2009 GlobalFoundries vs 2026/2027 Spin-off Intel Foundry Services

[VCT]

2009 GlobalFoundries vs 2026/2027 Spin-off Intel Foundry Services

Which is more challenging to survive?

I'd say IFS is more challenging because it is harder to secure customers for advanced nodes.
In contrast, GlobalFoundries' mature nodes have an easier time attracting customers.

 

[bashee_wang]

I am wondering what's the real status of Intel 18A, which said to be in mass production by this year. How is the yield? How is the quality? Pat said there are ~12 active customers are contact with Intel for 18A, which big customer included? Rumor said there is Nvidia, Microsoft, etc. If 18A is attractive enough, it might be the straw to grasp at during this stage.

 

[Y.H]

I think Microsoft already said they will use 18A. Not sure what’s the chip used or how big a contract.

 

[siliconbruh999]

I am wondering what's the real status of Intel 18A, which said to be in mass production by this year. How is the yield? How is the quality? Pat said there are ~12 active customers are contact with Intel for 18A, which big customer included? Rumor said there is Nvidia, Microsoft, etc. If 18A is attractive enough, it might be the straw to grasp at during this stage.

They gave the D0 at <0.4 this quarter

 

[Xebec]

2009 GlobalFoundries vs 2026/2027 Spin-off Intel Foundry Services

Which is more challenging to survive?

I'd say IFS is more challenging because it is harder to secure customers for advanced nodes.
In contrast, GlobalFoundries' mature nodes have an easier time attracting customers.

IFS because of volume also. A lot more fabs and complexity.. Also smaller potential customer pool (China is off the board today). Also in 2009 TSMC wasn’t nearly as dominant; the slowing/end game of scaling has a lot of consequences.

 

[count]

I think people fail to recognize how bad financial conditions were in 2009 when looking at GFS vs potential IFS spinoff compared to the overall macro today.

1. In 2009 we were in the worst of the financial crisis, and financing large deals was very difficult
2. AMD was in worse shape in 2009 than Intel is today. AMD stock price was down to $2. It was pretty questionable how much a wafer commitment from AMD would be worth in the long term. AMD did not have leadership in any market.
3. The world was in a recession and tech spending was way down, it wasn't clear yet how long it would take for the market to recover.
4. Semi's were not considered a hot thing, investors were focused on energy and natural resources
5. On the other hand petrostates were flush with cash and wanted to diversify

Compare that today where
1. Financial conditions overall are still pretty good (at least for now)
2. Intel may look like they are in bad shape on the surface, but they still have leading market share in several key markets, and a cash cow business in x86. A wafer commitment from Intel would be very meaningful.
3. Intel still has very good technology, although execution has been bad
4. There is a lot of excitement about semiconductors right now. Many of the hottest stocks out there right now are semi companies
5. Semi manufacturing has become a strategic priority for governments around the world

So it's a matter of perspective. I think a spin off makes a lot of sense and I'm quite sure there would be buyers.

Also Intel doesn't need a high price from these assets. One could argue that Intel could give the fabs away for free and it would still be a net positive.

 

[VCT]

I think people fail to recognize how bad financial conditions were in 2009 when looking at GFS vs potential IFS spinoff compared to the overall macro today.

1. In 2009 we were in the worst of the financial crisis, and financing large deals was very difficult
2. AMD was in worse shape in 2009 than Intel is today. AMD stock price was down to $2. It was pretty questionable how much a wafer commitment from AMD would be worth in the long term. AMD did not have leadership in any market.
3. The world was in a recession and tech spending was way down, it wasn't clear yet how long it would take for the market to recover.
4. Semi's were not considered a hot thing, investors were focused on energy and natural resources
5. On the other hand petrostates were flush with cash and wanted to diversify

Compare that today where
1. Financial conditions overall are still pretty good (at least for now)
2. Intel may look like they are in bad shape on the surface, but they still have leading market share in several key markets, and a cash cow business in x86. A wafer commitment from Intel would be very meaningful.
3. Intel still has very good technology, although execution has been bad
4. There is a lot of excitement about semiconductors right now. Many of the hottest stocks out there right now are semi companies
5. Semi manufacturing has become a strategic priority for governments around the world

So it's a matter of perspective. I think a spin off makes a lot of sense and I'm quite sure there would be buyers.

Also Intel doesn't need a high price from these assets. One could argue that Intel could give the fabs away for free and it would still be a net positive.

AMD IC design in 2009 was a lot worse than Intel IC design 2024. AMD IC design in 2009 was bad but didn't need a lot of money to survive.

However, although the money required to maintain AMD's manufacturing operations in 2009 was substantial, it was still considerably less than what Intel's manufacturing will require in 2025/2026.

