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

 

[mikk]

Yup it's hard to beat vaporware.

Both are vaporware, N3P and 18A. I highly doubt N3P will be competitive with 18A though.

 

[Ajay]

Well, Semi manufactures engaging in trash talking! LOLZ!

 

[DrMrLordX]

Both are vaporware, N3P and 18A. I highly doubt N3P will be competitive with 18A though.

N3P will be out before 18a, and N3P will hit high volume much quicker than 18a.

 

[Ajay]

N3P will be out before 18a, and N3P will hit high volume much quicker than 18a.

Yeah, but as to the rest, we will have to see. Unless TSMC has a spy inside Intel's R&D fabs.

 

[DrMrLordX]

Yeah, but as to the rest, we will have to see. Unless TSMC has a spy inside Intel's R&D fabs.

Who knows, they might. But yes, wait and see, I suppose. We won't have to wait that long to see N3P in action though. It's a bit sad that it'll take TSMC maybe two node revisions (N3B, N3E) to finally get N3 right (N3P), but in the end N3P should wind up being a pretty good node.

 

[H433x0n]

TSMC Places Confidence In 3nm Node, Compares It To Intel 18A: Better Maturity, Cost & Delivery Time

TSMC believes that it will retain its process leadership over Intel's 18A node with its 3nm process owing to better maturity & cost.

wccftech.com

Tsmc felling the pressure? .. no chance they can beat 18A with power via with 3nm finfet 🤔💻

I won’t comment on the performance characteristics of N3P vs 18A but the fact that CC Wei is commenting on it could be interesting. I don’t recall him ever doing that before with Intel or Samsung.

 

[H433x0n]

Who knows, they might. But yes, wait and see, I suppose. We won't have to wait that long to see N3P in action though. It's a bit sad that it'll take TSMC maybe two node revisions (N3B, N3E) to finally get N3 right (N3P), but in the end N3P should wind up being a pretty good node.

N3P isn't supposed to be production ready until 2H 2024 per Anandtech, which means we probably won't see products on it until 1Q 2025.

TSMC says that N3P will be production ready in the second half of 2024.

The reason why he's using N3P as the comparison is because they're both supposed to be hitting production at roughly the same time (assuming no unexpected issues arise) which makes it TSMC's direct competitor to 18A.

 

[Frenetic Pony]

N3P isn't supposed to be production ready until 2H 2024 per Anandtech, which means we probably won't see products on it until 1Q 2025.



The reason why he's using N3P as the comparison is because they're both supposed to be hitting production at roughly the same time (assuming no unexpected issues arise) which makes it TSMC's direct competitor to 18A.

The comparison is also ridiculous marketing speak. Nothing about being more performant, denser, or using less energy, you know the stuff that indicates progress and that people care about other than price. So it's just a blank faced nothing statement at best indicating nothing and at worst indicating TSMC knows it's going to be behind Intel at least for a bit and hoping to distract people from that up front.

 

[H433x0n]

The comparison is also ridiculous marketing speak. Nothing about being more performant, denser, or using less energy, you know the stuff that indicates progress and that people care about other than price. So it's just a blank faced nothing statement at best indicating nothing and at worst indicating TSMC knows it's going to be behind Intel at least for a bit and hoping to distract people from that up front.

Well, the term PPA encompasses the node characteristics (power, performance & area).

His wording outright says to me that N3P is behind since he phrased it like this:

N3P technology, demonstrated comparable PPA to 18A, my competitors' technology, but with an earlier time to market, better technology maturity, and much better cost.

While when he was talking about N2 he was much more confident and said this:

In fact, let me repeat again, our 2-nanometer technology without backside power is more advanced than both N3P and 18A

He used the term “comparable” when talking about N3P while emphasizing costs and tech maturity (not performance or process tech). His statement about it being available before 18A isn’t even correct since both are publicly scheduled to be manufacturing ready 2H 2024. If N3P was a superior node he would say that outright like he does when he speaks about N2. Even with his claims about N2 it’s pretty vague but he’s at least more aggressive with the language.

To me, his statements backfired since the tech press and investors view them as clear market leaders. To say that your upcoming major node is just “comparable” isn’t a good look.

 

[DrMrLordX]

N3P isn't supposed to be production ready until 2H 2024 per Anandtech, which means we probably won't see products on it until 1Q 2025.

