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]

You cant extrapolate potential transistor switching speed gains above 5.7GHz by looking at gains achieved above 3.5GHz. The electrical and physical bottlenecks are completely different. Capacitance and inductance losses, not to mention heat, become exponentially more difficult to overcome with as you increase frequency and increase density. Its basic physics.

Yup Mobile vendors has room that they extracted just getting to 5Ghz require 18W they must had to jank voltage quite high to achieve this.

Oh no? Double shrink doesnt necessarily mean jack whatsoever for fmax. Im not trying to be right just for the sake of being right, Im being realistic. I dont expect 100MHz bump, I expect 300MHz-500MHz fmax bump. I just said I wouldnt be very surprised if 100MHz was all we got if we didnt have N2X.

Quintuple shrink
GloFo 32nm > TSMC N7: 5.0GHz (FX9590) > 4.7GHz* (3950X) 6% Loss

Double shrink
Intel 7 > TSMC N3B: 6.2GHz > 5.7GHz 500MHz 8% Loss

LNC was wider than RPC but -500 Mhz is like 8-9% frequency difference which totally destroyed the IPC improvement of LNC.

Single shrink
Intel 4 > TSMC N3B: 5.1GHz (185H) > 5.1GHz (288V) Equal

Same as above

Zero shrink
TSMC 7nm: 1.905GHz (5700XT) > 2.581GHz (6700XT) 35% Gain

Different uArch?
What would be fair comparison would be
GLC -> RWC. Intel 7 vs Intel 4
RPC -> RWC+ Intel 7+ vs Intel 3(Intel 4+)

 

[CouncilorIrissa]

Oh no? Double shrink doesnt necessarily mean jack whatsoever for fmax. Im not trying to be right just for the sake of being right, Im being realistic.

Quintuple shrink
GloFo 32nm > TSMC N7: 5.0GHz (FX9590) > 4.7GHz* (3950X) 6% Loss

Double shrink
Intel 7 > TSMC N3B: 6.2GHz > 5.7GHz 500MHz 8% Loss

Single shrink
Intel 4 > TSMC N3B: 5.1GHz (185H) > 5.1GHz (288V) Equal

Zero shrink
TSMC 7nm: 1.905GHz (5700XT) > 2.581GHz (6700XT) 35% Gain

The first example is comparing two CPUs that have next to nothing in common. Irrelevant.
2-3: Irrelevant again, you're comparing Intel's nodes to TSMC's as if the clock speed isn't the single redeeming feature about 10nm ESF/Intel 7. It's kind of Intel's whole shtick.
The fourth one is also irrelevant, because the only thing it proves that you can gain frequency via *uarch changes as well as process improvements*. RDNA1 is also a really poor example, considering the silicon was barely functional and designed on pennies by a skeleton crew.

You cant extrapolate potential transistor switching speed gains above 5.7GHz by looking at gains achieved above 3.5GHz. The electrical and physical bottlenecks are completely different. Capacitance and inductance losses, not to mention heat, become exponentially more difficult to overcome with as you increase frequency and increase density. Its basic physics.

And I never projected 30% gains seen in QC/Apple/Arm designs to translate 1:1 to higher clocked designs of AMD and Intel as I explicitly stated just a few posts ago.

I dont expect 100MHz bump, I expect 300MHz-500MHz fmax bump.

On that we agree at least, my expectation is around 8-10% increase.

 

[marees]

An industry official stated, "The production cost per wafer for the 2nm process exceeds $20,000, and Samsung Electronics likely offered a lower price to Tesla to enhance competitiveness." They added, "Amidst this, the 2nm process yield remains sluggish at 20%, indicating that Samsung is focusing on publicizing the reference externally, even if it means incurring a loss."

Samsung slashes 2nm prices by 33% to steal customers away from TSMC​

Chip designers can get 2nm capacity at Samsung for much less.

with production lines sitting empty due to a lack of orders, the company has to take some drastic steps to ensure a return on its investment.

Discounts may be necessary to drive up business​

According to reports, Samsung has decided to cut its 2nm wafer pricing down to $20,000, a 33% discount compared to TSMC's expected $30,000 per wafer price.

