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

[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.

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

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

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

 

[A///]

Sorry, wasn't clear. Why would Apple pay Intel for wafers that it decided not to use?

Why would Apple pay for critical fab time it cannot use because it was bought up by another party?

 

[Doug S]

Sorry, wasn't clear. Why would Apple pay Intel for wafers that it decided not to use?

I find it unlikely to the extreme that Intel bought capacity ahead of Apple, given Apple's long relationship with TSMC. I also find it unlikely that Intel prepaid 100% of the cost, and if they paid less than 100% it would have been on "use it or lose it" terms. Or at best "we will consider a trade of that prepaid deposit for some future capacity at our convenience".

Most likely whatever past deal Intel may have had, they made a new deal with TSMC "hey we don't need this capacity now can we trade it for future capacity" and TSMC said "not on a 1:1 basis because we made our investments based on you having purchased this capacity and now it is going to waste".

Why would Apple be interested in making a deal with Intel, when Apple would have already made a deal with TSMC? If Intel really had bought capacity ahead of Apple (which again I think is just stupidly unlikely as TSMC would want to keep their most important customer happy, and they would know Intel was not going to be a long term customer) then Apple would have already made arrangements with TSMC to use N4P if they really couldn't use N3 for A17. Why would they want to bail out Intel from their own mistakes? They would hurt TSMC by canceling an N4P order to take over Intel's N3B order.

The biggest hint that Apple didn't buy their capacity from Intel is that they are paying TSMC for known good dies. TSMC has never made a deal like that before, they did it because of the low yields of N3B. Something they would not have known about when Intel was rumored to have made a deal with TSMC for N3 capacity. So Apple had to have dealt with TSMC, not Intel.

 

[DrMrLordX]

I'm sure Apple paid Intel to get that allocation. Would lessen the sting.

Apple probably paid TSMC for the extra capacity, not Intel.

I find it unlikely to the extreme that Intel bought capacity ahead of Apple,

Apple is a preferred customer. When TSMC rolls out a new node, it's generally understood (at least by us outsiders) that TSMC puts money down on R&D and capacity for the new node - in this case, N3 - assuming that Apple will be among the first to take wafers. And of course Apple showed up as predicted.

Intel is not a preferred customer. If they want capacity, they can a). wait until preferred customers get their fill or b). pay TSMC in advance to make it worth their while to build out additional capacity specifically for (in this case) Intel. Intel chose option b), and TSMC actually converted a research fab to a production fab to (partially) fill the initial order Intel put in for wafers. Intel paid them around $10 billion if I recall correctly.

Since Intel requested a delay in taking wafers, that left TSMC in the position to idle the fabs they had built out to meet Intel's order or to sell the capacity to someone else. And sell it they did - to Apple. Intel has no dog in this fight since TSMC now doesn't need to provide Intel with anything until a future date despite Intel having already paid them in full for the capacity.

 

[Doug S]

Apple probably paid TSMC for the extra capacity, not Intel.



Apple is a preferred customer. When TSMC rolls out a new node, it's generally understood (at least by us outsiders) that TSMC puts money down on R&D and capacity for the new node - in this case, N3 - assuming that Apple will be among the first to take wafers. And of course Apple showed up as predicted.

Intel is not a preferred customer. If they want capacity, they can a). wait until preferred customers get their fill or b). pay TSMC in advance to make it worth their while to build out additional capacity specifically for (in this case) Intel. Intel chose option b), and TSMC actually converted a research fab to a production fab to (partially) fill the initial order Intel put in for wafers. Intel paid them around $10 billion if I recall correctly.

Since Intel requested a delay in taking wafers, that left TSMC in the position to idle the fabs they had built out to meet Intel's order or to sell the capacity to someone else. And sell it they did - to Apple. Intel has no dog in this fight since TSMC now doesn't need to provide Intel with anything until a future date despite Intel having already paid them in full for the capacity.

Your claims are contradictory. You first say "...assuming that Apple will be among the first to take wafers. And of course Apple showed up as predicted." then claim that TSMC sold the extra capacity they built for Intel to Apple. If Apple "showed up as predicted" they would be using capacity that TSMC had already planned on building, not the extra capacity they built for Intel.

It beggars belief to think that TSMC had a new node coming, Apple said "no thanks we will stick with an older node" and meanwhile Intel is paying $10 billion for TSMC to build out new capacity especially for them, then when they no longer want it TSMC goes to Apple saying "hey we have this unused capacity Intel doesn't need, you sure you want to stick with that older node?"

