Intel Corporation (INTC) Earnings Call Transcript & Summary

Good morning. Why don't we go ahead and get started? I'm going to give some opening comments before our first keynote address from Pat Gelsinger, the CEO of Intel, but my name is John Pitzer. I am the semiconductor capital equipment analyst here at Credit Suisse, and it's been my job for the last, I think, 15-or-plus years to start off the conference with some welcoming remarks. And I think, along with myself and all of my colleagues, and I'll run through that list, it's Phil Winslow, it's great to have him back at Credit Suisse on the software side, with his partner in crime, Fred Lee. We also have Doug Mitchelson and Sami Badri in the telecom services and equipment space, Stephen Ju doing Internet and then Tim Chiodo on the fintech side. And we're all just couldn't be more thrilled to be able to host this conference live. It's great after a 1-year hiatus to be able to get owners and companies back together face-to-face. I think I had 4 consecutive dinners last night in the semi space, and both investors and companies were just ecstatic to be able to actually sit down in a room, read body language and have real deep conversations. As much as I love Zoom and we've all been zooming for the last sort of 24 months, it is great to finally get back on stage. As I was thinking about my remarks this morning, I was thinking back to the last time it was on stage here and just how much things have changed. And I always tend to look at things through the semiconductor lens. It's sort of an occupational hazard. But again, back in sort of 2019, the last time we were here, the S&P was at 3,200 on its way to 2,200 sort of at the peak of the COVID fears in March of 2020. We're now about 45% higher than those December '19 levels. Same thing with the NASDAQ. Was it sort of 8,500 on the way to 7,000. We're now approaching 16,000 on the NASDAQ. So clearly, tech has done extremely well kind of over the last 20 months. If I look at my space, the index that's nearest and dear to my heart, it's the semi index, and I think we're up about 125% from December '19 levels. We're up almost over 200% from the March lows of 2020. And so clearly, the last 20 months have been trying for all of us, I think in general, what we've been arguing is COVID didn't really change the tech trends that were going on, but they did codify and accelerate a lot of the trends that we've been trying to showcase at this conference over the last several years. And we've always sort of positioned this conference as well timed. We think it gives you a last look into a company's calendar fourth quarter. But most important, it kind of gives you the opportunity just to sort of recap what happened year-to-date. And even more importantly, what's best fit for your portfolio next year and in the years to come. And it really is the intent of this week to kind of get a more personal setting between investors and managers to really get those touch points that allow you to think a little bit more deeply than what stocks might do over the next sort of 1 to 2 weeks. I think the key question we're all trying to figure out is what's the new normal. I think that, that's going to be, I think, topical on people's mind this week. I know as I think about the semiconductor space, we spend a lot of time thinking about supply chain, supply chain constraints, whether or not COVID will mark the end of 20 years of globalization, what supply chains might look like going forward. We've sort of argued that they're likely to become a little bit more sovereign, a little bit more redundant as a result, probably a little bit more automated and a little bit more intelligent. That clearly plays well into the semiconductor capital equipment space because one of the things that's happened and whether this is U.S.-China tensions or just some of the vulnerabilities that COVID exposed on the supply chains is world governments are now much more focused on actually domestic production of semiconductors, and that is driving things like the CHIPS Act here in the United States. There's similar legislation going through the EU right now. And it's going to be sort of a brave new world about where supply chains are actually going to migrate sort of over time. The point of this conference this week is just to give you guys opportunities to talk with management. If there's anything that I or my colleagues can do to help facilitate that, don't hesitate to ask. I'd be remiss not to also thank the events team at Credit Suisse. This is an incredibly complex conference to put together, even more so in a year like this with COVID. And so they deserve a big sort of thank you, as does The Phoenician staff here, which always does a great job. And with that, I think I'm going to end my opening comments there and probably welcome Pat Gelsinger on to the stage so we can go through the first fireside. Pat? Good to see you.

Good to see you. How's Mr. Pitzer today?

I'll put you in the middle.

Okay.

So Pat, the format here is pretty straightforward. We're going to have a conversation. Near the end, there will be a mic going around the room. So if anybody in the audience has any questions, please feel free, raise your hand and we'll try to get you the mic. I know that I needed to read a housekeeping item for you. So I will do that. Today's discussion includes forward-looking statements, which are subject to risks and uncertainties. Please refer to Intel's SEC filings available at intel.com for more information on the risk factors that could cause actual results to differ materially.

With that out of the way, Pat...

I feel so much safer.

I always like to kind of make my first question a little bit open-ended. You're almost, I think, 10 months on the job officially. A lot of changes have happened sort of over those 10-month period. Maybe you can give yourself an assessment of the changes, the direction you're trying to put the company. And importantly, as you answer that question, what is the real value proposition you see for Intel for investors?

