Discussion Leading Edge Foundry Node advances (TSMC, Samsung Foundry, Intel) - [2020 - 2025]

Page 238 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
DisEnchantment

DisEnchantment

Golden Member
Mar 3, 2017
1,779
6,798
136
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.

1587737990547.png
N7 performance is more or less understood.
1587739093721.png

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 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 www.anandtech.com

1587739615344.png

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
 
Last edited:
  • Like
Reactions: Smithad49, Io Magnesso, Kaluan and 6 others
J

johnsonwax

Senior member
Jun 27, 2024
472
676
96
MoogleW said:
Benefit is a measly 20% density gains come on
Click to expand...
Considering how big an effect the AI datacenter are having on power grids and pricing, it's nothing to sneeze at. The only thing harder than securing leading node capacity right now is getting grid transformers, and they might be willing to pay a lot to get the former to avoid the shortages in the latter.
 
R

regen1

Senior member
Aug 28, 2025
363
456
96

BEIJING, Oct 9 (Reuters) - China's commerce ministry added 14 foreign organisations to its "unreliable entity list", it said in a statement on Thursday, restricting their ability to carry out commercial activities within the world's second-largest economy.
Click to expand...
TechInsights, a prominent Canadian tech research firm, and nine of its subsidiaries including Strategy Analytics were among those blacklisted.
Last October, TechInsights took apart a Huawei AI processor and found a chip made by Taiwan Semiconductor Manufacturing Company (TSMC) in it.
The U.S. had imposed export rules on the Shenzhen-based Huawei in 2020 to halt shipments of foreign-produced items to Huawei that are the direct product of U.S. technology or software, including TSMC's chips
Click to expand...
TechInsights added to the list for revealing Huawei using TSMC lol ??
 
Last edited:
  • Haha
Reactions: Elfear, DZero, Doug S and 1 other person
5

511

Diamond Member
Jul 12, 2024
5,452
4,879
106
people miss out on the lower power at the same performance part why have 20% more performance when you can lower power by 30-40% and use more compute.
 
  • Like
Reactions: DKR
poke01

poke01

Diamond Member
Mar 8, 2022
4,872
6,202
106
Will TSMC give in?

Apparently Sam Altman is pressuring TSMC to give up fab space for his “AI chip”
 
Josh128

Josh128

Golden Member
Oct 14, 2022
1,542
2,295
106
marees said:
18a got a customer ?

https://twitter.com/x/status/1979221349307396510
Click to expand...
Thats pretty damned big news if its a new development. Also, Charlie D. is a clown.

**EDIT: That article is old af. Its from 2/2024 and says that 18A will commence production at the end of 2024, lol. We still dont have the first FIRST PARTY chip out of it. MS likely cancelled that contract if thats what it was led to believe.
 
  • Like
Reactions: marees
J

Joe NYC

Diamond Member
Jun 26, 2021
4,216
5,824
136
Patrick Moorehead and Dulan Patel both said on Twitter that the story (of Charlie D. Retard) is false.
 
  • Like
Reactions: Grazick
R

regen1

Senior member
Aug 28, 2025
363
456
96
Microsoft reportedly have their share of issues with design challenges regarding MAIA 2(200) and may be more. Even if something of MAIA(200?, 280?, 300?)was to come on 18A family it won't be likely before H2 2027.
 
H

Hitman928

Diamond Member
Apr 15, 2012
6,754
12,500
136
LBT's comments on Intel's 18a yields show again that he is willing to be transparent where Pat would smoke screen at best, if not simply lie knowing that yields weren't great and hope the engineers could make miracles happen to not expose him.

"However, the ramp-up of 18A processors will be slow due to relatively low yields, and Intel will not expand available 18A capacity at a rapid pace. Currently, 18A's yields are not yet comfortable from a commercial point of view."

 
  • Like
Reactions: Tlh97, adamge, Elfear and 2 others
DrMrLordX

DrMrLordX

Lifer
Apr 27, 2000
23,223
13,301
136
@Hitman928

Looks like he's signaling that 18a will never have good yields and that 18ap will (since 18ap should be a 2027 node). I could be misinterpreting this, but once 18ap is out, will anyone bother with 18a anyway?

edit: okay it was Anton Shilov saying those things soooo whatever.
 
