Huawei Prepares EUV Scanner for Sub-7 nm Chinese Chips

[TumbleGeorge]

As an end user of overpriced hardware, I don't care about the formal reasons why ASML is a monopoly. Of course, the prices are dictated by many more factors, it is not only the price of the machines that is reflected in the price of the products. But I think the FTC should still find a way to ensure healthy competition at the cutting edge of technology using all legal measures.

 

[JensenHuang]

ASML has a monopoly when it comes to 14nm and lower. This monopoly is maintained by price fixing and US pressure and US sanctions. This is more than just one company this is about Chip/technology wars.

 

[evernessince]

It isn't a monopoly, other companies make the same products. There is just no one competing at their level.


Anyway good for the Chinese, we need competition. The problem is that ASML depends on a lot of other companies around the world to make their machines. Zeiss being the most notable.

I think people tend to focus on whether something is a monopoly or not. Ultimately the most important factor is how healthy the market is. Is competition able to get into the market and compete, is innovation being driven, are prices competitive for customers, ect.

Ultimately if a company is dominating a market without using their leverage to prevent others from entering the market while driving innovation that's acceptable.

That said, markets with one dominating player should be subject to frequent review as there is a large chance for undesirable situations to develop under the surface. A company that is not abusing it's position should have no problem with this.

 

[Vayra86]

OK, I can agree with that to a large extent. It's ridiculous because it has lasted so long and still there's no competitor apparently in sight. But what should ASML do? Share enough information with Nikon to enable them to build an EUV scanner too? It would still take Nikon several years to actually build the components and the system, and also to build the supply chain.

That's the whole thing...

Even if ASML would provide a licensing structure to its EUV technologies, they would honestly just be giving their exisiting supply chain more pressure. The fact is, many of the stuff inside these machines simply isn't produced in the vast majority of the world, and is built on yet another stack of unique technologies.

I say, GL Huawei, you do you, I think the vast majority of this message is political. China is losing even its access to DUV scanners, and now they 'poof!' have an EUV design going? Its the ultimate middle finger at 'Murica, and nothing else. So: Good luck! We'll see those fantastic chips soon, won't we?

In other news, I heard China 'cured Omicron' by simply not calling it a problematic disease anymore. What was grounds for a lockdown a month or two ago, is now nothing but a cough and a sneeze. All is well here boys, you can move on now. We're talking about an autocratic regime here desperately trying to save face. Place that lens (! )) on the EUV development here and you really know what's what.

Covid cases explode in Beijing leaving city streets empty and daily life disrupted | CNN

Empty streets, deserted shopping centers, and residents staying away from one another are the new normal in Beijing -- but not because the city, like many Chinese ones before it, is under a "zero-Covid" lockdown.

edition.cnn.com

The reality though is that a few million will die here, after several years under stringent lockdown policies. Sounds comfy

 

[Bomby569]

I think people tend to focus on whether something is a monopoly or not. Ultimately the most important factor is how healthy the market is. Is competition able to get into the market and compete, is innovation being driven, are prices competitive for customers, ect.

Ultimately if a company is dominating a market without using their leverage to prevent others from entering the market while driving innovation that's acceptable.

That said, markets with one dominating player should be subject to frequent review as there is a large chance for undesirable situations to develop under the surface. A company that is not abusing it's position should have no problem with this.

completely agree, the only issue here is oversight to make sure they don't abuse their position.

ASML has a monopoly when it comes to 14nm and lower. This monopoly is maintained by price fixing and US pressure and US sanctions. This is more than just one company this is about Chip/technology wars.

China announced recently the mass production of 14nm.

Shanghai confirms 14-nm chips being mass produced now - Global Times

www.globaltimes.cn

They are the only one how can they be price fixing? i don't think you know what that means.

US pressure and sanctions?! What! That could be a barrier to cpu manufacturing, sure. But lithography?! Even if the Chinese got the machines how would that help them making new machines? having one to copy?!

