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

Page 43 - 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
1,601
5,779
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
15,427
7,846
136
Standard business things. As noted Intel would probably get banned in China if they went ahead with the deal without China's approval.
Well, this is a sticky wicket:
Intel's main problem with the acquisition of Tower Semiconductor is the stance of China's State Administration for Market Regulation (SAMR), which suspended the clock in its review of the transaction in January. Intel is about to get grants from the U.S. government's CHIPS and Science fund on the condition that it will not invest in China for the next 10 years after the receival. Meanwhile China wants Intel to keep investing in Chinese assets and is trying to leverage this by potentially blocking the transaction.

Intel's future is pretty much stuck between the foreign affairs desires of two countries in an economic war.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Hmm, why does China have to approve? I didn’t think Tower had any fabs in China.

China has such influence in the American market that a certain American gaming company banned someone from competing just by saying something Pro-Taiwan.

Granted, that particular company had been retards for a while, but it shows you a glimpse of things, to come.
 

Joe NYC

Golden Member
Jun 26, 2021
1,930
2,269
106
Its ~35% of Intel revenue at stake which AMD will happily pick up, they wont risk getting kicked out of that market for non compliance.

Intel is between the rock and the hard place here. Last time Intel did something that offended China, which was some sort of ban of buying items from a controversial Chinese province. Intel reversed that within days, after thee heard from Chinese officials.

Sacrificing China sales would be nearly fatal to Intel. Scaling down the Foundry Services (by passing on Tower acquisition). that is supposed to save Intel's manufacturing arm is disadvantageous, but not fatal.
 
  • Like
Reactions: BorisTheBlade82

Ajay

Lifer
Jan 8, 2001
15,427
7,846
136
Intel is between the rock and the hard place here. Last time Intel did something that offended China, which was some sort of ban of buying items from a controversial Chinese province. Intel reversed that within days, after thee heard from Chinese officials.

Sacrificing China sales would be nearly fatal to Intel. Scaling down the Foundry Services (by passing on Tower acquisition). that is supposed to save Intel's manufacturing arm is disadvantageous, but not fatal.
IDK. Pretty close to fatal. Intel desperately needs higher wafer output to remain competitive. Developing leading edge process nodes on a stagnant or declining wafer production will increase their R&D costs per 100k wafers to unsustainable levels They’ll just get priced out of the market compared to TSMC and wind up having to go the way of GloFo.
 

A///

Diamond Member
Feb 24, 2017
4,352
3,154
136
not a surprised. everything is down at the mo. how long it lasts is anyone's guess and most economists have something you go toilet with for brains and talk rubbish.
 

Ajay

Lifer
Jan 8, 2001
15,427
7,846
136
FWIW TSMC's revenue was down 15.4% YoY in March. Jan and Feb were solidly up so that's quite a change.
Global economic downturn, inflation, soft demand for electronics (especially phones and PCs). Supposedly, hyper-scalers are in a 'digestion' mode. Fun times.
 

Doug S

Platinum Member
Feb 8, 2020
2,249
3,479
136
The sudden collapse of SVB and people waiting around to see if there are other shoes to drop has caused
Global economic downturn, inflation, soft demand for electronics (especially phones and PCs). Supposedly, hyper-scalers are in a 'digestion' mode. Fun times.


Don't forget the crypto crash affecting both bitcoin ASICs and GPU sales, both of which consume a lot of high revenue leading edge wafers. We haven't even seen the potential issues from the SVB failure tightening loans (the first signs of that on the commercial real estate market started appearing in the last week of March)
 

A///

Diamond Member
Feb 24, 2017
4,352
3,154
136
economy was already in free fall, svb's failure and acquisition accelerated everything. there's some unintended benefits to this freefall. Construction supply prices have gone up but raw materials have come down with exception of wood and premade concreate mix. My estimates for ordering 5 tons of nutrified top soil have come down.
 

DrMrLordX

Lifer
Apr 27, 2000
21,616
10,824
136
Looks like Intel Foundry Service and ARM are working together to bring ARM IP to IFS on 18a, starting out with mobile processor cores

That's certainly interesting. Does that mean Intel can "design" (read: implement reference design(s) provided by ARM) an ARM core for a customer to use on 18a?
 
Jul 27, 2020
16,145
10,232
106
Introducing the Intel ARMageddon series of processors.

Low power (max TDP 500W. You can re-use your old 1kW PSU!)

An array of cores on the same die (low power, medium power, high power. We got em all!)

ARM and x86 compatibility at the same time in the same OS!

Fanless (if you don't mind running radiator heatpipes from your PC to your window)

GPU not included (sorry, we ran out of die space)
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
That's certainly interesting. Does that mean Intel can "design" (read: implement reference design(s) provided by ARM) an ARM core for a customer to use on 18a?
Foundry customers generally want hard IP macros wherever possible. This is basically Intel and ARM announcing that they will work to offer hardened ARM IP on 18A. Really, this is a prerequisite for them to be taken remotely seriously in foundry. Having said that...
Note the date on this particular article.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,540
14,495
136
Foundry customers generally want hard IP macros wherever possible. This is basically Intel and ARM announcing that they will work to offer hardened ARM IP on 18A. Really, this is a prerequisite for them to be taken remotely seriously in foundry. Having said that...
Note the date on this particular article.
Does a 2016 article have any real meaning in 2023, 7 years later ?
 

Geddagod

Golden Member
Dec 28, 2021
1,148
1,005
106
Foundry customers generally want hard IP macros wherever possible. This is basically Intel and ARM announcing that they will work to offer hardened ARM IP on 18A. Really, this is a prerequisite for them to be taken remotely seriously in foundry. Having said that...
Note the date on this particular article.
Doesn't this also heavily suggest Intel 20A comes with HD cells? A 'full' node unlike Intel 4?
 

Exist50

Platinum Member
Aug 18, 2016
2,445
3,043
136
Does a 2016 article have any real meaning in 2023, 7 years later ?
That article is about an effectively identical announcement from the last time Intel tried courting foundry customers. Just pointing out that it's to be expected, and signifies little in and of itself.
Doesn't this also heavily suggest Intel 20A comes with HD cells? A 'full' node unlike Intel 4?
No? The announcement is for 18A. Intel's not offering 20A to IFS. Very likely that 20A has similar limitations as Intel 4. 18A is probably the Intel's first serious offering as part of IFS.
 

Geddagod

Golden Member
Dec 28, 2021
1,148
1,005
106
That article is about an effectively identical announcement from the last time Intel tried courting foundry customers. Just pointing out that it's to be expected, and signifies little in and of itself.

No? The announcement is for 18A. Intel's not offering 20A to IFS. Very likely that 20A has similar limitations as Intel 4. 18A is probably the Intel's first serious offering as part of IFS.
That article is about an effectively identical announcement from the last time Intel tried courting foundry customers. Just pointing out that it's to be expected, and signifies little in and of itself.

No? The announcement is for 18A. Intel's not offering 20A to IFS. Very likely that 20A has similar limitations as Intel 4. 18A is probably the Intel's first serious offering as part of IFS.
Oops, read it as 20A. That's mb.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,601
5,779
136
N5 --> N3 --> N2, 30% efficiency gain per gen. Not bad.
Interestingly N2 has better stats this year than last. By the time it launches in 2025 it will be another major bump over N3.

Quite an arsenal for TSMC, tons of capacity from N4P, N3E and N2 in 2025. Getting weary hearing about increasing wafer prices every gen .
Lets see a fab price war in 2024+. Who can sustain the longest?