Advanced Micro Devices, Inc. (AMD) CEO Lisa Su Presents at 22nd Annual Credit Suisse Technology, Media & Telecom Conference (Transcript)

|
About: Advanced Micro Devices, Inc. (AMD)
by: SA Transcripts
Subscribers Only
Earning Call Audio

Advanced Micro Devices, Inc. (NASDAQ:AMD) 22nd Annual Credit Suisse Technology, Media & Telecom Conference November 27, 2018 12:00 PM ET

Executives

Lisa Su - President and Chief Executive Officer

Analysts

John Pitzer - Credit Suisse

John Pitzer

My name is John Pitzer. I'm the semiconductor analyst here at Credit Suisse. It's my pleasure to introduce Dr. Lisa Su to my immediate right, who is the President and CEO of AMD. We've got about 20-25 minutes to host a fireside chat in this environment. We also have a mic roaming the room. So if you have a question -- if any questions please raise your hand, we will try to get that to you. With that, Lisa, first, let me thank you for coming to the conference.

And I guess the first question I have sort of been asking, I think it's particularly germane with AMD is -- talking a little bit about the longer term strategy of the company and more importantly, you have gone through sort of a three or four year repositioning of the company to really -- take this company back into sort of where it was 15-20 years ago as far as competitive products and potential to gain shares. Maybe you can talk a little bit about the strategy and kind of where you are today?

Lisa Su

Absolutely. So John thanks for having us. We are very happy to be here. And I would say, look we are excited about where AMD is today. My four year anniversary as CEO was in October, and it's been a long four years. We've gotten a lot done, but the whole purpose and our strategy is to focus on what we are really good at. And we are a high performance computing company with high performance CPUs and high performance GPUs, it's all about the products. It's all about our product competitiveness. We have made some significant progress on the technology side and that comes with our CPU architectures and our GPU architectures.

And I think we have also made significant progress on the business side. In 2017, we grew the company mid-20s percent, in 2018, as we've guided there would be another mid-20s percent. So I think we are making a strong progress on top-line as well as bottom-line, but we really are still at the very beginning of what we believe is a multi-year product cycle because of some of the decisions we have made in the technology and the architectures.

John Pitzer

Maybe we can talk a little bit about EPYC a little bit more detail. I think you have done a very good job over the last several years of setting reasonable expectations, and then, beating them. I think the one that the investment community has been most focused on for this year is kind of that market share target number for EPYC exiting this year. It seems like you are on track for that. Maybe you can update that? But, I guess more importantly as you move into next year with Rome and EPYC 2 and kind of the process leadership that you should have on the manufacturing side. How do we think about the next five to 10 percentage points of market share after you achieve kind of the target that you put out there already?

Lisa Su

Yes. So John servers are certainly the most strategic market for us. And when we look at data center overall both CPU, GPU, I mean it's just a very large market, significant profit pool and we are very underrepresented in share. We always believed when we started this journey that our server story was a three generation story. So as we know about the enterprise market they don't move that fast, but once they move there is tremendous capability.

So our first generation EPYC I think has done well. There has been a lot of work done with the cloud vendors as well as some of the OEMs and the enterprise accounts to really optimize AMD’s architecture. We set a goal of mid single-digit share by the end of 2018 and we're on track for that goal. So we feel good about that. But again that's not a destination, that's a point in time. We actually believe with the competitiveness of our Rome product in 7-nanometer that the next 5%, post the first 5% will actually come faster. And that's just a result of, number one, people are much more familiar with our architecture. So that's really important. And number two, I think we're -- our product is positioned very well both from -- and not just from a process technology side. We are very, very pleased with our positioning in 7-nanometer. And I think that was a very good decision for us to align with TSMC there, but on our architecture also I think we've made some really good decisions in terms of breaking up the product into these chiplets that really allow us to optimize the CPUs and the IO separately. So we're very excited about our product positioning. But as I said, it's a three generation race and we feel really good about the first generation we're about to launch second generation here in 2019 and there's a ton of work that has already been done on the third generation.

John Pitzer

Look Lisa you recently announced AWS as offering EPYC as a cloud instance. Congratulations. It sounds like that was really the anchor to get to the mid single-digit kind of target that you had for market share exiting sort of Q4 of 18. The question I get sort of often is just given how lumpy hyperscalers can be, is the move from five to 10 or 5 to higher going to be somewhat linear or could you actually see some bouncing around and how do you think about to the extent that you've exploited one hyperscaler, is the next thing for EPYC really enterprise or is that another hyperscale. And as you move into Rome is that the enterprise product or how do we actually think about that progression?

