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

 

[mikegg]

Musk, OpenAI, & US govt to invest further in Intel ???

(Musk complains that chips have become a bottleneck for Tesla)

https://twitter.com/x/status/1986585317995171980

Musk isn't wrong that chip demand will only grow from here. It's not going to slow down. I can't see a future world where we need fewer chips than in 2025.

Him saying that they need to build a cutting edge chip factory that produces as much as TSMC Is just him pumping Tesla stocks. He knows he can't do it. Building an electrical car is child's play compared to a cutting edge fab.

However, I can see him buying out Intel completely or merging with Intel in some way.

 

[Thibsie]

Musk isn't wrong that chip demand will only grow from here. It's not going to slow down. I can't see a future world where we need fewer chips than in 2025.

Him saying that they need to build a cutting edge chip factory that produces as much as TSMC Is just him pumping Tesla stocks. He knows he can't do it. Building an electrical car is child's play compared to a cutting edge fab.

However, I can see him buying out Intel completely or merging with Intel in some way.

Musk buying Intel is the end of Intel as we know it. Death of x86 competition with AMD IMO.

 

[Josh128]

Musk buying Intel is the end of Intel as we know it. Death of x86 competition with AMD IMO.

Why would you say this with Musks track record of consistently producing incredible manufacturing and engineering results in the companies he runs? Even Jensen was amazed at the speed at which
xAI built their compute cluster and says Elon gets sh*t done. Whether or not you like the man for his personal/political views is irrelevant. His results speak for themselves.



Jensen on xAI

 

[Thibsie]

Lol, really?
If only 'cos opinions might get in the way (as happened to Tesla)
Second, 'cos he doesn't know sheet about semiconductors.
Third 'cos he would keep it for himself and his businesses and absolutely not for our greater good.

You really think Musk would run a company providing components for others?
Think again.

 

[coercitiv]

Why would you say this with Musks track record of consistently producing incredible manufacturing and engineering results in the companies he runs?

Because you judge his track record by ignoring the things he does to sabotage some of his own companies. Example below from a few years ago:

According to Isaacson, Musk became impatient with the leadership at Twitter when they explained to him that it would take careful planning and time to move more than 700 servers from Sacramento, California to a much cheaper facility in Portland, Oregon. Instead of waiting for an orderly process to be put together, though, Musk decided to hastily move the servers with a motley crew consisting of himself, his cousins, a smattering of Tesla and SpaceX staff, and a moving company whose staff wanted to be paid in cash. Musk was so impatient that at one point he shimmied himself under a server floor and used a pen knife to open up an electrical cabinet in order to unplug the server.

They managed to move the servers in three days, but their methods were extremely unorthodox. They didn’t wipe the equipment of personal data before the move nor did they swaddle the servers in protection, horrifying facility workers.

“For the next two months, X was destabilized,” Isaacson wrote.

“In retrospect, the whole Sacramento shutdown was a mistake,” Musk told Isaacson. “I was told we had redundancy across our data centers. What I wasn’t told was that we had 70,000 hard-coded references to Sacramento. And there’s still shit that’s broken because of it.”

Maybe you know the "what and why" happened back then, I'll write a short summary here for everyone else as well: Musk bought Twitter in large part with his own money, some other investors, and around $13B in debt that got offloaded to Twitter. This meant that after the acquisition Twitter found itself in a very difficult financial situation, which got exponentially worse as Musk wreaked havoc on the social platform, scaring off advertisers. The company was drowning on the loan pressure alone, so Musk effectively gutted everything he could find to lower spending: from personnel to server costs. The infrastructure was in such bad shape that he could not use it to support republican candidates via live streams. (he tried, it failed)

Now you could argue he eventually saved the company, but don't forget he was the pathogen who almost killed it and left it disfigured, a former shadow of itself.

Fake promises and knee-jerk reactions don't work in semi manufacturing. Intel tried this for a decade now, and the results speak for themselves.

And there’s still shit that’s broken because of it.

 

[johnsonwax]

Musk isn't wrong that chip demand will only grow from here. It's not going to slow down. I can't see a future world where we need fewer chips than in 2025.

But more chips doesn't mean more money spent on chips. It's an aggressively deflationary market in the sense that the cost of a compute unit falls really quickly. That means there's a path there where you are increasing compute usage but decreasing money spent on that compute. That really complicates how the industry finances work, especially in light of AI which is adding a FOMO premium.

Him saying that they need to build a cutting edge chip factory that produces as much as TSMC Is just him pumping Tesla stocks. He knows he can't do it. Building an electrical car is child's play compared to a cutting edge fab.

However, I can see him buying out Intel completely or merging with Intel in some way.

I mean, Musk could borrow against his future trillion dollar paycheck and buy Samsung - their market cap is only half a trillion. The AI market is nothing if not self-fulfilling via recursive financing.

