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

DisEnchantment · Apr 24, 2020

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.

---------------------------------------------------------------------------------------------------------------------------------------------------

FEEL FREE TO CREATE A NEW THREAD FOR 2025+ OUTLOOK, I WILL LINK IT HERE

Doug S · Mar 5, 2025

igor_kavinski said:
TSMC might be doing this deal as a deterrent to any geopolitical disaster. Let's say worst comes to worst.

By the time that happens, TSMC figures they have at least one leading edge fab in the US so they have no problem destroying anything of value in Taiwan. Maybe part of the deal is that the best TSMC researchers, scientists and engineers can be transferred to the US overnight as soon as signs of a worsening geopolitical situation appear. So now the invader not only loses the equipment but also the people to build that stuff back up.

The invader "enjoys" being the dictator in their extended province without any economic profit. So with these precautions in place, would it be worth the gargantuan cost of an invasion with almost nothing to gain in return? Maybe they end up deciding it would be useless in practice and that might exactly be the strategy here.

That's only true so long as Trump (or someone who thinks like him) is in the White House. It is probably reasonable to assume that he would abandon Taiwan just like he's abandoning Ukraine, and China would be able to walk in.

So game theory probably says at least PLANNING fabs, or building shells, makes sense. But maybe they never get equipped, if they are mostly intended as a backup plan to move TSMC equipment to if it looks like China is going to try to retake Taiwan (remember, there will not be an "invasion", they would institute a naval blockade of Taiwan) If they don't need them they could sell the shells to Intel or Micron.

DrMrLordX · Mar 5, 2025

Doug S said:
So game theory probably says at least PLANNING fabs, or building shells, makes sense. But maybe they never get equipped, if they are mostly intended as a backup plan to move TSMC equipment to if it looks like China is going to try to retake Taiwan (remember, there will not be an "invasion", they would institute a naval blockade of Taiwan) If they don't need them they could sell the shells to Intel or Micron.

The first fab they built in AZ is already getting yields on N4-family nodes higher than what they get in Taiwan:

TSMC Arizona Semiconductor Yields Surpass Taiwan Facilities

TSMC’s Arizona semiconductor fab exceeds Taiwan yields, marking a pivotal step in U.S. chip manufacturing and tech independence.

www.electropages.com

Sorry if someone already pasted this link, but hey maybe we need to remember that these Arizona fabs will be useful irrespective of who does what geopolitically.

511 · Mar 5, 2025

DrMrLordX said:
The first fab they built in AZ is already getting yields on N4-family nodes higher than what they get in Taiwan:

TSMC Arizona Semiconductor Yields Surpass Taiwan Facilities

TSMC’s Arizona semiconductor fab exceeds Taiwan yields, marking a pivotal step in U.S. chip manufacturing and tech independence.

www.electropages.com

Sorry if someone already pasted this link, but hey maybe we need to remember that these Arizona fabs will be useful irrespective of who does what geopolitically.

Well it was a 3-4 year old process the initial yield ramp and issues are ironed out in Taiwan

LightningZ71 · Mar 5, 2025

It also helps to be in a less seismically active area...

DrMrLordX · Mar 6, 2025

511 said:
Well it was a 3-4 year old process the initial yield ramp and issues are ironed out in Taiwan

True, but you'd think the improved yields would be mirrored in all operational facilities.

QuickyDuck · Mar 6, 2025

DrMrLordX said:
True, but you'd think the improved yields would be mirrored in all operational facilities.

New fab, new equipment, well calibrated. It should have better yields, shouldn't it?

fastandfurious6 · Mar 6, 2025

yes it's like "gen 2" N4 fab

also only 4% better yield

DrMrLordX · Mar 6, 2025

QuickyDuck said:
New fab, new equipment, well calibrated. It should have better yields, shouldn't it?

I guess? Maybe it's better equipment.

511 · Mar 6, 2025

DrMrLordX said:
True, but you'd think the improved yields would be mirrored in all operational facilities.

That is what I want to know what the benchmark was at what point of time

igor_kavinski · Mar 6, 2025

DrMrLordX said:
True, but you'd think the improved yields would be mirrored in all operational facilities.

Or it's just a PR statement to make the US look great for more projects of such nature.

511 · Mar 6, 2025

igor_kavinski said:
Or it's just a PR statement to make the US look great for more projects of such nature.

