Intel Corporation (INTC) Deutsche Bank's 2024 Technology Conference

[Daniel Nenni]

Patrick Gelsinger
And as you think about the turnaround story for Intel, I'd really like to say that we've now begun phase two. You know, in phase one of the transformation was this rebuilding of the company and being able to get back to process and product leadership and laying out our IDM 2.0 strategy. And I think about that as the most significant rebuilding of the company since the memory to microprocessor, transition.

And this idea that we had to get back to process leadership. We described that as our five nodes and four-year journey. And we see the finish line in the site. And since we released the fifth of those, right, the PDK for Intel 18A, we've seen a market uptick in activities in the industry. We now have about a dozen customers that are actively engaged with us around that PDK.

We have eight product tape-ins that we expect to finish by the middle of next year. And of course, Panther Lake and Clearwater Forest, our first client and server product. And I'm happy to update the audience that we're now -- for this as a production process, we're now below 0.4d0 as defect density. As we touched on last night, Ross, this is now a healthy process that we're looking forward, and we'll start production wafers with Panther Lake before the end of the year. So, we're on track to deliver that.

We're seeing an uptick in activity from the industry since the PDK 1.0 went out. And as we've also talked about in our advanced -- in our foundry strategy, our advanced packaging remains a huge differentiator. And since the beginning of the year, we have about a tripling of our deal pipeline in advanced packaging. So that continues to be the on-ramp for many of our foundry customers.

https://seekingalpha.com/article/4717982-intel-corporation-intc-deutsche-banks-2024-technology-conference-transcript?

 

[Daniel Nenni]

Patrick Gelsinger
The second piece that's been disappointing is just the -- we underestimated, I underestimated the amount of heavy lifting beyond producing good wafers the EDA, the IP ecosystem that needs to get enabled to bring designs on to the foundry. So those have been the two areas that in this current environment have been a bit harder than I would have expected.

This is funny. Pat hired experienced people from TSMC and Samsung Foundry that know how hard building an ecosystem is. From what I have heard for the last year, and I heard it again today at the GF event, the Intel PDKs are not comparable to TSMC's. It is a very high bar that Samsung has missed over and over but in the foundry business the best PDKs win, absolutely.

 

[jim_22222]

Patrick Gelsinger
The second piece that's been disappointing is just the -- we underestimated, I underestimated the amount of heavy lifting beyond producing good wafers the EDA, the IP ecosystem that needs to get enabled to bring designs on to the foundry. So those have been the two areas that in this current environment have been a bit harder than I would have expected.

This is funny. Pat hired experienced people from TSMC and Samsung Foundry that know how hard building an ecosystem is. From what I have heard for the last year, and I heard it again today at the GF event, the Intel PDKs are not comparable to TSMC's. It is a very high bar that Samsung has missed over and over but in the foundry business the best PDKs win, absolutely.

It is not his fault. INTC should have sent him to a foundry as an apprentice to learn some PDKs for a few years. Instead, he was kicked to a storage/software firm, where he had been gaining irrelevant experience for ~ 10 years.

 

[Maximus]

Patrick Gelsinger
The second piece that's been disappointing is just the -- we underestimated, I underestimated the amount of heavy lifting beyond producing good wafers the EDA, the IP ecosystem that needs to get enabled to bring designs on to the foundry. So those have been the two areas that in this current environment have been a bit harder than I would have expected.

This is funny. Pat hired experienced people from TSMC and Samsung Foundry that know how hard building an ecosystem is. From what I have heard for the last year, and I heard it again today at the GF event, the Intel PDKs are not comparable to TSMC's. It is a very high bar that Samsung has missed over and over but in the foundry business the best PDKs win, absolutely.

Wasn't he supposed to be aware of all these challenges and their solutions before establishing the IDM 2.0 strategy?
Did he simply assume that Intel's manufacturing capacity was sufficient and technology was good so they can start the foundry business?"

