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

Page 19 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
TSMC's N7 EUV is now in its second year of production and N5 is contributing to revenue for TSMC this quarter. N3 is scheduled for 2022 and I believe they have a good chance to reach that target.

1587737990547.png
N7 performance is more or less understood.
1587739093721.png

This year and next year TSMC is mainly increasing capacity to meet demands.

For Samsung the nodes are basically the same from 7LPP to 4 LPE, they just add incremental scaling boosters while the bulk of the tech is the same.

Samsung is already shipping 7LPP and will ship 6LPP in H2. Hopefully they fix any issues if at all.
They have two more intermediate nodes in between before going to 3GAE, most likely 5LPE will ship next year but for 4LPE it will probably be back to back with 3GAA since 3GAA is a parallel development with 7LPP enhancements.


1587739615344.png

Samsung's 3GAA will go for HVM in 2022 most likely, similar timeframe to TSMC's N3.
There are major differences in how the transistor will be fabricated due to the GAA but density for sure Samsung will be behind N3.
But there might be advantages for Samsung with regards to power and performance, so it may be better suited for some applications.
But for now we don't know how much of this is true and we can only rely on the marketing material.

This year there should be a lot more available wafers due to lack of demand from Smartphone vendors and increased capacity from TSMC and Samsung.
Lots of SoCs which dont need to be top end will be fabbed with N7 or 7LPP/6LPP instead of N5, so there will be lots of wafers around.

Most of the current 7nm designs are far from the advertized density from TSMC and Samsung. There is still potential for density increase compared to currently shipping products.
N5 is going to be the leading foundry node for the next couple of years.

For a lot of fabless companies out there, the processes and capacity available are quite good.
 

MasterofZen

Junior Member
Jul 12, 2021
15
16
41
PhD vs MS is not always a good metric for research. PhD types are mainly in-the-box thinkers but are great at documentation. You need your rare creative types to push new ideas, and your PhDs to prove them valid. If you spend your career in documentation it does not bode well for their creative streaks. Ultimately what ideas get explored are up to decision makers that also have to count the pennies.
What the heck are you talking about? I would agree that R&D are not necessarily all PhDs or PhDs are not prerequisite for doing good R&D work. But to say PhDs are just good at documenting and validation is just pure B.S. I bet all major inventions in semicon. (CMOS, MOSFET, FinFet etc.) are mainly invented by PhDs.
 

MasterofZen

Junior Member
Jul 12, 2021
15
16
41
Why does the degree matter? Even someone with a good undergraduate or masters degree is probably overqualified for monitoring a single piece of equipment all day long, especially if the MTBF is high.
Yes, it's possible that they are over-qualified, but how do you know. My point is that what this guy said is totally wrong, so his conclusion is not credible. Unless there is some other evidence, we don't know whether they are over-qualified.
 

maddie

Diamond Member
Jul 18, 2010
4,722
4,624
136
What the heck are you talking about? I would agree that R&D are not necessarily all PhDs or PhDs are not prerequisite for doing good R&D work. But to say PhDs are just good at documenting and validation is just pure B.S. I bet all major inventions in semicon. (CMOS, MOSFET, FinFet etc.) are mainly invented by PhDs.
I guess it might be described thus.

There are some that are great so they get PhDs and those that get PhDs so that they become great.
 

MadRat

Lifer
Oct 14, 1999
11,908
228
106
Or he merely supported the notion to prove new ideas takes a PhD backing it to become credible. What person in charge is going to run with ideas not proven? They want sure choices to take to the implementation stage.
 

eek2121

Platinum Member
Aug 2, 2005
2,904
3,906
136
What the heck are you talking about? I would agree that R&D are not necessarily all PhDs or PhDs are not prerequisite for doing good R&D work. But to say PhDs are just good at documenting and validation is just pure B.S. I bet all major inventions in semicon. (CMOS, MOSFET, FinFet etc.) are mainly invented by PhDs.

This conversation has gone off the rails. Want me to really blow your mind? Degrees, in general, do not equate to knowledge, ability to learn, or talent. Some of the brightest engineers you will ever meet don't have a degree of any kind. How about we get back to foundry talk?
 

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
As we all already expected it looks like Intel will indeed be restricted by the ability of ASML to deliver the necessary hardware.

Chipmakers’ multibillion-dollar expansion plans will be constrained by a shortage of critical equipment over the next two years as the supply chain struggles to step up production, according to one of the industry’s most important suppliers.

The warning comes from Peter Wennink, chief executive of ASML, which dominates the global market for the lithography machines used to make advanced semiconductors.

