All right, why don't we get started with the next presentation? We're very honored to have Pat Gelsinger, the CEO of Intel, on stage with us. Before I get started, Intel has asked me to read their safe harbor statement.
The exciting safe harbor.
Exactly. Thankfully, it's not the entirety of that. 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 in Form 10-K and other filings with the SEC for information, 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 corresponding GAAP financial measures.
Don't you all feel better now?
I know.
Yeah.
We're so safe now. So now we can take that slide down.
Yes.
There we go. So, Pat, first and foremost, thank you so much for coming. It's been an exciting few years since you rejoined Intel, but before we get into the transformation, I want to talk a little bit about just the current macro trends. We've been through quite a boom and bust cycle over the last few years, during COVID, after the lockdowns, et cetera, et cetera, with differing paces of downturns and recovery times. So what's Intel's view on where we are in the market today? You know, kind of by your end market revenue segments.
Yeah. Yeah, good. And, you know, overall, you know, as we, you know, look at the, you know, the current environment, obviously, we've been through a strong bust, boom, cycle. As I'd say, pretty much everything that we laid out when we started this journey 2.5 years ago. We're 2.5 years into the transformation, you know, now, has sort of gone the way I would have expected at the time in terms of rebuilding the company. And in fact, if we go to the next page just for a second, right, I'll just quickly summarize this. Can we go...? Yeah. Right. And, you know, we said, "Hey, we got to rebuild our execution machine." And that follows, you know, two aspects: getting the product healthy again.
And I'll say, "Hey, we're now hitting or, you know, beating all of our product schedules." We have to get our process technology healthy again, and I'll say, "That's, you know, on track, our five nodes in four years." You know, we're going to build this new strategy, IDM 2.0, and seminal to that is the foundry. So you know, that's now well underway. We have to rebuild the financial disciplines in the company, right? You know, and, smart capital, you know, operational efficiencies and starting to do the value unlocks, like we've done with Mobileye and now our mask operation, IMS. And of course, you know, the second big thing, we weren't expecting this major economic cycle, and we weren't fully expecting this huge AI boom, right? And now have to fully embrace that as well.
But when you put all that together, I say, "Hmm, you always have some uncertainties, some aspects of unexpected," you know, but we're executing this strategy. You know, from the current market expectations, you know, some things, hey, you know, we said the client business would be healthy again, and the client business is healthy again. You know, inventory levels are solid. Our market share is improving. You know, we're about to go into a next major product, right, with Meteor Lake, you know, which will usher in the AI PC generation, a Centrino-like moment. So we feel good about where we are, and we feel good about the roadmap of capabilities going forward. You know, data center, obviously a lot of scrutiny on that.
You know, we said Q3 was down a bit from Q2 as we work through inventory. I'll say incrementally, hey, it's right where we thought it would be at this point. You know, Sapphire Rapids, you know, Xeon Gen 4, you know, very healthy ramp, crossed 1 million units, more and more AI use cases, you know, driving Sapphire Emerald, the next generation and next year's roadmap, very healthy. You know, in our networking business, we've had to fight through a lot of inventory, you know, challenges in this economic cycle. You know, taking all of those things together, you know, at this point, we say we're above the midpoint of our guide for Q3. You know, so we're feeling good about Q3 as that progresses.
The overall market, right, when you think about it, obviously inflation, you know, Ukraine, you know, U.S.-China dynamic, and I'll say some relative weakness in the enterprise market. You know, against that, though, hey, we're, like I say, above our midpoint, you know, of where we expect to be at this point, and overall, we're feeling pretty good about the progress that we're making to, you know, be able to deliver Q1 above expectations, Q2 expectations, you know, and Q2 to Q3 growth. So overall, we feel pretty good.
Perfect. Thank you so much for that near-term stuff. Now, I know in your seat, you care about the near term, but almost everything you do is focused on the long term and taking care of that. So let's dive a little bit deeper into the transformation topic that you have up on this slide. And, and the base of the transformation and everything else, in my opinion, at least, has to come from the manufacturing side.
Mm-hmm.
So five nodes in four years, or as John has talked about, maybe even six. Talk a little bit about where you are with that. Are you still going to intersect with leadership at 18A, and what are the next steps?