 

[DrMrLordX]

I mean, Musk could borrow against his future trillion dollar paycheck and buy Samsung - their market cap is only half a trillion. The AI market is nothing if not self-fulfilling via recursive financing.

S. Korea would not allow the sale. That's their baby.

 

[jdubs03]

I mean, Musk could borrow against his future trillion dollar paycheck and buy Samsung - their market cap is only half a trillion. The AI market is nothing if not self-fulfilling via recursive financing.

You speak of this like it is a forgone conclusion.
When has Elon Musk delivered on any of his roadmaps for the last 6 years?

 

[Thunder 57]

Lol, really?
If only 'cos opinions might get in the way (as happened to Tesla)
Second, 'cos he doesn't know sheet about semiconductors.
Third 'cos he would keep it for himself and his businesses and absolutely not for our greater good.

You really think Musk would run a company providing components for others?
Think again.

While I have no faith in this "Terrafab" and think he is talking out of his ass, the not use things for the greater good is just BS. SpaceX has put NASA to shame with reusable rockets and looks to do the same thing compared to the super expensive SLS. And Starlink is used by people around the world and other countries are racing to launch their own alternatives.

He gets a lot of shit because of his politics and that's fine. Doesn't discredit everything that's related to him though.

 

[Thunder 57]

@coercitiv

“In retrospect, the whole Sacramento shutdown was a mistake,” Musk told Isaacson. “I was told we had redundancy across our data centers. What I wasn’t told was that we had 70,000 hard-coded references to Sacramento. And there’s still shit that’s broken because of it.”

I can't speak to moving the servers as I don't know about it but I will say the above is comically stupid.

 

[QuickyDuck]

Nah, fabs aren't rocket science /s

 

[Doug S]

While I have no faith in this "Terrafab" and think he is talking out of his ass, the not use things for the greater good is just BS. SpaceX has put NASA to shame with reusable rockets and looks to do the same thing compared to the super expensive SLS. And Starlink is used by people around the world and other countries are racing to launch their own alternatives.

He gets a lot of shit because of his politics and that's fine. Doesn't discredit everything that's related to him though.

SpaceX is the ONLY thing he's attached to that isn't hyped to death with ridiculous promises, like the "Tesla will drive from SF to NYC on its own in 2017" that still hasn't happened and isn't close to happening. Maybe Gwynne Shotwell is good at keeping him at arms length and letting the engineers there engineer without him getting in the way?

The only "greater good" I see him going for is his own. If he was interested in the "greater good" he wouldn't be ramming through a $1 trillion compensation package for himself at Tesla. Pretty sure he's rich enough already. If it really was just about wanting more "control" he'd have his compensation come via the creation of a new class of share that gives him more voting more without more actual ownership, like Zuck did at Facebook to insure he would always remain in control of it. But he didn't do that, because he thinks he's going to win at life if he becomes the first trillionaire.

 

[Joe NYC]

Because you judge his track record by ignoring the things he does to sabotage some of his own companies. Example below from a few years ago:



Maybe you know the "what and why" happened back then, I'll write a short summary here for everyone else as well: Musk bought Twitter in large part with his own money, some other investors, and around $13B in debt that got offloaded to Twitter. This meant that after the acquisition Twitter found itself in a very difficult financial situation, which got exponentially worse as Musk wreaked havoc on the social platform, scaring off advertisers. The company was drowning on the loan pressure alone, so Musk effectively gutted everything he could find to lower spending: from personnel to server costs. The infrastructure was in such bad shape that he could not use it to support republican candidates via live streams. (he tried, it failed)

Now you could argue he eventually saved the company, but don't forget he was the pathogen who almost killed it and left it disfigured, a former shadow of itself.

Fake promises and knee-jerk reactions don't work in semi manufacturing. Intel tried this for a decade now, and the results speak for themselves.

BTW, traits of successful entrepreneurs and CEO are sense of urgency and impatience.

Upside typically outweighs downside, as Mush demonstrated in that stunt. Service disruptions were minimal / almost unnoticeable, and cost savings were astronomical. Like 75% cost savings.

Twitter is now generating $1.25 billion in profits (although most of it is consumed by acquisition related interest payments). Prior to acquisition, in 2021, Twitter was losing $221 million per year.

That's ~1.5 billion turnaround in profitability while the revenue (due to various boycotts and sabotage) is down 50%.

That's nothing short of legendary success story.

 

[MoistOintment]

Twitter is now generating $1.25 billion in profits (although most of it is consumed by acquisition related interest payments). Prior to acquisition, in 2021, Twitter was losing $221 million per year.

That's ~1.5 billion turnaround in profitability while the revenue (due to various boycotts and sabotage) is down 50%.

