Intel Aktie

Welcome to this session at our Bank of America Global Technology Conference. I'm Vivek Arya from BofA Semiconductor and Semicap Equipment Research Team. I'm really delighted to have the team from Intel join us this afternoon for this fireside discussion, David Zinsner, Executive Vice President and CFO -- just wanted to make sure, not exit Executive Vice President and CFO; and John Pitzer, Corporate Vice President of IR and Treasurer. And as always, I'll start with actually a disclaimer statement then go to my questions, but please feel free to raise your hand if you have a question.

So I'll just start with the disclosure. Before we begin, please note that today's discussion may contain forward-looking statements that are subject to various risks and uncertainties and may reference non-GAAP financial measures. Please refer to Intel's most recent earnings release and annual report on Form 10-K and other filings with the SEC for more information on the risk factors that could cause actual results to differ materially, and additional information on Intel's non-GAAP financial measures, including reconciliations where appropriate to the corresponding GAAP financial measures. I'm glad I didn't have -- I remember all that.

Okay. Wonderful. So with that, really warm welcome to you, Dave and John. Really glad that you could join us. Before we kind of get into all the exciting stuff about CPUs and IFS, Dave, you've been on this almost a 14-month journey, right, when Lip-Bu joined the company. Maybe just walk us through what are kind of the strategic directions that you thought about at that time? What are the actions? Because one thing that we have been surprised us from the outside is when Lip-Bu joined is we thought Intel would go to a more product-focused company versus a manufacturing company. And it's actually been -- one could debate more of the right mix versus being more manufacturing focused, right, than before. But walk us through your perspective, right, at the same time?

Yes, I think when Lip-Bu started as he -- or maybe before he started as he's making the assessment around why does Intel exist, I think we saw 2 things. One, a legacy of building CPUs over decades in x86. And second, a prowess to call it in manufacturing, both in terms of front end and back end. And those are the things that we have done over the history of Intel. And obviously, execution issues are prevalent, but it's kind of the core underpinnings of the company. And as he comes in and looks at that, you look at -- it's largely just kind of treading along on those 2 aspects, you get a GDP, GDP-plus level revenue growth, then you get pretty good margins, and that's what you live off of. And by the way, that's how Intel operate, at least for the prior decade.

But then there's 2 dynamics that are going on as he comes into the scene, and one felt earlier than the other. One is on the manufacturing side, there is this clear need for a resilient and more secure supply chain, and there's a need for a company that can provide that for logic that is based in the U.S. that does development and does the manufacturing in the U.S., which we have that position. And then as we progressed going through the month, earlier months of Lip-Bu's tenure, it became apparent that the CPU market, which ordinarily would be this kind of like bump and a long growth business is actually going to have explosive growth as a result of AI, and I know we'll go into those 2 things.

And so the market is now moving in the favor of Intel. This then becomes, well, okay, why is Intel not being successful? And it's really an execution issue at its core. And if you step into, okay, it's an execution issue. Why is there an execution issue? And then there's a number of different aspects of it, but it is basic element it's a cultural thing. We had a cultural problem at Intel. And so I think more than anything, that's what Lip-Bu addressed early on while he was there, while he was still learning a lot of the things about how the business is run. He was working on culture from day 1.

And you can kind of see it in the things that he did early on. We had 12 layers of management. He collapsed it into 6 layers of management. We had, I think at our peak, over 400 vice presidents, he collapsed that to 200. We had over 100,000 employees. He collapsed that to under 80,000 employees. And when he came in, he made a comment, which I think was maybe the biggest issue around culture at Intel, which is he said, "Hey, if we have a problem and you come to me, then we have a problem and it's something we have to work out. If there's a problem, and you keep it to yourself, then you have a problem and you're going to get fired."

And that really drove -- that, combined with this restructuring of the organization and the collapsing the layers and so forth, really drove a level of transparency that we just didn't have before. And of course, when you're dealing that in the early months of that activity, it sucks because you are finding out a whole lot of information about stuff that's going bad that you never realize was going bad. But it allowed us to kind of work through a lot of the issues early on that has improved our execution dramatically as exemplified by the results of last quarter and guide for this quarter.

And then on top of that, the other thing I think he really wanted to drive is kind of a financially disciplined company. And of course, music to a CFO's ears, that started with the balance sheet. We strengthened the balance sheet significantly in his early months in the company. And then on top of that, really worked the financials of the company to improve the cost structure, improve margins. We took a bunch of expense out on operating expenses that also kind of built it. So now we're actually in a little bit more of the blocking and tackling, quite honestly, of the business, which just drive relentless execution on the product portfolio, whether it be in terms of products coming out in client and data center or that's delivering process and improving yields at the right clip, bringing out the next process, improving those yields at the right clip.

Maybe one last thing I think he did and actually, quite honestly, hasn't even put the finishing touches on it, but we're close is he was looking for the best out there in everyone's individual discipline come on to the company. And I'm sure, early on, it was a challenge for him to get recruiting. As the stock started to perform, that helped out a lot. But we have put a number of new people in seats at the CEO staff level, let's call it, people that are world-class in their specific area, a lot of them obviously coming from the outside with a kind of a unique fresh perspective. That said, there is plenty of talent inside and he identified the ones that had the talent inside, elevated those people into positions that allow them to improve execution, improve the company and so forth.

Got it. So big kind of culture change.

On the mix, I think it's pretty balanced, by the way, between products and manufacturing or foundry. In fact, I'm sitting on a lot of the meetings that Lip-Bu conducts and I would say it's probably 50-50 as to what we spend our time on, 50% of the time we're spending on things related to the product business, 50% of the time we're spending on things that are related to the foundry business.

Got it. So let's first talk about the product side. And as you mentioned, the CPU growth has kind of caught this industry, I guess, from the outside caught us by surprise. What I'm also surprised by is just the pace of how the TAM expectations have been upgraded, right, like 60 to 100 to 120, and now the latest is 200. But it is Intel call in this spectrum? And more important to that, Dave, how does one conceptually -- is there a simple unit time ASP math so we can understand that number?

Yes. I think I would say we have scenario planning that we do, and we have a few different possible scenarios as to how things play out in the CPU space, in particular in the data center space, which I think you're talking about. I think probably the easiest answer to that is probably hard to articulate exactly which number we're going to hit. I think it's going to be a big market.

Obviously, the ratio of CPUs to GPUs is growing meaningfully as we get from training to inference, inference to agentic and multiagent and reinforced learning. So it's just going to drive a lot of CPU requirements. Quite honestly, big is enough. I mean we've got enough demand out there that if we can do a good job executing on the ramping of supply. We should have no issue with growing our revenue meaningfully in the data center space.

Got it. One thing that there seems to be at least some consensus on is that if one were to segment that server CPU market, right, there is kind of your traditional enterprise type market. then you have the sort of head note, what I would describe as within the AI cluster. And then you have the standalone agentic CPU racks. Is that the way you think about? And is there a simple way to kind of -- so how would you segment -- even if there isn't like an absolute number, whatever it happens to be $100 billion, $200 billion, is there a kind of percent by x percent, y percent segmentation?

Probably not something I want to kind of double-click into at this point. I would say as you look at it from an enterprise to hyperscale level, it used to be enterprise was above hyperscale. And obviously, that shifted meaningfully with what hyperscalers are doing in terms of AI workloads. And then as it double clicks, probably difficult to make the percentages or I should say, predict good percentages for the individual pieces.

All of those markets are going to be big markets. There's no question about that. We're not segmenting it that way. We just want to sell CPUs into as many of those markets as possible. I would say what we are really good at is single-threaded performance. So things that operate better as a single single-thread performance are areas we're going to have, I think, a good position and a good participation in.

Got it. Now the CPU to GPU ratio, right, without going into the technical aspect of it, would you agree that the biggest part of the CPU growth might be in places where it is kind of bundled with the accelerator, whatever that accelerator happens to be, right, whether it's a GPU or CPU or whatever. So unless Intel has a competitive accelerator, does that limit -- is that a right pushback, you feel?

Yes. Well, we are bringing out Crescent Island. So we are in the accelerator business. We're also partnered with NVIDIA to operate with them. So I think that will clearly be a significant market, and we think we have a good position to participate meaningfully in that area. But there'll be -- there's a lot of work that CPUs have to do besides just what's going on in the core rack system around storage and orchestration that sits away from all of those -- that workload, and that's also going to be a significant market. It has to be because CPUs, in most cases, are the only way to efficiently manage that activity.

Got it. Now one thing, Dave, just because you were in other companies, right, where shortages really played a big part of showing the growth, but then over a longer period of time, as soon as the shortage went away, right? Pricing and other factors changed right very quickly. So from what we are seeing in the CPU market, right, today, if you can make a CPU, you can sell it, right? At least that's how we perceive it.

But do you think it's easier to resolve shortages in server CPU because we're not talking about different kinds of products, right? We're just talking about capacity allocation type decision. So let's say your foundry competitor were to make a different capacity allocation decision. Does that impact your CPU opportunity? Like how much of this is just the here and now of a shortage phenomena versus a large secular growth market?

In terms of demand or in terms of pricing?

In terms of the demand and pricing, right. Also, that matters.

Well, I think demand, just talking to -- I talked to all the CFOs of the hyper scalers and look at, a, what they think in terms of their overall CapEx, what they think of their CPU percent of that total CapEx. One, it's clear over a multiyear period, they see demand going up meaningfully. And two, they also see the percentage of that CapEx going up. So I think I feel pretty comfortable that this year, next year, the following year, the demand is going to be strong. I'm not worried about that.

On the ASP front, I mean what we try to do is we want to give the customer a product that performs to a level that they give us commensurate pricing that goes along with that, and that's mainly our focus. I mean, this isn't the memory Market. My last company, where things supply understrips demand and pricing spikes and it goes the other direction. That's not this market. I'd say we do see some inflationary pressure. So we are seeing ASP increases. That is obviously going to impact the top line over the course of the near term. But largely, we're not driving our ASP decisions off of supply/demand imbalances.

What we are doing is trying to drive more long-term agreements with customers. So we're locking in a price, for sure. We're locking in a volume commitment. And then that enables us to do a better job of planning out our capacity and making sure when we're investing in capacity, we're going to see customers take that supply when it comes off the line.

Got it. So in Q1, what we saw was your server CPU business grew, I think, in the 20%, 25% kind of range, mostly through ASP. I think that's the data we got. So some of these LTAs and other discussions, do you think that kind of goodness and pricing that is still to come? Or have you already seen it?

Yes, there will be price -- I would say, keep in mind on the data center side, core count is going up. So ASPs, by virtue of that go up. You know what I mean? Everything you put more -- if ASP per core is roughly stable, you're going to see ASPs go up in data center regardless just because of the core count. But that said, even on an ASP per core basis, that had generally been a dynamic of falling pricing. It has certainly stabilized. And in certain cases, we are seeing like-for-like ASP per core pricing going up. And I think that has a solid path towards realizing that over the longer term.

Got it. Now the competition, right, first, within the x86 price side, I think your competitor has outlined a growth rate of almost like 70% for this year. And they are saying that almost 2/3 of that is units rather than pricing, and they are talking about preferential allocation from their foundry. And then on the ARM side, and I realize you can participate in ARM through IFS, and we'll come to that question. But how do you think about the competitive landscape, Dave? Like does Intel have a good competitive part that can withstand competition in x86 and ARM? Or do you think that is still under...

Yes. I think you said it best. You could -- if you just stamped something and called it a CPU right now, it probably would sell. So in the near term, it's all about supply. And we are ramping supply as we speak. There will be a meaningful increase in supply of 3 and 18A over the course of the next, call it, quarters or so. That will ramp up to meet what we're seeing from a demand perspective. We're even actually this year going to see Intel 7 go up in terms of wafer starts, at least initially and then probably next year, I think we can start to wind it down a little bit in '27 and allow Intel 3 and 18A to pick up the slack.

Longer term, it's going to be about products. We have to deliver products that are performing. On client, I think we're in a stellar shape. You could argue right now, maybe we were a little bit weaker on the desktop side. But 18A on the notebook side is ramping, it is ramping as fast as -- I mean I think it's the fastest ramping product we've had at least in the last 5 years.

5-plus years, yes.

On the client side. So I mean, it's selling as much as we can produce. And our ramp in general has been quite good, mainly because 18A is now making really good progress. on yields month-to-month, which is helping a lot. And we also are ramping facilities now both in Oregon and in Arizona, which is helping us, I guess, a lot. And then what was the other part of the question?

Yes, just competition versus ARM.

Yes, right. Yes. And then on the data center side, I would say we have. In certain cases, we have a very strong product. In other cases, particularly in multi-threading, not as good. And when you look at Diamond Rapids, which is the product after Granite that will come out, that does not have multi-thread, SMT basically. And so we will have it back in Core Rapids, which will be the next product. So this was back to kind of cultural changes.

This was Lip-Bu listening to customers, hearing what their needs were and then adapting the product portfolio and product road map to what he was hearing. This is an example of that. We've made steady progress in terms of improvement. I think Granite is certainly better than prior retention core will definitely be better than prior products. And Diamond, in its swim lane, not counting the SMT, will be better than Granite.

So we'll make steady improvement. I think we have a really good road map. We brought in a new person from the outside, what was it like 6 months ago or something like that, Kevork. He's doing a good job like tearing things down and making sure that we're back to executing at a high level. It helps that we have leveraged our own nodes in that area. And now those nodes are becoming more performing to 18As, what they call PPA price. Our power performance area is much better than Intel 3. So as we kind of migrate our process nodes, we should see continually better performance.

Recently, we saw NVIDIA also announced their client portfolio at Computex with the N1X that they are designing but with MediaTek. Do you think of that as kind of competition at the high end of the market? I know a lot of it was kind of advertised in advance, so it should not have been a surprise one way or another but do you see that as kind of incremental competition in the demand?

Yes, of course we always look at ARM-based solutions nice competition in the client space. I think we've shown we can do pretty well against those solutions given our share and performance. So I feel really confident about our portfolio. but we take all competition seriously. And we aim to always bring out the next product to be better than the product before.

I just might add on remember last fall, I validated the x86 ecosystem with their collaborative agreement with them across both client and data center. And as we made that announcement last fall, we kind of told the investment community. This isn't going to stop us from wanting to bring our own discrete GPU to market. just like it's not going to stop NVIDIA from wanting to bring their own discrete CPUs to market. You should really think about it as an additional arrow in their quiver. But to Dave's point, we feel pretty confident about not only the value of the x86 ecosystem but our road map there as well.

I think the other really important point as you think about the client market is all of the buzz that we've been talking about today is really around server CPU. Client is going to have a big statement to make in the AI infrastructure as well because we sort of on the last earnings call, we talked about CPU density, we stopped at agentic and multiagent. Beyond agentic and multiagent, you're going to have physical AI and you're going to have AI at the edge. And a lot of that demand is going to be serviced by client-based CPU platforms, and that's going to be a significant growth -- a new growth opportunity for us. And we would expect that because it's such a great growth opportunity to see additional competition over time.

And by the way, we're already strong in the edge. We've been doing edge for a while. So that obviously gives us an advantage. And I would just say Panther Lake's already in physical AI solution.

And you recently made a hire, right, from Qualcomm, right?

Alex.

From that market, okay. Let's talk about IFS. If you could just help us guess at the record straight, Dave, which is where is 18A in terms of its yield on an absolute basis? Is it comparable to where your foundry competitors are? If not, when can it get to that level where you can say, no questions asked, this is as competitive as you can find?

I would say it this way, I don't know, early last year, I think the challenge around 18A was 2 things. One, we tried to do too much at once. And it took a while to get that settled, and we heard that from customers. And I think second is we were trying to play performance and yield and trying to improve both at the same time. It was like trying to fly the plane and fix the wing at the same time basically. And what we -- when I was king for a day, I took Naga and I put him over TD and Manufacturing. And then they really just focused on first, stabilizing performance. And so they stabilize performance. Then once you've got your performance stabilized, then all you do is you work yield every month.

The second thing that we did when Lip-Bu joined is we really opened up our data to our vendors to really help us learn things that we could do to improve yield and that made a dramatic difference. Surprisingly, it seems like 101, but there was a resist back to the culture thing, there was a cultural resistance towards providing significant amount of data. to vendors. Once we fixed that, we really started to get some feedback into what we could do to improve. And then it was just our team just grinding it out every month. And probably was a little bit up and down in the first couple of months. But after that, we made basically what industry standard improvement over month would look like we were seeing that improvement.

And then, of course, we extrapolated that okay, at the end of '25, we want to be here; at the end of '26, we want to be here; at the end of '27, we want to be here. And I would just -- and ultimately, the goal is to get to the yields that generate great margins is kind of out in this end of '27 time frame. I would tell you that based on the progress we've made to date now, we are likely going to pull in those milestones by at least a quarter. potentially even a little more. So that, I think, is a testament to, hey, things are actually going better than planned and have been over the course of the last the last several quarters.

That said, there's more in front of us, and we've got to do more to drive the right level of margins. Right now, we look at Panther Lake kind of on a consolidated basis, those margins aren't to the level that they're at least neutral, if not, hopefully, over time, accretive to gross margins of the total company. So more work to be done. But the progress to date has been very good, and the yields are actually at a respectable point at this point.

Got it. So just to understand, pulling by a quarter means that if in the past, your yield progress would have led to achieve breakeven in IFS by the end of '27. Now you can pull it by a quarter. Is that...

Yes. Relative to the milestones we set forward set forth right, exactly. Now I would just say we have a more aggressive plan for 14A than 18A. And when you look at -- kind of it's obviously early and it's its maturity. But when you look at kind of yield and performance measures at this point in time and maturity of 14A compared to that same moment in time for 18A, we're ahead. So we're already ahead on 14A. This is a way more an industry standard PDK that we're bringing out for 14A.

So now the advantage is it all the stuff that I said that we bit off more than we can chew on 18A, and it really took some time. Now it's just a little bit of a rinse and repeat. I mean engineers are going to hear this and go bonkers on me, it's a lot -- it will be a lot easier to do 14A because it's just using a lot of the gate all around and backside power and so forth that we implemented in 18A.

Understood. One thing I heard recently is apparently, the thought of targeting Intel's financial model to the Rule of 45. Did I hear that correctly?

Absolutely.

And if I were to look at just where consensus expectations are, right, well this and next year, it's low double-digit kind of growth. and kind of high single-digit, low double-digit kind of operating margins. My simple math suggest you're not getting to that. So what is the right time frame in which Intel can strive to those kind of aspiration?

It's a multiyear aspirational goal. I'd start with that. But there are a number of things that we should see that improve the profitability of the business. Already, I think we got OpEx to a level where we can grow revenue at a faster rate than OpEx growth. maybe we can even hold it more flattish and that should drop leverage. And of course, as you grow revenue in a business that's largely fixed on the cost of sales, we have an opportunity to improve margins over time.

And then on top of that, I would just say that as we execute better and products come out on time performance, we have an opportunity on the pricing side to drive better margins. And we've talked about this on the foundry front that we'd like to get to breakeven as we exit 2027. The only reason at this point, we see that we would not get to breakeven on the foundry side is we're just so more wildly successful than we expected on foundry, and that just drives a whole bunch more fixed start-up costs to the business that could impact it.

But I think there's a great opportunity to drive this Rule of 45, and we have the advantage of having both the manufacturing and the product margins and margin stack goes in a way that should give us a real opportunity to hit that. I would just -- Lip-Bu's been pretty focused on this measure and his expectation is those that are doing these longer-range plans have a probable path towards getting to this Rule of 45.

I just might add, I think the important part of the Rule of 45 is I think Lip-Bu's philosophy on creating value for owners is growing profitably with stress on a growth profitably. If you look at our long-term CAGR, and I'm sure you've looked at it, it's like 1990 to 2020, we were a 3% CAGR top line company.

If you look at the opportunities that we have in front of us today, not only with the resurgence of the CPU TAM but quite frankly, the opportunities we have in ASIC, the opportunities we have in advanced packaging, the opportunities we have in foundry, we are going to be a solid double-digit growth business structurally going forward as we execute. And I think that's sort of the key point thinking around the Rule of 45. It's not just about the profitability you can drop to the bottom line. But I think Lip-Bu is really intent on driving a much better growth profile for the company.

Got you. The last question, I'll actually try to jumble in 2 questions, let's see if I can. So first, Dave is that -- most successful kind of foundries have capital intensities, right? That could be in the mid-30s or even higher. So that's kind of one part of the equation because if your operating margins get into the 30s, I mean, CapEx intensity might be there. So that's one aspect of the question.

The second is that with the fab buyout, your leverage ratio has now gone back up to, I think, close to 4x. So how are you managing these different trade-offs there you would like to grow faster but that had involved much higher capital intensity and takes you away from free cash flow, right, John mentioned. So how are you kind of balancing all these factors?

Yes. I mean, it could be in those intensity levels. I actually think we should, probably longer term, operate more in the 20s in terms of CapEx intensity. That's what we talked about in the last couple of years as we were looking at it. In the near term, CapEx intensity could rise. And again, it will be a high-class problem because it will mean that we have a lot of growth opportunities, particularly on the foundry side that would drive that level of CapEx intensity.

We do have the advantage as a company of -- again, with this margin stacking, that gives us an opportunity to generate good cash flow, irrespective of the fact that we have to have a significant amount of CapEx outlays. That's different than many other companies that have to fund their CapEx through just the wafer revenue. But again, in the near term, could it be affected because we're successful, and we're winning more customers on the foundry side that require more CapEx, yes, possible.

I would say the one thing that we have coming into this what is hopefully a really good growth period is we've already made a lot of the investments on the space side. So we have multiple clean rooms that are within a year or 2 of being ready that felt like a boat anchor there a year ago, and now feels like glad we have that because now all that we're talking about is equipping it. And we've done, I think, a pretty good job of giving suppliers on the equipment side good line of sight as to what kind of range we might be looking for in terms of equipment. So we've got the right -- we're in our right place in line to make sure that we get all the equipment we need to manage upswings that we might experience on the demand side.

We'll have to close it there. I could have gone on for another 45 minutes, but thank you so much. Thanks for joining.

Thanks. Appreciate it.

Thanks. Bye.

And now, ladies and gentlemen, please give a warm welcome to the CEO of Intel, Lip-Bu Tan.

[Foreign Language]

[Foreign Language]

[indiscernible] like steps. This is the Elephant Mountain and so 1,000 steps to 184 meter. And so I survived. And I came down in one piece. So I'm here. So -- but if I walk a little bit slower than you know, I'm exhausted. But anyway, it's a beautiful view. I mean I highly recommend all of you to do that.

Yes. I think, first of all, I think delighted to be here, and this is an important event. And I'd like to get started. And about nearly 6 decades ago, a group of brilliant and highly motivated engineers and venture capitalists, including Arthur Rock, Don Valentine and many others. They found companies like Intel, Apple and others, broadly set the motion, the largest economic known to mankind, create what became known as Silicon Valley. And this is a very exciting time, and that is a very same ambition and mindset with the semiconductors make across the ocean despite the creation of Silicon Island right here in Taiwan.

And I have been very fortunate to associate with the creation of semiconductor industry in Taiwan 40 years ago. Because 40 years ago, Minister KT Li [indiscernible] invite me to lay the foundation of venture capital concept in Taiwan. It's a very new concept and people put money with you and you play the money and then you share the profit of 20%, but you don't share 20% of the losses. So this is a very unique venture capital concept. But I managed to do that.

And then with the help of Minister KT Li and then Development Fund, and I set up my venture fund. And about the same time, Morris Chang from TI came back to E3 and then set up the TSMC. So it's a very exciting time that I'm being involved in this whole Science Park and Synchro Science Park and the foundation will become the Silicon Island. And so I really see the benefit from OEM, ODM, from design to manufacturing is all here in Taiwan.

Taiwan PC ecosystem has played a critical role in Intel growth and success. In fact, last year, I was here to celebrate Intel 40th anniversary in Taiwan. I want to thank all of the suppliers, partners and customers for 40 years of partnership with Intel.

As partnership continues to grow, -- thank you. As partnership continue to grow, and stronger every year. It has been the year since I step in the role of Intel's CEO, to be more precise, 14 months to being CEO of Intel, and it may be the first CEO can speak Mandarin. And in fact, a couple of customers of Taiwan, very important partner for us said, Lip-Bu it's also very unusual if a CEO can drink liqour with us. But anyway, I'm part of this community. Execution has always at the top of my list.

To do so, we had to bring focus back to the core. At our heart, Intel is an engineering company. And that's what I decided from day 1, I came to become CEO of Intel. I have all the engineering report to me. So their understanding to really drive the engineering -- drive success in engineering performance. Our customers and partners are always already seen the shift in the Intel showed up. Now we are just getting started. And so stay tuned. We have a journey in front of us. Opportunity ahead is enormous. And our job is to stay focused, execute and deliver.

Every year, Intel shipped 100 millions of -- hundreds and millions of SoC, orchestrating silicon across every industry, working tightly with our partners' ecosystem across each layer of the stack, from silicon to SoC to system and to software. This generates trillions of dollars in value across 4 core compute ecosystem. First, personal computers. Second, edge, agentic AI and later physical AI. Third, foundational data centers. And finally, emerging intelligence centers that will power digital agents of the future. Each of these ecosystem represent generational opportunity. And increasingly, each of these will need purpose-built CPU, GPU and ASIC solutions catered for specific workloads and application.

The silicon we are building now will be for human use and the digital agent use. Let us begin with the ecosystem that started all. To talk more about the PC ecosystem, please join me to welcome to stage our new leader for Client Compute and Physical AI, Alex.

Thank you.

Thank you.

Thank you, Lip-Bu. Little story about me. The first time I came to Taiwan was the year 1990. And I was fresh out of school, first job, first international trip, first East Asia country I've ever been to, and I came right here in Taipei. And I quickly realized even back then that the desire to grow, the mindset of win-win and cooperation is all over all of our customers and our partners here in Taiwan. And so it quickly became one of my favorite countries to visit and work.

And now forward the clock, 36 years. I'm here in my first international trip with all the great people here at Intel. And where do I come? Right here to Taipei. Thank you.

I can't wait to plan and build the future with you guys. So let's get started. We have a lot to cover. Intel has continuously increased the pace of progress across all PC segments. Workstations, desktops, creators, gamers, premium and mainstream laptops. Every major segment, every major segment is driven by an Intel system solution.

With that vast coverage in mind, we're adding another dimension to scale these products even more effectively. The Intel 18A process is now at full scale. We have a full lineup of products with hundreds of design wins to prove them. At CES, we launched the Core Ultra Series 3, Intel's first product built on 18A process technology. It's setting a new standard for premium mobile performance and battery life.

The Ultra Series 3 enables great user experiences across any tasks with a very fast response CPU, a highly improved GPU, low-power processing NPU and the latest multimedia capabilities. It's a perfect blend of IP, performance and power for any AI and agentic experience. It's allowing us to lead the way to transform every PC, every PC to an agentic-capable platform.

Today, more than 300 designs are shipping across consumer and commercial segments, over 300. And to scale these capabilities even further, we've taken the latest Core Ultra IPs and specifically tailored them for the mainstream market. The result, the Intel Core Series 3 introduced in April. Let me repeat, we just introduced this in April, and it's already scaled up to 70-plus designs. That brings -- thank you. Thank you. That brings the total series lineup to nearly 400 designs in just a few short months. Now that is massive scale.

Let's look at some of the capabilities of the core Series 3, and we can start with battery life. I can go over these numbers here that are printed and I can talk about how we measure them and things like that. But I have a question for you guys. How long is your day? 10 hours, 12 hours, 14 hours or more like us at Intel. But the great user experience is, if your PC lasts longer than your Workday, and that's exactly what we're delivering in all segments. And we support ample number of ports for all of your connectivity needs. Unlike some of our competitors who only have one USBC interface, but I'll let you be the judge of that one. Yes.

The goal of this Core Series 3 is to bring premium feel and experiences to incredibly thin form factors for mainstream PCs. And you don't have to take my word for it. You can look at this wall of incredibly designed sleek PCs that are here. And all of it is thanks to you, our partners and our customers, and we couldn't have done it without you. So Please, the round of applause.

Isn't it awesome? Super light, super thin, really great.

Okay. Now the next proof is scaling 18A IP into growing markets and the fastest-growing portion of the PC market is the handheld gaming. Let's take a look.

[Presentation]

This is the ARC G3. I think a beautiful chip, more beautiful than what was presented yesterday at a keynote Okay. The G3 is derived from the Core Ultra Series 3 and the ARC G3 is a tuned high-performance GPU specifically for handheld gaming, and it's providing great battery life. The performance tested across multiple games is consistent and stable versus competition. We are more than 40% faster, 40% faster. And at the same performance, we're half the power. And on top of that, we're running all AAA games at 1080p resolution, many of them above 120 frames per second. Now that is giving gamers a great user experience.

All of these devices will be available later this month and this is just the beginning. We're going to have plenty more designs coming throughout the year. Thank you.

Okay. It is indeed true that Intel has a leading lineup of processors. And with the versatility of the 18A process technology, our newest offerings, we're bringing powerful performance and efficiency to scale across the breadth of premium, mainstream and handheld gaming segments.

These same fundamentals, the same IP, the same capabilities can deliver far beyond the PC ecosystem. The demand for our processors at the edge has been booming. As you've seen, we've already taken existing product lines and pivoted them into adjacent markets, enabling our customers to grow their businesses. Now the edge is demanding the latest products from Intel. And that's why this year, we're taking our latest Series 3 products into the edge business with over 130 designs in multiple verticals.

For large-scale edge business, our customers need the best technology and chipsets easy-to-use reference designs and appropriate software stacks. And at Intel, we've done all of that. We have over 4,000 edge ecosystem partners deploying into such verticals such as manufacturing, robotics, retail and many more.

For those of you here at COMPUTEX, you can see some of that at the Pavilion. Given that capability, given the IP, given the chipsets, given the scale that we have. There's a massive opportunity ahead of us across many segments of physical AI. It's projected to be a $25 trillion market by 2050, and it will leverage all of our scale in the PC ecosystem.

Physical AI form factors will take shape across key industries as you see behind me. We will continue growing these markets with the same strategy of leading IP and chipsets, complete reference platforms of end user hardware and applicable software stacks enabling our customers to expand into new physical AI form factors and applications. And indeed, this will be our future.

Now back to you, Lip-Bu. Thank you.

Thanks, Alex. AI is profoundly impacting the way we use our devices. A major focus area for us is the use of AI on device. Together with partners, we are at the forefront of advancing intelligence. To tell you more about it, let me welcome on stage my close friend and Founder, CEO of Perplexity, Aravind.

Well, Aravind, welcome. You and I have been talking about hybrid compute for a while. The reason why are clear, the privacy cost performance. And let's talk about how to make this work.

Yes. So in February, we launched Perplexity Computer. Computer is an AI operating system. It creates a team of agents, uses up to 20 different AI models, and it orchestrates across models, tools and files in one single system. The agent harness inside computer is model-agnostic, perfectly balancing intelligence, accuracy, privacy and cost is the orchestration problem it solves. And so this allows you to run smaller models locally on the Intel Core Ultra Series 3 GPU. And so for the first time ever, we work together to create hybrid agentic inference. And so what we are showing today is just the start. Hybrid agenic inference is how we maximize token value per watt per user.

So should we show them how it works?

Yes. All right. So here it is. Let's say I'm an associate at a private equity firm, and I'm working on something that has a confidential project codename, Project Falcon. Here is the query. So think of it as me trying to understand if a certain private company is worth $1.1 billion, and I'm feeding it confidential deal materials. The work begins on the laptop.

It sees that Project Falcon has private deal room files and an NDA, a local leverage buyout financial model, a wide board diagram, and bilingual transcripts that are very confidential, you don't want these materials to be shipped to the server. So what the local model does on the Core Ultra Series 3 is it first decides is all very important work and shouldn't be sent to the server. It reads the files, classifies what is sensitive and what is not. And then computer decides what should leave the device and what shouldn't. And each of these things is done with local AI.

The orchestrator can spin up additional agents as necessary. And so if you need a research agent to bring in outside file materials against the local model without exposing any private files, that's what you want in the hybrid system. And so computer ARC acts as one single system, brings all inputs and outputs together. And so let's actually skip and see what the actual results would be.

All right. So the result is a document, a research report and supporting data, and it's been created by agents on large cloud-based models, keeping your sensitive information only on your device. And so all your local device models will take care of the private side and the server side models will take out of other things. True hybrid inference orchestration.

This is the architecture we both believe in. And the future is more compute in the data center and more compute on the local machine.

And so I think of this as a big milestone for engineering on both the agent harness AI side as well as the chip side. And so it's been really fun to partner with you and Intel on this. So thank you so much, Lip-Bu.

Definitely. Thank you, Aravind, and looking forward to continue our partnership.

