Intel Corporation (INTC) Earnings Call Transcript & Summary

Hello, everyone. This is Vivek Arya. I'm the Senior Semiconductor Analyst at Bank of America Securities. I'm absolutely delighted and honored to have Navin Shenoy, Executive Vice President and General Manager of the Data Platforms Group; and Trey Campbell, Head of Investor Relations at Intel, join us this morning to share their insights about Intel and the computing market overall. So welcome to both. Really glad to have you, really appreciate your taking the time. And what I'll do is I'll go through a list of questions. But to the audience, if you need to ask questions, please feel free to send them to me through that Veracast panel that you have in front of you. But with that, let's get right to it.

Navin, maybe just as background, you were recently promoted to lead Intel's Data Platform Group. Tell us about what that means for you, your responsibilities, but also importantly, what that means for Intel's strategy in the computing and the data center market.

Sure, sure. Great. And it's great to be with you all. I think maybe just some context. If you think about what's happening in the world that we are all kind of living through right now, you were seeing a once in a decade, maybe once in a multiple decade acceleration of analog to digital, if you will. And everybody, all companies, governments, society at large, is trying to figure out how to exploit the power of data, to use data to create value. And in that world, we see just a massive increase in demand for computing, for networking, for storage, for memory. And data centers, homes, cars, factories, all increasingly need computing power and the resulting need to move, process and store that data. And so we, Intel has been on this journey over the last several years to align our assets to capitalize on the shift from being a historically PC-oriented company to being a data-centric company. And while we still are on a journey here, this is going to be a multiple-year journey, you saw in the first quarter that our data-centric revenue surpassed our PC-centric revenue. And so we're making progress, and there's just a ton of opportunity for us. So in that context, we formed the Data Platforms Group about 1.5 years ago or so. And Vivek, I'm not sure it was a promotion, but it was a reflection to try to bring together all of our data-oriented businesses: our cloud business, our enterprise business, our networking business, our programmable solutions business, our FPGA product portfolio, our connectivity group, which has all things, Ethernet, SmartNIC and switch inside of it and our big bets in and around AI. So we wanted to bring together that end-to-end portfolio into one place, have a consistent methodology to bring products to market and to have a consistent experience with our customers.

Got it. Maybe starting with some near-term questions, Navin, and then we'll go to the longer-term drivers. So near term, the COVID pandemic has, of course, created a lot of, I think, positives and negatives for the semiconductor industry. What have you seen so far in terms of the impact on Intel? You mentioned the very strong results of the Data Center Group. How is the visibility as you look into your business over the next several quarters? And then just longer term, what do you think the experiences we are going through, how do they change the way you think about your business longer term?

Sure. In some ways, everything's changed. And in other ways, nothing's changed. There's been lots of disruptions around the globe, just supply chain-wise, with COVID. But it really hasn't changed our long-term strategy in one bit. If anything, it gives us more conviction that the need to move, process and store data is sort of moving from a nice-to-have to an essential. None of us could be doing the things we're doing right now without digital technology. So it increases our conviction. And as we've diversified our portfolio into this larger TAM, this expanded TAM, $300-billion TAM that we've been going after, it sort of insulates us a little bit from any shocks in any one business. But let me just maybe walk through a few of the dynamics. On the supply chain side, we certainly did start to see impacts of COVID right around Chinese New Year. As Chinese New Year got extended, many ODMs and other supply chain partners saw impacts. We were a little bit fortunate. Our factories remained operational through Chinese New Year. We were -- we had some supply shortages, so we kept our factories running. And so we didn't have a bunch of people leave for Chinese New Year and have to deal with trying to figure out how to get them back to work. So we were able to mitigate any non-Intel-related COVID-19 supply chain disruptions and also take advantage of the fact that our CPU factories and assembly test factories were running full tilt. And you saw -- and you've seen us talk about how we've achieved 90% on-time deliveries in the first quarter with modest impacts in some of our more adjacent businesses where we didn't maybe have full control of the supply chain. So I give our supply chain team just a ton of credit for doing an amazing job in business continuity planning and just continuing to operate despite all the challenges that we saw country-by-country around the world. On the demand side, we saw significant strength in the first quarter. Certainly, in the data center business across the board, whether that was microprocessors or FPGAs or networking or storage, we saw a very strong Q1. We [ were so strong ] in Q1 in the PC business so we saw the work-at-home, learn-at-home dynamic play out really globally. And we were [ to say it ] -- at our earnings in Q1 that we expect that dynamic and that strength to continue through the first-half. On the data center side, coming into the year, we were already calling for a strong Q1 and COVID just acted as tailwind, especially the cloud and networking infrastructure build-out. And we see that very strong trend in Q1 continuing in Q2 and I would say we'll probably see some of that strength going into Q3 as well. The Client business also called for a strong first half coming into the year and that really only accelerated as people were working from home and the kids were learning from home. And I'd expect that to sort of continue in the second quarter as well. I'd just say that the second half of the year is very difficult to predict. Historically, there's pockets of our business that have been relatively correlated to GDP but the GDP down trend that we see is different than the past, given that it's health-related and given that many people will still be working from home for some extended period of time. So it's difficult to sort of know how that's going to play out for us in the second half. I'd say we're probably mostly cautious about on-premise enterprise IT budgets in the second half and probably mostly cautious about the Client part of the business that is relatively indexed to GDP in the second half. But I think we might have some tailwinds as well with 5G, with cloud, build-outs continue to be relatively strong. Some of our other segments, our IoT segment, was weaker in Q1. The retail segment, the transportation segment, the industrial segment that our IoT business is indexed to was weak. I would expect that to still be a bit of a headwind in the second quarter and into the second half. Mobileye, our autonomous driving automobile-oriented business. Strong adoption and design wins and super bullish about Mobileye over the long term, but I think given the automotive sector's outlook, I think there's going to be some impacts there as well. So there's puts and takes. Net-net, nothing really changes in terms of our long-term view that the world is going to need more computing, networking, storage, memory over time.

