Chinese semiconductor industry

ougoah · Oct 2, 2021

It's very impressive China is already able to do, completely on its own using its own materials and tools, down to 28nm and having verified processes for 14nm and 7nm. It's only barely been 2 years in to tech war and it's closed more than half the previous existing gap.

If we look at the trendline rather than snapshot, this is exceptional given the circumstances. But then again East Asians do rule this field and innovate so much of it whether it is East Asians in Taiwan or East Asian living and working in the US. It's just a question of whether or not a state backed effort has the efficiency and organisational ability to allow Chinese people to cover the ground effectively enough despite every attempt by the US to thwart them. And I mean every attempt. They cannot escalate these bans any further since they've maxed out what they can play and already devouring their own vassals with much disdain from South Korea and Taiwan.

There is still much work to be done to get to leading edge fab and comprehensive tech self reliance in fab. It's achievable and a thing of pride since the club literally consists of four countries right now and three of those are in East Asia (I'm not counting Taiwan since it's not recognised by the UN lol soz but Taiwan is "China" like it or not... even Taiwanese consider themselves true Chinese lol... dunno what they think of the rest of the Chinese who remained on the mainland with "coHmUNissSem").

ougoah · Oct 2, 2021

If PRC's own foundries are producing 28nm and supplying over 40% of China's own market demand for semiconductors (of all sorts) as opposed to a much lower figure in the past, that is mighty impressive. Comparing it to 10 years ago or even 3 years ago.

Now that doesn't indicate much on how far we're going with the journey to 5nm, silicon based, but this is ALL Chinese. The US can force global companies to remove all their tooling and materials (if there are even any left or to begin with) and this figure is still a base minimum of how much Chinese foundries can supply China with. The rest either requiring or desiring leading chips from Taiwan and not banned from buying. Contrast with the only Indian foundry where removal of foreign equipment means they can produce 0 milligram of chips.

Quite an elite club if you ask me and one where Indian government is offering foreign companies billions of funding to just set up foreign production of foreign chips, using foreign skills and equipment, none of which will go to India or be allowed access by India's own "fabs" of which they really have none that are using any of their own equipment. If ever sanctioned, they could not even produce 250nm.

I think TSMC was looking at that option and have yet to make a choice.

Well just like how Chinese solar cell companies produce some components in South East Asian nations, it does mean core tech is given or mastered by producing nations. There is much more work to that than what many seem to think where once manufacturing arrive, you immediately mastered the core tech. Even if TSMC produces in India, Indian fabs are still at zero domestic fab tech until they master the physics and produce all the core equipment and master the processes themselves.

sinophilia · Oct 2, 2021

China is already competitive in design with HiSilicon even if it doesn't meet all criteria just yet. You're going to see more and more companies in that class and quickly, and they are going to improve even more despite already being competitive on the world stage. Already too late for the US to do anything about that. I mean HiSilicon's revenue before COVID was the same as all of AMD.

Same with OSAT, already competitive worldwide not just domestically with a Chinese company (JCET). Too late for the US to do anything about this either. Will be easy for China to dominate this sector going forward.

As for SMIC (fabrication) it's not even a decade behind anymore. It's not neck and neck (US companies and ASML won't provide equipment which obviously limits <7nm volume production but that's not guaranteed to be a bottleneck when they're ramping up apparently for that in 2 years or less with indigenous equipment), but not a decade behind either. 14nm was volume manufactured in 2013 by leading-edge foundries so we're talking <8 years behind.

And in EDA China has Empyrean which is similarly less than a decade behind the leaders (Synopsys & Cadence). Hard for the US to do much more than it already has done yet Empyrean is still closing the gap quickly. Hard to measure in this field but I'd also say less than a decade behind here.