 

[blueone]

Patrick Gelsinger
The second piece that's been disappointing is just the -- we underestimated, I underestimated the amount of heavy lifting beyond producing good wafers the EDA, the IP ecosystem that needs to get enabled to bring designs on to the foundry. So those have been the two areas that in this current environment have been a bit harder than I would have expected.

This is funny. Pat hired experienced people from TSMC and Samsung Foundry that know how hard building an ecosystem is. From what I have heard for the last year, and I heard it again today at the GF event, the Intel PDKs are not comparable to TSMC's. It is a very high bar that Samsung has missed over and over but in the foundry business the best PDKs win, absolutely.

I know Pat hired some senior people from Samsung, like Hong Hao, but Corp VPs don't create PDKs themselves. I wonder how many people with actual PDK development experience Intel hired? My guess is the number easily rounds to zero.

 

[jim_22222]

Wasn't he supposed to be aware of all these challenges and their solutions before establishing the IDM 2.0 strategy?
Did he simply assume that Intel's manufacturing capacity was sufficient and technology was good so they can start the foundry business?"

If you were to take on the task of building a boat and traveling around the world, would you know where the storm would be strongest?
Many folks in this industry believe in the following quote:

A good plan violently executed now is better than a perfect plan executed next week.

George S. Patton

 

[blueone]

It is not his fault. INTC should have sent him to a foundry as an apprentice to learn some PDKs for a few years. Instead, he was kicked to a storage/software firm, where he had been gaining irrelevant experience for ~ 10 years.

When PG left Intel he was an SVP. You seriously believe he'd become a PDK apprentice? And in Intel, which did not have anything like a foundry at the time? You can't be serious.

 

[jim_22222]

When PG left Intel he was an SVP. You seriously believe he'd become a PDK apprentice? And in Intel, which did not have anything like a foundry at the time? You can't be serious.

they could send him to tower jazz for example. which is better than nothing.
still paid as svp.
this is not one of those Elon Musk businesses.
otherwise they can just invite Elon to take over.

 

[blueone]

Wasn't he supposed to be aware of all these challenges and their solutions before establishing the IDM 2.0 strategy?
Did he simply assume that Intel's manufacturing capacity was sufficient and technology was good so they can start the foundry business?"

PG has never been a foundry customer, so I doubt he had any strong opinions. What it looks like is that he hasn't engaged any of his internal engineering teams to figure out what they really want. For example, it's been a long time since I worked for Intel, but the CPU teams back then didn't use licensed IP. Only the I/O teams, like networking did, and it doesn't smell like he engaged them.

 

[KevinK]

but in the foundry business the best PDKs win, absolutely.

And it even goes beyond the PDK. TSMC has set the bar that all the flows and IP the customer needs, must be ready and fully qualified via silicon test chips, both at the unit level and at the SoC level. Very long lead and expensive to do this ahead of customers.

 

[tooLongInEDA]

Patrick Gelsinger
And as you think about the turnaround story for Intel, I'd really like to say that we've now begun phase two. You know, in phase one of the transformation was this rebuilding of the company and being able to get back to process and product leadership and laying out our IDM 2.0 strategy. And I think about that as the most significant rebuilding of the company since the memory to microprocessor, transition.

And this idea that we had to get back to process leadership. We described that as our five nodes and four-year journey. And we see the finish line in the site. And since we released the fifth of those, right, the PDK for Intel 18A, we've seen a market uptick in activities in the industry. We now have about a dozen customers that are actively engaged with us around that PDK.

We have eight product tape-ins that we expect to finish by the middle of next year. And of course, Panther Lake and Clearwater Forest, our first client and server product. And I'm happy to update the audience that we're now -- for this as a production process, we're now below 0.4d0 as defect density. As we touched on last night, Ross, this is now a healthy process that we're looking forward, and we'll start production wafers with Panther Lake before the end of the year. So, we're on track to deliver that.

