Discussion Leading Edge Foundry Node advances (TSMC, Samsung Foundry, Intel) - [2020 - 2025]

[DisEnchantment]

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.


N7 performance is more or less understood.


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

AnandTech Forums: Technology, Hardware, Software, and Deals

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

www.anandtech.com

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.

AnandTech Forums: Technology, Hardware, Software, and Deals

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

www.anandtech.com

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.

---------------------------------------------------------------------------------------------------------------------------------------------------


FEEL FREE TO CREATE A NEW THREAD FOR 2025+ OUTLOOK, I WILL LINK IT HERE

 

[Win2012R2]

Intel has the highest ARM License anyway

Yeah, much good it did to it, what about AWS, Google, Microsoft who extensively use ARMs licensed server cores - they are the only real users of that stuff, so what would ARM do then going forward: no more licensed cores, buy finished stuff instead?

 

[DZero]

What did they pull - all server chip licenses?

No, I mean, they will defy the other makers by making themselves ARM processors. The biggest loser will be Mediatek and UNISOC due not having something in house compared to Qualcomm, Apple, Samsung or Huawei.

 

[Win2012R2]

No, I mean, they will defy the other makers by making themselves ARM processors

ARM processors for what market - server presumably because all other markets pretty much closed?

It will be bad for Intel, crazy to give Fabs for that, on other hand ARM's made CPUs will be far more expensive than what hyperscalers get otherwise.

 

[511]

ARM processors for what market - server presumably because all other markets pretty much closed?

It will be bad for Intel, crazy to give Fabs for that, on other hand ARM's made CPUs will be far more expensive than what hyperscalers get otherwise.

The stake is a mere 2% they don't have any significant voting power.

Yeah, much good it did to it, what about AWS, Google, Microsoft who extensively use ARMs licensed server cores - they are the only real users of that stuff, so what would ARM do then going forward: no more licensed cores, buy finished stuff instead?

IPU and other stuff Intel uses ARM quite a lot

 

[Win2012R2]

The stake is a mere 2% they don't have any significant voting power.

Yeah, that's nothing, making ARMs own CPUs (does not matter where) is everything tho.

IPU and other stuff Intel uses ARM quite a lot

Small fries, everybody uses ARM a lot, but Ampere sized server chips are a different matter.

 

[511]

Yeah, that's nothing, making ARMs own CPUs (does not matter where) is everything tho.

If it fills Empty Fabs than why not

 

[Win2012R2]

If it fills Empty Fabs than why not

Question is: who is going to buy ARM's own server chips which they'd expect to sell at 50%+ margins, same people who did not buy Ampere?

 

[DrMrLordX]

Not that this is an/the ARM thread, but allegedly ARM will be increasing their licensing fees (potentially by a lot) now that they see Amazon et al doing so much with in-house designs.

 

[Thibsie]

I guess some vendors will be happy, huh.
Arm gains more market share but the bigger it gets, the bigger the target on its back.

 

[511]

anyway back to foundry topic we will get an update with Clearwater forest/18A in like 4-5 days

 

[Doug S]

No, I mean, they will defy the other makers by making themselves ARM processors. The biggest loser will be Mediatek and UNISOC due not having something in house compared to Qualcomm, Apple, Samsung or Huawei.

That won't affect Mediatek or other SoC vendors regardless of whether they have custom cores, because ARM has no modem. There isn't any market in the Android world for SoCs that don't include a modem, because then you have to either fab your SoC at Samsung to use theirs or bend over for Qualcomm. Google is the only one taking that route, because they're able to find enough people to buy Pixels at iPhone/Galaxy prices despite a midrange SoC - something no other Android OEM could pull off.

 

[Hitman928]

US government to own 9.9% of Intel with $11.1B investment, though $2.2B of that Intel already received through the CHIPS act.

https://seekingalpha.com/news/4488523-intel-confirms-massive-11_1b-investment-by-us-government?mailingid=41257371&messageid=2900&serial=41257371.22365&source=email_2900&utm_campaign=rta-stock-news&utm_content=link-1&utm_medium=email&utm_source=seeking_alpha&utm_term=41257371.22365

 

[MS_AT]

US government to own 9.9% of Intel with $11.1B investment, though $2.2B of that Intel already received through the CHIPS act.

https://seekingalpha.com/news/4488523-intel-confirms-massive-11_1b-investment-by-us-government?mailingid=41257371&messageid=2900&serial=41257371.22365&source=email_2900&utm_campaign=rta-stock-news&utm_content=link-1&utm_medium=email&utm_source=seeking_alpha&utm_term=41257371.22365

Isn’t everything funded by the already awarded grants? Like the gov got it for basically for agreeing to follow up to give what it already promised...