That changes nothing. It'll be earlier to market than 18a, especially when it comes to production at significant volume. And that's saying something, considering it's an actual foundry node that will be available to multiple customers in this timeframe. 18a is supposed to be Intel's second major foundry node (first being Intel 3), but I severely doubt they'll be ready for many customers by 2H 2024 on that node. Intel's first Intel 3 product will be hitting shelves in Q2 2024 if they hold to schedule (BIG IF) and that's an internal product. N3P will be available to foundry customers in about the same timeframe that we should reasonably expect Intel 3 to be made available in the same fashion.

 

[H433x0n]

That changes nothing. It'll be earlier to market than 18a, especially when it comes to production at significant volume. And that's saying something, considering it's an actual foundry node that will be available to multiple customers in this timeframe. 18a is supposed to be Intel's second major foundry node (first being Intel 3), but I severely doubt they'll be ready for many customers by 2H 2024 on that node.

No, it's incredibly unlikely there will be any products in 2H 2024 on 18A. It's supposed to be manufacturing ready by that time, so there should be a 1 or 2 quarter lag between manufacturing ready and customers getting their wafers.

Intel's first Intel 3 product will be hitting shelves in Q2 2024 if they hold to schedule (BIG IF) and that's an internal product. N3P will be available to foundry customers in about the same timeframe that we should reasonably expect Intel 3 to be made available in the same fashion.

That's not what that article says. Per TSMC it's scheduled to be "manufacturing ready" in 2H 2024. This means best case scenario there will be products on shelves Q1 2025. You may be thinking of TSMC N3E, which is supposed to be manufacturing ready 2H 2023, so in that case it'd be feasible for it to be out in products ~Q1 / Q2 2024.

 

[lightisgood]

Intel's first Intel 3 product will be hitting shelves in Q2 2024 if they hold to schedule (BIG IF) and that's an internal product.

You may be thinking of TSMC N3E, which is supposed to be manufacturing ready 2H 2023, so in that case it'd be feasible for it to be out in products ~Q1 / Q2 2024.

N3E has not entered to HVM yet.
At Intel's 3Q23 earning, I would like to see if Intel 3 has entered to HVM.
Anyway, there are bit different release dates between Intel 3 and N3E...

Dr. C. C. Wei
N3E has passed qualification and achieved performance and yield targets and will start volume production in fourth quarter of this year.

 

[DrMrLordX]

No, it's incredibly unlikely there will be any products in 2H 2024 on 18A.

You're right on that. Intel will have Intel 3, not 20a or 18a ready in 2H 2024. And again that's for internal products which are the pipecleaners and first to receive wafers from that node. It's unclear when 18a will be ready for IFS customers, but it should be somewhere in the 2025/2026 timeframe.

That's not what that article says. Per TSMC it's scheduled to be "manufacturing ready" in 2H 2024.

There's no way 18a products will be going into production in that timeframe. That's when 20a will be ramping up for a Q4 2024 launch. No, I'm not thinking of N3E.

 

[mikk]

N3P will be out before 18a, and N3P will hit high volume much quicker than 18a.

You can't know this because both are vaporware, we don't even know what chips will be first on N3P. Dismissing unavailable nodes from Intel with vaporware and hyping unavailable TSMC nodes with will be is a biased one. N3B started HVM in late Q4 2022 and first chips came to market in late Q3 2023. Exact timeframes for N3E and N3P are not known yet. And even if N3P comes 2Q before 18A, so what? I'm not even arguing against it, I just have doubts that N3P can compete against 18A in performance (density/efficiency/speed). The big message is that N2 is definitely coming later than 18A, probably much later considering that TSMC avoids talking about it. Intel seems to deal with GAA and backside power much earlier than TSMC.

 

[igor_kavinski]

I think TSMC would be a better bet at this point than Intel. TSMC at least has products on a more advanced node selling RIGHT NOW.

 

[Tigerick]

You can't know this because both are vaporware, we don't even know what chips will be first on N3P. Dismissing unavailable nodes from Intel with vaporware and hyping unavailable TSMC nodes with will be is a biased one. N3B started HVM in late Q4 2022 and first chips came to market in late Q3 2023. Exact timeframes for N3E and N3P are not known yet. And even if N3P comes 2Q before 18A, so what? I'm not even arguing against it, I just have doubts that N3P can compete against 18A in performance (density/efficiency/speed). The big message is that N2 is definitely coming later than 18A, probably much later considering that TSMC avoids talking about it. Intel seems to deal with GAA and backside power much earlier than TSMC.

Intel "seems" also mean nothing if there are no products coming out of it. Intel also claimed unquestioned leadership with 18A process and High-NA machines. Now that Intel has delayed usage of High-NA, do we still believe the PPA claims of 18A? To me 18A should be renamed to 20A+ because it seems there won't be much advancement than 20A.

