Korean media: Intel intends to team up with Samsung to compete against TSMC

[XYang2023]

According to a report from South Korean media outlet *Maeil Business Newspaper*, U.S. semiconductor company Intel (INTC-US) has sought to establish a "foundry alliance" with Samsung Electronics, laying the groundwork to catch up with TSMC (2330-TW)(TSM-US).

Citing sources from the semiconductor industry, the report stated that senior Intel officials recently requested to meet with Samsung's top executives, conveying CEO Pat Gelsinger's desire to personally meet Samsung Electronics Chairman Lee Jae-yong to discuss "comprehensive cooperation measures in the foundry sector."

Since the establishment of Intel Foundry Services (IFS) in 2021, Intel has signed contracts with Cisco and AWS but has not attracted major clients relative to its investment. Samsung Electronics established its foundry division in 2017 and began attracting customers, but it still lags significantly behind TSMC.

According to TrendForce, TSMC held a 62.3% share of the foundry market in the second quarter, while Samsung Electronics had 11.5%. Notably, TSMC controls 92% of the market share in advanced process technologies such as 3nm and 5nm.

The report suggests that if the "foundry alliance" between Intel and Samsung is realized, it could lead to comprehensive cooperation in areas like process technology exchange, production facility sharing, and research and development.

Samsung Electronics has a technology called "3nm GAA" that can enhance performance and power efficiency in advanced processes, while Intel has technologies such as Foveros, which can combine chips made from different processes into a single package, and PowerVia, which can improve power efficiency. These technologies could be collaboratively developed for AI, data centers, and mobile application processors (AP), where high performance and low power consumption are crucial.

Moreover, Samsung Electronics has manufacturing plants in the U.S., South Korea, and China, while Intel has plants in the U.S., Ireland, and Israel, meaning they could potentially share facilities or win orders together if necessary. Especially with increasing U.S. and EU control over high-tech semiconductor exports, regional production may be required.

Kim Hyung-joon, head of next-generation smart semiconductor business (and a former professor of materials engineering at Seoul National University), stated that if the foundry alliance between Intel and Samsung Electronics is established, the synergy potential is limitless. However, given TSMC's dominant position, it would be difficult to expect immediate major impacts.

Both Samsung Electronics and Intel stated that they "cannot confirm" whether high-level meetings will take place.

韓媒：英特爾有意聯手三星 合作對抗台積電 | 鉅亨網 - 美股雷達

據韓媒《每日經濟》報導，美國半導體公司英特爾 (INTC-US) 已尋求與三星電子建立「代工聯盟」，為追趕台積電 (2330-TW)(TSM-US) 奠定基礎。

news.cnyes.com

 

[samwilde]

You saw it speculated here in this forum first by our forum members. I think it actually makes sense. Samsung used to have a process design alliance with IBM and AMD where they shared process design costs. But now that IBM and AMD bailed out on having fabs and Globalfoundries seems stuck in a process time warp, with no EUV in sight, having Samsung make a similar deal with Intel would make more sense.

Not that I expect a deal. Even the sources for this article are kind of dubious.

 

[Daniel Nenni]

Intel 18A has Samsung 2nm beat. Does Samsung even have a HNA-EUV strategy? Is Samsung giving up their GAA process? FinFets conquered GlobalFoundries, GAA conquered Samsung?

I honestly do not see how this would work unless Samsung signs on to use Intel leading edge processes moving forward. Even then that would mean Samsung is declaring defeat and I do not see that happening.

What else could this mean? If you combine the internal design groups from Samsung and Intel that is a pretty big number. If you add some external customers from the NOT TSMC market that would be a viable foundry business, absolutely.

 

[Mooredaddy]

According to a report from South Korean media outlet *Maeil Business Newspaper*, U.S. semiconductor company Intel (INTC-US) has sought to establish a "foundry alliance" with Samsung Electronics, laying the groundwork to catch up with TSMC (2330-TW)(TSM-US).

Citing sources from the semiconductor industry, the report stated that senior Intel officials recently requested to meet with Samsung's top executives, conveying CEO Pat Gelsinger's desire to personally meet Samsung Electronics Chairman Lee Jae-yong to discuss "comprehensive cooperation measures in the foundry sector."

Since the establishment of Intel Foundry Services (IFS) in 2021, Intel has signed contracts with Cisco and AWS but has not attracted major clients relative to its investment. Samsung Electronics established its foundry division in 2017 and began attracting customers, but it still lags significantly behind TSMC.

According to TrendForce, TSMC held a 62.3% share of the foundry market in the second quarter, while Samsung Electronics had 11.5%. Notably, TSMC controls 92% of the market share in advanced process technologies such as 3nm and 5nm.