Lisa Su

Yes. So when we look at it, first of all, we're very, very pleased with the partnership with the AWS. It is certainly a major milestone for us not just that they're deploying AMD, but they're deploying AMD in their most popular instances. And that's really important for us to get into the volume portion of the market and it is a key piece. Now we also have a number of other cloud vendors who have announced with us including Microsoft, Baidu and Tencent. The way we look at it sort of the path from 5% to 10% market share does include cloud deployments and I would say broader cloud deployments.

So again the way some of these customers work is, they try this out and let's call it, the first set of instances, gotten comfortable with the technology and we expect that that will broaden as we go into the second generation with Rome. There will be a little bit of bouncing around as always in a market like this, but I think we feel very, very good about the positioning of Rome for both hyperscale customers and let's call it a broader set of workloads as well as enterprise customers who again take a little bit longer to make a decision. But once they do it's very good business. And we've recently actually done very well in some HPC sort of deployments which we think are key for our competitive positioning.

John Pitzer

And then Lisa, just relative to Rome maybe you can help level set the audience. Remind us again your sampling now, when do you expect to see those samples start to turn in to recognizable revenue, how do we think about that growth curve?

Lisa Su

Yes. So we are sampling to a lot of customers right now. There's a lot of work going on to validate both on our customer side as well as on our side. In terms of specific launch date, it will be in 2019, but we haven't been more specific than that. But what I will say is that you should expect revenue to be strongly second half weighted. And with that notion in place there is a good balance between our first generation and second generation revenue in 2019.

John Pitzer

And then, as we look at your server and HPC opportunity a lot of the emphasis has been put on the EPYC product line for obvious reasons. But you've also done some good things on the core graphics side of the business. Can you talk a little bit about graphics specifically into the hyperscale data center server market?

Lisa Su

Yes. I think the graphics data center opportunity is equally exciting. And if we look at that that's a market phenomenon. So the market is growing tremendously. There are lots of new applications coming on board. And then, I think also our technology is getting much more competitive. So we recently just launched our 7-nanometer GPU called MI60 that will start shipping in volume here this quarter. And when we look at the applications we have hardware based virtualization, so anything that can use virtualization things like cloud gaming, things like desktop as a service, things like HPC, things like machine learning, those are all good applications for our product. And we view it as, we're growing from let's call it a small base, but it's a growth driver for us over the next two to three years.

John Pitzer

You mentioned earlier kind of how leverageable your design architecture has been with the chiplets. I've asked you this question multiple times, to me if you look at the progression of the semiconductor industry it used to be that CapEx was the single largest burden that a chip company had. You sort of solved that problem when you went fabless. But you kind of solved at the same year that R&D overtook CapEx as sort of the biggest single scale burden for the chip industry. And one of the questions that I have, I also get asked a lot is, just relative to your R&D budget and you compare that to your two largest competitors how can you compete as effectively as you have? And are there market opportunities that you're missing out on, if you had more R&D dollars?

Lisa Su

Yes. The way I look at it is, we are a company with let's call it a $75 billion TAM and we're $6.5 [billion] [ph] or so. So look we have a lot of opportunity to grow and as we grow we will grow R&D as well. Now I think what we've been able to do over the past few years is make really good decisions. I think we've made really good decisions about where to differentiate, how to differentiate, where you don't need to differentiate. And frankly part of that has resulted in significant innovation. I think our chiplet strategy will end up being the right strategy even for companies that are much larger than us. When you really look at how do you put together leading edge processors. This is sort of the best of both worlds.

So I think there's a lot of innovation that goes into it. And clearly as our revenue footprint grows, we will also grow our R&D footprint.

John Pitzer

I didn't want to -- I wished I'd never had to say this word again but cryptos back into the forebrain of a lot of investors right now with a recent report from a competitor where they had the lower expectations around GPU inventory. Maybe you can spend a few minutes talking about your exposure to crypto and kind of how you see the GPU inventory in the marketplace today and whether or not you think you've accurately sort of adjusted for that in kind of December quarter guide.

Lisa Su

Yes. So I think we talked about the impact of crypto on our Graphics business in our third quarter earnings release. And certainly there was an impact in the third quarter. I think if you look at the underlying nature there was a set of over shipments in the first half of the year. So there was excess inventory that came in the first half of the year that we need to work off here in the second half of the year.

In terms of our view of the market, it's pretty much what we expected. So it's pretty much what we discussed on our third quarter earnings call. We think it will take a couple of quarters to clear. So beyond the fourth quarter I would expect a similarly depressed graphics channel selling in the first quarter of 2019. But this is not something that is intrinsic to the Graphics business. Okay. I still view gaming as a very, very good market. Gamers are continuing to buy GPUs. We just launched our RX 590 mainstream graphics card and it actually did quite well.

We looked at Black Friday sales also. We actually did quite well. I think what we're seeing is gamers are being very discerning about what they buy. And so I think we feel good about how we're positioning our products. And clearly as we work through this channel inventory, we believe that the gaming cycle will continue to be a good growth driver longer term.

