RBC Capital Markets Global (TIMT) Intel Conference 2025 Nov 18

I'm the semi-analyst, and we're thrilled to have Intel with us today, along with John Pitcher, the head of the CIR team, and the Vice President of the Treasury. Before we begin, John asked me to read a forward statement, so I'll give it a shot. Before we start, please remember that today's discussions may contain forward statements that are subject to various risks and uncertainties and may relate to non-capacity financial measures.

 

Please refer to Intel's most recent earnings update and Form 10-K annual report and other filings with the SEC for further information regarding risk factors that could cause actual results to differ materially and additional information regarding Intel's non-capacity financial measures, including reconciliations where appropriate for the corresponding non-capacity financial measures. Thank you for that response; I appreciate it. Yes, it wasn't easy, but I managed.

 

Thanks for joining, John. It's been eight months since Lip Bu Tan joined, and a lot has changed. They've gone through a major restructuring.

 

There's been a lot of news coming out in the last three months, mostly positive. So maybe we'll start with that. Now that the restructuring is done, the balance sheet is in good shape, and we've also partnered with NVIDIA, which has been quite positive, let's talk about what's on the table for management over the next six to twelve months.

 

Perhaps we can discuss the top two or three priorities for management going forward. Yes, that's a good question. And by the way, thank you for inviting me, and I appreciate the time we've had with everyone in the room.

 

I believe that what Lip Bu Tan is trying to achieve in terms of transformation is very much cultural. So, the top priority—there's a lot we're trying to do at the business unit level—but all of this is based on the foundation of having the right culture. And I think Lip Bu Tan has talked about this in several places.

 

He really wants to return Intel to being an engineering-centric, customer-centric organization. And I think that, as part of the restructuring, we've done a lot to simplify the organization, cut red tape, and make better, faster decisions. And I don't think we're ever truly finished with the right culture.

 

And so I think if Lip Bu Tan were here, he'd still be stressed that the right culture is really the foundation for changing a lot of what we need to change at the business unit level. Now, I know that wasn't your question. Your question was really, at the BU level, what do we focus on? I think the number one priority is really having a good, successive release of Panther Lake, which also coincides with an 18A performance increase.

 

I'm really excited about the fact that we'll have our first SKU by the end of the year. I'll be putting a plug-in out for people in Vegas at CES in January. You'll hear the CCG team talking a lot about Panther Lake, and our OEM partners will be talking a lot about Panther Lake, and we're really excited about that launch.

 

I think that being with Intel Foundry, launching an external Intel 14A customer is clearly quite critical over the next six to twelve months. I'm sure you'll have some questions about that. And then, as we focus on the business units, I think the number one priority is really establishing a roadmap between CCG and DCAI that leads to merchandise gains and gross margin improvement.

 

I know you have some questions there, so I won't elaborate. And then, I think ultimately, it's really about executing our accelerator strategy in AI, which is on both the ASIC and GPU sides. Okay, that's a good start.

 

Perhaps we can start with the partnership with Nvidia because that's generating a lot of buzz. Nvidia is investing $5 billion in Intel, and at the same time, you're going to be part of NVLink Fusion. Can you talk about your current position regarding AI headnode CPUs? And how this might change it? And perhaps improve it? And also, if you can, what kind of work are we talking about here? Because Nvidia seems to be doing very well with its own ARM CPU, and now it's entering into a partnership with you.

 

How do you see that opportunity? Are these different jobs that x86 will object to? Is there anything you can tell us about the timing and the potential opportunity? Yes, that's a good question. We couldn't have been happier with the two announcements we made with Nvidia on the Q3 calendar. One was the data center and customer collaboration, and the other was the $5 billion investment, which we expect to close by year-end.

 

Regarding the collaboration, I think it was very important to do that, and it was the culmination of more than a year of work. There were a lot of delays between both companies at the fundamental engineering level to make that collaboration happen. And I think it really supports at least two things.

 

One is the criticality of the x86 ecosystem. As much as there are new applications around AI, AI is still fundamentally on top of and running on a computer architecture that has been x86 for the last 40 or 50 years. And I think Nvidia has made a great effort to address the value and reach that x86 ecosystem.

 

The other really important support is that it's a multi-generational agreement. So, you can imagine the amount of delays Nvidia engineers have made looking at our roadmaps, both in clients and in the data center, across multiple generations, and frankly, they liked what they saw. Now, to your specific question, what does it make us? I think we have a very strong position today in headend CPUs going into AI-accelerated servers.

