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 · May 7, 2025

lightisgood said:
Yeah, I know this too.
However, I don't know what about yield of GAAFET w/o BSPDN in Intel and it looks like Samsung's is bad for 3 years.

Yeah at leas Intel has given DD publicly and the DD can be calculated at this point of time but Samsung hasn't given anything. TSMCs DD are the best as of rn.

511 · May 9, 2025

@dullard look what i found regarding 18A and targets.

Thibsie · May 10, 2025

Lol, with graphs axis marked (should I say unmarked?) this way, Intel may say anything 'cos they can mark the axis later how it fits for them and say 'I told you'.

(This applies to others too of course, just talking about Intel right now)

If this was an attempt at showing how much Intel marketing is BS, well, Intel just won a gold BS medal.

511 · May 10, 2025

Thibsie said:
Lol, with graphs axis marked (should I say unmarked?) this way, Intel may say anything 'cos they can mark the axis later how it fits for them and say 'I told you'.

(This applies to others too of course, just talking about Intel right now)

If this was an attempt at showing how much Intel marketing is BS, well, Intel just won a gold BS medal.

When TSMC does this it is fine but when Intel does it it's not okay nice also this is a foundry event TSMC showed same graph in their event.

Thibsie · May 10, 2025

511 said:
When TSMC does this it is fine but when Intel does it it's not okay nice also this is a foundry event TSMC showed same graph in their event.

That why I said it applies to others too. Care to read post or just want to victimize ?

eek2121 · May 10, 2025

Tons of people have thrown shade at Intel for the lack of results regarding recent process improvements without understanding the absolutely gigantic investment in terms of time/money development costs, not just i. terms of R&D, but in terms of building a sales funnel and developing a base of clients.

Further, the reason there are no products using Intel 18A is because it takes quite a while to build a new product for a new manufacturer on new tooling. Intel has tried to use industry standard tooling, however, no standards body exists, they are also having to lay off people left and right due to being behind for 10+ years, and also being under political pressure.

It may sound like I am defending them, maybe I am, maybe I’m not. I’m no fan or investor, I just know how big of an investment this is. I also know that 18a is in maturity in terms of yields, libraries are lacking, and with no standards and decades working with a closed ecosystem, they have to probe the market the hard way. That being said, they are onboarding clients. Some of the names will absolutely surprise you. Just don’t hold your breath. The biggest names won’t launch any 18A product before 2027 or 2028.

TSMC is actually more terrified than they will ever admit. If you think the reason is that they care about chips act dollars, think again. Those dollars are being canceled, and despite complaints, US efforts are proceeding as planned…and tariffs have very little to do with it. Most big tech companies have signed deals with Intel or are in discussions with them at least.

Thibsie · May 10, 2025

eek2121 said:
It may sound like I am defending them, maybe I am, maybe I’m not. I’m no fan or investor, I just know how big of an investment this is. I also know that 18a is in maturity in terms of yields, libraries are lacking, and with no standards and decades working with a closed ecosystem, they have to probe the market the hard way. That being said, they are onboarding clients. Some of the names will absolutely surprise you. Just don’t hold your breath. The biggest names won’t launch any 18A product before 2027 or 2028.

Good. We need Intel to be competitive. Despite what I think about the company, it is way more useful living rather than dead or uncompetitive.

eek2121 said:
TSMC is actually more terrified than they will ever admit. If you think the reason is that they care about chips act dollars, think again. Those dollars are being canceled, and despite complaints, US efforts are proceeding as planned…and tariffs have very little to do with it. Most big tech companies have signed deals with Intel or are in discussions with them at least.

Oh I don't expect TSMC to show anything, that'd be bad business. We'll see if they've been smarter than Intel or just as dumb.
In recent history they've been smart. Intel was smart too, but look where they landed.

