Advanced Micro Devices, Inc. (AMD) Earnings Call Transcript & Summary

Good morning, everyone. I'm Stacy Rasgon, and I cover the U.S. semiconductor sector at Bernstein. Before we get started today, I'd like to say a word on how our interactive Q&A will work for the session today. [Operator Instructions] Also we're working with our partner, Procensus, to do live polling on the presentation to gauge investor sentiment on AMD after the session ends. A link to Procensus is also on the left of your screen. Again, this will be a live poll with immediate access to the results. For those that choose to do, please take the 60 seconds afterwards to fill it out. Okay. Now that all that's out of the way, it's my great honor to introduce our guest this morning, the President and CEO of Advanced Micro Devices, Dr. Lisa Su. So AMD, I think there's no company in my coverage, and probably no other company in the entire industry, that fully encompasses the true meaning of transformation quite like this company does. It is not that long ago that the question of long-term viability was an honest debate. The stock traded at a sub-$2 price at that point. But sometimes it takes adversity to bring up the best in us. And during that period, the company placed, I would say, some well considered and chose even gutsy bets on their forward product portfolio. They've redoubled their efforts on new architectural innovations, returned to a focus on high performance. They turned their process technology from an anchor to a set of wins and are now in a leadership position on that front. The balance sheet is now pristine, solidly net cash. The overall company is now consistently profitable, and they've gone fully on the offenses now with the primary controversy no longer run whether they're going to succeed in taking share, but rather how much they're going to take and how quickly it's going to happen. And our guest today has been here for all of this. She's been leading the company for the past 5.5 years as this transformation has unfolded. And I'm sure she's going to tell us today why she thinks that they're just getting started. So with that, it gives me great pleasure to welcome Dr. Lisa Su with us today. Lisa, thank you so much for joining us today.

Great. Thank you, Stacy. Thanks for having us, and we're happy to be here.

Great. So Lisa, in this format, I do like to stay away from near-term stuff, if I can. Today, given what's going on, I'm not sure I can avoid it. But before we dive in, I do want to start a little higher level and more open-ended. So again, you're on your 6-year tenure as CEO, with some admittedly remarkable changes in the profile of the business over that period. I guess, can you just talk a little about the position the company was in when you took over? What challenges did you anticipate at the time? And what actually surprised you? What was harder and what was easier maybe even versus what you had expected? And what are the guiding principles now that are determining your path as we go forward from here?

Yes. Absolutely. Well, again, thank you, Stacy, for having us. When you look at AMD, I mean, we're a company that's 51 years old, and we've always been about technology and pushing the envelope, and we've had great engineers. And what I saw, certainly, when I became CEO -- when I joined the company in 2012 and when I became CEO in 2014 is we perhaps did not have a strong execution bent in the sense that when you're in technology, you can't miss a product cycle, and we had, had a few missteps. So as we looked at what we had to do, and it was not just me but the team together, we said, "Look, this is a long game. We have to play a long game." It is a 5-year journey with any architecture, and that doesn't even necessarily mean that you've hit the peak. But what we've said is we're a technology company. We're great at engineering when we make the right bets, and we had to make some important decisions. And so our focus was around product execution; around manufacturing strategy, we needed to get a consistent manufacturing strategy around balance sheet. So it gave us enough time to do what we needed to do. And then there's a lot of earning customers' trust, and that was a big piece of it. So when I look at the various pieces of what has been important is I'm incredibly proud of the engineering team because we have built what I believe is a very repeatable execution machine. And that doesn't mean we're going for home runs every generation, but it means we're going for doubles and triples that keep adding up. And that's what you see, that's what you see with the Zen architecture. The first generation of the Zen architecture was good. The second generation is, I think, great. And frankly, we're getting ready to launch third generation of Zen, which will be even better. But that was the way we thought about it. This is going to be a multigenerational, multi-year journey. On the back end of it, we had to fix the balance sheet. We've taken down a lot of debt. And we've also had to really earn customers' trust and belief in our road map, and I'm proud to say that we've done quite a bit of that. But as you started with -- we're just getting started. I mean this is technology. It's absolutely a long game. And as we think about sort of what we need to do, there's a lot of opportunity to continue to consistently grow the company.

Got it. I do want to get a little bit into the near term now as much as I don't want to, but I think I need to. Obviously, COVID has been a huge issue on multiple dimensions for everybody in the industry right now. Just talk a little bit how it's impacting your business, right, both from a revenue standpoint as well as supply chain perspective. What are you seeing right now? And how is your visibility trending now?

