Discussion Leading Edge Foundry Node advances (TSMC, Samsung Foundry, Intel)

Page 15 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Saylick

Golden Member
Sep 10, 2012
1,643
2,074
136
They don't really have any leverage to force any of the above companies to cooperate with this. Sticks won't work, so they'll need one hell of a big carrot. And they know going in China's goal is to steal their IP, so why would they believe any lofty talk about "collaboration"?
The leverage would be "If you don't work with us, we'll make it difficult for you to make money in the Chinese market". That strategy has worked in the past, where companies had to partner with a domestic Chinese entity in order to even do business in China.

We'll see how this plays out for now. So far, the language does not suggest China will strong-arm companies to partake, only to provide financial backing for collaboration. If no one bites at the carrot, perhaps we'll see China reveal their true hand.
The committee will encourage collaboration between foreign and Chinese companies and research institutions. It will also invite foreign companies to set up development or manufacturing bases by working with local governments and providing funds. It also envisions offering financial backing to Chinese companies seeking to acquire overseas semiconductor-related companies.
 
  • Like
Reactions: Joe NYC and Tlh97

Ajay

Lifer
Jan 8, 2001
11,401
5,160
136
They don't really have any leverage to force any of the above companies to cooperate with this. Sticks won't work, so they'll need one hell of a big carrot. And they know going in China's goal is to steal their IP, so why would they believe any lofty talk about "collaboration"?
They already have all the IP - they’ve stolen it. Of course, that does nothing for them outside of the Chinese market because of international laws. What they really need is access to external talent to give them the practical skills of running a process development program to completion (HVM) and access to the high tech equipment and legal access to IP for external markets. These Joint Ventures (where China insisted on 51% ownership by the Chinese partner) are antithetical to the broader interests of international commerce.

Western governments have and should continue implementing sanctions and laws to protect their own semiconductor manufacturers from these predatory practices and ensure a secure and stable marketplace. We have to do this to protect our own companies capitalist behaviors from dooming themselves by succumbing to the temptation of short term gains offered by the Chinese.
 

maddie

Diamond Member
Jul 18, 2010
3,861
2,943
136
They already have all the IP - they’ve stolen it. Of course, that does nothing for them outside of the Chinese market because of international laws. What they really need is access to external talent to give them the practical skills of running a process development program to completion (HVM) and access to the high tech equipment and legal access to IP for external markets. These Joint Ventures (where China insisted on 51% ownership by the Chinese partner) are antithetical to the broader interests of international commerce.

Western governments have and should continue implementing sanctions and laws to protect their own semiconductor manufacturers from these predatory practices and ensure a secure and stable marketplace. We have to do this to protect our own companies capitalist behaviors from dooming themselves by succumbing to the temptation of short term gains offered by the Chinese.
These are questions for us to answer.

1) Do we believe that the Chinese or any other large technological economy cannot create unique IP?
2) Is any human population naturally inferior in IQ so as to prevent their technological advancement?
3) Are their any benefits to larger populations in approaching many problems?

Assuming that a Gaussian distribution as applied to IQ is correct, can't we also assume that a population 4X can produce 4X the number of IQ (x) to work on solving hard problems?

The Chinese ARE using these methods to advance their development more rapidly and cheaply, BUT all the thinking in trying to isolate anyone will eventually backfire. What happens if, and most probably when, they are ahead? Do we then demand that they share their superior tech? How is this supposed to work?

Gut reactions can end in the opposite result intended. This is a difficult issue to solve.

Some might see this as defending the Chinese, although I don't believe I am, but for sure, we need to think differently.

I'm editing this this to add what might be a very disturbing link. Apologies to those offended.:


Quote:
"Long before the United States began accusing other countries of stealing ideas, the U.S. government encouraged intellectual piracy to catch up with England’s technological advances. According to historian Doron Ben-Atar, in his book, Trade Secrets, “the United States emerged as the world's industrial leader by illicitly appropriating mechanical and scientific innovations from Europe.”