Samsung slashes 2nm prices by 33% to steal customers away from TSMC

Samsung foundry is reportedly undercutting TSMC's 2nm wafer pricing by $10,000 in a bid to drive up business.

www.sammobile.com

 

[511]

TSMC N2 is not 30K where are the clowns getting this info 😂😂

 

[jdubs03]

Oh wow I’m surprised no one has posted this (maybe it’s somewhere else and I missed it heh).

Intel stock pops on news company is in early talks to add AMD as a customer
https://www.cnbc.com/2025/10/01/intel-stock-amd-chips.html

 

[Josh128]

Oh wow I’m surprised no one has posted this (maybe it’s somewhere else and I missed it heh).

Intel stock pops on news company is in early talks to add AMD as a customer
https://www.cnbc.com/2025/10/01/intel-stock-amd-chips.html

Priced in at 2:00PM central time. All this hinges on Intel actually getting a working process.

 

[adroc_thurston]

All this hinges on Intel actually getting a working process.

Working is only part of the problem.
It has to be good.

 

[DavidC1]

Bulls**t. Show the numbers. Theres no way I'll believe that of the approximately 50% of the smartphone owners on the planet, more than 1 out of every 20 or 30 have an M3 Max system, unless you provide some kind of proof. I believe its significantly less than 1 out of every 100, closer to or even exceeding 1 out of 1000.

BS? What's BS about that? Why not actually search sales for yourself? Truth is greater than FEELINGS.

I gave you the numbers. It's your turn. Also, iWatch is a fraction of iPhone sales, so are iPads. Both combined are at most half of iPhone

 

[branch_suggestion]

Working is only part of the problem.
It has to be good.

How good does it need to be for a Mendo replacement?

 

[adroc_thurston]

How good does it need to be for a Mendo replacement?

Chromebooks are Very Dead so no replacement there.

 

[regen1]

Charlie's take on AMD fabbing @Intel rumors

Is AMD fabbing at Intel Foundry?

A few days ago, a financial note went around saying AMD is going to fab chips at Intel Foundry.

www.semiaccurate.com

Spoiler

A few days ago, a financial note went around saying AMD is going to fab chips at Intel Foundry. What has SemiAccurate heard about this deal and what does it mean?


Lets do this in the format of a Q and A session to save time.


Q: Is AMD fabbing at Intel?


A: TLDR is no.


Q: Longer answer?


A: No. No no no no, no <word unsuitable for a family site>ing way, no no no.


Q: How do you know?


A: I asked multiple people on both sides who know and everyone said it is BS. None even hinted that there is something in the works, all said the same hard no.


Q: So why did multiple financial oriented analysts and sites say it?


A: The best answer SemiAccurate has heard from multiple sources is that it appears that someone was trying to cover a short. We are not financial folk here so we can’t say but that was the opinion of several finance folk we talked to.


Q: Anything else?


A: SemiAccurate is saddened by the idiocy of the net, several of our peers, and the echo chamber who puked this BS back as fact without doing the merest shred of research, you know their job.


Q: And this surprises you why?


A: It doesn’t.


So there you have it, complete BS. SemiAccurate has covered Intel Foundry many times in the past, outed customers, and shot down rumors. Don’t believe random idiots who possibly have ulterior motives, it feeds the beast.S|A

 

[johnsonwax]

A: The best answer SemiAccurate has heard from multiple sources is that it appears that someone was trying to cover a short. We are not financial folk here so we can’t say but that was the opinion of several finance folk we talked to.

Back in the mid-aughts one of the better ways to make money was to listen to Jim Cramer on his CNBC show. He was also running a hedge fund at the time just like now, and right before options expiry if he floated a rumor, you could check where interest in options were at that time and based on whether the rumor was positive or negative, figure out what side of a trade he was trying to cover with the rumor and bet against that. I bought 3 cars off of that kind of thing. Apple was a favorite of his going way back, and he's only gotten more open in terms of pumping/dumping since then since clearly regulators don't give a shit that he's manipulating the market.

This sort of thing happens all the time and given how big a short target Intel was, I wouldn't find this the least bit surprising. SemiAccurate may be disappointed that the rumors get picked up, but good lord, this has been a thing for decades.

 

[Thibsie]

Oh wow I’m surprised no one has posted this (maybe it’s somewhere else and I missed it heh).

Intel stock pops on news company is in early talks to add AMD as a customer
https://www.cnbc.com/2025/10/01/intel-stock-amd-chips.html

Way too early for that to happen IMO.