Yes. No, there's no question that this year has been about managing through the various phases of COVID-19 across the world. I think in the first quarter, we had a lot of focus on what was going on in China and what was going on in Asia with the supply chain. I would say that we worked through those issues relatively quickly. I think our supply chain is relatively resilient, and so that hasn't been much of an issue here in the second quarter. I know some of our customers are continuing to work through some component issues. But overall, I would say that the supply chain, although there have been lots of moving pieces, has actually ended up relatively resilient. When we look at the demand environment, there are a bunch of puts and takes, right? So the first quarter, we saw weakness in China, particularly around offline channels and places where you needed to have stores open. That's recovered nicely here in the second quarter. As we look at the overall demand environment, I would say that the demand for computing has been relatively good. We see strength in cloud data centers. We see people accelerating some of their demand. If you think about collaboration, what we're doing here, if you think about just resiliency of the infrastructure for work from home, there's been strong demand there. And there's actually been strong demand in the PC segment, and we looked at it as is it a very short-term thing. It has continued to be stronger than, perhaps, one might expect for the second quarter given typical seasonality. But overall, we've seen some positives. We continue to watch for some demand softness. There is a thought process that the second half, you might see some consumer demand softness. I think it's too early to call that yet, but that's certainly something that we're watching very closely and just looking at how business environments go and how CapEx spending will go in some of the businesses that are more impacted by the macroeconomic conditions. But overall, I would say it's been a very dynamic environment, but we've been able to manage through it pretty well on both the supply and the demand side.

Got it. That's helpful. I want to move to some of the product stories now. I think I want to start with client. You've done very well in desktop, especially in the DIY market. I think, overall, if I just use the Mercury data as a proxy, I mean, your shares probably doubled or more off the bottom. I think your ASPs have probably doubled or tripled off the bottom. What is the -- is that the full opportunity within desktop? Like are you saturated at this point within the DIY market? Like how much more opportunity is there for you to generate more upside on that piece of the market, whether through share repricing or both?

Yes. Actually, it's a very interesting way of looking at it because if you look at the PC market, I mean, the PC market has really ended up being a very strong market overall. And if you think about it, $30 billion of processors are being sold a year. And for us, we were quite underrepresented. Now if you look at the share progression for us in the PC market, starting with the desktop DIY market, that is where we have seen the strongest uptake, so our market share is quite strong in the DIY market. And the reason for that is, frankly, the people who build their own computing systems are purist in terms of technology, right? They buy the best processor. They do all the research. They go on to Tom's Hardware or NanoTech and figure out what's the best value. And so we've done very well. I don't believe we're all at all "saturated," though. I think we're actually -- when you look at desktop, there's a DIY market for enthusiasts. And then we see system integration, which is another important part of the market where people buy prebuilt specialty systems. We see desktop OEM as an opportunity for us, which is part of the larger OEM business. I think we're very pleased with our desktop competitive position. I mean it's extremely strong, whether you're talking about at the very, very high end with what we do, with 16-core processors or even -- we recently introduced some sort of more mainstream processors at $100 or so. So I think desktop is a good market. I think it will continue to be a share opportunity for us, and we view it as an anchor of our competitiveness because we're winning the hearts and minds of people who are really computing-savvy, and that's important for the rest of the business.

Got it. And I was just reading into notebooks. So notebook is a different kind of market. It's primarily 100% an OEM market rather than a DIY. The games have been more modest. Obviously, the road map for notebook has been a little behind desktop as well. You just introduced new products. What can we expect there in terms of, like, share expectations as well as ASPs? We haven't seen the same kind of ASP lift in notebook either. Is there opportunity there? Do you expect that to play out ultimately the same way that we've seen in desktop?

