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

 

[Josh128]

This can happen even Jim Keller said so but everything else yeah not happening.

lol what? Intel is not hitting 1 trillion market cap without everything else happening PLUS more that we have no idea about yet. Im talking 18A being good, 14A being great, and Nvidia, Apple, et al using it for products.

 

[511]

lol what? Intel is not hitting 1 trillion market cap without everything else happening PLUS more that we have no idea about yet. Im talking 18A being good, 14A being great, and Nvidia, Apple, et al using it for products.

They can hit it though if what you said happens.

 

[Josh128]

They can hit it though if what you said happens.

Absolutely, but thats a huge freaking "if".

 

[511]

Absolutely, but thats a huge freaking "if".

Yup but it's a possibility unlike the other Cope Intel bagholder are seeing.

 

[Joe NYC]

That's pure cope to say the least ain't no way happening in 5 years best US can get is 25-30% by 2030 . If TSMC/Intel and Samsung plant go online

I think you missed the point. Which, I noticed, 100% of Intel fans on Twitter also do.

The US companies can export, to the rest of the world, purely Taiwan fabbed chips. No restrictions on that. That's 65% of global demand.

To supply the US market, which is 35% of the global demand, Lutnick is talking about 40% (and maybe 50%). 40% or 35% is 14%. So TSMC would need to produce 14% of its capacity in the US. The other 21% of US demand can still come from from TSMC fabs.

Taiwan TSMC fabs would supply 65% + 21% = 86% of total demand.
US TSMC fabs would supply 14% of total

I asked Grok to project what percentage of advanced chips would come from the US at the end of all 6 phases being complete. ("Advanced" may not be perfectly defined).

Grok projection was that all 6 phases in Arizona can max out at 30k wafers per month each, for total of 240k wafers per month, and at the completion date of all 6 phases, TSMC would be producing, globally ~950k wafers per month using "advanced" nodes, so well in excess of 14%.

The mid point, 3 phases being completed, and I have seen estimates of late 2027 to 2029. So half of the total in the US and also less of total TSMC capacity should meet the 14% US domestic target.

 

[johnsonwax]

I wouldn't rule out Intel at $1T market cap. Tesla hit it with a less compelling growth prospect. That's not an endorsement of Intels ability to execute, rather an endorsement of investors eagerness to jump on every hype train that rolls by. That's only magnified by a large enough fraction of investors believing in Trump magical powers to bend reality to his will, which are always part of the narrative.

I will note the situation with the Chinese bound AI/GPUs, which are limited by US export restrictions which include a requirement they be fabbed in the US. Tariffs are an indirect market tool, but that's a very direct one. For now, China has rejected most of those chips on their side, but that's a tool that can be used quite heavily if the buying nation is on board. Tends to be a bit of a catch-22, as I think the administration is discovering.

 

[DrMrLordX]

@johnsonwax

There is also the very real possibility of the entire AI bubble popping, rendering the AI situation irrelevant.

 

[johnsonwax]

@johnsonwax

There is also the very real possibility of the entire AI bubble popping, rendering the AI situation irrelevant.

Yeah, but this a recurring cycle of tech hype. AI is the most recent (and worst) but the market is a weighing machine so when those dollars flow out of Nvidia and Oracle they'll flow into something else. The US has gutted so much of their other high return industries, real-estate can hardly absorb more investment than it already has, I have to think it'll just find a different tech corner to dump into, so why not a domestic fab hype cycle? Don't get me wrong, I've thought for a long time that the US should in fact grow in that direction, but wishing doesn't make it happen and there's no real strategic effort at work to get there.

 

[Doug S]

I wouldn't rule out Intel at $1T market cap. Tesla hit it with a less compelling growth prospect.

Tesla's CEO surrounds himself with a reality distortion field that Steve Jobs would have been jealous of. It isn't worth 1/10th of that market cap based on any of its current businesses. It is all fantasy about robotaxis and robots. Whatever people might say about Lip-Bu Tan, no one would suggest he's going to hype Intel to $1 trillion bubble cap by talking about imagined future successes like they've already happened.

Intel's all time market cap high was around $300B, set in 2000 during the heights of the dot com bubble. It came close to that a few times since but never matched it. During most of those 25 years Intel utterly dominated the desktop, laptop and server CPU market, and lead the world in process technology. So even if it did all those things it still wouldn't be worth close to $1 trillion, it would have to do more. Do more like taking a big chunk of the AI market - and do it before that bubble bursts!

 

[511]

Tesla's CEO surrounds himself with a reality distortion field that Steve Jobs would have been jealous of. It isn't worth 1/10th of that market cap based on any of its current businesses. It is all fantasy about robotaxis and robots. Whatever people might say about Lip-Bu Tan, no one would suggest he's going to hype Intel to $1 trillion bubble cap by talking about imagined future successes like they've already happened.

Intel's all time market cap high was around $300B, set in 2000 during the heights of the dot com bubble. It came close to that a few times since but never matched it. During most of those 25 years Intel utterly dominated the desktop, laptop and server CPU market, and lead the world in process technology. So even if it did all those things it still wouldn't be worth close to $1 trillion, it would have to do more. Do more like taking a big chunk of the AI market - and do it before that bubble bursts!

Intel future hinges more on It's Fabs/Products than AI

 

[DrMrLordX]

I have to think it'll just find a different tech corner to dump into, so why not a domestic fab hype cycle?

AI is driving fab utilization. A pop of the AI bubble will reduce fab utilization and produce overcapacity on existing nodes, as well as those planned for the near future. Mobile and cloud growth is comparatively stagnant, meaning that neither sector is going to experience some sudden increase in demand for hardware that can fully absorb all the fab capacity currently committed to AI accelerators and associated hardware.

 

[Doug S]

Intel future hinges more on It's Fabs/Products than AI

Its future survival depends on its fabs/products, yes, but any fantasy about reaching $1 trillion market cap would require a bubble.

 

[511]

Getting Fab customer ougta be enough

 

[Henry swagger]

Tesla's CEO surrounds himself with a reality distortion field that Steve Jobs would have been jealous of. It isn't worth 1/10th of that market cap based on any of its current businesses. It is all fantasy about robotaxis and robots. Whatever people might say about Lip-Bu Tan, no one would suggest he's going to hype Intel to $1 trillion bubble cap by talking about imagined future successes like they've already happened.

Intel's all time market cap high was around $300B, set in 2000 during the heights of the dot com bubble. It came close to that a few times since but never matched it. During most of those 25 years Intel utterly dominated the desktop, laptop and server CPU market, and lead the world in process technology. So even if it did all those things it still wouldn't be worth close to $1 trillion, it would have to do more. Do more like taking a big chunk of the AI market - and do it before that bubble bursts!

Ai and good marketing brings value to wallstreet not sales and revenue

 

[Joe NYC]

AMD is going to up their orders for TSMC N2 capacity. Lots of TSMC N2 (Venice, Mi450) will go into the deal AMD signed with OpenAI

 

[marees]

AMD is going to up their orders for TSMC N2 capacity. Lots of TSMC N2 (Venice, Mi450) will go into the deal AMD signed with OpenAI

Is MI450 on 2nm ?

What is MI350 on ?

 

[511]

Is MI450 on 2nm ?

N2/N2P

What is MI350 on ?

N3P