Among those sniffing out innovations across the Atlantic was Harvard graduate and Boston merchant, Francis Cabot Lowell. As the War of 1812 raged on, Lowell set sail from Great Britain in possession of the enemy’s most precious commercial secret. He carried with him pirated plans for Edmund Cartwright’s power loom, which had made Great Britain the world’s leading industrial power.

Halfway across the Atlantic, a British frigate intercepted Lowell’s ship. Although the British double-searched his luggage and detained him for days, Lowell knew they would never find any evidence of espionage for he had hidden the plans in the one place they would never find them—inside his photographic mind. Unable to find any sign of spy craft, the British allowed Lowell to return to Boston, where he used Cartwright’s design to help propel the Industrial Revolution in the United States."


This is what countries do.
 
Last edited:

LightningZ71

Golden Member
Mar 10, 2017
1,266
1,284
136
Of course it's what countries do. Countries also work diligently to protect their own IP from being stolen. I expect China to behave no differently than they have in the past when they have superiority in any given area. To give the example of their behavior in the South China Sea, where they have unquestioned military dominance, and their construction of artificial island fortresses and well documented bullying of neighboring countries away from what the UN has recognized as international waters to be used as fisheries, it is but one instance of their expected behavior in the semiconductor industry once they have a technical lead, or at least parity. It is widely assumed that they will flood their home market with cheaper products than international ones, and, distribute those products overseas to areas that won't tariff them, destroying the market for the legitimate holders of the IP in question.

To go any further than this would be to make this thread way more political than it already has become.

I have no doubts that the people of China are smart enough to go through the technical development cycle to get where TSMC is today. It's just going to take time, iteration, and a lot of resources. They can shorten that by just brokering deals that invite foreign companies to sell their future competitiveness in the name of short term profits. Granted, I firmly believe that China will eventually gain technical leadership in chip manufacturing one day soon. They have the will, the resources, the manpower, and a highly educated workforce (often educated overseas) to see these projects through to the end. The question here is timetable. How long will it take them? They can produce, in volume, the equivalent of the ".22" process node domestically, and are, to my knowledge, in initial low volume with ".14/.12" nodes as well. It's the .7 equivalent nodes where they are apparently a good bit behind for commercial volume.

It's a matter of time...
 

moinmoin

Diamond Member
Jun 1, 2017
3,442
4,821
136
I honestly think this is nothing worth to worry about.

The area China wants to enter is leading edge foundry business. First issue: Unless CPP is content with older gens the only companies of interest that could help there are TSMC, Samsung and maybe Intel.
  • TSMC is in Taiwan, CCP publicly pushes another serious agenda there, fat chance of any serious collaboration to happen.
  • Samsung is in South Korea, also close by in the region, plenty unlikely.
  • Intel is in the US, and Intel tries to get plenty gov subsidies to make the US catch up with the former two, yeah let's collaborate with China, not.
Second issue: The cost in this business is enormous. China already randomly invested billions to get its foundries moving, but had to realize that that amount of money is still too little and the knowledge needed is much more than they could get.

Third issue: Massive shortage of all resources involved. As the industry moved toward ever higher specialization the amount of companies that can actually produce the required components decreases. And each of them is a potential bottleneck in how much can be produced, or even be told to no longer work with China at some future point.
  • Foundries: TSMC, Samsung, Intel.
  • Foundry hardware: ASML.
  • Lithography lenses/mirrors: Zeiss.
  • etc.
China is learning the hard way that entering existing highly specialized leading edge markets is very hard and expensive.

In my opinion China's best chance of building something of its own is researching and developing new approaches and compete with them against the existing market. But that's the long way, the ridiculous mismanagement China allowed to happen with all these shuttered local foundry efforts shows China so far didn't know how big any of the efforts involved in this undertaking actually are.

 

maddie

Diamond Member
Jul 18, 2010
3,861
2,943
136
Of course it's what countries do. Countries also work diligently to protect their own IP from being stolen. I expect China to behave no differently than they have in the past when they have superiority in any given area. To give the example of their behavior in the South China Sea, where they have unquestioned military dominance, and their construction of artificial island fortresses and well documented bullying of neighboring countries away from what the UN has recognized as international waters to be used as fisheries, it is but one instance of their expected behavior in the semiconductor industry once they have a technical lead, or at least parity. It is widely assumed that they will flood their home market with cheaper products than international ones, and, distribute those products overseas to areas that won't tariff them, destroying the market for the legitimate holders of the IP in question.