Yep. And you know, we feel, you know, obviously, Intel 7 is done. Intel 4, with Meteor Lake, you know, is nicely on track. We'll be volume ramping that, which means I'm volume ramping those wafers now to have, you know, volume products out in Q4. So Intel 4, looking great. Intel 3, with Sierra and Granite, really being led by the server products, you know, looking great. As we said in our data center update earlier in the year, you know, we you know, essentially accelerated those schedules for Sierra and Granite, so they're looking very good for you know, first half for Sierra Forest, you know, and shortly thereafter for Granite. So that's looking very good, and the products are getting good acceptance.
So, you know, seven, four, and three, you know, 20A, right, begins with Arrow Lake, you know, our next generation client product, and that also is looking very good. We've taken up the volume expectations on the internal factories there. There were some rumors about that in the industry. They were, you know, just false, right? We've increased the volume expectations in our roadmap for Arrow Lake, so that's looking very healthy. I'm looking forward to showing that off in the near future. You know, if I'm gonna ramp that, you know, next year, I got to be able to start showing it pretty soon. You know, maybe there's events coming up in the near future, Ross, where we'll do that. But,
Innovative.
Yeah, maybe. I got to think about that.
I'm bad.
Yeah, our innovation conference in September. And then 18A, though, is sort of the critical, critical completion point, right? When we sort of finish the five nodes in four years, get back to transistor leadership. And I would say, and I was just in Oregon yesterday with our technology development teams, looking good. You know, we feel like we're solidly on track with 18A. We're getting, you know, the... You know, and, and this, you know, for foundry customers, hey, they need good, solid PDKs. They need confidence that we can do this. You know, the amount of interest that we're getting in this area is progressing very well for it.
The technology itself, you know, is progressing very well, and as we said, we believe, you know, that that will be manufacturing ready end of next year, giving us leadership position in 2025. We're making good progress for our internal products, like Clearwater Forest, our next generation client products, you know, are all in the latter design phases, but also foundry customers. And, you know, in fact, you know, we are very pleased that, you know, the, as part of our smart capital, you know, strategy overall, you know, it's build shells, you know, get incentives, investment tax credit, U.S. and EU Chips Act programs, but also customer prepays. You know, and we've now received a large customer prepay for 18A capacity.
So now customers are getting confident enough that they're putting dollars on our balance sheet to accelerate our 18A capacity. So quite excited about that. You know, so we've, I've, you know, begun a process to accelerate the build-out of Arizona, right? And, based on that, customer prepay. So Arizona is great for us. Others might have difficulty. We love it, you know, so we're accelerating our construction build there. We'll accelerate the capacity buildup, that we have in Arizona. And the CHIPS Program Office submission, right, is now in the CHIPS Program Office hands for the Arizona site, the Oregon site, and the next two projects, Ohio and New Mexico, will soon be in their hands, as well.
So overall, everything is coming together, as we've said, and, you know, this, customer prepay really is a, you know, a strong, exclamation point on momentum for 18A and the manufacturing capacity for that.
I have a couple follow-up questions on the technology, but one clarification from that data point that you just said on the prepayment. I know it's always difficult for you to be able to talk about those foundry customers-
Yeah
... by name, by year, volume, all those sorts of things. But the fact that they're prepaying is a good sign of their commitment, and might be the best sign that you're able to give for a while. Is that true?
Well, you know, we would say, you know, in this regard, you know, we want to get public with some of the customer names over time. You know, we still hope to do that this year, but getting a significant customer prepay is, you know, a pretty profound, you know, statement of affirmation of their confidence that they're putting, you know, prepays on our balance sheet for us to not only commit capacity to them for 18A, but to accelerate, you know, our build-out as well. And, you know, it's just a great timing, for that, particularly as we're in front of the CHIPS Program Office, because they're asking the question: "Will you be able to fill this capacity?" Well, I have customers lining up and giving me prepays for it, so it's sort of like, okay, that's a pretty profoundly positive statement.
Not just for Arizona, not just for 18A and Foundry, but also, you know, in Washington, D.C., for the CHIPS office.
So a couple kind of tactical or at least chronological questions. So Intel 4, Meteor Lake, those wafers started months ago-
Yep.
-already.
Ramping like crazy, and that's... and again, that's also our first EUV node.
Right.
Right, which is an important technology statement.
