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.

 

[eek2121]

So Meteorlake compute tile hasn't been taped out yet? Takes a year so if they want to make it H1 like some people here are expecting they only have few short months to do it assuming it takes exactly 12 months to a product.

This is why I don't like the new "taped-in" terminology. Makes it sound like they are doing more than they are actually doing rather than in the old days where they only said things when tape outs occurred.

Also what is the server chip on the Intel 3 process?

2022 H2 - Sapphire Rapids Intel 7
2023 H2 - Emerald Rapids Intel 7
2024 H2 - Granite Rapids Intel 4? 3?

So Granite Rapids is Intel 3 rather than Intel 4 now? Or they don't need Emerald Rapids anymore? At least good to know that there's something big on Intel 3, otherwise that process would have been dead as a dodo bird.

Update: Ok wow so it looks like Granite Rapids is moving to Intel 3 from Intel 4. Good for them.

It sounds like Meteor Lake is no longer a “limited release” like some here thought. It will launch in H2 2023 as the 14900k while Arrow Lake will launch as the 15900k in H2 of 2024. At least that is my interpretation.

 

[igor_kavinski]

Culture Clash at TSMC's Arizona Fab - EE Times Asia

Will management and workers find a middle ground as the chip giant's newest U.S. fab ramps production?

www.eetasia.com

TSMC is often likened to a well-disciplined army. The company has the perhaps undeserved reputation for using Ph.Ds to monitor a single piece of equipment on a production line.

“Basically, one machine does not require a Ph.D to look after it,” said Lai I-Chung, president of The Prospect Foundation, a think tank run by the Taiwan government. “But since we have these over-educated engineers managing the process, they can deal with problems onsite very quickly. That’s how the competitiveness of TSMC really emerged.

“We are fielding our Ph.Ds as foot soldiers, but actually many of them could be colonels. That kind of culture wouldn’t work in the United States.”

The hope is that TSMC can improve its management skills so engineers can eventually work eight-hour shifts. In a 2020 interview Lai said, “I do not think that TSMC is ready for that.”

So their secret to success is more PhDs and longer working hours?

 

[jpiniero]

I just came here to say that.
For 20A: "expected to ramp in 2024" became "manufacturing-ready in the first half of 2024"
For 18A: "in development for early 2025" became "manufacturing-ready in the second half of 2024"

What does "manufacturing ready" even mean? Risk production? Pre-risk production?

 

[DrMrLordX]

It sounds like Meteor Lake is no longer a “limited release” like some here thought. It will launch in H2 2023 as the 14900k while Arrow Lake will launch as the 15900k in H2 of 2024. At least that is my interpretation.

Or maybe Intel is lying.

 

[eek2121]

Or maybe Intel is lying.

To investors? You know that is a federal crime, right?

 

[DrMrLordX]

To investors? You know that is a federal crime, right?

They got away with Cannonlake, so don't be so quick to assume that they don't have some plausible deniability to back up their claims.

 

[IntelUser2000]

Culture Clash at TSMC's Arizona Fab - EE Times Asia

Will management and workers find a middle ground as the chip giant's newest U.S. fab ramps production?

www.eetasia.com

So their secret to success is more PhDs and longer working hours?

That makes it sound pretty terrible working for them. That's a BAD thing because they are saying we need to trade people's lives for corporate profit. Great, $100K+ spent on education but you end up being a lowly manager. Sounds great for humanity.

Also what's that about not needing PHDs but making it sound it was the PHDs that made the company successful.

 

[DrMrLordX]

Great, $100K+ spent on education but you end up being a lowly manager. Sounds great for humanity.

If you get pay commensurate with your abilities doing a menial job, where is the problem?

 

[IntelUser2000]

If you get pay commensurate with your abilities doing a menial job, where is the problem?

That's not what the article seems to suggest. That they could be colonels but are rather foot soldiers? And won't fly with the way US culture works?

What is "US" culture in this case? Because exploiting employees by underpaying is pretty much worldwide. That TSMC's culture is like a well paid foot soldier?

