You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Intel is facing tough times after its stock plunged 60% in 2024 and it ousted its CEO Pat Gelsinger. Lip-Bu Tan has been tapped as new CEO for the semiconductor company as it has struggled with both chip manufacturing and chip designing. Unlike Intel, companies specialized in either designing chips–like AMD or Nvidia–or manufacturing chips–like TSMC. WSJ explains how a mistake Intel made in the mid 2000s has snowballed into an overwhelming obstacle for the tech company that Lip-Bu Tan will have to face.
A good summary of the conventional criticisms of Intel.
You need semiwiki to explore the other side, the opportunities:
-Vertical integration is growing in many industries. Intel represents “old is new again”.
-Intel graphics business has always been misunderstood. They own integrated graphics in a way AMD and NVIDIA can’t challenge. Qualcomm, maybe, distantly.
-After Intel shuts down 3 or more older fabs, which is a historical process that comes around every 10 years or so, they will start to recover financially
I will never forget that Intel in the Pat era guided, at least for this Foundry play, that 2026 would be the beginnings of the external customer revenue ramp. Any time before then was always going to be this hellish grind to bootstrap a whole new business in the most capital intensive industry in existence.
From what I’m seeing in all the threads here on SemiWiki and the great analysis of Intel’s Foundry conferences, it seems like they are, more or less, on track with bootstrapping this business.
I’m sure you’re all aware of what the S-curve in business looks like. At the start of the curve it looks like you’re dead from a revenue perspective. I’m reminded of Tesla’s car sales in 2017/2018 before the 3 and Y took off… it’s just a question, like it was with Tesla, of if Intel can financially survive to get out of the bottom of this S-curve and hit that exponential ramp.
I will never forget that Intel in the Pat era guided, at least for this Foundry play, that 2026 would be the beginnings of the external customer revenue ramp. Any time before then was always going to be this hellish grind to bootstrap a whole new business in the most capital intensive industry in existence.
From what I’m seeing in all the threads here on SemiWiki and the great analysis of Intel’s Foundry conferences, it seems like they are, more or less, on track with bootstrapping this business.
I’m sure you’re all aware of what the S-curve in business looks like. At the start of the curve it looks like you’re dead from a revenue perspective. I’m reminded of Tesla’s car sales in 2017/2018 before the 3 and Y took off… it’s just a question, like it was with Tesla, of if Intel can financially survive to get out of the bottom of this S-curve and hit that exponential ramp.
they are not on track to what Pat said was going to happen. They are on track to what Dan and others said was realistic. essentially external is minimal until late 2027.
Now the finances don't line up. Product revenue also dropped a ton since 2021 (FYI other companies did not). 2021 plan was to add 20B more per year in revenue and add 10B more in Capex/spending. Actual revenue went down by 20B. Plug that into a spreadsheet.... its not pretty. To do the capex, Intel sold off partial rights to their two most advanced fabs and changed depreciation model so depreciation lasts longer but is better short term. 18A Ramp does not improve the spreadsheet in 2026.
IMO LBT will fix it but it needs to be a different Intel.
This may be an understatement but I still think Intel should stay together (design and manufacturing). Altera, Mobileye, etc... should be cut loose. Lip-Bu is a clever man and Intel will require a clever strategy moving forward. Copying Nvidia, AMD, etc... is not the way.
Intel also needs to shut the hell up. Intel should not have blasted BSPD and other innovations to the world before leveraging them inside. I told Intel when Pat G joined that they should speak softly and carry a big stick. That is Lip-Bu.
This may be an understatement but I still think Intel should stay together (design and manufacturing). Altera, Mobileye, etc... should be cut loose. Lip-Bu is a clever man and Intel will require a clever strategy moving forward. Copying Nvidia, AMD, etc... is not the way.
Intel also needs to shut the hell up. Intel should not have blasted BSPD and other innovations to the world before leveraging them inside. I told Intel when Pat G joined that they should speak softly and carry a big stick. That is Lip-Bu.