Yes. And obviously, it's been, as we like to call them inside of the company, a torrid pace over the 9.5 months since I've been back. And coming in, we laid out a clear new strategy. We said, boy, we're going to be an IDM 2.0. We're going to redouble on manufacturing. We're going to open the doors to Intel intellectual property and Intel fabs to become a foundry. We're going to reorganize the company. We have our 6 business units now. We'll be laying out a clear strategy for each of those. We'll certainly talk more about those at the coming Analyst Day. We've brought in new talent to lead those business units. So we've reversed the talent trends internally. We've given a clear, I'll say, road map of how we're going to execute to the industry. We've redoubled our investments in process, packaging technologies, gotten that back on a solid footing, maybe the most negative indicator of our poor execution over the last couple of years. And now that's on or ahead of schedule, as I said on the last earnings call, Intel 7, Intel 4, Intel 3, Intel 20A and Intel 18A on or ahead of schedule. We are working on rebuilding the culture of the company for it. And all of those things are well underway and at the last earnings call, we chose to proactively come out to The Street and say, here's the financial model that's consistent with that strategic model that we laid out. We didn't have to come out that aggressively. We said, no, we want to come out, lay that out so investors know exactly what the strategy costs, but also a picture for the return model associated with that over time. And I feel good now, right? Our stock went down 10%, and I feel great, right? A CEO should never say something like that. But now we have coherence between the financial model, the strategy and the execution of the company. And over the 9.5 months, we've just been firing one after the other. We laid out the strategy. We laid out our process road map. We laid out our innovation strategy. We have launched major new products along the way. I'll say, we've just been giving breadcrumb after breadcrumb after breadcrumb, and that's what we're going to keep doing between now and the Analyst Day and beyond so that we rebuild the investor confidence in this iconic company. And for it, this is a tremendous opportunity for investors as I look at it. right? Because, hey, this is the lowest multiple semiconductor company by far, right? We have this extraordinary set of assets. We've laid out a strategy going into the future that builds on those assets. And hey, we're out not to increase the value of this company, but dramatically increase the value of this company as we look forward. It's going to take some investment, but pretty clear-minded view of the strategy and execution to get there. That's the path.

Well, Pat, you know better than almost anyone that the cornerstone of Intel's historical success has been the manufacturing muscle. That's also been the muscle that's let the company down over the last sort of 5-plus years. One of the questions that I get asked most frequently is, hey, you're embarking on sort of a Moore's Law mission of 5 nodes in 4 years in the wake of having very difficult times at 14, at 10 and at 7. From the outside looking in, how can you give investors confidence that you can execute to this strategy, on the manufacturing side alone?

Yes. And for that, the -- as we've talked about, the yields on 10 are now looking healthy, and you're seeing those volumes shift very well. At 7, Alder Lake is now shipping great product, getting great reviews in the marketplace, the clear leadership product in client. We said Intel 4, our Meteor Lake product, right, is looking good. And I've said Intel 3, 20A and 18A are ahead of schedule. And also maybe as a point of affirmation there, we're having our foundry customers like Qualcomm, who are looking at these leadership nodes who are very discerning customers, right? You don't get a design like that arbitrarily, right? They take us through our paces and our technology teams through our paces to be able to say, "Yes, we're going to start using that as a foundry customer on the angstrom era 20A, 18A process technologies." So I think we're giving the market solid proof points. Now again, right, many years of bad execution, don't make up for in 9 months. But I think that's a lot of proof points in a short period of time that the manufacturing machine is back on track. Also, I'd say that this team is now working for their pride, right? Here is the most iconic technology development team in history. This is the team that created strained silicon, that created the high-K/metal gate, that created FinFET. Every major innovation that's happened in semiconductor transistor technology for the last 30 years came from this team. They screwed up big time. They are now back to restore their honor in the marketplace. They are so fired up. If I told them they had to take a day off, I couldn't get them to do it, right? It just is impossible at this point because we said we are betting on them big time. We also laid out, you sort of say 5 nodes in 4 years. What, are you freaking nuts? Moore's Law has never done this. And what we've done is, essentially, we've created a tick-tock model in the process development team where the Intel 10 and 7 team is distinct from the Intel 4, 3 team, is distinct from the Intel 20A, 18A team. So I now have tick-tock teams. We've capitalized them. We've given them more engineering capacity, running more experiments, so we don't have the serial impact that we've had. When we came out boldly at our Innovation Conference, and I said we're going to maintain, if not go, super Moore's Law based on 4 things. One is EUV, right? We now have lithography headroom as we haven't had for the next decade. Second, we have the new transistor structure with RibbonFET, right, and new power delivery and metallurgy framework with PowerVia so we can now deliver power, extract it, be able to do it, and 3D packaging, right? And with that, today, and I happen to just bring it along -- fancy that. I actually have a show-and-tell. This is the 100-billion transistor Ponte Vecchio, right? And what we said is we're going to maintain, if not go, super Moore's Law, such that we will have our first trillion transistor chip by the end of the decade. Just think about it. 1 trillion fricking transistors on 1 package, but we're going to do so by leveraging 2.5 and 3D packaging technology. So not only is it increases, right, in the per die transistors, but in the number of transistors possible in advanced packaging as well. So we've laid out a path between now and 2030 to have a trillion transistors. And as the -- I'll say, the bearers of Moore's Law, we are committing to be the standard bearers for it and let the industry catch up, right? Because we're on an aggressive mission to be able to, I'll say, back to the heritage of technology development and manufacturing capacity.