Yes. Stacy, the way to think about it is we were actually very deliberate in how we rolled out our road map. The idea is, look, we can't do it all at one time. And frankly, that's not the best way to introduce things. And so we prioritize server and desktop first from a performance capability standpoint. And then, notebook is a large market. And frankly, notebook is what is the most exciting part of the market right now because everybody wants to buy a notebook, right? We've seen that shift, that notebook is actually quite a bit stronger than the desktop business just given people want the flexibility. And we've made good steady progress, actually. I'm very pleased with our notebook business. I think, this year, with the introduction of our Zen 2 based notebooks with Renoir, we are extraordinarily competitive. It's also now 3 years into sort of the platform cycle for OEMs. And so the OEM business requires one that we work closely with our partners, HP, Lenovo, Dell, the rest of the OEMs. It also requires that retail channels get to know us and get to know the quality of the product. And it requires sort of maybe the less expert user, sort of the regular PC buyer to also know AMD. And I would say that we've made great progress here, but we are a little bit, I would say, time shifted from the desktop story. And so I do believe that there is strong opportunities for us to grow notebook. I think that's especially true. When you look at our mix of business in notebook, right it's very consumer-focused. So a lot of people know AMD from a consumer standpoint, and we are investing heavily in the commercial part of the business. And when we look at some of the new platforms coming out from our partners, like HP and Lenovo, we're actually really excited about our commercial lineup and I see that as another multiyear growth vector. And I think it really stems from the product competitiveness, but then also the trust that the customer set has in the AMD road map and the AMD brand really selling in commercial as well as consumer.

Got it. What does the competitive environment look like as Intel starts to ramp 10-nanometers in a more deliberate way? It seems like they've finally gotten that processor where, at least, it's good enough that they're doing it. They've got Tiger Lake coming out in the second half, competing directly with Renoir. Like how do you see that sort of competitive environment play out?

Yes. So look, we always count on our competitors being strong. And we've been sort of thinking that 10-nanometer would be the competitive point for quite some time. Our road map is very strong as well. And I think we have -- again, back to this track record of multigenerational capability and being very dependable for the OEMs. So I understand that the competition is going to be stiff. I think we feel very good about what we have beyond Renoir. And so the competitive environment, from our standpoint, continues to look solid over the next couple of years.

Let's talk a little bit about that road map, and I'm going to fold in one of the questions. It's actually at the top of the list from the investors watching. The Renoir at this point is on 7-nanometer, and you've been sort of first to that node. I think you've talked about Zen 4. I'm not sure what the code name is of this one, the Zen 4 is supposed to be on 5. I don't know that you've given -- I think from the Analyst Day, you've kind of said that sometime before 2022, I think, is what the time line is. There's been some press reports overnight that TSMC is accelerating 5-nanometers in the second half of 2020, and that AMD is potentially doing that. Can you just give us maybe a little bit more of your views on the timing of road map for Zen 4, especially as it relates to 5-nanometers in process?

Yes. Well, that's a little bit far out, Stacy. So what I would say is...

Can you just give us more generally, like anything you can talk about in terms of like how you're viewing the process road map as it relates to any of competitiveness?

Yes. So being on leading-edge process technology is important for us. We've always believed that as long as we're sort of close, I think we're good. From the standpoint of where we see 5-nanometer, I think 5-nanometer is an important node. It will be one that we will use quite heavily in our road map. I'm not ready to talk about timing yet, but I will say that Zen 4 is deep in design, and we're very collaborative with TSMC. The way to think about it is the process nodes usually start with mobile, and mobile is usually a simpler process from the standpoint of just the performance it's trying to get. And then we come in with high-performance features. And our collaboration with TSMC has given us a lot of capability in optimizing for high performance. So yes, 5-nanometer will be important. Zen 4 as well as our GPU road map, we'll be using 5-nanometer, but we'll talk about timing as we get a little bit closer.

Got it. Okay. And then share targets, I'm going to get to the server share target in a moment when we talk about servers. But I don't think you've ever actually given like a concrete share target in client. Although I think, historically, your shares speak that where your notebooks were probably, what, high teens, close to 20%, desktops were probably in the mid-20s maybe. How do you think about the evolution of current share targets? Can we work toward or even potentially even exceed those? Is there any reason why we couldn't get back to share targets that look like that?

A couple of years ago, I think we -- our commentary was, look, there's no reason we shouldn't get back to historic share targets, and that was when we were, let's call it, in the low double-digit range. If you look at our share today in both desktop and notebook, we're now approaching that high-teens range. I think the way we think about the market we don't have, let's call it, a cap on our share. I don't think there's a belief that, hey, if we get to 20%, that's all we can get. I don't think that's the way we think. I think we think that the products are really, really good. I mean they are best in class in many aspects. The relationships are really good. We have work to do to get people more familiar with AMD, so we recognize that there's a go-to-market effort that has to layer on top of all the good technical work. But I don't see a particular share target. What I see is continued growth. And we believe that the PC market can grow for us -- the PC market can grow. Even though the overall market may not grow, we think our revenue can grow, let's call it, mid-teens over the next 3 or 4 years. So that implies a certain amount of growth in the market share.

Okay. Got it. So bigger slice, even if the pie gets smaller.