 

[PhantomTaco]

ASML has a monopoly when it comes to 14nm and lower. This monopoly is maintained by price fixing and US pressure and US sanctions. This is more than just one company this is about Chip/technology wars.

14nm and lower? Where do you derive that from? 14nm and lower =/ EUV lithography being used. In fact, plenty of 14-10nm nodes only use immersion lithography (DUV) with stuff like SAQP (Self Aligned Quadruple Patterning). In fact, Intel's commercially available processors today (to my knowledge) all don't use EUV layers at all. Keep in mind that the node/nm nomenclature at this point is literally nothing more than marketing terms, they don't really tie into much of anything real world.

As an end user of overpriced hardware, I don't care about the formal reasons why ASML is a monopoly. Of course, the prices are dictated by many more factors, it is not only the price of the machines that is reflected in the price of the products. But I think the FTC should still find a way to ensure healthy competition at the cutting edge of technology using all legal measures.

What exactly is the FTC going to do? ASML isn't imposing any barriers whatsoever to others entering. They came up with a consistent method for generating enough EUV wavelength light to be used for lithography, and started selling the machines along with service contracts. They haven't been strongarming through litigation to prevent others from entering, or much of anything really. As an end user of overpriced hardware, I think what you and others should realize is that the tools these companies make are so advanced, and so EXPENSIVE to even develop/manufacture that they don't even really make most of their money on selling the hardware. It's keeping these things functioning in the form of service contracts. As for an end user of overpriced hardware, speak with your wallet. If you already are, that's awesome because that's how changes happen in the end: when bottom lines are impacted.

 

[JensenHuang]

completely agree, the only issue here is oversight to make sure they don't abuse their position.






China announced recently the mass production of 14nm.

Shanghai confirms 14-nm chips being mass produced now - Global Times

www.globaltimes.cn

They are the only one how can they be price fixing? i don't think you know what that means.

US pressure and sanctions?! What! That could be a barrier to cpu manufacturing, sure. But lithography?! Even if the Chinese got the machines how would that help them making new machines? having one to copy?!

China is very good at reverse engineering, many of China's breakthroughs are a result of IP and technology theft. Though they have announced 14nm, China is well documented at lying about their advancements.
The big players had massive stocks of 28nm wafers years ago, TSMC sold this for cheap to Chinese companies undercutting SMIC. This industry is full of manipulation, and the chips wars have just heated up
The US did the samething to Japans semiconductor industry as well.

 

[RandallFlagg]

ASML has a monopoly when it comes to 14nm and lower. This monopoly is maintained by price fixing and US pressure and US sanctions. This is more than just one company this is about Chip/technology wars.

Nikon can do under 14nm already.

The NSR-S635E ArF Immersion Scanner, developed for use in 5nm node processes for high-volume semiconductor manufacturing, realized mix-and-match overlay (MMO*) of 2.1 nm or below and productivity delivering throughput capabilities of over 275 wafers per hour.

Semiconductor Lithography Systems | Product Technology | Nikon About Us

www.nikon.com

 

[onemanhitsquad]

China is very good at reverse engineering, many of China's breakthroughs are a result of IP and technology theft. Though they have announced 14nm, China is well documented at lying about their advancements.
The big players had massive stocks of 28nm wafers years ago, TSMC sold this for cheap to Chinese companies undercutting SMIC. This industry is full of manipulation, and the chips wars have just heated up
The US did the samething to Japans semiconductor industry as well.

Your post is fact!...the Chinese haven't had an original idea since spaghetti.

 

[Ultra Taco]

I lived in mainland China since 2005. At every level society there's IP theft and illegal use of patents and technology. I am actually related to Chinese people who have factories that infringe on the patents of Western companies and sell to other developing nations who don't care if they're stealing illegally manufactured tech. Literally billions are stolen from the U.S. daily from small businesses, to public schools, all the way into government bureaus.

I think that monopoly observations about ASML are legitimate and accurate, they dictate the market. However they developed their technology in house and through decades of R&D.