Pat, you mentioned sort of in your opening comments the reset off of the Q3 earnings call. And I think a lot of the reset was welcomed by the investment community. The gross margin reset, I don't think was a surprise. The CapEx numbers that you put out for next year of $25 billion to $28 billion, I don't think was that much of a surprise. I think what kind of rubbed investors the wrong way was the long-term growth rate that you threw out there of 10% to 12%, given that, that's sort of been a growth rate that you've really only hit once or twice in the last 10 to 15 years. Can you perhaps unpack sort of how you get to that long-term CAGR of 10% to 12%? What's coming from what I would call the core businesses today versus some of these new businesses that you're getting into?

Yes. And if you look at the 10% to 12%, obviously, it's the -- we gave out a 5-year guidance. So it's in the back end of the 5 years that we expect to start seeing those numbers. But you can put it essentially into the 2 buckets, right, I'll call the core businesses and the growth businesses. And if you unpack the 10% to 12%, it's about half and half, right, between it, not trying to be too precise. We're going to give more clarity on this at the Investor Day in February. But then if you look at the core business, PC and servers, for that, there is nothing dramatic in the PC business. We do believe that we have a systemically larger PC business. And every day, we have a new COVID variant. I think it reinforces this, right? Just the industry isn't going to dip, right? It's going to maintain a structural larger. We also see that we have market share opportunities in the PC as we have more of this capacity come online, and we have more BOM opportunities as well. So we'd say, "Hey, I mean, it's going to be a growing market, but nothing dramatic for that." The data center business is a stronger growth market. And again, we've had some market share losses. So we see those stabilizing. And over the horizon, we see our opportunity to regain share, and the data center market is estimated over the 5-year period to be a 10% to 15% grower. So in that, hey, we're going to see some amounts of recovery as we gain share. So essentially, half of the 10% to 12% is in those 2 business areas with more of that growth being in the data center business, but stability and modest growth in the PC business. But then I have 4 growth businesses, right? And our position in accelerated graphics and high-performance compute, very low, right? Essentially, I'm starting at 0% share with a good product, going to gain share, grow rapidly. Our Mobileye business, we've had 35% growth rate in that business for multiple years in a row. It's now approaching $2 billion. This is now getting to be a real growth contributor for us. Our networking business, we have a very substantial position. And with 5G edge computing, we see a very solid growth there. And finally, the foundry business. We don't see a lot contributing in the next couple of years. It takes years to get designs. But as you get to year 4 and 5, it's going to be a contributor. So about half of that growth comes from these new growth businesses that we've laid out. The last thing I'd say there is if the process of the manufacturing machine is good, it's a rising tide for all of the boats, for the core business and for the growth businesses as well. And when we put out a number like that, hey, I'm going to drive the team to be significantly better than that number as well. We're not arbitrarily putting it out. We've built a 5-year plan. We believe it's achievable or I wouldn't have said it.

Well, Pat, just to follow up on that, the $25 billion to $28 billion, I think the easy sort of concern out there is that you're spending and building out capacity to a demand curve that looks very aggressive. And if it doesn't turn out to be right, you've kind of straddled the company with a lot of structural costs that are going to hurt margins for years to come. I thought it was interesting at our dinner last night when you kind of talked about what was driving the CapEx for next year. It sounds like a lot like the core business, just trying to catch up with the current lack of supply in the industry. So maybe you can talk about next year's CapEx and how you see CapEx trending over multiple years as you go after that 10% to 12% growth rate?

