Chinese semiconductor thread II

[OptimusLion]

The sanctions from the US are coming again! More than 80 Chinese chip electronics entities are included in the list, and the total number has exceeded 1,300.

This is really nothing new. Self-reliance in science and technology is the only way forward!

 

[siegecrossbow]

They have truly learned nothing from the past few years.

Please, Log in or Register to view URLs content!

They were supposed to keep China below 14 nm I thought and now they are moving the goal post? Do they want a participation medal to go alone with this to make them feel better?

BTW look at what the mediocre phone is doing to Apple…

Please, Log in or Register to view URLs content!

 

[Wrought]

The lag effects are very long and so it’s hard to observe outcomes until it’s too late for policymaking.

The argument that the export controls don’t work because they solve a coordination problem within China’s private sector is fundamentally one about time preferences with the classic response that the controls lock China out of downstream markets that utilize semiconductors, and denying China the benefits from trade.

I don’t think there’s per se a “right” answer as much as you have fundamentally different orientations and that results in substantially different policymaking choices.

Sometimes I think other people give you too hard of a time, but this is some real bullshit right here. The export controls don't work because they are failing to achieve their stated objectives. Evidence of Chinese progress is all over this thread, while US efforts to onshore a domestic semiconductor industry have struggled to find their footing, much less achieve new breakthroughs. Being charitable, one could say they are at least putting time and money in the right direction. But this is quite simply wrong:

US efforts to maintain a technological edge on China are working — no matter what Beijing says.

Now it's one thing to claim, as you claimed, that China is paying a steep price for its progress. That much is undeniably true. But it's an entirely different thing to claim that it's making no progress at all—that Huawei and SMIC are engaged in propaganda efforts as opposed to genuine technical advancement. If these US officials were talking about how well they are using the limited time that export controls have bought them to seek new breakthroughs and keep themselves ahead, that would be questionably true, but at least they would be talking about reality. Instead of this fantasyland where export controls represent a magical hard stop that China is somehow stuck at forever.

The "right" answer is to not try to achieve something which is impossible, and then pat yourself on the back in defiance of all reality.

 

[chgough34]

Sometimes I think other people give you too hard of a time, but this is some real bullshit right here. The export controls don't work because they are failing to achieve their stated objectives. Evidence of Chinese progress is all over this thread, while US efforts to onshore a domestic semiconductor industry have struggled to find their footing, much less achieve new breakthroughs. This is quite simply wrong:

The Sullivan speech referenced making the technological gap “as large as possible” in AI, semiconductors, and clean energy. Semiconductors are an input to “AI” and it’s pretty clear they view “semiconductor” process as being about node counting. If you take some of the reports about the Nvidia export controls harming the development of ML in China, then there is an arguable space that it’s hurting China and thus “working”, and it’s clearer more so in the euvl space since SMIC will likely have its euvl roll-out delayed by a decade due to US actions; not to mention scaling time, so there it’s also “working”.

Now it's one thing to claim, as you claimed, that China is paying a steep price for its progress. That much is undeniably true. But it's an entirely different thing to claim that it's making no progress at all—that Huawei and SMIC are engaged in propaganda as opposed to genuine technical efforts.

Opportunity costs are not “free” - you open up multiple questions here - what would China/where would China be absent the controls in downstream sectors? Would Chinese firms in downstream sectors have larger scale and network effects? Etc. these are gains from trade that are not easily (or perhaps not at all) quantifiable but they do fall into the remit of keeping a technological gap “as large as possible”.

if absent the controls, China would have closed the gap more than China currently has (however defined), then the controls have worked in keeping the gap “as large as possible”.

 

[ansy1968]

Sometimes I think other people give you too hard of a time, but this is some real bullshit right here. The export controls don't work because they are failing to achieve their stated objectives. Evidence of Chinese progress is all over this thread, while US efforts to onshore a domestic semiconductor industry have struggled to find their footing, much less achieve new breakthroughs. Being charitable, one could say they are at least putting time and money in the right direction. But this is quite simply wrong:



Now it's one thing to claim, as you claimed, that China is paying a steep price for its progress. That much is undeniably true. But it's an entirely different thing to claim that it's making no progress at all—that Huawei and SMIC are engaged in propaganda efforts as opposed to genuine technical advancement. If these US officials were talking about how well they are using the limited time that export controls have bought them to seek new breakthroughs and keep themselves ahead, that would be questionably true, but at least they would be talking about reality. Instead of this fantasyland where export controls represent a hard stop that China is somehow stuck at forever.

The "right" answer is to not try to achieve something which is impossible, and then pat yourself on the back in defiance of all reality.

He is planning to join the US mental gymnastic squad, so he is practicing his routine move here.

 

[Wrought]

The Sullivan speech referenced making the technological gap “as large as possible” in AI, semiconductors, and clean energy. Semiconductors are an input to “AI” and it’s pretty clear they view “semiconductor” process as being about node counting. If you take some of the reports about the Nvidia export controls harming the development of ML in China, then there is an arguable space that it’s hurting China and thus “working”, and it’s clearer more so in the euvl space since SMIC will likely have its euvl roll-out delayed by a decade due to US actions; not to mention scaling time, so there it’s also “working”.

