. It still has to import the majority of its resources to support not only for exports but also for its growing middle class.
> It could try to print massively, but then its cash will be toilet paper, and for a country that mainly imports its resources, that would be devastating
China is currently running a massive trade surplus, they could print a ton and still be fine.
Not enough, especially with the increasing trade sanctions and tariffs coming from every other country against China. There's a reason why there are Chinese state employees not having been paid for 8 months, or Chinese citizens unable to withdraw a few hundred dollars from state banks.
There was some talk of Shanghai proposing a new rejuvenation tax for residential buildings this summer. The backlash was so great, that the government quietly backtracked.
> Chinese state employees not having been paid for 8 months
I haven't heard of this before. Do you have a source (pref. one that doesn't start with "radio free" haha)? China's a massive country and "Chinese state employees" can mean all sorts of things
There's a liquidity issue for some heavily in debt local governments. It's also much complex because central government is taking this as a chance to pushing unwelcome structural changes. There's a quick fix have been done before - central government can simply swap local with central government debt and they still have a lot of fiscal room to do so. But many economists are against the approach this time because there's a issue long aware that central government doesn't actually able to control local government's budget but their debt considered backed by central government's unlimited credit.
What make things worse is the political incentive is local government has to grow economy and they can achieve that by borrowing money for investment. Even if the investment return is almost 0 there's a short term GDP boost (just think it as giving money to local workers, similar to Civilian Conservation Corps).
So now what they are pushing is force local government sell unused state assets they owned, cutting civil servant (hurting local consumption and local GDP), limiting their borrowing capacity and all other government reforms. Most importantly they want to break the expectations that local government debts are back by central government which distorted the market a lot in past decades.
Localities are having cashflow issues, since all the VAT basically goes to the central government and without property taxes they rely on land transfer taxes. That could be playing out harshly in lower tiered cities, but I honestly have no idea what is going on ATM.
Any claim about china has to be prefixed with some at least, since there are lots and lots of people.
How do you know they can't print enough when they haven't tried printing money? They could print and thus increase internal consumption until they have a zero trade balance, which would help a lot of their issues, it is unclear how much but it would help.
That’s 15 years. Why not for another 15 years? I mean eventually things might crash, but as the saying goes, “economists have predicted 9 out of last 5 recessions”.
I hear this a lot, I hear it about Russia too, but it also just feels like wishful thinking, it's really hard to know what the reality is, but I guess that's how authoritarian countries work.
If you go to Russia you'll hear the same about rotting Europe and USA. If you bother checking old news it was going on for more than hundred years. Imperialists are still alive, so will be China, don't worry. But, if it will be a nice shock for the rest if they take Taiwan...
It depends if you include SOE debt as a part of China's national debt or not. One thinking is that SOEs are almost private companies, but everyone assumes their loans are backed by local governments at least. Kind of a mess when you try to sort out what is private and public in China's debt. Should the debt of state owned railroads who heavily bond out the development of the HSR network be counted as public debt, or private debt since it is held by private entities? Considering that default is still a possibility (only BJ - SH is profitable), the government will probably have to transfer these to public debt eventually (or alternatively tell lots of private chinese citizens their retirement is SOL).
When talking about national debt we include both private and public debt. By this measure China's debt stands at around $45 trillion while the US' stands at around $70 trillion. China also has a larger GDP (PPP) and 4.25x the population. However you spin it, you'd be hard pressed to defend the claim that China is facing a worse debt crisis than the US is
Maybe you are using a different chinese definition? If so, we are comparing apples and oranges. By the English definition of national debt, China should be pretty good since most debt in China is offloaded to localities or SOEs, or are hidden off books in some weird local lending scheme far away from Beijing.
> China is really really broke, from central government, to local government, to enterprise, to individuals. from 11 trillion off the book debts for local governments, to record number of Chinese blacklisted for debt defaults. from 7 trillion lost in stock market value in 3 years, to close to 20 trillion wealth wiped out in real estate.
