Intel Reports Fourth-Quarter and Full-Year 2024 Financial Results

[XYang2023]

So what is the market valuation for a capital intensive company with the P&L of IFS. It would seem like the valuation might be similar to what GF "paid" IBM for its foundry.

Say at worst, when it breaks even, it will not drag down Intel's overall valuation.

If IFS is profitable, its valuation should be 1-2x of its asset value. That means 100-200B.

 

[VCT]

This is the legacy of PG. I think the current approach is to untangle the mess he created to recover as much value as possible.

If Global Foundry wants to buy it, it means it has value. Then it is logical for Intel extract as much value from it as possible.

Again the overall goal should be to recover the company's financial status and return to a growth trajectory.

Are you willing to put your own money to invest in IFS at 50% discount of IFS's net value?

 

[XYang2023]

"A product company and the second largest foundry" was the same description used for Intel when they first attempted to offer foundry services, and they failed. It's worth noting that Intel had a superior process and a higher market share at that time.

Now, when considering giving designs to Intel, it's not just about process and price, but also the opportunity cost. Potential customers for 18A and beyond cannot afford any misses, whether in performance, timeline, or volume. Intel has a lot of work to do to earn trust, which is TSMC's strongest advantage.

Also Intel needs the fab capacity to maintain lead in the market. By outsourcing to TSMC, it would lose that lead. The question is about balancing.

 

[XYang2023]

Are you willing to put your own money to invest in IFS at 50% discount of IFS's net value?

I don't need to say anything more about my view. By the way, I did dislike what PG did before the earnings by announcing that he took a position in Nvidia. So many people took a position in Intel because of him, and I don't quite understand why he is doing that now. Even if it is due to being fired, I don't think that is appropriate.

 

[VCT]

I don't need to say anything more about my view. By the way, I did dislike what PG did before the earnings by announcing that he took a position in Nvidia. So many people took a position in Intel because of him, and I don't quite understand why he is doing that now. Even if it is due to being fired, I don't think that is appropriate.

I mean it's really difficult to find investors for IFS. Its won't be a one time investment. It will be like Japanes Rapidus. After your initial investment, highly possbile they will ask you to put in more money 2-3 years later.

 

[XYang2023]

I mean it's really difficult to find investors for IFS. Its won't be a one time investment. It will be like Japanes Rapidus. After your initial investment, highly possbile they will ask you to put in more money 2-3 years later.

I think they now have the option to either double down on it or simply roll it out. If they try to serve only their own needs, they might be able to manage the fabs. However, if Intel also tries to accommodate unknown clients, it will require significant capital.

At the moment, there is no need to further overcommit, as they already have empty shells available. If there are significant clients lining up for 18A capacity due to tariffs, they can decide to expand to meet that demand. Otherwise, Intel should focus on its own business.

 

[siliconbruh999]

I think they now have the option to either double down on it or simply roll it out. If they try to serve only their own needs, they might be able to manage the fabs. However, if Intel also tries to accommodate unknown clients, it will require significant capital.

At the moment, there is no need to further overcommit, as they already have empty shells available. If there are significant clients lining up for 18A capacity due to tariffs, they can decide to expand to meet that demand. Otherwise, Intel should focus on its own business.

I agree with this they should build according to demand not expectations they should keep 1 shell just in case but not multiple their products needs to be competitive as well the only products that were competitive were Raptor Lake/Lunar Lake /Alder Lake. In recent years and B580 as well

 

[XYang2023]

I agree with this they should build according to demand not expectations they should keep 1 shell just in case but not multiple their products needs to be competitive as well the only products that were competitive were Raptor Lake/Lunar Lake /Alder Lake. In recent years and B580 as well

They all came from the client division.

 

[siliconbruh999]

They all came from the client division.

Yes DCAI is a lost cause it's need to reorg and restructuring cause it's in a bad state the only good thing were GNR/SRF in last 5 years

 

[Maximus]

Also Intel needs the fab capacity to maintain lead in the market. By outsourcing to TSMC, it would lose that lead. The question is about balancing.

This is tricky. We already see how Intel retreat from Pat's Germany and Isreal fab plan. I find it misleading to highlight the benefits of both outsourced and insourced manufacturing while concealing the financial and capacity conflicts. On the other hand, TSMC excels in capacity planning by constructing fabs based on actual orders.

I think they now have the option to either double down on it or simply roll it out. If they try to serve only their own needs, they might be able to manage the fabs. However, if Intel also tries to accommodate unknown clients, it will require significant capital.

At the moment, there is no need to further overcommit, as they already have empty shells available. If there are significant clients lining up for 18A capacity due to tariffs, they can decide to expand to meet that demand. Otherwise, Intel should focus on its own business.