To go any further than this would be to make this thread way more political than it already has become.

I have no doubts that the people of China are smart enough to go through the technical development cycle to get where TSMC is today. It's just going to take time, iteration, and a lot of resources. They can shorten that by just brokering deals that invite foreign companies to sell their future competitiveness in the name of short term profits. Granted, I firmly believe that China will eventually gain technical leadership in chip manufacturing one day soon. They have the will, the resources, the manpower, and a highly educated workforce (often educated overseas) to see these projects through to the end. The question here is timetable. How long will it take them? They can produce, in volume, the equivalent of the ".22" process node domestically, and are, to my knowledge, in initial low volume with ".14/.12" nodes as well. It's the .7 equivalent nodes where they are apparently a good bit behind for commercial volume.

It's a matter of time...
This, for sure, is one topic where politics and technology will be closely interacting.

It's going to affect Intel and their grants from the Government.
AMD most probably will be looking at not being so reliant on Taiwan.
How will this affect upcoming products, 4-5 yrs out?
Samsung and GloFlo has a golden opportunity here for growth, apart from the general shortage of silicon capacity.
Will TSMC relocate a large % of their production overseas?
This is going to affect the entire industry as the present escalating tensions have only started their steep rise fairly recently (yrs) when compared to product design cycles.
 
  • Like
Reactions: Tlh97

Exist50

Senior member
Aug 18, 2016
717
668
136
TSMC is in Taiwan, CCP publicly pushes another serious agenda there, fat chance of any serious collaboration to happen.
There is a ton of collaboration between PRC and ROC businesses. The former is usually the latter's biggest customer.
 

maddie

Diamond Member
Jul 18, 2010
3,861
2,943
136
I honestly think this is nothing worth to worry about.

The area China wants to enter is leading edge foundry business. First issue: Unless CPP is content with older gens the only companies of interest that could help there are TSMC, Samsung and maybe Intel.
  • TSMC is in Taiwan, CCP publicly pushes another serious agenda there, fat chance of any serious collaboration to happen.
  • Samsung is in South Korea, also close by in the region, plenty unlikely.
  • Intel is in the US, and Intel tries to get plenty gov subsidies to make the US catch up with the former two, yeah let's collaborate with China, not.
Second issue: The cost in this business is enormous. China already randomly invested billions to get its foundries moving, but had to realize that that amount of money is still too little and the knowledge needed is much more than they could get.

Third issue: Massive shortage of all resources involved. As the industry moved toward ever higher specialization the amount of companies that can actually produce the required components decreases. And each of them is a potential bottleneck in how much can be produced, or even be told to no longer work with China at some future point.
  • Foundries: TSMC, Samsung, Intel.
  • Foundry hardware: ASML.
  • Lithography lenses/mirrors: Zeiss.
  • etc.
China is learning the hard way that entering existing highly specialized leading edge markets is very hard and expensive.

In my opinion China's best chance of building something of its own is researching and developing new approaches and compete with them against the existing market. But that's the long way, the ridiculous mismanagement China allowed to happen with all these shuttered local foundry efforts shows China so far didn't know how big any of the efforts involved in this undertaking actually are.

Second issue: The cost in this business is enormous. China already randomly invested billions to get its foundries moving, but had to realize that that amount of money is still too little and the knowledge needed is much more than they could get.

Do you really think that they have no idea of where to concentrate resources? Hell, quite a few of us on this forum could do better than that.
 

maddie

Diamond Member
Jul 18, 2010
3,861
2,943
136
Read the article I linked.
What I summarized as important.

Let’s first summarize the major elements of the HSMC heist that unfolded between late 2017 and early 2021 (which may read like a movie script, not real life):

According to a report from Caixin (in Chinese), in just the last three years, the number of semiconductor startups in China have increased by more than 50%. Moreover, the amount of investment has increased 6-times, from $946 million USD in 2018 to $6.16 billion USD in 2019. For the first half of 2020, investment has reached $8.46 billion USD.

