Discussion Leading Edge Foundry Node advances (TSMC, Samsung Foundry, Intel)

[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.

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.

 

[SiliconFly]

ASML fires back at accusations that its next-gen High-NA EUV chipmaking tools are too expensive

Says their insertion into production flows is inevitable anyway.

www.tomshardware.com

Funny when ASML says, "Talk to Intel if you wanna know the real costs of double patterning!"

ASML has every right to do so. When TSMC moved to EUV, the bean counter geniuses in Intel decided to stay with their outdated DUV machines. The idea was they'll stick to double patterning (and even quad) to not only get more out of their outdated investments, but also avoid making new expensive investments in EUV. The rest is history.

Now that Intel has ordered High-NA EUVs well ahead of TSMC, ASML wants to take a shot at TSMC for not ordering High-NA EUVs (similar to what Intel did long back).

 

[DrMrLordX]

And why is IFS having excess capacity? At the meantime, TMSC is having EUV machines constraint for N3E process cause as I listed here, TSMC has received almost all business from OEMs including Intel. Intel so far has been pretty much confirmed to order 3 EUV machines in order to produce LNL, ARL 8P+16E and ARL 8P+32E.

Meanwhile, IFS is having more than 10 EUV sitting idle without orders....At first I thought MTL would start Intel's EUV aggressive roadmap, but all signs are indicating IFS is not capable of producing competitive SoC. Sorry if there is Intel's shareholders in this forum, better run ASAP...

IFS doesn't even have any of its major nodes online yet (Intel 3 and Intel 18a).

 

[igor_kavinski]

ASML wants to take a shot at TSMC for not ordering High-NA EUVs (similar to what Intel did long back).

Except TSMC may have an actual plan and reason for not doing so.

 

[Doug S]

Also that Semianalysis article came out weeks (or months?) ago, so any possible damage to the stock price has already been done.

Stock prices are fluid. You can have bad news that cases a dip, then a counter to that bad news that undoes that dip.

There might even be an incentive to leave "bad news" like that Semianalysis article laying there over the new year - maybe the stock has already run up in 2023 even with that dip and they want a lower starting point in 2024 since they know it will be a more difficult compare to EOY.

There's also a myriad of complicating factors from upper management who may have bonuses tied to stock prices in some way - often ESOPs allow purchasing shares at a discount based on the lowest of beginning/end of year pricing. Or maybe they are issued stock options based on the Jan 1 price, or have bonuses tied up in how much the stock price increases in a given year. ASML is not based in the US so they may not have the problems in short termism management thinking we too often see here.

It is also possible they simply didn't hear about the Semianalysis article right away, then between the holidays and such just didn't get around to coming up with an official response until now.

 

[Ajay]

The slow rollout of the CHIP acts funding is likely playing a large roll in this delay as well. Funds were expected to be allocated before the end of last year IIRC

Yes it, it is changing the economics of starting a new fab - especially as capital costs have gone up.

 

[Tigerick]

IFS doesn't even have any of its major nodes online yet (Intel 3 and Intel 18a).

The design stage begins three years before HVM of new process, that's why there is news about upcoming Nova Lake being manufactured by TSMC's N2 process. That's mean atm N2's PDKs are ready for Apple, Intel, AMD and NV to begin design of new chips based on PPA of N2 specification.

IFS is at final stage of completing their PDK for upcoming three new nodes, whatever name they would be called....5NI4Y would be over after IFS announces new nodes this month, and have you wonder why Intel still not announcing any new customers?

 

[Ajay]

IFS is at final stage of completing their PDK for upcoming three new nodes, whatever name they would be called....

And finally using industry standard tools and workflows - as opposed to Intel's proprietary in house tool chain.

 

[maddie]

Except TSMC may have an actual plan and reason for not doing so.

Nah. Obviously everyone knows that TSMC is clueless in achieving leading edge nodes unlike Intel.

 

[DrMrLordX]

The design stage begins three years before HVM of new process, that's why there is news about upcoming Nova Lake being manufactured by TSMC's N2 process. That's mean atm N2's PDKs are ready for Apple, Intel, AMD and NV to begin design of new chips based on PPA of N2 specification.

IFS is at final stage of completing their PDK for upcoming three new nodes, whatever name they would be called....5NI4Y would be over after IFS announces new nodes this month, and have you wonder why Intel still not announcing any new customers?

IFS is new, and their track record is kinda iffy. They have to prove themselves, and part of that will be proving that their foundry nodes are actually worth using versus the perenially-available and cheap Samsung, and the less-available and much less cheap (and superior) TSMC. One of the ways they can do this is to roll out a successful IDF product on one of their foundry nodes (like Intel 3) to show off how great your future products can be on the next foundry node. 18a is the one Intel probably hopes to sell the heck out of to major players looking for at least an alternative to Samsung. And if Intel can get double-patterning to work as a stopgap solution for High NA EUV, all the better.

That's why I pointed out that we haven't even seen anything on Intel 3 (or 18a) yet. Sure, presumably, the PDKs are already available, unless IFS 2.0 is such an organizational failure that they can't offer that to customers.

 

[Hitman928]

IFS is new, and their track record is kinda iffy. They have to prove themselves, and part of that will be proving that their foundry nodes are actually worth using versus the perenially-available and cheap Samsung, and the less-available and much less cheap (and superior) TSMC. One of the ways they can do this is to roll out a successful IDF product on one of their foundry nodes (like Intel 3) to show off how great your future products can be on the next foundry node. 18a is the one Intel probably hopes to sell the heck out of to major players looking for at least an alternative to Samsung. And if Intel can get double-patterning to work as a stopgap solution for High NA EUV, all the better.