Another thing that doesn't make sense is if your education is allowing something that less educated employees can't do, then you can't call it a menial job, since you are actually providing value to work.

"Basically, one machine does not require a Ph.D to look after it,”

You know how that sounds like right? Doesn't need it but needs it?

The reason "foot soldiers" and "menial employees" don't get paid a lot is because they don't provide much in terms of thinking. If they did, then the perception would be different. The article is making it confusing if it indeed is the case as you suggest.

 

[DrMrLordX]

The article is making it confusing if it indeed is the case as you suggest.

It's not entirely clear what value the company derives from overqualified employees minding the daily operation of machines. Might be occasional boosts in productivity due to occasional prevention of edge-case failures. Hard to say.

 

[moinmoin]

Fabricated yields at Samsung? Yikes...

https://twitter.com/x/status/1496127871869456389

 

[Saylick]

Fabricated yields at Samsung? Yikes...

https://twitter.com/x/status/1496127871869456389

I guess Samsung took the word "fab" a little too liberally.

Big yikes, though. We all knew Samsung didn't have the best processes in the world, but this is just bad. Suffice to say, confidence in their 3GAE process must take a blow as well.

 

[DisEnchantment]

Fabricated yields at Samsung? Yikes...

I am not sure how credible the report is, but if true, the fact that they willing to do such a thing could lead to something good coming out of it.

Frankly it would be weird that an investment of hundreds of billions of dollars over a few years would not have routine audits.
It is just fraud waiting to happen if no such things are done regularly.
To me sounds like the usual audit of large investments at major corporations, I have seen many of those in our organization too.
Normally executives gets asked lots of questions and papers/the books/project report cards etc would need to be in order, and we do trial runs of the auditing process before the external auditor comes to check us.
The top management or the board would usually specify additional focus of the audit as well if not already part of the project auditing and accounting perimeter.
And we also usually have external consulting firms involved in the process.

 

[moinmoin]

I am not sure how credible the report is, but if true, the fact that they willing to do such a thing could lead to something good coming out of it.

Frankly it would be weird that an investment of hundreds of billions of dollars over a few years would not have routine audits.
It is just fraud waiting to happen if no such things are done regularly.
To me sounds like the usual audit of large investments at major corporations, I have seen many of those in our organization too.
Normally executives gets asked lots of questions and papers/the books/project report cards etc would need to be in order, and we do trial runs of the auditing process before the external auditor comes to check us.
The top management or the board would usually specify additional focus of the audit as well if not already part of the project auditing and accounting perimeter.
And we also usually have external consulting firms involved in the process.

If the report is indeed factual the audit should indeed lead to good improvements.

I can imagine this happening since for Samsung's foundries compute dies is just a secondary business, with memory still being the main business by some distance. And the latter is easier and faster to yield successfully on new nodes while the former is mostly "only" a prestige race against TSMC and Intel. But yeah, everybody should be happy for the money working more efficiently as a result of this.

 

[Doug S]

One only needs to google 'Samsung executives jail' to see that the company isn't run by the most honest bunch, with a half dozen or so top executives including the CEO having served time in jail recently (or still behind bars in some cases)

 

[igor_kavinski]

Is Samsung's 8nm yield woes one of the reasons for there not being enough Nvidia RTX hardware to go around? Who is fabbing the RTX 3050?

 

[jpiniero]

Is Samsung's 8nm yield woes one of the reasons for there not being enough Nvidia RTX hardware to go around? Who is fabbing the RTX 3050?

No it's because of miners (and scalpers selling to miners)

 

[igor_kavinski]

https://www.tomshardware.com/news/tsmc-3nm-yield-problems-may-derail-amds-cpu-plans

So now AMD is in trouble?

Also,

https://twitter.com/x/status/1496203886822105090

Easy die-shrinks!

 

[DrMrLordX]

Is Samsung's 8nm yield woes one of the reasons for there not being enough Nvidia RTX hardware to go around?

Probably not. But if you look at some of their 5LPE products - notably the Snapdragon 888 and Exynos 2100 - it might provide some clues as to how those SoCs wound up being as underwhelming as they were.