Opportunity costs are not “free” - you open up multiple questions here - what would China/where would China be absent the controls in downstream sectors? Would Chinese firms in downstream sectors have larger scale and network effects? Etc. these are gains from trade that are not easily (or perhaps not at all) quantifiable but they do fall into the remit of keeping a technological gap “as large as possible”.

if absent the controls, China would have closed the gap more than China currently has (however defined), then the controls have worked in keeping the gap “as large as possible”.

The points you are raising are not without merit, in a vaccuum, but they also aren't the ones voiced by Sullivan and co. in the article I cited. You are steelmanning their argument. They claim Chinese achivements are merely propaganda, not real technical progress. Which is complete and utter bullshit.

If US officials made your arguments, they would have a better case. They didn't, and so they don't. And you making these arguments in their stead doesn't speak well for your objectivity either.

 

[manqiangrexue]

The Sullivan speech referenced making the technological gap “as large as possible” in AI, semiconductors, and clean energy.

Ironically, that is true because China is pulling ahead and making that gap larger by the day. The only way one could say that the US has any advantage in any of those fields is by counting all of the tech in the world outside of China as American and even then, semiconductors would be the only area, due completely to the Netherlands.

Semiconductors are an input to “AI” and it’s pretty clear they view “semiconductor” process as being about node counting. If you take some of the reports about the Nvidia export controls harming the development of ML in China, then there is an arguable space that it’s hurting China and thus “working”, and it’s clearer more so in the euvl space since SMIC will likely have its euvl roll-out delayed by a decade due to US actions; not to mention scaling time, so there it’s also “working”.

The only way that they can be seen as harming China is if you believed that forcing a student to do his own work to truly begin his path to development rather than lazily copying others is harming him.

Opportunity costs are not “free” - you open up multiple questions here - what would China/where would China be absent the controls in downstream sectors? Would Chinese firms in downstream sectors have larger scale and network effects? Etc. these are gains from trade that are not easily (or perhaps not at all) quantifiable but they do fall into the remit of keeping a technological gap “as large as possible”.

if absent the controls, China would have closed the gap more than China currently has (however defined), then the controls have worked in keeping the gap “as large as possible”.

Absent the controls, China would be reliant on foreign foundational technology and have a large weakness open should sudden war/hostilities break out. In that situation, China might seem further ahead but that would only be an empty lead which would be vulnerable to a quick chopping down should all the controls hit at once in a critical situation. Fortunately, the US is incapable of doing anything quick so they ended up trickling in the sanctions as China progressed, essentially turning the controls into a training course to show China the value of getting it done by yourself and China is probably the only country in the world with the inner power to graduate this traning course and end up indigenous at a world-leading level.

The only way the US controls would have "worked" is if the politicians only thought in 4 year cycles and had no qualms trading a slightly weaker China now for a much more powerful China in the future when their careers are already over. I wouldn't put such stupidity past them.

 

[chgough34]

The points you are raising are not without merit, in a vaccuum, but they also aren't the ones voiced by Sullivan and co. in the article I cited. You are steelmanning their argument.

I’m not - it’s an extension of this argument he made 2 years back (

Please, Log in or Register to view URLs content!

). The BBG article doesn’t give good context

They claim Chinese achivements are merely propaganda, not real technical progress. Which is complete and utter bullshit.

Yeah I agree lol.

If US officials made your arguments, they would have a better case. They didn't, and so they don't. And you making these arguments in their stead doesn't speak well for your objectivity either.

Hence my point on whether the export controls are “working” depends on your orientation
- Are they “working” to improve US-China relations and thicken commercial relationships? Absolutely not
- Are they “working” to freeze China’s semiconductor industry? Absolutely not
- Are they “working” to keep the U.S. technological lead “as large as possible”? Maybe

 

[Pkp88]

That is going to happen sooner than later, there are many other litho companies entering the space, specially in the packaging and power market, that will put pressure in SMEE to focus more in front end lithography machines and that is really bad news for ASML because the Chinese government could push for the use of SMEE machines.

there is also the probability of NAURA and ACM expanding the offering with dry DUV litho machines, given that China is building the most comprehensive lithography subsystem ecosystem in the planet, unless these suppliers are exclusive to SMEE could allow other players to enter market.

That what you can see, the internet only allows to scan the surface but things in China go deeper, I am pretty sure that SMEE has advance immersion machines already running and patterning wafers, their recent announce give an insight, particularly the wording of Holistic metrology that means that this tool is mean to work in a close loop with the immersion scanner, the coating developing tool, AEI and ADI metrology tools, DJEL EBI tools.
View attachment 134911
View attachment 134912

Why they haven't announced yet? I think SMEE is facing pressure from SMIC, YMTC, ASML and others to keep their tools quiet in order to avoid more bans and export controls.

Would be good to discuss this further. The range of 55-7nm feels like a quite telling point of capabilities

 

[pbd456]

The sanctions from the US are coming again! More than 80 Chinese chip electronics entities are included in the list, and the total number has exceeded 1,300.

This is really nothing new. Self-reliance in science and technology is the only way forward!


View attachment 134992

These sanctions come right after Sullivan wang meeting. So it seems that Sullivan doesn't get any concessions