We talk a lot about the exciting new developments in the PC, edge and physical AI space. I want to take a few minutes to talk about the foundational IP that power all these advancements.

Let us talk about x86. When most people think of general purpose computing, they think x86. And that is a good reason for that. x86 is architecture that has powered data center for nearly 5 decades. And the leadership continues. According to the IDC expect 8 out of the 10 servers installed through 2030 to be x86 base, powering modern computing from foundational to emerging intelligent use cases.

Intel pioneered most of the breakthrough architecture innovation, that have enhanced x86 over the last 4 decades, starting with 8086 that become the foundation of modern computing. If you can see the chart. Today, we have 2 flagship CPO costs, P-cores and E-cores. One optimized for performance, one. The other is for efficiency. These are the Intel's most advanced CPU cores with the accelerator building -- built in specifically for foundational workloads like security. Our x86 cores power our PC, client, edge portfolio, and also power our data center and AI portfolio. Under my leadership, we are committed to building the best CPU cores in the world. And we will enhance -- ensure that the most compute-intensive workload run best on x86.

Next, let us talk -- thank you. Now let us talk about how x86 is enabling foundational data centers. To tell you more about it, let me invite on to the stage, Kevork.

Thank you, Lip-Bu. Wow, it's so great to be here. Specifically in this point in our history, global history, collective history; and b, at COMPUTEX with a blue badge. So I'm very happy, I'm very humbled to be here to share with you some of the innovations that we have. So let's talk a bit about -- see what this AI thing is about.

So when we say foundational, we mean the workloads that keep the world running. So currently, we have data centers, and there's a number of items and workloads and entities that run on these data centers. So for example, we have 5G networks that keep us connected. We have databases that keep our data safe. We have cloud services that power our daily lives. So we expect and demand for these workloads to grow in size and capacity between now and 2030 from 80 gigawatts to about 100 gigawatts.

And most of you involved in this domain understand the extent of this type of an expansion. These workloads are broad. They are mission-critical. So attention -- special attention has to be taken when running them. But also, they require performance, efficiency security and resiliency. And we can't emphasize enough all these 4 factors. That is why we are excited to have Intel Xeon 6+ introduced at COMPUTEX this week. It has 288 E-cores, a massive 576 megabytes of L3 cache built with our Intel 18A technology, and we can't emphasize enough the value of Intel technology that brings to data center products.

But most importantly, it delivers efficiency and density, which enables our partners to save very precious real estate, have more compact servers and the racks. So this is a leadership compute for the next era of cloud and network infrastructure. So Xeon 6+ launches with the strength of our ecosystem that's been built over decades and decades of data center development, both from a hardware but also from a software and infrastructure perspective. Moreover, our ODM partners are bringing Xeon 6+ solutions to the market today. So these range from full rack scale deployments to server-level designs.

Xeon 6+ joins our lineup of data center processors next to our already launched Xeon 6 based on P-cores. Both of these category and class of solutions delivers new performance and choice for all the enterprises whose infrastructure backbone is built on x86 and Xeon. This is critical for enterprises that need to increasingly balance preparing for AI workloads, but at the same time, running their day-to-day mission-critical applications.

So let's switch gears and talk about how Intel was certifying the deployment of intelligence at scale. It's undeniable that enterprise infrastructure today will have to evolve to keep up with the AI demand. Recent research forecasts that AI inference workloads are expected to become 40% of all data center power demand and much more than they are today. So we have these 2 paradigms where we have the foundational data centers keep on running their traditional workloads. But at the same time, they have to figure out ways of building their infrastructures to serve intelligence at scale. And this is where Intel and Xeon 6s come in.

Now -- up to now, training split the data center into 2. So on one hand, we have CPU-led enterprise infrastructure. And the other hand, we have GPU-heavy AI factories. And that was very clear divide for a while, right? And we've all been accustomed to that reality. But as agentic AI moves into real workflows, data tools, governance, the needs change. The next wave is not just about training models. It is about putting AI to work.

So let's look at why Agentic AI changes the infrastructure equation. The way AI inference works is straightforward. We take a prompt, it gets fed into an LLM where it spends the most time reasoning about the prompt. I mean we've all seen this. We've done this thousands of times and outcomes and answer. In this case, a lot of time, we spent computing the large language model, which is mostly GPU and compute-intensive.

Now the way Agentic AI works is radically different. It's given goals rather than prompts. So we've all seen the different types of loop that people are running on this Agentic AI. It's also very iterative in nature, but also prompted by automation. And thinking, planning, acting and reflecting our natural way of these agents interacting with us. As it works, it uses tools, reads and writes files, checks rules and other aspects that were in the traditional real of CPUs and x86. And then for each step, the type of underlying compute needs is very different, and we'll show that in a bit. This is particularly important as agents scale up their work, spawning new agents that work concurrently.

And the category and the complexity of agents are going to be very different depending on the complexity of the work. That's the main reason that there's such a rapid increase in CPU demand. For agenetic AI, the CPU orchestrates the show. Now what we're seeing is we're also seeing the balance and the ratio of 1 CPU to 8 GPU and more is coming much closer to parity. So let's take a look at the real example. John?

Thanks, Kevork. You talked about how Agentic AI is changing the compute requirements. Let's take a look at a real example. I have a traditional AI inference set up on the left-hand side of the screen. Let's send a request. Write a Python function that calls an open AI compatible chat completions API. The model gets the response, generates code and sends the request back. Take a look at the slider on the top of the screen, GPU dominates nearly 7:1 GPU heavy.

In contrast, let's take a look at an genic AI system. Across the top, look at the pipeline stages, Green is GPU work. Blue is CPU work. Linking is happening on our Xeon 6+ processor with the efficiency course. Webfetch and compile is happening on our Xeon 6 performance cores. And unit testing is coming back and running on our Xeon 6+ efficiency cores, the right class CPU for each stage of the pipeline. Take a look at the slide or across the top again. we're near parity, but CPU heavy this time. What's this look like when we multiply that by millions of queries a day. As you mentioned, each Xeon 6+ processor has up to 288 cores, that's 576 cores per 2-socket server. When we look at that from a rack scale perspective, that gives us over 36,000 cores per 32 use of compute space.

Thank you, John. Well, this is pretty amazing and some data to ponder on. By far, the density of CPU we showed has the highest density per rack ever but also looking at the number of agents, these are the new metrics that are emerging. We can safely say that, that particular rack can run up to 150,000 agents. So good news to all the CIOs in the audience.

Now we are very expensive GPUs can be -- can see more utilization because of our solutions. Now both Xeon 6 with P-cores and E-cores are built on for intelligence at scale. There are different cores, of course, but we've seen the workloads that require very high-performance cores, pushing the frequencies, but also there's a need for a very high-density power-efficient cores. So we've seen all the workloads, we run all the analysis, and we are delivering these solutions to all of you. Now having said that, we are working with our customers and partners to make sure that each solution is tailored to your needs. So I'd like to welcome Lip-Bu back on stage to talk about the server and rack sale solutions that our partners are working on. Thank you.

Thank you, Kevork. It is great to see the momentum in the data center. As we look forward, to see that is for intelligence at scale. Discrete compute alone is not enough. Our customers are asking us to think of system level to help them serve real agentic workloads at scale. It pushed us to rethink how we deliver our compute beyond the socket and to the rack. That is why we start the initiative called Rackscale Blueprints. Working with ecosystem partners to develop Rackscale Blueprints built on open standards. So customers can rapidly scale their intelligent infrastructure with confidence without proprietary workarounds.

Behind me, as you can see, 2 examples of these blueprints. One is for agentic performance based on Intel Xeon 6 with P-cores. The other is agent density with the Intel Xeon 6+ with e-cores. We are working closely with our partners ecosystem, including Foxconn, SambaNova to expand our Rackscale offering.

Let me call on stage one of our partners, Chief Product Officer of Foxconn, Jerry Hsiao to talk about how we partner on Rackscale solution.

So excited to be here today, wonderful product, an amazing event.

Jerry, Intel and Foxconn have been working together to many decades. And Foxconn has been instrumental in driving technology innovation in Taiwan and around the world.

Yes, that's right, Lip-Bu. I'm proud of the work we have done together from AI servers to data centers and to age computing altogether.

And today, we're excited to announce the next step in our partnership. Intel and Foxconn are working together to develop rack scale products built upon Intel Xeon processes. Together, we will focus on exploring the development, integrations and commercialization of differentiated rack scale AI infrastructure solution, leveraging complementary architecture to address diverse AI workload requirements.

Yes, together, we will continue to deepen and expand our partnership unlocking new opportunities ahead. Through this collaboration, we will deliver system-level AI solution to our joint customers enabling more integrated and scalable computing environment. This makes -- it marks an important step ahead, and we look forward on valuing more in the near future. Thank you, Lip-Bu.

Today is an exciting milestone for our continued partnership with Foxconn. Jerry, thank you for joining us.

Fantastic. Thank you for having me.

Thank you to the many partners in the audience today that is helping to bring this rack scale vision to life, providing choice through the ecosystem power, we do not believe one size fits all approach for intelligent centers. Each enterprise will run unique workloads. So their infrastructure needs will also need to be unique and purpose-built, as you can see from the screen here. Just look at the server in front of me. This is a whole series of partnership we have.

[Presentation]

Intel is working with a lot of partners to provide service, rack scale solution, designed to fit your existing infrastructure ready for AI at scale, as you can tell in front of you. We see token usage exploding. Agents now consume 1,000 time more tokens than single event reasoning. In addition to building the best CPUs, it is critical that we deliver compute solution optimized for token consumption and token generation.

The bottom line, AI at scale will require heterogenous computing. To this end, Intel recently announced a partnership with SambaNova. To talk more about this, let me call to the stage founder CEO of SambaNova, Rodrigo Liang.

Rodrigo, welcome to joining me today.

Thank you, Lip-Bu.

And over the next few months, we have announced a few updates on our joint development partnership. Can we talk a little bit more about the work that Intel and SambaNova are doing together?

Absolutely. We've been busy. Earlier this year, we announced a multiyear collaboration to deliver high performance, cost-efficient inference solutions based on Xeon infrastructure. We've been building something really special and excited to show you today. This is the SN50 SambaRack, we announced earlier this year. Rack scale AI infrastructure built for agentic workloads. It uses Intel Xeon 6 processors with SambaNova SN50 RDUs and shipping to customers later this year.

Today, we're also excited to demonstrate the world's first heterogeneous disaggregated inference using SambaNova's RDU with Intel's CPU and NVIDIA GPUs. What you're about to see is the same prompt, the same model running side-by-side, 2 different stacks. So the one on the left is GPUs, RDUs and CPUs disaggregated inference. And this one on my right is GPUs on their own. They both get fed the same prompt in the same model, just different stacks. The disaggregated inference stack is taking off, and what's happening here is you have the Xeon 6 processors doing all the tooling execution. You have SambaNova of RDUs doing the decode and generating all of the tokens.

And then you've got the GPUs performing the prompt caching and the faster prefill reducing overall time. When all 3 chips are working together you dramatically reduce the end-to-end latency and the agents for the fastest need for agentic AI. On the other side, the GPU stack is still working away. So the initial results of our work is disaggregated inference is the GPUs, the RDUs, the CPUs, that's the fastest. And artificial analysis in our test found to be 2x to 3x faster than just the GPUs alone. And this gives us an early look at how fast this can be.

Rodrigo, the most exciting part about all this is that we have tremendous customer interest in these solutions.

Absolutely. So let's see who comes next. So Lip-Bu, turning it back over to you. Thank you so much.

Thank you. To continue this conversation, I'm delighted to invite my good friends Robert Smith is Vista Equity Partners, Chairman, CEO and the partners and Robert Smith, to tell you more about how they plan to use these racks from Intel and SambaNova, Robert?

Great. Thank you. Lip-Bu good to see you.

Same here. Thank you so much for joining me. Yes. So Robert, AI is driving huge demand for computing, and it's reshaping the silicon system software and at all at once. What are you seeing and hearing from the enterprises that you work with?

First of all, I'm excited to be here at COMPUTEX to join you with this wonderful event. For us, it's been quite incredible. There's been a huge focus right now to bring AI to enterprises around the globe. We want to make it usable. We want to make it impactful for the organizations that we work with. We have over 90 portfolio companies and well over half of them have now have converted to agentic solutions. And with over 750 million users of our software, that really translates to over 10 billion agents. That's why we've launched Vector Core computer, VC2, with our partners at Cambium Capital to offer the world's first commercially available architecture for disaggregated inference. This novel agentic neocloud is built to deliver the fastest enterprise inference throughput of any architecture to date. The demo you just witnessed with Rodrigo was conducted live in our Los Angeles data center, and we have over 50 deployments planned in the U.S., which were targeted to convert existing data centers to inference data centers.

This is very exciting. And as we saw from Rodrigo a few minutes ago, we are already starting to see strong momentum for these offerings. Can you talk a little bit more about how Intel Vector Core compute and our partners like SambaNova are bringing this aggregated inference to life?

Of course. I'm excited to share that first Together AI is the first commercial customer and is excited to use this architecture as a service to accelerate inference workloads. We expect many of our enterprise software companies and their customers to quickly follow as the demand for inference keep growing, and this is -- has to be -- and it is more efficient than anything they previously have had access to.

Most critically, VC2 is built and utilizes the SambaNova stack, which is an air-cooled data center. We believe it will deliver what enterprise customers and communities are asking for. which is reliable, low latency, low-cost inference at scale. Partnering to advance AI is one of the best ways to develop this transformational technology. making it usable and economically viable for enterprises worldwide.

We are excited about that. Thank you for joining me today and delighted to have you here.

Always a pleasure, Lip-Bu. Thank you. Congratulations, 14 more months. We're excited to see what you're doing.

Thank you. As you just saw from Rodrigo and Robert, picking the right silicon architecture for your needs is critical. For enterprises today, there's a broad range of architectures to choose from as large workloads increasingly become strategic asset for companies. They are increasingly looking for silicon built around their exact needs.

Next, I would like to invite Srini, a semiconductor design veteran and a leader of our purpose built silicon team to talk more about the work we are doing in this area. Srini?

Thank you so much. A very good afternoon to you guys. Purpose-built silicon it is. This has been a journey that the industry has been using almost for the last decade or so, and especially the hyperscalers have tapped into this to its full potential and shown us the benefits in every way possible.

Lip-Bu, you challenged just to see, in this space, given the fantastic assets and the breadth of assets that we have at Intel, how could we be relevant to this, not just be focused on the stuff that we do internally, how do we bring this out to the external world and do something more. With that said, we had a proposition. We've been working on it. And today, I'm very happy to share a couple of good outcomes that we have.

The first, on the hyperscaler side, we have Google and Intel. Google and Intel have gone into a partnership wherein Intel is delivering what is called as the infrastructure processing unit, I would call it Intel processing unit, actually. But infrastructure processing unit, which is a piece of silicon, very vital for hyperscalers performance and the journey continues. And by the way, this is a deployment today. So it is not just something that we are doing, but it's already designed and being deployed.

While this is working on, we've been -- Intel as a company, has been pretty active in the telco market. And in this telco market, another marquee customer, Ericsson, has been partnering with us and Ericsson chooses us, wherein we deliver or Intel delivers the next-generation infrastructure silicon for -- at a global scale for them across the board. This just gives you a very sneak preview at the highest level to see the kind of work that's happening in the purpose-built silicon space, which is a very exciting space and more importantly, a high-growth space.

And I was just thinking what better place than COMPUTEX, and Taipei where custom silicon really is the name of the game here to announce that Intel has officially entered this market. So looking forward to working with many of you guys and see how we can be relevant to you some of your aspirational goals on silicon, okay? Thank you, Lip-Bu.

I'm super excited about all these partnerships that you announced and more to come.

Yes, absolutely more to come. Thank you so much.

Thank you. The work Srini and the team are doing with purpose-built silicon is really important. I'm super excited to be partnering to build custom silicon with many leading edge companies as well as some of the most dynamic startups across the industry vertical.

I would like to highlight some of this partnership today. One of the most exciting areas where we can deploy advanced silicon is biomedical engineering. For years, emulating the functionality of the human brain has been the holy grail of computing. One company that is in the forefront of brain inspire computing is Echo Neurotechnologies. Let us hear more from Eddie Chen, Founder, CEO of Echo Neurotechnology and also one of the world's best neurosurgeons.

I'm Eddie Chang. I'm a neurosurgeon at UCSF and Co-Founder of Echo Neurotechnologies. For decades, AI has been brain-inspired, meaning borrowing ideas from neuroscience at a distance. Neuromorphic computing has carried that vision the furthest. It built silicon around the brain's core principles like spikes sparse, communication, memory and compute all in the same place. That architecture is right. But what's been missing is direct evidence of how the brain actually performs the computation. That's now within our reach.

For the first time, we can study how the human cortex computes language in real time at the resolution where computation actually happens. This opens a whole new possibility. Algorithms that are not just brain-inspired but new ones that are trained on the brain activity itself, measured against the brain itself. That's the shift in our collaboration with Intel. Together, we're developing brain trained algorithms for streaming speech that approach the efficiency of biological computation. The payoff runs both ways. AI that's faster, lighter, and closer to how we actually think and new tools to restore speech to people who have lost it.

Together with Intel, we're building AI that learns from the most powerful computer ever discovered, the human brain. We're excited about what's ahead. Thank you.

Thank you, Eddie. I'm amazed by the work you are doing. I'm confident that our work together will help lay the foundation of highly efficient AI computers in the future. Another company doing work at the cutting edge of biology is Greenstone Technologies. We are partnering with Greenstone to establish scalable reference architectures applicable across the broader life science imaging ecosystem. Dr. Joseph Wu is the Head of Cardiology at Stanford and the Founder, CEO of Greenstone. Let's hear from him.

Hello. My name is Joseph Wu, and I'm a Professor of Medicine and Director of the Stanford Cardiovascular Institute as well as the Co-Founder, Greenstone Biosciences. Thank you so much for including me in COMPUTEX.

Intel and Greenstone are working together to speed up the development of new medicines. Our partnership combines state-of-the-art human genetics and biology from Greenstone with advanced AI computing for Intel so that we can scale data processing, storage and analysis. Greenstone has built the world's largest biobank of human-induced pluripotent stem cells.

From just 10 ccs of your blood, we can make your brain, heart, liver, kidney, gut and any type of organoids in your body that are genetically identical to the patient. This would then allow us to test existing and new medications more quickly and at a lower cost. I believe the companies in the human biology and AI computing will help shape the future of biomedicines in the next decade. And this is why we're so excited about the partnership between Intel and Greenstone Biosciences. Thank you very much and enjoy the event.

[Foreign Language]

Thank you, Joe. I'm amazed by the work you are doing and I'm excited about the potential of our partnership. Another key partner is Hitachi. They have a wide range of capabilities that help accelerating our work, our plan around foundry tools and quantum computing system.

Let us hear from Hitachi CEO, Tokunaga-san.

Hello, COMPUTEX. I'm Toshiaki Tokunaga, CEO of Hitachi. For decades, Hitachi and Intel have worked together to solve key challenges for society. And today, we are bringing our strength even closer by combining Intel's advanced computing with Hitachi's industrial strength in the physical world, we will create intelligent solutions that will benefit both businesses and society. Thank you, Lip-Bu. I look forward to our future together.

Thank you, TT. We are really looking forward to working with you. Finally, if you look at we have the brains inspired computing, biomedical medicine and then energy. The last one is industrial automation. Finally, one partner I would like to hear -- like you to hear from is known for their pioneering work in industrial automation. Let us hear from my very good friend, Roland Busch at Siemens.

As a customer of Intel, we all know that global semiconductor demand has hit a high record. In 2023, Siemens and Intel already joined forces to meet it. And now we are taking our collaboration to the next level. We are expanding our partnership across the entire value chain, from design to manufacturing, to chip applications in Siemens products. We improved design quality through EDA, automation and software solutions built with agent AI. We partner on all areas of the manufacturing process, including product life cycle management, automation, electrification, quality and sustainability. And what makes this even more relevant for us, the chips created in this value chain will be used in our own Siemens products. Looking forward to what's coming up.

Thank you, Roland. We are delighted to expand our long partnership with the Siemens Group. I'm looking forward to disclose more about this partnership in the coming months. And we are working with several other partners to keep pushing the boundary of what is possible. I would like to close by returning to where we start our conversation. The opportunity for Intel and for our partners is immerse. PC, edge, agent, physical AI, data center and emerging intelligence center from silicon to SoC to system and applications. This opportunity is only made possible by all of you, look at the list and the largest ecosystem of partners, suppliers and customers.

Intel is an iconic company. We laid the foundation of modern day computing, and we are proud of our heritage. But we do not want to rest on our honors and glories. A year ago, I joined as a CEO. I challenged my team to work with me to build a new Intel. That is exactly what we are doing. We are not incumbent by the past while building something wonderful. It has been a year of transformation for Intel. We ramp our 18A to high volume with multiple products.

We are executing well on our advanced packaging milestones. We made tremendous progress on engaging customers and building our foundry business. We introduced new SoCs for all major compute platforms from premium mobile to high-density cloud and 5G. We are rebuilding and strengthening partnerships across the ecosystem. We are double down on creating new business opportunity across existing and emerging domains. We are working at the forefront to reimagine computing and make it highly efficient for the AI era.

And this is just the beginning. I'm super excited to continue executing at hyperspeed.

[Presentation]

Ladies and gentlemen, this concludes the Intel COMPUTEX 2026 keynote. Thank you for joining us this afternoon. To witness the future of technology, we look forward to seeing you again at future Intel events. And please don't forget your personal belongings.

All right. Good afternoon, and thank you for attending the second day of JPMorgan's 54th Annual Technology, Media Communications Conference. My name is Harlan Sur. I'm the U.S. semiconductor, semiconductor capital equipment analyst for the firm. It is my pleasure to welcome Lip-Bu Tan, Chief Executive Officer of Intel.

Lip-Bu, one of the semiconductor industry's most accomplished leaders, right, bringing decades of experience, spanning venture capital, EDA and advanced chip design as Founder of Walden International and former CEO of Cadence Design Systems, he built a track record of disciplined execution, customer-focused transformation.

Lip-Bu received the Robert N. Noyce Award in 2022, which is the semiconductor industry's highest honor, right? Since taking the helm at Intel in March of 2025, Lip-Bu has fundamentally reshaped the company, flattening its management structure, strengthening its balance sheet and retaining and attracting great talent.

So Lip-Bu, thank you very much for joining us today.

Thank you for inviting me.

Yes, absolutely. I'm going to read the safe harbor statement very quickly. Before we begin, please note that today's discussion may contain forward-looking statements that are subject to various risks and uncertainties and may reference non-GAAP financial measures. Please refer to Intel's most recent earnings release and annual report on Form 10-K and other filings with the SEC for more information on the risk factors that could cause actual results to differ materially and additional information on Intel's non-GAAP financial measures, including reconciliations where appropriate to the corresponding GAAP and financial measures.

So with that, Lip-Bu, again, thank you for joining us today. It's been around 14 months since you've taken over as CEO. The team has certainly executed extremely well under your short tenure. I remember the first earnings call in front of investors in April of last year, right? And this is my guidepost to monitoring the execution of the team. But you came out last April and you summarized your goals very clearly.

Number one is best products always win. Intel needs to focus on designing best-in-class products, redefine the portfolio for emerging AI workloads. Number two, approach the portfolio with full stack systems focus. Number three, build a world-class foundry franchise, and that means building trust with customers; and four, strengthening the balance sheet, right? So 14 months later, give us your view on the team's progress on these goals and more importantly, your confidence on executing to your strategy mid-to-longer term?

Yes. Thank you for the questions. So let me address the 4 areas that you asked for. One is the balance sheet. When I took over, the balance sheet is pretty horrible. And so I tried to recruit some talent. They said they're almost a bankrupt company. Why should I join you? I love you, but not this one. So I really focus on the balance sheet to strengthen that.

So first of all, I'm delighted in President Trump and Howard Lutnick from Commerce Secretary and let me have a chance to explain my vision. They're delighted. They converted some of the CHIPS program money and then to convert into equity. And so we're delighted to have their support. And also my long-time friend, Jensen from NVIDIA besides we're going a deep collaboration, multi-products, multi-years of engagement and a very good engagement. And he said, Lip-Bu, I also want to support you with $5 billion investment.

And then also my old-time friend, SoftBank, Masa, used to be on this Board and he said, Lip-Bu I also like to help you. So I think 3 of them helping me. And so far, knock on wood, I made money for them, and they're quite happy with it. So I think overall, I think strengthens our balance sheet. I think that's very important. And because of that, also, we drive more successful cash flow and operations. We're able to divest this -- bought back this stake to Apollo and so that we can really have less dilution on the EPS. So that's very helpful.

So second, from the back, you mentioned about the foundry and I want to be a trusted foundry partner. So this is very important. At the end of the day, foundry is a service business. We really need to build the trust with the customer. Customers have to trust you enough based on their revenue, they provide you the -- we have to produce a wafer of good quality. Otherwise, they're going to attack their impact to their revenue.

So I think that's something is a lot of responsibility for doing that. And then for doing the foundry business, first of all, not just to improve your yield, improve your defect density, the cycle time, make sure that you can on time to deliver the customer service.

The other part, very important, you need to have all the right IPs. So for example, if you have a foundry customer for mobile, you need to have low-power IP. If you're working with the big data center or cloud infrastructure, you need to have high-speed connectivity, 3D IP to serve them. So those IP in some way, you are working to 2 team of the customer. One team is a supply chain they can use your foundry. The other team, basically the design team. They have to make sure that your PDK, your IP is ready. They can really design the chip based on your foundry nodes and foundry capability.

So there's a lot of faith to do that. So I think overall, we're making great progress. We can talk more about it. We make great progress on 18A and then our Panther Lake depends on that. Now I have a volume in the wafer to support my Panther Lake. We announced in CES and we have 200 design wins.

And then we're finally able to get the product volume to support them. The customer is quite happy with that. And then the other part is clearly the 14A. And we announced in Q1, we have 0.5 PDK so that they can do the test chip to look at our yield and see whether they can, over time, to really design their product and fabricate with us.

So we are making good progress. The holy grail is 0.9 PDK. Right now, we're looking at October to the outside customer. Internal customer will be earlier so that we make sure that we really kind of clean the pipe, make sure that we're doing right, make sure that we can sell with good quality.

So good news. I think we are knock on wood. We have multiple customers engaged with us and to really define what product what foundry location want to be, what kind of capacity we need. So I think that's something my philosophy, I don't disclose the customer. If customers want to disclose, we will support that. But I think something that's important to work with them. And then some of them worry about our competitors know that we are working with us. So we try to keep it highly confidential while we kind of gradually win their heart. So that's on the foundry side.

In terms of the products, clearly, I mentioned earlier, the long term, you basically have the best product.

That's right.

And so immediately from day 1, I have all the engineering report to me. And it's very important. So they are understanding where we went wrong, what are the mistakes we make and how we can correct that. So from day 1, March 18, 2025, my first day to work, and I have my big customer had dinner with me, and they told me that the product you was far behind, 25% behind and your product is so complicated and you are behind the schedule to deliver.

And so those are the things I took in, and I addressed that. And now our weekly call, and we are talking about big areas, big ideas to work together. So I think we kind of win the trust of them that we can come up with a better product by PC client with Panther Lake is a good one.

In terms of the data center, we used to have the leader. We have 90-plus percent market share, and we lost it. And so a lot of mistakes were made. And then I still remember one of the big customers told me, Lip-Bu, these are the 14 area that you guys screwed up and I take note of every single one that Lip-Bu that we are so delighted you are the CEO. You are so humble. In the past, we are so proud Intel that you don't even listen to us. Actually, you lectured us, and we quietly design you out. But now you're back, you become the CEO. We love to have you.

And then one thing about you, you are a very good listener, you take notes. And then less than 1 year, they decided, Lip-Bu, we're going to go big time with you, and we're going to announce a couple of areas that we are working together. So stay tuned. I think that's something that across the board, a lot of customers said Lip-Bu, we are so happy you are the CEO. We have worked with you at Cadence Design. You underpromise, over delivering, you work hard, you listen to us, you put together a world-class team to do that.

So I think all in all, I think we are making good progress to come out. And of course, as you know, some of the products, if you design, you take about 12, 15 months to come out.

That's right.

So right now, we're working on it. Hopefully, some of the new product, KLA product can come out, can really meet the schedule and meet the timetable and the performance. And one thing about timetable, I have a culture right now, I just implemented. I have to be A0 to production. Now A0 is when you tape out first-time parts. And Intel don't have that culture. So I kind of said well, first-time part, A0, B0, you keep your job. Anything above that, you're fired.

So that culture people initially thought that I'm just joking. And now I'm starting to implement, they're starting to say, okay, Lip-Bu, they are very serious. You really look into the -- all the design, all the bugs that we've tried to fix and then all the IP that we use to make sure that we certify and make sure we do that before we go to tape out. And so those are kind of the culture we need to have.

And then finally, I think most important is to delight the customer. And that's why I spend a lot of time, listen to customer and the customer, customer, understand what they want and make sure that we meet their requirement.

No, that's perfect. And as we've talked to some of your customers and as we follow the company over the past 14 months, I would also agree that under your leadership, the team has certainly improved its execution very meaningfully and articulated its strategy very well.

So now it's just a matter of continuing to solidly execute on a go-forward basis. And as you put in place these targets and goals, I think one of the most impactful and important decisions when you first joined was to flatten the organization to accelerate decision-making, right? And so can you discuss the response from employees, particularly engineers? Now that you've been operating under this more streamlined model for a few quarters, can you just discuss some of the early benefits that you see from this change?

Yes, it's a big change in the 14 months. And so first of all, I think it's very important. Over the years, I think they built so many layer of management. I still remember some areas like 12 layer. And I said, well, this is not good. We need to bring it down to 5 layer. And then the main reason is the more accountability.

And then secondly, the speed of making decision rather than so many business review. And then the most important is the middle management, they love meeting. And so I basically cut the crap of the meeting. And then if you have a meeting, you got to be what is the agenda, what are the things that we want to decide and then what are the things that we want to bring me up to speed, if not, they canceled meeting.

Yes.

And so I think that speed of decision-making instead of sometimes it can be one way to decide. Sometimes we did in Intel is take more than 1 year. And at the end, the people decide they don't even know what is the problem. And so I think this is something that we just have to eliminate. So right now, I kind of work as a kind of start-up environment. Everybody moved fast and then some of the customers really like that. It's a little, I call it the speed of light. So we have to make decision quick. If it's a mistake, we correct right away. And we don't to delight the customer.

In the past, customers complained to us, we are not responding, we are not listening and even lectured them.

Yes.

Now in 24 hours, I want to know what the bad news, what are the customer. If the customer tell me something that you didn't tell me, you're in deep trouble. And so that kind of transparency hold the people accountable and sometimes things get screwed up and I say, who responsible for it? Everybody pointing fingers. Nobody wants to take responsibility. So now I said, well, I pick all the right people, you are responsible for your job and you give me daily or weekly update. And then you make a problem, you let me know and then share the problem, bad news first. Good news, you don't have to tell me later on, we can celebrate. But bad news, tell me first. If you have tell me the bad news, it's our problem. We fix it together.

So that's kind of the culture we changed. And then so far, the feedback from customers is really happy to hear that. And then secondly, I think for the engineering leaders used to be kind of third class citizen. Nobody listened to them. Now they have direct line with me. I go down to 8, 9 layers down to find the real truth of the information so that we can really make some decisions good for the company. And I said the example is basically do what is best for the company and take away your personal interest and agenda, and let's look at what is best for the company.

And in fact, one thing is very similar to Cadence when I took over, same thing I didn't tell. When I look at the performance review, everybody outstanding outperform. And I said what is this? So I changed it unless the company performed, nobody get outstanding performance, unless it's super, super outstanding and include myself. Even I tell my Board, I'm just acceptable. And that set the tone for the culture of the company. Let's work together. I used to be a basketball player. The only way we win the business is the team work together, not the individual's goal.

Intel brings to market some of the most complex, what we call system on a chips or complex compute chips to the market, right? And I believe historically, Intel operated under a model that perhaps excessively relied on the use of internally developed intellectual property or IP, right? Given your background and tenure as CEO at Cadence, for example, how are you looking to shape how the engineering teams potentially integrate third-party IP, third-party design tools to expedite, improve productivity of these most complex designs that you're bringing to the market, as you mentioned, minimizing tape-outs while still champion internally developed IP for the superior competitive advantage that you're targeting, right?

Very good question. I think Intel has a culture, a lot of talent in Intel. And the problem is each one have a silo, they work on something. And some of them is not differentiating. You use the industry standard accepted practice on IP or EDA tool. And you were shocked when I came on board, we have a big team on EDA. So I say, why are you doing that? Unless you have something differentiation, you should just embrace the industry standard.