Hopefully, the pie doesn't get too much smaller. Actually, there are some that are saying that the pie might get bigger again because everybody needs more PCs. The idea of...

I bought 2 in the last 8 weeks, so maybe.

I appreciate that. I hope they were AMD, Stacy. But -- look, I think there is a view that PCs have actually seen a resurgence when you think about just how important they are, right? I mean they enable so many things. And so this whole -- this idea that you need one PC per person, not one PC per household, I think, is a longer-term trend that we're watching as well.

Okay. Let's shift over to servers. So obviously, I think when investors think about the investment case in AMD, that's probably at the top of the list. I think it makes sense. It's the area where your market share was the lowest. I mean it was -- not that long ago, it was almost 0. I mean we're not there anymore, right? It's been good. And I think when people are looking for the forward, I mean, certainly, that's where the biggest opportunity is. I mean you launched Rome -- when was Rome launched, August, I think, last year? So it's...

Yes. That's right, August '19.

More advanced and actually better product than Naples. And you've been sort of -- you've been on the Zen architecture now within service second generation, but over about 3.5 years. I guess can you talk a little bit about what's changed from Naples to Rome? Like how has the customer uptake of Rome been different from Naples? And what can we expect as we drive to Milan, which is -- my understanding is Milan is supposed to come out, I think, Forrest has told me 4 to 5 quarters after Rome, but you've kind of put Milan kind of into this year, competing against Ice Lake. How should we sort of think about the progression across those different generations of parts as you drive towards your share target? I think the share target right now is still low double digits by the -- you can give us your view on where the share -- on the targets, but maybe a little bit on that progression would be helpful.

Sure. So no question. Data center is the most strategic part of our business. And the way I look at data center is, again, I said it takes like 5 years to really see an architecture bloom, and we're a little bit more than halfway through that. From our viewpoint, we've actually seen very good progress in the server market, although it's been steady, and some have said, "Well, why isn't it faster?" I think the way to think about it, again, is we're not looking for, hey, let's take share at a point in time. We're looking at how do we build really a franchise around people optimizing and designing to our EPYC processors. So Naples was a good product. It really helped build the ecosystem. And in Naples, what we saw is we had a lot of customer interest, but they were kind of doing -- sort of trying us out, understanding how the AMD architecture works, and they used us on some instances. What we're seeing in Rome, I really do see Rome as -- first of all, it's an extremely competitive product. When you look at it versus Naples as well as versus the competition, it's up to double the performance in certain workloads. And so it's positioned very, very well. What we see across the market is -- in the cloud data center market, we're seeing just significant interest in both public-facing instances as well as internal workloads for some of the largest cloud vendors. We see good uptake in Enterprise, and then we're winning a lot in HPC. So when I look at the progression of the server business, it's actually going a bit like what you saw in the PC business, but it's taking a bit of a longer time just given that the Enterprise business tends to move a little bit slower. That being the case, what we see is really good signs of customers realizing the value proposition. Like what makes me happiest is when a customer says, "Wow. I can see that, and I want to now use you on all of these other workloads." And that's sort of the phase that we're in with Rome. It's going to be a workhorse for us. Even beyond this year, we believe Rome has a lot of applicability. I think Milan adds -- Milan is our Zen 3 based product. It will be our third generation of EPYC. It is doing very well in the labs. It's doing very well in early customer trials. And we see it at the end of this year sort of on that track of, we said, 4 to 5 quarters. We should be starting to ship at -- in the fourth quarter.

Got it. What is the value proposition that you do? Is it just more cores, more I/O, more memory bandwidth? Is it something broader than that in terms of how you're providing to your customers? Like, what...

Yes. So yes, if you look at the value proposition for Rome, it is more cores. It is more memory bandwidth. It's -- you get to do the -- do your optimization, so we're very flexible in how customers choose to put things together. If you're more I/O intensive, you're going to really like Rome because you get a chance to connect a lot of I/O to your choice of CPUs. If you're very CPU-intensive, you're going to really like Rome because you have a lot of cores to deal with if you're very CPU-intensive. And I think the important piece of it is people don't understand exactly how much optimization that can be done in the data center environment. So we see large cloud providers now sort of optimizing their software and their capabilities to make Rome operate better. And so we're actually seeing improvements as we go through deployment, and that's really important. So I think the key there is continuing to work with our customers, continuing to expand the number of workloads. Our goal is not to have a small share of a customer's business, but actually be a large portion of their business because the value proposition is strong. And so that's what we're working on in the server business.