We're seeing an uptick in activity from the industry since the PDK 1.0 went out. And as we've also talked about in our advanced -- in our foundry strategy, our advanced packaging remains a huge differentiator. And since the beginning of the year, we have about a tripling of our deal pipeline in advanced packaging. So that continues to be the on-ramp for many of our foundry customers.

https://seekingalpha.com/article/4717982-intel-corporation-intc-deutsche-banks-2024-technology-conference-transcript?

Slightly off the main topic. Has anyone noticed just how often Pat starts his sentences with "And ..." ? Almost a 50% hit rate in this extract. Relentlessly banged into us at school never to start with "and" or "but". But I still do it ...

But encouraging that he's almost completely migrated from "I" to "we".

 

[tooLongInEDA]

they could send him to tower jazz for example. which is better than nothing.
still paid as svp.
this is not one of those Elon Musk businesses.
otherwise they can just invite Elon to take over.

But he *left Intel* didn't he ? He wasn't *sent somewhere* by them to get more experience. Think you're in danger of rewriting history here.

 

[siliconbruh999]

But he *left Intel* didn't he ? He wasn't *sent somewhere* by them to get more experience. Think you're in danger of rewriting history here.

He left cause board chose paul/bk instead of him iirc

 

[Maximus]

Patrick Gelsinger
Yes. And let's start on the negative side, the two things that have surprised me in the foundry journey. One is that the hideous memories of COVID supply chains have diminished, right? And I was on the phone with Secretary Remondo of Department of Commerce, right? And she was lamenting that people aren't more concerned about the resilience of their supply chains, right? And I said, me too, right? It's like 3 years ago, oh my gosh, right? Now it's like, oh, we're okay with our Asian supply chain. So that's been surprising that, that hasn't remained higher priority, right, for customers just say, disappointing.
----
This part is funny.
The reality is that when Intel asks, "Are you interested in Intel's IFS, which has less geographic risk?" everyone says yes. But when they ask, "Are you going to place orders with IFS?" no one says yes.

 

[siliconbruh999]

Patrick Gelsinger
Yes. And let's start on the negative side, the two things that have surprised me in the foundry journey. One is that the hideous memories of COVID supply chains have diminished, right? And I was on the phone with Secretary Remondo of Department of Commerce, right? And she was lamenting that people aren't more concerned about the resilience of their supply chains, right? And I said, me too, right? It's like 3 years ago, oh my gosh, right? Now it's like, oh, we're okay with our Asian supply chain. So that's been surprising that, that hasn't remained higher priority, right, for customers just say, disappointing.
----
This part is funny.
The reality is that when Intel asks, "Are you interested in Intel's IFS, which has less geographic risk?" everyone says yes. But when they ask, "Are you going to place orders with IFS?" no one says yes.

On this supply chain part i find it funny if some incident happens everyone will be left blank face like in covid
Anyways for the D0 part is 0.4 good i found a slide from tsmc N7/N5 while searching online considering 18A was going to be HVM ready in 2H24

 

[BruceA]

It is not his fault. INTC should have sent him to a foundry as an apprentice to learn some PDKs for a few years. Instead, he was kicked to a storage/software firm, where he had been gaining irrelevant experience for ~ 10 years.

It is his fault! He has some external people who could and can tell him what is required and what the gap and problems are. Sadly given how things are going and what we hear and see the listening and change isn’t happening fast enough

 

[tooLongInEDA]

It is his fault! He has some external people who could and can tell him what is required and what the gap and problems are. Sadly given how things are going and what we hear and see the listening and change isn’t happening fast enough

Indeed, I think he hired some of those people to lead the foundry effort. They're on at least their second leader there. The obvious question is "did he hire the right people" ? But you're right, there are plenty of external sources of information - EDA partners, potential customers - who could have (and likely did) tell Intel all this.

I'm fairly sure that all these people would have flagged the critical importance of the "heavy lifting" to build the EDA and IP ecosystem. Indeed, Intel's own statements over the past 2 years reflected that. And their previous foundry effort around 6-8 years ago (around the time of Intel's original 10nm process) would have taught them exactly the same things.

Given all this, I find Pat's statement quite astonishing.