Yes. If you think about the $25 billion to $28 billion, that's a net number. We hope to do more than that based on capital offsets from government investments as well as customer prepays and other investment criteria for that. But the vast majority of the $25 billion and $28 billion is just to feed the core business. We have -- in some regards, I haven't lost share to AMD. I've given share to AMD because they didn't have capacity, right? So a lot of this is just catch up to a growing market and years of underinvestment. So the bulk of the $25 billion to $28 billion is just driven by the core business and catching up to the demand curve in the core business. And on top of that, we are overinvesting for the 5 nodes in 4 years, a very conscious decision to get back in front of it. And as you said, it's a rising tide for all boats. And then we're building some flexibility in capacity, but that's the smallest of those 3 components for the long term. And shells are very effective things to do, right? If I create a shell, I invest essentially $2 billion in the first 2 years of a project to create the opportunity to fill it with the second $8 billion, right, of equipment buildout at the end of it. So it's actually a very capital efficient way to start creating optionality in the business. So we call that smart capital. And if I had some free capacity today, I would be building it out, and I'd be building more share today. One of the standard rules of the road of Intel is you always have a spare shell. We don't -- that's part of why we're in the semiconductor crisis today. Nobody had any spare capacity around, including us. So we're going to create some of that option value over time. Now as we're on this journey, as I like to say, "Boy, I would lust for 1 wafer of spare capacity, much less any reasonable amount of spare fabs in the future." But let's pretend for a second that we actually have overshot. What am I going to do with spare capacity in '25, if in fact, I've overshot the market? And I'll just tell you, there's no way that we have spare capacity between -- in '22, '23 and '24. But let's get to '25 and '26, and we say we've overshot a little bit. I have 3 uses for that spare capacity. One is go get more market share, right? If I have leadership products and leadership process, I'm going to do so with good margins. Gain market share back. Second is, I've also built into our business model that we use foundries, right, external foundries as well. So if I have too much capacity, I'm going to pull wafers back from the external foundries and run them internally at better margins. Third, I go win more foundry customers. So I have -- and these are good margin foundry customers as well, just like we're seeing in the leading edge foundry market. So in the off chance that I have any spare capacity, I have 3 tremendous uses of that, that are highly margin and capital and cash flow efficient as well. So to me, those concerns, right, I lust after the day that one of those might actually materialize.

Pat, I've got a bunch of questions around the buckets of revenue you have. And I'll start with kind of the core business and within that, the PC market. I think there's a lot of investors out there that look at the tremendous growth we've seen in PC units during the pandemic. I think we've gone from 260 million back in 2019. I think we're going to be close to 350 million units this year.

Yes. Million units today.

And there's sort of a view that, that's got to be the PC market over earning. Now that's not a view that the industry holds. You look at Dell's earnings and HP's earnings a couple of weeks ago, still very, very solid. I'd like to joke that before COVID, work-from-home meant logging on, doing e-mails, but not a lot else. During COVID, we actually had to do real content creation. And so maybe the value of that notebook, that PC just became much more significant in that environment. Help us understand why you don't think that this is sort of an over-earning environment, but maybe a new kind of structural floor for the PC unit market?

Yes. And we do believe that sort of 350 million, that's the floor, if you would, for this market. And what you see, what you -- there's several factors inside of it. One is the refresh rates have been elongating. And now we're starting to see sans the demand pressure. We're actually seeing a desire to pull those back, right? And lots of that is driven by this dramatic increase use cases in the COVID environment. We're also coming into a Windows 11 refresh cycle as well. Corporations are way overdue on their refresh cycle. So we're seeing a lot of things pent up for the refresh. We're also seeing that this PC density, as we call it, is very real, right? Mom's working from home, dad's working from home, kids are working from -- or schooling from home as well that you're seeing that pressure for increased density. And you have the lowest retail channel inventory levels in history, right, ever in the PC industry. We're at 0 to minus 1 day of inventory through the entire channel structure, right? And all of the OEMs are saying the same things. They see no perishability of the backlog that they're seeing as well. And you've heard that from all of the majors as well and that's before they start really delivering on any of the corporate or Win 11 upgrades cycles associated with it. And we're also seeing globally that there continues to be more penetration for markets in second and third tier markets as well as they need to be educating online and learning online. And I think the continuing COVID comment will continue to increase those market penetration points as well. So that's why in our models, we've assumed an IDC now is at 2-ish percent for next year. We're slightly better than that for next year, but we see a sustainability to that, I'll say, slightly positive growth rate against that 350-plus million baseline for multiple years into the future.

Against that industry backdrop, it seems like you also have perhaps a stronger product suite. Alder Lake is the first sort of big/little core on x86. You've got the desktop part out. The notebook part is coming. It's nice to see. I think it's the first time in several years that I've seen third-party sources like Tom's Hardware doing benchmarks where that's kind of excelling vis-a-vis your competition. Can you talk a little bit about how important the Alder Lake product cycle is? And what your expectation is for share position in PCs over the next year or 2?

Yes. We do believe, and I think the market indicators that we were slight positive share gainers for the last 2 quarters in a row. So we've stemmed the share losses. We started to regain share, and that's before Alder Lake, right? Now with Alder Lake, we now have own question leadership products. And that's some of the benchmarks that's anywhere from 10% to as much as 50%, 60% advantages. This is like the heydays of Intel, where we have the best product, we have capacity, we go win share. And that's exactly what we're going to do with the portfolio of the Alder Lake products coming on, and the launch is going super well. All of the analysts have said this is a killer product. We also expect with that, that some of the graphics and media capabilities of that are comfortably better than the alternatives. And that's before we add our discrete graphics products that are coming in Q1. So overall, we just see as we're coming into a good cycle for PC market share growth. And again, I expect there is no limitation other than our capacity to grow market share in the next year or 2.