John Pitzer

One question I get asked, it's our opinion that we're probably too early on the content side for ray tracing to be really important in the near term, but with NVIDIA’s new architecture around Turing, how do you think about your ability to compete with ray tracing? I think you guys do it vis-à-vis software emulation. Is that an effective solution? Is there plans to try to bring out a more silicon based solution, or how do you think about the share dynamics as ray tracing becomes a bigger part of the content side of the gaming?

Lisa Su

Yes. And if I take a step back and just talk about the overall GPU market I think we believe we will be very competitive overall and that includes the high-end of the GPU market. Obviously, there are some new products out there from our competition. We will have our set of new products as well and we'll be right there in the mix. As it relates to ray tracing in particular, I think it's an important technology. But as with all important technologies it takes time to really have the ecosystem adopt. And we are working very closely with the ecosystem on both hardware and software solutions and expect that ray tracing will be an important element especially as it gets more into the mainstream frankly of the market.

John Pitzer

Switching gears maybe a little bit to Ryzen, I think you are now in the marketplace about 17 months or so with the Ryzen product. One of the criticisms that I've had, I think we've had is maybe the share gains haven't come as quickly. And I think that the right rebuff that you've always given is, you've got to wait for the product portfolio to kind of build out and for us to get into all the different segments of the PC market.

Can you give us a little bit of an update as where you are in that process and especially given that the only other competitor in that space is having some supply constraint issues? How do you think about your ability to gain share in an environment where there are supply constraint?

Lisa Su

Yes I have to say I'm really, really pleased with how Ryzen is doing in the market. There was some of that criticism the first couple of quarters John and I did say hey look guys it takes time. And if you look at how our portfolio has rolled out, it was desktop first, desktop DIY market and desktop OEM and now the notebook share has really started to show an inflection. What you're seeing is the platforms are really coming to market in broader scope. And my belief is, as you go through 2019, you're going to continue to see that momentum around Ryzen put aside any supply constraints and those kinds of things because those are temporal. I mean what's important is, the product is very strong. I think the OEMs are also really understanding the value of AMD in their portfolio and so we're going to see more platforms in 2019 versus 2018 and we continue to believe that the Ryzen momentum will continue as we go into 2019.

John Pitzer

The gaming console business has been a really important part of your growth story and free cash flow story. For me it's probably one of the harder parts of your business to try to model especially because we're getting a little bit long in the tooth of the current generation of consoles. How should we think about that market in 2019? And in general how should we think about seasonality of your business?

Lisa Su

Yes. That's good because I do agree that consoles have been a very important piece of our business. It's a large piece of our business, but it is also hard to model. So the way I would look at it is, we will be in the seventh year of the cycle as we go through 2019. So consoles will be down for us and they should be down sort of double digits and that's the way to think about it from a unit and revenue standpoint. However, when I take a step back and look at the semi-custom model for us not commenting on any specific customer, we see beyond 2019 that the semi-custom business returns to growth and actually significant growth for us. And I view it as a good business. It's a good business. It's a sticky business. It's really nice to have a design win that lasts seven years and I think it's been good for us. But from a modeling standpoint and a seasonality standpoint, we certainly see that 2019 will be down. And in general the first half of the year is lower than the second half of the year. And so we'll have to deal with that seasonality.

John Pitzer

Any insights you can give us on new platforms and timing?

Lisa Su

We feel very good about our semi customers.

John Pitzer

Appreciate it. Lisa your story is much more about the ability to gain share in already established markets and it's much less about what the market growth is in those markets. That said every three or four years this conference becomes dominated by the cycle question and the health of the industry and I think clearly we're in the 10th year or next year we will mark the 10th year of expansion post the financial crisis for the economy. And there's a lot of crosscurrents going on right now.

I'd love to get kind of your perspective on just end demand in general and I guess as you answer the question, do you think you can hit your growth targets even in a weaker overall market because you make up for it with market share.

Lisa Su

Yes, John. I do believe that we are somewhat unique from the standpoint that as I said earlier we have a very large TAM and we're underrepresented in market share. And so regardless of where we are in the cycle, our growth stories around market share gains and that's going to continue to be the main driver for a number of years. As it relates to the cycle and as you talk to my peers, I think you'll see that there are some pockets of caution that are out there in certain end markets.

I will say though that the PC market has fared pretty well. I mean it's probably fared a little bit better than people thought. And then overall data center, the expectation is, we will continue to be a growth market regardless if there's little pockets of air that are there. So I remain very positive on the end markets overall. And our focus is market share, market share, market share.

John Pitzer

Can you talk a little bit about your efforts in AI? I mean it's clearly a new workload still relatively small in the scheme of things, but the fastest growth area and while you have a massive opportunity even before you touch to sort of the ai workload pie. What are you doing behind the scenes around AI machine learning inference acceleration?