 

As Nvidia has done more with Grace and what will be Vera, there were some questions investors were asking about our ability to maintain that position. And I think the disadvantage we had before this collaboration is that both Grace and Vera, I believe, use the NVLink integration that we now have with this collaboration with Nvidia. So, the way the data center portion of the relationship is going to work is that we're going to offer them a custom Xeon component that will then be integrated into their system, and they'll be responsible for bringing it to market, and we'll reap all the benefits of having that NVLink manufacturing with our custom Xeon.

 

Looking at the client side, I think we clearly have an opportunity to build a whole new class of PC parts that we're very excited about. The way that relationship will work is that they'll be offering the graphics component at a loss, meaning the client will pay them for the graphics, but we'll be responsible for integrating that graphics with our CPU and bringing it to market. The reason we have this loss-making agreement is that we didn't want the same economics surrounding this collaboration with Nvidia that we have today, for example, with Lunar Lake, where we have embedded memory, which is essentially zero gross margin revenue. That's why it works better for us to move down the loss path.

 

And we're very excited. We believe it's in such high demand, both in the data center and the PC market, and we're looking forward to getting the product to market as quickly as possible. Yes, Max. So, to be clear, this is a custom CPU that you're going to sell directly to Nvidia; it's not going to a third-party hyperscaler.

 

That's correct. And they will integrate it and have the responsibility to go to market at the system level. Understood.

 

And you don't talk about time or potential. We haven't. I think, clearly, this is a key focus.

 

I think Lippu and Jensen are having weekly meetings with the team to do deep dives, so our aim is to bring this to market as quickly as possible, but we don't have a timeline. Understood. And then, on the PC side, the client side, you have your own graphics development.

 

Today you sell a lot of integrated GPUs. So when you say you're going to bundle Nvidia's RTX with Intel CPUs, what does that mean? Are you targeting a particular market with Nvidia's solutions, or what about your own internal strategy? I think we'll continue to pursue our internal strategy. Just like on the data center side, Nvidia will continue to pursue its strategy armed with Grace and Vera.

 

Time will tell what portion of the market this will ultimately cover, but we're going to be able to bring a new level of graphics performance to a notebook-type computer. Clearly, initially, we were targeting the high end, but we aspire to expand our market reach later as we develop this relationship. Okay, that makes sense.

 

Perhaps, going back a bit here. Speaking of Intel's overall AI strategy. On the one hand, you already mentioned that there are many opportunities in the x86 CPU.

 

And in the computer market, you already mentioned the possibility of shipping 100 million AI computers this year. But at the same time, when it comes to XPUs, it's supposedly a $1 trillion market, and you're losing ground on that. What's the strategy? Where do you think you can trade that market? What's the internal approach? Anything you can tell us? I'm glad you asked the question the way you did because we often pivot to our AI strategy, which is really our accelerator strategy.

 

And to your point, AI is driving much of what we're doing in the PC market today and in the traditional server market. It's also a major driver of what's happening with the manufacturing business on the sales and advanced packaging sides, which I'm sure you'll ask about later. But specifically on the accelerator side, I think we're still in the process of bringing our strategy to full market.

 

Lippu has talked about the idea that we're looking at a specialized GPU to target the inference segment of the market. If we look at what's happening in the hyperscale training market, we believe that market is already well-served by NVIDIA and ASICs, so it's really an opportunity to go after agentic AI, physical AI, and AI that's truly optimized for inference. That makes sense.

 

Customization at ASICS was also mentioned in the last call. Are we talking about ASICS doing something similar to what Broadcom and Marvell are doing? Yes, I think we're talking about something similar to what Broadcom and Marvell are doing today. I'd like to remind you that we are currently an ASICS company.

 

We have multiple ASICS holdings in the network, more than any other, which are now under Shreeny's permission in the core engineering group. So ASICS isn't new to us. We have many of the building blocks we need to be a broader player here.

 

They haven't just been optionally managed, and I think that's one of the reasons Lipu brought Shreeny in from Cadence. Not me, right? No, you. To come in and actually run that business.

 

And this will be via x86. It will also be available for accelerators. And, frankly, it's somewhat agnostic.

 

Could it be x86? Yes. Could it be assembled? Absolutely. Does it need to be built in our factories? It might.

 

Does it have to be built in our factories? No. So I think there's a great opportunity here. And frankly, one of the first things Lipu did when he joined the company as CEO was to go on a customer listening tour.

 

And one of the things he identified is that, while customers are doing a lot of ASICS activities, they aren't entirely satisfied with their current suppliers. So we believe we have a great opportunity here. It will take time.

 

And I want to be very clear, we're on a journey, but we're quite optimistic about the assets we're bringing 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. And Marvell comes from IBM's legacy.