Undoubtedly, not all of these investors know what they are investing in; most of them don’t. Similarly, not all of these startups will succeed; most of them won’t.

Failures are inevitable for something as hard as semiconductor manufacturing. Learning about why failures happen is an important necessity to a new industry’s growth and maturity.

Based on 36Kr’s calculation, the “HSMC trio” stole in total 12.4 billion RMB (~$2 billion USD) in three years from the three-course meal of district government investment, contractor deposits, and bank loans.


This is an analysis of a single entity. there will be others like this, but not all, and some will succeed. This is generally accepted by most venture capitalists (not the fraud). It's too easy to generalize and portray a narrative that the audience wants to hear. My concern is that it's too easy to fall into a self delusional mindtrap. We will always be the best.
 

moinmoin

Diamond Member
Jun 1, 2017
3,442
4,821
136
What I summarized as important.

Let’s first summarize the major elements of the HSMC heist that unfolded between late 2017 and early 2021 (which may read like a movie script, not real life):

According to a report from Caixin (in Chinese), in just the last three years, the number of semiconductor startups in China have increased by more than 50%. Moreover, the amount of investment has increased 6-times, from $946 million USD in 2018 to $6.16 billion USD in 2019. For the first half of 2020, investment has reached $8.46 billion USD.

Undoubtedly, not all of these investors know what they are investing in; most of them don’t. Similarly, not all of these startups will succeed; most of them won’t.

Failures are inevitable for something as hard as semiconductor manufacturing. Learning about why failures happen is an important necessity to a new industry’s growth and maturity.

Based on 36Kr’s calculation, the “HSMC trio” stole in total 12.4 billion RMB (~$2 billion USD) in three years from the three-course meal of district government investment, contractor deposits, and bank loans.


This is an analysis of a single entity. there will be others like this, but not all, and some will succeed. This is generally accepted by most venture capitalists (not the fraud). It's too easy to generalize and portray a narrative that the audience wants to hear. My concern is that it's too easy to fall into a self delusional mindtrap. We will always be the best.
It's not an analysis of a single entity, it's an analysis of the current internal market in China and how HSMC as an example is affecting it. This is to me the important part of the article:

"the two main chokepoints to China’s semiconductor ambition: access to advanced equipment, hiring technical talent. The greed of the “HSMC trio” has managed to shoot China in the foot in both areas, permanently damaging the prospect of other semiconductor upstarts that may actually be trying to make chips.

Regarding advanced equipment access, by luring Chiang and using his reputation to buy ASML’s DUV equipment (only to loan it off for more cash), ASML may never sell to another Chinese semiconductor startup, with or without sanctions. ASML’s reputation is damaged by this transaction. It doesn’t sell to any company that has the money to buy; it evaluates the buyer’s technical capabilities first to make sure the buyer can make proper use of its product. ASML toured HSMC prior to agreeing to sell its equipment, and lauded HSMC’s team as the best it has seen in Mainland China, mostly on the strength of Chiang’s reputation and the engineers he was able to attract. ASML now must feel like a fool. But the bigger fool here is China. ASML had a record 2020, selling 258 lithography equipment, primarily to TSMC. ASML doesn’t need the Mainland China market; China urgently needs ASML’s products.

Regarding hiring technical talent, a chilling effect will permeate the industry, where experienced engineers from Taiwan will think twice about joining a mainland startup, since even the reputation and judgment of Chiang cannot protect you from scam artists. This situation cuts off the best and most obvious source of talent, given the cultural and language similarities between Mainland China and Taiwan. Poaching talent from South Korea and the United States would be much harder. And training domestic talent will take much (much) longer.
"
 

Doug S

Golden Member
Feb 8, 2020
1,178
1,716
106
The leverage would be "If you don't work with us, we'll make it difficult for you to make money in the Chinese market". That strategy has worked in the past, where companies had to partner with a domestic Chinese entity in order to even do business in China.