WSJ explains how a mistake Intel made in the mid 2000s has snowballed into an overwhelming obstacle for the tech company that Lip-Bu Tan will have to face.
Seems like this simpleton WSJ narrative misses the most salient details of that mistake in the mid 2000s. Otelini chose poorly with the iPhone - they gave up leadership on leading edge volume, to maintain margin leadership. Short term good for stock, long term bad for a manufacturing company.
-After Intel shuts down 3 or more older fabs, which is a historical process that comes around every 10 years or so, they will start to recover financially
What TSMC sold to customer is 100% markup (50%+ profit).
While TSMC US is 10% costly than TW.
Which means if Intel fab's close to TSMC US, they should have at least 40% margain.
they are not on track to what Pat said was going to happen. They are on track to what Dan and others said was realistic. essentially external is minimal until late 2027.
I get what you’re saying, but my point about S curves is that the ramp is initially very slow and the inflection point to huge growth is hard to pinpoint exactly. IMO “external revenues doesn’t start until 2026” and “external revenues doesn’t start until late 2027” are not very different from an S curves ramping perspective.
Question is whether the company can survive until the inflection point, not whether that inflection will ever come, and to this I fully agree with your point here:
Now the finances don't line up. Product revenue also dropped a ton since 2021 (FYI other companies did not). 2021 plan was to add 20B more per year in revenue and add 10B more in Capex/spending. Actual revenue went down by 20B. Plug that into a spreadsheet.... its not pretty. To do the capex, Intel sold off partial rights to their two most advanced fabs and changed depreciation model so depreciation lasts longer but is better short term. 18A Ramp does not improve the spreadsheet in 2026.
The cratering of Intel’s DCAI business (while CCG continues to carry the water as it always has) was not exactly foreseen (though perhaps should have been) when this Foundry ambition was laid out… surviving until the S curve’s inflection point is much, much harder now.
I get what you’re saying, but my point about S curves is that the ramp is initially very slow and the inflection point to huge growth is hard to pinpoint exactly. IMO “external revenues doesn’t start until 2026” and “external revenues doesn’t start until late 2027” are not very different from an S curves ramping perspective.
Question is whether the company can survive until the inflection point, not whether that inflection will ever come, and to this I fully agree with your point here:
The cratering of Intel’s DCAI business (while CCG continues to carry the water as it always has) was not exactly foreseen (though perhaps should have been) when this Foundry ambition was laid out… surviving until the S curve’s inflection point is much, much harder now.
I have published external revenue forecasts based on the S-curve. 100% agree it should have inflection
however:
1) Intel did not get the commitment in 2021 or 2022 that was expected. the few plans made disappeared.
2) Intel advanced packaging decreased over the past 1-2 years. And that was supposed to be Intels easiest growth.
3) Intel has not shown plans to have product strong external ramp in 2026 or 2027. They are not where the powerpoint forecast said they would be in 2021,2022
4) Intels first breakthrough GAA BSPD process was cancelled.
5) given the increasing uncertainty about Intel foundry future, does that make people more or less likely to commit volume to Intel?
So the when on the S and even the "IF" on the S seems like a valid concern
Is it increasing uncertainty? I personally think the uncertainty is decreasing: PDKs are more refined, EDA are fully featured on 18A (or its derivatives) and have announced 14A early enablement, presumably their "foundry services" have their @&#$ together more and more each day, they have a very successful "foundry guy" as their CEO now, Tariff Man is running the executive branch... So what makes you say "increasing uncertainty"? I'm genuinely curious.
Of all the trade deals brewing at the moment, the one most important to Intel is US-TW. And this trade deal will certainly be tough on TSMC. Until we know details, the uncertainty will remain.
If the TW-US tariff remain at 10%, and the operating cost differential is 10% for apples-apples US-TW fab costs, maybe that is enough for Intel to succeed. I tend to think its not enough to force 50% of TSMC customers to switch though.