The report suggests that if the "foundry alliance" between Intel and Samsung is realized, it could lead to comprehensive cooperation in areas like process technology exchange, production facility sharing, and research and development.

Samsung Electronics has a technology called "3nm GAA" that can enhance performance and power efficiency in advanced processes, while Intel has technologies such as Foveros, which can combine chips made from different processes into a single package, and PowerVia, which can improve power efficiency. These technologies could be collaboratively developed for AI, data centers, and mobile application processors (AP), where high performance and low power consumption are crucial.

Moreover, Samsung Electronics has manufacturing plants in the U.S., South Korea, and China, while Intel has plants in the U.S., Ireland, and Israel, meaning they could potentially share facilities or win orders together if necessary. Especially with increasing U.S. and EU control over high-tech semiconductor exports, regional production may be required.

Kim Hyung-joon, head of next-generation smart semiconductor business (and a former professor of materials engineering at Seoul National University), stated that if the foundry alliance between Intel and Samsung Electronics is established, the synergy potential is limitless. However, given TSMC's dominant position, it would be difficult to expect immediate major impacts.

Both Samsung Electronics and Intel stated that they "cannot confirm" whether high-level meetings will take place.

韓媒：英特爾有意聯手三星 合作對抗台積電 | 鉅亨網 - 美股雷達

據韓媒《每日經濟》報導，美國半導體公司英特爾 (INTC-US) 已尋求與三星電子建立「代工聯盟」，為追趕台積電 (2330-TW)(TSM-US) 奠定基礎。

news.cnyes.com

Like dumb and dumber teaming up.

 

[XYang2023]

Like dumb and dumber teaming up.

I think teaming up is better than competing against each other. It is similar to the recent news that Intel and AMD teamed up to work on future x86 specifications.

 

[fansink]

Like dumb and dumber teaming up.

Worse, it'd be like Intel and Samsung teamed up.

 

[Xebec]

Could this be a trade of IP? some cash exchanged one way or the other?

In many ways Samsung and Intel Foundry compete, but in many other ways they're also complementary (Intel doesn't seem to plan to fab memory).

 

[MKWVentures]

Didnt some other company that starts with an "I" have a foundry partnership with Samsung? Deja Vu all over again?
It actually would be a great partnership strategically since they are head to head competition now.

There is a reason that Pat calls is "IBM2.0" Patton, Obuckley, Samsung partnership .... is IBM a good model to follow? Was IBM foundry successful?

 

[XYang2023]

Would it make sense from a capital spending point of view if the collaboration results in better factory utilization, fewer fab constructions, and fewer machine procurements?

 

[delong.height]

I honestly do not see how this would work unless Samsung signs on to use Intel leading edge processes moving forward. Even then that would mean Samsung is declaring defeat and I do not see that happening.

As the original creator of this concept, I'm not sure where Intel management saw my post, either here https://semiwiki.com/forum/index.ph...ntel-and-to-intel-management.20914/post-73868 or

https://www.reddit.com/r/wallstreetbets/comments/1f8vmnv

But either way, Samsung has probably no way but to accept it. They just cannot acquire any new customer, even Nvidia deal. Nvidia is trying on all front because of their supply is still limited while the demand is huge, and it doesn't matter who would fab their product, their customer don't care and believe in their brand.

This tweet is from Dylan today. https://x.com/dylan522p/status/1848557416264208461 And one must question is Samsung able to continue to deliver the next process node when the current one in production looks immature for so long.

And to answer Daniel's question, yes, I do see that coming. Samsung has done that in the past with Exynos, the switching to use Qualcomm Snapdragon in all if not majority of the lineup. It wasn't like that when they are still capable. They have ditched that in the past. And I have no wonder that it won't happen now.

Not to mention, if Intel wants to, they can unleash perhaps 80% of 18A capacity that was previously planned for Ohio and Arizona, and port all to Samsung because it cost less to manufacture in Asia, not to mention it's Intel's process. It's basically compatible with the one that's going to be manufactured in Arizona. Both Intel and Samsung knows what they can get vs they do it solo. If they partner, the utilization in both foundry increases, as customers who order in Samsung can directly port the design to Intel in case of geopolitical event, and Samsung can absorb customers who are a competitor of Intel, giving Intel the ability to benefit from its competitors, like Nvidia or AMD without having to go public for its IFS. There is a lot less need to spend on Capex dollars, which is going to be worse for almost all semiconductor equipment companies, but neither Intel nor Samsung should care.

Would it make sense from a capital spending point of view if the collaboration results in better factory utilization, fewer fab constructions, and fewer machine procurements?