 

What does Intel bring to the table in the market? I think the two biggest benefits we have are, first, the x86 ecosystem that we've been investing in for decades. And there's clearly importance to that. Even if you look at the ARM server market today, ARM has been very successful in supporting the internal work of hyperscalers.

 

It has been significantly less sequential for some of the external jobs. There's no reason why there couldn't be an x86 ASIC to try and address that market for some of the hyperscalers. I think the other big advantage we have here is the knowledge base.

 

And I think that's going to become critically important as you start thinking about some of the next-generation products these hyperscalers want to bring to market. That's right! That's right! There's a lot of talk about even companies like Broadcom potentially making IRAC-level solutions. Obviously, you've historically done some reference designs in that area.

 

But do you feel you have all the pieces that are going to go into building an IRAC system at this point? Or do you think you need to acquire some of them? That's a good question. I think we have many of the IP blocks we need. That doesn't mean we have them all.

 

It doesn't mean we have to buy them. We might license them. So stay tuned for that.

 

And again, I want to be clear. We're excited about this opportunity, but we'll have time to go and execute them. I understand.

 

And, obviously, we're talking about a balance sheet that's quite 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'm guessing, you know, would you start entirely with M&A as part of your artificial strategy? Because it seems like there are quite a few companies and solutions that could be quite interesting.

 

Yes, that's a good question. And again, I'm not going to address any of the specific speculation that might be circulating today, but I think there are clearly many forces that are bringing Lippu into the CEO role. His tenure at Cadence, his relationships with customers, suppliers, and, frankly, competitors, and his knowledge of the ecosystem through his investments in Walden.

 

And I think when the Board elected Lippu to be CEO, they chose him because they wanted to leverage all those strengths. So, if you think about our M&A strategy, I wouldn't say that. I think, quite frankly, that Tuckins might make some sense.

 

I also think partnerships could make some sense. One thing I think Lippu and Dave have in abundance is pragmatism. And I think they're going to be very pragmatic about how we get out there and pursue this strategy.

 

That makes sense. And then, going back a bit to the long-term service demand, it's been quite healthy despite concerns about the rise in ARM sales or AMD sales, etc.

 

You talked about supply constraints, etc. But tell us about the service side of things. AMD presented its analysis day and gave us some targets regarding market share.

 

They're targeting more than 50% of the sale. So I'd like to hear your thoughts on how you're thinking not only about this quarter or the next quarter, but also about the next two or three years. What's the strategy to stop those supply losses and, perhaps, potentially recover some supplies? Yes, we've been quite focused and clear on this.

 

We don't want to be competitive on our sovereign route. We have work to do. I'm sure you saw that we hired a new head of the central data group.

 

Recently, Kevork, coming from Qualcomm, and before that, NXP, I think he brings a lot of strength and knowledge from transistors to SoCs to systems, which will be important as we rebuild the roadmap for that, for the production frontier. Also, I want to make sure it's clear that we're referring to the server market as if it were a market.

 

There are many different segments in the server market today. And frankly, we've been very happy with the ramp-up in Grand Rapids. It's still early days on the ramp-up.

 

If we had more transistors, we'd have more revenue. So we're going through a pretty strong supply situation. I think when you think about the ramp-up from here, one thing that's important to highlight is that, as you know, these product design cycles last multiple years.

 

So, when Lipu took over in March of this year, a lot of the Grand Rapids were already set up from a design perspective. And, of course, I think he and Comborg are trying to make changes around the lines to make it better. And I think, when you think about the different flavors of the Grand Rapids, I think at the top we feel pretty good about where we are competitively.

 

But really, and I think Lipu touched on this a bit on the Q3 earnings call, it's really in Grand Rapids where I think we have the opportunity to take a clean approach and implement some of the major IP blocks that Lipu and Comborg believe are necessary to deliver a competitive product. Understood. And maybe we can touch on the margins a little bit.

 

Data center margins, if we compare them to the margins of clients who are doing quite well. On the other hand, the data center sector is a bit depressed compared to historical trends. So what's driving it? Is it more of a cost issue when it comes to profits, or is it more R&D spending? Whichever factor it is, I think... Yes, I mean, listen, it would be very clear.

 

I don't think we're satisfied with our margins for any customer, or for the data center today, or for the company overall. So we have work to do, and I think we have a plan in place to show improvements in expense margins throughout 2026 and beyond. That said, in the data center, I think that while we've been trying to achieve performance, we've been prioritizing performance over cost, and we haven't really been as cost-efficient as we could be.