That's why I pointed out that we haven't even seen anything on Intel 3 (or 18a) yet. Sure, presumably, the PDKs are already available, unless IFS 2.0 is such an organizational failure that they can't offer that to customers.

I know that at least select Intel 3 customers have the PDK.

 

[H433x0n]

First announced customer for 18A. There’s supposed to be at least 3 more. I’ve got a bunch on 2 of the other 3.

Mediatek and probably OpenAI.

 

[Hitman928]

First announced customer for 18A. There’s supposed to be at least 3 more. I’ve got a bunch on 2 of the other 3.

Mediatek and probably OpenAI.

Seems to be a small (< 1000 employees) IP company (i.e., they do design work but don't really produce many products themselves). Also seems to be a subsidiary of UMC but not sure as they may have been fully spun off at some point. Might be somewhat of a by-product of Intel's recent deal with UMC for their 12 nm production. I think success for them would be to merely show that you can successfully design with IFS and provide some catalog IP for the process, but they won't be pushing much volume at all by themselves.

Mediatek could be a huge deal if Intel could get an agreement from them for a major product line. OpenAI is a software company. I know they're exploring doing their own AI hardware, but I don't expect much on that end any time soon.

 

[igor_kavinski]

TSMC is now the world's largest semiconductor maker by revenue, beating Intel and Samsung: Analyst

Bigger than Intel, Nvidia and Samsung.

www.tomshardware.com

Intel is barely above Samsung!

 

[FlameTail]

TSMC is now the world's largest semiconductor maker by revenue, beating Intel and Samsung: Analyst

Bigger than Intel, Nvidia and Samsung.

www.tomshardware.com

Intel is barely above Samsung!

Samsung continues to sink like the Titanic

 

[igor_kavinski]

Samsung continues to sink like the Titanic

Until it rises from the depths of Hell to conquer all!

Coming soon to theaters near you!

 

[FlameTail]

Just like how the captain of the Titanic bet too much on the strength of it's hull. Eventually the ship struck an iceberg, the hull failed, and the ship went down to the seabed;

Similarly, Samsung bet too much on their 3nm GAAFET.

 

[desrever]

TSMC is now the world's largest semiconductor maker by revenue, beating Intel and Samsung: Analyst

Bigger than Intel, Nvidia and Samsung.

www.tomshardware.com

Intel is barely above Samsung!

Its comparing Intel as a whole to TSMC and Samsung's chip division. Kind of a dumb comparison. If it was only for foundry side, TSMC would be way bigger and so would Samsung.

 

[Tigerick]

Its comparing Intel as a whole to TSMC and Samsung's chip division. Kind of a dumb comparison. If it was only for foundry side, TSMC would be way bigger and so would Samsung.

Yes, that's the points. If Intel and Samsung's revenues are based on foundry business, they are both below $20 billions last year...

Samsung of course has to exclude memory business...

 

[FlameTail]

Samsung of course has to exclude memory business...

🤣

 

[DrMrLordX]

Similarly, Samsung bet too much on their 3nm GAAFET.

It's more complicated than that. Samsung is a massive government-backed conglomerate that makes everything from semiconductors to washing machines. They've had several problems in the semiconductor market that led to losses, and 3GAE is just one of them. Though apparently they've turned the ship around, at least in DRAM:

Samsung turns profit from DRAM again at tail-end of rough 2023

Demands for memory chips and smartphones are expected to gradually recover in 2024, Samsung says.

www.zdnet.com

As we all know, they're pretty deep into the DRAM business, and that has been a sore point for them over the past year which has . . . maybe less to do with their foundry misadventures. Samsung probably expects DRAM revenue to help prop up other parts of their semiconductor business, and when that doesn't happen (due to losses in a normally-lucrative DRAM market where they can use older/different nodes for their memory products), then maybe they have to lean more on their government ties and non-semiconductor units for survival, which is never a good thing.

 

[FlameTail]

It's more complicated than that. Samsung is a massive government-backed conglomerate that makes everything from semiconductors to washing machines. They've had several problems in the semiconductor market that led to losses, and 3GAE is just one of them. Though apparently they've turned the ship around, at least in DRAM:

Samsung turns profit from DRAM again at tail-end of rough 2023

Demands for memory chips and smartphones are expected to gradually recover in 2024, Samsung says.

www.zdnet.com

As we all know, they're pretty deep into the DRAM business, and that has been a sore point for them over the past year which has . . . maybe less to do with their foundry misadventures. Samsung probably expects DRAM revenue to help prop up other parts of their semiconductor business, and when that doesn't happen (due to losses in a normally-lucrative DRAM market where they can use older/different nodes for their memory products), then maybe they have to lean more on their government ties and non-semiconductor units for survival, which is never a good thing.

The AI boom for sure guarantees a steady and high demand for DRAM.

 

[moinmoin]

Samsung probably expects DRAM revenue to help prop up other parts of their semiconductor business

It's pretty clear that Samsung's semiconductor business wouldn't exist at all without DRAM, so of course the economic state of that is affecting everything else.

 

[DavidC1]

That's why I pointed out that we haven't even seen anything on Intel 3 (or 18a) yet. Sure, presumably, the PDKs are already available, unless IFS 2.0 is such an organizational failure that they can't offer that to customers.

Ericsson silicon is on Intel 4 and was introduced in November.

 

[igor_kavinski]

Ericsson silicon is on Intel 4 and was introduced in November.

Link?

 

[Saylick]

Link?

Ericsson processors built on 'Intel 4' process node beat Meteor Lake to deployment

IDM 2.0 is finally getting somewhere.

www.tomshardware.com