“Next year and the year after there will be shortages,” Wennink said. “We’re going to ship more machines this year than last year and . . . more machines next year than this year. But it will not be enough if we look at the demand curve. We really need to step up our capacity significantly more than 50 per cent. That will take time.”

ASML’s machines are used to etch circuits into silicon wafers. “It is the single most critical company in the semiconductor supply chain,” said Richard Windsor, tech analyst at Radio Free Mobile. “It is the printing press of silicon chips.”

Wennink said ASML was assessing with its suppliers how to increase capacity. It was not yet clear the scale of investment required, he said. ASML has 700 product related suppliers, of which 200 are critical.

His comments come as the semiconductor industry accelerates investment in new production to meet a global shortage of chips and surging demand. Analysts expect the market to double to $1tn by 2030.

Intel last week said it would invest roughly €33bn in manufacturing and research in Europe, rising to €80bn by the end of the decade, depending on demand. It has also announced plans to invest $40bn to expand chip manufacturing in the US.

The US chipmaker is racing to catch up with the industry leader, Taiwan’s TSMC, which is investing more than $100bn over the next three years. Samsung has said it will invest $150bn by the end of the decade to expand production, part of an estimated Won510tn ($421bn) to be invested by more than 150 South Korean companies, according to the government.

The US and Europe are also planning tens of billions in support for chip manufacturing, in an attempt to reduce their reliance on Asian manufacturers.

Pat Gelsinger, chief executive of Intel, acknowledged that the equipment shortage posed a challenge for the company’s expansion plans. He said he was in direct contact with Wennink on the shortages, and Intel had sent its own manufacturing experts to the company to help accelerate production.

“Today this is a constraint,” he told the Financial Times. But he stressed that there was still time to resolve the issue. It would take two years to build the shell of the chip factory. “Then you start to fill it with equipment in year three or four,” he said.

Wennink agreed there was still some time to expand capacity in the supply chain, as many of the new manufacturing facilities would not come on line before 2024. But this would not be simple. For example, the most complex component of ASML’s equipment was the lens, made by German manufacturer Carl Zeiss.

“They need to make significantly more lenses,” Wennink said. But first the company would have to “build clean rooms; they need to start asking for permits; they need to start organising the building of a new factory. Once a factory is ready, they need to order the manufacturing equipment; they need to hire people. And then . . . it takes more than 12 months to make the lens.”
 
Last edited:

maddie

Diamond Member
Jul 18, 2010
4,722
4,624
136
As we all already expected it looks like Intel will indeed be restricted by the ability of ASML to deliver the necessary hardware.

Chipmakers’ multibillion-dollar expansion plans will be constrained by a shortage of critical equipment over the next two years as the supply chain struggles to step up production, according to one of the industry’s most important suppliers.

The warning comes from Peter Wennink, chief executive of ASML, which dominates the global market for the lithography machines used to make advanced semiconductors.

“Next year and the year after there will be shortages,” Wennink said. “We’re going to ship more machines this year than last year and . . . more machines next year than this year. But it will not be enough if we look at the demand curve. We really need to step up our capacity significantly more than 50 per cent. That will take time.”

ASML’s machines are used to etch circuits into silicon wafers. “It is the single most critical company in the semiconductor supply chain,” said Richard Windsor, tech analyst at Radio Free Mobile. “It is the printing press of silicon chips.”

Wennink said ASML was assessing with its suppliers how to increase capacity. It was not yet clear the scale of investment required, he said. ASML has 700 product related suppliers, of which 200 are critical.

His comments come as the semiconductor industry accelerates investment in new production to meet a global shortage of chips and surging demand. Analysts expect the market to double to $1tn by 2030.

Intel last week said it would invest roughly €33bn in manufacturing and research in Europe, rising to €80bn by the end of the decade, depending on demand. It has also announced plans to invest $40bn to expand chip manufacturing in the US.

The US chipmaker is racing to catch up with the industry leader, Taiwan’s TSMC, which is investing more than $100bn over the next three years. Samsung has said it will invest $150bn by the end of the decade to expand production, part of an estimated Won510tn ($421bn) to be invested by more than 150 South Korean companies, according to the government.

The US and Europe are also planning tens of billions in support for chip manufacturing, in an attempt to reduce their reliance on Asian manufacturers.

Pat Gelsinger, chief executive of Intel, acknowledged that the equipment shortage posed a challenge for the company’s expansion plans. He said he was in direct contact with Wennink on the shortages, and Intel had sent its own manufacturing experts to the company to help accelerate production.