That's right.

So we -- I have decided to hire Srini from Cadence. I promised Cadence, I said that no more hiring. This guy I really need to have. And he's not an EDA guy. He's really the designer. So I have him run the central engineering. You will be shocked. I have 5 different organizations have different EDA effort and budget. And number is so huge. And then I said to wait a minute, why are we doing that? So I said, well, Srini, you come over to the central engineering is ASIC central design team. And then all the people we have cut by at least 2/3 of it and of the people gone and then consolidate the design, the budget to focus on EDA, what we really need.

And then internal, we still lose some, but something that is very different the industry don't have. And then so unless that otherwise just use industry standard. Same thing for foundry. We have to really embrace the standard. And that's why in the Intel Foundry days, I highlight I have all 4 EDA leader CEO on stage with me. We are embracing them. They are important partner, just like Jensen embrace them. And that's something that we do the same.

Got it. And going back to the partnerships that you talked about in terms of your discussion on shoring up the balance sheet, right? You've signed important partnerships with industry leaders and industry innovators, NVIDIA, SoftBank, Google, SambaNova, Ericsson, Obviously, the U.S. government as it relates to U.S. supply chain assurance and resiliency. Terafab is another good example of that. And you've also strengthened your partnership, as you mentioned, with the value chain companies like Cadence and Synopsys, right? And it's a great endorsement and reflection of the importance of Intel in the technology value chain, right?

And so you highlighted the financial benefits from some of these partners. But how has Intel been able to sort of leverage these partnerships in driving towards your product excellence, foundry leadership aspirations? How have these partnerships helped along those lines?

Yes. Good question. I think one of them is basically I'm a long-term guy. I like to build the company 5 years, 10 years from now. And then you have to look at from long term, where you want to go and then take steps to get there. And for example, full stack and then rather than just work on the silicon network, you have to build the software, you have to build the memory optimization, you need to really drive the platform so that you can provide the chassis to the customer. And so I think all this -- along the way, you can't do it alone.

So you need partners to help and also need partners to support you in the ecosystem support. So those is very important to have, but you have to be very selective, which are the right partners like Jensen is a good example, old friend of mine. He's done such a great job in building up the platform. I still remember many years ago, he asked me to support CUDA and then he's no more semiconductor. He's a platform company, giving credit to deliver.

And so I think in some way, I'm so happy from day 1, asking him what can I help you? And he turned around and said that you're not helping me. I'm going to help you as a friend. So I'm delighted to have that collaboration with him. And then likewise, for example, Google, we have a big announcement. They are the important player and then the Gemini 3 and 4 are really stand out, and then we want to embrace. And also some of the silicon, they are working with us, and then we are deep partnership with them.

So something that we like to have the kind of win-win for both. They're successful, and we will be successful together. And so we're going to see more and more of this partnership and I strongly believe in partnership. And one thing good I bring in to the table to help Intel, one is my network of contacts. And secondly, I'm very strong with the partnership with the customer. They built trusting relationship with me for a long time. So I think with that on the foundry customer side, from the product side and even some of the customer customers tell me what the product should be, then I help my customer to really deliver that together with them.

And so I think all in all, I think it's create a win-win partnership. I think that's the best thing so that we can create value and rather than compete with each other.

Absolutely. Let's focus over to the topic of AI and accelerated compute. And again, I'm going to rewind back to the April 2025 earnings call where you first -- when you first joined. And I believe as a part of your opening commentary in terms of defining your goals, you also mentioned the opportunity of Agentic AI. It's coming, and it's going to create a lot of opportunity for Intel and for the semiconductor industry.

And interestingly enough, by the second half of '25, we saw that inferencing inflection, right, moving past training workloads. And because obviously, your customers' customers move from training to monetization, right, one shot to Agentic-based workloads. So we saw that complexity increase, we saw the inflection of inferencing over training exactly as you predicted back in April. The market is now seeing the reemergence of CPU, right, as the key contributor to the AI computing fabric with the ratio of server CPUs potentially reaching parity with the number of accelerators in the data center segments of the market.

Can you just unpack for investors like the type of conversations you're having with customers on this front? How you see CPU intensity progressing over the next few years and puts and takes based on sort of various types of Agentic applications?

Yes, it's a very good question. Clearly, this Agentic AI kind of forecast that it was.

That's right.

And that turned out to be very accurate and also very helpful. We thank you for that. And then in some way, that's where the drive the CPU demand. And then used to be a training is 1 CPU to 8 GPU. And then now in the Agentic AI with all the agents, people -- I work a lot with the frontier model, some of the start-up are involved. They all tell me Lip-Bu, CPU actually is more useful, even single threaded. And he said that the most important is the reinforced learning, the orchestration of all the different agents and then also the optimizing for some of the workload and CPU is even more important.

So I can starting to see not just my wishful thinking, customer has said to me Lip-Bu, more like 1:1. And now even some of them tell me it's 4:1. So 4 CPU to 1 GPU, so the inference and agent. So I think that part, I think, is very sometimes in life, you need some breaks. And this is a good break for me. And so CPU become high demand, and I try to make sure that we can meet the requirement from the customer.

And then the other part, the next frontier is going to be the physical AI. And that's also coming up. So in a way, it's more than human, now you have to look at the compute requirement for managing the agent from the personnel and also to the commercial side.

That's right.

And that market is quite big. And so I think there's opportunity for me to work on the CPU and then drive new architecture to drive the purpose-built workload to optimize for particular workload going to be. And then same thing with accelerator and then make sure that we have that. So back to the AI, we have 10 years ago, I got 2 ideas because I thought GPU power hungry. So I back this wafer scaling. This is Cerebras adjustment part and I was Series A investor. And then the other one I back is the dataflow architect that is SambaNova. And that's why we have a deep partnership with SambaNova, the x86 CPU and GPU and with SambaNova accelerator or RDU, we call it, and then that we can go together and then win some of this cloud infrastructure play and with much lower power requirement. So those are the things that you have to think way ahead. And right now, we are thinking the next big frontier is physical AI.

So recently, I just hired Alex from Qualcomm. He used to lead compute and mobile, wearable and physical AI. I come over here to help me to build the physical AI, all the way full stack from silicon optimized to all the way to the software to the system platform engineering and drive the solution for robotics or digital workers. So I think those are humongous opportunity to address this whole Agentic AI, millions of agents and worker -- digital worker. I think this is a new era that we can really capture the opportunity.

Yes. And we were talking about it last night. I remember, when it comes to physical AI, a lot of the use cases are in areas like industrial, automotive, edge AI. And we were talking about it last night, but we call these the embedded segments of the market, right? And in embedded, interestingly enough, Intel still holds very, very strong market share. So I think it's a big opportunity from the perspective of there's a lot more intelligence and compute opportunities with physical AI. But in this case, Intel is starting from a position of very, very strong leadership. Would you agree with that?

I agree with that. And in fact, not just embedded and also the FPGA. And so we spin out the Altera. The CEO is a very good friend of my Raghib is very good. I kind of played the role to recruit him to be the CEO. And the FPGA together with the embedded, and then we can really drive some of the new physical AI and then will be a very exciting opportunity for us.

So I think what we've been hearing from certain shareholders is that they like the focus on the CPU architecture, continue to drive the road maps very aggressively. Obviously, we'll be talking about the foundry in a little bit. But the market is a little bit concerned that as you continue to focus on building back leadership in these 2 areas that they wonder that does Intel have the additional resources and capabilities to also execute a strong AI accelerator family of products, right, and try to move towards that full stack sort of AI sort of implementation.

We know that you have your Gaudi-based accelerator family of products. You talked about the partnership with SambaNova for inferencing applications. But help us understand the team's strategy around accelerators and your ability to compete with merchant GPU or ASIC XPU that are already out in the marketplace.

Yes. Good question. So I think we need to do both. And on the CPU side, we have to do more architecture change to purpose-built silicon. And then on the accelerator side, we're going to team up with SambaNova to do that. And then we also have an internal plan in the AI development and so that we can come out kind of leapfrogging to some of the areas that can drive some performance benefit. In the Agentic AI, physical AI, I think there's a new area that we can really capture.

In some areas, we are behind and no point to just do catch up to do something new. And so I think we try to get into areas that less crowded and something that we can differentiate in terms of performance or power or software that we can do that. So I think that we have to do. And then the other part is the system level and the full stack. And then some of the area I don't have is the I/O high-speed connectivity. Even though earlier, I invested in Astera Labs and Credo, they both went public very nicely.

And then Cadence, I bought a company called [indiscernible], basically has a high-speed SerDes. So right now, I think it is moving to optical now. So we need to look at what are the optical area that we can disrupt. And then same thing, just besides CPU, GPU accelerator, we also look at the memory. Memory becomes a big, big shortage and has become a bottleneck. And then how do we use CPU and memory to really couple together to make full optimization of the memory. That's something that we're working with the memory player and also we're going to recruit some of the talent.

We used to have memory business. We've sold it to SK Hynix. So right now, we have to figure out a way to really secure some of the memory requirement to serve our customer. So those are the things that we are focused on.

If we turn to your manufacturing and your foundry business, can you just give us a progress report on your 18A, 18AP, next-gen, 14A process technologies? Qualitatively, yields, cost, manufacturability, maybe compare that relative to your prior generation nodes, but also maybe importantly, compare it to TSMC's metrics that I'm sure you guys track internally as well?

Yes. Good question. So I think the Intel manufacturing used to be our strength. But we lost some of them. And then secondly, we never do the foundry service business. So if a culture change, I need to have a separate team that really know how to serve the customer. And that's why I think you'll see that some of the people are hired for Samsung Foundry, Shan Han, know him for 15 years to come over to help me. And then secondly, we're also going to look at how can we improve the yield. And then from the 18A, good news is much better than I took over. I can see 7% yield per month improvement.

And then right now, it's ahead of schedule compared to end of the year target. And then 14A, we are engaging with customers with real engagement. And then the IP block are also gradually building it up. So I think overall, still a long way to go. I mean, TSMC, Morris took them at least 10 years to build what they have.

That's right.

And so it's not going to be overnight. But good news, 14A, my risk production is 2028 and volume production in 2029 is about the same time as A14 for TSMC.

That's right.

And then now I'm starting to look at the 10A, 7A, the road map. People don't go to you just one node. They're looking for the road map for the future. So we want to build a long-term business. And then we can drive the efficiency, the defect density and then we can go to that Rule of 45, how to drive the operating efficiency, the profitability, cash generation. So those kind of a multiyear milestone that we are going there.

And you talked about -- you released your 0.5 process development kit for 14A beginning of this year. I know we know multiple customers that are designing test chips, running them through Intel's fabs right now. But you also have your 18AP process, which is your current generation process, which you are opening up for foundry this year -- opening up for foundry as well. I know you had anticipated early design commitments in the second half of this year, but we are in an overall very, very tight foundry supply situation. Geopolitical dynamics are still pretty volatile. And so I would think that some of your customers want an accelerated time line. But has the Intel team already secured design commitments on 18AP and/or 14A? Or is it still a little bit too early?

Yes. The 14 and 18A, when I took over, I decided to give my do [indiscernible]. And so in a way, I kind of focus on internal, which that Panther Lake is taking more successful. And now customers here that we are improved. So they're asking about 18AP. And then we are engaging with a couple of customers looking for the supply with resilience and robustness rather than depend on one single vendor, foundry.

So I think we have opportunity there. And the 14A, we already engaged with multiple customers. And my philosophy, don't announce ourselves. If a customer want to announce, we welcome them to announce. We support them to do it. And so I think we're making good progress. The other part, it's also a very happy surprise for me. We have a very good advanced packaging. And so this is something that a couple of customers come to me, EMIB-T is so good. We want to use that.

And in fact, right now, as you know, some of the substrate material is very shortage, 4 from Taiwan, 2 from Japan, they're all asking us to prepay the substrate commitment. And we ask our customer, if you are serious to use our EMIB-T , can you help me on the substrate prepay? They jump on it. So they meant that they show the commitment, they really want our technology. And this is not a few million, it's billions in the next few years.

So I think it's exciting with the advanced packaging, with the foundry and now we have all the new CPU architecture come in. And I think we are in a very unique area that we're going to go after some of the ASIC business so that we can really purpose built for some of the customers in a very fast turnaround time and the best performance to optimize all this together. So that's something that's very exciting for me.

Great execution and great opportunities in front of the Intel team. Lip-Bu, thank you very much for participating today.

Thank you.

Thank you for standing by, and welcome to the Intel Corporation Earnings First Quarter Earnings 2026 Earnings Conference Call. [Operator Instructions] As a reminder, today's program is being recorded.

And now I'd like to introduce your host for today's program, Mr. John Pitzer, Corporate Vice President of Investor Relations. Please go ahead, sir.

Thank you, Jonathan, and good afternoon to everyone joining us today. By now, you should have received a copy of the Q1 earnings release and earnings presentation, both of which are available on our Investor Relations website, intc.com. For those joining us online today, the earnings presentation is also available in our webcast window. I am joined today by our CEO, Lip-Bu Tan; and by our CFO, David Zinsner. Lip-Bu will open with comments on first quarter results, as well as provide an update on the progress we're making on our strategic priorities. Dave will then discuss our overall financial results, including second quarter guidance before we transition to answer your questions.

Before we begin, please note that today's discussion does contain forward-looking statements based on the environment as we currently see it, and as such, are subject to various risks and uncertainties. It also contains references to non-GAAP financial measures that we believe provide useful information to our investors. Our earnings release, most recent annual report on Form 10-K and other filings with the SEC provide more information on specific risk factors that could cause actual results to differ materially from our expectations. They also provide additional information on our non-GAAP financial measures, including reconciliations where appropriate, to our corresponding GAAP financial measures.

With that, let me turn things over to Lip-Bu.

Thank you, John, and good afternoon, everyone. Q1 results demonstrate continued and steady progress across the business, reflecting strong demand for our products. and disciplined execution to expand available supply. Revenue, gross margin and earnings per share were all above the high end of guidance. marking our sixth consecutive quarter of exceeding financial expectations. Even as we improve factory output, Demand continues to run ahead of supply for all our businesses, especially for Xeon server CPUs, where we expect sustained momentum this year and next. Intel 3 base Xeon 6 and Intel 18A based core series 3 products are now in full volume production ramp, and each represents the fastest new product ramp in 5 years. We are maximizing and optimizing our factory output to meet customer needs. It is our top priority.

Intel is now a very different company than when I first joined over 1 year ago. We have taken and continue to take deliberate steps to rebuild Intel into more competitive and more profitable company. Our cultural transformation is well underway and we are embracing our roots as data-driven, paranoid and engineering-centric company. We are also listening closely to our customers. and putting them at the center of everything we do. In our processes, some of the most vital assets necessary to be successful and to flourish in this era of extraordinary opportunity for the semiconductor industry. With a stronger balance sheet, new leadership team, resulted and motivated workforce and a renewed focus on engineering execution, we are turning our attention squarely towards innovation to capture opportunities in the near term and to position the company for robust growth in the long term, driven by tremendous demand for AI, the semiconductor industry TAM is now approaching $1 trillion. Intel is well positioned to benefit from this demand with 3 strategically important assets.

Our x86 CPU franchise, our advanced packaging technology and our vast manufacturing network. Artificial intelligence is now moving into the real world towards a more distributed inferenced and reinforced learning workloads, like agentic, physical AI and robust and edge AI. This shift is now beginning to show up in our results. as I want to spend some time on this today. For the last few years, the story around high-performance computing was almost exclusively about GPU and other accelerators. In recent months, we have seen clear signs that the CPU is reinserting itself as the indispensable foundation of the AI era. CPU now serves as the orchestration layer and critical control plane for the entire AI stack. This is not just our wishful thinking.

It is what we hear from our customers, and this is evident in the demand profile for our products. Xeon server demand is seeing strong and sustained momentum. Customers are deploying server CPUs along accelerators in the ratio that is moving back towards CPU. The accelerator remains central to frontier AI, and we will continue to participate, innovate and partner in that category. Our recent announcement with SambaNova Systems is an example of such partnership on hectogeneous compute architectures. But the backbone of AI computing in production remained a CPO anchor architecture. That is good news for the x86 ecosystem. It is great news for Intel. And it is a structural reason, I'm confident that CPU franchise will continue to be a meaningful growth engine for the company in the years ahead, not just quarters ahead.

Turning to Intel Foundry, the accelerating deployment of AI infrastructure, creates a meaningful opportunity for us as we continue to build our external foundry business. I'm pleased with the progress we have made in foundry technology development over the last year. Even though I will continue to remind you, this will be a long journey for us. We have made steady progress with Intel 4 and Intel 3 and 18 AEs are now running ahead of the internal projections, representing a meaningful inflection in our execution and our factory finished good output. We also continue to make steady progress on our advanced packaging technologies, including additional growth in customer backlog in the quarter. On Intel 18 AP and Intel 14A, we continue to be encouraged by our external engagements. Intel 14A, maturity, new and performance are outpacing Intel 18A at a similar point in time and we continue to develop PDKs with multiple customers actively evaluating the technology.

Their partnership has been critical and their feedback has continued to help us define the technology so that we can cater to their needs. We expect to see earlier design commitments emerge beginning in the second half of 2026 and expanding into the first half of 2027. And I'm particularly pleased that our progress today has driven us to land more of our own future product types on Intel 14A as well. At the time when advanced wafer capacity is in the short supply. This enable us to have better control over our supply chain. Intel has pioneered nearly every major innovation that has enabled dimensional scaling and high-volume manufacturing of silicon transistors over the last 6 decades. We have always been willing to take measured risks that have eventually passed away for step function improvements in transistor density, cost, power and performance.

As we look to continue challenging the state quo, I can think of no better partner than Elon Musk. We recently announced our partnership with SpaceX, xAI and Tesla to support Terafab. Elon and I share a strong conviction that global semiconductor supply is not keeping pace with a rapid acceleration in demand. We are excited to explore innovative ways to refactor silicon process technology looking for unconventional ways to improve manufacturing efficiency that will eventually lead to a dynamic improvements in the economics of semiconductor manufacturing. A year ago, the conversation about Intel was about whether we could survive. Today is about how quickly we can add manufacturing capacity and scale our supply to meet enormous demand for our products. This is a fundamentally different company today, and we still have a lot of work ahead. I would like to take this opportunity to thank our many customers, partners, and our hard-working employees across the world for their contributions towards building a new Intel. I remain firmly convinced of and focus on the opportunity ahead for Intel.

With that, I will pass it to Dave.

Thank you, Lip-Bu. We delivered robust Q1 results, reflecting strong demand and better-than-expected available supply. We also benefited from improved product mix and pricing actions in part to offset higher costs. First quarter revenue was $13.6 billion, $1.4 billion above the midpoint of our guide. Q1 revenue would have been meaningfully higher but demand continues to outpace our growing supply. Our collective AI-driven businesses now represent 60% of revenue and grew 40% year-over-year. These results reflect real and deliberate changes we have made to be more responsive and accountable. This quarter, our teams work directly and diligently with customers to reach mutually beneficial outcomes in weeks, not months. We value the partnership and support shown by our customers, partners and suppliers as we work to navigate this environment together.

Non-GAAP gross margin came in at 41%, approximately 650 basis points ahead of guidance due to the combination of higher volume, which included previously reserved inventory mix and pricing. In addition, better yields on Intel AT&A offset some of the higher costs we always incur in the early part of ramping a new node. We delivered first quarter non-GAAP earnings per share of $0.29 versus our guidance of breakeven on higher revenue, stronger gross margins and continued spending discipline. Q1 EPS included a roughly $0.06 onetime gain in interest and other. Q1 operating cash flow was $1.1 billion, with gross CapEx of [ $5 billion ] in the quarter and adjusted free cash flow of minus $2 billion.

Moving to segment results. CCG revenue was $7.7 billion, down 6% sequentially and better than our expectations. Even with improved factory output, demand outstrip supply against a client TAM that remains resilient despite industry-wide component shortages and inflationary pressures. Our AI PC revenue grew 8% sequentially and now represents greater than 60% of our client CPU mix. Operating profit for CCG was $2.5 billion, 33% of revenue and up approximately $300 million quarter-over-quarter on improved mix and product margins, sales of previously reserved inventory, better AT&A yields and lower operating expenses. Within the quarter, CCG launched Core Ultra Series 3 and expanded our offerings across consumer, commercial and edge. This has proven to be our strongest product launch in 5 years, delivering better performance per watt, stronger integrated graphics and more capable on-device AI features, all while maintaining our broad ecosystem of compatibility that partners and customers value.

In Q1, CCG also expanded the reach of our core family by launching the Intel Core Series R processor, which brings the latest IP, modern features and all-day battery life to the mainstream for the first time. We're enabling a new class of mainstream systems that once again set the standard for everyday computing. PC AI revenue was $5.1 billion, an increase of 7% sequentially and 22% year-over-year. well above expectations and reinforcing the strong year of growth for PC AI we signaled 90 days ago. Strength continued across all segments and customers. as investments in CPUs are accelerating to support the evolution of AI from foundational training to inference and from inference to Agentic. We also saw strong ASIC growth with revenue up more than 30% sequentially and nearly doubling year-over-year.

Operating profit for GCAI was $1.5 billion, 31% of revenue and up approximately $292 million quarter-over-quarter on improved product margins better cycle times and yields, especially on Intel 3 and lower operating expenses. Within the quarter, DCAI signed multiple long-term agreements including Google, supporting our view that the current business momentum is sustainable. In addition, Xeon 6 was selected as the host CPU for NVIDIA's DGX Rubin NVL8 systems and Xeon remains the most deployed host CPU due to its industry-leading memory, security and networking orchestration. Lastly, DCAI also established a multiyear collaboration with SambaNova to design a next-generation heterogeneous AI inference architecture, combining SambaNova RDUs and Intel Xeon 6 processors.

Intel Foundry delivered revenue of $5.4 billion, up 20% sequentially on increased EUV wafer mix driven by Intel 3 and significant growth in 18A. External foundry revenue was $174 million in the quarter. Intel Foundry operating loss in Q1 was $2.4 billion, and improved $72 million quarter-over-quarter as better yields across Intel 4, 3 and 18A, drove higher gross margins. This was mostly offset by increased operating expenses associated with an intentional step up in Intel 14A investments to support both internal and external customer evaluations. As a reminder, Intel Foundry carries the bulk of the costs associated with the early ramp of Intel AT&A, We expect Intel Foundry's operating loss to improve through the year as AT&A continues to ramp into volume and yields improve further.

Within the quarter, Intel Foundry delivered output above our expectations drove steady improvements in yields and met key 14A milestones. Intel Foundry also added to its backlog of advanced packaging services and announced a multiyear expansion of our back-end facilities in Malaysia. This expansion will help support the committed demand that will begin to convert to revenue in 2027. Turning to all other. Revenue came in at $628 million and was up 9% sequentially due to a strong quarter for Mobileye. Collectively, the category delivered an operating profit of $102 million.

Now turning to guidance. As we look ahead, we remain mindful that the macroeconomic and geopolitical environments are dynamic, views on global growth policy and trade continue to shape customer behavior and investment decisions. In addition, constraints and rising prices around key components like memory, wafers and substrates are driving higher costs that could impact demand for our product at some point in the year. We're prudently planning for PC demand to weaken in the second half of the year and expect the full year PC unit TAM to be down low double-digit percent in line with industry peers and experts. Offsetting this, near-term customer order patterns remain very robust across all of our businesses. In addition, our confidence in the sustained growth CPUs driven by the AI infrastructure build-out is growing. Our outlook for server CPU demand has improved over the last 90 days and we expect a strong year of double-digit unit growth for the industry and for us with momentum extending into 2027.

Combining all of these factors, we're guiding Q2 revenue to a range of $13.8 billion to $14.8 billion, up 2% to 9% sequentially. As we work hard to support the need of all of our customers, we expect sequential revenue growth in both CCG and DCAI on improved supply and a full quarter of pricing actions with DCAI, up double digits. At the midpoint of $14.3 billion, we forecast a gross margin of 39%, a tax rate of 11% and EPS of $0.20, all on a non-GAAP basis. Our Q2 gross margin guide declined modestly from Q1 due to a meaningfully larger contribution from Intel AT&A, still early in its ramp and some inventory benefits in Q1 that aren't expected to repeat in Q2.

Before I close, I'll share some additional insights on the full year. We expect our factory network to continue increasing available supply in the third and fourth quarters though at a more measured pace than we anticipated 90 days ago, reflecting the base effect of much stronger-than-expected first half output. We also expect 2026 revenue on a half-on-half basis to follow the seasonal trends experienced over the last 10 years with servers above and PCs below. We were very pleased with Q1 gross margins, and we will continue to push for gross margin expansion. It's my top priority. Our foundry team is delivering consistent yield and throughput improvements across all process nodes, which will help gross margins. With that said, Intel AT&A is still early in its ramp and rising input costs, especially in memory, present growing headwinds in the second half that we need to overcome.

For OpEx in 2026, we have been directionally targeting $16 billion that are likely to be higher due to inflationary pressures, variable compensation and targeted investments we are making to capture the opportunities ahead. The drive for efficiency is core to the new culture Lip-Bu is creating, and we will remain laser-focused on finding additional operational improvements and maximizing ROI on all of our investing activities. We forecast capital expenditures in 2026 to be flat to last year versus our prior expectation of flat to down. reflecting increased capacity investments to support committed demand and a continued emphasis on improving fab productivity and output. We now expect expenditures to be roughly equal across the year and still to be heavily weighted towards the equipment that directly grows wafer outs to support growth this year and next.

We recently closed the transaction to repurchase the 49% equity interest in the investment in Fab 34 in Ireland, a highly accretive deal, allowing our shareholders to participate in the full economic benefits from a fab just now hitting its stride. As a result, we now expect noncontrolling interest, or NCI, to net to approximately $250 million in each of Q2, Q3 and Q4 of this year. and be approximately $1.1 billion for 2027 and 2028 on a GAAP basis. Lastly, excluding the buyout of the Fab 34 joint investment, we still expect positive adjusted free cash flow for the full year. As a reminder, we funded our purchase with approximately $7.7 billion in cash and $6.5 billion in new debt. We remain committed to retiring all $2.5 billion of maturities as they come due this year and all $3.8 billion in 2027.

In closing, Q1 was a strong quarter financially and operationally. All demand signals continue to emphasize the growing and essential role of the CPU in the AI era, and the unprecedented demand for leading-edge wafers and advanced packaging to realize the vision of driving silicon-based intelligence to the edge efficiently and at scale. Our confidence is growing. We have the right team and the broad IP portfolio needed to solve our customers' most pressing economic challenges and drive long-term value for our shareholders.

With that, I'll turn it over to John to start the Q&A.

Thank you, Dave. [Operator Instructions] With that, Jonathan, can we please take the first question?

Certainly. And our first question comes from the line of Ben Reitzes from Melius.

Congrats on the quarter and this is good news for the country too. With regard to my question. The first 1 is on [indiscernible]. If you could just talk about the Google deal and there's a comment I believe in the release that you signed other [indiscernible] and how do they get make better ability of the environment data.

Yes, good question, Ben. I think let me describe. Google is 1 of the multi long-term contract agreement in Q1. And this is significant. Google, we have the Xeon IPO and building a long-term trusted partnership, which is very important for us. It's a very just evidence of a strong demand for our CPO and some of the ASIC business that is important for us. And this is a good example of and how we win in AI infrastructure build-out. And then stay tuned, we -- at the right time, we announce other contracts. Dave, anything to add?

Yes. Maybe just to add, most of these agreements are structured with volume and pricing and are usually somewhere between 3 and 5 years. The Google one, I think both parties wanted to see an announcement. In some cases, customers want to keep that confidential, and we respect their desire to maintain confidentiality. So some of them we just didn't announce. It's a win-win, I think, in a lot of ways. We get a good understanding of the volumes that we can build into our assumptions around supply is good for the customer because they know where the supply is coming from, and they get a good sense of what pricing they can expect.

Ben, do you have a brief follow-up?

With regard to CapEx, is there anything in there with regard to investing in foundry customers? Or is that still not there? And when do you think we'll hear more about that in the CapEx figure?

I'd say let me unpack CapEx just for a minute. So what we're now calling it is flat year-over-year. initial thinking was that it was going to be down. I think we kind of moved it last quarter to flat to down. And now I think we're looking at flat. And that is really a function of the current demand environment we're seeing. One thing to keep in mind, in the last few years, a lot of our CapEx spending was space. And I think we're actually in a pretty good position in the space. We wanted to have white space salable to move into when needed. And I think when I both feel like we're in a good place. So we actually will be bringing the space spend down pretty materially, even though the total is flat.

And so what that means is the tool spend is actually increasing pretty significantly. In fact, tool spending will be up year-over-year 25% or so. And so that's, I think, a function of the fact that we just see a lot of demand, and we want to make sure we're catching up on the supply front. And then as we get into next year, we'll have a better sense, I think, of what CapEx looks like for next year. As it relates to external customers on the foundry side, our expectation in which we've been pretty consistent on this through almost, I think, a year is that we thought that customer signals would be more concrete in the back half of this year and into early next year.

And so as we kind of pull that information together, combined with our own requirements, which are growing over time here. That will give us a good sense of what supply we need over the next few years, and we'll be putting the spend in place. Lastly, maybe just to tack on I think our relationship with the equipment vendors is quite strong. And so I think we have a pretty good ability to flex as needed. Naga and Lip-Bu are in regular engagement with all of the CEOs of the equipment suppliers. And so I think we'll be able to manage and course correct as necessary as we get a better sense of the supply dynamics for us, both internal and external and move our capacity accordingly.

And our next question comes from the line of Ross Seymore from Deutsche Bank.

A couple of questions. First on the CPU side of things, the server CPU side. Can you talk about how Intel's positioned competitively? Is the strength that you're seeing more that the market demand is just that high? Or do you believe that your product line actually has some competitive differentiations versus either other x86 competitors or ARM offerings?

Yes, Ross, I think good question. So first of all, I think the feedback from the customer. CPU is very important when they move from training to inference. The inference side, I think in terms of orchestration, control plan and also managing all the different agent with data, CPU is much more efficient. So I think the ratio of CPU to GPU used to be 1 and 8, and now it's 1 to 4 on and I think towards priority or even better. So I think that demand is very strong. And then secondly, I think address your issue question. I think clearly, we are kind of continue to refine our road map in terms of, at the end of the day, it's the best product win. We have a lot of changes in terms of the CPU architecture change to focus on optimize for the different workload. And then the other part is, I think we have a very big advantage.

We're not just at the CPU. We have advanced packaging and foundry, we can really effectively driving some of the changes more quickly to serve the customer in terms of their different workloads. So I think overall, I think it's an exciting time that we call it the XPU. Besides CPU, we are quietly building up the GPU with a new hire and also we are moving into the acceleration and so that we can serve the customer from the edge and then to the physical AI and then really drive some of the new initiatives to drive the competitiveness.

Ross, maybe 1 other thing to add is that just -- it's obviously early in the Granite Rapids life cycle here. But so far, the early traction has been quite good. So at least that's a positive step for the data center CPU business. .

Ross, do you have a follow-up question?

Yes, I do. Lip-Bu, following up on something you said in your preamble where you said a year ago, Intel was trying to survive and now it's all trying to scale the supply, and that's a very positive change year-over-year. how does the business model and the spending behavior strategically changed in that? Dave talked about increasing CapEx a relatively small amount, maybe $17 billion up to $18 billion this year. But if you're scaling supply is under demand across the board. Is that something that you can handle with just improving yields or does structurally CapEx need to go up? And maybe call into question that you're not going to spend on 14A until you actually get customers thesis that you've said in the past.

Yes, good question. So I think in terms of spending, like Dave mentioned earlier, we are -- over the last year, we drive a lot of efficiency, driving a lot of layer of management team. And then now we are really focused on I spent a lot of time meeting with customer. So we're understanding the workload, understanding the engineering side, how do we drive improvement in terms of the architecture, the execution in terms of tape-out and design to really drive efficiency there. And then more important, I think, is talk to the foundry side -- now clearly, we really drive the yield improvement. We see a very nice yield improvement on the AT&A. And then the 14A we already have the 0.5 PDK available and now we are aiming for the 0.9 PDK. That's where customers are going to decide which product, how much volume capacity we need to have. And then besides us driving the yield -- we're also driving the improvement in the cycle time so that we can really meet the customer demand and timing that they request and then really optimize for them. .

And our next question comes from the line of Stacy Rasgon from Bernstein Research.