Did Intel simply under-resource the effort ? They're certainly on record as spending quite a lot with Cadence and Synopsys to help build this ecosystem.

 

[BruceA]

Patrick Gelsinger
Yes. And let's start on the negative side, the two things that have surprised me in the foundry journey. One is that the hideous memories of COVID supply chains have diminished, right? And I was on the phone with Secretary Remondo of Department of Commerce, right? And she was lamenting that people aren't more concerned about the resilience of their supply chains, right? And I said, me too, right? It's like 3 years ago, oh my gosh, right? Now it's like, oh, we're okay with our Asian supply chain. So that's been surprising that, that hasn't remained higher priority, right, for customers just say, disappointing.
----
This part is funny.
The reality is that when Intel asks, "Are you interested in Intel's IFS, which has less geographic risk?" everyone says yes. But when they ask, "Are you going to place orders with IFS?" no one says yes.

Why is this a surprise ? No one will hitch their business to IFS, look what happened at 10nm to Intel!

Wait and see is the only wise choice and trust in TSMC till they falter or Taiwan gets invaded and oops

Show me the silicon and yield and maybe Intel will get the secondary customers, 2026 will be a telling year and if 5N4Y and IFS is real

 

[KevinK]

Why is this a surprise ? No one will hitch their business to IFS, look what happened at 10nm to Intel!
Show me the silicon and yield and maybe Intel will get the secondary customers, 2026 will be a telling year and if 5N4Y and IFS is real

I think the problem is different than 10nm. If you read the quote carefully, Intel can show the yield and the silicon, at least for Intel products, but their foundry enablement isn't as broad and deep as it needs to be to really play at the leading edge (per Pat "the EDA, the IP ecosystem that needs to get enabled to bring designs on to the foundry"). Back in 2014-2015 we had an "complete enablement roadmap" for 16/14nm for Intel ICF (the last incarnation of foundry), Samsung and GlobalFoundries that would put them on par with TSMC, but all drastically downsized what they were willing/able to do (EDA flow validation) and pay for (IP). ICF had other issues related to funky Intel flow, as well. But bottom line is that TSMC sets to bar really high due to the long-long-term evolution of their enablement and 3, 4 and even 5-way learning with EDA and IP companies. Yes, there were enablement projects that brought together 5 different companies to create friction-free, validated solutions for the end customer.

 

[hist78]

I think the problem is different than 10nm. If read the quote carefully, Intel can show the yield and the silicon, at least for Intel products, but their foundry enablement isn't as broad and deep as it needs to be to really play at the leading edge (per Pat "the EDA, the IP ecosystem that needs to get enabled to bring designs on to the foundry"). Back in 2014-2015 we had an "complete enablement roadmap" for 16/14nm for Intel ICF (the last incarnation of foundry), Samsung and GlobalFoundries that would put them on par with TSMC, but all drastically downsized what they were willing/able to do (EDA flow validation) and pay for (IP). ICF had other issues related to funky Intel flow, as well. But bottom line is that TSMC sets to bar really high due to the long-long-term evolution of their enablement and 3, 4 and even 5-way learning with EDA and IP companies. Yes, there were enablement projects that brought together 5 different companies to create friction-free, validated solutions for the end customer.

It's a very different mindset and operations between the IDM, like Intel, and the Foundry, like TSMC.

From TSMC 2023 Annual Report:

"In 2023, the Company manufactured 11,895 different products using 288 distinct technologies for 528 different customers."

It's not easy for any manufacturing businesses, including semiconductor, to deal with such volume and complexity.

Many years ago I heard a speech given by a CEO of a major global DVD manufacturer. They manufacture, package, and fulfill the DVD orders for major studios in different regions, countries, and languages worldwide. He said don't think it's an easy business. The volume and variety can cause a lot of troubles.

For example, one client complainted that a family with young children decided to have a fun Friday night cartoon watch party. To their surprise that the whole movie was R rated content instead of a children friendly cartoon. Although the title and description were totally correct on the case and DVD.