“Today this is a constraint,” he told the Financial Times. But he stressed that there was still time to resolve the issue. It would take two years to build the shell of the chip factory. “Then you start to fill it with equipment in year three or four,” he said.

Wennink agreed there was still some time to expand capacity in the supply chain, as many of the new manufacturing facilities would not come on line before 2024. But this would not be simple. For example, the most complex component of ASML’s equipment was the lens, made by German manufacturer Carl Zeiss.

“They need to make significantly more lenses,” Wennink said. But first the company would have to “build clean rooms; they need to start asking for permits; they need to start organising the building of a new factory. Once a factory is ready, they need to order the manufacturing equipment; they need to hire people. And then . . . it takes more than 12 months to make the lens.”
In my opinion, all predictions being made now are about to be rendered useless. The present supply chain for many products are about to be upended. ANYONE expecting the unfolding disruption in trading of commodities and correspondingly production is still viewing the world through pre-war lens. Over this year it will become obvious to all. I'm very pessimistic about this.
 
  • Like
Reactions: Tlh97 and Saylick

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
Wennink agreed there was still some time to expand capacity in the supply chain, as many of the new manufacturing facilities would not come on line before 2024. But this would not be simple. For example, the most complex component of ASML’s equipment was the lens, made by German manufacturer Carl Zeiss.

Sometime in the past couple of years, I watched a long detailed video on Carl Zeiss and the later half of the documentary focussed on their work in the Semiconductor arena. EUV mirrors appeared to be just as challenging as creating EUV lasers. The deviations from their prescribed curvature must be controlled to a much more stringent degree than mirrors for NASA's space telescopes. The alignment between the multiple mirrored elements along the optical path must also be held to a precision that seems impossible. This can only be done slowly with great diligence and constant measurement of each component and of each step of assembly. Honestly, I was shocked that we humans were able to do this with such large and heavy components - and yet we have.
 

moinmoin

Diamond Member
Jun 1, 2017
4,933
7,619
136
Sometime in the past couple of years, I watched a long detailed video on Carl Zeiss and the later half of the documentary focussed on their work in the Semiconductor arena. EUV mirrors appeared to be just as challenging as creating EUV lasers. The deviations from their prescribed curvature must be controlled to a much more stringent degree than mirrors for NASA's space telescopes. The alignment between the multiple mirrored elements along the optical path must also be held to a precision that seems impossible. This can only be done slowly with great diligence and constant measurement of each component and of each step of assembly. Honestly, I was shocked that we humans were able to do this with such large and heavy components - and yet we have.
Yeah, I thought the ending quote and the last sentence of the article were especially humbling in this regard:

"“They need to make significantly more lenses,” Wennink said. But first the company would have to “build clean rooms; they need to start asking for permits; they need to start organising the building of a new factory. Once a factory is ready, they need to order the manufacturing equipment; they need to hire people. And then . . . it takes more than 12 months to make the lens.”"
 

Saylick

Diamond Member
Sep 10, 2012
3,084
6,184
136
Yeah, I thought the ending quote and the last sentence of the article were especially humbling in this regard:

"“They need to make significantly more lenses,” Wennink said. But first the company would have to “build clean rooms; they need to start asking for permits; they need to start organising the building of a new factory. Once a factory is ready, they need to order the manufacturing equipment; they need to hire people. And then . . . it takes more than 12 months to make the lens.”"
Yeah, I've said it a few times in this forums as well, that even if Intel throws money at ASML they would still have to wait a long time before they get a substantial quantity of EUV machines. That's what happens when Intel literally didn't buy more than a paltry amount of EUV machines each year (source: look at ASML earnings reports, then look at revenue by continent). Meanwhile, TSMC and Samsung have been making orders year over year. The back log of orders can't simply be bypassed because those are legal obligations; no amount of money from Intel can let them jump ahead of another customer who's already put down their money months if not years before. Lastly, Intel says they're adopting High NA EUV first before everyone else, mostly as a PR move to be honest. They might get the first machine, but TSMC will likely get the 2nd one within the same week if not month. Good luck to Intel if they want to catch up to TSMC. They'll need to make twice the orders for those expensive High NA EUV machines if they want to catch up, and even then, that would take many years after production of said machines start. The intersection of the two lines representing the EUV output of each company won't be before 2025 guaranteed. Likely not before 2027 I bet.

This tweet from Dylan wasn't a surprise for me, knowing the above:
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
@Saylick

You aren't the only one that's been saying this, but of course you are correct. It's good to see someone putting at least a few detailed points in there to drive home the argument:

Intel doesn't have enough EUV machines to compete with even their past products on Intel 4 or Intel 3.