So the first one, I did want to dig into the segment outlook and I guess the implications for gross margin. So you said data center up, I guess, double digits, so that puts it up, I guess, 40% year-over-year or something like that. assuming PC sequential or year-over-year similar up maybe low single. I guess I'm just surprised the gross margins, I understand the inventory benefit in Q1, but it feels to be gross margins are still -- are probably flat, excluding that inventory benefit, maybe even down a little bit. I guess I'm just surprised, given the magnitude of the server growth especially given the AT&A yields are supposed to be improving. So are they still low enough that the AT&A mix is just completely offsetting that? Or I guess any color you can give us in more detail on the gross margin drivers in the near term would be really helpful. I'm just a little confused.

Yes. Okay. So obviously, I don't have your model in front of me. But the -- I'd say the gross margins we will see some benefit from pricing. We got a little bit of pricing benefit in the first quarter, but I would expect us to see some more meaningful improvement in the second quarter. That's certainly going to help mix, I don't know, it's going to plus or minus be in the ZIP code. Yes, data centers like it's obviously going to grow faster, I think, but I'm not sure that mix is going to drive much. But 18A is going to be a pretty decent headwind to our gross margins. And if you look at Panther Lake volume increases, it's going to be going up I don't know, 6 or 7x in the second quarter relative to the first quarter. And while the gross margins are improving in Panther Lake quarter-to-quarter, it's still below the core average. So when you have that big a shift in the mix with gross margins below the corporate average, it kind of weighs down on the gross margins.

But I mean, we're roughly in the ZIP code of what Q1 was like anyway. So I'm not too concerned about it. In the back half of the year, we'll see you have some of those dynamics helping us. I'd say the 1 cautionary concern I have on gross margin in the back half of the year is just some of the materials have gone up in terms of costs, substrates are going up, glass. We've got memory going up, as you know. So those things offset some of the improvements that we're having through the year. Longer term, I'm still hyper focused on gross margins. And I think we have elements of the road map in the right place in terms of cost structure, certainly on client, definitely seeing improvement on the foundry side, we've got more work to do on the data center front, but our goal is to get the gross margins up quickly.

Stacy, do you have a follow-up question?

I do think. To push on the PC a little bit. So you said industry volumes probably down double digits. So it's going to be even worse I guess, in the second half, given you guys are running pretty strong in the first half. I guess do you expect your full year client revenues to be down consistent with that industry outlook? Or is pricing helping you or hurting you? Is share helping you or hurting you? I think about the -- how do we think about the shape of your client business in the wake of that industry forecast?

Yes. Good question. Well, 1 thing you've got to kind of separate is when we talk about the industry, we're generally talking about consumption and when we're -- obviously, that's different than our billings because of the inventory movements at customers. We are not going to be as impacted as the industry TAM because we expect partly in terms of pricing a little bit, but also because of inventory replenishment at the customer level. So I think from a modeling perspective, if we -- whatever we get to in 2Q is probably what we run the rest of the year roughly. So it's going to be kind of flattish from Q2 onward from a revenue perspective, at least is how we're modeling it.

Our next question comes from the line of Timothy Arcuri from UBS.

I want to ask just about the evolution of your foundry model. And of course, you're pursuing typical foundry customers, but it seems like Terafab is a little bit of a different deal and maybe even like a process licensing agreement. I wouldn't normally ask about 1 particular customer, but he did talk about it yesterday. So I'm wondering if you can just talk about that? Like is that going to be a typical foundry arrangement? Or are you sort of going to possibly turn the keys over unlike an entire fab to the FX?

Yes. Timothy, thanks so much for the question. So I think on the 4A, I think we are making great progress in terms of yield and cycle time. And clearly, we are engaging with multiple customers, in heavy engaging and especially my style is under promised over delivering. So we have no plan to announce the customer unless the customer want to announce it, and we are supporting that. So I think back to the tariff fab, clearly, Elon and I, we believe that global supply chain is not keeping pace with the rapid acceleration and the demand. And so we both share the vision that we're going to learn a lot together, exploring the innovative way and then in the process of the manufacturing Saying all this, clearly, it's a very broad relationship. And then we would update you as we go. Clearly, this is a very exciting customer to work with, and we have multiple other customers we are engaging. Stay tuned. We'll update you when we can. .

Tim, do you have a follow-up?

I do. Yes, Dave, is there a way to sort of quantify like how much demand you're sort of missing out on? Like how much are you undershipping the market still -- is it like as much as 10%. So if you were unconstrained, you could -- revenue would be like 10% higher. Is that like a reasonable number?

I probably not want to put a specific number. Let's just say it starts with a B. It's -- so it's meaningful.

And our next question comes from the line of Vivek Arya from Bank of America Securities.

For the first one, I just wanted to understand the server CPU TAM growth this year. I think, Dave, you mentioned up double digit. I was hoping you could help kind of tighten that. Is it in [ 10 15 ], [ 15 20 ]? And then how much ASP expansion do you expect this year also, right? And what I'm really curious to understand is the new server TAM growth that you have, how does this compare versus what you thought 6 months ago, just so that we can get a better sense for what does this agentic CPU workload mean in terms of incremental unit and ASP growth?

I mean I think when we're talking about the market, we're generally talking about units 6 months ago, we certainly expected it -- well, I'd say probably at that point, we were thinking it would be up instead of down from a units perspective. Now obviously, it's going to be up meaningfully. I'll probably leave it to the industry analysts to come up pinpoint the exact number. ASPs, we have moved, obviously, to offset some of the cost increases we've seen over the last couple of quarters. But it's not the biggest driver, obviously, of our revenue outlook. We think the unit volume is going to be the biggest driver. Now that's on an ASP per core basis. Obviously, core count is increasing significantly in the data center CPU space. And so we get the lift on a -- as core count increases, we get the lift on the ASPs from that and that obviously is meaningful. .

Vivek, do you have a follow-up question?

Yes. maybe the follow-up, Lip-Bu for you on server CPU competition. So both when we look at competition versus x86 against AMD, do you think you are gaining share? Do you expect to gain share again? And then broader, I think the competition against ARM because NVIDIA is planning to launch a stand-alone never CPU rack, Recently, we heard Amazon talk up their Graviton option. I think Google yesterday said they would launch Axion and connected with every TPU. So just kind of near term, how do you look at competition versus AMD in x86. And then kind of medium to longer term, how do you see the competition develop against ARM?

Good question, Vivek. I think clearly a couple of things. One, the CPU is in great demand right now, I think we all enjoy that. And then in terms of the our product road map, we have been fine-tuning the last year. Typical new chip come out, it will take about 12 to 18 months. We are laser-focused on execution, multi trading, I think we are putting in. So we're going to have [indiscernible], have the multi-trading then we can compete effectively with AMD. And we try to accelerate that [indiscernible] ahead. And then the other part is we're also looking at some of the architecture CPU and GPU architecture. I've been recruiting some of the top talent to really refine the new product to effectively competing against. In terms of your second question about ARM, clearly, we know a lot, a lot of respect for them. Renee is a good friend of mine. They have a licensing model. they have been very effective.

And of course, they have continued to raise the loyalty fee. And then now they also have the silicon team and building the silicon to compete and then provide us reference. And clearly, some of the Amazon, Google, they are using that. That's not news. [indiscernible] not led back to Amazon. So very familiar with that. Good news is, I think, clearly, we have this OEM customers are working with us. And then we also have a long-term relationship with some of these important customers. So the road map from Granite Rapid to Diamond Rapid and then Colorado Rapid is coming up strong. We very like our portfolio. And then on the server side, we also look at the -- beside the 86, and we also have SambaNova partnership so that we do the data flow architecture to really driving that all together. And we already have some success on that. And then the other part, we also recruited some of the top talent. [indiscernible] used to be running the ARM data center server chips. So he know very well. And then Srini used to be working with me at Cadence, we optimized for all the ARM products for the customer. So he knows how to optimize on the requirement in terms of tweaking the performance to meet the requirement. So I think all in all, I think we have the team -- we have the technology road map. I think we're going to be, over time, going to be a very effective competitors to them.

And keep in mind, Vivek, that beyond the product side, we have another bite at the apple or maybe multiple bites at the apple on the foundry side. We can provide customers with advanced packaging. We can provide customers with wafers. So we have a pretty strong breadth of offerings to customers to help support their CPU needs or AI needs in the marketplace.

Our next question comes from the line of C.J. Muse from Cantor Fitzgerald.

I guess I was hoping you could look through how you're planning to drive increased output through the second half of the year. How much of it is yields? How much of it is cycle times? How much incremental wafer fab equipment as well as outsourcing to TSMC. Would love to kind of get a feel for that please.

Okay. So first and foremost, we are increasing wafer starts and all 3 of our nodes Intel 107, Intel 3 and 18. More meaningfully, the EUV nodes of course, but even in Intel 107, we'll be increasing wafer starts this year. So that's a key component of our ability to meet demand. That said, Lip-Bu has pushed the team really hard to provide more supply, the old-fashioned way with better yields and better throughput. And I think that's largely how we got it in the first quarter. So we can expect him to do that through the year, and I think that will be a meaningful contributor to our output. Of course, we use outside foundries as well, and we flex them as needed as well. We have -- Lip-Bu's got great relationships with the external foundries. And so he's able to leverage that to help us in that area as well. So it'll be certainly a component of it as we move through the year. .

Just to add on to that, I mean TSMC is a very important partner for us. Morris and CC have a decade of friendship. And then clearly, we know our product group will decide which is the best foundry. So I think we're going to use a multi-foundry approach our own internal and also external. And so we really have good relationship, continue to build from both sides to benefit the customer.

C,J.,do you have a follow-up?

I do. I guess I would love to kind of level set where we are on the advanced packaging front. You talked about rising backlog. Anything you can share in terms of what that number looks like, revenue targets this year or next number of customers? Any help there would be great.

Yes. I mean I think I've actually said this in the past. We have been really pleased with our traction there. And I think maybe naively, I had thought that these opportunities would come in the hundreds of millions of dollars level. But so far, what we're seeing is that their demand is more in the billions of dollars per year kind of level. So this is going to be a big part of the foundry revenue as we get through this decade. And the good news is advanced packaging really is a differentiated offering for us, and it does a lot for the customer in terms of allowing them to use larger reticles. So there's real value to the customer. And as a result, we get very attractive pricing relative to some of the other areas of the foundry business. So we'd expect this to be at least foundry at corporate average or foundry average gross margins over time.

Our next question comes from the line of Srini Pajjuri from RBC.

My first question on the 18A yields, Dave. Obviously said deals are better than you expected and it looks like they're improving further. But at the same time, it is still a headwind to your gross margins. So if you could give us some context as to how much better they are. And also, as we go through the year, when do you see -- I guess, when do you expect that being a headwind to your gross margin, when do you expect that, I guess, at least near to gross margins.

Yes. I think T&A yields are somewhat a closely guarded proprietary piece of information for us. So we don't typically -- I would just say, Lip-Bu had a target as we came into the year. for the end of this year, and we're probably going to hit that probably the middle of this year. So he's done a very good job working the team to drive a better response there. And of course, that carries on to next year's expectations around yields. I think as we get towards the end of the year, on a kind of a full picked basis. This is combining the product margins and the foundry margins will be in a relatively decent place in terms of the gross margins at Panther Lake. We've got more work to do at the foundry level to drive the gross margins to where want to be. That's going to be multiple, multiple quarters before we get those to be foundry average gross margins. But it's tracking better than we expected, which is good. I think we have focused a lot on yields.

We -- Lip-Bu brought in a lot of talent into that space to really focus on it. and we brought in external partners that are particularly good at -- including metrology that's helped us execute better there, and we're starting to see the benefits of that this quarter.

Srini, do you have a follow-up question?

Yes, John. And then on the ASIC business, Dave, I think you said it doubled year-on-year. If you could maybe help us with what is included in that? I believe it's IPU, I just want to get a better sense how big it is. And as we look out to the next few years, what's the strategy for that business to grow? I mean, are you going after the classic ASIC in terms of XPUs or is this more some of the adjacencies like what you're doing with IPU? Any color would be would be helpful.

Thank you, Srini. I think it's a good question. So I think this ASIC business, sometimes I call it the purpose built silicon optimized for specific workload customer want and that's very important. I spend most of my time besides our focus on engineering improvement, is spending a lot of time with customers. So it's very important to understand that the workload and then how do we drive the workload and then to purposely tailor for their requirements. And clearly, it's very important to really have the right strong IP portfolio to be able to do that. And also, we are a very unique position. We have the CPU, XPU, and then we also have the advanced packaging and also advanced processing so that we can really optimize for the customer That, I think, is a very exciting opportunity. It's a fast-growing area. And then clearly, we have something unique to offer. And we are already engaging with a couple of customers the feedback from the customer is very excited, very positive to work with us. And I think stay tuned on that 1. The next 5 years going to be a fast growing for us.

I think 1 thing that people have been surprised about is how big the business is already. It's at a run rate that's north of $1 billion already. And I think Lip-Bu and his partner Srini have barely gotten started in terms of what we can make from that. So it's got a really strong base in which to grow meaningfully from.

Our next question comes from the line of Joshua Buchalter from TD Cowen.

Congrats on the very strong numbers. Actually, I want to follow up on the next question from earlier. You gave some metrics on the near-term CPU TAM I think investors are directionally really struggling with how to model the CPU demand for agentic workloads. Any help you can provide us longer term about how we should think about growth, whether in terms of units, cores, gigawatts CapEx? Just anything you can give us? I mean [indiscernible] is the $100 billion number that ARM gave reasonable in your view?

Yes. Maybe I'll start and Lip-Bu can pile on here. I think 1 statistic that we look at is the ratio of CPUs to GPUs. And if you look at training solutions, they're generally running in the kind of to GPUs to 1 CPU. As we look into inference, it's probably getting into like the 3 to 4:1 kind of level. And as you get into Agentic and multi-agent, it's one, potentially even flip in the other direction a little bit. So that's 1 way to think about it. As you think about the growth rate now going forward, it's going to become a significant part of the AI TAM. Keep in mind also, I would just say data center, obviously, is where we're focusing a lot of our conversation on. But there is going to be AI and particularly CPU opportunities in a lot of different areas. The client space is, of course, as I already said, migrating towards AI PC in a more meaningful way. there's edge computing, there's physical AI, specifically, all of those, I think, can benefit a lot from CPUs because of the nature of the power consumption relative to the performance. And so those areas could have even more explosive growth than the data center space. .

Yes. Just to add on to it, I think you have to think about the full stack. And so you address some of this agentic AI and later on, the physical AI, how the CPU optimize working together with the foundation model how to optimize that and using the data to really drive the massive opportunity in the agenetic AI, I think inference is going to be a much bigger market. And the physical AI is another big market. So I think that is an opportunity for us. And it's hard to quantify. But I think as you go, we will update you on that. .

Josh, do you have a quick follow-up?

Yes, sure. And then as we think about your capacity tightness, the leading-edge foundries are also quite tight as well. Has this driven any near to medium-term share gains? And longer term, how important is your captive capacity to winning business with customers on a multiyear basis?

I mean, obviously, all the supply right now or the lion's share of the supply is all internal. But we do expect, obviously, to win customers over time. I think I don't know if there's anything else you want to add on. Yes.

Then our final question for today comes from the line of Aaron Rakers from Wells Fargo.

I want to go down the path of the supply side as well. I think in prior quarter's discussion was that you were reallocating maybe some supply of wafers from [indiscernible] data center. And I think the notion was that maybe 1Q would kind of be the peak degree of constraints. As you rolled up your guidance for this current quarter, I'm curious of how you would frame the level of constraints that you see in the guidance this quarter and maybe does the back half improve that dramatically?

Supply will go up in the second quarter. It's going to go up every quarter now going forward. I would say we certainly were in our lowest point in terms of supply probably in the first quarter relative to the rest of the year. But what we were able to do in the first quarter was go through finished goods inventory and find opportunities to sell product. We didn't think we would be able to move. It was either despeced product or it was legacy product we had shelved and then worked with customers and found opportunities for them to leverage that technology in their system. So that helped out a lot. I'm not sure we have that benefit in the second quarter. So obviously, we will scrutinize our finished goods inventory to see if we can find some opportunities. But for the most part, what we're relying on from a volume growth perspective, Q2 versus Q1 is going to be increasing supply. .

Aaron, do you have a quick follow-up?

I do. I guess as a follow-up, I know you talked about kind of the path of the server CPU and the competitive dynamics there. And I can appreciate that you're not giving kind of descriptive timing of kind of road maps for future generation products. But how do we think about like the progression of your CPU on the server side, Xeon to Diamond Rapids, Core Rapids and really kind of closing that gap with simultaneous multi [indiscernible]. Just any kind of color on the cadence of the road map would be helpful. .

Thank you. I think clearly that the demand is strong. And that we fine-tune the road map. I think clearly, we highlight the Diamond Rapid after the Granite Rapid we have. And then the Colorado Rapid is the next 1 that we're multi-trading. And then -- so I think that is our road map. We're laser focused on execution. And then meanwhile, we're going to use our ASIC business to really drive some of the customer requirement. We can built some purpose built silicon in the short term. And that is a huge opportunity for us. We have all the unique assets that we can really provide. So I think those are kind of road map and execution. I think year 2026, I call it the year of execution. We are improving the execution. Back to Dave mentioned about the yield, productivity, the cycle time, make sure that supply chain, we can catch up with the demand. .

With that, I would like to thank everyone again for joining us today. It has been the eventful first year for me at Intel. It is gratifying to see our progress even as we know we have a lot more to do. I'm looking forward to see many of you at the JPMorgan conference in May and at Computec in June.

Thank you, ladies and gentlemen, for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.

Great. Welcome back, everybody. I'm Joe Moore, Morgan Stanley Semiconductor Research. Very happy to have with us today, David Zinsner, CFO of Intel.

Before we begin, on behalf of Intel, please note that today's discussion may contain forward-looking statements that are subject to various risks and uncertainties and may reference non-GAAP financial measures. Please refer to Intel's most recent earnings release and annual report on Form 10-K and other filings with the SEC for more information on the risk factors that could cause actual results to differ materially and additional information on non-GAAP financial measures, including reconciliations, where appropriate, to the corresponding GAAP financial measures.

That was awesome.

Yes. Welcome back.

Thanks.

So I guess it's been a little more than a year since the CEO transition. And it seems like there's a lot of style differences and a little bit of substance in terms of the way you're approaching some of these opportunities, but still basically sort of a very similar strategy to where you've been. So can you just talk about the big picture? Is this major course corrections or small changes?

Yes, I think you've got it right. I think mainly it's course correction. Although I would just say Lip-Bu is pretty focused on making sure we're not getting ahead of things in terms of our investment. So he wants to see real signals that we can count on from a demand perspective before we're putting investments in place. And you clearly saw that in the last couple of calls with investors.

In addition, what he's done is he's really simplified the organization in a lot of ways, reduce -- obviously, we reduced the headcount, but we also significantly reduced the layers within the organization. So he is closer to the lowest levels. And I think that's better. Decision-making is obviously better, but he also has more access to what's going on in the field, and that's helped out a lot.

The third thing that he's done that has been, I think, really helpful is Intel, I think, traditionally has been a company that kept everything within the four walls of Intel, did not expose a lot of data and information to partners. And of course, you do that because you want to maintain some confidentiality around things that are proprietary. But what you lose is the ability for those partners to help you improve the -- whatever, in the case that I'm talking about right now, more like the process.

And so he has opened up a lot of data to our partners. And I think that has really been one of the key factors in getting yields to start to improve meaningfully on 18A. We've got several partners that help us in terms of improving yields and yet they were dealing with like little to no data. And so that -- I think that was a big improvement he's made.

He also -- obviously, since he has spent a lot of time investing in the space and in particular, more recently investing in the AI space, he has a unique view in what AI workloads look like across the ecosystem and what we could do that could be helpful that isn't directly competing with those that are entrenched competitors, but really things that are orthogonal or that really would make a big difference that really only Intel could do. And so our AI strategy, I think, is forming in a way that perhaps is significantly different than it might have without him.

Interesting. Yes, it seems like you've put a lot of focus back on the core business a little bit with AI is kind of more in the future, Foundry is a little bit more in the future and you're really focused on kind of the core product...

Well, there -- I mean, we clearly have work to do to get our product portfolio to where it needs to be. We've got work to do to get our processes to where they need to be. So that is the most significant thing that we need to accomplish. And Lip-Bu is very focused on that. Like I said, he's very into the details, very -- he has kind of a style of kind of open communication that I think has allowed us to get more insight into where are the issues and how we can improve them.

Great. And on the topic of process technology, Panther Lake being out on 18A, looking like a good part, seems like a really important proof point after the sort of 5 nodes.

Yes, for sure. And like I said, there's kind of a glide path you kind of move along -- that you would expect to move along from a yield perspective. And we're now at or even slightly better that glide path across our 18A process. So things look good. We're also -- there's a lot of volatility, even though you have a yield, like some wafers are yielding a lot less and some are yielding a lot more. He's actually focused a lot on trying to minimize the volatility wafer to wafer, and we've made good improvement there. So the process looks good. And I think we would expect a pretty steady yield progression as we go through this year, probably a bit ahead of schedule, quite honestly.

And then Panther Lake as a part has been well received, obviously, particularly around battery life in particular. And so demand has been robust there. In fact, I think our bigger challenge is making sure we get all the supply to the customers. But that's a great proof point to customers -- external customers that 18A is a good process. And while Lip-Bu was, I think, thinking that we probably should focus on 14A as a foundry node and make 18A really just an internal node, now that we've got -- seen some real progress there, I think he's now starting to recognize that this is actually a good node to offer to external customers as well. And we've been getting some kind of inbound interest in 18A-P as a foundry node. So that's -- I think that's pretty positive.

Great to see that progress with Panther Lake. So maybe talking about the demand side, start with servers. And I know it's tricky because you've had some supply challenges, and so it's always hard to assess demand. But it seems like there's a growing consensus that CPUs are really benefiting from AI and that agentic things, once they're crafted with GPUs, the agents are run on CPUs and there's a lot of workloads. So how durable are you seeing that incremental demand?

Yes. I think it's -- I think that's -- the CPU has become cool again this year. And we've long believed that CPUs needed to kind of stay along with the GPUs in these data centers. And yet like a lot of the spending was -- had moved towards GPUs and CPU on a unit basis was even coming down. So -- but we knew...

Lisa Su may have used the exact same words...

What's that?

Lisa Su said CPUs are cool again.

Really. So anyway, so I think that this back half of the year of last year, where we really started to see the pickup in demand really showed that, hey, actually, as you move into, like you said, agentic, as we're looking at better orchestration away from just running the LLM, into the orchestration aspects, all of that has to be run on CPUs. And we're seeing the benefits now of that.

So units are up -- I think units were up last year, something like -- from a TAM perspective, something like mid-20s percent year-over-year. And I think it's going to be up again pretty meaningfully this year. We're starting to see customers come in, in that space asking for long-term agreements, that should tell you that there's legs to this. They're looking at this over a 3- to 5-year basis and want to lock in supply with us.

And your strongest position in server tends to be with more enterprise-centric cloud -- enterprise-centric clouds or enterprise-centric on-prem. But are you seeing also demand side from the traditional kind of cloud hyperscalers? And how do you sort of juxtapose that with ARM starting to encroach and...

Yes. Yes, across the board, the demand is strong. I guess I would say x86 is a really strong ecosystem. And the customers and those -- the users on the other side of this generally have spent a long time maturing their networks with an x86 architecture. So I feel very good about our opportunities. There'll be other competitive technologies that make progress, I'm sure, mainly because the demand is going to be so high and supply will be constrained. But I feel really good about the x86 position. And in fact, when -- if you look at the recent announcement we had with NVIDIA to partner between our CPUs and GPUs, that really was, in a lot of ways, an endorsement around the x86 ecosystem.

Yes. Maybe if you could talk about that deal. It seems really important to NVIDIA as well. They've talked about bringing rack scale to ecosystems that aren't comfortable with ARM, and that's a big part of why they're doing that deal. I mean, how big an opportunity is that for you guys? [ Also, ] sort of focused on that is like, well, there's definitely a foundry thing. It seems like it's really around the...

No, no. This was a really -- this was a product.

CEOs said it but...

Yes, it's a product-driven engagement between the two CEOs. It was, as you say, a lot of customers looking for an x86 solution. And I think NVIDIA wanted to be able to offer that. Of course, for us, it's great to marry our CPUs with their GPUs. They're the best in the business for sure. And so that opportunity, I think, gives us a pretty good option to be able to capture, I think, meaningful growth from that business. And it's both a data center and a client product that we're talking about. Obviously, it's a couple of years out before those products come to market. But so far, the progress has been great.

Yes. Yes. Okay. Great. Competitiveness within server, Lip-Bu, I think, has actually done a good job of bringing a lot of humility to those conversations, which -- and I mean that sincerely that I think building credibility with the channel is really important and the cloud guys, but it's also sort of been open that Diamond Rapids didn't do everything that he wanted it to do, and we need Copper (sic) [ Coral ] Rapids next year to sort of retake share. Does any of that matter right now? I mean you're coming out of such a supply-constrained situation that you should have a nice rebound as we get through that. How do you think about market share in the server...

Yes. I mean I think from a from the perspective of what we can capture, supply is going to be our greater challenge certainly this year and probably even next year. But that said, we want to bring out products we're proud of. And I think Lip-Bu, when he came in, he looked at the Diamond Rapids road map, didn't like that it didn't -- wasn't capable of multi-threading. And there were reasons those decisions were made. But Lip-Bu, I think, spends a ton -- this is probably another thing that I think he's really good at is listening to customers and taking that back and developing a road map based on that. And he felt really strongly that we needed to be able to provide that. So we'll have this hole in Diamond Rapids, but it's actually Coral Rapids, not Copper Rapids.

I'm sorry. Too many rapids, I guess.

No, kidding. Coral Rapids will -- we're going to look to try to bring that in as early as we possibly can. There's only so much we can do about that. But in the meantime, there are parts -- it's not like everything shifts over to a given product anyway. We're still selling some of our products that we've introduced 5 years ago. So we'll have plenty of opportunity to, I think, address the market with our portfolio.

Great. And so then you talked about the supply constraints, which have been severe. Can you -- I know you've gone through this a lot, but just help us understand why they're so severe right now when the growth is not that high. And then it seems like Q1 is the worst of your undershipping relative to demand and improves from here. Can you talk about that path?

Well, I mean, first, we kind of -- we build our plans based on the demand we get from customers. And we started this, obviously, as others did see the signals in the back half of last year that demand was going to be stronger. The challenge is the lead times. By the time you start wafers and get those through, it's a couple of quarters before those products come out. And so there's a time lag to it.

In the meantime, we were operating off of inventory that we had, but we've leaned that inventory out to the point where the first quarter really, there isn't much in the way of finished goods. Lots of WIP on the inventory level, but not much in the way of finished goods to be able to support that demand. So as we come out into the second quarter, things will improve.

I would say wafers is part of the challenge around supply, but the industry is suffering shortages all over the place. So memory, of course, you know about; substrates, they're short; T-glass, it's short. I mean there's a lot of things that are short across the board that all have to get caught up as we progress through the year. But we feel pretty good that we'll be able to bring more and more supply online every quarter and things should have a relatively rapid improvement as we get through the year.

But you do feel like you're supply constrained for a while?

Certainly this year. Yes.

Okay. But that view that Q1 is the kind of worst under-shipment, that's [ still the view? ]

Yes.

Okay. And then with regards to clients.

By the way, maybe one other thing I'd just say is it's not like we're sitting on our hands. We're operating these fabs at above 100% in terms of what they're spec-ed to supply. But there's an ability to get it even more above 100% if we can, and we're -- and that's what we're focused on is [ squeeze out ] every wafer we can from the system this quarter.

I mean there was a bit of a transition where it seemed like Intel 7 was kind of coming down and then it became -- Intel 7 became the binding constraint for a lot of your business. Is that part of this as well that you sort of have -- there's more demand on the Intel 7?

Yes. I mean there's demand on -- I'd say things are tight across all of our foundries and our processes. And keep in mind, we're ramping two processes at one time basically, which is not typical for Intel. We have Intel 3 and 18A both ramping and trying to catch up on the yields and so forth. So that clearly is driving some of it.

That said, Raptor Lake, for example, which is an Intel 7, 10 process has great demand on it. And you got to play the algebra here of what -- how much capacity you want to add on, on that node, given that it's an older node and you're really trying to migrate customers into Intel 3 and 18A. So we're kind of managing the shell game there.

It will probably -- as we've talked about, we are going to be required to increase our wafer capacity on Intel 10, 7, but we're going to try to push customers also to take products that are on those newer nodes to smooth things out. The other thing is the more we can drive Panther Lake on 18A since it's only client, that does free up the other nodes a bit. So obviously, we have more demand than we have supply on Panther Lake, but the more we can ramp that, the better we will be across all nodes.

Great. And on the topic of client, I mean, you guys did talk about memory being something of a constraint to market growth this year. That seemed like at the time, it was kind of more applying common sense to what's happening rather than something that you're seeing. But like just can you update us on that? Does memory continue to be a concern for end demand?

Yes. I think -- my expectation is that memory is going to be short all through this year and probably all through next year. And given the importance of memory, into data center and in particular, into AI workloads, it is going to create a more significant constraint in the client space for sure. And so we talked about this on the call, our expectation is that in the back half of the year, they're going to suffer probably from the lack of memory.

And so we've built that into how we're planning the year and what capacity we want to bring online on given products. And we talked that we were going to kind give up some of the small core in the client space, push into the mid and high end of the client space, but then try to push as much as we can into data center to make sure we can alleviate the constraints there.

Okay. Okay. So you're sort of working through these constraints, but they're still a factor, probably...

Still a factor.

In all the businesses because you're going to build less for the client business as well.

Correct. Yes.

Okay. And then I want to get to the external foundry prospects. But just if we think about the foundry losses that are kind of weighing on you, a lot of the path to breakeven is not external foundries. A lot of that path to breakeven is just internal. Can you update us on that? Like how do we get those losses down? And if it's just because you're charging this product group more for wafers, are we sure we're going to see the improvement that, that implies?

So it's -- there's a meaningful improvement in margins in '26 versus '25 for foundry. We -- part of that is that we had a lot of like start-up costs that we were running through the P&L. And now we're at a point where we're just ramping like kind of fabs that we've built out. And so our expectation is that we should see some pretty good improvement in what we call our other cost of sales line, this period cost line, where a lot of those start-up costs show. So that should help meaningfully in terms of gross margins for foundry.

On top of that, as you point out, the newer processes, yes, they get a better ASP, but mostly it's around their cost structure is better for their performance. And so the notion is that, yes, they'll be charging more for those wafers, but because it's working -- to charge more for those wafers and the product company part of the business gets that benefit. So that will also help a lot in terms of margin.

Now obviously, we're in the really early innings of 18A ramping in the fab. So those margins are negative right now. And as they become more of the mix, it actually pulls the margins down a bit. But as we progress through this year, certainly as we go into next year, those margins get better and better, which will also help.

So I think we still feel good about our original guidance, which is we expect to exit '27 at breakeven operating margins for foundry. The only caveat to that would be we're planning a certain level of external foundry wins in that number. If it's stronger, it actually could count against us on the profitability side because we need to invest more, which would push the profitability out a bit, but it will be a good problem to have because that means we're going to have more demand in the out years beyond that.

So on that kind of process, you sort of -- it seems like you're still committed to the potential of being a major foundry supplier, but you're not going to spend the money until you have the customer commitments. Is that a fair statement? And where does all that stand?

Yes. So to be clear, we are spending the money on all the R&D development especially with 14A, and that includes the CapEx investment for the R&D in 14A. So there's a pilot line that's got to be put in place. We're investing in all of that. What we're holding back on is the high volume and when to put that in place based on what we do in terms of customer wins on 14A.

The engagements have been good. So I think we're cautiously optimistic that this will be a successful node. We also have internal demand on 14A. So even still, I think we've got to think about that as part of the CapEx cycle. In addition, we have seen more engagement from external customers on 18A-P as well. So we're starting to see some demand there, which also could go hand-in-hand with this. So the likelihood is we'll start to see some of the stuff fall into place in the back half of this year, maybe some of it into early next year. And then we're going to have to make an assessment as to what that means from a CapEx perspective.

And the timing has always sort of been risk production, '28; volume, '29. I know there was some confusion maybe on the call, but that's still...

That's still the case. Now that's more a function of what customers want. We can pull it in. For our internal demand there, we have the ability to go risk production in '27, which is likely what we're going to do. So I guess if a customer wanted to do that at the same time frame, we could do that.

And risk production can have useful output...

But right now, from what we hear, when customers -- remember, they got to spend a lot of money when they're moving to a new node. And so they're going to want to make sure that they're layering that into their product road map in a way that generates the best ROI. And when they do that, that seems to suggest that it's more a '28 to '29 time frame that they really need these kind of wafers.

But there was no delta around the quarter, like that...