Got it. So I do want to touch on the share question because this is one that is top of mind. So you have this share target, which is -- at the time it was, I think, mid-single digit end of '18 and then low double digit kind of middle of this year. I don't actually like the share target. I don't like how it's defined. It is what it is. I'm not sure you'll like it either, frankly, but...

By the way, Stacy, I'm not sure I like it either because we keep talking about it every time we talk. But yes, please ask your question.

Like how should investors just going to be thinking about the trajectory of share gain from here? Because it has, at least, been said you've taking share every quarter year -- every quarter for 3 years, like we've seen share gains. And it hasn't necessarily inflected, but it has been very steady. And it's been up every single quarter. How should we think about the trajectory of that as Milan rolls in, as new innovations come in, as the process -- I think the process gap probably stays there as you move over to 7-plus and into 5. How should we be thinking about that?

Yes. So let us talk about the share target because I do want to make sure that we're clear about that. So what we were trying to do is give some idea of what the length of the cycle would be, and that was the reason for the share target. And our share target was actually 10% by the second quarter of 2020. It was 4 to 6 quarters after launch, and 6 quarters is the second quarter of 2020. We're on track to that. We have good visibility to that. I think we feel good about what we're seeing. In terms of how you calculate share, it does depend a lot on the TAM. And so without getting into all the gory details, I think you can calculate it different ways. What we're trying to do is show a trajectory. Now let's talk about sort of where we see the business going. I actually believe 2020 is an inflection point for the server business. I feel it, and I feel it as it relates to customers now giving AMD the mind share to take some of their highest priority workloads to Rome. And so we're working very hard on that. You've seen one quarter. I think as we go through the second quarter and the second half of the year, we're going to transition from share targets to more of a percentage of AMD revenue because that will give you a better idea of the progression. But overall, look, I feel good about it. I mean, this is a multiyear journey. We have -- beyond Zen 3, we have Zen 4 and Zen 5 in the pipeline. I think customers see the value of having us in their data center, and we appreciate that. And so we're going to continue to make progress. And our goal is to earn the repeat business, not just the business of today. So the server business will continue to be very strategic. I probably won't give you another share target. But what we did say, though, and it gives you an idea of what we think the size of the businesses can be, when we did our Financial Analyst Day a couple of months ago, we said we saw the server -- or the data center business for us being upwards of 30% of overall AMD. And today, it's about mid-teens. So it's going to grow significantly over the next 3 or 4 years.

Got it. So that data center business is not just CPUs. That includes GPUs as well. Can you talk a little bit about what you're doing on the data center GPU front? And this is a business that we only just started to see it layer in a little bit like last year. And like -- give us a little bit of idea of the applications that are driving it and maybe even some of the differentiation that you bring versus -- again, this is another market where there's another sizable incumbent that you're coming in. So what are you bringing into the table on this? And kind of where is that business today? Where is it going?

Yes. So high-performance GPUs are very strategic for the way we think about the business in data center. And the way to say it is the business is still relatively small, so our data center business is predominantly CPUs with sliver GPUs. We had some early wins, and it's been public what we've seen in cloud gaming with Google and Stadia. We had some other cloud gaming, desktop-as-a-service type wins. When you look more broadly at the business, we see that the linkage between the design of the server CPU and the data center GPU can actually bring a lot of synergy. And so we're quite proud of our high-performance computing wins in the large supercomputer bids. We've won 2 out of the 3 large supercomputers for the U.S. at Oak Rouge National Labs and Lawrence Livermore. These are very, very large installations. They're both aiming to be sort of the top of the top 500 supercomputing list when they come out. And we won those as a result of, yes, our CPU road map, but also not just our GPU road map, but how we put those 2 together. So I think we're going to see more optimization of the components as we go forward. I think HPC is a very important market for us. It's one that is also very technical, and so people will buy based on what they see in the technology. And then we also view machine learning and AI as an opportunity for us, albeit we have some work to do to get the software in sort of shape for DML opportunity.

Got it. Can we talk a little bit about those hyper -- those HPC opportunities, Frontier and El Capitan? How should we think about timing and opportunity size? And even if -- I don't know if that makes sense to think about it this way, that the CPU versus the GPU contribution, it does sound like it's really the 2 of them that was driving the win in the first place. Like how do we think about those opportunities?