I'm concerned that Intel cannot sustain a healthy financial state by solely focusing on its own business, especially if only their CPU products remain competitive, to offset the high costs of developing advanced nodes. AMD has built a strong reputation and competitiveness over the past few years. Even if Intel has some excellent products on the horizon, the opportunity to charge premium prices like in the past may no longer exist. This is why Pat needed to present a grand IDM 2.0 vision to make the story convincing.

 

[XYang2023]

This is tricky. We already see how Intel retreat from Pat's Germany and Isreal fab plan. I find it misleading to highlight the benefits of both outsourced and insourced manufacturing while concealing the financial and capacity conflicts. On the other hand, TSMC excels in capacity planning by constructing fabs based on actual orders.


I'm concerned that Intel cannot sustain a healthy financial state by solely focusing on its own business, especially if only their CPU products remain competitive, to offset the high costs of developing advanced nodes. AMD has built a strong reputation and competitiveness over the past few years. Even if Intel has some excellent products on the horizon, the opportunity to charge premium prices like in the past may no longer exist. This is why Pat needed to present a grand IDM 2.0 vision to make the story convincing.

I don’t see that as a retreat. It was PG who went around the world looking for subsidies to build fabs without fully understanding actual demand dynamics and the challenges of integrating third-party IP support into its processes.

He has taken a similar approach in other scenarios. For example, during COVID, he over-invested in PC business which led to later write-off.

At the moment, the bar is quite low for Intel. To turn the company around, it doesn’t need to charge customers premium prices. Instead, Intel needs to generate a reasonable margin and build from there. Ideally, it should drive more products to IFS to improve overall margins for the group.

 

[kevin01]

This is tricky. We already see how Intel retreat from Pat's Germany and Isreal fab plan. I find it misleading to highlight the benefits of both outsourced and insourced manufacturing while concealing the financial and capacity conflicts. On the other hand, TSMC excels in capacity planning by constructing fabs based on actual orders.

I think Trump's tariff plan will help Intel in the next few months. It is kind of tricky, though, because currently most Intel products are manufactured at foreign fabs (Israel, Ireland) too. We will see.

I'm concerned that Intel cannot sustain a healthy financial state by solely focusing on its own business, especially if only their CPU products remain competitive, to offset the high costs of developing advanced nodes. AMD has built a strong reputation and competitiveness over the past few years. Even if Intel has some excellent products on the horizon, the opportunity to charge premium prices like in the past may no longer exist. This is why Pat needed to present a grand IDM 2.0 vision to make the story convincing.

Intel seems to lack visionary leadership capable of anticipating future industry needs and proactively developing solutions. In contrast, NVDA's CEO, Jensen Huang, has demonstrated remarkable foresight. His acquisition of Mellanox, anticipating the critical role of high-bandwidth ultra-fast interconnects in accelerated computing, and his attempted acquisition of ARM, while unsuccessful, were both strategically brilliant moves.

What is Huang betting on? He seems to think that Moore's Law is ending, and he is focused on advanced packaging and system-level integration.

To compete effectively in the DCAI space, Intel needs to adopt a similar approach, anticipating industry needs 2-3 years out rather than simply reacting to current trends or making something slightly faster.

What are the future needs? Perhaps bigger memory? Computation needs arising from new AI reasoning models?

 

[Maximus]

Intel seems to lack visionary leadership capable of anticipating future industry needs and proactively developing solutions. In contrast, NVDA's CEO, Jensen Huang, has demonstrated remarkable foresight. His acquisition of Mellanox, anticipating the critical role of high-bandwidth ultra-fast interconnects in accelerated computing, and his attempted acquisition of ARM, while unsuccessful, were both strategically brilliant moves.

What is Huang betting on? He seems to think that Moore's Law is ending, and he is focused on advanced packaging and system-level integration.

To compete effectively in the DCAI space, Intel needs to adopt a similar approach, anticipating industry needs 2-3 years out rather than simply reacting to current trends or making something slightly faster.

What are the future needs? Perhaps bigger memory? Computation needs arising from new AI reasoning models?

The issue is whether Intel still has the confidence to make bold moves in its product lines. Does Intel have a strong internal culture that encourages employees to think innovatively? Having foresight and reacting promptly requires action before the need becomes obvious.

Intel discontinued Ponte Vecchio after delivering it to the National Lab, canceled Riato Bridge, and now uses Falcon Shores for internal research (a better term for kill). These failed attempts could hinder the courage needed to take bold actions.

 

[DanX]

Maybe Intel should sell its fabs to TSMC to become a dedicated CPU/AI design company and use domino logic technology to gain some advantage.

