Samsung talking smack about passing TSMC.

[ta152h]

Samsung Can Overtake TSMC In 5 Years Says Executive

Samsung believes that catching up to TSMC is possible in five years, and memory will overcome NVIDIA's AI GPUs. Take a look.

wccftech.com

They better not forget Intel ...

 

[blueone]

I think talking smack has become a thing in the chip industry. TSMC talks smack about how expensive it is to build & manufacture in the US. Intel talks smack about how it's going to achieve process superiority. IBM is claiming to have developed the world's first 2nm process. So Samsung is just following the crowd.

 

[ta152h]

I think talking smack has become a thing in the chip industry. TSMC talks smack about how expensive it is to build & manufacture in the US. Intel talks smack about how it's going to achieve process superiority. IBM is claiming to have developed the world's first 2nm process. So Samsung is just following the crowd.

Don't forget about Gelsinger talking smack about the geopolitical situation, and how it favored fabs in places other than threatened areas.

 

[blueone]

Don't forget about Gelsinger talking smack about the geopolitical situation, and how it favored fabs in places other than threatened areas.

Yeah, like Israel.

 

[Jert]

The world champion of smack talking does not live in Seoul. He lives 200km up north, in a place called Pyongyang

 

[Daniel Nenni]

Samsung Can Overtake TSMC In 5 Years Says Executive

Samsung believes that catching up to TSMC is possible in five years, and memory will overcome NVIDIA's AI GPUs. Take a look.

wccftech.com

They better not forget Intel ...

Another ridiculous article from a finance reporter.

"However, Samsung aims to close the technological gap between it and the Taiwanese firm by employing Gate All Around (GAA) transistors for the 3-nanometer process, the leading edge semiconductor node, in 2023."

I have not heard of one Samsung 3nm tape-out or even a design start. The EDA and IP companies seem a bit quiet as well. Samsung started 3nm production a year ago, right?

Samsung Begins Chip Production Using 3nm Process Technology With GAA Architecture | Samsung Semiconductor USA

Samsung is the one and only production of 3-nanometer process node, applying GAA transistor architecture in the world.

semiconductor.samsung.com

Samsung has yet to learn that trust is key in the foundry business.

 

[jorgequinonez]

The world champion of smack talking does not live in Seoul. He lives 200km up north, in a place called Pyongyang

Any ideas what kinds of fabs or process nodes Rocket Man has?

 

[ta152h]

Any ideas what kinds of fabs or process nodes Rocket Man has?

World's First Wood Transistor Chops Out at 1 Hz

It can't run Doom yet, but it's made from renewable materials.

www.tomshardware.com

Well, some scientists just released the first wood transistors. They currently run at 1 Hz. I'm sure with this breakthrough technology we'll be seeing more integrated circuits (and fewer trees) in North Korea. Plus, they might be suitable as eating utensils, if done right.

Naturally (forgive the pun), there will unexpectedly pertinent questions, like "Is that an orchard or a CPU?".

 

[Daniel Nenni]

Any ideas what kinds of fabs or process nodes Rocket Man has?

China might be helping him a little bit.

 

[Maximus]

Samsung Electronics to Invest KRW 133 Trillion in Logic Chip Businesses by 2030

Samsung Electronics, a world leader in advanced semiconductor technology, today announced that it will invest KRW 133 trillion by 2030 to strengthen its

news.samsung.com

Samsung at 2019 Apr: We will be No.1 in logic chips by 2030.
----
Reality:
2019 Q1 foundry market share: TSMC 48.1%; Samsung 19.1%
2022 Q4 foundry market share: TSMC 58.5%; Samsung 15.8%

 

[Daniel Nenni]

Samsung Electronics to Invest KRW 133 Trillion in Logic Chip Businesses by 2030

Samsung Electronics, a world leader in advanced semiconductor technology, today announced that it will invest KRW 133 trillion by 2030 to strengthen its

news.samsung.com

Samsung at 2019 Apr: We will be No.1 in logic chips by 2030.
----
Reality:
2019 Q1 foundry market share: TSMC 48.1%; Samsung 19.1%
2022 Q4 foundry market share: TSMC 58.5%; Samsung 15.8%

TSMC will be 60%+ market share in 2025 thanks to N3. If Intel Foundry get's their ecosystem and PDKs ship shape Samsung will fall even more.