Do you have a source on that? My understanding is that Twitter was merged with xAi, so how do we know the standalone finances of "Twitter" any more?

 

[Joe NYC]

Do you have a source on that? My understanding is that Twitter was merged with xAi, so how do we know the standalone finances of "Twitter" any more?

Most of the sources cite info from 2 paywalled Bloomberg articles:

Some estimates for 2025 revenue are up $200m from $2.7 to $2.9 billion

https://twitter.com/x/status/1888194818544918614

 

[Josh128]

RE: The whole "Elon sucks" narrative, Huang said what he said about him, it is what it is, thats not up for debate. As far as how hypothetically taking a stake in Intel could go, it seems to me that some of the main problems at Intel of the last 10 years was extreme bloat, too much jockeying and work politics, and lack of focus on their bread and butter, and the hubris that they could never be touched. IMO Musk's management style would be perfect to deal with the first three issues, at least. I'd bet anyone here that Intel stock would skyrocket if news broke that Musk purchased a large portion of the stock and was given a seat on the board of directors.

 

[511]

RE: The whole "Elon sucks" narrative, Huang said what he said about him, it is what it is, thats not up for debate. As far as how hypothetically taking a stake in Intel could go, it seems to me that some of the main problems at Intel of the last 10 years was extreme bloat, too much jockeying and work politics, and lack of focus on their bread and butter, and the hubris that they could never be touched. IMO Musk's management style would be perfect to deal with the first three issues, at least. I'd bet anyone here that Intel stock would skyrocket if news broke that Musk purchased a large portion of the stock and was given a seat on the board of directors.

Tan already fixed the politics and messing around issue he cut bloat as well

 

[marees]

Samsung getting crypto mining ASIC orders...

Samsung Begins Closing Gap with TSMC, Wins 2nm Order from Chinese Mining Firm [Kang Hae-ryeong's Tech & the City]​

Posted 2025.11.16 16:26 Modified 2025.11.16 16:31

According to the semiconductor industry on the 16th, Samsung’s foundry division has received production orders for 2 nm application-specific integrated circuits (ASICs) from Chinese cryptocurrency mining companies MicroBT and Canaan. ​

​

MicroBT and Canaan are the No. 2 and No. 3 players in the global mining machine manufacturing market. The No. 1 player, China’s Bitmain, works with TSMC. It is reported that MicroBT and Canaan turned their eyes to Samsung Foundry after finding it difficult to continue working with TSMC, which is overflowing with orders.​

​

Samsung Electronics has already begun production of MicroBT’s ordered volumes. For Canaan’s mining-machine chips, the company plans to load the first 2 nm wafers (300 mm, 12-inch) into the production line early next year and start deliveries in the second half of the year. The two companies’ chips will be produced at Samsung Foundry’s S3 line in Hwaseong, Gyeonggi Province, which is equipped with 2 nm facilities.​

​

The delivery volume for the two companies is known to be around 2,000 300 mm (12-inch) wafers per month. This corresponds to about 10% of Samsung Electronics’ total 2 nm production capacity.​

​

Samsung Electronics plans to ramp up investment in the 2 nm line at its Taylor plant in Texas, its U.S. production base. Starting in the second quarter of next year and running through 2027, the company plans to bring in equipment that will allow it to produce more than 15,000 2 nm wafers per month.​

​

Regarding the 2 nm contracts with MicroBT and Canaan, Samsung Electronics stated, “We cannot confirm details about our customers.”​

[ë¨ë] ì¼ì±, TSMCì 격차 ì¢í기 ìëâ¦ä¸ ì±êµ´ìì²´ì '2ëë¸' ì주 [ê°í´ë ¹ì íí¬ì¤ëìí°]

[ë¨ë] ì¼ì±, TSMCì 격차 ì¢í기 ìëâ¦ä¸ ì±êµ´ìì²´ì '2ëë¸' ì주 [ê°í´ë ¹ì íí¬ì¤ëìí°] , ê°í´ë ¹ 기ì, ê²½ì

www-hankyung-com.translate.goog

https://twitter.com/x/status/1989961169696698794

 

[511]

>Random chinese mining firm
> closing gap with TSMC
Modern Journalism

 

[DavidC1]

Samsung's the worst in naming process lol. "2nm" is a joke. Also, Intel had cryptomining chips and it had almost no impact.

Tan already fixed the politics and messing around issue he cut bloat as well

Tan has nothing on cutting out bloat compared to Musk. He'd probably cut 70% in the first quarter.

His way of running is chaotic but it seemingly works in the end. Tesla single handedly made electric cars viable for mass use, when previous to that it was a joke that was barely above electric bicycles. They had problems of automating too much for a while and Tesla learned to make a better balance of it. It's learning-as you go approach and there's little alternative to it when you are at the forefront of things.