We haven't changed it. And in fact, when you look at -- we plot out yield and performance even as early as now on 14A, and we look at it relative to where it was 18A at this time frame, where was Intel 3 at this time frame. And we're actually ahead. So good news is that actually, we could be seeing better -- one would hope, better performance for yield than we had in the prior node.

So it kind of comes back to the execution is good, but there's more like [ cumulity ] about time frame.

For sure. And we've got to make sure that these customers have unique requirements that go along above and beyond just the process in terms of deal terms and so forth. So we've got all that pulled together to make it work.

Can you talk about EMIB and the advanced packaging technology as a Foundry offering? It seems like that's a lot bigger than you originally thought it was.

Yes. So good point. Ironically, this is probably the more interesting part of the Foundry business today, quite honestly. There's -- of course, there's supply shortages, but EMIB is actually great from a technology perspective relative to what's out there competitively. You can get 30% more reticle space on EMIB or EMIB-T. So this looks like a great offering for us, and we've gotten, I think, really good engagement from customers on this business.

So originally, when I was thinking about it and talking to investors, I was calibrating everybody to, hey, this is like a -- think about these wins in the hundreds of millions versus wafer wins, which would be in the billions. That's the way you should think about it. And I've since revised that because we're actually at the close to closing some deals that are in the billions of dollars per year in terms of revenue on packaging. So lo and behold, this is going to be, I think, a really good business for us. And...

And [ size, ] we're clearly talking about AI ASICs...

Yes, of course. And that's what's driving a lot of this because of the advanced packaging, this is what makes this so interesting in terms of an offering. And I think we're likely to see some announcement potentially even before that back half of the year in this part of our business.

The gross margins for the overall Foundry, we were targeting at about 40%. That's kind of where we want to have our run rate gross margins land. And I think a lot of people think about, okay, wafers get better margins than packaging, and that's the way we should think about it. In reality, of course, there's -- you got to get the technology up and you got to get the scale up and so forth. But when it's running at kind of normal state, the margins in this business should be just as good as the wafer margins.

Okay. Okay. And you used to have some sort of lower end, lower-margin packaging business that...

We did. That was a little bit like COVID-driven. And we weren't -- that wasn't a business we were looking to get in. We did a solid for some customers. But this stuff has got real capabilities and real advantages to customers that they're willing to pay for.

And billions of dollars in what time frame?

Could be out in that same time frame that I talked about, potentially a little bit earlier than the wafer business.

Okay. Cool. Gross margins overall, it seems like you have a lot of things going in the right direction over the course of the year to the extent that revenue can rebound, you get through these capacity constraints. You talked about start-up costs coming down. Just any update on thoughts on gross margin?

Yes, I think we should expect to see gross margins improve through the year. And my current thing is, we got to get gross margins to start with a 4. And once we're there, we can talk about where the gross margins go after that. But there's no reason when you look at these businesses, if they're competitive, if the cost structure is in the right place, with our margin stacking advantage, there's no reason why the margins shouldn't be significantly higher than they are.

But we've got near-term issues or -- near-term headwinds that are because we're driving so many nodes through at the same time. We've got a few products that aren't at a competitive cost structure at this point that we've got -- that we have to improve. So -- but I feel pretty good about -- I mean, I think that's one thing -- another thing that Lip-Bu has really brought is a lot more focus on balancing, bringing out products that have performance with products that actually have a cost structure that can be competitive as well.

And so you'll see that in the road map as we progress. Wildcat Lake has got a great cost structure. That's part of the Panther Lake offering. Nova Lake's cost structure is significantly better than prior products on the client side. And then on the data center side, I think if we look at the Coral Rapids, you'll start to see a much different cost structure for that, that will help drive better profitability.

Okay. And that time frame, Core Rapids is more second half next year?

Core Rapids is out beyond this year. So I don't -- this year, it's about kind of just getting Panther Lake up in terms of yields. Wildcat Lake is margin accretive. So that should help out meaningfully as well. Obviously, the revenue drives -- we get fall-through, so the revenue improves, gross margins as well.

We have a few things that kind of will work against us. We're selling a lot of Raptor Lake and some of the older server products that had better cost structures. So we'll -- our mix probably won't be as good this year, which will be the headwind to gross margins. But beyond this year, I think that starts to improve as well.

Okay. And then puts and takes with regards to CapEx, given everything that's going on. You have a decent amount of tool spending this year. You have a lot of shelf space if you need it. How do you think about all of that?

Yes. I mean exactly, we guided flat to down, call it, flattish in terms of capital spend. You had a pretty -- we have a pretty big step off of spend on the clean room side, the space side, and we're actually increasing our tool spend this year versus last year, not surprisingly. We've got all the space, now we're tooling it out.

I think we're pretty locked in, quite honestly. I mean most of the POs had to be put in place already to drive the CapEx dollars for this year. Depending on when tools turn on and so forth might influence a little bit the CapEx. But for the most part, I think we have pretty good line of sight where CapEx will land this year.

Right now, we're in the throes of trying to figure out, okay, now what does the CPU demand that we're seeing on the data center side look like over a multiyear period? What can we expect from a client perspective? And then what sort of wins are we going to get on the Foundry side that we'll layer all that in and build out a CapEx plan over the next few years that manages the capacity.

Great. So I have one more question, and then I'll open it to the audience. So OpEx, I think you talked about a $16 billion a year. Where are you investing more from an R&D perspective? What are the focus areas?

I mean, not surprisingly, Lip-Bu kind of went back to look at -- make sure we got the core working appropriately. So a lot of the allocation is to the core products and data center and client, making sure that they're performance driven, have the right cost structure. We are also obviously investing in the AI space in solutions that we think can be unique and competitive in the marketplace.

So that means kind of finding an area where you're sort of competitive rather than trying to take on NVIDIA directly?

Exactly. And we talked about the fact that we have our own now ASIC business, which actually was -- we had actually a set of products that were ASIC products. We pulled together and kind of built the business around that. That business is going to grow significantly this year versus...

Yes, you talked about $250 million a quarter. Is that more of the -- traditionally, I think of you guys that's around the comm infrastructure space and stuff like that.

Comm infrastructure is part of it, but we have IPUs selling into the hyperscalers that I think is going to be a good product this year. So I mean, we have a foundation. Now is it down the fairway directed at AI? Not yet. I mean we still have to work on building out the IP, make sure we understand what customers really want us to do, and building products that solve those problems. But I kind of like where that business is going to go just in the year that we've kind of reformed it.

Great. Well, let me see if we have questions from the audience.

If not, maybe we could just touch on the Board changes last night. It seems like you've sort of developed a Board that has a lot -- over time, a lot more semiconductor expertise than what you had 2 or 3 years ago. Can you just talk about the change and anything that, that may signal to us?

Yes. So our most recent hire into the Board, Craig Barrett, obviously, well known in the semiconductor space. and brings a significant technical capability along with -- I mean I've worked with him now for a few months, and it's been pretty amazing how the depth of his understanding about the markets and so forth. So I think that was a great addition.

Frank, the Chairman is stepping down, been at the company for 17 years, led a lot of changes that needed to happen. So it's not -- I think after 17 years, not unusual that somebody would like to hang up the saddle there. And excited for Craig to come on to be the Chairman. I think he's going to be great. He had a strong relationship with Lip-Bu in the past. So that's also, I think, helpful that they kind of see the world in a very similar way and I think can work together to make the company successful.

Yes. That's great. We have a question up in front. Can you wait for the mic.

In addition to the yield comments you've made, you've talked about increasing throughput. And could you elaborate on that and what the opportunity is there?

Yes. So -- and there's actually two parts to the throughput. There's a front-end and a back-end kind of aspect of the throughput. As you might imagine, when you're trying to like lock yields down and get on a kind of a steady path, you somewhat give up on the throughput a little bit to make sure that you're delivering yield. So our days per mask layer is kind of how we measure it, has kind of eked up a little bit just to get the yields to start to follow along the right progression.

Now that we're there, I think looking at days per mask layer and trying to shrink that, I think, makes a lot of sense. When you look at our number competitively and even year-over-year, it's not where it should be. So there's certainly clearly opportunity there. And I think that's one of the ways, hopefully, we can improve the supplies as we progress through the year is just by improving the throughput.

The other -- so that's in the front end. The back end also, quite honestly, isn't at peak performance. as we build out what we need across our network, we have maybe allowed some of that to atrophy a little bit. So Naga and Lip-Bu are very focused on how we can improve the back-end cycle times to get product out once -- through the wafer portion more expeditiously.

So I think we had another question. Really quick one, if you could.

Yes, really quick one. Sorry. Perhaps I was a little late into the presentation. Perhaps you have already commented on that. Could you please comment a little bit about your 14A process and especially when do you expect the PDK to be out?

So we have obviously less mature versions of 14A out today. Customers are working with it using -- running test chips through 14A. Our expectation as we get towards the end of the year, we'll be at a maturity level for 14A that we'll get actual customer yay or nay on the process, and we'll go from there.

All right. We're out of time. Thanks so much.

All right. Take care. Bye.

You have a very unique record in my life. Do you know what that is? The speed at which I have developed a friendship with you, I don't think I've developed with anyone else. And it's not a function of me. I think it's a function of just how good you are at doing the kind amount of things that you do on a regular basis, but, thank you for coming here.

Thank you for inviting.

Tell us a little bit about Intel is a national Treasurer. Intel has gone through some tough times. You are bringing it back again to life, but there's so much that needs to get done. Walk us through the current state of Intel. Where do you think it is? Where do you think it should go? How should people think about it?

Yes. I think, first of all, I joined the Board for 2 years and learned somethings going on, a lot of good friend of mine from Intel. And it took me a while to decide to take this role. And I say, no, that's not easy. A lot of my friend told me that don't do it. And we have a great reputation in the venture investment that you love it and to take on this one.

But finally, I decided this is an iconic company. It's very important for the industry and also for United States. So I finally convinced my wife, let me do it one more. And after cadence and he supported me to do it one more time. And just to update, I think I just passed 10 months coming to 11 months.

And I kind of call it matching off the map. So something that some of the surprises, I don't even know you have to kind of reacting at a real-time basis and learn from it. So in some way, I think it's good to have. And then just update. And we have a very complex company. You have a foundry that's strategically very important. And then the other part is the product. And then -- so clearly, you need to balance the two. And then the other part is.

Do you believe your foundry business becomes a general purpose foundry that's outside of just your products?

Yes. And so that's the intent. And we have the 18A. And when I took over, the yield is quite poor. And so I really get all my friends to help from PDF solution to KLA in the equipment to make sure that we have that 7% to 8% yield improvement per month.

8% of yield improvement per month.

That's right. And that will be the best practice. And it took me a while. Finally, now I seeing 7% or 8% yield per month. So in a way, it's very helpful that you open up the kimono and let people come and help you. And then -- so I think one is 18A. We just announced the Penta Lake, and then I can really count on my fab to produce that. And then interesting enough, a couple of customers are knocking on my door now and say, hey, it seems like your 18A is doing well. We want to be part of that.

So I think we are delighted to see the customer want to see that, but we are laser-focused on the 14A. And that will be -- it's very important. It's 1.4, and this is the most advanced. And we're going to be in the risk production 2028 and then we'll be in volume production '29. And like any foundry, I learned a lot, and you need to have not only the yield improvement, you need to have the variation, so they're more predictable. And you also need to have the final product that what is the yield look like and the spin and all the detail.

And second part, you need to have all the IP available. If you want to have mobile customer, you got to have all the low-power IP to have. Otherwise, you can't serve the customer. So that's something that is missing in the past. So I really put something in place and so that we can have both and then serving the customer. Good news is a couple of customers are very excited. 0.5 PDK this month, we will have that and the customer can use a test chip to work on us. And so I think good news is a couple of customers are engaging heavily, and we're looking forward to serve them. Hopefully, second half of this year, I will see the volume commitment from the customer, which product they want to have, and I can serve them.

On second half, you'll see the volume commitments on the foundry or on your core products?

On the customer. So in the way, they need to let me know what products and what volume they need. Then I can -- and one thing to -- people will see because, first of all, I'm not going to announce any customer. This is confidentiality. We need to support the customer. And what we're going to do is when you see me starting to put money into subscript is a very important material. And then secondly, some of the CapEx equipment I need to have and scaling it. I mean that I have real customer I've committed to it. That's kind of discipline I have.

So on the fab side, to get to mass scale, what does mass scale look like for you? And what's the time frame that you feel like mass scale would happen where you could be a formidable fab for third-party chips?

Yes. So it's very challenging to operate the 2 because one is on the product. You have to drive innovation, drive the product road map. The other part is really a service business. Culture has to change. And in some way, I kind of tell my team, this is a business of grinding. It's not like you have good innovation, good technology, you people will come.

You have to grind until something that you prove the yield, they can count on you to make the production and then their whole revenue, profit depend on it. There's a lot of trust and responsibility, and we have to earn it. So I think those are the things that scaling is important. And that's why I always ask customer, give me the biggest product that you have, the most important product. give me 5%, 10%, 20%, 50%, let me earn it and earn your trust. And then many of them, I work with them when I was running Cadence, so they trust me. So somewhere that I think back to your point, trust is very important for AI. So I think something that even more important in foundry is the customer trust.

Now if you think about the constraints that are there right now, do you think that the energy constraint is with power, cooling, all of that is more severe or right now is manufacturing capacity, which one limits growth first?

I think in terms of the AI, the biggest challenge, I think, for a lot of my customer is memory. Memory, actually, there's no relief as far as I know, when I talk to only 3 key players, 2 of them I talked to very frequently. And then they told me that Lip-Bu, there's no relief until 2028.

Why is that?

Because I think this whole AI is suck up a lot of memory. Our mutual friend, Jensen tonight, you talk to him. And clearly, he needs a lot of memory for his login and the next-generation product. And so I think it's very important to have that memory. If anything going to slow down going to be the memory. So I think is one.

Secondly, I think it is clearly from the compute side, I was very happy to hear that customers are all crying for more products, and I didn't prepare the production enough to meet their requirement. And so I think people starting to find out that in application, CPU actually is more useful in terms of performance for all the compute requirement. And in fact, a friend of mine, he mentioned that Moore's Law double used to be 3 to 4 years, and now it's like 3 or 4 months. So the increase of compute is increasing so much. And right now, my biggest challenge is focus on our production, our supply chain, make sure we can meet the requirement.

Almost every CEO, they call me, Lip-Bu, can I have more? I'm your friend. I'm your customer, the most important customer, I want to have more of that. So I think somehow it's kind of encouraging for me to see that compute has become very important. And then the other part, very important to me is beside memory, beside the compute and then the other part is the thermal we just talked about, the cooling technology.

So in terms of some of the high-performance processor, either GPU or CPU, sometimes you have to guide the gigahertz down because your thermal issue and power management. So those are the ones that you want to find new way of cooling in terms of liquid cooling, air cooling is not cutting it now. So liquid cooling, micro cooling, emergent cooling. So I think those become very important. The other part that you also mentioned, I think something that Cisco has is really good is that interconnect is getting become very critical. It used to be more in the copper and like Kratos Semiconductor and Astera lab by investing, but now moving to optical...

And so optical become very critical. And so that's something that is a new wave because the speed and the density become very important. So that's another issue. The other part of the constraint is the software. And then so if you really want to address some of this like the Moore's Law, you need to really look at the whole full stack. So not just the silicon and you need the cooling and then you need to have interconnect and then the other part is the software. The open source, I'm very strong belief in it.

And then, of course, the crude compatibility is very important. And even more important, right now, it's a cluster of GPU or CPU and then it becomes the clustered manager, the software. And right now, sometimes if we have a problem, you don't even know where to look for it. And now there's a lot of new start-up coming to me, they said, Li-Bu, that will be an area that we want to work on so that we know exactly how to solve some of the clustered issue that you're going to face. Kubernetes is great, but you cannot find the real issue there.

So I think those are kind of interesting being a venture capitalist and running CEO of Intel, it gives me that kind of perspective of what are the new things coming and quantum computing is another big area coming up is around the corner. And so something that is interesting to see after AI, we talk about agentic AI, the next big wave is the physical. The next thing is going to be quantum. So I think that's kind of exciting.

And so you talked about open source and talk to us a little bit about like your perspective on open source in the United States? How important is it? What needs to happen?

Yes, very good questions. I'm very passionate about the education. And I'm involved with MIT, CMU and Berkeley and others. The key part is one thing that I worry a lot is the fundamental research we're starting to not continue. A lot of university, the best professor being recruited out to Asia to Europe.

And so I think the fundamental research, some of the public companies like Cisco and ourselves, we have some short-term midterm requirement. You cannot invest into long-term 10 years, 20 years from now. And so that part, to me, is an outcry on foundation research. The other part is the open source. I'm a big fan of open source. And frankly speaking, a couple of people that good -- very knowledgeable people that told me, Lip-Bu, we are behind China now. And so that's something that we have to pay attention. I think DeepSeek is just a wake-up call.

But don't you think like to some degree, some of the models that are emerging in China are just pure distillations or the models that are here? And so at some point in time, if all you're doing is distilling the models, then the moment -- they can't really get that much further ahead. Or you don't think that's the case. They'll actually figure out a way to start creating original?

Yes, very good question. I thought that initially, I thought that they will be falling behind because they don't have the access to the most advanced GPU from NVIDIA, most advanced processor from many different players. But you will be surprised. And recently, I was just try to recruit some of the top talent in CPU architect. And I found out that Huawei have 100 CPU architect, top notch. I was shocked.

And so when I talk to them, why are you going there? And they said, even though we don't have access to the best tool like EDA tool from Cadence and Synopsys, but we have the pan way to do it, and we can do it. And I said what about ASML equipment that you guys don't have. But I said well, we are quietly building on it. And so to me, is they are just shortly behind us. And then if you're not careful, they just -- they ahead of us.

And what they've done is they've spent the calories on infrastructure optimization because they don't have the right chipset?

Correct.

And so like if they are at 7-nanometer and we are 2-nanometer, what they're offsetting that with -- and they have unlimited power and they've got engineering capacity that actually creates the caloric redirection towards making sure that you you drive a certain level of infrastructure optimization.

That's correct. I think the one thing that we are behind is -- I think all the AI cloud, you need a lot of power, right? And then if you look at U.S., the regulatory approval, and it just takes longer to get that. In China, if they decided to have it, they quickly can get all the approval and get it done. So that's something that we have to pay attention. We may...

You're doing this within -- you're investing in companies. What -- do you see open source development changing in the U.S. materially? Or do you feel like it continues on the foundation model side where it's going to be largely closed source and not open rate's open source models?

Good question. Some companies, they call it open source. And when they become successful, they become closed source. And then so they block it out. So I think we have to continue encouraging open source. I think that's the best way of not repeating everybody's effort and so that we can really drive more successful improvement and quicker.

You almost need a business model where something else is able to go out and fund that, because if you go just with open source, the training costs are so high that the economics don't work out, whereas in China, it's different because the government subsidizes a lot of that.

That's correct.

How do we handle that? What's the game theory on our side? What do we do?

I think a couple of my friends are putting a lot of effort in terms of rebuilding that open source community and even funding some of this -- the best AI researcher to really do that, they even from a new institute rather than just university to fund some of this program. And I think I really encourage a jump on it because this is something that badly needed. Same thing for semiconductor, I used to be the only guy venture investment and until this AI take off, everybody knocking on my door what is the next big deal that you are doing.

And until this AI take off, everybody knocking on my door. What is the next big deal that you are doing. And so I think overall, and I'm glad to see all my brother and sister from the VC side starting to put money to VC to co-invest with me. I think in some way, I think we have to continue doing that. And then the other part is on the foundry side, it's very important to have a U.S. manufacturing foundry. That's why I decided to come in and double triple down. It's a long-term business, but we just have to really do that because the industry need that and the U.S. need that.

So I think overall, not just the process, also the advanced packaging and that becomes -- really become the bottleneck and something that we have to really continue to drive system wafer kind of packaging arrangement. And that's why we are investing in some of that and then some of the new technology to drive the packaging so that we can really help enable the AI driver and then to really take off.

Does Intel build GPUs in the future?

Yes. And I just hired the Chief GPU architect. And then so he's very good. I'm very delighted he joined me. And it takes some persuasions. And then I told him that not just CPU, GPU is also very important, different or different application workloads. And so you have to really optimize. So I'm not hung up with just Xeon 86 and also embrace RISC-V and you got to be more flexible. And the key thing is more that it's kind of more the software layer that you have to drive down from call it the software 2.0. It's define the software, you dry the hardware.

So Intel is going to build CPUs, they're going to build GPUs. They're going to have their own foundry to build those CPUs and GPUs, and you're going to have a foundry that's going to be made at scale available to other manufacturers, and you will make sure that you partner with the other GPU providers.

That's right. And some new material like the glass is a very good insulator. So we have doubled down on glass. And then the other part is artificial the diamond. And that's another good glass. So I think you have to go through the different chemical table and find out what are the new material like gallium nitride is very good for RF and switch area. And so that is interesting. So I think we have to look at different new material we have to look at CMOS is a little bit run out of steam. So you need to look at some of the new material. And that's why it's very important for me is I have continued to have that curiosity. So starting to look at some of these new things, things are a little bit short term, midterm and longer term and how you're going to make the impact to the industry.

So Lip-Bu, you've got enterprise audiences over here. You've got CIOs, you've got Chief Information Security officers, there's 10 million people that are probably watching -- going to be watching this. What advice would you have for people on how they should be thinking about AI and how they should be thinking about scaled infrastructure build out. What would you say to them?

Yes, it's a very good question. So I think we all get excited and also a lot of pressure to adopt AI and then to the enterprise. So I think it's very important to think about what is the problem you try to solve. What is the outcome you want to look for. And then -- and also some of them that just like Intel, we have a very legacy IT infrastructure.

I just recruited the best CIO that I can find and I brought her on board. And I kind of told her that, look at -- I think it's a good time to look at the foundation, what are the changes we need to make. You don't just put on top of the old legacy, it's not going to work. And then you have to really undo that and then embrace some of the new AI tool pro function by function to really drive the augmentations. But more important is what are the process that you cannot drive the metric to success.

And then frankly speaking, a friend of mine, we're just talking actually is MIT professor on the MIT CEO Advisory Board. And one just mentioned about -- if you look at all the AI, the study that he study is that basically the productivity of the global economy is growing very small percentage. So we still have to adopt more broadly so that we can really see the productivity go up with agentic AI with agent and enroll...

So we still have to adopt more broadly so that we can really see the productivity go up with agent AI with agentic and then roll actually

Actually the lowest it's been in compared to back in the 1800s it was very eloquent...

I was shocked to see the productivity that he showed me the curve. I said, wow, I thought that would be much higher. So I think that's something that we have to think about what is the end outcome that you want to have. And then go back to foundation, how do we drive that -- and then meanwhile, give ourselves enough accountability that able to really drive the productivity of the enterprise that you can measure, you can present to your Board, hey, by doing this, I invest this new technology, actually improve my productivity and improve our revenue growth.

Lip-Bu, Intel is a national treasurer, so are you. Thank you for being here.

Thank you for inviting.

Thank you for standing by, and welcome to Intel Corporation's Fourth Quarter 2025 Earnings Conference Call. [Operator Instructions]. As a reminder, today's program is being recorded.

And now I'd like to introduce your host for today's program, Mr. John Pitzer, Senior Vice President, Investor Relations. Please go ahead, sir.

Thank you, Jonathan, and good afternoon to everyone joining us today. By now, you should have received a copy of the Q4 earnings release and earnings presentation, both of which are available on our investor website, intc.com. For those joining us online today, the earnings presentation is also available in our webcast window.

I am joined today by our CEO, Lip-Bu Tan; and our CFO, David Zinsner. Lip-Bu will open with comments on our fourth quarter results as well as provide an update on the progress we are making on our strategic priorities. Dave will then discuss our overall financial results, including first quarter guidance, before we transition to answer your questions.

Before we begin, please note that today's discussion does contain forward-looking statements based on the environment as we currently see it, and as such, are subject to various risks and uncertainties. It also contains references to non-GAAP financial measures that we believe provide useful information to our investors. Our earnings release, most recent annual report on Form 10-K and other filings with the SEC provide more information on specific risk factors that could cause actual results to differ materially from our expectations. They also provide additional information on our non-GAAP financial measures including reconciliations where appropriate to our corresponding GAAP financial measures.

With that, let me turn things over to Lip-Bu.

Thank you, John, and thank you all for joining us today. 2025 was a year of solid progress. Over the last 10 months, we established the foundation for new Intel, a more focused and execution driven company. We simplify our organization and greatly reduce bureaucracy to improve efficiency and accelerate decision-making. We also recruited new leaders from the outside and empower key leaders from within. We strengthened our balance sheet for strong new partnership and deepened relationship with existing as well as new customers. I'm encouraged by my conversation with our customers and partners around the world. I'm hearing a clear, consistent message. They see the progress we are making. -- they won intel at the table as they navigate their own transformations. The opportunity in front of us is meaningful and significant. The era of artificial intelligence is driving unprecedented demand for semiconductor across the entire compute landscapes from AI accelerated and traditional data centers into the network and enterprise domains all the way out to client and edge devices.

Rapid deployment of AI workloads across this diverse environment will require hectogeneous silicon solution, leveraging CPU, embedded NPUs discrete and integrated GPU, ASIC and XPUs. In addition, we will need to see innovation in the software stack along with new emerging technologies like photonics, memory interfaces, interconnect and Quantum to name just a few.

The breadth of our IP and know-how across silicon design, system level integration, wafer manufacturing and advanced packaging uniquely position us to capitalize on these AI-driven trends, capture sustainable profitable growth. This will not happen overnight, and our execution needs to continue to improve. While we will stay humble as we address the work ahead and we will never be satisfied.

Our Q4 was another positive step forward. Revenue, gross margin and EPS were all above our guidance. We delivered these results despite supply constraints which meaningfully limited our ability to capture all of the strengths in our underwriting markets. We are working aggressively to address this and better support our customers' needs going forward. Looking ahead for 2026. We will continue to position Intel to capture the significant growth opportunity AI presents across all our businesses. We will do this by strengthening our client franchise advancing our data center, AI accelerator and ASIC strategies and continue to build trusted U.S. foundry.

Let me start with our Core x86 franchise, which remains the most widely deployed compute architecture in the world. The deployment of AI is only amplifying the importance of x86 from orchestration and control planes to influence edge workloads and agentic AI.

In our Client Computing Group, we strengthened our position in both consumer and enterprise notebooks with our Core Ultra Series 3 lineup, formerly known as Panther Lake. And build on our most advanced Intel 18A manufacturing process. We committed to deliver our first Series 3 SKU by the end of 2025, and we exceeded that commitment by delivering our first 3 SKUs.

While we still have work to do, I'm encouraged by the steady progress on our Intel 18A use, and Naga and his team remain laser-focused on additional improvements as they ramp Series 3 into the high volume needed to meet strong customer demand. Our client momentum was on full display at earlier this month, where we formally launched Series 3 with our OEM partners, powering over 200 notebook designs Series 3 will be the most broadly adopted and globally available AIPC platform we have ever delivered.

Along with our next-generation Nova Lake coming at the end of 2026. We now have a client road map that combines best-in-class performance with cost optimized solutions. Giving me confidence that we are on the path of 45 market share and profitability in both notebooks and desktops over the next several years.

In addition, PC become an important part of the AI infrastructure. The search in AI workloads is driving massive demand for data centers, but cloud capacity alone cannot meet the scale of influence needed, especially in the power constrained environment. This accelerating the push toward hybrid AI splitting workloads between cloud and client, which offers clear advantages in performance, cost and control. We are working closely with ecosystem partners to seamlessly enable hybrid AI, and we are encouraged by the opportunity to grow the installed base and accelerate the refresh rates over time.

Let me now turn to DCI to support our AI objectives, I believe that our traditional server and accelerated road maps must advance together. To reinforce the alignment I centralize our data center and AI businesses under Havok, ensuring tight coordination across CPUs, GPUs and platform strategy.

Demand for traditional service continues to be very strong, and we are focused on ramping available capacity to support the meaningful uptick we are seeing including partnering with key customers to support their needs beyond 2026. The continuing proliferation and diversification of AI workloads is placing significant capacity constraints on traditional and new hardware infrastructure, reinforcing the growing and essential role play in the AI era.

This is and will continue to benefit the ongoing ramp of Granite Rapids as well as our mainstream products Sapphire and Emirate Rapids. We have also made decisive changes to simplify our server road map focusing resources on the 16-channel Diamond Rapids and areas to accelerate the introduction of Coral Rapids, where we can. Recorded Rapids, we will also reintroduce multi-trading back into our data center road map. We also continue to work closely with NVIDIA to build a custom Xeon fully integrated with their NVLink technology to bring best-in-class x86 performance to AI cost notes.

Over the last several quarters, we have been developing a broad AI and accelerated strategy that we plan to refine in the coming months. This will include innovative options to integrate our x86 CPU with fixed function and programmable accelerator IP. Our focus is on the emerging wave of AI workloads, reasoning models, genic and physical AI and influence at scale, where we believe Intel can truly disrupt and differentiate.

Our long-term ambition is clear to rebuild Intel as a compute platform of choice for the next era of AI-driven computing, grounded in world-class engineering and accelerated road map and a renewal culture of execution. We are also building momentum in ASICs as customers seek purpose-built silicon for AI networking, cloud workloads. Our combination of design services, IP building blocks and manufacturing capabilities position Intel well to resolve specialized problems at scale. This is not a new area for us, although this is one that I'm committing significantly more focused resources and investment dollars, including leveraging my own experience at Cadence Design supporting and growing this market.

Finally, we remain focused on the long-term objective of building a world-class wafer and advanced packaging foundry, anchor in trust, consistency and execution. As I have said before, building a foundry business will take time and considerable effort and resources. While still early in our journey, we have hit some earlier important milestones worth highlighting.

We are now shipping our first products built on Intel 18A the most advanced semiconductor process developed and manufactured on U.S. soy. As stated earlier, yields continue to improve steadily as we work to ramp the supply needed to meet strong customer demand. In addition, Intel 18AP continued to progress well. and we are engaging with internal and external customers on this note, delivering our 1.0 PDK at the end of the last year.

Intel 14A development remains on track. We have taken meaningful steps to simplify our process flow and improve our rate of performance and yield improvement. We are developing a comprehensive IP portfolio on Intel 14A, and we continue to improve our design enablement approach. Importantly, our PK are now viewed by customers as industry standard. Engagements with potential external customers on Intel 148 are active. We believe customers will begin to make firm supplier decision starting in the second half of this year and extending into the first half of 2027.

We also have the opportunity to provide strong differentiation in advanced packaging, particularly with IMP and IMT. We are focusing on improving quality and yield to support customer desire for ramps beginning in second half of 2026.

In closing, I reflect on 2025. I'm proud of the resilient commitment our team has demonstrated. We exit the year with a stronger foundation and clearer group map for 2026 and beyond. The opportunity ahead is meaningful and significant as AI-driven computing expands all the markets we serve. But I'm also mindful of the challenges ahead of us and transparent about the areas that we are doing well and areas we need to improve. In the short term, I'm disappointed that we are not able to fully meet the demand in our markets. My team and I are working tirelessly to drive efficiency and more output from our fabs, while yields are in line with our internal plans, they are still below what I want them to be. Accelerating yield improvement will be important lever in 2026 as we look to better support our customers.

As I said earlier, we are on the multiyear journey. It will take time and resolve, but my team and I are committed to rebuilding this iconic American company and increasing the long-term value for our shareholders.

I would like to thank my team for their hard work over the course of the last 10 months. I look forward to updating you on our progress as we continue this journey together, including hosting an Investor Day in the second half of this year, at our headquarters in Santa Clara.

Let me now turn it over to Dave to walk through our financials and business trends in more detail.

Thank you, Lip-Bu. We remain encouraged by the fundamental drivers of demand across our core markets. Fourth quarter revenue was $13.7 billion at the high end of the range we provided in October. We experienced strong growth across all our businesses, benefiting from the AI infrastructure build-out with AIPC, traditional server and networking revenue, all up double digits sequentially and year-over-year.

Q4 marks the fifth consecutive quarter of revenue above our guidance even as we navigate industry-wide supply constraints for our key products. Non-GAAP gross margin came in at 37.9% and approximately 140 basis points ahead of guidance on higher revenue and lower inventory reserves, partially offset by increased mix of outsourced client products and the early ramp of Intel 18A to support the launch of core Ultra Series 3 code name, Panther Lake.

We delivered fourth quarter non-GAAP earnings per share of $0.15 versus our guidance of $0.08 and driven by higher revenue, stronger gross margins and continued spending discipline. Q4 operating cash flow was $4.3 billion with gross CapEx of $4 billion in the quarter and positive adjusted free cash flow of $2.2 billion. NVIDIA's $5 billion investment closed in Q4 as expected. For the full year, revenue was $52.9 billion, down slightly year-over-year due to constraints across our own manufacturing network and with external suppliers, which limited growth, especially in the second half.