The only real serious gap here is in the Semicaps (equipment). The US dominates it (deposition, removal, cleaning, doping, etc.), with large single-segment monopolies by Japan (resist processing) and Netherlands (lithography). This is the area where China is struggling the most and I'd love to hear from people who know more about it. I've heard of the EUV problem potentially being solved as early as 2024, not sure if this is inside knowledge or can be backed up with publicly sourced info, but the other aspects of equipment are rarely talked about so would love to see more about that too.

ougoah · Oct 2, 2021

sinophilia said:
China is already competitive in design with HiSilicon even if it doesn't meet all criteria just yet. You're going to see more and more companies in that class and quickly, and they are going to improve even more despite already being competitive on the world stage. Already too late for the US to do anything about that. I mean HiSilicon's revenue before COVID was the same as all of AMD.

Same with OSAT, already competitive worldwide not just domestically with a Chinese company (JCET). Too late for the US to do anything about this either. Will be easy for China to dominate this sector going forward.

As for SMIC (fabrication) it's not even a decade behind anymore. It's not neck and neck (US companies and ASML won't provide equipment which obviously limits <7nm volume production but that's not guaranteed to be a bottleneck when they're ramping up apparently for that in 2 years or less with indigenous equipment), but not a decade behind either. 14nm was volume manufactured in 2013 by leading-edge foundries so we're talking <8 years behind.

And in EDA China has Empyrean which is similarly less than a decade behind the leaders (Synopsys & Cadence). Hard for the US to do much more than it already has done yet Empyrean is still closing the gap quickly. Hard to measure in this field but I'd also say less than a decade behind here.

The only real serious gap here is in the Semicaps (equipment). The US dominates it (deposition, removal, cleaning, doping, etc.), with large single-segment monopolies by Japan (resist processing) and Netherlands (lithography). This is the area where China is struggling the most and I'd love to hear from people who know more about it. I've heard of the EUV problem potentially being solved as early as 2024, not sure if this is inside knowledge or can be backed up with publicly sourced info, but the other aspects of equipment are rarely talked about so would love to see more about that too.

I'm more keen for the next. China shouldn't bother honestly that much with 5nm and so on since there isn't much point beyond 5nm. The improvement to performance is minimal and unnecessary for 90% of applications. Even if a producer in smartphones must have competing specs, they can simply design with more room or with stacking? It's important for Chinese domestic foundries to get to 14nm and even 7nm but beyond that, stacking, wafer bonding etc are all good enough to make competitive specs and these are only for the products that absolutely must have competing top end specs. For 90% of other uses even 28nm is MORE than good enough. India uses 200nm+ for instance where they don't need top performance and the Indian foundry achieved that using 1990s foreign tooling and they also do not have any wafer bonding or stacking technologies unlike China which innovated and pioneered them due to necessity.

Chinese research should be invested in finding alternative principles and even though lab scale experiments and alternatives may have shown some promise, there's a big time gap between verified and mass produce-able. This is where the main effort should really focus on rather than catching up to Netherlands + German + Taiwan + South Korea + Japan + USA in comprehensive supply chain reaching parity down to 2nm or whatever is possible. It's going to cost a lot of man hours and resources to catch up to the unified supply chain of multiple tech competent nations and their well established niche fields within fab.

Weaasel · Oct 2, 2021

ansy1968 said:
@BlackWindMnt I know bro but is there a middle ground? From the profit it had gain and the knowledge it hold, it can do its owned research and replaced those equipment? My gut feeling is Samsung is trying to go that route and may take at least 5 years, with some investment as part of their $100 billion dollars program. The time of pushing the envelope of Moore"s law smaller node maybe delayed as the cost multiply and escalated therefore the right money is to invest and owned the critical equipment. After reading and browsing the net I learned a lot of equipment used are manufactured by the Taiwanese themselves, only the EUVL and the materials used are foreign, so they had a great start to begin with.