Switching gears to the core data center business, a bunch of questions here. One, how do you think about the share outlook there over the next year or 2? It seems like that unquestionable leadership, it's perhaps maybe the second half of '23 into '24, correct me if I'm wrong, but how do you see Ice Lake and then Sapphire Rapids sort of rolling out? And how do we think -- how should we think about share over the next 12 to 24 months?

Yes. Ice Lake wasn't our greatest product, but it closed some of the gaps. Sapphire Rapids closes all the gaps. So we're now ahead with Sapphire Rapids coming in Q1 of next year. And on some of the benchmarks, it's close to AMD. On other benchmarks, it's unquestionably the best product. And like the AI performance of Sapphire Rapids is now 50% of the highest-end NVIDIA specialty processor, right? It is dramatic. It's hundreds of percent better than AMD's alternative at this point. So we're clearly having some very strong points of advantage with it. But AMD, we expect them to respond next year, so it's going to be a little bit nip and tuck over the next couple of years. And as you suggest, it's going to be out a couple of years until we're back to unquestioned leadership in the data center. It takes a bit longer in the data center side. The development cycles are longer. And as our process technologies get better and better, our server road map takes a little bit longer to take advantage of that. But we're on a path for capacity and for product leadership. It's going to be more of a market share battle over the next couple of years. But we're limited in capacity and so is AMD. So in that, I think it's going to be a fairly, I'd say, a stable market in terms of share and price points, but we do see ourself able to start regaining our footing in that market. And we've gone through many quarters of share losses. I see that flattening and then a recovery cycle in front of us as the products get stronger. And we have a solid road map after Sapphire Rapids to build on those strengths and bring more of the product capabilities into the market.

And Pat, I think investors focus almost exclusively, for good reasons, on the share shifts that are going on in the data center. But I think we lose sight of the fact that you're still shipping 8.5 out of every 10 servers into the marketplace. I'd love to get your perspective just on the overall market for next year. It seems like Ice Lake being delayed did perhaps create some pent-up demand in the market. How are you viewing just the server market for 2022 and beyond?

Yes. We see the market as a double-digit grower, right, in the teens in a sustained growth rate looking forward, more and more penetration of cloud going forward. We are in a very healthy cycle of enterprise upgrades now for enterprise data centers. But the third market is the one I'm actually most excited about is edge, right? As we start seeing 5G edge -- and a little quiz question for you is what share of the inference market does NVIDIA have?

Relatively small.

Relatively small, and our share is very high, right? And right now, the inference market is about 1.5x bigger than the training market. As edge computing emerges, that grows faster and we expect to see the inference market grow substantially faster than the training market over the next couple of years. And that's an area of strength, area of strength for our products like FlexRAN, 5G and edge, area of strength because we're a much better inference platform. You don't need the special purpose, high power, expensive accelerators, you get it off of your standard Xeon platform. So we see that being an area of strength for us as we look and will be part of our network and edge business going forward for that area of the market. So overall, we see the data center market in the 3 aspects: the cloud, the enterprise and the edge to be very healthy growth rates. And given the breadth of our portfolio, many of these investments in the open platform, we feel like we're pretty nicely positioned.

We talked about x86 competition in servers. You mentioned a little bit about accelerator competition in the data center market. One of the questions I often get asked is the idea of general purpose compute versus more customized silicon. You're seeing a lot of more hyperscale companies trying to build their own specialized silicon, optimized silicon for certain workloads. I'm wondering if you can help us kind of understand your view of how big this customized silicon market might become? And what's the risk? And quite frankly, what's the opportunity for you?

Yes. To some degree, with our foundry strategy, we've taken the debate off the table. You want to do innovation yourself in silicon? Come on down. We're happy to give you the richest portfolio of technologies, both ours and what you would do to co-optimize and co-innovate together. So the first thing is we've taken any religious discussions or premonitions off the table right? And where actually like Amazon is one of our early foundry customers for exactly that. Come on down, let's go innovate together. That said, where does a cloud guy want to invest their capital, their best engineering capacity? It's not in silicon, right? It's in software and services, right? That's where they get disproportionate value differentiation. So this is actually a lousy use, and multiple, right, of the top cloud guys have said this, "I don't want to do this kind of investment. The only reason I did it is I lost confidence in you, Pat, Intel delivering a sustainable TCO value proposition to us over time." And in that -- we have engineers, right? I can go engineer this myself. And if the only way that I see to give TCO value at my massive cloud scale is to do some of the engineering myself, great, I'll go do it. That's exactly what they've done with some of their custom programs, but it's actually a pretty lousy use of their capital and engineering if I'm doing my job. So what we have to do is reprove to them that we're going to do our job, right? We're going to give nice steady road maps of TCO improvements on the core platform, and I'm going to give them the opportunity to co-innovate and design with us on our foundry services. We've opened up x86, something we've never done before. It's very much a Microsoft, Satya-like move, right? We said, "Hey, we're no longer religious about this, right?" Foundry supports RISC-V, ARM and x86. And for the first time in history, we've opened the x86 architecture up that will go co-innovate with the industry and customers as well. And we're getting great response to that strategy. Obviously, new designs like that will take a couple of years to materialize. But the response probably of the 100-plus foundry customers that we had, somewhere between 1/3 and 1/2 of them are interested in using x86 as part of their design. Very good indicator.

