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

Page 40 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
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.

1587737990547.png
N7 performance is more or less understood.
1587739093721.png

This year and next year TSMC is mainly increasing capacity to meet demands.

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.


1587739615344.png

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.
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
The cost numbers there contradict some other sources.


In light of "SemiAnalysis's" history of claiming unsubstantiated rumor as fact, inclined to take that with more than a grain of salt.
Maybe I’m missing something, but he is saying that N3B will be 40% higher than N5. He was calling out a Digitimes report that estimated $20k/wafer. 40% higher seems a bit crazy, but N3E will be lower cost.
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136

dullard

Elite Member
May 21, 2001
24,998
3,327
126
There is one thing that really bothers me about Tom's Hardware numbers: the depreciation.

Suppose I have a $1000 fridge/freezer combo to make ice. Suppose I have a lemonade stand. Suppose I sell 10 lemonades. Then according to Tom's analysis, I charge each customer $25.90 in depreciation in my freezer (plus whatever it costs for the lemons and sugar and labor). That isn't how markets actually work. I charge each customer what I can actually get based on the market's supply and demand. If I can't cover the cost of the lemons and sugar and labor, then I stop selling the lemonade. The price of the freezer never comes into play, like all sunk costs, it shouldn't be a factor.

I'm not saying that 5 nm is cheap, but RetiredEngineer's table greatly exaggerates the effect of depreciation. It totally neglects the impacts of supply and demand.
 
  • Like
Reactions: Saylick

Doug S

Platinum Member
Feb 8, 2020
2,203
3,405
136
Maybe I’m missing something, but he is saying that N3B will be 40% higher than N5. He was calling out a Digitimes report that estimated $20k/wafer. 40% higher seems a bit crazy, but N3E will be lower cost.

Perhaps Digitimes was talking about N3E since that's the N3 that will actually be used? Supposedly TSMC is looking at like 1000 wpm for N3/"N3B"; no one cares how much it costs because no one is using it.
 
  • Wow
Reactions: lightmanek and Ajay

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
Perhaps Digitimes was talking about N3E since that's the N3 that will actually be used? Supposedly TSMC is looking at like 1000 wpm for N3/"N3B"; no one cares how much it costs because no one is using it.
1kwpm - well, Apple isn't using it then.
 

Doug S

Platinum Member
Feb 8, 2020
2,203
3,405
136
1kwpm - well, Apple isn't using it then.

Well not for iPhone 15, clearly. So either Apple is using N3E or N4P there. They could use N3 for M2 Pro/Max, if they are looking to release Macs containing those this spring. Though N3's problems may affect them (especially since they don't bin on frequency) so the result might be rather underwhelming.
 
  • Like
Reactions: Ajay

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136

TSMC customers cutting orders?

Well, I am not surprised. But why is AMD also cutting orders? I thought they were supply constrained with Zen 3 CCDs
Because the recent past is often held as the new normal.

AMD capacity constraints means producing more low end products is sacrificing margins as too few wafers. TSMC will keep raising prices as demand outstrips supply. Sound familiar?

They are still being written and are wrong. The few voices stating that prices will fall are right, it's just not a quick (1-2)Q process, sort of still early in the stages of grief thing.

A new normal is being born and it's going to be ugly for most.

What was so funny is this line at the end, "Market observers generally remain optimistic about the demand for advanced chips returning to normal in 2023." Classic denial phase.

What planet are these people living on? I understand that most get their information from certain trusted sources, which in my opinion they have been corrupted and when reality, as always happen, insist on imposing itself on us, we resort to increasingly crazier explanations. Good luck with that.
 
  • Like
Reactions: KompuKare

moinmoin

Diamond Member
Jun 1, 2017
4,934
7,619
136
Foundry revenues to fall? So this means chips should vome down in price right?
Can go both ways. If lower prices entices more customers to order from foundries again chip could get cheaper in production cost (though whether consumer will see any of that...). But if that doesn't work foundries may have to actually raise the prices further to get sufficient return for running the factories.
 

Ajay

Lifer
Jan 8, 2001
15,332
7,792
136
It’s not looking good. Intel has suspended plans for a FAB in Marburg, Germany and an expansion in Oregon. Samsung logic needs a new significant customer to sign up soon. There is just so much assistance the SK government can give to Samsung if their market share keeps spiraling down.
 

Doug S

Platinum Member
Feb 8, 2020
2,203
3,405
136

Foundry revenues to fall? So this means chips should vome down in price right?

No, it means they will invest less in future capacity. Maybe TSMC will undo the price increases they put in place a year ago (or was it two years ago?) in response to limited capacity, but don't expect big price cuts on logic wafers.

Now DRAM and NAND, that has been falling in price and will continue to fall, until the reduced future investment catches up and we're back in shortage again. That's the boom/bust cycle they've always gone through.
 

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
Can go both ways. If lower prices entices more customers to order from foundries again chip could get cheaper in production cost (though whether consumer will see any of that...). But if that doesn't work foundries may have to actually raise the prices further to get sufficient return for running the factories.
Go 1st principles.

Why is this happening? Falling sales are due to end customers getting squeezed financially, not due to lack of demand for new stuff. Every prediction claiming 2nd half rebound is delusional, so I guess we'll have to wait a few quarters before the dummies writing all those articles get their light bulbs finally switched on.

