Intel Employees "Very Optimistic"

[Artificer60]

I think there's no point in making empty promises. PG said many things, but just look at where Intel is now...

This is the very reason I want something I can measure progress against. I want to see that what we are hearing from Intel is not more of the same. Given Lip-Bu Tan's past record I'm inclined to believe he will do what he says, but I want a way to validate that belief.

I'd also point out the two things in your list above that are measurable (product launches and IFS breakeven) were laid out under Gelsinger.

 

[hist78]

Thanks for your detailed analysis/info on TSMC's fab-size strategy. Perhaps the Arizona-fab is the new US-version of a connected-GigaFab coming into existence? See also your earlier thread about this, I was not aware of it, my apologies:
https://semiwiki.com/forum/threads/tsmc-phoenix-arizona-fab-site-plan.16066/

Very impressive to scroll through a couple of recent 2025-photos where you can see the shell of Phase 2 done; the groundwork of building Phase-3 started end of April-2025:
https://maps.app.goo.gl/t6U4F29pQezG7fR8A

Here again the link on the story about "super-Arizona-CEO" Ying-Lang Wang:
https://cwnewsroom.substack.com/p/key-figure-behind-tsmc-us-expansion-yinglang-wang
Ying-Lang Wang has long overseen TSMC’s fabs in Southern Taiwan Science Park, the center for 5nm and 3nm production—a critical growth driver that helped TSMC pull ahead of Intel and Samsung in recent years.

In April 2023, he was urgently sent to the U.S. to take over the troubled Arizona fab, successfully leading the team to begin mass production of 4nm chips earlier this year.


Indeed a formidable activity by TSMC in Arizona, in combination with the new packaging plant and "R&D"-center in the future at that site.

TSMC refers to its Arizona project as Fab 21, which includes at least six planned phases. Each phase is designated for a different process node, with Phase One designed to produce approximately 24,000 300mm wafers for N4 per month. If you combine the capacity of all six phases, it would qualify as a GigaFab by TSMC’s standard. However, using Intel’s definition, these would be considered six individual smaller fabs.

 

[Daniel Nenni]

Nope, I don't know a soul there. His performance at Cadence was nothing short of amazing. It is what gives me hope he can turn Intel around.

I feel like you may have missed my point. So far Lip-Bu Tan has essentially given a list of goals (build trust, deliver promised products/services on time, etc.). However, he has given no timelines nor elaborated on how he measures progress towards these goals. That is the only thing I have an issue with. How am I to determine if his plan to achieve these goals is on track, ahead or behind?

I see your point but it seems to me a bit naive. Cutting a big part of a Intel is a painful process, one that should not be taken lightly. Why would Lip-Bu layout detailed plans to people who will no longer work there? Or are you talking about something like Pat's 5N4Y marketing schtick? Was that really motivating for people in the manufacturing trenches?

 

[dkr1986]

This is the very reason I want something I can measure progress against. I want to see that what we are hearing from Intel is not more of the same. Given Lip-Bu Tan's past record I'm inclined to believe he will do what he says, but I want a way to validate that belief.

I'd also point out the two things in your list above that are measurable (product launches and IFS breakeven) were laid out under Gelsinger.

Pat Gelsinger inherited inferior node process tech (10nm) & inferior product road map (Sapphire rapids on DC CPU - Alder lake was good on client side but didn't address the notebook battery life aspects) and managed a business under external pressure from degrading PC business after post COVID boom due to inventory glut & loss of DC CPU business due to hyperscalers extending their server useful life post COVID & emergence of generative AI getting all the hyperscale capex.

Lip Bu Tan on the other hand is inheriting a competitive node process tech (even if 18A is only N3P comp) & competitive product road map (DMR, PTL, NVL) in Intel's core markets. He is also coming in just in time for Windows 10 EOL PC refresh cycle and at the end of that post COVID server useful life extension. All he has to do is drive operational efficiency and get customers to sign on for foundry. (& Add competitive AI products to get some revenue)

 

[dkr1986]

are you talking about something like Pat's 5N4Y marketing schtick? Was that really motivating for people in the manufacturing trenches?