Those all seem like things that would be really, really important to someone so immersed in the assumptions of the Western capitalist system that they've confused them for the laws of nature.
But I'm not convinced they'd be so constraining for China.
> SMIC has been able to poach talent by throwing mountains of money at them. This might not be possible going forward. Not being able to work on EUV could be career suicide, and I’m not so sure China can be so frivolous with their money anymore. They have to start raising taxes to compensate for the loss of land lease revenue.
I would be extremely surprised if China wasn't throwing mountains of money at building their own EUV lithography machine so they could totally cut out ASML.
I would also be extremely surprised if such a project wasn't top priority for them, and would be one of the last things to be cut "to compensate for the loss of land lease revenue."
> Yet in 2017, after an investment of $6.5 billion in R&D over 17 years, ASML’s bet began to pay off.
Assuming China's starting from square one, takes the same amount of time and money to develop EUV, and the government pays 100% of the cost, it looks like the yearly cost (382 million) would amount to about 2/3 of their current green energy subsidies ($607 million in 2022 https://www.reuters.com/business/energy/china-sets-2022-rene...)
> I’m not sure SMIC will be able to keep up when these efforts bear fruit. Eventually they’ll need EUV to unlock further progress.
The issue with SMIC and fabs in China in general was never the technology. If only the technology ban was the real problem.
I mean there were ASML EUV setups sitting there for years doing nothing...
> SMIC has been able to poach talent by throwing mountains of money at them. This might not be possible going forward.
That wasn't exactly the tactic and with the comparatively low wages in South Korea and Taiwan, it'd still likely be doable. It's not like they poach everyone - just a few key figures.
Article is misleading, this is in terms of chip design, not in terms of equipment manufacturing which is what they need to scale or improve yield. SMIC has the equipment they bought before sanctions and no ability to make new comparable or better equipment. So really, this is just SMIC squeezing a little more out of what they already have. Their yield still sucks and can’t be dramatically improved.
Also, 7nm mainstream cpu made by TSMC was released 2018
That was the idea with Russia and Germany - did not really work out (but german machine tools continue to churn out weapons in Russia, bought from hydrocarbon profits...).
China chasing TSMC is probably a fool's errand. They might catch up for a bit, but they'll always be renters in the houses that ASML builds (and they can be evicted/sanctioned out at any time[1]).
Rage Against the Machine said it best: "F** tha G-ride I want the machines that are makin' em."
Call me when they get serious about their domestic lithography business.
Your link says that they are also chasing ASML. And if they invade Taiwan, it will take the rest of the world a couple of years to replace TSMC, regardless of ASML.
They might be chasing, but I haven't seen anything showing that they're even close to the same time zone as ASML right now.
It seems like China invading Taiwan doesn't make any strategic sense until they have a serious alternative to ASML. If they invade they'll never lay eyes on ASML hardware again. However, if they have their own serious domestic ASML competitor, plus a bunch of domestic foundries that could pick up TSMC's orders after an invasion, they have the world over a barrel.
Does the teardown consider yields (reading it, they just include an aside comment)? It is easy to make cheap, it is easy to make fast, the hard part is making cheap and fast (the same goes for jet turbine, the other side of China's tech push).
> U.S. export curbs on chip technology have spurred China to become more self-reliant in producing semiconductors.
everyone needs to be more self-reliant in producing semiconductors
how far behind is the US on TSMC? because I want that Arizona supply chain to be up and running Chairman Nao!
literally nobody should be basing any of their decisions on the possibility of China invading China. No wonder TSMC makes up dumb excuses about their issues deploying in the US, they need to stall and delay the obvious, that's the only... chip... they have.
Is it just me or are we past the stage where being at the absolute apex of chip development is strategically important? For at least a decade, chips have been fine and even pretty often overpowered for the vast majority of applications. You want to have access to chips, yes, but who cares if they're five years out of date?