 

[DrMrLordX]

Your claims are contradictory. You first say "...assuming that Apple will be among the first to take wafers. And of course Apple showed up as predicted." then claim that TSMC sold the extra capacity they built for Intel to Apple. If Apple "showed up as predicted" they would be using capacity that TSMC had already planned on building, not the extra capacity they built for Intel.

Wrong, Apple is taking both: the initial planned capacity, and a share of what was added to meet Intel's order.

 

[Doug S]

Wrong, Apple is taking both: the initial planned capacity, and a share of what was added to meet Intel's order.

Huh? Where is Apple's extra demand coming from? Unless they originally planned to use A17 in only the Pro models but will now use it across all models (which we'll find out next week) I don't see where the additional demand could possibly materialize.

 

[Mopetar]

Sorry, wasn't clear. Why would Apple pay Intel for wafers that it decided not to use?

It depends on the cost. Intel was contractually obligated to pay for what they had booked. Maybe the contract has a use it or lose it type clause, but Intel would have to pay some kind of cancellation fee or just don't get their money back if they paid in advance. TSMC would then be free to sell that allocation to someone else.

It's also possible that Intel has the right to sell their allotment to another company under the same terms where the new owner would be on the hook to pay TSMC. Intel might be willing to sell at a price under a cancellation fee and Apple might be willing to buy at a price less than what TSMC charges per wafer if they think they can use the extra wafers.

It's also possible that Intel paid a cancellation fee and that TSMC sold to Apple at a discounted rate as no one using the wafers is less money than someone using them at a discounted rate. Given the timeline for all of this, Apple was probably the only company that could use the wafers. Almost anyone else wasn't planning on having access until next year and wouldn't have anything ready to go into production.

 

[A///]

SMIC 7nm N+2 in Huawei Mate 60 Pro - Uncovering the innovation inside the chip | TechInsights

Following the weekends breaking news by TechInsights, our Disruptive Event Briefing is now available to access for FREE (Registration Required).

www.techinsights.com

anyone got full access to the report? from the summary brief it looks like smic knocked it out of the park.

 

[qmech]

SMIC 7nm N+2 in Huawei Mate 60 Pro - Uncovering the innovation inside the chip | TechInsights

Following the weekends breaking news by TechInsights, our Disruptive Event Briefing is now available to access for FREE (Registration Required).

www.techinsights.com

anyone got full access to the report? from the summary brief it looks like smic knocked it out of the park.

The summary literally only says that it's the most advanced *Chinese* manufacturing node, capable of "full SoC elements like SRAM" (i.e. not a research node).

No other specific claims are made.

There are no comparisons to other 7nm nodes, no claims of being better than or even matching e.g. N7 or Intel7, nothing.

 

[A///]

The summary literally only says that it's the most advanced *Chinese* manufacturing node, capable of "full SoC elements like SRAM" (i.e. not a research node).

No other specific claims are made.

There are no comparisons to other 7nm nodes, no claims of being better than or even matching e.g. N7 or Intel7, nothing.

Intel 7 is a 10nm++ node. The summary is there to entice you to buy a subscription. If they wanted to point out that smic was lying they would have.

 

[qmech]

Intel 7 is a 10nm++ node. The summary is there to entice you to buy a subscription. If they wanted to point out that smic was lying they would have.

Intel7 matches TSMC's N7 node quite well in terms of density and critical dimensions. In other words, if Intel7 is a 10nm node, so is N7.

SMIC itself hasn't really made any substantive claims that I have read regarding the N+2 process.

 

[DrMrLordX]

Huh? Where is Apple's extra demand coming from? Unless they originally planned to use A17 in only the Pro models but will now use it across all models (which we'll find out next week) I don't see where the additional demand could possibly materialize.

Apple to Buy TSMC's Entire Supply of 3nm Chips for 2023

Apple will receive all of TSMC's first-generation 3-nanometer process chips this year for upcoming iPhones, Macs, and iPads, according to...

www.macrumors.com

That article might clear things up a bit. It looks like TSMC is running fabs at less than full capacity? But the article doesn't exactly clarify on this point.

 

[Ajay]

Apple to Buy TSMC's Entire Supply of 3nm Chips for 2023

Apple will receive all of TSMC's first-generation 3-nanometer process chips this year for upcoming iPhones, Macs, and iPads, according to...

www.macrumors.com

That article might clear things up a bit. It looks like TSMC is running fabs at less than full capacity? But the article doesn't exactly clarify on this point.