Dunno about 4LPE yet but we'll see.

Also AMD hasn't announced any designs for N3 yet, not sure why people are talking about them? Intel is investing billions to get early wafers of N3, though.

 

[geoxile]

Is Samsung's 8nm yield woes one of the reasons for there not being enough Nvidia RTX hardware to go around? Who is fabbing the RTX 3050?

I'm pretty sure Ampere is one of the fastest growing nvidia generations on steam's hardware survey in recent history, despite everyone complaining about stock. If there were low yields they must have improved it very quickly.

 

[soresu]

Is Samsung's 8nm yield woes one of the reasons for there not being enough Nvidia RTX hardware to go around? Who is fabbing the RTX 3050?

I doubt it, Sammy's 8nm is just a refined 10nm process so it would have made news long before now if it was a yield trainwreck.

Of course though, nVidia are well known for comparatively huge dies, so lower yields on more advanced processes are to be expected - it's one of the reasons that smaller chiplets are the only sane way forward on future fab nodes.

 

[DisEnchantment]

Intel picks Magdeburg in Germany for new European chip factory

Intel INTC.O has chosen the east German city of Magdeburg as the site for a new multibillion-euro European chip factory and will make the decision public on March 4, a person familiar with matter told Reuters.

www.reuters.com

Intel to build new fab in eastern Germany. Official announcements to come soon I guess

(TSMC also rumored to follow suit with a fab in EU but nothing concrete so far.)

Within the next 3 years: 100B+ USD TSMC, 100B+ USD Samsung, 40B+ Intel. Some estimates put TSMC's investment from 2022 - 2024 to be in 120B+ USD range with a staggering official commitment of 44B USD in 2022 alone.
From history, the record was held by Samsung with 93B USD over 2017-2020 years and they have been the top spender from 2004 to 2021 barring one year in 2009.
2022 could go to TSMC with Samsung a close second.

Samsung's investments comprises of the memory business as well, which decreases their overall available capital expenditure on logic but nevertheless they have shown they can sustain their investment over two decades and it also means they can sustain R&D capital expenditure even with years of trailing behind TSMC.

The sheer magnitude of Samsung’s spending over the 2017-2020 timeperiod ($93.2 billion) is unprecedented in the history of the semiconductor industry!

https://www.icinsights.com/news/bulletins/Samsung-And-TSMC-Seeking-To-Spend-Their-Way-To-Worldwide-Domination-Of-Advanced-IC-Technology/

With lots of fabs being planned I really hope there really is some consistent demand otherwise these investments would be really painful to sustain and not sure if these giants can recover their investments.
But nevertheless it is relieving to see more fabs planned outside of SK and Taiwan, which is a bit unnerving since we are always one disaster away from chip meltdown.

 

[MadRat]

Silicon production is unlikely to reduce any decade soon.

 

[Frenetic Pony]

Given that even TSMC is now (supposedly) having problems advancing nodes, I wonder who'll be the first to have CNFETS on their roadmap? That appears to be the prime candidate for a silicon alternative given their extreme perf/power metrics. And with demos of them being built on industry standard manufacturing practices and turing complete chips built out of them afaik the biggest blocker is just producing enough high quality CNTs for low enough cost.

Versus paying $50 billion (or thereabouts) per 1.5nm fab even "risky" alternatives have to seem a better bet at some point in the not too distant future.

Silicon production is unlikely to reduce any decade soon.

I mean, that doesn't guarantee there won't be an oversupply. But then an oversupply is good for us, yay low prices!

 

[Ajay]

Intel picks Magdeburg in Germany for new European chip factory

Intel INTC.O has chosen the east German city of Magdeburg as the site for a new multibillion-euro European chip factory and will make the decision public on March 4, a person familiar with matter told Reuters.

www.reuters.com

Intel to build new fab in eastern Germany. Official announcements to come soon I guess

(TSMC also rumored to follow suit with a fab in EU but nothing concrete so far.)

Hmm, Magdeburg is a pretty small city (~ 1/4M pop). Guess Intel will be importing a lot of talent from greater Germany.