Full year non-GAAP gross margin was 36.7% up 70 basis points on reduced period charges. Full year non-GAAP EPS was $0.42, up $0.55 year-over-year on lower period charges and improved operating leverage. Specifically, non-GAAP OpEx of $16.5 billion was down 15% versus 2024 as we executed actions to reduce complexity and bureaucracy in the business and drive improved execution. For the full year, we generated $9.7 billion in cash from operations and made $17.7 billion of gross capital investments with capital offsets of approximately $6.5 billion. Although adjusted free cash flow was minus $1.6 billion in 2025, we produced $3.1 billion in the second half as cash from operations more than doubled half-on-half.

We exit 2025 with $37.4 billion of cash and short-term investments, bolstered by further monetization of Mobileye the completion of our stake sale of Altera to Silver Lake accelerated funding from the U.S. government and investments by the SoftBank Group and NVIDIA. In addition, we repaid $3.7 billion of debt. Looking back, 2025 marked an important year of progress against our key priorities, even as we know we have more work ahead.

Internally, we reorganized and rightsized the team to become customer-centric and engineering-focused while shoring up our balance sheet to give us more flexibility to pursue our goals. We've navigated a market that has shifted from tariff-driven uncertainty in the first half to an intense AI-driven demand environment constrained by supply in the second half. 2025 demonstrated the staying power of the x86 ecosystem across client and data center and the importance of our manufacturing assets as we launched Core Ultra Series 3 on Intel 18A, the most advanced process fully developed and manufactured in the United States. Both create a firm foundation on which to build the new Intel.

Moving to segment results. Intel Products Q4 revenue was $12.9 billion, up 2% sequentially. CCG revenue was down 4% quarter-over-quarter even as AIPC units grew 16%, and DC AI was up 15%, reflecting strong demand for traditional server compute. These results reflect our efforts to balance our constrained supply with strong data center demand while maintaining support for our client OEM partners.

Where possible, we're prioritizing our internal wafer supply to data center and leveraging an increased mix of externally sourced wafers in clients. CCG revenue was $8.2 billion and in line with our expectations. We estimate the client consumption TAM was greater than 290 million units in 2025, marking 2 straight years of growth of the post-COVID bottom in 2023 and the fastest TAM growth since 2021.

Within the quarter, CCG launched 3 SKUs of Series 3 ahead of our expectations of one. Performance reviews have been extremely favorable, with up to 27 hours of battery life, a 70% gen-on-gen improvement in graphics and performance on industry standard benchmarks that is 50% to 100% better than peers. TCAI revenue was $4.7 billion, up 15% sequentially, above expectations and the fastest sequential growth this decade.

Revenue would have been meaningfully higher if we had more supply. While the market continues to benefit from more power-efficient CPUs, stimulating a refresh cycle, all indicators point to the growing and essential role CPUs will play within hyperscale and enterprise AI data centers as inference-driven AI usage expands. The world is shifting from human prompted requests to persistent and recursive commands driven by computer-to-computer interactions. The CPU central function coordinating this traffic will drive not only traditional server refresh, but new demand that grows the installed base. In addition, due to the networking demand for the AI infrastructure build-out, our custom ASIC business grew more than 50% in 2025, 26% sequentially and reached an annualized revenue run rate greater than $1 billion in Q4. This strength provides our ASIC team a solid base to pursue a $100 billion TAM opportunity.

Operating profit for Intel Products was $3.5 billion, 27% of revenue and down approximately $200 million quarter-over-quarter on an increased mix of outsourced products and seasonally higher operating expenses. Intel foundry delivered revenue of $4.5 billion, up 6.4% sequentially on increased EUV wafer mix. EUV wafer revenue grew from less than 1% of wafers out in 2023 to greater than 10% in 2025. External foundry revenue was $222 million in the quarter driven by projects with the U.S. government and the deconsolidation of Altera. Intel foundry operating loss in Q4 was $2.5 billion and $188 million worse quarter-over-quarter driven by the early ramp of Intel 18A.

Within the quarter, Intel foundry met key T&A and 14A milestones. With the official launch of Core Ultra Series 3 and Intel Foundry is the only semiconductor manufacturer in the world shipping gate-all-around transistors with backside power for revenue. These advanced wafers are rolling off our production lines in Oregon and Arizona here in the United States. Finally, our continued progress on Intel 14A demonstrates our commitment to research and develop the world's most important technology on U.S. soil.

Turning to all other. Revenue came in at $574 million and was down 42% sequentially due to the Q3 25 deconsolidation of Altera. The primary components of all other in Q4 were Mobileye and IMS. Collectively, the category delivered an operating loss of $8 million. I'm pleased with the early momentum at Altera as an independent company with a new leadership team. They're industry-leading programmable fabric, developer productivity-driven software tools and a large installed base positions them well to drive long-term value creation.

Now turning to guidance. During the second half of 2025, we supported strong demand for our products with intra-quarter wafer production and inventory on hand. As we enter 2026, our buffer inventory is depleted and the mix shift in wafers towards servers, which began in Q3 will not come out of fab until late Q1 '26. As a result, and as we stated last quarter, our internal supply constraints are most acute in Q1.

In light of these dynamics, we are forecasting a Q1 revenue range of $11.7 billion to $12.7 billion. The midpoint of $12.2 billion reflects a lower end of seasonal Q1. Within the Intel products, we forecast a more pronounced revenue decline in CCG than in DCI as we continue to prioritize internal supply to our server end markets. We expect Intel foundry revenue up double digits quarter-over-quarter, helped by continued mix shift to EUV wafers and Intel 18A pricing.

At the midpoint of $12.2 billion, we forecast a gross margin of approximately 34.5% with a tax rate of 11% and breakeven EPS, all on a non-GAAP basis. Gross margin is down sequentially due to lower revenue, increased AT&A volumes and product mix.

Let me take a few moments to provide some color for your full year 2026 model. First, from a revenue perspective, we expect our factory network to improve available supply beginning in Q2 and for each of the remaining quarters in 2026. Within the server market, customer feedback and our own market intelligence points to the likelihood of a strong year of growth for DCAI. Finally, client CPU inventory is lean, and there is excitement for Series 3. In contrast, over the last several months, industry-wide supply for key components like DRAM, NAND and substrates has come under increasing pressure due to intense demand to support the rapid expansion of AI infrastructure.

Rising component pricing is a dynamic we continue to watch closely, especially relative to the client market and could limit our revenue opportunity this year. We target 2026 operating expenses of $16 billion. We expect noncontrolling interest or NCI to net to approximately $325 million in Q1 and be approximately $1.2 billion for the year on a GAAP basis. NCI is expected to grow meaningfully again in fiscal 2027. Our share count is forecast to be 5.1 billion shares and grow in line with our stock-based compensation going forward.

As we think about our capital expenditures for 2026, we're working to balance our ability to drive capital efficiencies with our need to respond to the demand signals we're receiving. Previously, we said CapEx would be down but are now planning for a range of flat to down slightly and for expenditures to be more weighted to the first half.

As a reminder, CapEx in 2026 would be to support demand in 2027 and beyond. We expect to generate positive adjusted free cash flow for the full year, and we're planning to retire all $2.5 billion of maturities as they come due this year.

I'll wrap up by saying that Q4 was another solid quarter to mark our fifth consecutive quarter of overdelivering to our guide. We exit 2025 increasingly confident in the long-term sustainability of the end markets we serve. We believe our improved balance sheet, thanks in part to the trust of our strategic partners, combined with the strong talent we have will enable us to meaningfully participate in the next wave of computing as the industry pushes for returns on their AI investments. I look forward to providing you our shareholders an update on what this future means to you at our Analyst Day later this year.

With that, I'll turn it over to John to start the Q&A.

[Operator Instructions]. With that, Jonathan, can we please take the first question?

Certainly. And our first question for today comes from the line of Ross Seymore from Deutsche Bank.

I guess the first one is 2 quick parts. It's both on supply. In the short term, are the yield improvements and the other actions you're taking sufficient to address just typical seasonality throughout the year given that usually the first quarter would be the low point on revenues. And then perhaps more importantly, longer term on the supply side, you guys seem more confident in your 14A, your 18A, the customer engagements, your internal road map when would you decide to loosen up the reins on the CapEx side of things so that you could address structurally higher demand going forward with more internal supply?

Thanks, Ross, for the question. So on the short-term supply, certainly, improving yields and throughput are a great driver of supply increases. In fact, it's got a great ROI to it because it doesn't require any incremental capital. So that's what Lip-Bu and I are actively working on to improve and we're reasonably confident that there's a good trajectory there.

That said, when you look at CapEx, it's a little bit more nuanced than just it's going to be flat to slightly down. It's actually down significantly in space. So we're spending a lot less in space. We think we have a good footprint in terms of clean room. And what we're devoting more of our dollars to is tool. So we are ramping up tool spending quite a bit in '26 relative to '25 to address this supply shortfall as well.

And in fact, every quarter, we're seeing kind of wafer start increases pretty much across the board across Intel 7, Intel 3 and 18A. So all 3 of them every quarter improve and get better to address the supply. Like I said, we think we -- things will certainly improve in Q2, but we won't be completely out of the woods here. But as we progress through the year, we think things will get better and better. Do you want to take it for anything?

Yes. On 14A, Lip-Bu has been very direct with us on all of this. He does not want to spend on capacity on 14A only spend on the kind of TD spend or R&D spend associated with 14 even in the fab until we have customers secured. We've talked about the likelihood of our customers on 14 their window to secure or for us to secure them will be in the back half of this year and in the first half of next year. And so once visibility improves there, we'll start to unlock the spend on 14A.

I can just add a little bit more. I think on the yield improvement, which we see in the 7%, 8% yield improvement per month. I think it's more in focus in the variation, make sure that we can be more consistent delivery and also the defect density at the end so that we can ship quality refer to the customer. So I think all those are very important for our PC client, Panther Lake and then also for the 18A and 14A development. So I think, all in all, I see improvement, but still not quite to the industry-leading standard yet.

Ross, do you have a quick follow-up?

Yes, I do. The gross margin side, Dave, you gave some puts and takes for the full year on a bunch of different metrics, but you didn't mention gross margin. How should we think about that? And forgive me, at the 40% to 60% incremental, I know you guys give kind of -- you can drive a truck through. So anything with perhaps just a little more precision directionally, it would be helpful.

I'm not even sure the math is that good on this time around. When you look at Q1, the gross margin decline there's 2 main components. Obviously, revenue coming down with a largely fixed cost business is going to affect gross margins. But the other piece of it is Panther Lake, while the cost structure improves from Q4 to Q1, it's still dilutive to the corporate average, and it's a bigger percentage of the mix. So it actually has a relatively negative impact on gross margin. So that's partly why we're guiding down. There's some mixed things going on as well.

I think as we progress through the year, the 2 things should benefit us. One, we improve our supply, Ergo, that should improve our revenue picture. On top of that, paper like cost structure gets better and better. Look, we talked about the incremental improvement every month we're going to see in yields. We're working on the throughput as well. So those 2 things combined, that should help in terms of cost structure and make this more of an accretive product to us as opposed to a diluted product. And I think that will be a lot of what is the story around gross margins for the year. There's still mix and mix can go in any different direction depending on how things play out. But what we're largely focused on for the kind of the next 12 months is driving the cost structure of these products that we're building. to improve the margins. We know that 34.5% is by no means an acceptable level of gross margins. And we're actively working it first to get it to 40% and then once we get it there, we'll move to a new target.

And our next question comes from the line of Tim Arcuri from UBS.

Dave, I'm wondering in the guidance, you're guiding to [ 12.2 ], but I'm wondering if you can kind of pro forma that for us like if you could meet all the demand, what would the kind of unconstrained guidance be for March?

Yes. It's a squishy figure to figure out, Tim, but I would tell you that if you look at kind of that 12.2 relative to the 13.7, we posted in the fourth quarter, and look at normal seasonality. It's in the range of seasonal, but it's at the low end of that range of seasonal, we'd be well above seasonal if we had all the revenue or supply, I should say, to hit the revenue.

Tim, do you have a follow-up?

I do, thanks. Lip-Bu, there's obviously a lot of excitement about your foundry business. It sounds like we might get a few customer announcements maybe in the second half. But I just wanted to ask you, what do you define as like in that business. I think prior to you arriving, the mantra was sort of to be #2, the #2 foundry player by 2030. If you look at most of the forecast for that #2 player at somewhere in the range of like $30 billion and revenue up by that time. Is this still kind of a reasonable bogey that you're shooting for? And like what do you consider as a I'll come to that.

Yes. Thanks,. Good question. So I think we are determined to commit to drive the world-class foundry business. So I think, first of all, on the FDA, I think we clearly development is on track. We like what we see. And we simplify the whole process flow. Most important is kind of building up the IP portfolio so that we can serve the customer.

Some of the IP is very critical in order to serve the customer. The other part is on the yield and improvement. We see a trend of improvement, and we also see the valuations getting better. I think in the long run, I think clearly, we are kind of anticipating the -- we have heavy engaging in some of the key customers. We think that the second half of this year, they're going to indicate to us what kind of capacity firm commitment so that we can deploy the capacity CapEx to really build that. So I think all in all, I think it's a service business, We really built a trust and the consistency we're able to deliver. And then we have the PDK on the 14A first quarter, 0.5%. And then we're starting to engage with customers with the key products that they want to run with us. So I think all in process. I think in the second half of this year, we were able to have the commitment that we can really drive the scale of the operation.

Another early indicator, I think, of success in the foundry is going to be advanced packaging. And we'll start to see that revenue come in even before we start to see meaningful wafer revenue. And I think as I was talking to investors over the past kind of 12 to 18 months, I was thinking that those opportunities would be measured in hundreds of millions of dollars and wafer opportunities would be measured in billions.

I'd say some of the early customer engagements suggest that we'll be well north of $1 billion on many of these opportunities for advanced packaging. So they're way more exciting than even I had expected. And it's because we have really good technology there that's very differentiated and supports AI in a way that is particularly special.

I think the , I think, is a very big differentiator for us. And then clearly, we have a couple of customers willing to even prepaid the subscript because subscript is very big supply shortage and then they're willing to share with us. That means that show the commitment they are in working with us.

And our next question comes from the line of Joe Moore from Morgan Stanley.

Yes. I wonder if you could talk about server prospects -- you sort of talked about some of the challenges of Diamond Rapids not having symmetric multifitting and copper rapid is going to be important. Can you give us a time frame on copper Rapids and sort of what's your expectation for the potential for market share puts and takes as you wait for that to come?

Yes, good question. So I think, first of all, I would centralize the data center and AI under Havok that I hired in to help us to build that. we already built the team and then include some of the recruit some of the talent on board. I think the more important right now, we are laser-focused in 16 channels Diamond Rapid and we simplify the product road map. And then the other part is accelerate the introduction of Coral Rapid. And Coral Rapid will have we introduced the multi-thread into our data center workforce. So I think overall, we are very positive. The team is in place now. The road map is very clear, and we are very decisive for doing that. and the laser focus on the Diamond Rapid section channel and also accelerate the Coral Rapid introduction.

Joe, you have a quick follow-up?

Yes, sure. And just in terms of the mix for the rest of the year, are you able to move wafers towards data center away from PC? Is that something that you're thinking about and just -- it seems like the constraints are at their worst point in Q1 and get better, but I assume you're still constrained beyond that. Are you able to move the mix towards data center?

We're absolutely constrained, Joe. So what we're doing within client we're focusing on the mid- and high end and not as focused on the low end. And then to the extent we have excess, we're pushing all of that into the data center space to meet that customer demand. And I think you'll see some share adjustments based on that because our primary focus is to our main customers. And obviously, we have important customers in the data center side. We have important OEM customers. on both data center and client and that needs to be our priority to get the limited supply we have to those customers.

And our next question comes from the line of Ben Reitzesfrom Melius.

My first question is about seasonality throughout the year. So Dave and Lip-Bu. I mean you're subseasonal in the first quarter, you said you'd be well above seasonal if you had the supply. So what does that imply for 2Q to 4Q? Should we model that above seasonal or are the constraints in PC so much that we shouldn't, thanks.

Yes, I mean, we would expect to be better than seasonal through the year if we can get the supply to where we think we can as we get into 2Q. So that's correct.

Ben, do you have a follow-up question?

Sure. I wanted to ask about -- with regard to the hyperscaler situation in servers. It's -- in terms of your momentum there, is this mostly driven by hyperscaler? Or do you feel like the shortage is mostly impacting them, making you sub seasonal? Or is the enterprise demand also are you seeing it there?

Yes. I think let me answer that question. I think the hyperscaler is very important for us to scale the business, and I spent a lot of time with the hyperscalers. I think a couple of things. One, clearly, message from them is the CPU actually is driving a lot of that business in terms of the different workloads that they're driving. So I think it's very encouraging to see that they are willing to commit long-term agreement. So we really prioritize deployment. So that something is very positive.

And then secondly, I think they are very excited about working with us in terms of decide working on the silicon the software and the system level engagement, and that something is also very exciting. So all in all, I think they are very strong. Their workload, they shared with us what they're looking at and what can be helping them. And also the ASIC design also is an opportunity for us. They also want to build some of the purpose-built silicon with the Xeon CPU included. And also, they are very interested about overall, how do you use the advanced packaging and then to make it more complete. I think overall, I think it's a great opportunity for us to work with them.

Certainly. And our next question comes from the line of Stacy Rasgon from Bernstein Research.

For my first question, I wanted to dig a little bit into the segments. So I mean if I sort of run the math, like Mobileye is going to be up and the Altera foundry revenues maybe close to a couple of hundred million dollars I mean both DCIA and client to be down pretty meaningfully. And the client's down more -- I mean maybe DCIA is down high single digits and clients down like mid-teens. But I guess, number one, is that true?

And number two, like why should data center be down so much given where the demand is and given where you're prioritizing, like why would I expect data center units to be down? It seems like they have to be down pretty meaningfully in Q1.

Yes. I mean, both will be down as a function of supply. Obviously, we're shifting as much as we can over the data center to meet the high demand. But we can't completely vacate the client market. So we're trying to support both as best we can and obviously work our way out of this supply issue. I do believe that the first quarter is the trough. We will improve supply in the second quarter. And part of the challenge is that in the third and fourth quarter of '25, we lived off of supply, but we also had a reasonable chunk of fixed finished goods inventory to also work through. Unfortunately, that is now down to kind of 40% of what it was at peak levels. So we don't have that to rely on. So it's just literally hand to mouth, what we can get out of the fab and what we can get to customers is how we're managing it.

Stacy, do you have a follow-up question?

I do. I mean just to push on that a little bit. I mean, you guys have your own factories, like why are you in the inventory situation that you're in? And then -- I mean, even if you look at -- I get the whole idea about finished goods versus other stuff. But I mean like you have $11.6 billion of inventory. And yet, it's not in the right place at the right time to sit like how does that happen?

As largely -- I'll tell you, Stacy. I think the biggest thing is that we if you go back 6 months or so ago and looked at what the outlook was. Core count was absolutely looking like it would increase. but the units were not expected to increase. And every hyperscaler customer we talked to was signaling that. And obviously, it has rapidly increased over the third and fourth quarter. And in talking to a few of them right before this call, I got the feeling like it was going to be a story we'd feel for several years. And yes, the advantage we have is we do have our own fab so we can squeeze out supply as much as possible, which is what we're working on. but we directionally weren't managing the supply to an expectation that there would be unit increased that significantly in data center.

Our next question comes from the line of Vivek Arya from Bank of America Securities.

For the first one, Lip-Bu, I'm curious, when do you think Intel should start getting any credit for external foundry efforts because you mentioned that you might care of awards in the second half of this year. I assume that you start building the capacity for that, right, sometime late this year or next year. So when do you actually start to get a decent amount of revenue from those customers. And I think you mentioned that building this business will take incremental amount of resources. So what level of external foundry revenue does intend to call this business a success? And when do you get that? Is it '27? Is it '28? Is it later?

Yes. Good question. I think first of all, I think the engagement with our potential external customer on the 14A are very active right now, a couple of key customers working with us. We expect them to really go through the milestone basis on 0.5 ptk and then starting to look at the test chip and look at how is our yield performance. And it's going to be a process to work with them.

And then I think the second half of the year, then they're starting to satisfy, then they're going to asking us, okay, now this is the particular product we're going to run with your foundry and the production. And then we the indication to us and that's the time my discipline is until they have a commitment to the volume, then starting to really build and the foundry expansion so that we can meet their requirement. And then the other part petrol, they also give us a list of IP. If it's a mobile related, you have to be low power IP that we need to have. If it is data center related, it clearly will be the performance, the connectivity, all these things that we need to really get it ready so that we can serve the as a service as a customer to meet their requirement. So it's parallelly on the IP readiness and also our yield readiness that is think satisfied.

Okay. Now this product. This is a volume we're going to run with you. and that's how you're starting to build. So -- in terms of 14A, realistically in terms of, I call it, the risk production in the later part of 2027 and real production, volume production in 2028. That is similar to the same time frame as a leading foundry.

And I'd just say, we probably will be able to give you a lot more color around all of these things at the Analyst Day that Lip-Bu mentioned will have in the back half of the year.

Does that you have a follow-up question?

For my follow-up, I'm curious, what do you think is the server CPU TAM in 2026? How much of that is x86, -- how much of that is ARM? I mean if you are supply constrained, is the entire industry supply constrained? Do you think that whatever you can't supply all that market share is going to go to your x86 competitor and to everyone in the AI community that is building ARM-based servers. So like when do you think your supply -- how much does the market grow? And when do you think your server CPU supply constraints come off?

Okay. Maybe I'll start -- so I think this demand, what we're seeing is largely in x86 phenomenon because it's an upgrade cycle in a lot of ways around older networks that have to talk in some way to the AI systems and the performance is not where it needs to be. So I really view this as x86. Of course, we do have another competitor in the space, and we'll be jockeying for position from a market share perspective. I think we will make great strides on the supply as we progress through the year. So I wouldn't envision that to be the fundamental driver ultimately a market share. And it's really about products and Lip-Bu talked about getting to a 16-channel time we wrap it out. accelerating the introduction of Coral Rapids. Those will be the things that are most important for us as we look at market share dynamics in the coming years.

I think from my side, I think clearly, hyperscale and the high-end OEM, ODM is critical for us from the -- side. And then we're basically working with them. Their first choice is the CPU from Intel. And that is a very clear message from them. They will try to get as much as we can give them. And then I think that's the key home driver.

Our next question comes from the line of CJ Muse from Cantor Fritzgerald.

I guess a follow on Stacy's question around supply. Considering your bullish commentary and AI-led demand and how you're supply-constrained and your peers, TSMC and Samsung Taylor are aggressively slotting for equipment delivery. Do you worry that if you wait for late 2026 to place orders that the lead times then might be longer than you thought? And as part of that, why wouldn't you look to be more aggressive today?

Yes, maybe I don't know if this is the answer to your question, but I mean we are aggressively getting tools on Intel 710, Intel 3 18A, that is happening. And we will be increasing our wafer starts as aggressively as possible on those nodes. What we're holding back on is 14A because 14A is really linked to foundry customers, and it does not make sense to build out significant capacity there until we know that we have the customers that will accept that demand.

And so that's just the discipline we're going to have. I'd say the other thing with regard to a lot of this around supply is Lip-Bu's first focus and our short-term focus is we think we gain a lot of supply just by doing things better with our existing tools and footprint, getting yields improved, getting cycle times improved and we're aggressively working on that. And we think there's lots of opportunity to improve our supply just on those 2 things that don't require CapEx, quite honestly. And that's something that's probably more unique to us right now than to other foundries.

C.J., do you have a follow-up question?

Yes, John, and thanks, just to hit on the press release, you talked about demonstrating technical feasibility on High-NA for future HVM. And so just curious, is that something you're still contemplating for 14A? Or is that more of a 10A adoption?

It will be part of our 14A process. Of course, there will be different variants of 14A, but 18A is targeted at 14A.

And our next question comes from the line of Harlan from JPMorgan.

Liu, as you look forward to 14A, you talked about engineering engagements with customers. And typically, the largest fabless semiconductor companies in the world want to run their own test chips to assess the new foundry node, and they typically wait for PDK 0.4 or 0.5 before they start their ship designs, it looks like they can get started now on test chip design with the release of your PDK 0.5. So I guess the question is, have customers commence test chip designs or are they maybe even further along than that and customers are already running their own 14A test chips now? I mean, in order for them to make decisions on 14A in the second half, they need to be running their test chips fairly soon. So if you give us an update there?

Yes, good question. So I think you're absolutely correct. A couple of customer we are already engaging about the PDK 0.5 and they are looking at the test chip. And also more important is a specific product they're going to run to our fab foundry. And that one, we are working with them. And then, of course, they want to know the capacity, the pricing, those are all in the discussion right now. So that's why I mentioned the second half of this year, they're starting to satisfy, then they can starting to say, okay, now we need this volume and we need particular fab from yours to do it. And then also the other part is, do we have all the right IP to serve them. And so those are the things that we are paalprocessed with their supply chain and also their design teams to work with them. So this is a very complex steps and that we are very familiar, we are working in the right way.

Harlan, do you have a follow-up question, please?

On your server portfolio, I apologize if I missed this earlier, but cater for us or your core sort of cloud workload optimized server platform that was targeted to be the first server platform to use your 18A manufacturing and ramp first half of this year. But Lip-Bu talked about server road map changes to focus on more performance-focused products. So is the team still supporting Clearwater for us? Or just focusing now on Diamond Rapids? And has the team taped out or take in your next-gen Xeon 7 Diamond Rapids products and any preliminary views on Ramp timing of Diamond Rapids?

Yes. So to answer your question is yes, we continue to do this and support that. And what I mentioned about focus on the 16-channel of Diamond Rapid is kind of focus on the high end of the Diamond Rapids so that we can really later focus really providing a differentiating competitive products. And then the other part is that you mentioned in the past, this multi trading is very important in terms of driving the performance. And the one that we can really come out with, it takes time to have that and we're going to have that in the Coral Rapid.

Now the question mark is how can -- how much can we accelerate that they put in earlier customer is really excited about that. Lip-Bu, you put it up earlier. That's why I would be cooking this team see how far the way we can really drive that acceleration to bring the market to bring the customer earlier.

Jonathan, we have time for 1 more question, please?

Certainly. And our final question for today then comes from the line of Aaron Rakers from Wells Fargo.

I have 2, if I can fit them in. But on the memory side and what we're seeing in the market, I'm curious of how you guys are seeing customers react to memory? Is there a potential demand destruction in the PC market? Just any kind of curiosity on what you're seeing in that? And how impactful is memory pricing to the gross margin, given obviously, I think, stronger for longer Lunar Lake demand.

Yes, good question. So I think industry facing a very big challenge is the memory constraint and also the pricing. So I think we also listen to our customers, some of the bigger players and in the OEM and the bigger player in the hyperscale, they have more access into the memory locations. So we kind of hear from them. And then secondly, I think some of the smaller ones, they are really challenging to scramble to get the memory. So I think that will be very important for us. Dave and I how to allocate and also our sales great in how to allocate to the right customer. We don't want to have a CPU, we sent to them, but they are missing the memory, they cannot complete the products. So we try to do it correctly and then -- that is very important to have that intelligence and then also feedback from the customer to work with us and so that we can fulfill their requirement. Dave?

Yes. I think on the Lunar Lake side, I think we've got what we need based on the current forecast, of course, that could always tick up, and then we would need more memory, which would kind of impact gross margins. But we were relatively aggressive in terms of getting the memory early. So I feel like we're in a relatively good place there. Now that said, those margins are low because the memory is packaged. So that is an impact to our gross margins as well. But I think we're largely in the place we thought we would be a quarter or 2 ago.

Aaron, do you have a quick follow-up? SP130629864 I do, and I'll be really quick. I'm curious on the custom ASIC side, you talked about hitting $1 billion of run rate business. How do we think about the progression of that? And how broad is the customer base within that opportunity set?

Good question. I think Dave mentioned earlier, is a $100 billion TAM market opportunity. And we are delighted. We are already in the run rate of $1 billion. It's a robust demand and customers really excited about what we have in terms of the CPU Xeon and also the AI-related momentum. So they're more building a purpose build, silicon for AI and network and crow -- and that part, I think we continue to work on that. And also more important for them is also the advanced packaging, make that more compelling. And that's the advantage of Intel that we can do both to provide customer delight. And so that, I think, is a good opportunity for us. So with that, thank you again for joining us today. As we move forward, I remain focused on disciplined execution and deep collaboration with our customers to see the meaningful opportunity created by AI era. While we have a lot more to do, we are confident in the foundation that we built and the progress underway. We're looking forward to provide you another update in April.

Thank you, ladies and gentlemen, for your participation in today's conference. This does conclude the program. You may now disconnect. Good day.

All right. Welcome back to the Barclays Global Tech Conference. Pleased to have John Pitzer with us from Intel. I'm going to read some disclosures first and then we'll hop in. But before we begin, please note that today's discussion may contain forward-looking statements that are subject to various risks and uncertainties and may reference non-GAAP financial measures. Please refer to Intel's most recent earnings release and annual report on Form 10-K and other filings with the SEC for more information on the risk factors that could actually result -- that could cause actual results to differ materially and additional information on our non-GAAP financial measures, including reconciliations or appropriate to the corresponding GAAP financial measures. That's harder to do than you make it see.

You did it well, Tom. I appreciate the time this morning.

Yes. Thank you for joining us. So why don't we start on 18A? I understand it's always a tough question because everyone asks, give us mark-to-market, and it's difficult because they're obviously very customer-specific. But how are things progressing there? And then any updates on yields, landmarks, et cetera, that you can offer?

Yes, happy to help and I appreciate everyone joining today. On 18A, listen, our first product out on 18A is Panther Lake, it's internal product for PCs. We had committed to getting our first SKU out by end of year. The good news is we've done that. If you're in Vegas in January, stop by our booth, you'll hear a lot more on Panther Lake from us and probably from our OEM partners as well for the first time because while we're shipping, it's really CES, which is the launch event.

And you're likely to hear a couple of things about Panther Lake. Our OEM partners love the part and they want us to get them more volume. And I'm sure we'll talk about the supply constraints we're seeing in our business a bit later on. Now relative to Panther Lake and 18A yields, it's been a journey this year with, I think, a positive trajectory. I think when Lip-Bu joined the company in March, he was clearly unhappy with where absolute yields were.

I think he was even more unhappy that the progress on 18A yield was kind of half hazard. I think as you fast forward to today with a lot of work that Naga, Lip-Bu and the teams have done. I'd say we've gone from unhappy to unsatisfied because we'll never be satisfied with where our yields are on 18A, but I think that Lip-Bu and the team are pleased that we're now kind of seeing a consistent cadence of yield improvement on Intel 18A month on month on month.

And as Dave talked about on the earnings call, we will continue to see yields improve as we go throughout next year. And by the end of next year, into '27 will be at what we think are kind of industry standard yields. We never talked about absolute yield numbers. We had, obviously, a POR plan of record as we exited this year that we're on track to hit. So I think the progress has been good.

Sounds great. Yes, I was curious and people ask this often is under new leadership, what are specific changes that have kind of -- what are specific changes that new leadership has brought in? Is it equipment? Is it process? Is it culture? Like where are those changes? Because you would imagine you had a lot of smart engineers in the room before. What's changed?

I think a bit of all 3 of those things that you hit upon, I think Again, when Lip-Bu did his deep dive on 18A yields and was not happy. I mean the first thing he told the team to do is why are we engaging external customers with a product that's not good. Let's take those resources and kind of refocus on getting yields where they need to be for Panther Lake, which we did. And once we have an acceptable yield curve for Panther Lake, we can reengage with the external world on Intel 18A, which we're doing now on Intel 18AP and we're pleased with the maturity of the PDKs on AP.

I think in general, it was really incentivizing the teams and quite frankly, leaning more heavily on suppliers. I think Lip-Bu -- one of the reasons why the board hired Lip-Bu is because he has a tremendous network within the semiconductor ecosystem. And on day 1, he did something that we've never done, which is share our yield data with external suppliers, and I won't mention names, but they've been extremely helpful. And I think Naga has put a huge focus on this as well. And I think that the improvements we've seen as we've gone throughout 2025 are a reflection of those efforts.

You're leaning on some outside IP today. I think most recently, last week, you talked about 70-30, which you reiterated from kind of the earnings call. As you move forward, where is that headed with Nova? Can you talk about splits there?