That's helpful. You've mentioned foundry now several times. I've got a bunch of questions. As you know, there's some level of skepticism. I think there's also some level of confusion about what the strategy really involves. I think first, to take the confusion off of the table. There seems to be some out there that view the foundry strategy as sort of if we build it, they will come. And it worries people that you're going to go out and build a giant fab for foundry and wait for the customers to be there. Can you talk about how you plan to grow this business? I mean from my perspective, I could be wrong, this is almost like co-mingling foundry wafers with IDM wafers to start. And if at some point, it gets large enough, perhaps it deserves its own dedicated fab. But help us understand how you're going to grow that business from a capital perspective.

Yes. And I think you framed it well. In the '25 to '28, we already talked about the vast majority of that is for the existing business, some of that's for the catch-up and some of that is to create more flexible capacity, which may get used for foundry at that point. But we also see the smart capital perspective that, "Hey, I'm not going to build it without some expectations." We also see a lot of interest from customers to pre-invest, governments to support that business model, so we get capital efficiency for it over time. And the design cycles that are associated with foundry are fairly long. It takes a couple of years for a design to come on board. So it creates some shell capacity, start engaging in designs. Both of them start to come together in 2 years from now, whether we know at that point, we'll be making bigger commitments of capacity, but we're already going to have design commitments as well. So I actually consider it a fairly low-risk strategy at this point because I just need more shells, get more shell capacity put into the ground as quickly as possible, giving us more market share flexibility, more foundry flexibility and more flexibility to bring wafers back from our foundry customers as required. So a lot of, I'll say, strategic optionality in it. Now the foundry business as well, there's only 3 companies in the world that can do leading-edge process technology. And there's not going to be a fourth, right? The capital requirements and the R&D requirements are so extreme, it's not like somebody else is going to come running by and say, "John, I got a great business idea. All I need to do is have a $5 billion R&D stream and on the order of an invested $25 billion of capital, and I'll start doing foundry services 5 years from now," right? It's like -- I mean this is so extreme at this point. We're now in the cycle of basic business 101 that it's a consolidation cycle, not an expansion cycle in the number of players in the foundry business. Samsung, TSMC, Intel, the only 3 that can do it. And we've gotten tremendous geopolitical reinforcement in the U.S. and in Europe for the desire for our Western fab, right? And that's why we're seeing CHIPS Act, the European equivalent of CHIPS Act, and customers as well. As I was flying down here yesterday, The Wall Street Journal reported that there were 27 Chinese planes in Taiwan airspace yesterday. How do you feel about having your sole-source foundry capability in Taiwan right now? I mean this is a geopolitical risk. And as we've argued since the beginning of my tenure, the world needs a more resilient, geopolitically balanced supply chain. So we're seeing a lot of enthusiasm to support us moving into that market. I went into a major customer, one of our early foundry customers and they produced on the wall, made in U.S.A. with our 2 logos on it. That means they're like, "Please give us some optionality in this environment," right? And there is a lot of geopolitical concern around TSMC and single-source concerns. So a lot of support for that. But we, of course, got to do a good job. We have to create a good process technology, good foundry technologies. And I've also said that foundry makes our IDM better and IDM makes our foundry better. And how does foundry make IDM better? Well, it's getting us to leverage more industry standard IP, which makes my design teams more efficient. Industry-standard PDKs, which make the EDA tools more efficient. I get more efficiency in my internal design, but my internal design and process technology, TSMC has to fund all of this R&D for the foundry business. My foundry business gets it for free, right? I'm making all that IP available, right, of IP blocks, they get it for free. So I get to monetize existing investments at a scale that they don't. And as we look at the fab network, back to your beginning point, I expect that it's going to be quite a number of years until I have dedicated foundry fabs. Most of them are going to be flex fabs running a lot of our product wafers and intermingling it with foundry customer wafers to optimize, right? The best way to run a fab is 7 x 24, 101% loaded, right? So if I'm better positioned, right, by being able to do that, that will be a better financial outcome as well. Finally, foundry fixes a perpetual bug in the Intel business model because just about the time a factory was highly optimized, depreciated, running great, we'd flip it to a new process technology, right, and destroy the capital efficiency and cash flow cycle associated with it. Well, now I'm fixing the bug, right? I'm getting more monetization on the back end of those capital capacities, right, and that rich -- this is what has always funded TSMC's business is that rich long tail of foundry. So I'm now fixing a bug in the Intel business model. And I think as we get into the second half of the decade, it becomes a very rich cash flow opportunity out of those depreciated leading-edge capacity.