Oh Come on! 5N4Y strategy also came with increased investment in R&D for accelerated foundry node development process. It does not matter how many nodes were developed & cancelled with some delay, the end goal is catching up in process node tech which they have done more or less done. All things indicate 18A is N3P/ N2 comp node depending on which letter you focus on PPACT!

 

[nghanayem]

Was that really motivating for people in the manufacturing trenches?

1000%. Ann's (not Pat's) 5N4Y plan was a north star for people in the trenches to follow. Going back to Intel's long-standing status as the unquestioned process technology leader was top of mind and a measurable goal to drive your accomplishments towards. That goal is also what motivated ex Intel people who left from the toxic culture and underinvestment of the 2010s to come back from tool vendors, TSMC, Micron, SK, GF, etc. as well as folks who were laid off BK wanted to do cuts because of his failed mobile push.

 

[Daniel Nenni]

Oh Come on! 5N4Y strategy also came with increased investment in R&D for accelerated foundry node development process. It does not matter how many nodes were developed & cancelled with some delay, the end goal is catching up in process node tech which they have done more or less done. All things indicate 18A is N3P/ N2 comp node depending on which letter you focus on PPACT!

Lip-Bu's point is that you should overdeliver on your promises not underdeliver. TSMC has the same strategy. Which do you think Wall Street prefers?

1000%. Ann's (not Pat's) 5N4Y plan was a north star for people in the trenches to follow. Going back to Intel's long-standing status as the unquestioned process technology leader was top of mind and a measurable goal to drive your accomplishments towards. That goal is also what motivated ex Intel people who left from the toxic culture and underinvestment of the 2010s to come back from tool vendors, TSMC, Micron, SK, GF, etc. as well as folks who were laid off BK wanted to do cuts because of his failed mobile push.

Yes, Pat said Intel would beat TSMC and then he retired unexpectedly. Had Pat said Intel would beat Samsung Foundry he may not have retired unexpectedly. What about BK? He had so much Intel manufacturing and operations experience? He got fired as well and it was not due to his failed mobile strategy (Otellini botched mobile for Intel), it was BK's mishandling of 14nm. Of course Intel said it was because of a past consensual relationship. Just like Intel said Pat retired.

How are employees motived by lies, pipe dreams, and massive layoffs? We call it drinking the Corporate Kool-Aide and Intel is famous for it. Lip-Bu is not famous for it so get ready for change.

 

[XYang2023]

Before/After/During the layoffs? How many heads at the end of the project? How long to reach the end state?

I have not seen the details on this roadmap, nor even a timeline on when it should get rolled out.

Launch Dates? I've seen EOY for Panther Lake and I think Q1'26 for CWF. The others are still speculative unless I missed something.

This one is clearly measurable. No complaints here

How many should I expect? When?

As I said. Largely a list of goals lacking in clear milestones. Lip-Bu Tan may have something in mind, but he hasn't communicated that to the market. I agree it is fairly clear on what he wants to do, but in most of these cases I don't have a good way to measure progress towards the goals.

I think the reason you and some Intel employees lack a clear sense is because you don't follow Intel closely enough. How is it that I don’t even work for Intel, yet I have a better understanding of what’s going on?


Regarding hiring, DZ mentioned that one of Lip-Bu’s key priorities is to rebuild Intel’s talent pool:

There has already been progress:

https://twitter.com/x/status/1924444123479888043

As for the AI strategy, you can find it here:

https://twitter.com/x/status/1923121554314919999

On launch dates, I won’t elaborate because they’ve already addressed this during the investor meeting—emphasizing "execution." MJ and DZ also added more context, particularly about launching products on A-stepping silicon.


They’ve already stated that for 14A, they’ll need a partner. Lip-Bu has proactively visited all potential customers to gather both positive and negative feedback.

 

[dkr1986]

Lip-Bu's point is that you should overdeliver on your promises not underdeliver. TSMC has the same strategy. Which do you think Wall Street prefers?

Only thing Lip Bu promised so far is to "we are going to delight customers". Lets hope he overdelivers on that.