Then Intel 3, you should be starting those for Sierra Forest imminently, I would assume.
Well, yeah. Yes, yes, they will start soon. The production steppings will soon go in. Obviously, we have, you know, the A steps on Sierra and Granite that we've been broadly sampling customers going through the validation, and now it's both dialing in the process and any final refinements to the design, and then we'll soon be sending what we believe will be the production wafers into Fab on Intel 3.
Great. Then the longer-term one, and the real question on the technology side is, is that transistor leadership at 18A is interesting, not only because Intel catching up, of course, which is important, but also there is some significant transistor architectural changes on both the RibbonFET side, or gate-all-around, as well as on the PowerVia, the backside power.
Mm-hmm.
Talk a little bit about where you think you are relative to the competition on those to truly differentiate.
Yeah, and you know, remember, on both of these, this is new invention, right? You know, this RibbonFET, this is a new transistor structure. You know, if you think about you know, the planar transistor, right, you know, to high-k metal gate, strained silicon, FinFET, you know, moving the gate all around is a you know, once a decade, once every 15-year kind of transistor change, right? So it isn't just like we're saying, "Oh, okay, yeah, let's go create a new transistor." You have to invent new molecules, new physics, new chemistry, you know, work with the equipment vendors, and we're now to the point that we're quite confident that we have it all figured out, you know, for the Gate-all-around transistors.
Like I said, I was there just yesterday, and I view the gate-all-around transistor that we're now producing to be a work of art. And if you're a device physicist, you look at the SEM diagrams on this, and you just say, "That is freaking elegant." Right? You know, it is just looking good, you know, for us. And now it's... You know, of course, you know, creating an elegant transistor, but then getting it that it's volume, reliable, producible, performant, et cetera. You know, but we're in the final throes, you know, of that, as we said, to deliver transistor leadership next year. But the one that has really surprised the industry was backside power, right?
Being able to do this wafer level assembly, you know, processing the, you know, metal layers, flipping it over, connecting it to the backside and creating essentially a wafer sandwich, but then being able to deliver power through the backside network and, you know, some in the industry have said we'll get as much benefit from the backside power as you get from the new transistor, right? And this is an area that Intel is multiple years ahead of the industry, you know, as we've been, you know, analysts and others have come out from some of our disclosures there. So this is an area that I'll say has been a profound leadership statement, right? You know, hey, the transistor, you know, Samsung's working on gate-all-around, so is TSMC, and you know, ours is a work of art. You know, we'll see how theirs show up.
You know, but the backside power is now an unquestioned area of leadership, and it's really important because as you get to these very high performance, and particularly as we start seeing high performance AI chips, things like power delivery, IR drop, you know, and also moving all of that power network to the backside and giving you more signal integrity, you know, on the design, it's highly, you know, I'll say performant, but also highly designable, right? And that's what we're finding from some of the early foundry customers. They're saying, "Wow, I get a lot of area improvement," right? So because, you know, ultimately they're paying for, you know, foundry customers are paying for how many millimeters square is their design gonna take.
If I can get more millimeters square for their design per wafer, wow, then that makes my wafers that much more valuable to them as foundry customers. And some of the, you know, early designs there have been really quite impressive. The area improvements that they're able to get because of the backside power and the performance improvements because of a better metal, IR delivery and the new transistor, we're feeling really, really good about 18A. And of course, we're not done, right? You know, yesterday they just gave me all the stuff for the one after that and the one after that. So, you know, this is, this is a race, right, to the future of science, chemistry, physics, and manufacturing, and we're feeling good.
Why don't we stick with the foundry side, because you mentioned it before. Where do you think the biggest advantage Intel will have in taking market share in that will reside? Is it going to be from the technology side? Is it gonna be the geopolitical issues that we have in supply diversification? Where are you seeing the customers who are interacting with you showing most interest?
Yeah, the, you know, obviously, the geopolitical stuff, everybody wants more resilience in their supply chain. Check. You know, we are the company that provides that. You know, everybody wants good transistors, you know, good process technology, so they can build better products. Okay? You know, per the last update, we feel like we're making good progress, and we're getting a lot of interest from customers, in that respect. Obviously, everybody wants diversity in their supply chain. Right? I don't want to be incumbent in one place, somewhat irrespective, right, of the geopolitical aspects. You know, give me more, you know, choice, more diversity. You know, I want good cost, right, as well, right?