 

[gdansk]

Isn’t everything funded by the already awarded grants? Like the gov got it for basically for agreeing to follow up to give what it already promised...

It's hard for me to argue that isn't a better deal from the government's perspective. But maybe shortsighted.

 

[Hitman928]

Isn’t everything funded by the already awarded grants? Like the gov got it for basically for agreeing to follow up to give what it already promised...

Yes, but I think the government wasn’t going to be actually giving that money anymore unless Intel agreed to this.

 

[Josh128]

Wonder what the stock is about to do now? Is 11 billion enough to do anything?

https://twitter.com/x/status/1958998149231780303

 

[Win2012R2]

Is 11 billion enough to do anything?

Should be almost exactly enough to finance 11 billion worth of stock buy backs.

 

[jpiniero]

Wonder what the stock is about to do now? Is 11 billion enough to do anything?

Maybe a down payment on 14A.

 

[Saylick]

I don't see money as Intel's problem. It's that they lack the talent and grit to develop a node that other fabless designers want.

With the brain drain that's happened at Intel in the last few years, TSMC's engineers simply work harder and longer than their counterparts at Intel. As I understand it, if you are Taiwanese, working for TSMC is practically the equivalent of being a doctor or lawyer in terms of national prestige. TSMC is a household name on that island, and the employees know that their country's national security is on the line if they fail. That coupled with TSMC's customer service as a pure play fab and the Asian work culture makes it a hard combination to beat. I don't see how IFS will compete with TSMC at the cutting edge unless cutting edge nodes become so hard to develop that it becomes inevitable that TSMC stagnates.

 

[jpiniero]

I don't see money as Intel's problem. It's that they lack the talent and grit to develop a node that other fabless designers want.

Money is obviously an problem. You need mega volume and you need to sell the wafers at a very high price if you are serious about staying in and not just wasting "Investors"/Taxpayers money.

You are not wrong about the second sentence though.

 

[gdansk]

Hmm, money looks to be Intel's largest problem at the moment. The one time impairments have not been as one time as Intel claimed.

 

[Saylick]

I

Money is obviously an problem. You need mega volume and you need to sell the wafers at a very high price if you are serious about staying in and not just wasting "Investors"/Taxpayers money.

You are not wrong about the second sentence though.

Idk... I feel like money wouldn't be a problem if they didn't have technical issues with their nodes.

 

[gdansk]

I feel like money wouldn't be a problem if they didn't have technical issues with their nodes.

Hmm, I have to disagree. Samsung is having technical issues but since they are printing money in other markets it seems likely their logic fabs can weather the storm if the company chooses to do so.

Are their technical issues less severe? I'm not sure, maybe. But a major difference is many American companies are atomized for various off topic reasons. And so they need external investment if they want to make it through to the other side where they have external customers.

I still argue that now the fabs should be split off entirely so we don't reward Intel design for Intel fabrication failures. But ultimately both administrations have made decisions that ended up with billions flowing to Intel as a whole.

 

[DavidC1]

If I recall correctly, Sierra Forest came out in early-to-mid 2024 for a specific customer that had made a firm commitment to buy, with commercial availability of Intel 3 products coming later in the year. So TECHNICALLY it was like April/May 2024.

Lunarlake/Arrowlake came out late last year. We know in late 2023 18A would have been late 2025, and the CLW would be among the first movers. 2023 was Meteorlake, so a year later product is Lunar, and then another year is Panther. People were REALLY expecting 18A in early 2025? How does that even make sense? A year's worth of gap is expected. Otherwise it's skipped by Intel or the market.

Remember that 20A was supposed to fit in between too. It was 1 out of the "5N4Y". So 18A was always latter half of 2025. 20A Arrow in late 2024, and 18A Panther in late 2025.

 

[Joe NYC]

Wonder what the stock is about to do now? Is 11 billion enough to do anything?

https://twitter.com/x/status/1958998149231780303

Intel is issuing new stock, so this basically leaves Intel with extra cash and Intel shareholders with diluted ownership per share.