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

DisEnchantment · Apr 24, 2020

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 · Sunday at 12:11 PM

SiliconFly said:
18A is a 2025 node and shouldn't be compared to N2, considering N2 is a disaster this year. N2 doesn't have any significant products from noteworthy leading-edge high-volume customers like Apple, QC, etc this year.

Since Apple is set to use N2 next year, N2 compete with 18A-P next year. Both N2 & 18A-P have similar densities with 18A-P having higher performance than N2. 18A-P is clearly a superior node compared to N2.

N2 was never a H2 25 node and the tape outs for it were in November-December and Q1 25

Geddagod · Sunday at 12:35 PM

SiliconFly said:
18A is a 2025 node and shouldn't be compared to N2, considering N2 is a disaster this year. N2 doesn't have any significant products from noteworthy leading-edge high-volume customers like Apple, QC, etc this year.

Since Apple is set to use N2 next year, N2 compete with 18A-P next year. Both N2 & 18A-P have similar densities with 18A-P having higher performance than N2. 18A-P is clearly a superior node compared to N2.

I'm genuinely curious, if 18A-P is clearly a superior node vs N2, why do you think Intel would be using what is very likely N2 for the compute tiles in NVL?

jpiniero · Sunday at 12:50 PM

Geddagod said:
I'm genuinely curious, if 18A-P is clearly a superior node vs N2, why do you think Intel would be using what is very likely N2 for the compute tiles in NVL?

Intel is too poor to afford the needed capacity regardless of what they think of 18A.

Meteor Late · Sunday at 12:52 PM

Geddagod said:
I'm genuinely curious, if 18A-P is clearly a superior node vs N2, why do you think Intel would be using what is very likely N2 for the compute tiles in NVL?

Of course, because it is not a superior node.

Hitman928 · Sunday at 12:59 PM

jpiniero said:
Intel is too poor to afford the needed capacity regardless of what they think of 18A.

If this is true, why are they pitching as many external customers as they can to use the node?

jpiniero · Sunday at 1:04 PM

Hitman928 said:
If this is true, why are they pitching as many external customers as they can to use the node?

Sell the idea to Wall Street they are trying?

Hitman928 · Sunday at 1:19 PM

jpiniero said:
Sell the idea to Wall Street they are trying?

No, they are very actively working with potential customers and trying to sell space on the node, they just haven’t been able to convince any major customers yet. That’s why they paused their fab buildouts, they didn’t get the customer interest they were expecting.

SiliconFly · Sunday at 1:23 PM

Geddagod said:
I'm genuinely curious, if 18A-P is clearly a superior node vs N2, why do you think Intel would be using what is very likely N2 for the compute tiles in NVL?

Like they said recently, way too many products are on 18A/18A-P. PNL, NVL, CWF, DNR, etc. They just don't have enough capacity. External is not just an option, it's a necessity.

SiliconFly · Sunday at 1:27 PM

Hitman928 said:
If this is true, why are they pitching as many external customers as they can to use the node?

They don't have a choice. If IFS has to survive, they need to prioritize external customers. In fact, I think its a massive step they've taken trying to prioritize external over their own to a certain degree. Will help them in the long run (exp. when their own products shift to the next one and frees up capacity).

Geddagod · Sunday at 1:29 PM

SiliconFly said:
Like they said recently, way too many products are on 18A/18A-P. PNL, NVL, CWF, DNR, etc. They just don't have enough capacity. External is not just an option, it's a necessity.

PTL should get phased out by NVL.
CWF is even according to Intel a pretty niche product.
DMR is the other major product, but I mean, 18A would have had a year to ramp by then, and Intel is fabbing Intel 4/3 MTL+GNR+SRF currently, and as we see with the wafer capacity graph, the amount of Intel 3/4 volume they have rn is way less than the amount of Intel 18A/20A volume they have projected themselves to have rn. And the situation only improves later on.