You know, and, hey, we realize that, you know, we're gonna have to go win, right, on the value proposition that we present to our customers, and we think about it as power, performance, area, and cost. Right? And I got to give them good power performance. You know, they have to be able to build their designs very effectively, and then I'm gonna have to compete on cost as we go establish our presence here. So, you know, we are off on doing all of those, and we're feeling quite good, you know, for that. But the near term thing that's been most unexpected is our advanced packaging. Right? And this is proving to be a big on-ramp, you know, for foundry customers, because right now there's an industry-wide shortage, you know, of technologies like TSMC's CoWoS, right?
You know, our technology, Foveros, is a little bit different than that, but largely is solving the same problem, right, in a, you know, a very effective way. So right now we have, I'll say, everybody who's doing high performance computing, right, and a lot of that is driven by the AI surge, right, is looking at us to become a complementary advanced packaging partner, you know, for them. And, you know, I believe this will prove to be a big on-ramp for foundry customers, because if I can satisfy a near-term, urgent need, get them comfortable with us and trusting us as their foundry supplier, my ability then to ramp them as a major wafer supplier as well, I think is huge. And also, you know, when we call it on the slide here, you know, we call it an open systems foundry.
You know, I have a lot of capabilities that TSMC or Samsung doesn't have. I do high performance computing, I do advanced packaging, I do thermals and systems, I do, you know, IO and connectivity. I have a leadership position in silicon photonics and optics. You know, we have leadership technology, you know, at multiple levels that I get to bring to those foundry conversations that my competitors don't have. So I think this, you know, the idea of the open systems foundry will be very differentiated, you know, for us, and in this surge of how do I build high-performance computing? And if you think about, you know, the world of, I'll say, Xeons and data centers, it was getting a little bit boring. You know, it's like, okay, let's go from 32 cores to 48 cores. No, no, no, I need 56.
No, no, maybe 64. You know, it was getting reasonably boring. AI is anything but boring in terms of the amount of innovation it's bringing into data center scale, and so much of that is what Intel's good at. So I think it really opens up new avenues of interest from foundry customers, you know, to look at us for not just our products, but our semi-custom products and our foundational technologies.
So the last question on the foundry side, just because there's, we could go on that for hours, but we have a lot of things to hit on. When you talked about getting the costs right, you know, you can get the technology right, all those other services that are unique to Intel can be great, assuming you can get the service orientation-
Mm-hmm
... of a foundry into your culture. But the cost of building things in the Western world, as opposed to the Eastern world, is just inherently more expensive. Now, I know the CHIPS Act will bridge some of that gap, but where do you think that Intel needs to land to be a competitive foundry from a cost structure point of view for the customers, and yet balancing the margins that you need for your investors?
Yeah, and, you know, we, you know, as, as we look at it, you know, TSMC has established a market, right? You know, super clear. Remember, I'm a customer of TSMC, so I know exactly what their wafer costs are or what their wafer ASP is. You know, they're presenting to their N5 customers, to their N3, you know, customers, their budgetaries for N2. You know, we know what the target is, right, you know, for that. And then to my teams, it's like, okay, how do I hit that target? Because I need to be able to, you know, be, you know, delivering better than their cost at that level, right? You know, I have to be able to present to customers wherever they are, right?
If it's $10 per wafer, you know, I have to be 10- , you know, per wafer that we present to customers to earn our way into the business. So we know exactly what we need to do in terms of getting our costs there, right? Also then, right, you know, as we look at U.S. CHIPS Act and ITC, they are designed specifically to close the cost gap, right, with Taiwan and Korea. That's how they're designed. You know, that's built into our business case. We expect our first CHIPS Act dollars to appear this year. The ITC, the investment tax credit, is part of law, so that's built into our business plan, as well. And obviously, you know, we just got to go attack our cost structure. That's part of what we've said in terms of the internal foundry.
How many heads do I have per wafer, you know, start per month, right? Okay, you know, if I'm not competitive with TSMC, what does it take to get competitive? How do I put more AI, machine learning, telemetry, you know, to get my head count costs equal to theirs over time? You know, so these are all things that we are well underway. And again, right, you know, I get to sort of crawl my way up that margin stack from where we are. And the value creation for the Intel shareholder as I work my way toward a TSMC-like margin structure is extraordinary, right? You know, and if you do any form of calculation on how you value, you know, Intel today, we're getting credited zero, you know, for those factory capacities.