How does this relate to ASML, who does not need the Chinese market - certainly not for EUV which they would not be permitted to sell to China anyway. As for Intel and AMD, they already know their Chinese market is limited and will disappear before long, as China is deliberately going full speed ahead without them in multiple directions - they already have their own x86 clone, there's ARM and RISC-V, and longer term are putting a lot of weight behind Loongson.

So again, what could China threaten any of these three with to force them to a cooperate in a way that would allow for potential IP transfer? Also, even if they somehow stole the latest Intel and AMD designs it wouldn't do them any good - there are no fabs in China capable of making them. They could steal exact blueprints for the latest ASML EUV scanners but would still be unable to build them. They are so complex and tolerances are so tight that they'd need access to all of ASML's suppliers to get all the parts ASML doesn't make themselves, and all the maintenance/service data as well.

While China has plenty of capable engineers it will take them many years to make their own EUV scanners - look at how long ASML (and others who gave up) tried to crack EUV before they managed it. They can make a national priority all they want, I'd be shocked if they have the ability to mass produce chips using even a single EUV layer by the end of the decade. I suppose they could benefit if they were able to get their hands on a working one to take apart, and perhaps avoid going down some of the blind alleys ASML probably did on their way.

The west has learned their lesson about letting their premiere companies do business with China in a way that compromises strategic IP, and aren't going to allow it to happen again. If Intel said it was going to build a 2A fab in China, the US would put a stop to it before they broke ground.
 

maddie

Diamond Member
Jul 18, 2010
3,861
2,943
136
It's not an analysis of a single entity, it's an analysis of the current internal market in China and how HSMC as an example is affecting it. This is to me the important part of the article:

"the two main chokepoints to China’s semiconductor ambition: access to advanced equipment, hiring technical talent. The greed of the “HSMC trio” has managed to shoot China in the foot in both areas, permanently damaging the prospect of other semiconductor upstarts that may actually be trying to make chips.

Regarding advanced equipment access, by luring Chiang and using his reputation to buy ASML’s DUV equipment (only to loan it off for more cash), ASML may never sell to another Chinese semiconductor startup, with or without sanctions. ASML’s reputation is damaged by this transaction. It doesn’t sell to any company that has the money to buy; it evaluates the buyer’s technical capabilities first to make sure the buyer can make proper use of its product. ASML toured HSMC prior to agreeing to sell its equipment, and lauded HSMC’s team as the best it has seen in Mainland China, mostly on the strength of Chiang’s reputation and the engineers he was able to attract. ASML now must feel like a fool. But the bigger fool here is China. ASML had a record 2020, selling 258 lithography equipment, primarily to TSMC. ASML doesn’t need the Mainland China market; China urgently needs ASML’s products.

Regarding hiring technical talent, a chilling effect will permeate the industry, where experienced engineers from Taiwan will think twice about joining a mainland startup, since even the reputation and judgment of Chiang cannot protect you from scam artists. This situation cuts off the best and most obvious source of talent, given the cultural and language similarities between Mainland China and Taiwan. Poaching talent from South Korea and the United States would be much harder. And training domestic talent will take much (much) longer.
"
That's his take. Guess we will have to see, but money speaks loudly. I do think that tooling is a chokepoint.
 

maddie

Diamond Member
Jul 18, 2010
3,861
2,943
136
How does this relate to ASML, who does not need the Chinese market - certainly not for EUV which they would not be permitted to sell to China anyway. As for Intel and AMD, they already know their Chinese market is limited and will disappear before long, as China is deliberately going full speed ahead without them in multiple directions - they already have their own x86 clone, there's ARM and RISC-V, and longer term are putting a lot of weight behind Loongson.

So again, what could China threaten any of these three with to force them to a cooperate in a way that would allow for potential IP transfer? Also, even if they somehow stole the latest Intel and AMD designs it wouldn't do them any good - there are no fabs in China capable of making them. They could steal exact blueprints for the latest ASML EUV scanners but would still be unable to build them. They are so complex and tolerances are so tight that they'd need access to all of ASML's suppliers to get all the parts ASML doesn't make themselves, and all the maintenance/service data as well.

