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.

TSMC Radically Boosts CapEx to Expand Production Capacities, To Reach $14B For 2019

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.

Samsung’s Aggressive EUV Plans: 6nm Production in H2, 5nm & 4nm On Track

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]

Yeah but you have to understand it's not only US it's combination of multiple countries vs China

 

[DavidC1]

Yeah but you have to understand it's not only US it's combination of multiple countries vs China

When it comes to investments, US is a powerhouse because they are much more conservative in Asia/Europe. Also because it's multiple countries, there are lot more redundancies.

Note that rest of the world isn't under pressure to go another way either. China does because of the sanctions.

 

[Doug S]

• 

This assumes that there's only one way beyond DUV.

I guess you ignored the part where I said to assume this for the purposes of my question, despite it being the first sentence

Sure we can discuss what options China has besides the Rube Goldberg approach ASML took, but that's not what I was asking.

 

[511]

https://twitter.com/x/status/1935559937616470440

View attachment 126120

My post got wasted next time I am gonna post links only 😛

 

[jdubs03]

My post got wasted next time I am gonna post links only 😛

Heads up: it’s restricted.

 

[511]

Heads up: it’s restricted.

That's why I posted screenshot TLDR: Intel 4 vs 7 comparison was a A53 while Intel 3 to 18A comparison was A7XX core. I4 vs I3 is likely Redwood Cove

 

[Tigerick]

[News] Fujitsu Taps TSMC for 2nm CPU, But Flags Rapidus as Key to Supply Chain Diversification | TrendForce News

According to MoneyDJ, citing Nikkei, Fujitsu is developing a 2nm CPU named “MONAKA” based on designs from Arm. The chip is scheduled for commercializa...

www.trendforce.com

I am surprised TSMC is able to spare production of N2/N2P process to Fujitsu. Monaka should utilize 4 x 32-core = 144-core ARM server CPU. Before this news, it was rumored that Monaka is using Rapidus's 2nm process. Guess TSMC's Kungfu wins over Rapidus. The same situation happens to SF and IF; Samsung's Exynos 2600 is still struggling with SF2 process and Intel NVL's 8+16 compute tile is fabbed by N2 process with 4+8 compute tile being fabbed by Intel's 18-A process.

 

[511]

It's just mere 8+16 Compute tile/bLLC on N2 For Intel their entire product stack is on IFS Next year they are dual sourcing smartly like they should.

 

[igor_kavinski]

Next year they are dual sourcing smartly like they should.

I have yet to see a single smart move from them. They will still force people to eat hybrid cores. Either choose fat cores or the mont cores or make separate SKUs for both and let people choose. The only place where these hybrid cores need to exist is in laptops.

 

[Io Magnesso]

I have yet to see a single smart move from them. They will still force people to eat hybrid cores. Either choose fat cores or the mont cores or make separate SKUs for both and let people choose. The only place where these hybrid cores need to exist is in laptops.

It's difficult to make each of them because it's costly and costly. It's sad but...
It's hard to manage SKUs with different designs and completely different personalities…
I'm also interested in a desktop CPU that is only for E Core.

 

[511]

I have yet to see a single smart move from them. They will still force people to eat hybrid cores. Either choose fat cores or the mont cores or make separate SKUs for both and let people choose. The only place where these hybrid cores need to exist is in laptops.

You can say the same for AMD Zen6 it's an Hybrid Arch 😛.
Do you really think P and E core are issue with Skymont in ARL wait till you see Arctic Wolf 🤣.

 

[igor_kavinski]

You can say the same for AMD Zen6 it's an Hybrid Arch 😛.

Desktop Zen 6 is hybrid? Source?

On second thought, don't link to the source if you are talking about the 2 or 4 puny cores while the bigger cores are asleep. That's not similar to Intel's strategy of trying to save face with E-cores.

 

[511]

Desktop Zen 6 is hybrid? Source?