We all need those two to be healthy and competing.

maddie · May 10, 2025

eek2121 said:
Tons of people have thrown shade at Intel for the lack of results regarding recent process improvements without understanding the absolutely gigantic investment in terms of time/money development costs, not just i. terms of R&D, but in terms of building a sales funnel and developing a base of clients.

Further, the reason there are no products using Intel 18A is because it takes quite a while to build a new product for a new manufacturer on new tooling. Intel has tried to use industry standard tooling, however, no standards body exists, they are also having to lay off people left and right due to being behind for 10+ years, and also being under political pressure.

It may sound like I am defending them, maybe I am, maybe I’m not. I’m no fan or investor, I just know how big of an investment this is. I also know that 18a is in maturity in terms of yields, libraries are lacking, and with no standards and decades working with a closed ecosystem, they have to probe the market the hard way. That being said, they are onboarding clients. Some of the names will absolutely surprise you. Just don’t hold your breath. The biggest names won’t launch any 18A product before 2027 or 2028.

TSMC is actually more terrified than they will ever admit. If you think the reason is that they care about chips act dollars, think again. Those dollars are being canceled, and despite complaints, US efforts are proceeding as planned…and tariffs have very little to do with it. Most big tech companies have signed deals with Intel or are in discussions with them at least.

How will Intel deliver the volume you seem to indicate they might get? AFAIK, their wpm numbers are much lower than TSMC.

lightisgood · May 10, 2025

maddie said:
AFAIK, their wpm numbers

Show me, please.
I have no number.

511 · May 10, 2025

eek2121 said:
Tons of people have thrown shade at Intel for the lack of results regarding recent process improvements without understanding the absolutely gigantic investment in terms of time/money development costs, not just i. terms of R&D, but in terms of building a sales funnel and developing a base of clients.

Further, the reason there are no products using Intel 18A is because it takes quite a while to build a new product for a new manufacturer on new tooling. Intel has tried to use industry standard tooling, however, no standards body exists, they are also having to lay off people left and right due to being behind for 10+ years, and also being under political pressure.

It may sound like I am defending them, maybe I am, maybe I’m not. I’m no fan or investor, I just know how big of an investment this is. I also know that 18a is in maturity in terms of yields, libraries are lacking, and with no standards and decades working with a closed ecosystem, they have to probe the market the hard way. That being said, they are onboarding clients. Some of the names will absolutely surprise you. Just don’t hold your breath. The biggest names won’t launch any 18A product before 2027 or 2028.

If Intel Itself makes all of its products on 18A and break even by 27 that would be a good start how can you be a foundry and develop your product externally. this is a bad showing for the biz.

eek2121 said:
TSMC is actually more terrified than they will ever admit. If you think the reason is that they care about chips act dollars, think again. Those dollars are being canceled, and despite complaints, US efforts are proceeding as planned…and tariffs have very little to do with it. Most big tech companies have signed deals with Intel or are in discussions with them at least.

CHIPS Act is unofficially gone they laid off the entire teams xDd.

lightisgood · May 11, 2025

511 said:
CHIPS Act is unofficially gone they laid off the entire teams xDd.

I'm understanding that IFS was driven by RAMP-C, for external customer, not CHIPS Act.

511 · May 11, 2025

lightisgood said:
I'm understanding that IFS was driven by RAMP-C, for external customer, not CHIPS Act.

yeah i think so too except Microsoft they are Intel partner for too long intel must have cut MSFT some deal.

maddie · May 11, 2025

lightisgood said:
Show me, please.
I have no number.

You can search yourself, but I've seen TSMC = 2.5 M wpm & Intel = 0.9 M wpm from several years ago. The spread has gotten larger with time as seen by CAPEX differences. The other issue with Intel is only a fraction of this capacity is available for external customers. This will take a fair time to change in the most optimistic scenarios.

I would imagine that the present political situation is a strong factor in decision making.