While China has plenty of capable engineers it will take them many years to make their own EUV scanners - look at how long ASML (and others who gave up) tried to crack EUV before they managed it. They can make a national priority all they want, I'd be shocked if they have the ability to mass produce chips using even a single EUV layer by the end of the decade. I suppose they could benefit if they were able to get their hands on a working one to take apart, and perhaps avoid going down some of the blind alleys ASML probably did on their way.

The west has learned their lesson about letting their premiere companies do business with China in a way that compromises strategic IP, and aren't going to allow it to happen again. If Intel said it was going to build a 2A fab in China, the US would put a stop to it before they broke ground.
RISC-V is quite the uncertainty here. Not only free but the potential to displace X-86. A nightmare for the established players.

Greater Asia appears to coalescing into a common market. Even India is making nice noises. A 3-4 billion population with high growth rates. Sounds too good to ignore.They might not be able to force Intel or AMD, but don't forget the carrot approach.

EUV IS very complex but never forget the quantum difficulty leap in innovating vs following. It's not going to take as long to replicate, especially with active espionage.
 

Roland00Address

Platinum Member
Dec 17, 2008
2,113
213
106
EUV IS very complex but never forget the quantum difficulty leap in innovating vs following. It's not going to take as long to replicate, especially with active espionage.
Knowing a thing is not just the only reason why EUV is hard to do. It is the supply chain where one EUV machine uses parts from 4,000 different suppliers. You are getting parts from 4,000 different people to make your $140 million dollar machines per unit. And those 4,000 suppliers have their own suppliers as well.

And the machine does not work if you are missing the key part.

Put another way it is not just "knowing" how to do it, but having all those relationships with everyone else, and the everyone else is all over the world. Thus it is not just about money, or technology, it is also a legal knot and to a lesser extent a supply chain logistics problem.
 

Roland00Address

Platinum Member
Dec 17, 2008
2,113
213
106
Tower semi? Anyone know anything about them?
Ian just posted a 15 minute video on YouTube.

They are 1/60th the size of Intel with yearly revenues, but they are new customer relationships for they specialize in the foundry stuff similar to Global Foundries where you pay extra for something that is not a generalized foundry but is like RF or best low power and so on. Aka 70% of the market want a generalized homogeneous fab that is good and cheap, and Tower specializes in the 30% where you have specific needs and do not want the white bread homogeneity.

 
  • Like
Reactions: lightmanek

DrMrLordX

Lifer
Apr 27, 2000
19,351
8,116
136
So more of a competitor to, say, GF FDX nodes.

edit: oh, they do high-voltage stuff? Might be an automotive play. FDX nodes don't necessarily target high voltage ranges.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,171
3,021
136
It's likely a strategic investment where the existing relationship and knowledge that Tower Semiconductor offers in the foundry service is invaluable to Intel in ramping up their own, rather than viewing Intel going into 1um node products en masse.

Although they probably would use some of their fabs.

It also opens up the possibility to move existing non-Intel process products to Intel ones like Networking.
 
  • Like
Reactions: Tlh97

Hitman928

Diamond Member
Apr 15, 2012
4,075
4,808
136
So more of a competitor to, say, GF FDX nodes.
No, those are more advanced even. Tower has 45nm+ nodes with lots of non-bulk CMOS options. They've got some good stuff, just not things this forum would probably be too interested in, generally. I've actually used Tower Semi a few times and have an upcoming project slated to use them again. I'm really hoping Intel doesn't screw this up and make things difficult for their customers when trying to integrate Tower into Intel's 2.0 model.
 

NostaSeronx

Diamond Member
Sep 18, 2011
3,484
981
136
So now Intel can compete with all those FD-SOI CPUs based on Bulldozer that AMD are apparently working on? ;)
Intel already has 22FFL/22FFL+/22FFL++\NG.

Tower purchase isn't against FDSOI/RFSOI but rather SiC, GaN, etc. Which GlobalFoundries is just starting to get into.

The CPU core isn't derived from Bulldozer, but rather is just using Clustered Multithreading.
 

ASK THE COMMUNITY