If I start producing reasonable margin structures and don't even get all the way to TSMC, that is an extraordinary, you know, value creation story, value unlock, you know, for our shareholders over time. You know, we believe we're on track to do that. Obviously, five nodes in four years is expensive. You know, I'm putting a lot of capital, I'm churning capital rapidly, but as we start to get to those cost structures, we get to a normal cadence on a leadership transistor structure and get our costs in line, start realizing the benefit of U.S. CHIPS and ITC and EU Chips Act, you know, we think that we'll be in a very competitive cost structure. If I have the best transistors, you know, we're gonna do very well.
Part of that cost structure, discussion brings you into the Intel or the internal foundry model that you're-
Yeah
...putting in place. In the first quarter of next year, you guys are gonna split out your manufacturing operations, a P&L.
Yeah. Yeah. That's getting a huge discussion internally.
Yeah. Well, and that's—to me, the it's more important internally in the cultural change-
Mm-hmm
... than it is externally. We'll all look at the numbers and, you know, argue about what's good and bad and ask you 15 questions every time we see you on it.
Only 15? Is that all?
Yeah.
Yeah. Okay.
But to me, like the cultural importance in the company-
Yeah
... is changing. How have people embraced that? Because I know you're doing it now, even though you won't start reporting it until later.
Yeah, and with this, you know, let's just take, like, a little example. You know, my design teams now or my product teams are now being... you know, they are looking at their product decisions through the lens of a wafer price that's comparable to the TSMC price as they make their product decisions. They are making different design decisions, not because they're making them based on a wafer price, not on a wafer cost. That was deluding them into using two expensive technologies for different market price points. You know, so we were creating inherent inefficiencies by the bad economic model that we have of presenting cost, as opposed to a price that's a market price. A second example has been, and we now are charging our internal teams for wafer expedites and new product introductions, just like TSMC does.
And all of a sudden they're saying: "Well, why are we doing another stepping? You know, why are we doing another expedite on this?" And you'd sort of say: "Well, hey, Pat, that's sort of false economics." It's not false economics, because if you're expediting wafers in a factory, you are making the factory less efficient. You know, it has a real cost associated with it to say, "Oh, I did another stepping because I had a crazy design idea. Let's run it fast." That hurts the factory's economics. And as we start bringing a real internal economic model to all of these questions, it's causing us to get efficient, more disciplined on every dimension, you know, of how we operate internally. And that's a major cultural shift that's going to take us multiple years to fully realize all the benefits.
You know, that's part of what we said, $3 billion this year in cost savings, $8 billion-$10 billion over time. That's a key aspect of how we achieve that, because we're going to drive in all of my business units, well, what margins is AMD able to get, right? On that economic model, why can't you be as good as AMD, right? And so my foundry team, here's what the margins that TSMC is able to get. Why aren't you that good, right? And it starts driving, I'll say, the appropriate, you know, disciplines and accountability to how we operate the company internally, and it's really fun to watch. Right? And I get to sit somewhat as judge and jury on some of these trade-offs that are occurring, you know, between them, and that's too high of a wafer price. No, it's not.
This is the TSMC price, and, you know, it just unleashes good, productive energy to make the company great.
We have about 10 minutes left, and there's two topics I want to address. So one is AI, and then the last one is just wrapping this into some financials, given the audience here.
Yep.
I think we could spend an hour talking about each alone, but I'll try to make it quick. On the AI front, are GPUs crowding out CPUs? Is that a permanent trend? Just generally speaking, how is Intel going to address AI?
Yeah, and, you know, let's step back just a little bit before we go specifically to the data center piece of it. Right. Remember, AI, as we said, will affect everything. Right? You know, I have AI in my hearing aids. You know, we're going to be building AI into the PC clients, right? You know, AI will be at the edge, you know, footprints as well. So we will be building, you know, an AI and that, you know, we, you know, call this AI continuum picture here, right? We'll be launching the AI PC with Meteor Lake in the fall, you know, that now includes, neural accelerators, the NPU, as part of it. You know, so we see that occurring and, hey, could you go run that AI model on, you know, for your Zoom real-time language translation? Could you go run that in the cloud?