SiliconFly · Sunday at 1:42 PM

Geddagod said:
PTL should get phased out by NVL.
CWF is even according to Intel a pretty niche product.
DMR is the other major product, but I mean, 18A would have had a year to ramp by then, and Intel is fabbing Intel 4/3 MTL+GNR+SRF currently, and as we see with the wafer capacity graph, the amount of Intel 3/4 volume they have rn is way less than the amount of Intel 18A/20A volume they have projected themselves to have rn. And the situation only improves later on.
View attachment 116151

EUV shortage is still a thing for them. Unfortunately, they can't phase out Intel 3 early and repurpose the EUVs and other tools, cos it appears most of their products are set to use Intel 3 base tiles.

Makes me wonder, why do they even need Intel 3 base tiles? Why can't they use something like TSMC N6 base tiles that'll workout a lot cheaper and will also free up some of their EUVs for more advanced nodes. Compatibility issues with Foveros? Just sayin'.

Thibsie · Sunday at 3:07 PM

Hitman928 said:
No, they are very actively working with potential customers and trying to sell space on the node, they just haven’t been able to convince any major customers yet. That’s why they paused their fab buildouts, they didn’t get the customer interest they were expecting.

Yeah :
- 'We don't use our own node ourself but I swear to you it's pretty good
- ' yeah, right....'

OneEng2 · Sunday at 8:52 PM

511 said:
There is no issues with N2 yields but they are not as high as 18A from the presentation and Public data that was last shared for SRAM and Intel shared D0 in August

Aren't they? Not that TSMC has been that forthcoming with details of how that yield number they threw out was calculated, but neither has Intel.

Generally speaking, I trust the marketing crap from TSMC more than I trust the marketing crap from Intel.

511 said:
Someone leaked node rumors

https://twitter.com/x/status/1886038282569744394

We will see by the end of the year ..... OR NOT! Wasn't very long ago that people in this forum were speculating (based on Intel charts and statements) that CWF would be an early Q3 2025 release. BSPDN, GAA, and a completely new process all in one step.... and Intel to execute it perfectly?

I don't know guys. I don't think I would bet my house on this one.

SiliconFly said:
18A is a 2025 node and shouldn't be compared to N2, considering N2 is a disaster this year. N2 doesn't have any significant products from noteworthy leading-edge high-volume customers like Apple, QC, etc this year.

I think that 18A yields and process aren't doing that well. This is why CWF was canned for 2025 as it would have required large tiles to be produced on 18A (my SWAG). N2 on the other hand isn't being promised for 2025 product release AFAIK.

Thibsie said:
Yeah :
- 'We don't use our own node ourself but I swear to you it's pretty good
- ' yeah, right....'

I feel this way as well.

Fool me once? shame on you.
Fool me twice? shame on me.
Fool me three times? Man, am I naive.
Fool me four times? Ok, I'm an idiot.

When was the last time we were fooled again?

dttprofessor · Monday at 12:12 PM

SiliconFly said:
EUV shortage is still a thing for them. Unfortunately, they can't phase out Intel 3 early and repurpose the EUVs and other tools, cos it appears most of their products are set to use Intel 3 base tiles.

Makes me wonder, why do they even need Intel 3 base tiles? Why can't they use something like TSMC N6 base tiles that'll workout a lot cheaper and will also free up some of their EUVs for more advanced nodes. Compatibility issues with Foveros? Just sayin'.

This is too exaggerated. Intel7 is still operating in large quantities, turning off Intel3?

DavidC1 · Monday at 4:46 PM

@OneEng2 Clearwater Forest is delayed due to packaging issues. Either way, it doesn't look good, because delays make the product WORSE, never better. Delays happen because goals are missed and they are trying to un-miss it.

People around the Net are making an argument that tariffs and Chinese invasion will save Intel, but the reality is a serious war like that will plummet the demand for luxury goods such as computers, and will further the decline of Intel. It's an astonishing level of denial if you ask me. The world is at a precipice for potential end and somehow that is spun as positive.