Yes. So the timing of those installations are fairly public. You will see Frontier in the latter part of 2021, so late next year. And those will, -- like I said, those will be relatively large installations, so that ramp will go quickly. The business is actually more heavily weighted towards GPUs. If you look at the ratio, it's -- the ratio is such that it's more heavily weighted towards GPUs. It's hard core engineering, so it's going to be a ton of work to get it done. But we are extremely excited about it. The Lawrence Livermore opportunity is an early 2023 installation, so the majority of that revenue for us will come in 2022. So you can see, we have a nice growth path that will be meaningful for the GPU business in 2021 and 2022 just based on these 2 large installations. Now the way to think about it is winning these are like winning like the Halo products, and so it's really good validation for our products. Obviously, these products are still in development. But the validation, as we talk to now other opportunities in the space and, certainly, commercial supercomputing as well, I think it's just good validation. And there's a lot about that road map that we have yet to disclose. So we started a piece of that with our compute-optimized architecture at the Financial Analyst Day. You'll hear some more about that as we go through the next number of quarters. But it's a big investment. It's another big strategic investment that we expect to pay off over the next 5 years, but not necessarily over the next 2 quarters. That's the way to think about it.

Is this the primary reason that you bifurcated like the gaming architecture and the data center architecture in GPU? So we've got RDNA in gaming. Is it CDNA in data center at this point?

That's right, Stacy. Yes, yes.

What does it look like between those architectures? And again, was this sort of the -- was this the primary driver to do that, so you could actually just be more optimized for these types of applications or it's necessary to do this, it seems?

Yes. I think what you see is, if you want to be best in class, frankly, you have to be as efficient as possible on the silicon die. And there are some gaming features that are not needed in compute, and there are some compute features that are not needed in gaming. And we did decide to bifurcate the architecture. I think that, again, goes to our commitment to these markets. I think gaming is always going to be a very, very important market for us. I think compute, we see just a tremendous growth opportunity over the next 5 years. And so that bifurcation of the road map allows us to really optimize for the workload and for the use case. And yes, that's exactly why we did it.

Okay. Got it. So I want to touch a little on the gaming opportunity both PC game and GPUs as well as consoles. But I guess on the GPU on the PC side, it's going to be an interesting year for gaming GPUs. Both you and your competitor look like they have new products that are coming out second half, hopefully. You've been closing the gap. I know there was a period of time where the gap was fairly wide, especially on the premium tier. And over the last couple of architectures, I think that gap has been closing. You've been coming up. What do you see in the second half? What is the new -- I don't know if you're calling it Navi 2, I'm not sure what the name is...

Yes, we...

What is that actually doing in terms of improvements over the prior Navi? And how do you actually see that competitive environment evolving as we kind of move forward on this?

Yes. Stacy, the way we think about our graphics architecture is we actually use that graphics architecture across a lot of products, right? Yes, there's a lot of anticipation right now in the enthusiast market for our big Navi or our Navi 2 or first RDNA 2-type product, and that's important. Again, it's a Halo product. And enthusiasts love to buy the best, and we certainly are working on trying to give them the best. That architecture then goes through the entire stack. So if you think about it from mainstream GPUs all the way up through enthusiasts, and that architecture also goes into our game console products as well as our integrated CPU, GPU or APU products. So the investment in the architecture is very, very important for us across that entire consumer product portfolio. And yes, we're very excited about Navi 2. I think it's an important product for us. It's one that the teams are working very, very strongly on -- over the next number of months, and we look forward to that launch.

Got it, got it. How can we think about the prospect for hardware-enabled ray tracing? Is this something that we see on the consoles first and then in the PCs? Does it come together? Like how do we think about that?

Yes. So the RDNA 2 architecture that is being used on both the current generation game consoles that are going to launch shortly as well as our Navi product portfolio is the same. So that's an important feature, and you will see it. I don't know if you will say is it before or after, you'll see it in a similar time frame in terms of what we're trying to do. And the idea there is to, frankly, leverage the ecosystem. What we've seen is, look, ray tracing is very exciting, but you also need the game developers to have the application environment ready for ray tracing. So the fact that the consoles as well as the discrete GPUs will use a similar ray tracing architecture is actually a positive and, I think, is going to accelerate the adoption of ray tracing overall.

Got it. And to touch on the consoles a little bit, I mean, obviously, like this is -- they've been under pressure at the end of, like, a 7-year cycle, and you've got a new one now coming. How do you see this cycle potentially playing out versus the last console cycle? Do you think the peak could be higher? Do you think it lasts longer? And I guess just in terms of your own financials, like what does this look like? Are the margins better, the pricing? Like how does this compare versus what we saw for the last one, which, to be fair, was a very good business for you? It drove like a lot of cash flow. And it lasted a long time. It was a good -- it was actually -- it was a good thing. So like how does this one compare?