I hope China makes breakthroughs that shake up the monopolistic market we have now. Nvidia and AMD being good reasons why. There are good reasons to have selfishly altruistic hopes for our species:

China or any country making breakthroughs with tech, innovative scientific research, or social justice progress benefit humanity and the free-market PH balance.

 

[Wirko]

Nikon can do under 14nm already.

They say 5nm but as far as I know (whatever I know, I've learned from semianalysis and semiengineering), 5nm is barely possible with DUV. Not impossible but costly and slow, with lots of multiple patterning.

Canon is farther behind, this is their best:

FPA-6300ES6a | Canon Global

This page provides information on High Resolution / High Productivity KrF Scanner "FPA-6300ES6a."

global.canon

But I noticed Canon has "i-line Steppers for Back-End-Of-the-Line (BEOL) and Advanced Packaging Applications" with a 52 x 68 mm exposure field. That's the size of interposers they should be able to make. So interposers aren't limited by the same 26 x 33 mm size as chips are.

I think that monopoly observations about ASML are legitimate and accurate, they dictate the market. However they developed their technology in house and through decades of R&D.

It was ASML's customers that helped create this monopoly. Intel, TSMC and Samsung all invested in it in 2012, when it needed money to develop EUV.

 

[Vayra86]

14nm and lower? Where do you derive that from? 14nm and lower =/ EUV lithography being used. In fact, plenty of 14-10nm nodes only use immersion lithography (DUV) with stuff like SAQP (Self Aligned Quadruple Patterning). In fact, Intel's commercially available processors today (to my knowledge) all don't use EUV layers at all. Keep in mind that the node/nm nomenclature at this point is literally nothing more than marketing terms, they don't really tie into much of anything real world.


What exactly is the FTC going to do? ASML isn't imposing any barriers whatsoever to others entering. They came up with a consistent method for generating enough EUV wavelength light to be used for lithography, and started selling the machines along with service contracts. They haven't been strongarming through litigation to prevent others from entering, or much of anything really. As an end user of overpriced hardware, I think what you and others should realize is that the tools these companies make are so advanced, and so EXPENSIVE to even develop/manufacture that they don't even really make most of their money on selling the hardware. It's keeping these things functioning in the form of service contracts. As for an end user of overpriced hardware, speak with your wallet. If you already are, that's awesome because that's how changes happen in the end: when bottom lines are impacted.

Most DUV scanners are also produced by Western oriented companies, Japanese, EU or US.

Fact is, 'we' own this market.

 

[CovertGoat]

The ASML monopoly is ridiculous, and I don't understand how it is legal.
Governments of the world should force them to share the tech with at least one another (western) company, just like they did with Intel and x86

But anyway, gratz to Huawei for pushing the tech as well !

good idea if you want to be cut off from advanced semiconductors. Public funding in the west created EUV, other governments have zero right to the intellectual property.

huawei is attempting to use a LED, it’s not going to work.

 

[RandallFlagg]

They say 5nm but as far as I know (whatever I know, I've learned from semianalysis and semiengineering), 5nm is barely possible with DUV. Not impossible but costly and slow, with lots of multiple patterning.

Canon is farther behind, this is their best:

FPA-6300ES6a | Canon Global

This page provides information on High Resolution / High Productivity KrF Scanner "FPA-6300ES6a."

global.canon

But I noticed Canon has "i-line Steppers for Back-End-Of-the-Line (BEOL) and Advanced Packaging Applications" with a 52 x 68 mm exposure field. That's the size of interposers they should be able to make. So interposers aren't limited by the same 26 x 33 mm size as chips are.


It was ASML's customers that helped create this monopoly. Intel, TSMC and Samsung all invested in it in 2012, when it needed money to develop EUV.

That's why I said "under 14nm".

The complexity of getting high yield on ASML DUV systems at 10nm is why Intel had such a hard and long time getting there, but obviously it can be done, and done in volume with good yield.