If you think Intel's revenue level is bad, imagine when China blocks sales entirely. That alone will result in total revenue(not net) dropping 20% or so. This is assuming rest of the world is insulated in a vacuum, which won't be. Likely the total drop in revenue is going to be closer to 50%. That's bankruptcy in <3 years.

OneEng2 · Monday at 5:40 PM

DavidC1 said:
@OneEng2 Clearwater Forest is delayed due to packaging issues. Either way, it doesn't look good, because delays make the product WORSE, never better. Delays happen because goals are missed and they are trying to un-miss it.

People around the Net are making an argument that tariffs and Chinese invasion will save Intel, but the reality is a serious war like that will plummet the demand for luxury goods such as computers, and will further the decline of Intel. It's an astonishing level of denial if you ask me. The world is at a precipice for potential end and somehow that is spun as positive.

If you think Intel's revenue level is bad, imagine when China blocks sales entirely. That alone will result in total revenue(not net) dropping 20% or so. This is assuming rest of the world is insulated in a vacuum, which won't be. Likely the total drop in revenue is going to be closer to 50%. That's bankruptcy in <3 years.

Seems like CWF had a window to take back leadership in performance (although I still question the profitability of the Intel vertical integration model). Slip this until mid 26 and it will soon be up against an AMD Zen 6 part on either N3P or N2 (or N3P for the high performance version and N2 for the dense version?).

As for the geopolitics .... I totally agree. People (and politicians) seem to be blind to the cliff they are blithely sprinting up to. If people thought we had electronics inflation during COVID, man, that will look like a pretty tiny thing compared to what would happen in your scenario. Lets hope that someone somewhere wakes up and sees the cliff.

LightningZ71 · Monday at 6:03 PM

People really don't appreciate the bind that the West is in regarding computer supply chains. In a situation where things go sideways in Southeast Asia, the large regional power can easily interdict every supply chain that feeds the West either whole product or critical components for the rest of their products. It's not just processors either; all up and down the list of parts are ingredients that come from that area. Even for products advertised as being all Intel internally sourced tiles are process steps or subcomponents that come from overseas. AMD is essentially without product.

NTMBK · Monday at 6:11 PM

LightningZ71 said:
People really don't appreciate the bind that the West is in regarding computer supply chains. In a situation where things go sideways in Southeast Asia, the large regional power can easily interdict every supply chain that feeds the West either whole product or critical components for the rest of their products. It's not just processors either; all up and down the list of parts are ingredients that come from that area. Even for products advertised as being all Intel internally sourced tiles are process steps or subcomponents that come from overseas. AMD is essentially without product.

They can always go back to Globalfoundries and fire up the old Stoney Ridge designs! It was a terrible chip, but in the event of that sort of disaster at least they'll have something to sell. A functional laptop is better than no laptop at all. (Similar to the stuff that Russia puts out for internal use.)

DavidC1 · Monday at 6:14 PM

OneEng2 said:
Seems like CWF had a window to take back leadership in performance (although I still question the profitability of the Intel vertical integration model). Slip this until mid 26 and it will soon be up against an AMD Zen 6 part on either N3P or N2 (or N3P for the high performance version and N2 for the dense version?).

They are missing everything that could have been guaranteed successes. Falcon Shores, and now Clearwater Forest. Yea it does look like a 6 month delay. Q4 meant Nov/Dec, so Q1 means May/June. They have such execution problems that what they say is always towards the last few days left over. H1 means early June, and H2 means December. Why not Feb/March for H1 and August/September for H2?

Maybe it would have been fine but the recent revenue slips and the VSPs and layoffs have affected the development? If it's true it means the most optimistic part of Intel's plans all depended on the revenue being optimistic, which was unrealistic.