> For at least a decade, chips have been fine and even pretty often overpowered for the vast majority of applications
True, but that changed the last few years, next gen AI require vast amounts of compute, that isn't strategically important today but could be in a few years.
True, but how much has power consumption changed in the last few years? Enough to be a serious obstacle in the face of, say, a determined national defense effort? Serious question, I guess. I don't really know know how power consumption varies across the last few generations of chips, or among the chips you'd plausibly choose between for a warhead or something. But I'm guessing it's not a huge effect.
I mean just look at modern cpus and gpus. Power consumption gets cut in half between generations. So it can be a big deal but you are right about diminishing returns at some point it’s low enough.
I am of a similar mind. 95%+ of work done today is not limited by compute capacity. The remaining frontiers are already heavily distributed (datacenters for processing weather, spying on citizens, early universe simulations, AI model du jour, etc) so that you can buy a few more racks if you really need to get the process done sooner.
I already enable ECO mode (AMD power limitation) because the chip is fast enough without burning extra watts for a marginal performance improvement (which in my testing was incredibly minor relative to the additional draw).
Ok, fair, "who cares" was too broad. But "older chips would cut into our profit margin" is not the same as "strategically important". Would any of those large businesses go bankrupt if they had to use slightly older chips, or would they grumble and adapt (especially if their competitors were in the same situation)?
> Would any of those large businesses go bankrupt if they had to use slightly older chips, or would they grumble and adapt (especially if their competitors were in the same situation)?
It all depends on the business. If you compare AMD EPYC growth in 5 years, there are huge savings, maybe >200% (in extra cores, less power and faster cores). Is that profit margin not significant? I'm sure a lot of businesses relying on a fleet of servers would struggle.
If you're selling phones for example and Samsung/Apple have the latest chips whilst your local company doesn't it will soon matter. Today it might be close, but in 5 years? There's the you don't "need" it as a consumer but there's also a why would a consumer buy a vastly inferior product?
So who are you competing against and hence how do you adapt? You assume the competitors are in the same situation, but are they? A lot of companies compete globally.
If we focus on compute, the cores in the A14/M1 are excellent still, even in terms of efficiency. Mine holds up excellently. The iPhone 12’s 4GB RAM (6 on Pro) is its bottleneck.
The reason nearly no one wants the iPhone 12 is due to marketing, preferring to buy from Apple, and the 14’s new design, plus 13’s bigger batteries and 13 Pro’s 120Hz.
I sold a refurbished 12 Pro to a friend for very cheap and it should do well for a long time.
I don't understand why there are such extreme export controls wrt China. Isn't a more technologically capable China better for everyone because it increases the supply of high-quality parts? Is it just revenge for Chinese nationals stealing IP ?
Wouldn't sanctions just make them more antagonistic? If you integrate China into the world economy more they have more to lose when they invade. Imposing sanctions necessitates they become more self-reliant, making them more resistant to further sanctions if they were to invade.
This is too risky because China is growing too fast. In contrast, proposing some proposals that can be implemented quickly and take effect during the four-year term is more likely to be passed and reflect political achievements.
China was integrated into the world economy (and still is, albeit not to the same degree as it once was). It did nothing to curb the CCP's imperialist ambitions. That navy and the bases they're building in territories they hold no legal claim to could've only been built with money from foreign trade.
It's blatantly about international power. Technology is seen as a bottle neck to limit China's power on several fronts. Economic power, military power, scientific power, technological power.
TSMC is putting a lot of their R&D into next generation technologies like EUV, backside power, advanced packaging/chiplet stuff, glass substrates etc
I’m not sure SMIC will be able to keep up when these efforts bear fruit. Eventually they’ll need EUV to unlock further progress.
SMIC has been able to poach talent by throwing mountains of money at them. This might not be possible going forward. Not being able to work on EUV could be career suicide, and I’m not so sure China can be so frivolous with their money anymore. They have to start raising taxes to compensate for the loss of land lease revenue.