And just to note, TSMC "N7" is actually a 10nm node. I'm going to refer to the 'real' 10nm, which includes N7, N6, and Intel 7. This is actually the crossover point where it seems DUV can't be stretched much more.

Unlike the ASML systems, where ASML went for EUV on sub 14nm nodes, Nikon's equipment appears to be specifically designed to use DUV and multiple patterning to achieve higher density.

Nikon's claim 275 wafers/hour using multiple patterning - which is just a few percent below ASMLs top end DUV equipment's max speed which is usually hovering just below 300 wafers/hr.

Intel used ASMLs DUV systems, since Nikon's equipment wasn't out until 2019 and Intel was already making its first gen 10nm chips at that time (Ice Lake). ASML had EUV systems out two years earlier (2017) that could do 10nm, vs Nikon in 2019.

That's not really much difference.

Mainly my point was that there are in fact other competitors in this space, ASML is not the only one, and not everyone has to use an approach identical to ASML to achieve similar results.

As far as competing with China in this space, the west dominates right now. I'd be a lot more concerned about where people are going to get a MOSFET or LCD display if for example, trade with China were to stop for whatever reason.

We'd probably wind up with a glut of 7nm chips, which are useless without all those other not so sexy parts like capacitors, resistors, diodes, power switching transistors, LCD displays and so on that we *don't* make in the west.

Here's an interesting picture of Shanghai, 1990 vs 2010 :

 

[TumbleGeorge]

China is far ahead technologically than many are able to imagine and figured out. To realize such a thing, one must have a mind free from Western anti-China propaganda.

 

[aminy23]

As an end user of overpriced hardware, I don't care about the formal reasons why ASML is a monopoly. Of course, the prices are dictated by many more factors, it is not only the price of the machines that is reflected in the price of the products. But I think the FTC should still find a way to ensure healthy competition at the cutting edge of technology using all legal measures.

It's too niche.

Lots of companies make cars. Only SSC (Shelby SuperCars) makes commercial road legal cars that go over 300mph.

In the world you might have 5-10 shops equipped with a $1,000,000 machine that services a critical component of this car.

5-10 chip companies order a $10,000,000+ production machine from the company that makes the best one.

It's not a consumer product. No one (except maybe China/Iran/Cuba/North Korea/Russia) wants to invest billions of dollars to engineer a new one, and then sell 3-5 of them.

It's not a monopoly when you have the best of something - this occurs in every industry.

 

[RandallFlagg]

It's too niche.

Lots of companies make cars. Only SSC (Shelby SuperCars) makes commercial road legal cars that go over 300mph.

In the world you might have 5-10 shops equipped with a $1,000,000 machine that services a critical component of this car.

5-10 chip companies order a $10,000,000+ production machine from the company that makes the best one.
<snip>

ASML's top end EUV machines cost about $350M. Each.

Again I'd be a lot more worried about not so cutting edge stuff. The US and Europe have plenty of capability to make sub 10nm chips. China has nearly zero capability below 14nm right now, and despite some flag waving probably won't have 10nm for 3-5 more years. By then, the west will have moved on to 5nm. I mean the real 5nm, otherwise known as Intel 4/3 and TSMC N3, perhaps even 3nm otherwise known as Intel 18/20A and TSMC N2.

What we lack, is the ability to make LCD driver chips, MOSFETs, capacitors, transformers, and many other mundane parts.

That is why TI's expansion in Sherman TX is probably far more important than a new N5 fab in Arizona.

MSN

www.msn.com

 

[TumbleGeorge]

real 5nm,

You joke right?

 

[TheoneandonlyMrK]

Everything is extremely simple. It doesn't even need to be actively argued that only ASML has the necessary technologies for the bleeding edge, attracting investment and orders. Thus, competing companies have difficulties in raising active capital for the research, design and reorganization of production they need. Think of ASML as a black hole, with gravity engulfing everything in its reach.