 

I think that's one of the strengths Lippu is bringing back to the organization. This notion that to be competitive, it's not just about being competitive on performance. You have to be competitive on cost.

 

And we believe our PC roadmap starts to get there with Panther Lake, even more so with Nova Lake. And again, as I mentioned before, it's going to take some time on the data center side. Yes, yes.

 

So, to stay on the margin, you did quite well in the fourth quarter, 40% better than expected, but your guidance suggests a 3-4 point deviation sequentially. And then you made some comments about next year; what would the key indicators be? Perhaps we can review them, right? On the one hand, demand is very strong, and your supply is constrained. I think the price is generally quite healthy.

 

But on the other hand, you have new products that are coming up, and you also have some interchangeable advantages. So, when we think about the next four or six quarters, what are the two or three measures that could affect margins? Yes, very good question. I'll start by reiterating that we only guided one quarter out.

 

But, to your point, in the middle of our revenue, we guided the margins to about 36.5% in the quarter, which would be down about 350 points sequentially. Let me break that down for the audience. About 50 points of deviation is simply taking Altera out of the numbers.

 

Now that we've completed the sale of pricing to Silver Lake, we've deconsolidated that. Margins account for 50 of the 350 points that Altera is distributing. Of the other 300 points of deviation, I'd say it's roughly evenly distributed among three things.

 

One is the Intel 18a start, which is always quite expensive, especially since we're only getting diversions from Oregon. It won't be until early next year that we start getting diversions from Arizona with a much better cost structure. And then there's the pricing action we're taking on Arrow Lake and Lunar Lake to navigate this tiger supply situation.

 

And I think the tiger supply situation is going to be here for a while. We talked on the earnings call that Q1 was the peak of tiger supply, but it will remain beyond Q1. And while we try to navigate this, we're taking actions that are both uncritical and incremental to uncritical profits.

 

On the more pragmatic side, we're favoring servers over PCs, and likely occupying the lower end of the PC market. We're also raising prices on 10nm and 7nm Raptor Lake components due to the supply tiger situation. On the other hand, because we know we're shortening the market and are trying to do what's right for our customers, we're lowering prices on Lunar Lake and Arrow Lake to fill different parts of the PC stack so we don't overpower the market.

 

And those are the things we're taking on. I would say that of all of them, probably the most significant is Lunar Lake. As you know, we have memory embedded in Lunar Lake, and that creates some long-term challenges.

 

I would say that at the end of Q2 we would have been fairly confident in saying that Lunar Lake volumes would likely be increasing in Q3, floating sequentially in Q4, and then starting to decline after that. Now, with some of the demand models we're running, Lunar Lake is definitely going to grow sequentially in Q4 and is definitely going to increase year over year. And that creates some incremental challenges, but I think that's the best trade-off as we try to do the right things for our customers.

 

That makes sense. When do you think Lunar Lake will peak in terms of the overall mix? Let me be clear. Lunar Lake is a very small portion of the mix, but because of embedded memory and the cost of that memory, it doesn't need to be a significant part of the mix to have a detrimental impact on margins.

 

And isn't there supply constraint at the Lunar Lake frontier? Because it's generally... That's a good question. We clearly discussed on our Q3 call that we have supply constraint, but I also think there's supply constraint across the industry in general, whether it's substrates with T-Glass or, frankly, some of the memory concerns that surfaced late. And I think TSMC has done a good job as a supplier at the Lunar Lake frontier, but there is supply constraint there as well.

 

So, when you talk about supply constraints, we're not just talking about Intel 10 and 7, which is in-house, but also the supplies coming from TSM. I think we have better supply availability there because of some of the decisions we're making. Because remember, we're actively shifting our in-house supply from PCs into servers, largely because we're growing the server market at a wider margin than the PC market, so we want to make sure we capture that opportunity.

 

So, obviously, Panther Lake and the supply chain, AT&A, there's a lot of focus on that. Perhaps, looking at history, when you launch a new product like Panther Lake, how long does it take before that product becomes a higher percentage of the overall mix? When does that shift happen? Yes, I think it takes longer than people expect, and that's very clear. When you look at the expected ramp-up for Panther Lake next year—and remember, Panther Lake is just one part of the notebook.

 

If you compare that to the last two notebook launches we had with Arrow Lake and Meteor Lake, there's nothing unusual about Panther Lake's ramp-up as a percentage of the mix. We clearly want to do better on the margin side. I think the important thing is that when Lipu met in March, he wasn't satisfied with the earnings, and he was disappointed that the progress in earnings was erratic.