Yeah likely

fastandfurious6 · Mar 6, 2025

Intel reportedly sticking with TSMC as part of new long-term partnership strategy amid underwhelming 18A yields

Intel is seemingly shifting its strategy of no longer outsourcing wafers to third parties to instead embracing a longer partnership with TSMC. This could be due to the unfavourable yields from Intel’s own 18A process.

www.notebookcheck.net

Intel still bleeding money with no end at sight

GTracing · Mar 6, 2025

fastandfurious6 said:
Intel reportedly sticking with TSMC as part of new long-term partnership strategy amid underwhelming 18A yields

Intel is seemingly shifting its strategy of no longer outsourcing wafers to third parties to instead embracing a longer partnership with TSMC. This could be due to the unfavourable yields from Intel’s own 18A process.

www.notebookcheck.net

Intel still bleeding money with no end at sight

Clickbait. The article says that Panther Lake has poor yields and it's been pushed back to 2026. The earnings call directly contradicts both those points.

Panther Lake is on track to launch in the second half of this year. That launch date has not changed. We feel really good about the progress that we are making. In fact, if you look at where our yields are on Panther Lake today. They're actually slightly ahead at a similar point in time to Meteor Lake.

The other major point in the article is valid though, Intel is using TSMC wafers long term.

511 · Mar 7, 2025

They did not answer the % considering they went full on TSMC and it was only 30% wafer for them.
I assume they would want it less than 30% as for outsourcing they have always outsourced to TSMC

511 · Mar 9, 2025

Here is the TLDR

Key Takeaways

Intel aims to achieve operational profit breakeven for its foundry business by the end of 2027.
The company plans to monetize assets like Ultera and Mobileye to balance capital needs.
Intel’s 18A technology milestones include the launch of Panther Lake in the second half of the year.
Intel is focusing on AI acceleration with Jaguar Shores and advanced packaging capabilities.
The company expects PC unit growth of 3% to 5% this year, marking the best post-COVID growth.

Financial Results

Intel revised its CapEx guidance for the year from $20-23 billion down to $20 billion.
The company has over $50 billion in assets under construction, including equipment and infrastructure.
Intel’s deployed CapEx will increase year-over-year, exceeding $20 billion.
The transition to Intel 18A is expected to triple the ASP per wafer increase compared to cost per wafer.

Operational Updates

CEO transition: Michelle has more decision-making power regarding the product roadmap and TSMC wafer production.
Industrial policy: Intel aligns with U.S. manufacturing interests, investing heavily in domestic manufacturing and R&D.
Process technology: Panther Lake is set to launch later this year, with yields surpassing those of Meteor Lake.
Server products: Clearwater Forest’s launch is delayed to next year due to packaging issues, with samples shipping later this year.

Future Outlook

Intel aims to become the second-largest external foundry by 2030, targeting $15 billion in external foundry revenue.
The company anticipates 3% to 5% PC unit growth this year, the best post-COVID growth.
Intel expects 2024 to stabilize its market share in the server market.
AI strategy focuses on Jaguar Shores for AI acceleration and solutions across the compute continuum.

Q&A Highlights

Intel is addressing foundry challenges, including expectation setting and timing for Intel 18A’s PDK release.
The company is enhancing its foundry expertise by hiring experienced personnel from other foundries.
Intel Foundry Day will be hosted on April 29 at the San Jose Convention Center to connect with the ecosystem.

Here is the entire transcript for the call

https://www.investing.com/news/transcripts/intel-at-morgan-stanley-conference-strategic-focus-on-foundry-and-ai-93CH-3913018b

Doug S · Mar 9, 2025

511 said:
Intel aims to achieve operational profit breakeven for its foundry business by the end of 2027.

That's nice, but so long as the foundry is under the same corporate umbrella as x86 we can't know whether that's been achieved. The foundry and the x86 side aren't conducting real price negotiations until they are legally separate corporations.

511 said:
Intel aims to become the second-largest external foundry by 2030, targeting $15 billion in external foundry revenue.

They don't think they can even beat Samsung, despite how badly Samsung has bungled pretty much every process in the EUV era? OK I admit I don't know how much legacy foundry business Samsung has with 10+ year old nodes with may make the difference (since Intel's plan to acquire Tower Semi was blocked) So what I'd really be interested in is not total foundry revenue but foundry revenue for say 22nm and newer. If they can't beat Samsung on that metric by 2030 given their inbuilt advantages with government/defense contracts and all the tariff chaos we're seeing that's not good.