 

[blueone]

Another ridiculous article from a finance reporter.

"However, Samsung aims to close the technological gap between it and the Taiwanese firm by employing Gate All Around (GAA) transistors for the 3-nanometer process, the leading edge semiconductor node, in 2023."

I have not heard of one Samsung 3nm tape-out or even a design start. The EDA and IP companies seem a bit quiet as well. Samsung started 3nm production a year ago, right?

Samsung Begins Chip Production Using 3nm Process Technology With GAA Architecture | Samsung Semiconductor USA

Samsung is the one and only production of 3-nanometer process node, applying GAA transistor architecture in the world.

semiconductor.samsung.com

Samsung has yet to learn that trust is key in the foundry business.

I can't think of another industry where so many people positioning themselves as experts talk about so much they really don't know much about.

 

[Mooredaddy]

TSMC will be 60%+ market share in 2025 thanks to N3. If Intel Foundry get's their ecosystem and PDKs ship shape Samsung will fall even more.

So IFS is a non threat to TSMC but a huge threat to Samsung logic? Am I understanding correctly?

 

[nghanayem]

In the short-medium term anyways. If Intel achieves it's goal of sustained performance, power performance leadership, then I think TSMC will bleed some (even if at 18A that bleeding is only a few drops) as fabless firms would for the first time since 14LPP have a viable choice launching within a similar TTM as TSMC (if you are generous about the initial 7LPP ramp I guess you can also count that as a viable competitor to N7 family when that was TSMC's latest and greatest). Dan has on a couple of occasions said something to the tune of "intel can easily steal the not TSMC/2nd place pity orders, but if they succeed in their goal they can also snag some of the premium HPC market away.". If IFS can build a somewhat robust ecosystem and PDKs then TSMC might end up bleeding a fair bit more beyond the 18A/N2 era. However I still expect that Samsung will keep "the close enough to TSMC and way cheaper market". They can accept lower margins, and they have their MASSIVE scale and memory experience to allow them to run their fabs more efficiently than probably anyone else on the planet.

If I had to pull out the crystal ball I would say that if intel succeeds in being the number 2 foundry by 2030. My guess is that in 2030 60-80% of the IFS business would have been from the usual crowd of Samsung customers (many of whom are now on TSMC as of N5/N3) and most of the balance would be from products that would have pre-Samsung imploding have been on TSMC (with small amounts of share stolen from SMIC/UMC/GF with i16).

 

[tooLongInEDA]

The world champion of smack talking does not live in Seoul. He lives 200km up north, in a place called Pyongyang

He's certainly in the top two - and arguably met with the other one - visiting from the US - a few years back.

 

[Markwrob]

I agree with the idea that IFS will take share from Samsung first, rather than from TSMC. Samsung's foundary customers are already more comfortable (though still probably queasy) with a potential competitor making their chips for them. That trust thing is a strong disincentive with most of TSMC's customers.

After awhile, if things go well at IFS, some of the holdouts will place a few orders. If for no other reason than gaining pricing leverage with TSMC. Then some more orders. The process will be gradual. Will IFS be able to keep itself supported long enough to become profitable given their current build plans? I surely do not know.

 

[ta152h]

I agree with the idea that IFS will take share from Samsung first, rather than from TSMC. Samsung's foundary customers are already more comfortable (though still probably queasy) with a potential competitor making their chips for them. That trust thing is a strong disincentive with most of TSMC's customers.

After awhile, if things go well at IFS, some of the holdouts will place a few orders. If for no other reason than gaining pricing leverage with TSMC. Then some more orders. The process will be gradual. Will IFS be able to keep itself supported long enough to become profitable given their current build plans? I surely do not know.

That's kind of a long distance view of things, based on the fact Samsung sucks and is more vulnerable. I get it.

But, looking closer, NVIDIA has a somewhat rocky relationship with TSMC, and has used Samsung not too long ago, despite being a deficient process node. And have complained pretty publicly about the cost of wafers from TSMC very recently. And announced they are very interested in working with Intel foundries. If NVIDIA moves things to Intel, where are they moving them from? Not Samsung.