 

[hist78]

AMD IC design in 2009 was a lot worse than Intel IC design 2024. AMD IC design in 2009 was bad but didn't need a lot of money to survive.

However, although the money required to maintain AMD's manufacturing operations in 2009 was substantial, it was still considerably less than what Intel's manufacturing will require in 2025/2026.

In 2009 AMD design/product division had only one major competitor: Intel. Or we can add Nvidia as a major competitor in the graphic cards market.

In 2024 Intel has AMD, Nvidia, Qualcomm, Apple, MediaTek, Broadcom, Google, Amazon, Microsoft, Facebook, Ampere Computing, Cerebras, and Tesla as the major competitors. I probably missed several more.

 

[count]

In 2009 AMD design/product division had only one major competitor: Intel. Or we can add Nvidia as a major competitor in the graphic cards market.

In 2024 Intel has AMD, Nvidia, Qualcomm, Apple, MediaTek, Broadcom, Google, Amazon, Microsoft, Facebook, Ampere Computing, Cerebras, and Tesla to be the major competitors. I probably missed several more.

Not relevant to IFS spinout IMO, as long as Intel can offer a wafer volume commitment for the next 5 years, which I think they can.

AMD IC design in 2009 was a lot worse than Intel IC design 2024. AMD IC design in 2009 was bad but didn't need a lot of money to survive.

However, although the money required to maintain AMD's manufacturing operations in 2009 was substantial, it was still considerably less than what Intel's manufacturing will require in 2025/2026.

Scaled back capex would probably be an outcome of the spinout.

 

[hist78]

Not relevant to IFS spinout IMO, as long as Intel can offer a wafer volume commitment for the next 5 years, which I think they can.


Scaled back capex would probably be an outcome of the spinout.

"Not relevant to IFS spinout IMO, as long as Intel can offer a wafer volume commitment for the next 5 years, which I think they can."

All those major fabless semiconductor companies I listed are also Intel product division's competitors. Intel Foundry will be in a very difficult position to gain significant orders from them if Intel keeps Intel Foundry and Intel Design/Product divisions inhouse under the IDM model.

 

[nghanayem]

I think people fail to recognize how bad financial conditions were in 2009 when looking at GFS vs potential IFS spinoff compared to the overall macro today.

1. In 2009 we were in the worst of the financial crisis, and financing large deals was very difficult
2. AMD was in worse shape in 2009 than Intel is today. AMD stock price was down to $2. It was pretty questionable how much a wafer commitment from AMD would be worth in the long term. AMD did not have leadership in any market.
3. The world was in a recession and tech spending was way down, it wasn't clear yet how long it would take for the market to recover.
4. Semi's were not considered a hot thing, investors were focused on energy and natural resources
5. On the other hand petrostates were flush with cash and wanted to diversify

Compare that today where
1. Financial conditions overall are still pretty good (at least for now)
2. Intel may look like they are in bad shape on the surface, but they still have leading market share in several key markets, and a cash cow business in x86. A wafer commitment from Intel would be very meaningful.
3. Intel still has very good technology, although execution has been bad
4. There is a lot of excitement about semiconductors right now. Many of the hottest stocks out there right now are semi companies
5. Semi manufacturing has become a strategic priority for governments around the world

So it's a matter of perspective. I think a spin off makes a lot of sense and I'm quite sure there would be buyers.

Also Intel doesn't need a high price from these assets. One could argue that Intel could give the fabs away for free and it would still be a net positive.

Interesting argument. I think there are some strong counterpoints though:

Today 2) Intel's volumes are much higher than AMD back then, but nowadays intel products are mostly using older nodes with only minimal content on the most advanced process. As an example take MTL vs RPL. Similar die sizes but the intel 4 die is only like 20-25% of the total die area. So assuming intel can win back share and get back to an equivalent product volume to the pandemic era, then intel's most advanced wafer demands would conservatively be a quarter of what they were.

Today 3) Those two statements are incompatible. They are either good at executing to a technology roadmap or bad at it.

Today 4/5) When AD bought out AMD's fab it was one fab (and much of that capacity was 8in). Even today after two expansions Dresden has a similar 300mm equivalent capacity to intel New Mexico iso process. And NM is intel's smallest fab. Add in two "giga fabs" in AZ and OR, the similar sized IS fab, and the larger IR fab and intel is a FAR FAR FAR FAR larger beast to swallow than AMD and later chartered. GF also hemorrhaged money for a decade, cancel a China fab that was built practically for free, and had to give up all process TD to even turn profitable. Intel cannot stop process TD because the old nodes are unusable in a foundry context and reworking them would be as expensive as just starting from scratch.