511 · May 11, 2025

maddie said:
You can search yourself, but I've seen TSMC = 2.5 M wpm & Intel = 0.9 M wpm from several years ago. The spread has gotten larger with time as seen by CAPEX differences. The other issue with Intel is only a fraction of this capacity is available for external customers. This will take a fair time to change in the most optimistic scenarios.

I would imagine that the present political situation is a strong factor in decision making.

That would put TSMC at 1.33 Million Wafer per Month also Intel Oregon alone is 40K WSPM iirc

maddie · May 11, 2025

511 said:
View attachment 123620

That would put TSMC at 1.33 Million Wafer per Month also Intel Oregon alone is 40K WSPM iirc

Below is what I was referencing. The numbers quoted are for different sized wafers. Mine = 200mm & yours = 300mm. A 2.25X correction factor, so mine = 1.1 M wpm (300mm) in 2019.

Five Semiconductor Companies Hold 53% of Global Wafer Capacity - Semiconductor Digest

Leaders in memory IC and foundry production maintain strongest capacity presence.

www.semiconductor-digest.com

They also had this in the article. Unless you tell me that Intel has out-invested TSMC by multiples since, then I maintain that Intel's capacity is a fraction of TSMC.

Capacity at other semiconductor leaders, including Intel (817K wafers/month), UMC (753K wafers/month), GlobalFoundries, Texas Instruments, and STMicro, fell off rapidly from the top five.

oak8292 · May 11, 2025

maddie said:
Below is what I was referencing. The numbers quoted are for different sized wafers. Mine = 200mm & yours = 300mm. A 2.25X correction factor, so mine = 1.1 M wpm (300mm) in 2019.

Five Semiconductor Companies Hold 53% of Global Wafer Capacity - Semiconductor Digest

Leaders in memory IC and foundry production maintain strongest capacity presence.

www.semiconductor-digest.com

They also had this in the article. Unless you tell me that Intel has out-invested TSMC by multiples since, then I maintain that Intel's capacity is a fraction of TSMC.

Capacity at other semiconductor leaders, including Intel (817K wafers/month), UMC (753K wafers/month), GlobalFoundries, Texas Instruments, and STMicro, fell off rapidly from the top five.

The wspm type of information has been disappearing behind paywalls. TSMC Investor website has a historical operating speadsheet with approximate capacity and quarterly production numbers. The capacity is greater than 16 million 300 mm wafers per month. The max quarterly production post pandemic was about 4 million wafers in a quarter when everything was full it typically runs quite a bit less.

I haven’t found Intel numbers for awhile and don’t know what was included in the 2019 numbers. At that time they were in a couple of joint ventures. They were using Intel 68 in Dalian China for NAND and co owned a facility with Micron for Optane. Micron bought out Intel’s ownership in the Optane facility and shut it down. Fab 68 was sold to SKHynix. They have started at least one new fab in Ireland but they are also contracting a lot more wafers from TSMC.

maddie · May 11, 2025

oak8292 said:
The wspm type of information has been disappearing behind paywalls. TSMC Investor website has a historical operating speadsheet with approximate capacity and quarterly production numbers. The capacity is greater than 16 million 300 mm wafers per month. The max quarterly production post pandemic was about 4 million wafers in a quarter when everything was full it typically runs quite a bit less.

I haven’t found Intel numbers for awhile and don’t know what was included in the 2019 numbers. At that time they were in a couple of joint ventures. They were using Intel 68 in Dalian China for NAND and co owned a facility with Micron for Optane. Micron bought out Intel’s ownership in the Optane facility and shut it down. Fab 68 was sold to SKHynix. They have started at least one new fab in Ireland but they are also contracting a lot more wafers from TSMC.

Yep, recent info not openly available, but monthly?

oak8292 · May 11, 2025

maddie said:
Yep, recent info not openly available, but monthly?

Sorry, greater than 16 million annual wafer starts.