Yes. So we're very excited about this generation of game consoles. I think it's very likely the most anticipated set of products between Microsoft and Sony. People have their favorites, but we view it as just a great console upgrade cycle. There is a significant content. You've seen some of that from our customers talking about the amount of GPU horsepower that you see in these consoles. It's really amazing, the amount of technology you're putting in these consoles. We view it as a start of another long cycle. In terms of what we see, there is more content, so you would expect a bit higher ASP. But those ASPs will continue to see some incremental reductions over time. We expect the units to be similar to the last generation. I mean is it more or less? I mean who knows, but we expect them to be similar. I think the difference for us is we're really doing more to bring some of the ecosystem together, and I think that is just good for AMD graphics, right? The Radeon graphics brand, now you see in consoles, you see in PCs. We also have an engagement with Samsung on the mobile side, and that's all goodness. So consoles will be important for us as we ramp the second half of the year. I think, typically, you see the peak years actually be in year 3 or year 4. So you can see it continue to be a growth vector for us. From a margin standpoint, consoles are below our corporate average. It is something that we talk about to try to make sure investors understand. This is an area where the customer actually invests in the R&D. So they invest upfront, and that results in lower gross margins. But on the operating margin line, it's good cash generation for the company.

Yes. I'm not convinced the operating margins are necessarily below corporate average growth, sure, but...

Yes, the operating margins, depending on when you look, there was a time when they were above corporate average, but as the rest of the corporate average increases, I think they become closer to corporate average.

Yes, yes. Got it. I want to talk a little bit about sort of the financial transformation that's happened as well. You've given some long-term guidance. If I kind of work the math properly with something like $2.40, plus or minus, in, I think, 2023, but can you sort of discuss just the building blocks to get us there in terms of like growth by segment, margins like spending? Like how do we get from where we are today, which is incredibly better than where we were already? Like we're on good footing, but how do we get to where you see things going over the next few years?

Yes. It's an exciting sort of long-term model from my perspective. What we're trying to build is really a high-performance computing company at scale. And so the key tenets of the financial model are, let's call it, 20% plus CAGR in growth -- revenue growth over the next 4 years. If you actually do the math on that, from where we were in 2019, that's about double the size of the company, so that's building significant scale. That gives us more opportunity to invest, which high-performance computing is a large investment driver. So we're going to continue to invest in the business, but we will see some leverage on the model in terms of overall OpEx, let's call it, gradually, as a percentage of revenue coming down. And then we do believe that because the products are getting stronger, because our penetration in the markets will get stronger that we'll see growth in the gross margins. So we see gross margins on the order of 50% or greater. And when you put all that together, I think that gives you an exciting model for earnings power growth. But the overarching theme is growth in our target markets. I think we've chosen a set of markets that we are very excited about. It's an $80 billion TAM. Even at wherever we ended up in 2020, there's still a lot of growth opportunity for us. And we think that, even today, that TAM is showing to be even more important, right? It's very resilient to sort of the ups and downs of the macro. When you think about people need computing, people need more cloud capabilities, they need more work-from-home capabilities, people are buying more gaming systems. And so we've chosen some good markets. And now it's on us to execute both on the product and on the business front.

Got it. Lisa, we've got about 7 minutes left. I want to get to a few of the questions that we have online now from the people watching. Question on your supply situation. How is it trending? And do you have enough access to capacity to support upside, especially now that Huawei has potentially shut out of TSMC?

Yes. So we spend a lot of time on managing sort of the overall supply chain. I would say that the wafer supply continues to be tight, but we are getting what we need to achieve the growth expectations that we have. And we continue to have a very strong partnership with TSMC, as we go through sort of the puts and takes in the market.

Got it. There's a question to your dividend plans. I'm going to expand on this, actually. The question here says, do you have plans to pay a dividend in the next year? I would actually -- if I was going to phrase it, I say, just in general, you're obviously in a much better financial footing than you were. Like just what are your thoughts given that and where things are going around cash usage, capital structure and capital for whether dividends or buybacks or both at some point? How do we think about that?