Lisa Su

Yes. I think AI is extremely important, no question a huge growth opportunity. I think it's applicable to both our CPUs and GPUs. Our main focus in AI is not actually around hardware because we think our hardware is actually fairly well suited, but it is around software and trying to enhance the software ecosystem particularly around an open ecosystem that will bring more developers to the party.

So I think, again, when I look at GPUs for us in the data center, there are a number of areas that we believe we can grow including AI and it's a multi-year effort as we look at it from both the hardware roadmap and a software roadmap standpoint, it's probably the single largest area that we're investing.

John Pitzer

And then, one of the questions that I get quite often is okay, even if AMD hit their share targets, is there a risk that competition comes in and lowers pricing so that the gross margin isn't what they think it's going to be or is there a risk that as you gain share you reinvest at a higher rate in the business. How do you think about those two dynamics?

Lisa Su

Well, I think we've been very open about what our philosophy is in this area. So for reinvesting in the business, we will certainly invest in the business, but we will invest slightly slower than our revenue growth. So that we do have leverage from a profit standpoint.

I think on the margin question, I think we've made very nice progress in margins over the last couple of years. And I would say that we've been conservative in our expectations of what the competitive response may be. And no question look we're not going to underestimate the competition, the competition is very strong both technically and otherwise. We believe though that products will make a difference and particularly in the data center market. Data center market is probably not the most price sensitive market out there. I think performance is extraordinarily important and that's what our focus has been on for our Rome in our next generation of products.

John Pitzer

You've had a very strong ecosystem partners especially on the manufacturing side both global foundries and now TSMC. But can you talk a little bit about what the renegotiation of the WSA has done fundamentally for your business, for your roadmap for customer confidence?

Lisa Su

Yes. I think you really have to think about this again as -- a series of strategic moves. I think we renegotiate the WSA to allow some flexibility in moving products to TSMC and that was extraordinarily important for our server business just given the consistency that you require in the server business.

Most recently given some of the strategic decisions that have been made by Global Foundries, we have aligned more of our product portfolio to TSMC. I think that's a positive. I think TSMC is a very, very strong execution engine. Global Foundries will continue to be a strong partner for us in 14-nanometer. And we're working through a set of business discussions that are favorable to both sides, so that we can continue to be a long-term partner. And as well as utilize the best in the industry with 7-nanometer.

John Pitzer

Are there questions in the audience for Lisa. Large room and no questions very quiet people.

Lisa Su

They are absorbing your wonderful questions John.

John Pitzer

I'm sure that's not the issue. I guess still going back to sort of the data center market. How do you think about kind of the different architectures that are coming into play whether that be -- you were very early talking about heterogeneous compute, you've got sort of the full suite of IP on the CPU side whether that BX 86 or ARM. But as you think about acceleration being done by FPGA and quite frankly what's now become a vibrant private market in semi's again around ASIC companies to do AI or AI specific companies and/or your hyperscale customers doing their own silicon. How did this all play out in your end?

Lisa Su

Yes. I would say still our belief is heterogeneous architecture is the way to go in the data center. And that is a combination of CPUs, GPUs, some FPGA, some ASICs. I don't view that, for example, if there are CPUs, that means that GPUs are not important. I think GPUs will continue to be important. I think there are certain functions that are done better in fixed function and there are certain functions that are done better in processing and it's a huge, huge market. So I just view it as a huge market for us. And we will continue, I think our ability to optimize between CPU and GPU is a significant positive that has yet to really play out as we're continuing to really gain share in the data center market. But I think over the next few years that will become a more important part of our story.

John Pitzer

Architecture question I have for you, if you look at the architecture of EPYC, clearly IO access with external memory was something that you guys did a lot of work on understanding the importance of memory and memory side going forward. Your large competitor in that space has decided to get back into the memory space with the view that it's so important that it can help lock in kind of customers at the CPU level with things like Optane and crosspoint. How do you see that market evolving and if crosspoint or persistent memory becomes bigger, what's the strategy to compete there.

Lisa Su

We are not going to get into the memory market. I would say I do not believe that customers will enable themselves to get locked in that fashion. I think we have great relationships with all of the memory vendors and we will continue to have great relationships with them. I do think this ecosystem play beyond just memory. But how do you connect things at very high-speed with very low latency is a very important part of the system equation. And so we think about that a lot. That's our infinity fabric. That's how we architect the system. And we are very focused on the memory architectures as you said, but we're going to do it in partnership.

John Pitzer

Perfect with that we've come to the end of the session here. But I want to thank Lisa for joining us and everyone in the room for joining us. Appreciate the time.

Lisa Su

Very good. Thank you, John.

Question-and-Answer Session

Q - John Pitzer

Lisa Su