 

[DaRAGingLunatic]

Maybe Intel should sell its fabs to TSMC to become a dedicated CPU/AI design company and use domino logic technology to gain some advantage.

TSMC doesn’t want intel fabs. Why would they?

I see a bear on Intel.

Wont build capacity for clients it doesn’t have who wont go with intel due to lack of capacity. Capacity is expensive to build and due to bad financials doesn’t want to really expend more. What client will want to go with intel if intel is in such a bad state that you won’t know if they are going to be there doing fab in the future. Why try risk expensive masks with intel on a new process if you may have to switch to TSMC later down the line?

What about TSMC?

They are going to town on fabs. No risk other than China invasion.

 

[XYang2023]

TSMC doesn’t want intel fabs. Why would they?

I see a bear on Intel.

Wont build capacity for clients it doesn’t have who wont go with intel due to lack of capacity. Capacity is expensive to build and due to bad financials doesn’t want to really expend more. What client will want to go with intel if intel is in such a bad state that you won’t know if they are going to be there doing fab in the future. Why try risk expensive masks with intel on a new process if you may have to switch to TSMC later down the line?

What about TSMC?

They are going to town on fabs. No risk other than China invasion.

Shark attacks are rare, but they can happen. In fact, when they do occur, they are often fatal.

When assessing risk, people should consider both probability and impact simultaneously. In the context of geo-conflict, the consequences can be severe. For instance, TSMC might not be the TSMC we know today.

 

[TICin852]

Wont build capacity for clients it doesn’t have who wont go with intel due to lack of capacity. Capacity is expensive to build and due to bad financials doesn’t want to really expend more. What client will want to go with intel if intel is in such a bad state that you won’t know if they are going to be there doing fab in the future. Why try risk expensive masks with intel on a new process if you may have to switch to TSMC later down the line?

QC, NVDA, and others are all for the existence of "Not TSMC" business. Intel will win that business first - it won't be massive, but the Not TSMC group will throw Intel a few bones, especially at 18A. Intel already has confirmed that they should have an external customer taping out on 18A in 1H25, and I don't see any reason why this would change. I do believe that if Intel receives pre-pays and makes wafer agreements, they will be able to install equipment and meet client capacity, especially for 14A - which is when we will likely see larger volumes (if 18A is well-received and customers decide to take a chance with Intel).

 

[BruceA]

Maybe Intel should sell its fabs to TSMC to become a dedicated CPU/AI design company and use domino logic technology to gain some advantage.

That was what Bob wanted to do and he got fired and Pat went the opposite and got fired too! Is there a middle ground or is it only those two obvious directions that can work?

Actually the surest is cut the design choose the best fab and let Foundry stand alone which they simply can’t. The bridge - get TSMC to assume the fabs to assure Chip security for a US firm and after 18A migrate the fabs to TSMC technology while offering 14A if there are takers for it. I could argue TSMC and Ohio can be US safe enclave producers of advanced nodes that are geographically safe.

US government can help support more directly and this is just another expensive weapons program cost.

 

[siliconbruh999]

That was what Bob wanted to do and he got fired and Pat went the opposite and got fired too! Is there a middle ground or is it only those two obvious directions that can work?

Actually the surest is cut the design choose the best fab and let Foundry stand alone which they simply can’t. The bridge - get TSMC to assume the fabs to assure Chip security for a US firm and after 18A migrate the fabs to TSMC technology while offering 14A if there are takers for it. I could argue TSMC and Ohio can be US safe enclave producers of advanced nodes that are geographically safe.

US government can help support more directly and this is just another expensive weapons program cost.

The only one who didn't have plan was the board and why do we need to rely on TSMC so much if Intel's technology is getting close?
Japan is taking even more risk with Rapidus Intel's Plan seems less risky compared to that.
If I were to do something with regards to fabs I would just force x86 designs everywhere as cheaply as arm TVs/Smartphone and stuff just offer them better price vs Competitors on the remaining fab capacity.

 

[DaRAGingLunatic]

QC, NVDA, and others are all for the existence of "Not TSMC" business. Intel will win that business first - it won't be massive, but the Not TSMC group will throw Intel a few bones, especially at 18A. Intel already has confirmed that they should have an external customer taping out on 18A in 1H25, and I don't see any reason why this would change. I do believe that if Intel receives pre-pays and makes wafer agreements, they will be able to install equipment and meet client capacity, especially for 14A - which is when we will likely see larger volumes (if 18A is well-received and customers decide to take a chance with Intel).

Well, I guess the question is how many wafer starts per month can Intel do on 18A and 14A? What portion of those wafers are assigned to Intel, and how many wafers are left for everyone else, and are those wafers enough for a potential client to switch over and use?