Tesla is actually similar to Apple in a way where they had an idea, and no ecosystem existed, so they created the ecosystem to fit their idea. If you read about factories and processes for the iPhone and then megabattery facilities for Tesla, it's similar. Jobs didn't have the best reputation among employees either.

BTW, traits of successful entrepreneurs and CEO are sense of urgency and impatience.

This actually goes hand-in-hand with what WE expect of semiconductors and technology in general. We want the next CPU cores to be much better right? That requires tearing down and putting new stuff in, things that wasn't tried before. Now that isn't conducive to peaceful, serene living for the engineers and management, but that's the tradeoff you get.

You look at other sectors and it's the same. Look at the pictures for the core developers for CD Projekt Red. Between Witcher 1, 2, and 3, 2/3rds of the faces change. Chaotic. So chaos itself isn't an indication of lack of success. Because having all the details and precision in the product within a certain timeframe comes with sacrifices on behalf of the employees. So if you see the employees getting good compensation, and is comfortable with all the amenities, some sacrifices in quality may be seen.

Now I'm not excusing Musk. Because it seems we're coming to a point where we're having to sacrifice humanity for the sake of tech. It's like, "Is it worth it?"

 

[511]

Tan has nothing on cutting out bloat compared to Musk. He'd probably cut 70% in the first quarter.

Intel is not a small company manufacturing business has comparatively higher employees and you don't want to kill good talent

 

[DavidC1]

Intel is not a small company manufacturing business has comparatively higher employees and you don't want to kill good talent

TSMC is doing relatively well, but I don't think they are the epitome of that. They have 80K employees with much bigger 3rd party customer count. Intel can probably do it for 30K on that metric alone. Extract efficiency, and maybe 20K. Then you have 10K for everything else. Bare minimum just to make things work. By having less people, you communicate faster, cut down the garbage as well.

Like read about how efficient companies can be. Cyrix had 1/3rd of the development time. Made x86 processors for $10 million. But they really ran on bare minimum. Assembled test chips by hand. Ran simulations on overclocked, off the shelf, desktop platforms, rather than overpriced servers. They had 70 or so employees total, and 50 core engineering group. They quoted one guy that "would do the work of ten". On Youtube you have some guys doing reballing DRAM chips by hand, one solder ball by one solder ball using tweezers. The amount of patience and persistence is insane.

 

[Doug S]

So today Musk is out there complaining that he's being told it takes five years from start to finished wafers being output for a megafab, and that's the reason he thinks he needs to build his own fabs. It is laughable that his ketamine soaked brain thinks he'd be able to shorten that timeline. Maybe if it was factory stamping out something relatively simple like Cybertruck body panels I could see him being willing to overpay to compress the schedule and beat what third parties are telling him. But the idea he could build a leading edge fab faster than TSMC or Samsung, something he knows zip about and has only three players in the entire world - two of whom are falling behind... I mean, its just comical. I guess he's surrounded himself with so many yes-men there is no one to tell him what a clown he looks like saying stuff like this.

So long as he doesn't get a dime of taxpayer money I hope he tries. It will provide years of entertainment laughing at his abject failure, and maybe people will quit thinking he's some sort of uber genius who can do no wrong.

 

[johnsonwax]

So today Musk is out there complaining that he's being told it takes five years from start to finished wafers being output for a megafab, and that's the reason he thinks he needs to build his own fabs. It is laughable that his ketamine soaked brain thinks he'd be able to shorten that timeline. Maybe if it was factory stamping out something relatively simple like Cybertruck body panels I could see him being willing to overpay to compress the schedule and beat what third parties are telling him.

Cybertruck shipped 2 years late. The easiest thing in the world to do is the thing you know the least about.

 

[DavidC1]

So today Musk is out there complaining that he's being told it takes five years from start to finished wafers being output for a megafab, and that's the reason he thinks he needs to build his own fabs.

Intel themselves said that taking 24 months for a new process node was a choice motivated based on profit. Then gave comparisons to different timeframes.

12 months, and 36 months. 12 months is suboptimal because you aren't fully taking advantage of the revenue, and 36 months is well after the optimal profit point. Hence, they chose 24 months for new process technology. Intel took 3.5 years to go from Intel 7 to 18A. As chaotic it was, speeding up is to a point, possible.

He doesn't have technical knowledge about electric cars either. Yet here we are with Tesla, that single-handedly made electric cars viable to masses. Previous to that electric cars barely got 100km, and couldn't outperform a Prius, with a design that was solely optimized for weight and looked like a electric bicycle with a hood. Delays and problems they are facing doesn't change the fact his company achieves things that others couldn't. Yes, not all his claims and goals come to fruition. Yet maybe it's that ambition and dreams that allows it to get ahead of others, because others have much less of a target to attain.