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

 

[511]

That's the problem with someone reporting yield it entirely depends on what metric it is calculated and at what performance figures die size as well

 

[DZero]

That's the problem with someone reporting yield it entirely depends on what metric it is calculated and at what performance figures

They want performance but are risking to lose the advancements, they are not Huawei who are extremely limited but with SW pulls miracles
They are not Samsung, since Samsung really screwed up from 10nm to recently their 4nm which FINALLY become decent enough but has yield issues.

 

[511]

They want performance but are risking to lose the advancements, they are not Huawei who are extremely limited but with SW pulls miracles
They are not Samsung, since Samsung really screwed up from 10nm to recently their 4nm which FINALLY become decent enough but has yield issues.

cause performance sells it's in their DNA at one point they will freeze the process and than they will begin HVM sometime this year. Reuturs made this as exclusive btw.

 

[Geddagod]

Lmao Ian is coping

 

[LightningZ71]

Being a pipe cleaner makes sense on at least some level.... But, if that's the case, they won't make any appreciable profit off of NVL as those N2P die can not be inexpensive... both to acquire, and to ship to their packaging facility.

 

[Geddagod]

Being a pipe cleaner makes sense on at least some level.... But, if that's the case, they won't make any appreciable profit off of NVL as those N2P die can not be inexpensive... both to acquire, and to ship to their packaging facility.

Pipe cleaner for what lol

 

[Win2012R2]

So NVL is basically N2 with 18A nowhere to be seen apart from cheap Celeron level parts???

 

[Geddagod]

So NVL is basically N2 with 18A nowhere to be seen apart from cheap Celeron level parts???

Sounds like it.
Ofc SoC and iGPU tiles is also rumored to be internal, so there's still volume, but...

 

[jpiniero]

Pipe cleaner for what lol

It actually makes some sense to try to make some money off of 18A as they try to get the yield up.

 

[Win2012R2]

Sounds like it.

So this brings this question - will there be enough N2 capacity for Apple, AMD and now Intel?

 

[Geddagod]

It actually makes some sense to try to make some money off of 18A as they try to get the yield up.

That should have been PTL. And then should have been CLF. And DMR should be launching around the same time as NVL too.
Ian is just not trying to make his Intel contacts mad lol, that's the only reasonable explanation for why he is calling this a pipe cleaner.....

 

[511]

Being a pipe cleaner makes sense on at least some level.... But, if that's the case, they won't make any appreciable profit off of NVL as those N2P die can not be inexpensive... both to acquire, and to ship to their packaging facility.

why though what are you basing this on? If we take the leaked die even if compute tile is TSMC everything is Intel
iGPU Desktop - Intel 3 the same one used in PTL with 4 Xe3 Cores
HUB - 18A
PCD - Intel 4/3(guess)
Compute Tile -> TSMC N2

Now ARL in Reference
iGPU - N5
Soc - N6
IO -> N6
Compute N3B
packing is Intel

So this brings this question - will there be enough N2 capacity for Apple, AMD and now Intel?

Yeah TSMC capacity is based on customer orders

 

[511]

That should have been PTL. And then should have been CLF. And DMR should be launching around the same time as NVL too.

WCL-U NVL Hub Celestial i/dGPU would 18A capacity with their plan would be enough considering they are cutting CAPEX

 

[Win2012R2]

Yeah TSMC capacity is based on customer orders

It does not magically appear however in an instant, that order should have been made long time ago when Pat was in charge, and why would he do that if he bet the farm on 18A?

I would have thought Intel could not have ordered a lot of N2, AMD already gets first dibs on it and making MI400 chiplets would have profitability far higher than anything Intel would make with them, likely Zen 6 will be in same territory, so it follows AMD should be able to outbid Intel, big incentive to do so since it would limit competitors supply of chips that compete.

Intel is cooked, it's over.

 

[Geddagod]

WCL-U NVL Hub Celestial i/dGPU would 18A capacity with their plan would be enough considering they are cutting CAPEX

Makes more sense to scale volume down other parts than cut down the 4+8 die from a financial perspective... if it was actually all about volume from the start lol.

 

[511]

It does not magically appear however in an instant, that order should have been made long time ago when Pat was in charge, and why would he do that if he bet the farm on 18A?

cause the decision was made after him and before Lip Bu like Bionic said so likely our DEI experts CO-CEOs Dave and Mitchell

 

[511]

Makes more sense to scale volume down other parts than cut down the 4+8 die from a financial perspective... if it was actually all about volume from the start lol.

Intel is expecting 20K WSPM for 18A/AP by 2026 this was few months old info for example TSMCs plan is 60K WSPM N2.

 

[Win2012R2]

cause the decision was made after him

Source?

 

[jpiniero]

That should have been PTL. And then should have been CLF. And DMR should be launching around the same time as NVL too.

CLF and DMR will be paper launches I guess.

 

[511]

Source?

https://twitter.com/x/status/1954978484817891621

 

[Geddagod]

Intel is expecting 20K WSPM for 18A by 2026 for example TSMCs plan is 60K WSPM

And Intel already sunk a ton of money into even more 18A capacity, that they will esentially just not be using because no one (including themselves) wants more 18A capacity.
Also yes, Intel ships much less volume than TSMC. This was a thing even before 18A.

 

[511]

CLF and DMR will be paper launches I guess.

They will be as paper launch as GNR/SRF was

 

[Geddagod]

CLF and DMR will be paper launches I guess.

Nah, I think Ian is just coping.
CLF maybe, but Intel will be pushing a bunch of 18A volume no matter what. They have too, in order to capitalize of the tons of money they already poured into 18A fab construction.

 

[jpiniero]

They will be as paper launch as GNR/SRF was

You can buy both now, it just took forever.

At this point, if Intel was serious about recovering their server share, they need to fab at TSMC.

 

[Geddagod]

They will be as paper launch as GNR/SRF was

Hopefully DMR ramps faster than GNR seemed to have.
I just refuse to believe GNR allowed AMD to gain so much market share in such a short amount of time, based on their competitiveness LOL. Hopefully their Q1 market share update coming tomorrow? IIRC, would have better numbers.