Of course, you could. It's really expensive, really slow, high latency to go do that. You're, of course, going to push that to the edge, right? Are you going to do, you know, real-time human tracking in stores and manufacturing and supply chain locations, you know, with AI? Could you run that in the cloud? Absolutely. Are you going to? Absolutely not. You know, the laws of economics, the laws of physics, the laws of privacy, you know, will force it to the edge. So we really see this AI being pervasive. Now, of course, the hot topic today then, of course, is what happens in the data center. But on those other aspects, Intel has huge assets.
You know, I am going to deliver more AI tops than any other company in the world, you know, in 2024. Why? Because I put it into our volume products, right? You know, we're not talking about, you know, shipping, you know, millions of data center accelerators. We're talking about building hundreds of millions of clients and edge devices, right? You know, and hey, volume matters for what developers and ISVs take advantage of. So those volume plays are a natural area of strength for us. Coming back, though, to the data center, hey, you know, NVIDIA's, you know, hey, they worked hard, right? They have a great leadership position, you know, but right now, there's a bit of false economy there.
You know, huge demand, ASPs are, you know, high, you know, supply chains are constrained, you know, double, triple booking, you know, all these kind of things are underway, and lots of people are showing up, including us, you know, to compete, you know, for those. We've seen a rapid expansion of our Gaudi pipeline. We're building our supply chains to get much, you know, larger, you know, for our footprint there as we start competing, as well as others. You know, we'll start competing, you know, for those. And then the critical question is: how do CPUs versus GPUs play? Well, you know, I call it, you know, there's the three laws, right, of the data center. One is the laws of economics.
If I run a piece of software on a GPU at 10x the cost of the CPU, okay, you're going to run it on the CPU, right? And a lot of this is, you know, then driven by law number two, Amdahl's law, right? If you have a little bit of AI and a lot of general purpose, you know, then speeding up the little bit of AI, you know, workload by 100x doesn't improve the application because it's 10% of the total workload. So you end up with Amdahl's law effect and the law of inertia, right? These are decades-long application environments for security, regulation, privacy, all of those types of things, traditional areas of strength. You know, so I think right now there is a bit of euphoria, right? You know, we're sort of in, like, the third cycle of AI euphoria.
You know, if you think about the 50-year history of AI, and this one is big, right? I think, you know, it really is industry shaping, you know, what's possible. You know, but as we put more and more AI capabilities into all of our products, including Xeons, you know, and, you know, we expect somewhere between a quarter and a third of today's Sapphire sales are driven by AI use cases, you know, today, and that's ramping up nicely for us. So overall, I expect to see a more moderation there. We're going to be competing more for the GPU and accelerator, you know, but we also see workloads driving energy that will create opportunities for our CPU offerings as well.
Great. So AI is a positive driver? Is the key take?
Yeah.
Not a trade-off away from Intel.
Yep. And hey, yeah, they're doing well, right? We all have to give them credit, you know, for that. But, we're going to show up, and many of the technologies, you know, that we have on the table, memory, IO, power, cooling, you know, design, you know, advanced transistors, these are all going to play significantly, you know, for AI opportunities for Intel's products.... but also our foundry.
The last question I wanted to hit on in the last four or five minutes we have, was kind of wrapping all this together into a little bit of a framework, a financial framework. Your last analyst meeting was, you know, coming up on three years ago. I might guess that you might have one next spring as well.
Mm-hmm.
Just tends to be your typical timing, so maybe you'll update this then. But at the time, and since then, you've talked about good companies have kind of a 60/40 gross and operating margin model.
Mm-hmm. Mm-hmm.
Let's just focus for now on, on the gross margin side of things. At the time you laid that out, it seemed like it kind of implied roughly $100 million, $100 billion in revenues. You're about half of that, roughly speaking now. We think things normalize cyclically. You get up, you know, call it $65 billion, instead of 50-55, somewhere in that range. The question I have is: What is the revenue level necessary, just in round numbers, to get to that margin structure so the investors in the room can get the return that they're looking for? And then how do you get there? Is it market share? Is it more data center than PCG? Is it a huge foundry assumption? Just the moving parts in a general sense would be very helpful to people.