Last edited:
R

regen1

Senior member
Aug 28, 2025
363
456
96
Hitman928 said:
LBT's comments on Intel's 18a yields show again that he is willing to be transparent where Pat would smoke screen at best, if not simply lie knowing that yields weren't great and hope the engineers could make miracles happen to not expose him.

"However, the ramp-up of 18A processors will be slow due to relatively low yields, and Intel will not expand available 18A capacity at a rapid pace. Currently, 18A's yields are not yet comfortable from a commercial point of view."

Click to expand...
That's not LBT's comments, that's a remark from Anton Shilov in that article.

LBT's comment from the Tomshardware article:
"We are making steady progress on Intel 18A, we are on track to bring Panther Lake to market this year," said Lip-Bu Tan, chief executive of Intel. "18A yields are progressing at a predictable rate, and Fab 52 in Arizona, which is dedicated to high-volume manufacturing, is now fully operational."
Click to expand...

Intel CFO's comments:
"Yields are adequate to address supply, but they are not where we need them to be in order to drive the appropriate level of margins," said Zinsner. "By the end of next year we will probably be in that space, and certainly the year after that they will be at what would be an industry-acceptable level."
Click to expand...

And to put into context for the impact of a new node development on margin let's see TSMC's earnings call

Excerpt from TSMC Q3 Earnings Call:
Okay. Sunny, yes, it’s too early to talk about 2026. But you already mentioned about the N2 dilution. And as all the new node when they just come out, the N2 will have dilution in our gross margin in 2026. But at the same time, the N3 dilution is gradually coming down and we expect the N3 to catch up to the corporate average sometimes in 2026.
Click to expand...
Source: https://www.investing.com/news/tran...2025-shows-strong-revenue-growth-93CH-4291402

N3 family came up in 2023 just for context.

Various issues can come up during the initial phase of new node. This is not to say the issues at TSMC and Intel are exactly similar. But we need to give more time to assess the impact of the node. Anyway it seems the roadmaps of nodes and some of R&D have been reduced under LBT(that could also have impact(?)).
18A family is going to be a long node. Almost every product line esp. CPUs(server and client) and perhaps more products will have lots of 18A family utilization. Intel 10nm/7 is a margin destroyer as well. Ideally for them they need to gradually move from Intel 10nm/7 family towards their better nodes while the metrics of their nodes also progressing well.
 
  • Like
Reactions: pcp7, Tlh97, Elfear and 5 others
K

Khato

Golden Member
Jul 15, 2001
1,381
491
136
Yeah, current 18A PTL yields would maintain or slightly drag down gross margin rather than being in the >50% 'appropriate' level. And yields only got to the 'maintain' gross margin point recently. Just have to wait and see whether it takes a year of small iterative steps to improve yields or if some further breakthroughs occur.

It is nice to see the management messaging change to expecting it to be a year of small iterative changes.
 
R

regen1

Senior member
Aug 28, 2025
363
456
96
DrMrLordX said:
@Hitman928

Looks like he's signaling that 18a will never have good yields and that 18ap will (since 18ap should be a 2027 node). I could be misinterpreting this, but once 18ap is out, will anyone bother with 18a anyway?
Click to expand...
Don't think anything of that kind is getting signaled.
18A to 18A-P is supposed to be about a year of gap wrt HVM(may be a bit here and there) and they aren't that fundamentally different plus they are design rule compatible. Most of the learning and progress/improvements(including yields) on 18A should also go into 18A-P.
18A-P has some tweaks and progresses(which nets more perf/W and more performance range) plus has more IP libraries for various categories.
With time the PDK also should progress for 18A-P from where it was from 18A(but may be not as well(?) as it should have considering the reports of halting development of IPs with Synopsys on 18A? and some possible node related R&D cuts(?)).
In theory once 18A-P is out in full-scale most of the newer products should use it instead of 18A which is natural. It was the same with previous Intel's node progressions(+ nodes/SF/ESF/Ultra etc.).
 
Last edited:
  • Like
Reactions: DKR
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom