Intel splits Foundry from Design

[XYang2023]

This looks mostly for show to me. They still report to PG.

I think the answer is both yes and no.
They need to send a message to potential active investors.

At the same time, they implemented what we have been discussing here:

Intel’s board lacks the right experience:
-> By having a separate board, they can bring in people with the right experience to guide IFS.
Scale back:
-> Postpone European projects.
Split IFS:
-> They made IFS an independent entity but still under Intel.

 

[hist78]

Independent board of directors is pretty critical. They can appoint key people from customers to be directors, like imagine if Intel could get Hock Tan or Jensen Huang as a director.

I’m wondering how this 'independent board' actually works. Even with the board in place, Intel still owns 100% of Intel Foundry, contributes over 90% of its revenue, provides the majority of its CapEx funding, and Intel Foundry’s senior executives still report to Pat Gelsinger.

What’s even stranger is that if the independent board disagrees with Pat Gelsinger’s decisions or strategy, they would have to essentially organize a BBQ party with Intel’s actual board of directors to convince them to fire Pat.

It would be difficult to recruit high-profile and capable individuals to join this independent board if it lacks the true decision-making power of a normal company’s board of directors.

 

[kevin01]

I’m wondering how this 'independent board' actually works. Even with the board in place, Intel still owns 100% of Intel Foundry, contributes over 90% of its revenue, provides the majority of its CapEx funding, and Intel Foundry’s senior executives still report to Pat Gelsinger.

What’s even stranger is that if the independent board disagrees with Pat Gelsinger’s decisions or strategy, they would have to essentially organize a BBQ party with Intel’s actual board of directors to convince them to fire Pat.

It would be difficult to recruit high-profile and capable individuals to join this independent board if it lacks the true decision-making power of a normal company’s board of directors.

CEO is not a dictator. Pat's decision, if not well made (e.g. opposed by a large number of IFS board members), can cost him dearly.

The reason IFS is not spun off today, is because Intel is in its worst position in the past several decades. The timing could not be worse for major actions such as selling off operating assets, or spining off business units.

Today's actions are major wins for Intel's long term investors. Now just go get 18A and 14A delivered!

 

[XYang2023]

I’m wondering how this 'independent board' actually works. Even with the board in place, Intel still owns 100% of Intel Foundry, contributes over 90% of its revenue, provides the majority of its CapEx funding, and Intel Foundry’s senior executives still report to Pat Gelsinger.

What’s even stranger is that if the independent board disagrees with Pat Gelsinger’s decisions or strategy, they would have to essentially organize a BBQ party with Intel’s actual board of directors to convince them to fire Pat.

It would be difficult to recruit high-profile and capable individuals to join this independent board if it lacks the true decision-making power of a normal company’s board of directors.

They should select relevant people for the board. The board should be able to examine PG’s decisions.

The mix should be better than Mobileye’s:

Mobileye Board of Directors

Patrick P. Gelsinger
Christine Pambianchi, HR head of Intel
Christopher Schell, Sales head of Intel
Frank D. Yeary, Director of Intel
Saf Yeboah-Amankwah, Strategy head of Intel

In some other forums, Intel employees do not rate Christine, Christopher, and Saf very well (not to mention PG).

I also don’t think those people have relevant experience with self-driving.

Mobileye should have people like Andrew Ng as board members.

 

[hist78]

They should select relevant people for the board. The board should be able to examine PG’s decisions.

The mix should be better than Mobileye’s:

Mobileye Board of Directors

Patrick P. Gelsinger
Christine Pambianchi, HR head of Intel
Christopher Schell, Sales head of Intel
Frank D. Yeary, Director of Intel
Saf Yeboah-Amankwah, Strategy head of Intel

In some other forums, Intel employees do not rate Christine, Christopher, and Saf very well (not to mention PG).

I also don’t think those people have relevant experience with self-driving.

Mobileye should have people like Andrew Ng as board members.

"Mobileye Board of Directors

Patrick P. Gelsinger
Christine Pambianchi, HR head of Intel
Christopher Schell, Sales head of Intel
Frank D. Yeary, Director of Intel
Saf Yeboah-Amankwah, Strategy head of Intel"

It's beyond funny

 

[XYang2023]

https://www.thelayoff.com/t/1tTYo0ri

Maybe people here can suggest a board mix for IFS.

 

[XYang2023]

"Mobileye Board of Directors

Patrick P. Gelsinger
Christine Pambianchi, HR head of Intel
Christopher Schell, Sales head of Intel
Frank D. Yeary, Director of Intel
Saf Yeboah-Amankwah, Strategy head of Intel"

It's beyond funny

They should not do things like that. It is not PG's personal game.

It is a game against Tesla and Chinese competitors.

For IFS, it is a game against TSMC. They should be very serious.

 

[hist78]

CEO is not a dictator. Pat's decision, if not well made (e.g. opposed by a large number of IFS board members), can cost him dearly.

The reason IFS is not spun off today, is because Intel is in its worst position in the past several decades. The timing could not be worse for major actions such as selling off operating assets, or spining off business units.

Today's actions are major wins for Intel's long term investors. Now just go get 18A and 14A delivered!

Cost him what? Pat Gelsinger doesn't report to the Intel Foundry's board. Intel product division generates the majority of profit and has the conflict of interest to Intel Foundry.

 

[XYang2023]

Cost him what? Pat Gelsinger doesn't report to the Intel Foundry's board. Intel product division generates the majority of profit and has the conflict of interest to Intel Foundry.

In any case, it is a necessary step to make IFS independent. Most importantly, PG needs to do better, i.e., deliver what he promised asap. If he cannot do it, step down and also replace the board asap.

 

[siliconbruh999]

I think Pat or the CEO will be on the board but the rest of them should be customers i would like someone along the lines of hock tan mediatek Ceo and Arm ceo will be a good pick

 

[siliconbruh999]

In any case, it is a necessary step to make IFS independent. Most importantly, PG needs to do better, i.e., deliver what he promised asap. If he cannot do it, step down and also replace the board asap.

At least let him get products on 18A out in 2H25 if not Time to say Bye Bye

 

[nghanayem]

https://www.thelayoff.com/t/1tTYo0ri

Maybe people here can suggest a board mix for IFS.

Ecosystem folks like EDA, arm, etc are easy choices for a board. But I would assume it would be hard to get customers on the board. While I think it would be super sick to get Jensen, Cristiano, and Rick Tsai (doubly so from his time at TSMC) on the board, I am skeptical they would want to do that given how that would likely jeopardize their relationships with Samsung or especially TSMC when they are all so leery of how the tiger cub of the west will disrupt the current status quo. Maybe you could see UMC and Tower staff on the board. Between the fact that they don’t play in the same process technology sandboxes they aren’t really supporting a competitor and they already collaborate on manufacturing and mature node TD. Seems like a why not next step. People in the past have also fantasized about Intel leading a new common platform alliance like group. Now that intel is about to make good on their promise earlier this year to set up IF as a separate legal entity, I suppose that offers the possibility of a UMC or Tower to invest and get some % of the returns based on the number of shares they own pf IF. This hypothetical organization would then transcend a mere technology development alliance and turn into a business alliance (maybe not to dissimilar to the Renault–Nissan–Mitsubishi Alliance). If you really want to get wild with it, I suppose you could even go to an airline alliance like situation where they funnel each other’s customers to each other. UMC “owns” planar device space and intel “owns” the 3D device vertical. Speaking of co-ownership… Maybe Micron gets shares of IF to expand their TAM to logic, or Hynix gets back into logic by investing money in IF.

hock tan

Way overvalued. The only thing he is good at is laying people off (a skillset intel seems comfortable with) and coasting by on existing patients without actually creating any more innovations after buying a company (a death nail in any foundry where even legacy technology needs constant innovation). There is a reason Qualcomm fought so desperately to not get swallowed whole by Avago (as that is the end of any company innovating further as Tan squeezes it out in the name of maximizing ROI from the bare minimum of investment).

 

[Xebec]

Tower and UMC are the most Rational reasons to be optimistic.... very good partnerships from a strategic point of view. but those are long term strategies.

I'll assume you are correct on wins. So the questions will be what is the volume and when. Intel does not have planned volume today to load even one 18A/20A fab in 2026. we have plugged the numbers into the spreadsheet and a lot of work is needed to make 2027 even close to breakeven ... (as predicted).

Who are Samsungs top 5 customers? If Intel stole 50% from all of them instantaneously (other than Samsung). Would that add up to one fab? (it doesnt on my spreadsheet).

In simpleton terms, what kind of volume would it take to fill an 18A fab?

Is it more volume than an entire generation of Intel’s own desktop and mobile chips now (without igpu/soc)? or are you talking external customers only?

 

[ChrisGar]

Maybe people here can suggest a board mix for IFS.

I'd guess this would be more like an "Advisory Board." These boards can easily be ignored.

The main benefit could be to include a couple of major customers as part of the board -- with the idea it could help IFS win some business.

 

[MKWVentures]

In simpleton terms, what kind of volume would it take to fill an 18A fab?

Is it more volume than an entire generation of Intel’s own desktop and mobile chips now (without igpu/soc)? or are you talking external customers only?

lets say a 18A fab is 50k wafers per month. you can take chiplet sizes and estimate volume. The part people miss is that volumes do not ramp quickly unless you are doing a massive fab conversion. example: In Q4 2023 meteor lake was being outsold by raptor lake 8:1. in fact Alder lake was outselling meteor lake. and the benefit of chiplets is that you do not scale all parts of the chip... so some stay on 3 or 5 or 6.

Current Intel model is to get half a fab of foundry and half a fab of Intel internal on 18A in 2026. Both are behind and the impact on depreciation and cash flow requires a full fab.

Simple example: I could move the CPU chiplet for ALL Intel Client products (assuming ALL are chiplet based) into a little over 1/2 a fab (unless the yields are terrible). I could make ALL datacenter processors in One 18A fab. If these optimistic items were to happen, even the fastest conversion would mean this does happen until 2028 timeframe.

In order to need two fabs, you need ALL of the Samsungs (non-samsung) leading edge foundry business PLUS ALL Intel CPUs to be made on 18A PLUS Intel to start moving GPUs, SOCs to 18A. Or Intel needs to steal double digit leading edge share from TSMC.... which only happens IF TSMC massively messes up.

The Game changer on this is the WHALE. Amazon Fabric is not a whale, Microsoft is not a whale. government work is not a whale until 2030.
Qualcomm, Nividia, AMD, Apple, going all in is a whale... @Daniel Nenni knows far more than me on this part.

Feel free to do the math and tell me where I am wrong. we can review the spreadsheets

 

[kevin01]

lets say a 18A fab is 50k wafers per month. you can take chiplet sizes and estimate volume. The part people miss is that volumes do not ramp quickly unless you are doing a massive fab conversion. example: In Q4 2023 meteor lake was being outsold by raptor lake 8:1. in fact Alder lake was outselling meteor lake. and the benefit of chiplets is that you do not scale all parts of the chip... so some stay on 3 or 5 or 6.

Current Intel model is to get half a fab of foundry and half a fab of Intel internal on 18A in 2026. Both are behind and the impact on depreciation and cash flow requires a full fab.

Simple example: I could move the CPU chiplet for ALL Intel Client products (assuming ALL are chiplet based) into a little over 1/2 a fab (unless the yields are terrible). I could make ALL datacenter processors in One 18A fab. If these optimistic items were to happen, even the fastest conversion would mean this does happen until 2028 timeframe.

In order to need two fabs, you need ALL of the Samsungs (non-samsung) leading edge foundry business PLUS ALL Intel CPUs to be made on 18A PLUS Intel to start moving GPUs, SOCs to 18A. Or Intel needs to steal double digit leading edge share from TSMC.... which only happens IF TSMC massively messes up.

The Game changer on this is the WHALE. Amazon Fabric is not a whale, Microsoft is not a whale. government work is not a whale until 2030.
Qualcomm, Nividia, AMD, Apple, going all in is a whale... @Daniel Nenni knows far more than me on this part.

Feel free to do the math and tell me where I am wrong. we can review the spreadsheets

This is a useful analysis. Thank you, @MKWVentures

Let's use Intel's own estimate of what is going to happen (or rather, what they want to see happening). The timeframe of this chart is 2024-2030. The EUV nodes are basically Intel 3, Intel 18A, plus 14A.

Just by looking at the chart, IFS's plan is not super ambitious and seems achievable: the EUV capacity in 2030 is slightly larger than the current Pre-EUV capacity. And Intel stated that by 2030, they expect to have internal products use 70% of IFS capacity, while external uses 30%.

BTW, I am unclear about the point your are attempting to make. Are you suggesting that IFS will likely fail to find enough external customers, or are you suggesting the economics of IFS does not work, or something else?

 

[kevin01]

This is a useful analysis. Thank you, @MKWVentures

Let's use Intel's own estimate of what is going to happen (or rather, what they want to see happening). The timeframe of this chart is 2024-2030. The EUV nodes are basically Intel 3, Intel 18A, plus 14A.

Just by looking at the chart, IFS's plan is not super ambitious and seems achievable: the EUV capacity in 2030 is slightly larger than the current Pre-EUV capacity. And Intel stated that by 2030, they expect to have internal products use 70% of IFS capacity, while external uses 30%.

BTW, I am unclear about the point your are attempting to make. Are you suggesting that IFS will likely fail to find enough external customers, or are you suggesting the economics of IFS does not work, or something else?

Actually, let me put the detailed timeline in this chart. It represents 2024-2030. And by 2030, it desires to have a capacity of 150k EUV wafers per month.

 

[delong.height]

which win?

it's a multi-products, multi-billions of wafer and packaging deal for sure, that's what they announced. I do expect Amazon to do more in-house silicon that demonstrated their in-house design capabilities and save cost. Bad news for AMD and Nvidia for sure. But I'm pretty sure, as long as Amazon ask, all Intel's Ip will be accessible to them, including future GPU, CPU, FPGA as long as they use it for themselves.

 

[delong.height]

I have trouble seeing how this isn't just preparation to sell / spin-off Intel Foundry.

What are the tangible benefits of "greater independence for Intel Foundry"? What "independent sources of funding" would they be eligible for that Intel as a whole isn't already eligible for today? Wouldn't this diminish Intel's deal making power (e.g., like a unified PS6 design & manufacturing deal)?

And this may be me being pessimistic, but how does this in any way "enhance collaboration with Intel Products"?

so instead of owning both intel product, intel foundry, the companies who chose to invest in IFS only get a board seat in IFS. less scrutiny from government I guess

 

[delong.height]

I think the Amazon relationship is much more important and strategic than Sony's as it could convince other CSPs to start looking at IFS.

I do think that to be the case as well, imagine you got two choices, relying on merchant x86 CPU or develop your own with custom logic. I'm sure each CSP controls vast amount of data about how the silicon is being used, whether it be bandwidth, memory, the cores. Imagine getting that ability to tune based on your customers' demand and offer competitiveness that is above the market.

I'm pretty sure Intel would like all of the CSP to use any of its IP(whether it come from current-gen or future-gen) to build their own. It has two advantage.

1. Within x86 space, more spending allocation to be on IFS because it can offer a high level of customization.
2. Intel seemingly want to expand its x86 franchise, defeating the purpose of arm in data center.