And Pat, you somewhat anticipated my next question when you talked about the EDA and the software libraries, but the perception out there and perhaps the misperception is that the DNA at Intel is building relatively few parts in really large volume across a network of fabs. And that the DNA you need to be successful to be a foundry is running relatively few parts in a single fab really, really well. So how do you convince the investor that you actually have the core talent you need to properly run a foundry business?

Yes. And some of this is 5, 6, 7 years ago, when Intel attempted foundry before. We were so proprietary that we required the foundry customers to essentially jump this massive over into our tools, into all of our proprietary things. There was no industry standard. And over the last several years, and this predated me, I've accelerated it, but we've standardized many of the EDA flows, the PDKs associated with the process. We've essentially got ourself almost aligned with industry standards. And that enables this alignment of the industry forces with us. And this is part of the reason, hey, it makes my internal design teams more efficient as well because they get to cherry pick from these tool vendors and these IP blocks across the industry. At the same time, the thesis that you laid out is a myth, right, at least for the big foundry customers. These are big customers, right? You're running 1, 2, 3 fabs worth of material. It feels more like a microprocessor, right, when you look at a Qualcomm or somebody like that, running a very few number of SKUs at very high volume. In some respects, even more so, right? They have less SKUs running their business than I have running the Xeon business today. But there is a long tail, right? Essentially, you can take foundry customers and put them in these big massive customers, very front-end focus and the long tail of customers. And I do think there will be learnings for us on getting better at that long tail, but we've already industry standardized, as we've already said, many of our capacity, many of our tool chains, many of our customer and supplier relationships. So I don't see that as an insurmountable problem, but there will be learning there for us.

And Pat, I want to go back to something you said earlier, and I don't want to overstate it because I think it's still a relatively small number. But you mentioned customer prepays to help fund foundry CapEx. Is that in that '25 to '28 net number next year? And I'm assuming as you execute, you would expect that number to get larger over time. How meaningful could that be as a source of capital to help fund this business?

Yes, the '25 to '28 is our capital net of any prepays or government incentives. And we've taken a fairly modest view in that '25 to '28. Obviously, as we get into '23 and '24, CHIPS Act, European CHIPS Act gets funded, we would certainly hope that those numbers become more significant. And as half of the U.S. semiconductor industry today in terms of capacity, I'm going to be fighting for my unfair share of the CHIPS Act as we look -- I mean, my fair share for that. But we're going to be aggressive in our pursuits of it. Prepays of capacity, for instance you saw the GLOBALFOUNDRIES deal with Ford, just a couple of weeks ago. I mean it's become de jour, right, to be expecting some level of prepays on these very large capital investments for the large foundry customers as well. I expect my practices will match the industry practices. What's TSMC doing? What's GLOBALFOUNDRIES doing? We will do similar. And if you look at the GLOBALFOUNDRIES deal, for instance, this is what we're seeing generally in the industry, approximately 1/3 of the capital is being covered by the prepay requirements. So our simple math would be, hey, the CHIPS Act will fund up to 30% of a fab. Typically, there's state incentives at 10-ish percent. We'd expect similar numbers in EU and member states in the EU, so 30 plus 10, plus potentially up to 1/3 for prepays. Those would be the numbers that we're working to, to give more capital efficiency as we go build out that network. But obviously, there's a lot of work to do to get there.

And then, Pat, you mentioned Ponte Vecchio and getting back into the discrete graphics market. And again, some of the early benchmarks look really good around Ponte Vecchio, but the concern I always get is this isn't your first rodeo into the discrete graphics market. Why is it going to be different this time?

A couple of things. One is when Intel was, I'll say, dithering with being in and out of the discrete graphics market, it was before AI, right? Now there's much, much broader market implications associated with that movement of accelerated computing from integrated graphics, where we're the unquestioned market leader to discrete graphics to high-performance computing. And that continuum creates a much bigger business impetus. And essentially, that's what NVIDIA has done, right, without integrated graphics, but they've sort of followed that continuum of accelerated computing to leverage the platform, leverage the IP blocks. And we're going to follow a very NVIDIA-like playbook there. We're going to do so with much more openness, industry standard. Customers are very frustrated with proprietary CUDA. So we're going to drive many of the open software standards, Open Data Python, parallel C++, other things like that, that will be entirely open approaches, and we're getting tremendous interest. I think there's almost an insatiable demand for us to come into the marketplace now. I don't think I can catch up to the demand requirements that we've seen from big OEMs from markets for at least 2 years as we enter that market. And like you said, early indications, right, for, I'll say, undefended, unsupported leaks and rumors in the industry are quite positive.