24C Z6 +2LPE are considered Hybrid 🤣

 

[MS_AT]

24C Z6 +2LPE are considered Hybrid

Strix Halo was supposed to be hybrid, didn't turn out to be true, so until we see something more tangible that these LP cores will be in and what is actually low power about them, I think it's too early to say. Btw, on my colleague's Meteor Lake, Windows is doing quite a good job of keeping the compilation workloads away from LPE cores.

 

[511]

Strix Halo was supposed to be hybrid, didn't turn out to be true, so until we see something more tangible that these LP cores will be in and what is actually low power about them, I think it's too early to say. Btw, on my colleague's Meteor Lake, Windows is doing quite a good job of keeping the compilation workloads away from LPE cores.

It's certainly there tbh same as 4LP-E cores there is no other way to lower the ideal/low load power scenarios.
The problem is games sometimes in rendering and other scenarios I never faced this problem.

 

[igor_kavinski]

24C Z6 +2LPE are considered Hybrid 🤣

Why can't Intel do the same? Instead of trying to compete with AMD, they should just release Darkmont CPU with 2 tiny Darkmont LPE cores. They shamefully released a crappy Lion Cove CPU but releasing a Darkmont CPU is a complete no no? Such political hypocrites they are.

 

[511]

Why can't Intel do the same? Instead of trying to compete with AMD, they should just release Darkmont CPU with 2 tiny Darkmont LPE cores. They shamefully released a crappy Lion Cove CPU but releasing a Darkmont CPU is a complete no no? Such political hypocrites they are.

They decided Unified Last year so add 4 years and we should get what you want in 2028 with Unified Core.

 

[MS_AT]

It's certainly there tbh same as 4LP-E cores there is no other way to lower the ideal/low load power scenarios.

Sure, it makes sense for mobile, but on desktop, if they go Halo interconnect route, the power savings of LP cores will be dwarfed by power supply inefficiencies. But remains to be seen I guess.

 

[Io Magnesso]

Sure, it makes sense for mobile, but on desktop, if they go Halo interconnect route, the power savings of LP cores will be dwarfed by power supply inefficiencies. But remains to be seen I guess.

It's true that there aren't many examples, and the benefits may not be as good as a laptop, but even on a desktop, LP-E Core may be useful when you're in sleep.

 

[511]

don't forget HX laptop shares design with Desktop where they can be useful you can pack 52 cores and battery life near a strix point laptop.

 

[branch_suggestion]

https://www.reuters.com/business/retail-consumer/intels-new-ceo-explores-big-shift-chip-manufacturing-business-2025-07-02/

Moving the goalposts again.

 

[Io Magnesso]

https://www.reuters.com/business/retail-consumer/intels-new-ceo-explores-big-shift-chip-manufacturing-business-2025-07-02/

Moving the goalposts again.

Wonderful

 

[Joe NYC]

https://www.reuters.com/business/retail-consumer/intels-new-ceo-explores-big-shift-chip-manufacturing-business-2025-07-02/

Moving the goalposts again.

It says Intel is contemplating to no longer offer 18A to external customers, which could lead to a partial write off of development costs of the node - ranging from 100s of millions to billions.

Also mentions that some analysts assess the 18A process to be on par with N3. Which would be N-1 node vs. TSMC N2, available at the same time.

The original Gelsinger promise / pitch was that 18A would leapfrog TSMC.

 

[Io Magnesso]

It says Intel is contemplating to no longer offer 18A to external customers, which could lead to a partial write off of development costs of the node - ranging from 100s of millions to billions.

Also mentions that some analysts assess the 18A process to be on par with N3. Which would be N-1 node vs. TSMC N2, available at the same time.

The original Gelsinger promise / pitch was that 18A would leapfrog TSMC.

That being said, I don't think it will stop providing 18A externally.
The existing contract itself will be fulfilled, and after 14A is released, I don't think 18A will be exempt from working.

 

[Io Magnesso]

That being said, I don't think it will stop providing 18A externally.
The existing contract itself will be fulfilled, and after 14A is released, I don't think 18A will be exempt from working.

In the first place, there is a considerable loss when you stop providing it to external customers.
It's a waste
It's true that there aren't many 18A customers, but quitting seems to have the opposite effect.