Historical Information - Taiwan Semiconductor Manufacturing Company Limited

TSMC historical information

investor.tsmc.com

lightisgood · May 11, 2025

maddie said:
Below is what I was referencing. The numbers quoted are for different sized wafers. Mine = 200mm & yours = 300mm. A 2.25X correction factor, so mine = 1.1 M wpm (300mm) in 2019.

Five Semiconductor Companies Hold 53% of Global Wafer Capacity - Semiconductor Digest

Leaders in memory IC and foundry production maintain strongest capacity presence.

www.semiconductor-digest.com

They also had this in the article. Unless you tell me that Intel has out-invested TSMC by multiples since, then I maintain that Intel's capacity is a fraction of TSMC.

Capacity at other semiconductor leaders, including Intel (817K wafers/month), UMC (753K wafers/month), GlobalFoundries, Texas Instruments, and STMicro, fell off rapidly from the top five.

So what ?
I well know this numbers.
However, this is not comparison between 18A's volume and N2's.

511 · May 11, 2025

lightisgood said:
So what ?
I well know this numbers.
However, this is not comparison between 18A's volume and N2's.

Well from what I know the two fabs in Arizona are 40K WSPM I saw it in a intel slide somewhere.

maddie · May 12, 2025

lightisgood said:
So what ?
I well know this numbers.
However, this is not comparison between 18A's volume and N2's.

I thought I was shining some light on the issue. I guess light is not good after all.

RTX · May 14, 2025

oak8292 said:
Sorry, greater than 16 million annual wafer starts.

Historical Information - Taiwan Semiconductor Manufacturing Company Limited

TSMC historical information

investor.tsmc.com

How much is that if 350nm, 180nm, 130nm, 90nm, 65nm, 45/40nm, 32nm, 20nm not included?

511 · May 14, 2025

RTX said:
How much is that if 350nm, 180nm, 130nm, 90nm, 65nm, 45/40nm, 32nm, 20nm not included?

So have to calculate on basis of area of wafers

oak8292 · May 14, 2025

RTX said:
How much is that if 350nm, 180nm, 130nm, 90nm, 65nm, 45/40nm, 32nm, 20nm not included?

All of the 20 nm capacity went to 16 nm FinFET. If you look at the annual wafer capacity at the top of the spreadsheet they had a capacity of about 8 million wafers per year. So about half of the capacity is 16 nm and smaller.

I think they have added higher nm capacity at times and they shift capacity around. I think Intel always reused a lot of equipment through node shrinks and I think TSMC does some of that. I hear rumor that they ramp to about 120-150 Kwspm on each node. However depending on expected long term demand some of that may shift to the next node? All speculation on my part.

Capacity ramps take years and talking about relative capacity means taking a snapshot at a point in time. Intel provided a image with their expected capacity per node and there isn’t a flat line on it.

1nm Node by 2027: Intel Foundry Roadmap - techovedas

This marks the first time Intel has publicly acknowledged plans for a 1nm node, with production/development expected to begin in late 2027.

techovedas.com

Hitman928 · May 14, 2025

Intel CFO has now publicly stated what I’ve mentioned a couple of times, Intel has no significant volume customers right now committed to 18a.

We talk about a pipeline model, and we get some customers. Then we test chips and some customers fall out, and there's a certain amount of customers that hang in there. So, committed volume is not significant right now, for sure. But, I think we've got to partly prove ourselves a little bit with our own product and eat our own dog food here. And then it's likely we start to see more engagement around customers. I hear more customers wanting a second source.

Based on this quote and others, they’re not going to be spinning customer engagement or sales pipelines as if they are signed agreements anymore, so sounds like they are making an effort to be more transparent moving forward, so that’s good.

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

Golden Member

Platinum Member

Platinum Member

Golden Member

Platinum Member

Golden Member

Diamond Member

Golden Member

Diamond Member

Senior member

Platinum Member

Senior member

Platinum Member

Diamond Member

Platinum Member

Diamond Member

Member

Diamond Member

Member

Senior member

Platinum Member

Diamond Member

Member

Platinum Member

Member

Diamond Member