Yes. So when you say IP, it's really externally sourced wafers from external foundry suppliers. Yes. So roughly speaking, about 30% of the wafers that we get today are externally sourced. We've talked about Panther Lake starting the process of bringing more wafers internally. And so as you think about kind of our external strategy on the client side, it really started with Meteor Lake a couple of years ago, where 1 of the 4 tiles was outsourced. That was accelerated both with Arrow Lake and Lunar Lake, which is today our leading-edge parts on notebook desktop for Arrow Lake and notebook for Lunar Lake.

And those are at least on the compute tiles -- sorry, the logic tiles are 100% outsourced. We still do a lot of the base die and advanced packaging internally, but most of the active tiles are externally sourced. What we've said with Panther Lake is to the extent that Arrow Lake and lunar are 100% outsourced, about 70% of the tiles that we need for Panther Lake will be coming back in-house, which is a positive relative to filling fabs and driving better profitability.

I will remind people, Panther Lake is only a notebook part. As we move to Nova Lake, Nova Lake will cover the full PC stack of notebook and desktop and bring even more wafers back. And so we've kind of got a trajectory now today where we're going to be bringing more internally. Having said that, I'll also remind you, as we navigate this tight supply situation, we are shifting more of our internal capacity away from client towards server to serve that market. And we are leaning a bit more heavily on some of our external suppliers on the foundry side around Arrow Lake and Lunar Lake to augment some of the volume that we're shifting internally towards servers.

Yes, I want to keep going down the technology , but it's a perfect time to kind of interject on what's going on with supply charges right now. So obviously, you're moving product around because there's a strong market need on the server side. What are you seeing there that's driving that? And then is that a limiting factor for you guys in the near term? Or do you think that that's actually beneficial because if when things are tight, you have a little more flexibility on price that cannot be growing?

Yes. Let me take a step back and talk a little bit about how this year kind of unfolded because we've been in what we consider a strong PC market all year long. We came into the year thinking PC unit volumes at the industry level being about $270 million. we're now sitting at about [ $290 million, ] but we consistently saw upside in Q1 and in Q2 in a way we didn't see that upside in the server business. Now we were worried back then that a lot of that was tariff dynamic pull forward. I think as we went into Q3, 2 things happened on each side of the business.

On the PC side, we got more comfortable about sustainability. And on the server side, we just had meaningfully different conversations with our customers. They were significantly upsiding their forecast as we went throughout Q3 in a way that they weren't in the first half of the year. And that has created a situation where we're in tight supply.

We're undershipping demand today in both markets. I would say despite the fact that we're moving wafers actively away from client towards server internally, we're still likely going to undership server demand next year by a wider margin than we will undership PC demand, which gives you a sense of how strong the server demand market is today. And I'll reiterate what we said at earnings, Q1 is probably the peak of our supply tightness. We'll continue to undership demand after Q1, but I think Q1 is the roughest point of next year and things will get gradually better as we go throughout Q2, Q3 and the second half.

And what's driving that server dynamic? Obviously, you're seeing a lot more spend as it relates to AI. And I think that there was a misconception historically that an increased spend on GPUs would eradicate spend on CPUs. It seems like that attach rate is only increasing or it's kind of pulling along CPU with it. Are you seeing that? And is that just like in head nodes? Where is that application?

Yes, we're still trying to get our arms fully around what's driving this I don't want to beat ourselves up too much for missing this because, quite frankly, it feels like our customers missed it. But there's probably 3 dynamics at play. One is, as you mentioned earlier, over the last couple of years, most of our customers, especially in the hyperscale space, were rightfully focusing on AI infrastructure and they just delayed some refresh of their installed base of servers.

The second dynamic, which I think relates to the first is all of those customers are now severely power constrained. And one of the easiest ways to improve your power budget is to take a 5-year-old chip that's sitting in your installed base and replacing it with a brand-new chip because those brand-new chips are about 80% more power efficient than what's sitting in the installed base today. The third dynamic, which is the one that's, I think, hardest for us to try to get our arms around is whether or not agenetic AI is just putting additional demand on traditional hardware infrastructure. And it sort of makes sense because to date, AI has really been focused on frontier models in the LLMs, and that's a GPU spend. But that's the brain of AI, and that brain is being trained on data, which is somewhat agnostic to what you do every day.

To make that brain valuable to you as an individual, you've got to move your data onto that brain, which is really a RAG sort of architecture, which looks a lot more like traditional compute. And you've seen other areas in traditional hardware infrastructure that have benefited as well. And the one I always point to is nearline HDD demand is also very, very strong. If you look at what our customers are telling us, many of them are now asking for longer-term supply agreements on the server side. Many of those agreements are asking for stretch beyond next year. So there does feel to be some sustainability in this.

When you talk about the supply in server being more strained next year than it is in clients or at least undershipping market, the market more than in client. Should we think about the ability for you guys to impact pricing in a similar way? Or would there be other variables that would...

I mean potentially, I want to be clear, we feel a lot better competitively where we sit on our road map for client than we do for servers. So we still have a lot of work to do on the server road map front to get to where we want to be competitively. So clearly, as we think about these longer-term supply agreements with some of the CSPs, we're probably focused a little bit more on market share than necessarily on the pricing side of things. But obviously, in a tight market, the economics are economics.

Yes. Perfect lead into 14A next generation. So how are you progressing there customer engagements on 14A? And just remind us, I think prior regimes had talked around a very specific moment where you would be at parity with the rest of the market. I think part of when Lip-Bu took over and some of the comfort that investors got was, hey, let's just wait until we get there, and then we'll talk about it a little bit more. Maybe give us mark to market there.

Definitely a bit different style in communications with Lip-Bu. And as the guy who needs to interface with investors, I appreciate that style. I would say the early engagements on 14A are going well. We still obviously have work to do. But I do think it's a little bit of a different sort of mindset and engagement model at 14A than we had at 18A. And if I could point to kind of 3 things that are meaningfully different and probably helping us.

The 3 I would point to is during 18A, in the definitional phase of the node, we weren't really engaging with the external world. We were really only engaging with Intel products as the internal customer. And all of the decisions we made on the transistor level was really to optimize for them.

And it wasn't really to the development phase that we started to get some true feedback from external customers. In 14A, in the definitional phase we are engaged with external customers. And what that really means is we're getting earlier, more and better feedback and feedback that can actually influence how we develop the node a lot quicker. So that's absolutely helping us.

The second thing that's helping us is I tend to tongue in cheek say that we didn't learn how to spell PDK until about halfway through the 18A node. And we had a lot of growing pains in making sure that our PDKs were industry standard like. And we continue to miss the mark on 18A. We were late with our 1.0 PDK. When we finally got it out, if you did channel checks, some customers would say, it really wasn't a 1.0 PDK. The good news is we took all of those sort of lessons and we forward fed them, quite frankly, both into the 18AP PDK, which is maturing nicely, but also on the 14A PDK.

The third thing I'd point out is 18A, we were trying to bend physics for the first time in 2 new ways. We were moving from FinFET to gate-all-around, and we were introducing backside power for the first time. and that created a lot of complexity. The good news on 14A, there are clearly technical challenges on 14A, but this is a second-generation version of FinFET and it is a second generation of backside power. And so again, we can forward feed all the learnings on 18A. And what we have said and what the data shows is if you look at yield and performance today on 14A, and you compare it to a similar point of development on 18A, we are meaningfully further ahead on 14A than 18A.

Now having said all of that, we are still setting the expectation that as you think about our ability to land external customers, that opportunity really starts to open up in the second half of next year into the first half of 2027.

Got you. At earnings, you talked about a more flexible CapEx plan for calendar year '26, which sounded to me at least like it would move a bit higher. So as we're approaching the turn of the year, how should we be thinking about capital commitments and spend into next year? Is this really dependent on customer breadth of 18A? Or what other factors should we be thinking about?

Yes. So our CapEx guidance for this year on a gross basis is at $18 billion. I would say, at the end of Q2, we were emphatically saying it was going to be down next year. I think given some of the tight supply we have today, I think directionally down is still right. I'm not sure we're as emphatic as we were. And there's a couple of things to keep in mind. As we -- especially as we start to get into early next year given the lead time of actual -- from spend to ramp, most of next year's CapEx is really about 2027 capacity, not about 2026 capacity.

The other point that I think is important is I do believe we fundamentally feel like we can still run our fabs more efficiently. And we like having that tension in the business because again, we should be able to get more output on the same footprint. And I think Naga and Lip-Bu are driving the teams to absolutely do that. Offsetting that is it's not a great place to be when you're undershipping your customers. It's a bad position to put them in, and we want to be mindful of that.

And so I still think down is probably directionally right. But quite frankly, if we were plus or minus $1 billion around that $18 billion in either direction, it wouldn't surprise me.

Helpful. On the packaging side, you hear a lot more rumors around Intel Foundry and the packaging capabilities, which seem to be quite strong. I know that it is a smaller dollar value that you can bring to the market than some of the other large customers, but maybe focus on where you bring value and like why people should be paying a little bit more attention.

Before I answer that, one last point I want to make on CapEx for next year, we are not prebuilding for external foundry. And so when we win an external customer, we will need to come back to the market and increase our capital spending plans. Now there's probably 2 phases to that. I think the initial tranche of 14A customers, we should be able to support out of Arizona, which will be an increase in CapEx, but not a meaningful increase in CapEx. The more meaningful increase will come when we have to start accelerating Ohio.

So just think through that. On the packaging side, yes, we're pretty excited about our technology with EMIB and EMIB-T. We think EMIB allows you to do things around reticle size and sort of density that you perhaps can't do with CoWoS. And we're being a little bit cautious about being overly optimistic about this because if you remember a year ago, when we were here, we were pretty bullish about advanced packaging revenue for this year. That was much more about being CoWoS overflow with [ fervorous ] and quite frankly, we probably underexecuted a bit. The market probably brought up CoWoS capacity better than people thought, and that revenue opportunity didn't materialize. EMIB is a little bit different because we've got capabilities that we don't think you can get with CoWoS. We're starting to see indications from customer bases -- our customer base that agrees with that. The hope is, as we get into the second half of the year -- of next year, we can give you tangible proof by actually generating revenue through the P&L.

A couple more I want to hit on the data center side. So NVIDIA investment, that obviously frees you up from a capital perspective in the near term. I get a lot of questions about your priorities in the data center. Obviously, you're continuing to march down the path of having the best server product that's available. But when you're interacting with NVIDIA, there's a view that you're going to want head nodes compliant with NVL. And when that -- when you look at your priorities for the next year, how do you balance resources that are dedicated to a server road map that's Intel first versus Intel -- doing an NVIDIA product is still Intel first and that you had a great capital commitment associated with that, but how do you balance those 2?

We're going to walk and chew gum at the same time. I think we can do both. I'll remind you, it wasn't just the NVIDIA investment, which, by the way, hasn't closed yet. We're hoping to get that done either later this year or early next, depending upon the government shutdown, backlog being removed. But we had SoftBank, we also had the United States government. And so I think we did a lot in Q3 to kind of put the balance sheet in a much better place. Relative to that NVIDIA relationship, it's extremely important. I think that last time I heard, I think Lip-Bu and Jensen are meeting with the teams every other week.

In the off-week, Lip-Bu are meeting with the internal team. So there's a lot of focus on getting this right. The advantage for us on the data center side is it does give us a lock-in with NVLink and puts us sort of on a level playing field with Grace and Vera. We need to go off and prove that. But we think as a head node within NVLink, we are the best head node out there. Time will tell. I also think that what was important about NVIDIA was a nice endorsement of the X86 ecosystem, and it's also a multigenerational agreement. And so their engineers had to get comfortable not just with our current product road map.

But what's coming next and what's coming next after that. And they liked what they saw. And so we're going to work, I think, aggressively to get product out of the market, but we haven't given -- set a time line for that, and we're looking forward to that relationship only getting better over time.

Another one on the data center, you described the dynamic in which both across client and in server, shortages is one word, constraints is another word. You also talked on the last earnings call about [ N-1 and N-2 ] and how you're seeing kind of an increased flurry of demand there as well. Is that a structure just of not being able to get the leading-edge product because you're tight? Or is that because people are finding value in that kind of...

Yes, I'm glad you asked the question because I just want to clarify one of the things we said on the earnings call is that we're tight and it's most pronounced on [ 10 7. ] And because we said that there's sort of this narrative that people are only looking to buy our older parts and not our newer parts. That's not true. The constraints are most pronounced on [ 10 7 ] because that's still where the vast majority of our volume is. I'll take data center as an example.

Remember, our first [ non-10 7 ] data center part is granite, and that's on Intel 3. That's still very early in its ramp phase. And Granite is going well. It's just very early. Most of our volume is still on [ 10 7, ] which is why most of the supply constraints are on [ 10 7. ] And the demand surge we're seeing is not just for Granite, it's for Emerald Rapids and Sapphire Rapids as well. They are really good parts. I think if we have more Granite wafers, we'd be able to sell more Granite product. And so constraints are across the board, but most pronounced where we have most of our volume.

How does the dynamic in the data center and also in client impact gross margins next year? Because as you are more mature, I would imagine things can be beneficial, but then you also have the island dynamic as well, like maybe walk through gross margin trajectory.

Yes. I'm looking forward to a point in time where there's not 16 different things that influence gross margin. because there's a lot of complexity in the gross margin dynamic right now. Let me level set by just reminding you what we said at earnings is that roughly speaking, incremental gross margin rule of thumb for next year is in that 40% to 60% range. And that's the right way to think about it. Now when you kind of drill down into that, there are some dynamics that are going on that are gross margin positive. There are some dynamics going on that are creating some gross margin headwinds. On the positive side, we're deemphasizing the low end of the PC market next year, which should be gross margin positive.

When you look at Raptor Lake, there are SKUs where we are absolutely raising pricing, which should be gross margin positive. To offset that, though, we also are doing some demand shaping on Arrow Lake and Lunar Lake. And the reason for that is we know we're undershorting our PC customers by a certain amount next year, we don't want to undershort them by too much. And so as we move internal capacity from client to server, we're trying to augment that with more external wafers on Arrow Lake and Lunar Lake. And that is creating some gross margin headwinds as we do demand shaping. Probably the biggest being Lunar Lake because remember, at the end of Q2, I would have been fairly emphatically telling you that Lunar Lake volumes probably peak in Q3 are flattish sequentially into Q4 and then start coming down after that.

Given the way we're trying to manage this tight supply situation, that's no longer true. Lunar Lake is going to grow sequentially in Q4, and it's going to be up again next year on a year-over-year basis, more so in the first half than in the second half of the year. But especially with the embedded memory, that does create some margin challenges. Now it still puts us in that 40% to 60% range. But those are sort of the pluses and minuses that are going on. with gross margin.

Last point I'll make, we are still in the early ramp of Intel 18A. And clearly, in the early ramp of any nodes, yield is not where you want -- cost structure is not where you want that starts to get better every quarter. But I would say on balance, it's probably more of a headwind than a tailwind in the first half of next year, and then it reverses to a tailwind as we go into the second half of next year into '27.

Another minor thematic we're hearing about here and over the last couple of weeks is just memory shortages in general. You still have a large amount of exposure to consumer end markets, which require memory as well, so maybe not your problem, but your customers' problem. Are you hearing any kind of conversations from customers that you may see some muted demand trends with when it comes to memory?

Yes. I mean we're watching it closely, and that is part of the current customer conversation, albeit it is very secondary to the fact that we're not getting them the unit volume that they need. And so I know that there is concern out there that rising DRAM ASPs are going to have an impact on units. That's not the primary concern that customers have today, but it is a dynamic that we're watching. Quite frankly, given how tight we are if the PC market were a little bit less robust next year, probably wouldn't hurt us all that much, and it would be a little bit of a relief. Now having said that, I've gone back and looked at other periods of time where DRAM as the percent of the bill of material has started to get to uncomfortable levels. And if you look at historical times, the data is kind of inconclusive.

There doesn't seem to be a high correlation to units backing off or quite frankly, even mix changing all that much. I think we want to be respectful of the fact that it could be different this time, given how price inelastic the AI market is. Perhaps this lasts longer and goes higher. But historically, the data would suggest it hasn't been a huge influence on the overall PC market, but we're watching it as it unfolds.

Rotating to AI, I want to be respectful about you guys saying that things are on the come. Obviously, you've had a couple of products internally, stops and starts, but you said, hey, we're reassessing where we're going with this, and we'll have a very formal kind of view in the next couple of quarters. There's also news out around a potentially signed memorandum with SambaNova. Anything around your AI strategy, whether inorganic or organic that you can give us an update on at this time?

Yes. So when you say AI, I'm assuming you're saying AI accelerator because one of the things I keep to try to remind you, AI is an umbrella dynamic that's helping all of our businesses. We've got an AI PC strategy quite frankly, AI is helping us in advanced packaging. It's helping us in the wafer business. It's helping us in the traditional server. So let's talk a little bit about the accelerator market and how we're thinking about that. Lip-Bu has been fairly clear that as we think about the GPU portion of the accelerator market, we're not going after the LLM training, hyperscale accelerated data center.

We're really looking to be focused on a power-optimized GPU for inference as AI moves from the center out to the edge. We've got a lot of work to do there. So stay tuned. It's going to take some time. I think the other dynamic that Lip-Bu has introduced that's worth talking about is the ASIC portion of this business.

He, under his leadership, set up a structure called the central engineering group run by Srini, which is one of the individuals he recruited over from Cadence. And part of Srini's mandate is not just central engineering group, which is making sure that the teams are using IP blocks as efficiently as possible. He's also put in charge of a real ASIC business. And I think a lot of investors don't realize we've got a pretty vibrant ASIC business today. It's all in networking. We've got multiple customers there for doing SmartNIC ASICs. And that business is actually showing great upside similar to a lot of other networking companies helping to being driven by AI. I think the mandate that Lip-Bu has given Srini is to expand that business into an XPU attached like business, much like Broadcom and Marvell.

I think the other interesting dynamic that we have in this ASIC business is we can pursue it in a Broadcom Marvell type model. But given also Intel Foundry, there are a lot of hyperscalers that are looking to go directly to the foundry and to try to do that, to circumvent kind of the traditional model. We've got that opportunity as well. And one of the advantages we have is we just have a ton of system integration knowledge, which I think is very helpful as we go off and prosecute that. I want to set the right expectation. Both The GPU and the ASIC sort of businesses are going to take time to develop, but we're pretty optimistic about what they could do longer term.

Very helpful, John. Thank you for joining. We're running out of time here. I appreciate it. It sounds like some good things ahead.

Thank you. .

Thanks a lot. Appreciate it.

Okay. Good afternoon. I'm Tim Arcuri. I'm the semiconductor analyst here at UBS. And very pleased to have Intel. We have John Pitzer with Intel. John and I competed for many years, and this used to be John's conference. So anyway, John, you're back.

You've done a fantastic job, by the way. This has been a great day. So I appreciate that.

Perfect. So anyway, Okay, John. Before we start, I have to read a statement on your behalf. Before we begin, please note that today's discussion may contain forward-looking statements that are subject to various risks and uncertainties and may reference non-GAAP financial measures. Please refer to Intel's most recent release and annual report on Form 10-K and other filings with the SEC for more information on the risk factors that could cause actual results to differ materially and additional information on our non-GAAP financial measures, including reconciliations where appropriate to the correct GAAP financial measures. So thank you, John.

Thanks for reading that. I appreciate it, and thanks for the opportunity to talk to everyone today.

Perfect. So John, let's talk about just the shortages that you're experiencing right now. There is -- demand is definitely better. But part of it also is that you have tightness on Intel 10 and 7, and you are diverting products or you have a mix issue such that you're short on supply. So can you actually talk about both PC and server demand and also supply?

Yes. It's a good question, and thanks, everyone for joining this afternoon. I mean, listen, I think I'll break this down into the PC market and in the server market. I think we've been in a relatively robust PC environment all year, Tim. I think in the first half of the year, we were concerned that the strength was really tariff pull-in, people trying to get ahead of potential higher costs in the second half of the year by buying PCs in the first half of the year. I think clearly, as we went into Q3, that concern kind of dissipated. And we think there's some sustainability to what's going on in the PC market. I think more importantly, the bigger shift was really what happened in the server market in Q3. I mean, it was a relatively okay market in the first half of the year, but it wasn't sort of a very strong market in the first half of the year. As we went into Q3, I think clearly, customer conversations changed. Their demand signals changed. And it wasn't just near-term volume. And as we talked about on the earnings call, we now have multiple CSPs that are looking for longer-term supply agreements that stretch out beyond next year. And that has put us into an uncomfortable position, as you pointed out, that we now are undershipping demand, both on the client side and on the data center side. We said on the earnings call that the shortages are most pronounced in 7, 10, but really, we're short just about everywhere. The reason why they're most pronounced in 7, 10 is that still the mainstream of our volume today, both in client and in data center, which is why we called that out. We also talked about the peak about the supply shortages occurring in Q1, that's going to be the most difficult quarter for us to get wafers out. Things will get better beyond that, but we're still going to be short apply even after Q1.

And so we have a situation where there's multiple hyperscalers now asking for long-term agreements. Can you also talk about that?

Yes. I think it's simply that. I think one of the things that I get asked often, and we're trying to get our arms around is what's really going on with traditional server demand. And I think there's probably 3 dynamics at play. One, you have relatively easy compares for the last several years, the hyperscale community has not been refreshing their installed base and been rightfully going after AI infrastructure build. And so they put that a little bit on the back burner. I think the second dynamic, which is somewhat related to the first is, they're also very power constraint right now. And if you were to take -- and we talked about this on the earnings call, a server in the installed base that's 5 years old and you were to upgrade into a new server, that new server is going to be 80% more power efficient. And so there is a significant ROI to actually start more aggressively sort of refreshing the installed base. The third dynamic, which I think is probably the most interesting one is just, I think, as you see AI transformed from LLM training into a agenetic and inference, it's just absolutely putting demand on traditional infrastructure that I think surprised us. Quite frankly, it surprised our customers. And I think that's one of the reasons why we're seeing sort of a desire by the CSPs to get some longer-term supply assurances in place.

And do you think that, that drives some upgrades? Do you think that it drives an upgrade cycle?

I think clearly, it's driving capacity needs, quite frankly. I think as you think about the LLM and the training model, all the spend has been done for that. But the LLM is basically the brain of AI. It still needs your personal data for you to get real insights out of it. And that's really a rag model that is putting, I think, tremendous demand for traditional compute that has some sustainability to it.

John, one thing I hear, and you actually got these questions coming out of earnings was that, well, the customers aren't buying the new products. So there's some perception that customers are still buying the old products, that's part of all of this. Can you...

Yes. I understand the question. I would remind you that the vast majority of our capacity today is still on 7, 10 and that's why we're tightest there. Quite frankly, if we had more Granite wafers, we'd be selling more Granite. If we had more Lunar Lake wafers, we'd be selling more Lunar Lake. If we had more Arrow Lake wafers, we'd be selling more Arrow Lake. And so I think we feel pretty good about where we are in the AI PC transition. And we feel very good about where we are in the initial phase of the Granite ramp, which is our latest generation server part.

Great. So let's talk about 18A. And Panther Lake is out. Yields, I think, were initially unpredictable and low, but I think that's back on track. So can you just talk about 18A and progress. It seemed like Lip-Bu was a little more bullish on the progress last call?

Yes, it's a good question. I think that we feel good about the progress we're making on Intel 18A. We had committed to everyone to get our first SKU of Panther Lake out by year-end. We have done that. I encourage everyone to get to Las Vegas in early January because you'll hear us and more importantly, our OEM partners talk a lot more about the excitement they're seeing around Panther Lake. We'll ramp multiple additional SKUs in the coming months in the first half of next year. What I would tell you is, we're still not satisfied with where yields are. But relative to when Lip-Bu came back to the company in March, he was clearly unhappy with where yields were absolutely. And to the point that you made, he was probably more displeased by the fact that progress on the yields were very unpredictable, very haphazard. I think that he kind of tasked the team himself and Naga to really double down and focus on getting yields right for Panther Lake. And I see -- I think we're starting to see the benefits of that because at the very least, yields are still not at the levels we want them to be and they will continue to get better over time, as Dave talked about on the earnings call. But we are now in a position where we're seeing predictable improvement month-on-month, that's in line with what you would expect as an industry average.

Great. And then can you kind of dovetail that into 18AP? And you've made progress on the PDK milestones and the plan to leverage the 18A family. And you've said you're going to keep on leveraging that family for the next 3 generations of both client and server. So when would you expect the sort of to get more news on 18AP and for us to have more milestones?

Yes, it's a good question. I think there was a little bit confusing coming off the Q2 earnings call. 18A is both a node and it's a family. And I think when you think about 18A as a family, there's 18A, there's 18AP and there's 18APT. 18A is going to be mostly an internal node. As we've transitioned to 18AP, and we've seen good maturity on the PDK there, it will be used both for the internal customer, and we'll reengage external customers with that node as well. And we've got good engagement with customers going on now with Intel 18A. And like I said, the PDK maturity has come along nicely. In addition, 18APT will be our base [ dye ] for advanced packaging, and that will be used for both internal and external customers. And it's one of the reasons why on the Q2 earnings call, we kind of went out of our way to just tell people, we think this is going to be a very healthy, long-lived node that generates a very good ROI. I think Dave talked about peak wafer starts for 18A, really not happening until the 2030 time period. And so this is a capacity point that we think is going to be a good ROI for our investors.

Great. Can we just -- can you remind us what the outsourcing ratio is going to be? I think you said it's 30% this year. Can you remind us sort of some milestones, maybe the direction of it in '26 and '27?

Yes. So it's a good question. So what's been driving the outsourcing ratio this year is really the fact that both Arrow Lake and Lunar Lake are parts that are fully outsourced to external foundries. As we start to think about Panther Lake, that starts the process of actually bringing wafers back in-house. And what we've said in the past and it's still true today, is that about 70% of the logic compute -- the logic tiles for Panther Lake will be done internally. That's versus 0% for Arrow Lake and Lunar Lake, that gets even better with Nova Lake because remember, Panther Lake is just a notebook part. Nova Lake is also desktop. So as we move into the end of next year, into 2027, we'll get the added benefit of bringing desktop wafers back as well. Now what I will tell you is we try to navigate through this current period of tight supply, we are going to lean more heavily on our external foundry Partners around Arrow Lake and Lunar Lake. And you and I have talked about this. I think one of the bigger changes from Q2 to Q3 is at the end of Q2, I would have told you that Lunar Lake probably peaks from a volume perspective in Q3, it's kind of flattish in Q4 and then starts to fall off after that. I think now as we think about trying to augment the shortages in PCs as we move client wafers to data center, we're leveraging more Arrow Lake and Lunar Lake. Lunar Lake is going to grow sequentially in Q4, and it will be up next year year-over-year. And that has some challenges at the gross margin level because of the embedded memory that is really a pass-through cost for us.

Great, John. Can you also then extend that to 14A? How is the node development progressing? What are the next milestones we can expect on 14A?

Yes. We're pleased with the early engagement. As I've talked about in the past, simplistically, I think about every node having 3 phases, sort of definitional development and then high-volume manufacturing. 14A is a very different node for us because on the 18A node, in definition, we were only engaging with Intel products. And all the decisions we made at the transistor level was really to optimize for what they needed, which was a CPU tile. It wasn't really until the development phase that we actually started engaging external customers. And to be fair, many external customers, the fact that we optimize for Intel products didn't matter. For others, it did. And we underexecuted on 18A, and had we executed better, we probably would have had better results to show. I think importantly, on 14A, we are engaged with external customers in the definitional phase, and quite frankly, we're making optimization choices more to favor what they need than necessarily the internal customer. And what that means is we're getting earlier, more and better feedback about how we're doing on the node.

The other point that I'll make is, as you know, with 18A, we tried to bend physics in 2 new ways, right, moving from FinFET to gate-all-around and then implementing backside power. There's a lot of complexities on 14A, but at the very least, it's a second-generation gate-all-around. It's a second-generation backside power. And we've been very clear that if you look at yield and performance today on 18A and compare it at a similar point in time on 18A, we are meaningfully ahead on the 14A node, which is also encouraging.

And then the third thing that helps is kind of tongue-in-cheek like to joke that we didn't know how to spell PDK until we were about halfway through Intel 18A development. And quite frankly, we struggled a bit with getting our PDKs to be very industry standard like. We're not having that problem either with AP or with Intel 14A. And so I think that customers are pleasantly surprised that our PDKs are just much easier to use.

So what's the next most important milestone that we should be listening for on 14A?

It's really customer announcements, quite frankly. There will be incremental progress that we'll learn. But I think the big one is really, when can we get an external customer on either Intel 14A or Intel 18AP. And what we've said on that front is the time line to think about for when fabless customers have to start making hard decisions about 14A designs, that window opens up in the second half of next year into the first half of 2027.

And even at the same time, though, you said that even if you do sign a customer, you might not even tell us about it. So is there anything else tangible besides just to...

No. I mean, listen, I think we have a guiding principle that our job as a foundry is to try to make our customers successful. It's not to try to market their names. And clearly, I think if the customer would like to put out a press release, we're not going to be adverse to doing that. Quite frankly, I think we've had some good success with some advanced packaging customers that we haven't talked about. And I think that that's just par for the course of a good foundry. You don't see other external foundries talking about specific customers, and we're going to follow that example.

Great. I actually wanted to ask you about advanced packaging. So the external advanced packaging revenue is still pretty small, but we do hear about a lot more engagements there. And you've been very clear that the back-end engagements will lead the front-end engagements. So would we think we're about to see an inflection point here with respect to EMIB and EMIB-T given some of the tightness we're seeing for CoWoS.

Yes. I mean we're pretty excited about the technology. I mean if you go back and think about the journey we've been on in advanced packaging, we were pretty excited about the business about 12, 18 months ago, in large part because we saw a lot of customers coming to us to be spillover capacity because CoWoS was so tight. And for full transparency, we probably under hit the potential for that business. I think TSMC did a very good job increasing CoWoS capacity. We probably underperformed a little bit and getting Foveros where it needed to be. But the advantage of having that happen is it brought customers in the door and it allowed us to start moving from tactical conversations to strategic conversations. And when we did that, and we started to show them our road map against EMIB and EMIB-T, they were pretty excited about what they saw. And we do think that EMIB brings capabilities into the market that you just can't get with CoWoS. And so it's not a spillover. It's really a technology [ buy ] that people are focused on. We've got some good momentum there. I think the proof is in the pudding is to when we actually start to see revenue move through the P&L. And I think that's probably something that starts to ramp in the back half of 2026.

Got it. Let's talk about how you prioritize capacity between server and PC. Obviously, these hyperscalers are coming to you, wanting to have LTSAs and they're talking about that. Obviously, and you've also said that you're prioritizing wafers to server. And so -- but at the same time, PC is getting better as well, demand. So are you happy losing some PC share? You're okay with that through '26 because -- I asked that because AMD put up a pretty aggressive share gain target in a PC.

Yes. I wouldn't say happy, never happy in losing share. But I do think as we think about how we're going to navigate through this tightness in supply, we've been fairly clear that we're going to deemphasize the low end of the PC market. Quite frankly, that's an area where our closest competitor actually doesn't play. So it's going to be interesting to see how market share trends play on that front. But we're really going to focus on trying to optimize for revenue and profit share in the PC market and our tight capacity. Now we're doing things in that business today that are both gross margin accretive and also create gross margin headwinds as we've talked about. I wish our gross margin story was a little simpler. But for example, deemphasizing low end, positive for gross margins. On some of our lower-end SKUs, we are going to raise pricing, positive for gross margins. Offsetting that, though, we are going to work closely with our OEM customers because we know we're undershipping demand. And we will do demand shaping and change pricing on Arrow Lake and Lunar Lake that allows our OEM partners to bring those CPUs and into lower price points within the PC stack. And that obviously is going to create some gross margin headwinds that we need to manage through next year.

And just on that front, Lip-Bu has been kind of rethinking the entire data center road map. And I think Pat was maybe a little more bullish on peak on E-core. And the idea was originally that Diamond Rapids was going to stem some of those share losses. It sounds like now maybe that optimism is pushed out to Coral Rapids. Can you just talk about how the data center server road map is changing?

Yes, it's a good question. I mean, I'll remind people, we're still very early in the Granite Rapids ramp. And we actually do think Granite Rapid does a good job of helping to close gaps in many parts of the market. And so we should benefit from that over the next several years. As you think about Diamond Rapids, Lip-Bu has made proactive decisions to take some of the SKUs off of the road map, mainly at the low end, where he came to the determination along with Kevork, that we're just not competitive on cost and we're just not competitive on performance. And while it's not an ideal position to be in, it's actually very healthy for him to do that because it sends a very strong signal internally that we're just not going to accept subpar products. And it also frees up resources to go off and work on products that we think will be competitive. As you know, Tim, it takes multiple years from cradle to grave to get a design out the door and into a product. I think when you think about the opportunity that Kevork and Lip-Bu have to really embed their DNA into a server part, that first opportunity is probably going to be Coral Rapids.