Pat said Intel would beat TSMC

Pat never said that afaik. He said Intel will achieve unquestioned process leadership with 18A at the end of 5N4Y and in terms of foundry, he said they will become the second largest foundry. There is going to be an Intel 18A product end of this year (one PTL SKU). If that product achieves better performance than TSMC's leading edge process N3P\N3X at that time, then Intel likely does achieves that. Afaik, there is no N2 based products this year and AMD being the first N2 customer likely launches their products late next year (Computex 2026 at best). AMD's EPYC Turin launched in October 2024 iirc, so by the time TSMC's N2 PC\HPC products launch next year by AMD, Intel would have shipped millions of 18A silicon. imo. Remember C.C Wei claimed N3P is comparable to 18A on PPA based on their internal assessments and I still think he was overselling his product in that statement. Let's see how it goes.

 

[Daniel Nenni]

Pat never said that afaik. He said Intel will achieve unquestioned process leadership with 18A at the end of 5N4Y and in terms of foundry, he said they will become the second largest foundry. There is going to be an Intel 18A product end of this year (one PTL SKU). If that product achieves better performance than TSMC's leading edge process N3P\N3X at that time, then Intel likely does achieves that. Afaik, there is no N2 based products this year and AMD being the first N2 customer likely launches their products late next year (Computex 2026 at best). AMD's EPYC Turin launched in October 2024 iirc, so by the time TSMC's N2 PC\HPC products launch next year by AMD, Intel would have shipped millions of 18A silicon. imo. Remember C.C Wei claimed N3P is comparable to 18A on PPA based on their internal assessments and I still think he was overselling his product in that statement. Let's see how it goes.

Let's see how it goes? It has already gone.

This is the problem with Intel in a nutshell. Do you know how CC Wei knows that N3 is competitive with 18A? Because his customers told him so after evaluating the PDK. Do you not think that customers were part of his internal assessment? He cannot say that of course but I have been told the same. I'm sure Lip-Bu already knows this because he listens to customers.

But sure, 18A is the best and don't worry about TSMC N2 even though it has more design starts than N3 which was uncontested. Hopefully Lip-Bu cuts everyone at Intel who thinks this way otherwise I see little hope of a turnaround. Sorry to be so blunt but I really want to see Intel succeed.

 

[Daniel Nenni]

"Yes, Pat said Intel would beat TSMC"

Pat never said that afaik.

Pat did say that along with AMD is in our rear view mirror. Is AMD really in Intel's rear view mirror?

 

[dkr1986]

This is the problem with Intel in a nutshell. Do you know how CC Wei knows that N3 is competitive with 18A? Because his customers told him so after evaluating the PDK. Do you not think that customers were part of his internal assessment? He cannot say that of course but I have been told the same. I'm sure Lip-Bu already knows this because he listens to customers.

CC Wei made that comment in Q3'23 even before Intel launched 18A PDK 1.0 in July 2024 (Although I am not sure if that would have made any difference). That statement is probably made based on 0.9 PDK metrics. Maybe that is all they needed to make that statement but still Intel 18A ~ N3P is very good achievement imo.

I would be very cautious using the 18A external PDKs based assessments as a measure of 18A performance of Intel's own products. There has been reports that Intel's external PDKs are not good (as you reported here many times) and missing lot of libraries etc that could skew the real performance of Intel 18A. Intel Product design team on the other hand most likely has the know-hows to make this work for their own products as they seem pretty confident about Panther Lake competitiveness.

But sure, 18A is the best and don't worry about TSMC N2 even though it has more design starts than N3 which was uncontested.

As explained above, it is the PDK issues and immature ecosystem that is being a barrier for 18A design wins from external customers. Also not everything needs to be looked at from external foundry POV here, In 2025 & 2026, the important thing for Intel is still Intel's own 18A Products vs AMD's TSMC n-1 node Products. 5N4Y is not a strategy only laid out for IFS, it is also for the benefit of Intel's own products.

Let's see how it goes? It has already gone.

I personally would like to wait until 3rd party reviews of products on these node process to come out before I pass judgement on them. So again lets see how it goes when that 1 PTL SKU launches later this year.

Pat did say that along with AMD is in our rear view mirror. Is AMD really in Intel's rear view mirror?

That is one statement I think was one of Pat Gelsinger's worst moment as the CEO, imho. But IIRC, AMD was in the rear view mirror when that statement was made in end of 2021. It was Alder Lake (Intel 12th Gen) vs Zen 3 in client at least. Intel definitely claimed the Performance crown but it was short lived as Zen 4 launched later to reclaim (or as Pat would have said AMD passed Intel and left them in dust).