 

[511]

Amd can use Intel for advanced packing but that's it they won't move over to fabricate so soon.

 

[Joe NYC]

Charlie's take on AMD fabbing @Intel rumors

Is AMD fabbing at Intel Foundry?

A few days ago, a financial note went around saying AMD is going to fab chips at Intel Foundry.

www.semiaccurate.com

Spoiler

A few days ago, a financial note went around saying AMD is going to fab chips at Intel Foundry. What has SemiAccurate heard about this deal and what does it mean?


Lets do this in the format of a Q and A session to save time.


Q: Is AMD fabbing at Intel?


A: TLDR is no.


Q: Longer answer?


A: No. No no no no, no <word unsuitable for a family site>ing way, no no no.


Q: How do you know?


A: I asked multiple people on both sides who know and everyone said it is BS. None even hinted that there is something in the works, all said the same hard no.


Q: So why did multiple financial oriented analysts and sites say it?


A: The best answer SemiAccurate has heard from multiple sources is that it appears that someone was trying to cover a short. We are not financial folk here so we can’t say but that was the opinion of several finance folk we talked to.


Q: Anything else?


A: SemiAccurate is saddened by the idiocy of the net, several of our peers, and the echo chamber who puked this BS back as fact without doing the merest shred of research, you know their job.


Q: And this surprises you why?


A: It doesn’t.


So there you have it, complete BS. SemiAccurate has covered Intel Foundry many times in the past, outed customers, and shot down rumors. Don’t believe random idiots who possibly have ulterior motives, it feeds the beast.S|A

There have been maybe 10-20 stories that were all false and smell of financial / stock manipulation.

Just as one story is proven wrong / denied, a new fake story pops up. Old fake story is immediately forgotten.

 

[adroc_thurston]

Amd can use Intel for advanced packing but that's it they won't move over to fabricate so soon.

I've no idea for what exactly since AMD (heart emoji) loves ASE/SPIL.

 

[Joe NYC]

Amd can use Intel for advanced packing but that's it they won't move over to fabricate so soon.

Intel may have something to offer that is just on par with what others offer.

I wonder, why would AMD go to Intel for something that is barely on par with others have to offer, at prices likely higher than what others have to offer.

Tariffs could be one explanation, but AFAIK, the biggest packaging facilities of Intel are in Malysia. Some in the US, many other scattered around the world.

 

[adroc_thurston]

Tariffs could be one explanation

US is not the world.

 

[Joe NYC]

US is not the world.

Exactly. If you follow the Intel fan delusions on Twitter, they seem to miss this fact.

For example, when Commerce secretary Lutnick says he would like to see 40% to 50% of US domestic semiconductor demand met from US based plants, their reaction is that everyone has to crawl on their knees to Intel, and Intel is going to $1 trillion market cap.

What they don't realize that the US is ~35% of global demand, and to reach 40% of this demand being met from US domestic fabs, TSMC would need to operate 14% of its capacity in the US.

And that is already part of the current TSMC plan, as far as its advanced nodes. Target is for 3 Phases (of 6) of Arizona fab to be completed in ~2028 and they will all ramp up to full capacity of 30k wafers per year. It can be done without Intel.

 

[adroc_thurston]

their reaction is that everyone has to crawl on their knees to Intel, and Intel is going to $1 trillion market cap.

Not even that, they're just doing random stuff for %reasons%.

 

[511]

Intel is going to $1 trillion market cap.

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

 

[511]

I've no idea for what exactly since AMD (heart emoji) loves ASE/SPIL.

you forgot AMKOR now that amkor does EMIB

 

[Thunder 57]

Exactly. If you follow the Intel fan delusions on Twitter, they seem to miss this fact.

For example, when Commerce secretary Lutnick says he would like to see 40% to 50% of US domestic semiconductor demand met from US based plans, their reaction is that everyone has to crawl on their knees to Intel, and Intel is going to $1 trillion market cap.

🤣 🤣 🤣 I doubt even the people on Reddit are that simple.

 

[adroc_thurston]

you forgot AMKOR now that amkor does EMIB

AMD hasn't done much at Amkor for eons doe.
They're very in love with ASE/SPIL, as I've said.

 

[511]

What they don't realize that the US is ~35% of global demand, and to reach 40% of this demand being met from US domestic fabs, TSMC would need to operate 14% of its capacity in the US.

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