We’ll know it’s fair if TSMC is forced to close a few (more than Intel) of their older, less competitive fabs.
Seems like this simpleton WSJ narrative misses the most salient details of that mistake in the mid 2000s. Otelini chose poorly with the iPhone - they gave up leadership on leading edge volume, to maintain margin leadership. Short term good for stock, long term bad for a manufacturing company.
CEO number 1 goal is shareholder value at that moment in time.
Working for US Company you realise they are obsessed with this to the detriment of long term planning.
Our US bosses utterly bamboozled when we not impressed with their spiel when they tell us employee comes below Shareholders , God and Family in that order.
A good summary of the conventional criticisms of Intel.
You need semiwiki to explore the other side, the opportunities:
-Vertical integration is growing in many industries. Intel represents “old is new again”.
-Intel graphics business has always been misunderstood. They own integrated graphics in a way AMD and NVIDIA can’t challenge. Qualcomm, maybe, distantly.
-After Intel shuts down 3 or more older fabs, which is a historical process that comes around every 10 years or so, they will start to recover financially
"After Intel shuts down 3 or more older fabs, which is a historical process that comes around every 10 years or so, they will start to recover financially"
How significant do you think in terms of cutting down the cost?
I think Intel's old fabs have fully depreciated. If Intel Foundry wants to offer mature node manufacturing service, those fabs might be useful.
This is arguing semantics on whether 20A and 18A are different "process [nodes]", which IMO they are not.
Is it increasing uncertainty? I personally think the uncertainty is decreasing: PDKs are more refined, EDA are fully featured on 18A (or its derivatives) and have announced 14A early enablement, presumably their "foundry services" have their @&#$ together more and more each day, they have a very successful "foundry guy" as their CEO now, Tariff Man is running the executive branch... So what makes you say "increasing uncertainty"? I'm genuinely curious.
On 20A, the process was announced and repeatedly displayed. I'm not the one who called it the first GAA BSPD process and different from 18A. But we may see the same claim again soon.... get ready for "14A has always been our real focus".
Intel still doesnt have money to pay for foundry (please do the math) .... and the deals created to spin it off or partner up were not finalized. This will become clear when we see foundry finances going forward. Unless LBT writes off everything in 2025 (which he might do). Then 2026 and 2027 foundry finance will still be problematic.
CEO number 1 goal is shareholder value at that moment in time.
Working for US Company you realise they are obsessed with this to the detriment of long term planning.
Our US bosses utterly bamboozled when we not impressed with their spiel when they tell us employee comes below Shareholders , God and Family in that order.
On 20A, the process was announced and repeatedly displayed. I'm not the one who called it the first GAA BSPD process and different from 18A. But we may see the same claim again soon.... get ready for "14A has always been our real focus".
You are correct, but one could argue that the claims were all Pat Gelsinger's over-optimistic world view on full display. They were claims fueled by hope and not cold, hard rational thought behind them. No argument he made the claims, but given that he is no longer with Intel, is it still fair to judge them by his claims?
As to 14A being the real focus, I believe it should have been the focus all along. The only world in which a new offering from an unproven foundry player is going to come out of the gate and get significant adoption is the one in Mr. Gelsinger's head. I believe the follow on 18-AP and 18-PT(?) nodes and 14A will have a much better opportunity than 18A. Gelsinger may have set unrealistic expectations, but if a few players give Intel small runs on 18A and have a favorable experience, that is what Intel really needs right now. If you don't think Intel is learning a lot on the 18A offering and that their follow up offerings will give a better experience then I think you are mistaken.
Fitch Ratings had this to say about Intel's financials as of 6Mar25. "The company's liquidity is solid, supported by substantial cash reserves and credit facilities. However, Intel's financial metrics, such as EBITDA leverage, remain higher than its peers, with forecasts indicating gradual improvement by 2026."
While Intel's position is not good and there is risk, That would seem to indicate that your claims of imminent failure are a bit exaggerated.
.