That is the case, and why both party should pursue this strategy because the industry has voted with their feet that they are standing with TSMC. Reduce the resource while making sure both party get what they want. Samsung want to get a good chunk of fab business while Intel needs its IDM 2.0 to excel. And that's what I think make Intel and Samsung different. Samsung are willing to do anything they possibly can to increase its revenue, its profits, from multi-businesses, multi-industries. While Intel's goal is to produce any silicon that's cheaper than their competitor yet better. Franchisee 18A, Intel 3, and Intel 14A enable Intel to produce at a perhaps a much lower cost than their counterpart, AMD, Nvidia, Qualcomm, and Apple. That's is how IDM 2.0 should work in Pat Gelsinger's plan.

 

[semiman]

The issue is that they're focusing on different types of products. Both Intel and Samsung are IDM foundries. Intel is focusing on HPC and Samsung is on mobile products. Intel has more internal demands (200m HPC chips per year?), but mobile has more foundry market potential (1.2B per year). So there's no silver bullet for both of them. It's interesting to see what kind of synergy they're thinking.

 

[Daniel Nenni]

Would it make sense from a capital spending point of view if the collaboration results in better factory utilization, fewer fab constructions, and fewer machine procurements?

I think it is the only way to fill the fabs Intel and Samsung have in process. Samsung really is in bad shape:

Hundreds of Samsung engineers apply for SK Hynix etching job

Since Samsung’s lower-than-expected third quarter earnings and its apology for it, there is increasing concern that the tech giant’s chip business is facing a crisis.Amid this climate, there are more and more cases being reported where Samsung semiconductor engineers are moving to competitors or gov

www.thelec.net

I have a podcast coming up with a CHIPs Act friend so hopefully I can find out what happens when Samsung and Intel do not build fabs in the US as planned in the CHIPs Act applications.

 

[XYang2023]

I think it is the only way to fill the fabs Intel and Samsung have in process. Samsung really is in bad shape:

Hundreds of Samsung engineers apply for SK Hynix etching job

Since Samsung’s lower-than-expected third quarter earnings and its apology for it, there is increasing concern that the tech giant’s chip business is facing a crisis.Amid this climate, there are more and more cases being reported where Samsung semiconductor engineers are moving to competitors or gov

www.thelec.net

I have a podcast coming up with a CHIPs Act friend so hopefully I can find out what happens when Samsung and Intel do not build fabs in the US as planned in the CHIPs Act applications.

I meant that they should finish what they are currently building to reap the rewards of the CHIPS Act. However, they can delay starting any new projects, at least for now.

I suppose they are not in direct competition, as one is focused on logic and the other on memory.

If they can progress in this direction, it would significantly ease the strain on both their cash flows and balance sheets.

They could then pool their resources to compete with TSMC.

 

[delong.height]

The issue is that they're focusing on different types of products. Both Intel and Samsung are IDM foundries. Intel is focusing on HPC and Samsung is on mobile products. Intel has more internal demands (200m HPC chips per year?), but mobile has more foundry market potential (1.2B per year). So there's no silver bullet for both of them. It's interesting to see what kind of synergy they're thinking.

That's isn't true anymore. In order to be successful, they need to have on all front, whether it be automobile, hpc, personal compute, edge, mobile, you name it.

 

[delong.height]

I think it is the only way to fill the fabs Intel and Samsung have in process. Samsung really is in bad shape:

Hundreds of Samsung engineers apply for SK Hynix etching job

Since Samsung’s lower-than-expected third quarter earnings and its apology for it, there is increasing concern that the tech giant’s chip business is facing a crisis.Amid this climate, there are more and more cases being reported where Samsung semiconductor engineers are moving to competitors or gov

www.thelec.net

I have a podcast coming up with a CHIPs Act friend so hopefully I can find out what happens when Samsung and Intel do not build fabs in the US as planned in the CHIPs Act applications.

They should continue to build in the US, but for next administration, no money, no build out. And Biden administration have played these companies hard. Chip act being begging since 2021, yet still no pennies that's ever being distributed.

 

[delong.height]

I meant that they should finish what they are currently building to reap the rewards of the CHIPS Act. However, they can delay starting any new projects, at least for now.

I suppose they are not in direct competition, as one is focused on logic and the other on memory.

If they can progress in this direction, it would significantly ease the strain on both their cash flows and balance sheets.

They could then pool their resources to compete with TSMC.

Yup

 

[fansink]

Didnt some other company that starts with an "I" have a foundry partnership with Samsung? Deja Vu all over again?
It actually would be a great partnership strategically since they are head to head competition now.

There is a reason that Pat calls is "IBM2.0" Patton, Obuckley, Samsung partnership .... is IBM a good model to follow? Was IBM foundry successful?

Pat, trying to put a good face on the alliance.