Hmm, interesting. Makes some sense since TSMC N3E isn't derivative of N3B. Apple will be able to use the same macros in M3 designs as the A17.
(Well, the PDKs aren't compatible at least).

 

[moinmoin]

Huh? Where is Apple's extra demand coming from?

I'd say the extra demand is technically TSMC's due to the lower yield. They'd accordingly need more fab capacity to fulfil Apple's requested amount of good dies.

The rumoured decrease in wpm compared to previously projected seems very high, indicating yield is not the only manufacturing problem with N3B?

 

[Ajay]

I'd say the extra demand is technically TSMC's due to the lower yield. They'd accordingly need more fab capacity to fulfil Apple's requested amount of good dies.

The rumoured decrease in wpm compared to previously projected seems very high, indicating yield is not the only manufacturing problem with N3B?

The decreased WPM appears to be demand driven, not a technical problem.

 

[moinmoin]

The decreased WPM appears to be demand driven, not a technical problem.

Sure, but then the additional capacity previously set up for Intel wouldn't be needed for Apple anyway, which is what the whole discussion was about.

 

[Mopetar]

Apple did buy up what Intel had originally purchased, so presumably they should be running closer to capacity. There could be a little bit of a gap between when Intel was supposed to start production and when Apple actually does start using those wafers, but without any inside information it's impossible to know more.

 

[Panino Manino]

Maybe yelds at SMIC are better than expected?
Apparently there's another 7nm chip called Longying for cars.

 

[A///]

Intel7 matches TSMC's N7 node quite well in terms of density and critical dimensions. In other words, if Intel7 is a 10nm node, so is N7.

SMIC itself hasn't really made any substantive claims that I have read regarding the N+2 process.

And yet Intel's performance was poorer in thermals compared to TSMC's 7nm. You could give Intel N3 today and they'd still make it a firebox through poor design choices.

Pay for the report or don't. don't kill the messenger. Email techinsights and argue with them.

 

[A///]

Sure, but then the additional capacity previously set up for Intel wouldn't be needed for Apple anyway, which is what the whole discussion was about.

It might imply apple seeing a greater demand for phones and laptops/desktops in the very near future? consumer spending seems to be on the rebound.

 

[Ajay]

It might imply apple seeing a greater demand for phones and laptops/desktops in the very near future? consumer spending seems to be on the rebound.

No, lower. Apple is expecting to sell fewer iPhone 15s.

 

[A///]

No, lower. Apple is expecting to sell fewer iPhone 15s.

I saw that news article too. Seems like they say that every year and still have record sales numbers.

 

[DrMrLordX]

Sure, but then the additional capacity previously set up for Intel wouldn't be needed for Apple anyway, which is what the whole discussion was about.

Seems like a shell game on TSMC's part. Probably just so they don't fully idle any of their facilities.

 

[A///]

Seems like a shell game on TSMC's part. Probably just so they don't fully idle any of their facilities.

time is money as they say.

 

[Doug S]

Apple did buy up what Intel had originally purchased, so presumably they should be running closer to capacity. There could be a little bit of a gap between when Intel was supposed to start production and when Apple actually does start using those wafers, but without any inside information it's impossible to know more.

That still doesn't make sense. Exactly what are you claiming Apple is getting made in N3B now that they wouldn't have gotten made before they supposedly "bought up what Intel had originally requested". Are you claiming Apple was not going to make A17 in N3B originally? Or that A17 will be used in all iPhone 15s now but originally it was planned only for the Pro/Pro Max?

If Apple is buying extra capacity Intel had previously bought, they have to getting MORE N3B chips than they had originally contracted for. That means there is something Apple is getting made in N3B now that they previously would not have.

I think it is far more likely that TSMC originally had exactly two major N3B customers, one of them dropped out so they reallocated those lines for N3E which is where they have been strongly encouraging customers to go as it is the "long lived" node that fixes the yield issues with N3B. Apple is getting the exact same quantity of N3B chips they had always been planning on, they are not getting any of "Intel's allocation". To keep Apple happy with N3B's shortcomings, TSMC agreed to sell by known good die rather than per wafer, so that Apple doesn't have to accept the costs of TSMC's failure on N3B - which may have meant TSMC allocating more wafer starts to Apple than they originally planned but that's on them once again for screwing up N3B.

If it is true that Apple will be switching from N3B to N3E once N3E is available (i.e. in the middle of the A17 manufacturing cycle) then I'm guessing either TSMC is eating the cost for that (i.e. new mask sets and any required design rework) or TSMC requiring that transition midstream was the price for the KGD deal.