Coral Rapids. Okay. So John, let's just talk about PC. Do you feel like you're losing some ground in PC? I mean -- and how do you think about the -- you're fighting off AMD, you're fighting off ARM. How do you think about your PC road map?

I mean we feel pretty good about the PC roadmap. Quite frankly, I think that this year was a challenging year for us at the high end of the desktop market. We've been pretty, I think, transparent about that. I think a lot of the damage that has occurred there has already happened. I think that -- as you think about Panther Lake, I think it shores up our position in the notebook market. And as you bring Nova Lake late next year into 2027, we'll have a competitive part across the PC stack, across both notebook and desktop. And so we have very strong share there today. With Panther Lake next year, I think share stabilization, notwithstanding the fact that we might deemphasize the low end of the market, just to be mindful of that, that could impact kind of the unit share number, but share stabilization from a road map perspective is what I would think about. And as we bring Nova Lake to market, I think we are very happy with our competitive position.

And just on PC John, how do you -- you've been following these markets a long time. How do you think about memory price increases impacting the market? I mean it's historically been -- at least people think of it as being a fairly elastic market. So can you just talk about that?

Yes, it's a good question. I mean this surfaced in mid-November, with the market got very concerned about this. And so I spent time with the team kind of going through some of the historical data and just looking at DRAM as a percent of the BOM, how it's too over time. And what happens to units and mix when it gets to sort of a pain threshold because we're clearly going through that today. I thought what was interesting is, the data is not conclusive that it has an impact on either units or mix, quite frankly. I know it creates a lot of chatter in the marketplace but the supply chain is pretty resilient about figuring this stuff out. Now it's clearly a dynamic that we're hyper focused on. I will tell you today, if you go out and talk to our OEM customers in the PC market, what they're most concerned about is our inability to get them units. And so that's not a great place to be in, but at least a high-class problem and not a low-class problem. But we will monitor the situation very closely. And quite frankly, if on the margin, the market is a little bit less robust next year, in a market where we're having trouble meeting demand. I'm not sure it's going to bother us all much.

Great. John, let's talk about AI. Can you just update us on where the road map is? I know there was some optimism sometime back about Falcon Shores. And just talk about where the accelerator road map is?

Yes. So you're asking a specific question around AI accelerators. I'd like to remind people that AI kind of permeates almost everything we do. There's the AI PC strategy, there's clearly the uptick in traditional servers that's being driven by agentic AI and inference. I would say our advanced packaging is being driven by AI. Quite frankly, our wafer business is being helped by the AI trend as well. But specific to the accelerator market and the GPU side, we clearly have work that we need to do. I think Lip-Bu has defined a strategy that suggests the part of the market where we think we can make a difference and solve a customer problem is really on GPUs that are optimized for power, for inference. We're not going to go and try to duke it out in sort of the training market and the AI accelerator business. The other thing that we talked about on the last call, which is relatively new, is this new ASIC business that Lip-Bu has put under Srini. ASICs is actually an area where we have a pretty healthy business today. It's mostly in networking, but we have multiple customers there, and I think people would be surprised by the revenue run rate we're seeing in ASICs. And quite frankly, as you know, AI is driving a lot of demand for networking. And if you see the forecast for revenue for that business, it's following sort of the industry trend of what we're seeing in other networking assets driven by AI. I think what Srini is tasked to do is to build that into a real business. We've sort of treated it as a hobby. But I think he's going to be, I think, much more focused on trying to bring whether it's an ASIC CPU or an ASIC accelerator to the market. And I think we have a lot of the IP blocks that are needed. And I think Lip-Bu [ in ] his customer conversations has seen a real need here. So we're pretty excited about that opportunity.

Can you talk, John, you've been there now for several years. Can you talk about Lip-Bu coming on board? How has the company changed under Lip-Bu?

Yes, it's a good question. I mean at the end of the day, whether or not we're successful in our strategy or not, it's going to come down to culture and incentive structures. And I think Lip-Bu has come in and been very aggressive in trying to change culture and incentive structure. And I think some of the visible signs is, I think when he joined the company back in March, we had 11 or 12 layers of management. We've now taken that down to 5 or 6. And so he's flattened the organization tremendously. In addition, we sort of had an engineering group that didn't report directly into the CEO. It reported up through the business units. He now has engineering sitting on his executive staff. In addition, those engineers who never want to go talk to a customer, it's now mandated that whenever we have a customer conversation, there's an engineer there to talk with the customer. And so it's really getting back to kind of an engineering-focused customer-centric sort of culture with a lot of transparency. I mean, again, I think the decision that Lip-Bu made on some of the SKUs in DMR is very healthy for changing the culture of the organization because historically, if we were to miss milestones on new products, we usually just gave them exemptions. Saying no is a pretty powerful tool. And I think that that's kind of being viewed in spades, and I think it's healthy. In addition, last point I'll make is return to office happened on September 2. It's great to get people back in the office. We're back in about 4 days a week now, and it really is helping to drive incremental collaboration.

John, you used to be an analyst. So you sort of think about the company being on the inside. I'm sure you also think about it still in your old hat, too. So oftentimes, people argue, well, Intel should be broken up into the foundry and into a fabless chip business. How do you think about that? I mean it's easier said than done. You don't have any external customers now for the foundry. So it's not -- so it's kind of a moot point right now. But how do you think about that? Like what boxes have to be checked for you to get to that point if you were going to do that?

Yes. Listen, we're clearly taking steps to create that optionality, right? And a lot of that is being driven by engagements with external customers who would like to see more separation between the Foundry business and the rest of the company. And so we've got an advisory board today for Intel Foundry. We're moving that into its own legal entity. And so we continue to move down that path to give ourselves that optionality. To your point, financially, pragmatically, it's probably not something we can do today. You've all seen sort of the split out of the 2 P&Ls. And I think first and foremost, Intel Foundry has to get healthier financially. I think we're on a path to do that as we bring 18A wafers in. But I think from a longer-term perspective, if it's the right decision to create value to create more separation, I'm pretty confident that Lip-Bu, Dave and the Board will move in that direction.

And John, one of the big keys to improving profitability is to get off of Intel 10 and 7 and the wafer costs don't change very much, but the wafer prices go up a lot. So that helps margins in Foundry. However, the transfer pricing to the product business goes up alongside of that. So is the product road map in a good enough place that it's able to absorb those price increases when you think about this transfer pricing model.

Not as good as we would like. We have work to do there. I would say client probably further ahead than the data center road map. But I want to be very clear, as we move to Intel 18A, it's very accretive for Intel Foundry margins. It's also accretive to Intel corporate margins. To your point around the higher price that Intel products needs to pay for 18A wafers, it does create some mixed dynamics within Intel products that they need to go figure out. But net-net, it is absolutely a positive to bring wafer home.

But doesn't it? So it improves profitability in Foundry, but it brings profitability down in the product business?

Yes.

But you think it's still..

All else being equal.

Still a net...

It's still in net positive.

Got it. And then let's talk about CapEx, John. So I think the original plan was this year, it sounded like it was going to be down, and now it sounds like it's biased down. But can you talk about CapEx in light of you being short of supply?

Yes. I mean listen, we're trying to balance 2 dynamics. One, we still fundamentally believe that we can run our fabs more efficiently and just drive more output out of the current footprint, which would argue for CapEx being down next year. On the other hand, we've got a situation where we're under supplying the market, and that's a very uncomfortable position for us to be in and for our customers to be in, and we want to respond to that. I will say that I do think that Lip-Bu and Dave fundamentally are willing to run fabs probably structurally tighter than the historical model would. And we're trying to balance that a lot. I think you characterized it well. We guided gross CapEx to about $18 billion for this year. We were emphatically saying, down at the end of Q2. We were less emphatically saying that at the end of Q3, but I think that, that's still directionally right, but it's a plus or minus for next year at this point. Now I'll also point out we are not prebuilding capacity for external foundry. So if we were to win a significant external foundry customer, that would put upward pressure on our CapEx.

Got it. And then can you just talk about -- I'm not asking you to guide gross margin for next year, but can you talk about just the puts and takes? And is the 40% to 60% drop-through still the right way to think about gross margin as we kind of go into next year?

Yes. It's still the right rule of thumb. I'm looking forward to a point in time where the number of sort of levers in gross margins goes down because there's a lot of moving parts right now. And it is hard to forecast, especially because it is so mix dependent. I would tell you that the biggest sort of incremental change from Q2 to Q3 is something I mentioned earlier, which is just the outlook for Lunar Lake, which we thought would be peaking right about now. It looks like it's going to continue to grow as we augment of shortness of client supply and what we build internally with more external wafers. And that is going to put pressure. I think we've done a good job, I think, pre-buying a lot of the memory for Lunar Lake, especially given what DRAM pricing has done, but we still have some more do there.

And John, last question. How is the involvement of the U.S. government changed the behavior of the company? Has it changed anything at all?

No, I don't believe so. I mean, at the end of the day, even before the U.S. government was an equity holder in the company, they were a critically important stakeholder in the room. And you see that across the technology supply chain. The number of CEOs that are making visits to Washington, D.C. I think they're visiting D.C. now more than they're visiting New York City. I think having them as an equity holder was the right decision for us and for our shareholders, it fully aligns incentive structures. And we're glad that they're a good partner of ours.

Great. Well, John, we're out of time. Thank you again.

Thank you. Appreciate it.

I am Srini Pajjuri, I'm the semi analyst here. And we're delighted to have Intel today, John Pitzer, who heads the IR team there and also the CVP of Treasury. Before we start, John asked me to read a forward-looking statements, so I'm going to try this.

Before we begin, please note that today's discussions may contain forward-looking statements that are subject to various risks and uncertainties and may reference non-GAAP financial measures. Please refer to Intel's most recent earnings release and annual report on Form 10-K and other filings with the SEC for more information on the risk factors that could cause actual results to differ materially. And additional information on Intel's non-GAAP financial measures, including reconciliations where appropriate to the corresponding GAAP financial measures.

Thanks for that, Srini.

Yes, that wasn't easy, but, hey, made it. Okay. Yes. Thanks for joining, John. So it's been about 8 months since Lip-Bu joined and a lot of changes gone through a major restructuring. I would say, the past 3 months, a lot of news flow, mostly positive. So maybe start with that now that the restructuring is done, the balance sheet is in pretty good shape. And also you got into a partnership with NVIDIA. That certainly has been quite positive. So let's talk about what's on the plate for the management as we look out to the next 6 to 12 months. Maybe we can talk about what are the top 2 or 3 priorities for the management going forward?

Yes, it's a good question. And by the way, thanks for inviting me and appreciate the time this afternoon with everybody in the room. I mean listen, I think at our core, what Lip-Bu is trying to drive as far as the transformation is really cultural. And so the top priority, there's a lot that we're trying to get done at the business unit level. but it all really is built on a foundation of getting the culture right. And I think Lip-Bu has talked about this in several settings. He really wants to get Intel back to being engineer-focused customer-centric organization. And I think as part of the restructuring, we've done, I think we've done a lot to really simplify the organization, take bureaucracy out, drive, I think, better, faster decision-making.

And I don't think you're ever done getting culture right. And so I think if Lip-Bu were here, he would still stress that getting the culture right is really the foundation of changing a lot of what we need to change at the business unit level. Now I know that wasn't your question. Your question really was at the BU level, what are we focused on? I think priority #1 is really getting a good and successful launch of Panther Lake, which also coincides with yields and yield improvement. I feel really good about the fact that we're going to get our first SKU out by end of year. I'll put a plug in for people in Vegas in January at CES. You're going to hear the CCG team talk a lot about Panther Lake and our OEM partners talk a lot about Panther Lake, and we're really excited about that launch.

I think sticking with Intel Foundry, clearly, landing at Intel A external customer is pretty critical over the next, call it, 6 to 12 months. And I'm sure you'll have some questions around that. And then as we pivot to sort of the business units, I think job #1 is to really, I think, establish a road map in both CCG and DC AI that drives both market share gains and gross margin improvement. I know you've got some questions there. So I won't elaborate. And then I think finally, it's really prosecuting our accelerator strategy in AI. That's both on the ASIC side of the house. and on the GPU side of the house.

Okay. That's a good start. Maybe we can start with the NVIDIA partnership because that gets a lot of attention. So NVIDIA is investing $5 billion in Intel. At the same time, you're going to be part of NVLink Fusion. Maybe talk about what's your current position is when it comes to AI, head node CPUs and how this might actually change it, maybe enhances it. And also, if you can kind of touch upon what sort of workloads are we talking here because NVIDIA seems to be doing quite well with their own ARM CPU and now they're entering a partnership with you. So how do you see that opportunity? Are there different workloads that target? Just anything you can tell us about the timing and the potential opportunity of that.

Yes. It's a really good question, and we couldn't have been more pleased to make the two announcements we made with NVIDIA in the calendar third quarter. One was the collaboration, both on data center and client. The other was the $5 billion investment, which we hope to close by year-end. Relative to the collaboration, I think it was really important to get that done, and that was the culmination of over a year's worth of work. And so there was a lot of back and forth between both companies at the core engineering level to really drive that collaboration. And I think it really is an endorsement of at least two things. One, really the criticality of the x86 ecosystem.

As much as there are new applications around AI, AI still fundamentally sits on top of and leverages compute architecture that's been x86 for the last 40, 50 years. And I think NVIDIA saw value in getting locked into that x86 ecosystem. The other really important endorsement, this is a multi-generational agreement. And so you can imagine the amount of tire kicking that the engineers at NVIDIA did looking at our road maps, both in client and in data center over multiple generations. And quite frankly, they like what they saw, now to your specific question, what does it do for us? I mean, clearly, today, we've got a pretty strong position in head node CPUs going into AI accelerated servers.

As NVIDIA did more with both Grace and what will be Avera, there were some questions that investors were asking about our ability to sustain that position. And I think the disadvantage we had prior to this collaboration is that both Grace and Avera I think, utilize the NVLink integration that we now have with this collaboration with NVIDIA. And so the way the data center side of the relationship is going to work is we will provide them with a custom Xeon part that they will then integrate into their system and they will have the responsibility of going to market, and we'll get all the benefits of having that NVLink fabric with our custom Xeon.

If you look at the client part, I think that clearly, we have the opportunity to build really a new class of PC parts that we're pretty excited about. And the way that, that relationship is going to work is they will provide the graphics tile through bailment, which means that the customer will actually pay them for the graphic style. But we will be responsible for integrating that graphic style with our CPU and bringing it to market. And the reason why we had this bailment agreement is we didn't want to have the same economics around this NVIDIA collaboration that we have today, for example, with Lunar Lake, where we have the embedded memory which is really revenue at 0-calorie gross margin.

And so that's why it works best for us to move down this bailment path. And we're pretty excited. We think that it's TAM expansive both in the data center and in the PC market, and we're looking forward to getting product out to market as quickly as possible.

Yes, it makes -- so just to be clear, so, this is a custom CPU that you're going to sell directly to NVIDIA? This is not going to third-party hyperscaler?

That's correct. And they will integrate it, and they had the go-to-market responsibility on the system -- at the system level.

Got it. And you haven't talked about the timing or potential?

We have not. I mean, I think clearly, this is a key focus. I actually think that Lip-Bu and Jensen are having meetings every other week with the team to do deep dives. And so our intent is to get this to market as quickly as possible, but we haven't set a time line.

Got it. Got it. And then on the PC side, client side, I mean you have your own graphics development, right? Today, you sell a lot of integrated GPUs. So I mean when you say you're going to kind of package NVIDIA's RTX into Intel CPUs. How does that mean are you targeting a particular market within video solutions? Or what happens to your own internal tailwinds?

Yes, I think we're going to continue to pursue our own internal strategy. Just like on the data center side, NVIDIA will continue to pursue their ARM strategy with Grace and Avera. Time will tell as to what portion of the market this will actually cover, but we are going to be able to bring, I think, a new level of performance on graphics to a notebook type class PC, clearly initially targeting the high end, but there are aspirations that we can broaden the market further as we develop this relationship more.

Okay. That makes sense. Maybe switching gears a little bit here. I mean, talking about the overall AI strategy for Intel, right? On one hand, you just mentioned -- there's a lot of opportunity on the x86 CPU itself. And then I guess on the PC market, you guys talked about potentially shipping 100 million ARPCs this year. But at the same time, when it comes to XPUs, it's supposedly a $1 trillion market and you're missing out on that. What's the strategy? I guess where do you think you can intersect that market? And what's the approach internally maybe anything you can tell us about?

Yes, I'm glad you asked the question the way you did because oftentimes, we over rotate to our AI strategy. Being really our accelerator strategy. And to your point, AI is driving a lot of what we're doing in the PC market today and in the traditional server market. It's also a big sort of driver in what's going on with the fab business on the wafer side and the advanced packaging side, which I'm sure you'll ask about later. But specifically, on the accelerator side, I think that we're still in the process of kind of bringing our strategy fully to market.

Lip-Bu has talked about the idea that we're looking at really an inference specialized to go after the inference part of the market. If you look at sort of what's going on in the hyperscale training market, we think that market is already well served with NVIDIA and with ASICs. And so it's really the opportunity to go after a genic AI, physical AI and really optimized for inference is what we're targeting.

Okay. That makes sense. And then there was also a mention of customization and ACE again on the last earnings call. So just to kind of level set, are we talking ASICs similar to what Broadcom and Marvell are doing? Or is this something different? What's the approach here?

Yes. I think we are talking similar to what Broadcom and Marvell are doing today. I'd like to remind people, we're actually an ASIC company today. We have ASICs multiple wins more in networking than not that now fall under Srini's purview in the central engineering group. And so ASICs are not new to us. We have a lot of the building blocks we need to be a broader participant here. They haven't just been optimally managed. And I think that's one of the reasons why Lip-Bu Tan brought Srini in from Cadence, to...

Not me, right?

Not you. Just to be clear. Different Srini. To come in and really drive that business. And this will be across x86. It will also be in for accelerators. And quite frankly, it's sort of agnostic. Could it be x86, yes? Could it be ARM-based, Absolutely. Does it need to be built in our fabs, could be? Does it have to be built in our fabs, No. And so I think that there's a real opportunity here. And quite frankly, one of the first things that Lip-Bu did when he joined the company as CEO was go out on a listening tour to customers. And one of the things he identified is that while customers are doing a lot of ASIC activity, they're not fully satisfied with the suppliers that they have today. And so we think we have a real opportunity here. It will take time. And I want to be very clear, we earn a journey -- but we're pretty optimistic about the assets that we bring to this market.

What do you think is the differentiator here? Obviously, Broadcom would argue that they've been doing this for almost 30 years since the day I mean there's still a and Marvell, they come from IBM legacy. What is that Intel brings to the table that the market?

I think the two biggest benefits we have is one, the x86 ecosystem that we've been investing in for decades and decades. And there's clearly importance of that. I mean even if you look at the ARM server market today, ARM has been very successful for internal workloads at hyperscalers. It's been significantly less successful for some of the external workloads. No reason why there couldn't be an X86 base ASIC to try to address that market for some of the hyperscalers. I think the other real advantage we have here is just system know-how.

And I think that that's going to become critically more important as you start thinking about some of the next-generation products that these hyperscalers want to bring to market.

Right, right. I mean there's a lot of chatter about even companies like Broadcom, potentially doing Iraq level solutions, right? I mean, you obviously have done historically some reference designs on that front. But do you feel that you have all the pieces that go into building a rack system at this point? Or do you think you need to acquire any of those?

Yes, it's a good question. I think that we have a lot of the IP blocks that we need. It doesn't mean that we have them all. doesn't mean that we have to acquire. It could be that we license. And so stay tuned on that front. And again, I want to be just clear. We're excited about this opportunity, but it will take some time for us to go off and execute.

Got it. And then obviously, we talked about balance sheet being pretty healthy right now. There was some speculation in the market about potential M&A. I don't expect you to comment on that. But I guess would you completely rule out M&A as part of your AI strategy? Because it looks like there are enough start-ups out there and have solutions out there that could be quite interesting.

Yes, it's a good question. And again, I'm not going to address any of the specific speculation that might be in the marketplace today. But I think clearly, there are a lot of strengths that Lip-Bu brings to the role as CEO. His turnaround at Cadence, his relationships across customers, suppliers and quite frankly, competitors and his knowledge of the ecosystem through his investments at Walden and I think that when the Board chose Lip-Bu to be CEO, they chose him because they wanted to leverage all of those strengths.

And so as you think about our M&A strategy, I wouldn't rule it out. I think, quite frankly, tuck-ins could make some sense. I also think partnerships could make some sense. One of the things that I think Lip-Bu and Dave have plenty of is pragmatism, and I think they're going to be very pragmatic about how we go off and prosecute this strategy.

Makes sense. And then just switching gears a little bit to the near term. server demand, it's been pretty healthy despite the concerns about ARM gaining share or AMD gaining share, et cetera. You talked about supply constraints, et cetera. But talk to us about the road map on the server side. AMD hosted their Analyst Day and they gave us some targets about market share. They're targeting more than 50% share. So I'd love to hear your thoughts on how you think about it, not just this quarter or next quarter, but as you look out to the next 2 to 3 years, what's the strategy to kind of stop those share loss and maybe potentially regain some share?

Yes. We've been pretty front-footed and clear on this. We're not where we want to be competitively on our solar road map, and we have work to do. I'm sure you saw that we hired a new head of the data center group recently Kevork coming in from Qualcomm, prior to Qualcomm, NXP. I think he brings a lot of strengths and knowledge from transistor to SoC to system, which is going to be important as we rebuild sort of the road map on that on the product front. Having said that, I also want to make sure that it's clear that we say server market as if it's one market, there are a lot of different segments within the server market today. And quite frankly, we've been very pleased by the ramp of Granite Rapids. It's still early in the ramp. Quite frankly, if we had more wafers, we'd have more revenue.

So we are going through a pretty tight supply situation. I think as you think about the road map from here, one thing I think it's important to highlight is, as you know, these product design cycles are -- take multiple years. And so as Lip-Bu came in the door in March and this year, a lot of Diamond Rapids was already baked from a design perspective. Clearly, I think he and Kevork are trying to make tweaks around the edges to do better there. And I think as you think about the different sort of flavors of Diamond Rapids, I think at the high end, we feel pretty good about where we are competitively. But really -- and I think Lip-Bu talked about this a little bit on the Q3 earnings call, it's really Coral Rapids, where I think we have the opportunity to really have sort of a clean sheet approach and implement some of the key, I think, IP blocks at Lip-Bu and Kevork think are necessary to bring our competitive product.

Got it. And then maybe we could touch upon a little bit on the margin side of things. Data center margins, if you kind of compare client margins are doing actually pretty well. On the other hand, data center is actually somewhat depressed versus history. So what is driving it? Is it more on the cost issue when it comes wafers? Or is it more R&D spending? Just any color on that.

Yes, I mean, listen, I would be very clear, I don't think we're satisfied where our margins are for either client or for data center today or for the overall company. And so we've got work to do. And I think we've got sort of a plan in place to show improving gross margins as we go throughout '26 and beyond. Having said that, on the data center front, I think that as we've been trying to catch up on performance, we've been over-rotating to performance over cost. And we haven't really been as cost-efficient as we could be. I think that's one of the, I think, muscles that Lip-Bu is bringing back to the organization. this notion that in order to be competitive, it's not just being competitive on performance, you have to be competitive on cost. And we think that our PC road map starts to get there with Panther Lake even more so with Nova Lake. And again, as I pointed out earlier, it's going to take some time on the data center side.

Right. Right. Yes. So just to follow up on the margin front. You did pretty well in the quarter. You did 40% better than expected, but your guidance suggests 3, 4 points of decline sequentially. And then you made some comments about next year, what are the puts and takes? Maybe we can kind of revisit them, right? So on one hand, the demand is very strong, your supply constrained. I think the pricing is generally pretty healthy. But on the other hand, you have new products ramping and you also have some mix headwinds. So as we think about next 4, 6 quarters, what are the 2, 3 puts and takes that are going to impact the gross margins?

Yes. Really good question. I'll start by reiterating, we only guide for 1 quarter out. But to your point, at the midpoint of our revenue, we did guide gross margins to about 36.5% in Q4, which would be down about 350 basis points sequentially. Let me kind of deconstruct that for the audience. About 50 basis points of that is just taking Altera out of the numbers. Now that we've completed the stake sale to Silver Lake, we've deconsolidated that. Gross margins at Altera were above corporate average. So 50 of the 350 is Altera coming out. Of the other 300 basis points, I would say it's plus or minus equally sort of distributed amongst 3 things. One, the early ramp of Intel 1A, which is always pretty expensive, especially as we're just getting wafers out of Oregon, it won't be until the beginning of next year that we start to get wafers out of Arizona with a much better cost structure.

And then it's pricing action that we're taking on Aero Lake and Lunar Lake to kind of navigate through this tight supply situation. And I think the tight supply situation is going to be here for a bit. We talked about on the earnings call Q1 being the peak of tight supply, but it will persist beyond Q1. And as we try to navigate through this, we are taking actions that are both gross margin accretive and create incremental gross margin headwinds. On the gross margin accretive side of things, we are favoring sort of servers over PCs. And we're probably deemphasizing the low end of the PC market. And we are raising pricing on parts in 10- and 7-nanometer Raptor lake because of the tight supply situation. On the other end of the spectrum, because we know we're shorting the market, and we're trying to do what's right for our customers, we are bringing price points down on both Lunar Lake and Aero Lake to fill different parts of the PC stack so that we don't undership the market by too much.

And those are sort of the puts and takes. I would tell you of all the things probably the most meaningful is Lunar Lake. As you know, we've got the embedded memory in Lunar Lake, and that does create some gross margin challenges. I would say at the end of Q2, we would have been pretty confident telling you that Lunar Lake volumes probably we're going to peak in Q3, be flattish sequentially in Q4 and then start to decline after that. Now with some of the demand shaping that we're doing, Lunar Lake is absolutely going to grow sequentially in Q4. and it's absolutely going to be up next year year-over-year. And that does create some incremental challenges, but we think that's the right trade-off as we try to do the right things for our customers.

It makes sense. When do you think Luna Lake will kind of peak out in terms of -- as a percent of the overall mix?

Well, I want to be clear, Lunar Lake is a very small portion of the mix. But given the embedded memory and the cost of that embedded memory, you don't need it to be a significant part of the mix to have a detrimental impact on margins.

Right. And there are no supply constraints on the Lunar lake front because it's mostly...

Well, it's a good question. I mean we clearly talked on our Q3 earnings call that we are having supply constraints. But I also think that there's some fly constraints in general across the industry, whether that be substrates with T-glass or quite frankly, some of the memory concerns that have popped up of late. And I think TSMC has done a fantastic job as a supplier getting us wafers, but they're tight as well.

Okay. So when you talk about supply constraints, we're not just talking about Intel 10, 7, which is in-house, but also the wafers that are coming from TSM are tight as well?

I think we have better availability of supply there because of some of the decisions we're making. Because remember, we are actively moving our internal supply away from PCs towards servers. In large part because we're undershipping the server market by a wider margin than the PC market. And so we want to make sure that we capture that opportunity.

Got it. Got it. And then obviously, Panther Lake and the yields, T&A, there's a lot of focus on that. Maybe looking at history, when you launch a new product like Panther Lake, how long does it take before that product kind of becomes I guess, a higher percent of the overall mix. When does that crossover happen?

Yes, it does take, I think, longer than people suspect. What I will tell you because we haven't been that explicit. As you look at the expected ramp in Panther Lake next year. And remember, Panther Lake is just a notebook part. And so if you compare that to the last 2 notebook launches that we had with Aero Lake and Meteor Lake, there's nothing unusual about the Panther Lake ramp as a percent of the mix. We clearly want to do better on the gross margin side. I think what's important is when Lip-Bu joined in March, he was unsatisfied by yields and he was unhappy that the progress on yields was sort of erratic.

I think one of the things that's changed dramatically over the last 7 or 8 months, is we now have a predictable path for yield improvement. We've talked about in the past that the industry average yield improvement on a new ramp is about 7% per month. And we are now on that curve for Panther Lake, which is giving us some confidence as we launched the product this quarter. And like I said, if you go to CES in January, you can hear a lot more about that.

Okay. And then just to kind of follow up on the margin comments. I guess on the earnings call, you sounded optimistic about margins improving through all of next year and maybe by end of next year, kind of getting to at an accretive level. I think that was a term use. So I guess it depends on the yields and you talked about them right now. But longer term, given that this is a new process for you and assuming that the yield shakeout where you expect them to be. How should we think about your longer-term margin potential here? I mean, I don't know when the last time you guys gave us a long-term model. But are we talking a 5% handling front of gross margin? Is it a 6 handle? How should we think about that?

Yes, it's a good question. Let me first back up a little bit on the lead up to your question. When we think about Panther Lake, absolutely, as we go through next year, increasing volume will help us come down the cost curve. I'll remind you that initially on any new process, we take wafers from Oregon. Oregon is where we do all of our technology development and then move into quasi high-volume manufacturing. Those wafers tend to be pretty expensive. And most, if not all, of the Panther Lake wafers this year are coming from Oregon. As we transition into Q1, you'll start to see wafers coming in from Arizona has a much better and different cost structure, and that ramps throughout the year.

As far as the more important part of your question, the second half of your question on the long-term gross margins, I'm not going to get ahead of Lip-Bu putting out a financial model we've discussed internally about when is the appropriate time to have an Investor Day. Stay tuned on that front. My guess is it's going to be some time in the second half of next year. But clearly, there's nothing in sort of the way we're thinking about our business. that wouldn't suggest that we should have industry comparable gross and operating margins with our fabless peers. And quite frankly, with the margin stacking of being an IDM, we should do a little bit better than that.

Okay. We'll definitely look forward to Analyst Day. I want to switch gears to the foundry side. Obviously, it's a challenging business. And I think you talked about A potentially leading to a breakeven point at some point. I think in the past, you said maybe 2027, I don't know if that's still valid or not. And then also, you made -- I think since Lip-Bu took over, there's also a comment that if you don't get a material customer for 14, you might actually stop process node development. So my question is, is 2027 still the target for foundry breakeven?

Yes. So what we've said historically is that the ramp of 18A, mostly on the back of internal products should be able to drive Intel foundry to be op profit breakeven on a run rate basis exiting 2027. And that is clearly where NAGA is still driving the organization. I will give you an important caveat though. When we win a customer for Intel 14A, we will have to layer on expenses well ahead of getting revenue. And so I do think for transparency purposes, as a sort of customer traction materializes, it's likely to push out that end I'm thinking though most investors will be okay with that because it will be confirmation that we can actually stand up an external foundry.

Yes. It's probably a good problem to have. We've got a couple of more minutes. I want to see if there are any questions from the audience. Okay. We'll keep going. So again, the 14A process node comment that caught a lot of attention, right? But at the same time, you also said until 2030, you're good with AT&A AAP for your internal product, so you don't have to. You actually don't need 14 for internal products, right? So talk to us about if you were to hypothetically give up 14 development, what does that do to your I guess, OpEx or CapEx? How does that impact the business model?

Yes. I always hate the hypothetical questions because they only get me in trouble. I want to be very clear. We are all in on the development of Intel 14. And we're feeling good with the early engagements we're having with external customers. And I do think it's important to point out that 14a is, in many ways, a very different node from an external perspective than Intel A. I mean I think simplistically, and I'm sure that the teams in TD are going to give me a hard time when I get back to the office. But I always think about any node having 3 phases. There's sort of a definitional phase, the development phase and then the high-volume manufacturing phase.

On 18A, in the definitional phase, we were only engaging with Intel products. It really wasn't until the development phase that we actually started soliciting feedback from external customers. which meant that a lot of the choices we made at the transistor level was really to optimize for the internal product groups instead of external customers. In addition, it was really our first foray into understanding PDKs, process development kits. And we had some growing pains on getting our PDKs to be true industry standard. I think the big difference on Intel 14A is that we are in the definitional phase engaging with external customers.

And what that means is we're getting earlier, more and better feedback on how we're doing from those external customers at 14A than we did at 18A, and our PDK maturity is much better. And we are now bringing to market industry standard PD both of which help tremendously. I'd also point out that at 18A, we were changing from FinFET to get all around. We were also adding backside power. We were making major changes. At 14, it's a second-generation gate all around. It's a second-generation backside power. And we have stated and been very clear. If you look at where we are today on 14A on performance and yield versus a similar point of development on 18A, we're significantly further ahead on 14. So we're feeling very good about 14. Now to answer your question, on the Q2 earnings call, when we first introduced the risk factor around 14A, Dave did mention that maintenance CapEx was sort of that high single-digit billions number.

That's the way you should think about our capital spending as we get through the 18A capacity build-out, if 14A weren't going to happen, but I want to be very clear, we are all in on 14.

Got it. That's all the time we have. And thanks, John. Appreciate it.