Lacking EUV lithography machines and photoresists or/and other chemicals is a BIG DEAL if it cannot source them from a source that is inert to US sanctions and dictates. That is why I posted earlier that TSMC is hoping - and I believe that they are - that China will be able to produce EUVL or other lithography machines comparable in performance to the ones of ASML, with entirely Chinese made components. I will even add the photoresists and other chemicals to that. Japan currently dominates the market share for the production of photoresists used in UVL and especially EUVL, but if the tells Japan not to sell those products to any foreign entity - bar the US of course - Japan would comply... Unless an entity such as China, which can and will essentially do as it pleases with regards to the sales of tech that it posseses the wherewithal to produce - begins to produce photoresists and other chemicals comparable in quality to those of Japan. In fact, China having tech capabilities in semiconductor and IC chip manufacturing equipment and chemicals substances synthesis very comparable to Japan, frees Japan from the yoke of US dictats.

ougoah · Oct 2, 2021

Weaasel said:
Lacking EUV lithography machines and photoresists or/and other chemicals is a BIG DEAL if it cannot source them from a source that is inert to US sanctions and dictates. That is why I posted earlier that TSMC is hoping - and I believe that they are - that China will be able to produce EUVL or other lithography machines comparable in performance to the ones of ASML, with entirely Chinese made components. I will even add the photoresists and other chemicals to that. Japan currently dominates the market share for the production of photoresists used in UVL and especially EUVL, but if the tells Japan not to sell those products to any foreign entity - bar the US of course - Japan would comply... Unless an entity such as China, which can and will essentially do as it pleases with regards to the sales of tech that it posseses the wherewithal to produce - begins to produce photoresists and other chemicals comparable in quality to those of Japan. In fact, China having tech capabilities in semiconductor and IC chip manufacturing equipment and chemicals substances synthesis very comparable to Japan, frees Japan from the yoke of US dictats.

In these aspects (similar others) Japanese, Taiwanese, and Korean execs are actually hoping China develops comparable alternatives.

They know they'd be forced by the US if not, and if China has it, they at least maintain status quo and even within China they'd still have some share as opposed to zero (US ban order). This gives them time for their own abilities to develop further which is their hope. But the alternative of surrendering everything to the US and then making much less money is what they'd rather avoid. It's between a rock and a hard place. Long term, they consider the Chinese ones a threat. BUT when that happens, they at least maintain their markets outside China with orders also from Chinese sources since there is no way Chinese gov would force all Chinese buyers to only go for the Chinese brands. We've never seen that. At most a push to get numbers up and that's it. They'd still have a non zero market. If China fails in producing alternatives there, outright ban with zero market from China due to US action and a smaller global market along with the tech being raked in by the US. Rock and a hard place.

ansy1968 · Oct 2, 2021

Weaasel said:
Lacking EUV lithography machines and photoresists or/and other chemicals is a BIG DEAL if it cannot source them from a source that is inert to US sanctions and dictates. That is why I posted earlier that TSMC is hoping - and I believe that they are - that China will be able to produce EUVL or other lithography machines comparable in performance to the ones of ASML, with entirely Chinese made components. I will even add the photoresists and other chemicals to that. Japan currently dominates the market share for the production of photoresists used in UVL and especially EUVL, but if the tells Japan not to sell those products to any foreign entity - bar the US of course - Japan would comply... Unless an entity such as China, which can and will essentially do as it pleases with regards to the sales of tech that it posseses the wherewithal to produce - begins to produce photoresists and other chemicals comparable in quality to those of Japan. In fact, China having tech capabilities in semiconductor and IC chip manufacturing equipment and chemicals substances synthesis very comparable to Japan, frees Japan from the yoke of US dictats.

@Weaasel I know it's asking to much and in the real world seems impossible, but TSMC had a role in the development of ASML NXE3400i EUVL then with that knowledge it can help SMEE or CETC develop its EUVL? the Chinese had the wherewithal but lack the expertise and experiences and here TSMC can help. Everybody assume the American will prevail , well with the Mafia is after you then that is a concern. But both of them(TSMC 56% Samsung 17%) hold 73% of the logic chip market and 100% of SOC market 7nm and below, they can fight back especially if American company is being affected. And there is China they can leveraged that, I'm really intrigue on what @FairAndUnbiased had posted regarding the Prague Spring, this may be the infection point of American decline especially as the Chinese DUVL had its debut with optimism of an EUVL in 2 years time.