Samsung also claimed they are first to use GAA, but where is the product from old partners like nVidia and Qualcomm? In fact, we should be seeing first wave of TSMC's N4P process products, namely mobile SoC from Qualcomm and Mediatek. They are shipping in volume as flagship Android headsets.

 

[mikk]

Intel "seems" also mean nothing if there are no products coming out of it. Intel also claimed unquestioned leadership with 18A process and High-NA machines. Now that Intel has delayed usage of High-NA, do we still believe the PPA claims of 18A? To me 18A should be renamed to 20A+ because it seems there won't be much advancement than 20A.

What PPA claims exactly? What do you mean with it seems? Do you have a source for anything? High-NA was not an option for 18A since early 2022 if not earlier: https://www.intel.com/content/dam/w.../2022-intel-investor-meeting-process-tech.pdf

Since then nothing changed in any of the public figures, there wasn't even High-NA in the first announcement of 18A. You know nothing about 18A, so why do you say they should rename it to 20+? This is such a big nonsense, maybe ask TSMC for renaming some of their 5nm and 3nm nodes to a + instead of using 4nm and N3x because the advancement won't be big.

 

[DrMrLordX]

You can't know this because both are vaporware

TSMC would have to delay N3P to late 2025 for that to be true.

 

[Doug S]

You can't know this because both are vaporware, we don't even know what chips will be first on N3P. Dismissing unavailable nodes from Intel with vaporware and hyping unavailable TSMC nodes with will be is a biased one. N3B started HVM in late Q4 2022 and first chips came to market in late Q3 2023. Exact timeframes for N3E and N3P are not known yet. And even if N3P comes 2Q before 18A, so what? I'm not even arguing against it, I just have doubts that N3P can compete against 18A in performance (density/efficiency/speed). The big message is that N2 is definitely coming later than 18A, probably much later considering that TSMC avoids talking about it. Intel seems to deal with GAA and backside power much earlier than TSMC.

N3P is a tweak of N3E which is entering mass production any day now. 18A is a tweak of a FUTURE Intel node. Claiming N3P and 18A have equal "vaporware" status is like claiming Meteor Lake and Apple M4 have equal vaporware status because neither are shipping today.

N3B had a very long "HVM" to first chips because it is a very problematic node and it appears no one except Apple was willing to deal with it. N3E will not have a similar wait, we won't have to wait for A18 to see the first N3E chips. There will be many other N3E chips out months in advance of that (likely including M3)

 

[jpiniero]

N3B had a very long "HVM" to first chips because it is a very problematic node and it appears no one except Apple was willing to deal with it. N3E will not have a similar wait, we won't have to wait for A18 to see the first N3E chips. There will be many other N3E chips out months in advance of that (likely including M3)

Yeah. Everyone is using N3E. You could say though that since it sounds like the original N2 is going to be very unpopular like N3B... that most products will still be using N3 even well into 2026.

 

[DrMrLordX]

Yeah. Everyone is using N3E. You could say though that since it sounds like the original N2 is going to be very unpopular like N3B... that most products will still be using N3 even well into 2026.

N3P will likely have legs, sort-of like N4P.

 

[mikk]

TSMC would have to delay N3P to late 2025 for that to be true.

Nonsense. If you call an upcoming Intel node vaporware you have to call an upcoming TSMC node vaporware as well unless you are biased. You don't like this right because TSMC is the good guy and you hate Intel right? I got it.

 

[Doug S]

Yeah. Everyone is using N3E. You could say though that since it sounds like the original N2 is going to be very unpopular like N3B... that most products will still be using N3 even well into 2026.

How can you tell who is going to be using N2 now? Obviously N3 will still be heavily used once N2 is out just like N7 was still heavily used well after N5 was out.

 

[jpiniero]

How can you tell who is going to be using N2 now? Obviously N3 will still be heavily used once N2 is out just like N7 was still heavily used well after N5 was out.

That's the buzz. Most everyone will wait for N2P.

 

[H433x0n]

Nonsense. If you call an upcoming Intel node vaporware you have to call an upcoming TSMC node vaporware as well unless you are biased. You don't like this right because TSMC is the good guy and you hate Intel right? I got it.

The weirdest thing I’ve noticed on the tech enthusiast internet is the pro TSMC monopoly stance.

The current situation is arguably horrible for both tech advancement and geopolitics. I imagine some of it is people that dislike Intel and view TSMC as an extension of AMD.