Pat did say that along

I have never heard or seen him say that but you are the industry person. If there is an actual quote or interview that he said that, I am willing to accept I was wrong on that.

 

[XYang2023]

CC Wei made that comment in Q3'23 even before Intel launched 18A PDK 1.0 in July 2024 (Although I am not sure if that would have made any difference). That statement is probably made based on 0.9 PDK metrics. Maybe that is all they needed to make that statement but still Intel 18A ~ N3P is very good achievement imo.

I would be very cautious using the 18A external PDKs based assessments as a measure of 18A performance of Intel's own products. There has been reports that Intel's external PDKs are not good (as you reported here many times) and missing lot of libraries etc that could skew the real performance of Intel 18A. Intel Product design team on the other hand most likely has the know-hows to make this work for their own products as they seem pretty confident about Panther Lake competitiveness.

As explained above, it is the PDK issues and immature ecosystem that is being a barrier for 18A design wins from external customers. Also not everything needs to be looked at from external foundry POV here, In 2025 & 2026, the important thing for Intel is still Intel's own 18A Products vs AMD's TSMC n-1 node Products. 5N4Y is not a strategy only laid out for IFS, it is also for the benefit of Intel's own products.

I personally would like to wait until 3rd party reviews of products on these node process to come out before I pass judgement on them. So again lets see how it goes when that 1 PTL SKU launches later this year.

That is one statement I think was one of Pat Gelsinger's worst moment as the CEO, imho. But IIRC, AMD was in the rear view mirror when that statement was made in end of 2021. It was Alder Lake (Intel 12th Gen) vs Zen 3 in client at least. Intel definitely claimed the Performance crown but it was short lived as Zen 4 launched later to reclaim (or as Pat would have said AMD passed Intel and left them in dust).

I have never heard or seen him say that but you are the industry person. If there is an actual quote or interview that he said that, I am willing to accept I was wrong on that.

Around that time, PG said that Intel’s management consisted of the very best people in the industry. Meanwhile, Sandra from DCAI mocked Nvidia GPUs during one of Intel’s events. Also around that time, when Nvidia was demonstrating AI boundaries in their events. PG, on the other hand, chose to high light AI safety in a grandparent-grandchild conversation format in their tech event. Honestly, I think during his tenure, he focused on things that didn’t improve Intel’s competitive position.

From what I understand, he has always been that way. I think it is better for someone else to lead as CEO.

https://twitter.com/x/status/1930335255275413762

 

[Daniel Nenni]

CC Wei made that comment in Q3'23 even before Intel launched 18A PDK 1.0 in July 2024 (Although I am not sure if that would have made any difference). That statement is probably made based on 0.9 PDK metrics. Maybe that is all they needed to make that statement but still Intel 18A ~ N3P is very good achievement imo.

I have never heard or seen him say that but you are the industry person. If there is an actual quote or interview that he said that, I am willing to accept I was wrong on that.

I heard Pat say that at the first Intel Foundry Conference. I am considered media/analyst so I get access. But I heard Pat say a lot of things that did not jibe. He is a shoot from the hip type of person so that happens. Don't get me wrong I liked Pat. BK I did not like, we met at an Intel Developers conference. He was horrible with media/analysts. The problem is that most media/analysts have zero semiconductor experience and I have 40 years in the trenches so we don't necessarily see eye to eye.

CC Wei does not oversell. TSMC is the trusted foundry for a reason. Misleading customers is a very bad thing because they will know the truth soon enough. You have to understand the power of the ecosystem that TSMC has built. There are no secrets because we all know each other and we talk frequently. Me more than others since I manage the business side of SemiWiki and attend quite a few ecosystem conferences every year. My day job is an advisor to emerging EDA and IP companies. I help with business plans, funding, market strategy, getting established amongst TSMC's top customers, and I do exits.

Bottom line for me it is customers that determine what foundry process technology is the best based on PDK evaluations, test chips, and end products. TSMC N3 won by a landslide and TSMC N2 looks to be an even bigger landslide win. I can assure you Lip-Bu will use a similar metric because without paying customers it is a fail no matter how you look at it.