Sure the designs will be an improvement, but from the manufacturing side, there just isn't going to be much product to go around. Not compared to 10nm+/10SF/10ESF, and definitely not compared to 14nm variants.
 
  • Like
Reactions: Tlh97 and Ajay

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
Is there a source for this "10% of laptop volume" claim? That seems far lower than one'd expect. Especially since it only uses a little Intel 4.
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
Yeah, I've said it a few times in this forums as well, that even if Intel throws money at ASML they would still have to wait a long time before they get a substantial quantity of EUV machines. That's what happens when Intel literally didn't buy more than a paltry amount of EUV machines each year (source: look at ASML earnings reports, then look at revenue by continent). Meanwhile, TSMC and Samsung have been making orders year over year. The back log of orders can't simply be bypassed because those are legal obligations; no amount of money from Intel can let them jump ahead of another customer who's already put down their money months if not years before. Lastly, Intel says they're adopting High NA EUV first before everyone else, mostly as a PR move to be honest. They might get the first machine, but TSMC will likely get the 2nd one within the same week if not month. Good luck to Intel if they want to catch up to TSMC. They'll need to make twice the orders for those expensive High NA EUV machines if they want to catch up, and even then, that would take many years after production of said machines start. The intersection of the two lines representing the EUV output of each company won't be before 2025 guaranteed. Likely not before 2027 I bet.

This tweet from Dylan wasn't a surprise for me, knowing the above:
Wow o_O. I didn't know it was *that* bad! [that being Intel 4 wafer output]. I thought that using one small I4 compute chiplet would have allowed for a higher % of new laptop sales to be built with ML. This is pathetic. Guess Intel will be using TSMC N3 for a while.

ASML's backlog has lead to very long lead times from booking to delivery. It seems highly probably that your prediction is closer to the truth. It's not like TSMC is going to stop ordering EUV lithography machines so that Intel can catch up.
 

Saylick

Diamond Member
Sep 10, 2012
3,084
6,184
136
Wow o_O. I didn't know it was *that* bad! [that being Intel 4 wafer output]. I thought that using one small I4 compute chiplet would have allowed for a higher % of new laptop sales to be built with ML. This is pathetic. Guess Intel will be using TSMC N3 for a while.

ASML's backlog has lead to very long lead times from booking to delivery. It seems highly probably that your prediction is closer to the truth. It's not like TSMC is going to stop ordering EUV lithography machines so that Intel can catch up.
I got to stay on top of this stuff as an AMD investor, ya know? ;)

If I had to guess, Intel today has about 10-20 EUV machines vs. TSMC having like 60-80 machines. It's possible 10-20 EUV machines are enough just to supply Intel for their own Intel 4 demand, but it will be dependent on yields and the number of EUV layers Intel 4 uses. TSMC has years of EUV experience ahead of Intel: TSMC has shipped N5 in high volume for over a year now, while Intel hasn't even rolled out an EUV node for HVM. The small quantity of Intel 4 for Meteorlake mobile is probably just risk manufacturing, i.e. just to get some out to say they're shipping an EUV node for PR reasons. Dylan is right in that the next critical step, which is figuring out how to ramp to high-volumes, will be crucial. If Intel can't figure that out, it won't matter how many EUV machines they have because they wouldn't be leveraging them to their best potential.
 

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
Is there a source for this "10% of laptop volume" claim? That seems far lower than one'd expect. Especially since it only uses a little Intel 4.

It's a person who is an analyst that is supposed to have good (but unidentified) sources. I don't give much weight to any of these analyst types, but that's where it comes from. Some people who have actually followed him seem to think he's trustworthy :shrug:
 

oak8292

Member
Sep 14, 2016
81
67
91
This doesn't say much about the specific number of EUV at either TSMC or Intel but it does discuss the upcoming shortage of machines to fill all of the clean room space.

"I believe there will be demand for 20 more EUV tools than ASML can produce each of the next 3 years. To put that is perspective, ASML shipped 42 EUV systems in 2021 and is forecasting 55 system in 2022. Interestingly I saw a story today where Pat Gelsinger commented that he is personally talking to the CEO of ASML about system availability and admitted that EUV system availability will likely gate the ability to bring up all the new fabs."

 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
It's a person who is an analyst that is supposed to have good (but unidentified) sources. I don't give much weight to any of these analyst types, but that's where it comes from. Some people who have actually followed him seem to think he's trustworthy :shrug:
Call me skeptical then.

It would leave them essentially stuck with Alder Lake through till probably the latter half of '24.