 

I think one of the things that has changed dramatically in the last seven or eight months is that we now have a predictable path to improved earnings. We've talked in the past about the industry's earnings improvement on a new ramp being around 7% per month, and we're now on that curve for Panther Lake, which gives us confidence as we launch the product this week. Like I said, if you go to CES in January, you're going to hear a lot more about that.

 

Okay. And then, to continue with the comments about margins, I think on the earnings call you sounded optimistic about improving margins throughout the next year, maybe by the end of next year, and getting to a creative level. I believe that was the term you used.

 

I think it depends on the earnings, and you've already talked about them, but in the long run, given that this is a new process for you, and assuming the earnings are what you expect them to be, how should we think about your long-term potential here? I don't know when was the last time we were given a long-term model, but are we talking about a five-belt versus a six-belt? How should we think about it? That's a good question. First, I'm going to go back a bit to the suggestion in your question. When we think about Panther Lake, absolutely, as we move into the next year, the increased volume will help bring the cost figure down.

 

I'll remind you that initially, in any new process, we source flyers from Oregon. Oregon is where we conduct all our technological development and then move into high-volume flyer manufacturing. Those flyers tend to be quite expensive, and most, if not all, of Panther Lake's flyers this year are coming from Oregon.

 

As we move into Q1, you'll start seeing flyers coming from Arizona, with a much better and different cost structure, and that increases throughout the year. Regarding the more important part of your question, the second half, about long-term, high-volume flyers, I'm not going to go into a financial model. We've discussed internally when the appropriate time is to hold an Investor Day.

 

Stay tuned for that. My question is, it's going to be sometime in the second half of next year. But clearly, there's nothing in the way we think about our business that doesn't suggest we should have high-volume flyers comparable to the industry average with our fabulous peers.

 

And frankly, with the margin stagnation of being an IDM, we should do a bit better than that. Well, I'm definitely looking forward to Investors Day. I want to switch gears to the foundation portion.

 

Obviously, it's a challenging business. I think you're talking about 18A, potentially leading to a starting point at some point. I believe you previously said maybe 2027; I don't know if that's still valid or not.

 

And also, I think that since LeBoot retired, there's been talk that if you don't get a material customer for 14A, you might halt the development process. My question is, is 2027 still the target for the foundation? Yes, what we've historically said is that the 18A ramp-up, primarily on the back of internal products, should be able to bring the Intel Foundation to a profitable position based on the 2027 exit rate. And that's clearly where Naga is still leading the organization.

 

I'll give you an important caveat. When we win a customer for Intel 14A, we'll need to price it in properly before we see any revenue. So, for transparency reasons, I think that as 14A materializes customer traffic, it's likely to be pushed to the end of 2027.

 

I think most investors will agree with that, because it will confirm that we can raise an external foundation. Yes, that's probably a good thing. We have a couple more minutes.

 

I want to see if there are any questions from the audience. Let's continue. Again, the comment about the 14A trial didn't garner much attention.

 

But at the same time, you said that you're fine with 18A and 18AP for your internal products until 2030, so you don't need 14A for internal products, right? Tell us, if you had hypothetically abandoned the development of 14A, what would you do with your OPEX or CAPEX? How would it affect the business model? I always hate hypothetical questions because they just put me in a difficult position. I want to be very clear. We are all in favor of developing Intel 14A and we feel good about the major agreements we have with external customers.

 

I think it's important to point out that the 14A is, in many ways, a very different node from an external perspective than the Intel 18A. Simply put, and I'm sure the teams at TD will give me a hard time about this when I get back to the office, but I always think of any node as having three phases: the final phase, the development phase, and then the high-volume manufacturing phase.

In the final phase of 18A, we were only involved with Intel products. It wasn't until the development phase that we started soliciting feedback from external customers, which meant that many of the choices we made at the transistor level were to optimize internal product groups rather than external customers. Furthermore, it was the first time we began to understand PDK process development systems, and we encountered some growing difficulties in making our PDKs a true industry standard.

 

I think the big difference between Intel and 14A is that we're in the final stage and we're engaging with external customers. That means we're getting more and better feedback on how we're doing from external customers in 14A than we did in 18A, and the maturity of our PDKs is much greater. We're now bringing industry-standard PDKs to market, both of which help tremendously.

 

I also want to point out that in the 18A we switched from FinFET to Gate All Around and also added rear power. We made some big changes. The 14A is a second-generation Gate All Around, a second-generation rear power amplifier.

 

Y lo hemos dicho y ha sido muy claro que si miramos dónde estamos hoy en 14A, en performance and yield versus a similar point of development on 18A, we're significantly further ahead on 14A. So we're feeling very good about 14A. 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

30 first minutes