Yeah. And, you know, in terms of hitting those metrics, you know, we don't—we believe that we have a lot of cost work to do, right-sizing the business, margining the business, et cetera. So, you know, we don't believe we have to get to that revenue levels to get in that zip code to go get there, right? So I'll just say, you know, that's too high of a number for the aspirations we're setting on fixing the financial foundations of the company. Obviously, you know, we're out to grow the revenue line going forward. You know, we do see modest growth in the client business going forward. You know, we're now getting our market share back to historic levels.
We do think the AI PC will become an accelerant, upgrade cycles, but nothing, you know, I'll say heroic on the client side in terms of growth rate, but we do see that contributing to growth, you know, for the company. Data center, obviously, we're at a cyclic low, right? If you look at it, our products were uncompetitive, we lost share. We're now getting better products associated with that. You know, we believe that we'll start doing better on the data center business going forward. And we believe that AI accelerators and other aspects of the data center will be a meaningful contributor over a multiyear basis. In the networking, you know, business also, we're at sort of a cyclic low, you know, you know, for that business.
You know, again, we have a lot of technologies, you know, new products that we're bringing into that category, so we do see reasonable revenue growth there. I think, of course, you know, based on the earlier part of this conversation, how big we can grow the foundry business ends up sort of being that last big factor, you know, for us. You know, near term, as I say, we're seeing an extraordinary amount of interest for the advanced packaging technologies. You know, as we start winning some of these whale, you know, customers and, you know, the prepay was a, you know, a great proof point that we're making progress there. You know, I think that's going to, you know, start to really, you know, contribute. But, you know, foundry wafer customers don't materialize for several years, right?
Because you got to win their design, they got to get their designs done, and then they ramp, you know, the wafer. So that really starts contributing on a multi-year basis, but, you know, that's on track. So, you know, the idea, you know, of a much higher revenue Intel, that's why I'm here. But we have to also then underlay to it, you know, how do we get our, you know, basic financials well? The internal foundry model, you know, helps us get good financial accountability, fixing our core cost structure and our manufacturing operations. You know, obviously, as we start ramping, you know, the new nodes and we get past this rapid churn of five nodes in four years, hey, I'm getting more capital efficient, right, you know, as well. I'm not churning as much new capital that isn't producing free cash flow.
The final thing about the foundry model, you know, from our perspective, is that as it starts to scale, I am gonna keep those factories full forever, right? You know. Right? You know, because we're gonna-- instead of turning over those factories when they start getting good cash flow, you know, we're gonna be filling them, right, with other advanced packaging technologies. We're gonna be adding variants to the process technology for RF and wireless and mobile, you know, so that we're able to run those factory networks much longer, you know, with very healthy cash flows on the other side of it. And that's always been the gravy train of the foundry business, right? It's being able to run those depreciated assets on a much longer basis with good margin, wafers.
As we start layering that into, you know, the business, you know, I think this is an extraordinary, you know, value creation cycles for the shareholders, you know, as we pull this off. You know, I think appropriately, you know, we're 2.5 years into this journey. You know, when we laid out this bold strategy to start, you know, people should have looked at us and says, "Wow, you guys are, like, a little bit..." Right? Right, you know, as I joke, there's a, you know, there's a thin line between, you know, audacious and crazy, right, you know, from this, and, you know, which side of that line were you on, right, you know, for?
But as the evidence is mounting of five nodes in four years, of rebuilding the product execution, you know, the factory network coming online, you know, our ability to create this financial model of competitiveness with U.S. CHIPS, EU chips, and ITC and Skip and customer preface et cetera, all of the pieces are starting to come into place that whatever. You know, however skeptical you might have been 2.5 years ago, you have to be much less skeptical about our ability to pull this off. You know, and fundamentally, right, you know, we are rebuilding, you know, the iconic Intel, right, in the image of the Trinity, Grove, Noyce, and Moore. You know, secondly, we are making huge progress, you know, to rebuild the technology supply chains of the Western world.
You know, and fundamentally, this is the most important company to reestablish, you know, the geopolitics that we're interested in, you know, for decades to come. That's the mission that we're on, and I think we've made pretty good progress in getting that done in two and a half years.
Well, Pat, thank you so much for your views and joining us here at the conference today. Good luck with the continued success in the transformation.
Very good. Thank you, all.
Thanks, all.