Pat, I want to talk a little bit about Mobileye. It's probably one of the better AI assets you have inside of the company, if you view autonomous driving as an AI application, which it obviously is. There are a lot of investors out there that would argue the best way to monetize that asset is actually to spin it out. It's the most profitable, self-sustaining sort of autonomous driving company in the world. The people have pegged the valuation there at $30 billion potentially. Why keep it inside of Intel? Why is it more valuable inside of Intel than outside?

Yes. Clearly, we bought it for $15-plus billion. We've clearly created value by giving it more capacity, more credibility in the market, and I see a strong value creation cycle for it over the next couple of years. And a couple of examples of that. Clearly, right now, the auto industry, as everybody knows, has had extraordinary supply chain issues. Well, Intel gets to come in and fix the supply chain issues for Mobileye, right? We have the capacity to do that. Nobody else does, right? If I go to -- when I call up Jim Farley, Mary Barra, Oliver Zipse or something like that and say, "Capacity? Not an issue," right? We got it covered. That's a pretty powerful statement as they're making their decisions into the future. So we see the supply chain benefits. We're also -- while we haven't announced anything specific with regard to Mobileye, we have said that we will have an automotive-grade foundry capability as well. So we're clearly building out our foundry capacity for the automotive segment, which by the way, just one quick little factoid, the premium car segment today, which represents about half of auto revenue today is about 4% semiconductors. By 2030, it will be 20% semiconductors, a 5x increase in semiconductors in autos as they move to AV and EV. Wow. Probably -- and we expect it to be one of the fastest-growing segments for the semiconductor industry overall. And of that 20%, what percentage will be leading edge technologies? Almost all of the increment, right? So I mean this, again, just reinforces leading-edge capacity, right, leading-edge AI, vision performance, 5G, 6G connectivity, all of it reinforces the strength for us. So we see a very strong value creation cycle. And then some of our unique technologies like LiDAR, we're making those immediately available, and I expect that we're going to have a very strong position in the LiDAR car segment overall based on Intel technology. So those would be some of the examples where we see a lot of value creation over the next 4 or 5 years for Mobileye from being part of the Intel asset.

Pat, we're coming to the end of the session, but I wanted to give you the opportunity in case you think I missed something in my Q&A that you think is pretty critical to the overall narrative of Intel, especially as it pertains to the investment community. Any sort of concluding remarks?

Yes. And thank you for the opportunity to be here today. What we'd say is investors are looking today and sort of saying, "Is this dead money? Or is this huge multiple money as we look forward?" And I think over the 9 months, we've made a lot of progress. We're starting to give very substantial proof points to the marketplace that this is going to be a -- not a little expansion, a major expansion. When I took the job it's like take over to pivot a company, an iconic company that puts silicon in Silicon Valley. Okay, that's a tough job. Do it in the middle of a pandemic and in the middle of a global semiconductor shortage that's going to last several years, rebalance the global supply chain of the world around semiconductors across Europe and the U.S. soil and put us on a trajectory that we have both economic and national security interest, right, satisfied over time. What do you think? Is that a big enough job? I mean this is such a massive, right, assignment. At the same time, we need it. I mean this is so critical for the planet and for the industry. And over the 9 months, we made a tremendous amount of progress. We're seeing talent come back into the company. These 6 business areas that we've laid out are gaining real traction. We're delivering proof points. When I make statements like we're on or ahead of schedule on Intel 7, 4, 3, 20A and 18A, over the same period of time that the competitors have slipped their process technology, the gap is closing faster than I forecast at the beginning of the year for this. And we're seeing the good products like Alder Lake and the proof points coming here. We're seeing this enthusiasm for graphics and what we're doing in network and edge. 3 years ago, Huawei was going to, right, be listening to everybody's phone. Today, that's all happening on open rebrand based on the Intel platform. We are opening up. Now we're delivering the best benchmarks in AI performance for inference and for training, right? Before, we gave NVIDIA a decade advantage, right? We gave market share to AMD, and we're going to start getting it back, right, as we look to it. These things are starting to build a cadence of deliverable. And maybe the last comment is, I mean, we pushed the Investor Day to February. We did so very consciously, right? We have a very clear set of things that we're trying to get done. And we're going to give you a lot more clarity. These are going to become reportable segments as we look to next year so that we're going to give you, the investor, a lot more understanding of the business areas respectively, so that you can really judge us in the business that we're laying. But so far, it's been an incredible journey. It is that important for Intel, it's iconic heritage, for the nation and the industry, and we are well on our way to making it successful.

Great recap. And Pat, I really appreciate the time you spent with us this morning, and everyone in the room, thank you very much.

Thank you.

Good to see you.