And a side note, it will delayed the development of 3nm , 2nm or even the 1nm as the motivation had been stifle, enabling the Chinese to caught up, what a major change of event from 2 years ago, The American really want the Chinese to succeed in the Semiconductor field...LOL

BlackWindMnt · Oct 2, 2021

sinophilia said:
China is already competitive in design with HiSilicon even if it doesn't meet all criteria just yet. You're going to see more and more companies in that class and quickly, and they are going to improve even more despite already being competitive on the world stage. Already too late for the US to do anything about that. I mean HiSilicon's revenue before COVID was the same as all of AMD.

Same with OSAT, already competitive worldwide not just domestically with a Chinese company (JCET). Too late for the US to do anything about this either. Will be easy for China to dominate this sector going forward.

As for SMIC (fabrication) it's not even a decade behind anymore. It's not neck and neck (US companies and ASML won't provide equipment which obviously limits <7nm volume production but that's not guaranteed to be a bottleneck when they're ramping up apparently for that in 2 years or less with indigenous equipment), but not a decade behind either. 14nm was volume manufactured in 2013 by leading-edge foundries so we're talking <8 years behind.

And in EDA China has Empyrean which is similarly less than a decade behind the leaders (Synopsys & Cadence). Hard for the US to do much more than it already has done yet Empyrean is still closing the gap quickly. Hard to measure in this field but I'd also say less than a decade behind here.

The only real serious gap here is in the Semicaps (equipment). The US dominates it (deposition, removal, cleaning, doping, etc.), with large single-segment monopolies by Japan (resist processing) and Netherlands (lithography). This is the area where China is struggling the most and I'd love to hear from people who know more about it. I've heard of the EUV problem potentially being solved as early as 2024, not sure if this is inside knowledge or can be backed up with publicly sourced info, but the other aspects of equipment are rarely talked about so would love to see more about that too.

There was a research paper released last year December about a Chinese and German university having designed some high power laser that could be used for lithography on EUV level.

FairAndUnbiased · Oct 2, 2021

ansy1968 said:
And there is China they can leveraged that, I'm really intrigue on what @FairAndUnbiased had posted regarding the Prague Spring, this may be the infection point of American decline especially as the Chinese DUVL had its debut with optimism of an EUVL in 2 years time.

The key thing here is that TSMC and Samsung have high risk right now. They really cannot give up. If they give up their sales and customer data, here is what can potentially happen:

1. They can get dropped as fabs by customers in favor of Intel Foundry or GloFo, especially TSMC which operates many non-leading edge fabs for automotive, industrial and scientific customers which are competing directly with Intel Foundry and GloFo.

2. They can get demands for compensation or lawsuit threats for revealing trade secrets since revealing who your foundry is, can actually give competitors huge amounts of information and cause material losses.

3. If that data is shared with Intel and GloFo (which it surely 100% won't be, since this has never happened before and would be highly unethical

) then in combination with revenue and process node data, they can pull customers away from TSMC and Samsung especially for their non-leading edge fabs. Long term, they know which process nodes to target, saving tons of time.

So in general, if they submit now it will be just like breaking Toshiba's legs in the 1980's.

Overbom · Oct 2, 2021

I dont know why some people see TSMC as an "ally" or friend. TSMC has repeatedly taken action in order to handicap small Chinese Fab companies from growing and eventually competing with it

I dont care about TSMC or Samsung. As long as domestic EUV tool is made, then domestic Fabs should be set up and the foreign players should driven out and left with only about 10% of the domestic market in order to have competition

Foreign Fab companies have repeatedly shown that they are not trustworthy. TSMC is controlled by the US puppet Taiwan current Gov, while Samsung is controlled by S.Korea which is also a US puppet having US troops on its soil

Would you trust US puppets producing chips for your country? I don't.

Chinese semiconductor industry

ougoah

Brigadier

ougoah

Brigadier

sinophilia

Junior Member

ougoah

Brigadier

Weaasel

Senior Member

ougoah

Brigadier

ansy1968

Brigadier

BlackWindMnt

Captain

FairAndUnbiased

Brigadier

Overbom

Brigadier