That's your analogy too, it's EUV that was the black hole not ASML and that's the point.
We didn't get stuck at 14nm by accident, it wasn't without effort that DUV and sub 14nm didn't get done with EUV as sub 14nm was expected to use.
It was because making EUV happen at all pushed the science of what's possible beyond what was possible that light source alone took years to develop.
And even ASML accept there is plenty of refinement still necessary to push it's use since pelicles are still an issue with earlier degradation then Ideal.

And it's important to note that in many cases those negative comments about China's government were earned, we are not stupid, but we are also fair and I would guess in all cases people are just as negative in public about their own government, I know I am.

 

[TumbleGeorge]

the science of what's possible beyond what was possible

Hmm, explain that. What is practical possible mostly is problem of quantity and quality of engineering knowledge and experience than breakthrough in the sciences. I have not heard or read of the existence of any fundamental scientific discoveries in the last 100 years that would allow action beyond the laws of physics as we know them.

 

[TheoneandonlyMrK]

Hmm, explain that. What is practical possible mostly is problem of quantity and quality of engineering knowledge and experience than breakthrough in the sciences. I have not heard or read of the existence of any fundamental scientific discoveries in the last 100 years that would allow action beyond the laws of physics as we know them.

Did I say beyond the laws of physics?!.

No the edge of what's possible is often no where near the limit's of the laws of physics.

As I said creating the light source for the specific wavelength of light required which can output as required consistently and for a time span that isn't pointless.

The science of going beyond what's possible isn't the same as beyond the laws of physics.

What man or woman can possibly make is at the edge of what's possible in scientific and engineering terms.

ASML have had to push beyond that point continuously and consistently to be in the lead.

Plus new paradigm of engineering excellence had to be developed to create assemblies as accurate as required, including the many mirrors if you can still call them that.

 

[TumbleGeorge]

In this sense, only engineering experience has been pushed forward. There is no impact on the progress of specific natural sciences.

 

[TheoneandonlyMrK]

In this sense, only engineering experience has been pushed forward. There is no impact on the progress of specific natural sciences.

What sense, your showing none.

No one said ASML makes new physics.

I didn't.


I said that they're pushing what's possible for humans to make.

And creating systems and devices that humans have not before been possible to make at all.
And some of these, like the light source took a degree of innovation, a lot of time and a significant amount of money to develop.


Should they f£#@, be forced to split and pass that effort to someone else to make money off.

I have no shares, I'm a consumer only yet I see no sense in that.

 

[RandallFlagg]

You joke right?

Nope I'm not joking.

As a matter of fact, TSMC itself stated that N5 is not '5nm'.

If you aren't aware of this, you might want to strive to get better educated on these things.

But TSMC’s vice president of corporate research, Dr. Philip Wong, was keen to point out that after introducing his company’s latest node, despite a history of the node naming scheme actually having some relevance to the silicon features etched into the wafer, the node names are now effectively meaningless. So, while we might like to think that the N7, N5, and N3 names it’s using for its 7nm, 5nm, and 3nm nodes relate to the gate length of transistors, they’re effectively just brand names. ...

“It used to be the technology node, the node number, means something, some features on the wafer,” says Philip Wong in his Hot Chips 31 keynote. “Today, these numbers are just numbers. They’re like models in a car – it’s like BMW 5-series or Mazda 6. "

So N5 is just N5, it has nothing to do with 5nm.

Source (and there are many, many others) :

TSMC’s 7nm, 5nm, and 3nm “are just numbers… it doesn’t matter what the number is”

"These numbers are just numbers. They're like models in a car - it’s like BMW 5-series or Mazda 6."

www.pcgamesn.com

 

[TumbleGeorge]

As a matter of fact, TSMC itself stated that N5 is not '5nm'.

If you aren't aware of this, you might want to strive to get better educated on these things.

I'm exactly mean that. I watch on ITRS roadmap and I have an idea of what the actual numerical limits are in which the nodes are located.