I would also not be surprised if Apple did some parts at Intel as well. At least in an exploratory way, to ensure supply and possibly make some less strategic parts there. Qualcomm is already involved with Intel. AMD is safe, but do they actually sell anything anymore .

So, I think both are going to get hit a bit, if Intel is successful. Intel already doesn't suck like Samsung, their I7 parts are the fastest clocked products in that segment, and have no problems competing with AMD parts on N5. Samsung isn't really competitive at anything, except smack talking.

But, Intel has to be successful with their nodes, and while they have been on high performance nodes, their experience with making a node suitable for lower power/higher density is limited. No doubt they can do it, but can they do it competitively against TSMC? Not an easy bar to reach.

A lot of things can happen, I think it's premature to think TSMC will be largely unaffected. They may be, and Intel may be largely unsuccessful, but there are other scenarios that could play out too.

 

[GYHTBT]

Samsung has said the same thing repeatedly since entering the Foundry space in 2005.

 

[Maximus]

TSMC will be 60%+ market share in 2025 thanks to N3. If Intel Foundry get's their ecosystem and PDKs ship shape Samsung will fall even more.

第一季前十大晶圓代工業者營收季減近二成，第二季仍將持續下滑

TrendForce 研究顯示，受終端需求持續疲弱以及淡季效應加乘影響，第一季全球前十大晶圓代工業者營收季跌幅達 18.6%，約 273 億美元。本次排名最大變動為格羅方德（GlobalFoundries）超越聯電（UMC）拿下第三名，以及高塔半導體（Tower）超越力積電（PSMC）及世界先進（V...

technews.tw

No need 2025, tsmc has achieved 60%+ in 2023 Q1.
2019 Q1 foundry market share: TSMC 48.1%; Samsung 19.1%
2023 Q1 foundry market share: TSMC 60.1%; Samsung 12.4%

 

[Daniel Nenni]

That's kind of a long distance view of things, based on the fact Samsung sucks and is more vulnerable. I get it.

But, looking closer, NVIDIA has a somewhat rocky relationship with TSMC, and has used Samsung not too long ago, despite being a deficient process node. And have complained pretty publicly about the cost of wafers from TSMC very recently. And announced they are very interested in working with Intel foundries. If NVIDIA moves things to Intel, where are they moving them from? Not Samsung.

I would also not be surprised if Apple did some parts at Intel as well. At least in an exploratory way, to ensure supply and possibly make some less strategic parts there. Qualcomm is already involved with Intel. AMD is safe, but do they actually sell anything anymore .

So, I think both are going to get hit a bit, if Intel is successful. Intel already doesn't suck like Samsung, their I7 parts are the fastest clocked products in that segment, and have no problems competing with AMD parts on N5. Samsung isn't really competitive at anything, except smack talking.

But, Intel has to be successful with their nodes, and while they have been on high performance nodes, their experience with making a node suitable for lower power/higher density is limited. No doubt they can do it, but can they do it competitively against TSMC? Not an easy bar to reach.

A lot of things can happen, I think it's premature to think TSMC will be largely unaffected. They may be, and Intel may be largely unsuccessful, but there are other scenarios that could play out too.

I understand your logic but I look at it differently. QCOM and NVDA bounce between foundries, that is part of their company culture. Today they bounce between TSMC and Samsung. Right now they are at TSMC due to Samsung PDK failures but could bounce back at anytime. Next bounce, in my opinion, would be to Intel at a loss to Samsung.

Intel and Apple have a checkered past so I don't think they will bounce again. Apple had bounced between Samsung and TSMC but have been at TSMC exclusively since the iPhone 6+. AMD/Xilinx will stay at TSMC as well. AMD due to the Intel competition and Xilinx due to Intel competition. That is the problem with being an IDM foundry, you compete with your customers.

Intel and Samsung both have a foundry day coming up this month. It will be interesting to see the positioning. Intel had said they will be the #2 foundry in 10 years. Samsung has said they will be #1 on different occasions. Last I read it was by 2030. Of those two things happening I would bet it would be IFS as #2 by 2030. Any takers?