511 · Mar 9, 2025

Doug S said:
That's nice, but so long as the foundry is under the same corporate umbrella as x86 we can't know whether that's been achieved. The foundry and the x86 side aren't conducting real price negotiations until they are legally separate corporations.

They are already kind of separated and as for price negotiation they need to have another customer to negotiate 🤣

Doug S said:
They don't think they can even beat Samsung, despite how badly Samsung has bungled pretty much every process in the EUV era? OK I admit I don't know how much legacy foundry business Samsung has with 10+ year old nodes with may make the difference (since Intel's plan to acquire Tower Semi was blocked) So what I'd really be interested in is not total foundry revenue but foundry revenue for say 22nm and newer. If they can't beat Samsung on that metric by 2030 given their inbuilt advantages with government/defense contracts and all the tariff chaos we're seeing that's not good.

I mean Samsung counts internal volume as foundry revenue if Intel counts internal products as revenue they are already number 2 but yes I agree with you if they don't achieve it it should be a shame.

jdubs03 · Mar 10, 2025

Thoughts?

https://twitter.com/x/status/1899025325905871344

Count me as skeptical on the timeframe still since it’s such a step change with not that much time in R&D, and it easily could be homegrown hype. But I also saw Huawei has been hiring industry folks from the big names (not sure on how many but that would certainly help). When taking that into account compared to what I was thinking prior, they’re a few years ahead of the timeframe I was.

LightningZ71 · Mar 10, 2025

My read on this...
trial production this year. Small batch fine tuning production through next year. Volume production in 2027 for their first installation, but limited to a couple of product lines with limited wafer throughput. They'll start churning refined machine designs out in 2027 and hit notable volume levels in 2028/2029. TSMC first volume production of EUV nodes was N7+ in 2019. They'll beat the supposed 10 year head start by a year or two. They'll go for High NA EUV ASAP after that.

Not to be overly political, but, keep in mind that China has publicly stated that they want to be ready to go for any sort of Taiwan operation by 2027. Getting EUV up and running by that point means that they are less likely to care what happens to TSMC's production facilities. This does not bode well for the industry.

511 · Mar 10, 2025

it's not possible this decade it will be possible maybe next decade?

DZero · Mar 10, 2025

jdubs03 said:
Thoughts?

https://twitter.com/x/status/1899025325905871344
Count me as skeptical on the timeframe still since it’s such a step change with not that much time in R&D, and it easily could be homegrown hype. But I also saw Huawei has been hiring industry folks from the big names (not sure on how many but that would certainly help). When taking that into account compared to what I was thinking prior, they’re a few years ahead of the timeframe I was.

Even if the production ammount is not high, it will be a deadly blow to TSMC and USA.

Doug S · Mar 10, 2025

China hasn't even produced their own DUV machines advanced beyond the state of the art from 2010, and we're supposed to expect they can make EUV? Color me skeptical.

Anyone who followed the twisty path of EUV in the 10+ years it took to get from lab prototype to production worthy will recall how many potential shoals that ship can run up against. If they don't have a powerful enough light source they could have yields at 100% and it still wouldn't be useful for mass production because the exposure times would be so long they'd be processing a single digit number of wafers per hour.

LightningZ71 · Mar 10, 2025

With effectively unlimited capital for investment, they could realistically just go wide and build lots of even subpar machines, speed of production wise, to hit their desired throughput goals.

DrMrLordX · Mar 11, 2025

DZero said:
Even if the production ammount is not high, it will be a deadly blow to TSMC and USA.

Not really. In the timeframe that they could bring such machines online, cutting edge will already be on High NA EUV.

511 · Mar 11, 2025

It's just not about the machines what about the chemicals and all other stuff that is only possible due to collaboration in semi industry.
It is near impossible to make everything in house.

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

Golden Member

Diamond Member

Lifer

Golden Member

Platinum Member

Lifer

Member

Senior member

Lifer

Golden Member

Lifer

Golden Member

Senior member

Senior member

Golden Member

Golden Member

Key Takeaways​

Financial Results​

Operational Updates​

Future Outlook​

Q&A Highlights​

Diamond Member

Golden Member

Golden Member

Platinum Member

Golden Member

Senior member

Diamond Member

Platinum Member

Lifer

Golden Member

Key Takeaways

Financial Results

Operational Updates

Future Outlook

Q&A Highlights