If the designs that seemed promising are slipping, what about others? DMR isn't even mentioned so we're looking at probably late-2027.

LightningZ71 · Monday at 6:19 PM

NTMBK said:
They can always go back to Globalfoundries and fire up the old Stoney Ridge designs! It was a terrible chip, but in the event of that sort of disaster at least they'll have something to sell. A functional laptop is better than no laptop at all. (Similar to the stuff that Russia puts out for internal use.)

Can GF even produce a complete processor for AMD from end to end without involving any parts or steps in SE Asia? Do they have any nodes better than 12FF+ or their 22FDX line that's producing in any volume? We know 12FDX is doing something, but is it fully ready for products?

NostaSeronx · Monday at 6:26 PM

LightningZ71 said:
Can GF even produce a complete processor for AMD from end to end without involving any parts or steps in SE Asia? Do they have any nodes better than 12FF+ or their 22FDX line that's producing in any volume? We know 12FDX is doing something, but is it fully ready for products?

GlobalFoundries has been partnering with companies and getting funding to support full domestic production lines.
Amkor is the main partner for GlobalFoundries.
Porto, Portugal / Dresden OSAT : https://gf.com/gf-press-release/amk...-ribbon-on-strategic-cooperation-in-portugal/
Peoria, Arizona / Malta OSAT

It will only be a matter of time for other smaller details.

DavidC1 · Monday at 9:06 PM

NTMBK said:
They can always go back to Globalfoundries and fire up the old Stoney Ridge designs! It was a terrible chip, but in the event of that sort of disaster at least they'll have something to sell. A functional laptop is better than no laptop at all. (Similar to the stuff that Russia puts out for internal use.)

Why not fix and maintain used computers? Many are way better than Stoney Ridge generation.

NostaSeronx · Monday at 9:53 PM

DavidC1 said:
Why not fix and maintain used computers? Many are way better than Stoney Ridge generation.

Used computers are power hungry. Old products are always obsolete.

The circuit design of the 28nm generation is superior to 14nm/6nm/4nm. Those circuits can only be 1:1 ported to 28FDS/18FDS, 22FDX/12FDX.

Combine Family 15h's Dynamic CMOS, Family 16h's Low Power, Family 19h's Clustered FPU, Family 1Ah's Clustered Front-end. The only thing left is a clustered integer core. Basically, return back to what could have released in 2009.

Gen1 (avg, most common SKU) = 6W/4.5W .. $25
Gen2 (avg, . . .) = 8W~10W .. $50
Gen3 (avg, . . .) = 15W .. $100
FinFETs killed AMD's low power and low cost roadmap. 4nm part is likely to have a PPT of 30W/54W. Would really suck if another Energy Act of 2005 comes by. All new home-bound computers must consume LESS for BIG DATA!!!

511 · Monday at 10:39 PM

What about the Intel 16 capacity in USA it uses Industry standard EDA would be better than GF.
In case of war there will be a issues In supply chain for every company it's just Intel will take the least damage out of it's competitors.

Also in case of war Tsmc Arizona will be fought over by companies

RTX · Monday at 11:51 PM

NostaSeronx said:
GlobalFoundries has been partnering with companies and getting funding to support full domestic production lines.
Amkor is the main partner for GlobalFoundries.
Porto, Portugal / Dresden OSAT : https://gf.com/gf-press-release/amk...-ribbon-on-strategic-cooperation-in-portugal/
Peoria, Arizona / Malta OSAT

It will only be a matter of time for other smaller details.

Does AMD plan on using GF's advanced packaging facilities?

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

Golden Member

Golden Member

Golden Member

Lifer

Senior member

Diamond Member

Lifer

Diamond Member

Golden Member

Golden Member

Golden Member

Golden Member

Senior member

Senior member

Member

Golden Member

Senior member

Platinum Member

Lifer

Golden Member

Platinum Member

Diamond Member

Golden Member

Diamond Member

Golden Member

Member