Some here with adequate incomes seem to have a lot of difficultly imagining an alternate existence. Raising prices to compensate for falling sales due to economic stress is, frankly, incomprehensible to me, and if the end buyer has no share in the pricefalls than what has changed to increase sales. If Nvidia, AMD, Intel, etc do not cut prices then sales do not rebound. If sales do not rebound, then why should TSMC cut prices? The only way sales rise is if prices fall for the end consumer as the world economic situation is not improving soon. Raising prices further decreases sales. A poor decision if taken.

But after all, what do I know. Cheers.
 
  • Like
Reactions: KompuKare

insertcarehere

Senior member
Jan 17, 2013
639
607
136
Go 1st principles.

Why is this happening? Falling sales are due to end customers getting squeezed financially, not due to lack of demand for new stuff. Every prediction claiming 2nd half rebound is delusional, so I guess we'll have to wait a few quarters before the dummies writing all those articles get their light bulbs finally switched on.

Some here with adequate incomes seem to have a lot of difficultly imagining an alternate existence. Raising prices to compensate for falling sales due to economic stress is, frankly, incomprehensible to me, and if the end buyer has no share in the pricefalls than what has changed to increase sales. If Nvidia, AMD, Intel, etc do not cut prices then sales do not rebound. If sales do not rebound, then why should TSMC cut prices? The only way sales rise is if prices fall for the end consumer as the world economic situation is not improving soon. Raising prices further decreases sales. A poor decision if taken.

I think its a bit more complicated picture than that. Intrinsic demand for Silicon, in the form of CPUs+GPUs, has probably decreased/not increased as projected due to underlying demographic changes, namely an aging (but generally wealthier) population in higher-income countries who can generally buy these sorts of items. Not to generalize, but it shouldn't be controversial to say that demand in new computers (esp. gaming) is far higher among <35 year olds.

Compared to even 10 years ago, median age in high-income APAC countries (Japan/South Korea/Taiwan/HK/SG/China (arguably)) have all increased by roughly 5 years into the early/mid/late 40s. That is a large chunk of the market which is quickly aging away from buying new computers as a hobby. Western Europe is less severe in this regard but they are also "aging out" of the likely demand peak noticeably in the last decade. So unless demand perks up with the opening of another major market soon (India maybe), foundries and OEMs will increasingly choose to recoup R&D by charging higher prices moving forward.
 

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
I think its a bit more complicated picture than that. Intrinsic demand for Silicon, in the form of CPUs+GPUs, has probably decreased/not increased as projected due to underlying demographic changes, namely an aging (but generally wealthier) population in higher-income countries who can generally buy these sorts of items. Not to generalize, but it shouldn't be controversial to say that demand in new computers (esp. gaming) is far higher among <35 year olds.

Compared to even 10 years ago, median age in high-income APAC countries (Japan/South Korea/Taiwan/HK/SG/China (arguably)) have all increased by roughly 5 years into the early/mid/late 40s. That is a large chunk of the market which is quickly aging away from buying new computers as a hobby. Western Europe is less severe in this regard but they are also "aging out" of the likely demand peak noticeably in the last decade. So unless demand perks up with the opening of another major market soon (India maybe), foundries and OEMs will increasingly choose to recoup R&D by charging higher prices moving forward.
Good points.

Then we're in a positive feedback deathloop. Higher prices leading to less sales leading to higher prices. At some point a reevaluation has to happen and accept that changes in doing business are needed.

You either sell more often, charge more or a to a larger % of the population, to keep or increase sales volume. Companies seem to be departing that last part of the market as it's the most price sensitive and lowest margin segment. It's also the biggest volume that allows the sharing of R&D costs. Everyone seems to want to emulate Apple and go upmarket, not realizing that there's not enough room at the top. All that recent easy money has really done harm to thinking.

There is a truism being tested. The cure for high prices is high prices.
 
Last edited:

Doug S

Platinum Member
Feb 8, 2020
2,203
3,405
136
Why is this happening? Falling sales are due to end customers getting squeezed financially, not due to lack of demand for new stuff. Every prediction claiming 2nd half rebound is delusional, so I guess we'll have to wait a few quarters before the dummies writing all those articles get their light bulbs finally switched on.


Sorry I don't buy that. The reason sales are falling is because there was a huge burst of sales related to the pandemic. Lots of companies were buying laptops for employees so they could work from home. Schools were buying Chromebooks or laptops for remote learning. Server/cloud capacity needed to be beefed up to deal with these new types of remote loads. All those extra puchases were sales robbed from the future - a company who bought every employee a laptop in 2020 probably won't buy much of anything until 2025. Ditto for schools - many of which won't replace the Chromebooks at all because remote learning is a thing of the past.

Whether the economy will formally enter recession or not is irrelevant, if there is a recession it will be shallow, if there isn't then growth will be tepid. Interest rates will have a much bigger impact since those drive purchases of really expensive stuff that drives the economy far more than tech gizmos. It may be painful for some to have interest rates go up, but they are still below what was considered a normal level for decades until 9/11 instituted a 20 year period of abnormally low and often essentially zero interest rates. It is healthier for the economy in the long run to return to the mean, and hopefully we never see zero rates again.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
Utilization rate for Fab18 for N5 family is expected to fall below 70%. This additional to the drop in the utilization rate for Fab14/Fab15 for N7 family.
TSMC's capacity utilization for 5/4nm process nodes, which will fall to about 75% in the first quarter of 2023, may drop below 70% in the second quarter, according to industry sources.
TSMC already cut the planned investment of 44B to 36B.