Yes. So I mean, if you look at the overall balance sheet, I think that's one of the areas that we feel really proud of, the fact that the balance sheet is very strong, we're in a net cash position, we have paid down a lot of the long-term debt that we had. As we look forward, we do believe that free cash flow is an important metric to measure the company on. And we're going to continue to work up our free cash flow margin as an important metric. When you look at what we'll do, our priority is, first and foremost, to invest in the business. I think that's what you want us to do as a growth company. We will, as we build up those cash balances, think about what other things that we might do in terms of other shareholder return. But I really do think that we're sort of at the beginning of that wave, and so it's too early to talk about when you might do a dividend. But I think we should consider it as #1 priority, invest in the business, and then we will certainly look at what are good returns of capital. And that's something that we'll develop over the next couple of years.

Got it. That's helpful. A question here on the China situation. What are the potential risks with the ongoing escalation of the trade war and some of the other geopolitical issues with China?

Yes. It's certainly a topic for those of us in the semiconductor industry. From our standpoint, we've been sort of managing that since last year, with -- the tariff situation, we've been able to mitigate much of that with just diversity in our supply chain. We continue to watch the new regulations as they come through. We're certainly very committed to meet sort of the U.S. objectives and regulations around China. I think we've been able to manage through that. From our standpoint, we don't do a lot of commercial business in China. We do have some consumer business that's there, but we continue to watch that very carefully.

Okay. Here's an interesting one. How will TSMC building factories in the U.S. affect company decisions about chip fabrication?

Well, I think this is, again, all part of the broader question of U.S. manufacturing, which is on lots of people's minds right now. Well, I think TSMC investing in the U.S. is a good thing and, certainly, one of the things that we've said. Look, supply chain diversity, we know now is just really, really important. So we would certainly advocate for not just TSMC, but other suppliers of ours to consider investing in the U.S.

Got it. Question on your new win in NVIDIA's DGX displacing Intel. Just your thoughts on that. What drove that? Why did you do displace them? I've got my own views on that, but I'm curious to hear what you say.

Well, again, I think NVIDIA is a very, very strong engineering company, and they looked at what was available and decided what would be best for their system. Look, we're honored to be in the NVIDIA system. I think we are honored to win any high-profile system that is really pushing the envelope. So I think it's really a technical decision, more than anything else.

I mean was it just the PCIe 4.0 support? Or was it broader than that?

Again, I think NVIDIA has commented a bit on their selection, but I think the PCIe Gen 4 certainly helped. I think the number of cores and the performance capability certainly helped. I think one of the things you find just to feed lots of GPUs, you do need a pretty strong CPU there.

Okay. What are your thoughts on the competitive environment and potentials for price wars? And again, by the way, this is a question I get a lot when it comes to -- you are taking share, what is Intel going to do about it? So like how do you think about those risks? And how do you mitigate them?

Yes. So if you look at the various segments and the dynamics, there are actually some important things to recognize. First and foremost, price wars are not really company-friendly. So in general, I don't think people like that. We view it as it's a technology choice first. And certainly, in the server market, it's a technology choice first, okay? We were -- when we were uncompetitive, we -- you couldn't drop the price because the total cost of ownership wouldn't be made up on the other side. And what we see now is price is not the determining factor of a server buying decision. It's a factor, but it's not the determining factor. What we've seen in the marketplace is it's competitive. I mean there's no doubt about it that it's competitive, and we go head-to-head in certain opportunities. That being the case, though, we believe that our story is around technology, it's around the range of products that we have from sort of the very high end to more mainstream products, and I think we feel good about our ability to compete, while still growing our revenue and overall margins.

Got it. So Lisa, we have about 30 seconds left. I'm going to finish this the way I finish all of these. I'm going to give you your soapbox, 30-second pitch, why should investors buy and destock.

Look, we are very excited about what we can accomplish over the next 4 or 5 years. Our goal is to be a growth company, and our aspiration is to be one of the top growth companies in tech. And we think we're in the right markets. We have built a product execution machine that is very strong, and we've also built a set of customer relationships where we've generated trust in the customer base. So what that all translates into is revenue growth, margin expansion, profitability growth and, hopefully, a very exciting business over the next 4 or 5 years. So by the way, thank you, Stacy, for the time today. We appreciate spending some time with your audience.

As a reminder to the audience, again, we're doing live polling with Procensus. If you click on that link on the left side of your screen you'll see new window open up with the short poll. It will take you 60 seconds. You will benefit if you take it from real-time tracking of investor sentiment on AMD. With that, I think we'll close it here. Lisa, thank you so much for being with us here today. I've been thrilled to have you here. Thank you very much, and have a great day.

Great. Thanks, Stacy.

Take care. Bye-bye.

Bye-bye.