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

Page 61 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

DisEnchantment

Golden Member
Mar 3, 2017
1,777
6,778
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.

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.


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:

NostaSeronx

Diamond Member
Sep 18, 2011
3,808
1,289
136
I thought NIL had major issues with defect rates in any kind of production quantity throughput. Have those issues been solved? Is anyone using NIL in production today?
Defect density is currently low enough to do logic starts for older nodes.
defectdensity.jpeg
hp2xnm ~ 40~60nm metal pitch

This is good enough to start, 0.01 is needed for HVM w/ end process D0 being 0.3 less like TSMC Risk production.

3nm/2.1/1.5 hasn't been validated for defect density yet, which is the end target not the start target//
defectivitychallenge.jpg
canonnilpower.jpeg
The above is heavily pushing for 3nm (12nm hp etaby:2024) ~ 1.5nm (8nm hp etaby:2028-onwards)

Older nodes are good enough with larger than 20nm half-pitch being near 0.01. Of which, the same tool can be used for much later nodes.
12FDX HP-193i = metal1-metal4 is 56-nm pitch
12FDX HP-NIL = metal1-metal9 is 56-nm pitch
9FDX NIL = metal1-metal9 is 48-nm pitch
6FDX NIL = metal1-metal9 is 40-nm pitch
potental plus node here for 32 and 36-nm pitch from 6FDX
3FDX NIL(not 3nm logic node, but rather 5nm logic node) = metal1-metal9 is 28-nm pitch. // Adding ULK pores and insulation is very achievable with NIL.

The tool for HVM won't start full production till 2025 (Canon Japan[for china/japan] and Canon Texas[for america/europe]). NZ2C != HVM tool, so it isn't being used in anything that leads to products.

NZ2C maxed out is >200 wph, and average max speed is ~160wph.
The next tool skips over the NZ3C(6-cluster machine) goes for a 8-cluster machine, with a maxed out capacity of >400 wph, average max speed of ~320wph.
 
Last edited:

Dayman1225

Golden Member
Aug 14, 2017
1,160
996
146
I think its somebody else and not Ericson.


Seems like a big company...Nvidia?
Yeah I don’t think the customer Pat is talking about is Ericsson because Pat was being cagey with the name in the full interview, whereas Ericsson already announced they would be using 18a, don’t think it’s nvidia though.
 

Doug S

Diamond Member
Feb 8, 2020
3,198
5,490
136
dod has contracts with many american fabs not just intel.dod was using tsmc taiwan for some sensitive chips beforehand.

The DoD is big. There are some things that aren't that sensitive, others which are classified technology and could not be made outside the US or by a foreign owned entity.
 

A///

Diamond Member
Feb 24, 2017
4,351
3,160
136
The DoD is big. There are some things that aren't that sensitive, others which are classified technology and could not be made outside the US or by a foreign owned entity.
True, I know TSMC makes chips that go into the f35lii. iirc Bosch is contracted for some stuff that's low on the importance chain.
 

A///

Diamond Member
Feb 24, 2017
4,351
3,160
136
Did I miss something?

Also per your other remark, the DoD has used GlobalFoundries as well.
Yep gloflo, I didn't include them initially because I could not find an older article stating that other than a newer contract being signed.

exist's ability to post on the forum was cut short according to others.
 

Ajay

Lifer
Jan 8, 2001
16,094
8,111
136
Yep gloflo, I didn't include them initially because I could not find an older article stating that other than a newer contract being signed.

exist's ability to post on the forum was cut short according to others.
Intel also used to make radiation hardened CPUs for the DoD - but I don't know if that's still the case.
 

A///

Diamond Member
Feb 24, 2017
4,351
3,160
136
Intel also used to make radiation hardened CPUs for the DoD - but I don't know if that's still the case.
I did not know this! Very interesting. Do you have any reading material on that you could share?

Still curious about that mystery. His recent post history didn't seem to reveal anything out of the ordinary. He must have had a PM war with someone, probably someone with "powers".
who knows. nothing of value was lost.
 

A///

Diamond Member
Feb 24, 2017
4,351
3,160
136
That's not fair. He was one of the most knowledgeable members on the forum and it seemed like he worked in the industry. I think his contributions to the forum are a value add.
There's plenty of industry people on this website. Big whoop. This is as much as I'll discuss because the rules forbid talking about bans or suspensions or people with a habit of showing off their egregious behaviour.
 

eek2121

Diamond Member
Aug 2, 2005
3,370
4,987
136
I read a headline (didn’t click it) that said China was stockpiling EUV machines so if that is true I would’t bet against them using EUV in the future.

Yeah I don’t think the customer Pat is talking about is Ericsson because Pat was being cagey with the name in the full interview, whereas Ericsson already announced they would be using 18a, don’t think it’s nvidia though.
There was a recent slide posted earlier that indicated it was Ericsson. It also still has Arrow Lake under 20A, so sorry all you folks mentally hoping Arrow Lake will be on TSMC N3.
 

Ajay

Lifer
Jan 8, 2001
16,094
8,111
136
I read a headline (didn’t click it) that said China was stockpiling EUV machines so if that is true I would’t bet against them using EUV in the future.


There was a recent slide posted earlier that indicated it was Ericsson. It also still has Arrow Lake under 20A, so sorry all you folks mentally hoping Arrow Lake will be on TSMC N3.
They are DUV machines - which are about to be cut off because of new trade restrictions.
 

A///

Diamond Member
Feb 24, 2017
4,351
3,160
136
I read a headline (didn’t click it) that said China was stockpiling EUV machines so if that is true I would’t bet against them using EUV in the future.
twinscan nxt 2000. It does 7nm and 5nm but they're getting the 7nm version until January first of 2024. I saw some headlines of smic making headways a month ago but nothing concerete. us gov has now banned nvidia from selling their h100 and such to middle eastern countries including allies in fears it's a resell point for China. Saudi arabia and similar have been working deeply with china on ai for a while and the fear is legitimate. although I'm sure if china wants to get their hands on the hardware they'll find a way.
 
  • Like
Reactions: Tlh97

Dayman1225

Golden Member
Aug 14, 2017
1,160
996
146
There was a recent slide posted earlier that indicated it was Ericsson. It also still has Arrow Lake under 20A, so sorry all you folks mentally hoping Arrow Lake will be on TSMC N3.
I don’t think it’s Ericsson that Pat was talking about in that interview because he wouldn’t say the name of the prepay customers, whereas Ericsson is already a public customer, he would have said Ericsson if it was imo.
 

Saylick

Diamond Member
Sep 10, 2012
3,891
9,036
136
I read a headline (didn’t click it) that said China was stockpiling EUV machines so if that is true I would’t bet against them using EUV in the future.


There was a recent slide posted earlier that indicated it was Ericsson. It also still has Arrow Lake under 20A, so sorry all you folks mentally hoping Arrow Lake will be on TSMC N3.
I thought China wasn't allowed to even purchase EUV machines from ASML? The machines they might be stockpiling are likely older DUV ones, the ones that are good for 7nm-class chips.