Navin, just one second, I'm getting some feedback that the webcast is breaking out. Maybe, operator, could you just make sure that we are all -- everything is lined up properly? Operator?

I'm sorry, you have to hold for a moment.

I apologize for the interruption.

No problem. I can hear you just fine.

Yes, same here, but I got several messages saying that it's getting out. [Technical Difficulty]

You can continue now.

Okay. So let's try this one more time. Let me just do a quick check. Okay. Let me -- okay. I think we are all back up now. I apologize for that interruption. Maybe, Navin, because this was a very important question that several people had sent to me, maybe if you could just summarize. I think you made some kind of high-level comments about which specific end markets could -- you could continue to see the strength in maybe Q2 and beyond. So just the puts and takes that you described, if you wouldn't mind just going through it one more time because it's important, and I think people might have missed that. Thank you.

Sure. Yes. I was just saying on the demand -- this is COVID-19-related, so what's going on in the light of COVID-19. On the demand side, there's tailwinds and headwinds. Clearly, we saw strength in the first quarter on both PCs and in the data center. I expect that, that will continue in the second quarter in both businesses. Shelter-in-place dynamics are a tailwind for the cloud, for networking, for laptops, cloud applications that were just sort of nice to have a quarter ago, even online shopping, video collaboration. I mean those are just indispensable now. So as a result, we're going to see strength that we saw in Q1 continue in Q2 in those segments. And I was commenting that I can see how some of that strength might continue into the third quarter as well, especially in the cloud. For client, we called a strong first half prior to COVID, largely driven by the refresh to Win 10. And I think the work-at-home and learn-at-home dynamic could continue in the second quarter. All that said, I think we'll be relatively cautious on the second half. Our visibility in the second half isn't that strong at this point, and global GDP is expected to contract. So I would expect that discretionary spending, enterprise IT budgets are going to be indexed to that to some extent, particularly in the client business. But on the data center side, I think the cloud part of the business, again, relatively strong in Q2. Some of that could go into Q3. What happens after that is difficult to predict. IoT or Internet of Things or Intelligent Edge business, super bullish about that business over the long term. It's been a very strong growth driver for us for many quarters now, but we saw a relatively weaker Q1 there given the retail, transportation, industrial segment, indexing that we have in IoT. I expect some headwinds to continue there in the second quarter and beyond. Mobileye, very bullish about Mobileye, and they're having tremendous success in driving adoption. Design wins, making great progress in autonomous driving. So over the long term, we're very bullish about Mobileye. But over the near term, I expect the decline in the overall automotive sector to impact Mobileye as well. That's a high-level summary, Vivek.

Got it. Very good. So just one quick follow-up to that. There's been a lot made of this notion that there has been some kind of bubble in cloud spending in the first half given all the dynamics that everyone has heard of. Just in terms of the broader demand and just the cloud environment, do you think there is just the modest kind of puts and takes of just the lumpiness of CapEx? Or there is some other deceleration that we should be worried about? Because it seems like this demand is more secular rather than the cyclical aspect that I think people worry about.

Yes. I mean, look, even in -- the business has been lumpy historically. But the secular trend is very clear. So despite the sort of bumps along the way, the trend is up and to the right over many, many years. So will the cyclical trend smooth out or not? It's hard to say. I think many of the cloud service providers learned from what happened. If you think about what happened in 2018 heading into the first half of 2019, there was a big build throughout 2018. It started to slow down a little bit in the fourth quarter. And then the first half of '19, there was a pretty significant correction. And some -- as I talk to customers, some of those cloud service providers learned from that and felt like they pulled back more than they needed to, more than they should have given the long-term trends on the business. So the cost of having a little bit of extra compute infrastructure for the cloud service providers is completely worth it when you have situations like we have now which is a significant acceleration to digital infrastructure. And it's kind of across the board, right? It's not just the work-at-home dynamic driving demand. It's just digital transformation happening at companies around the world. It's consumers using digital infrastructure more and more and more for all aspects of their life, whether it's e-commerce or connecting with family members or gaming or whatever the case may be. So I think I have deep conviction that the cloud build-out that we've seen is a secular trend. And that while there may be some periods of time where there is digestion and periods of time where there's build-out, I don't expect that to change. Over time, I do think there may be a dampening effect on the magnitude of the ups and the downs.

Got it. Now let's get into the different parts of the data center business, starting with server CPUs. Extremely strong results over the last 3 quarters. If I recall correctly, the business has, I think, beaten at least the consensus expectations of almost 10% every quarter for the last 3 quarters running, and that's been with your 14 nm products. And here, you are on the verge of 10 nm products. So talk to us about what is the state of competition as you assess it? And then what benefit will 10 nm provide to you? And just a general update on where you are with the 10 nm server products.

Sure. Look, I think we've been saying for a while that we expect it to be an increasingly competitive environment. I've probably been saying that for 25 years that I've been at the company. And our view really hasn't changed on competition in any fundamental way over the last quarter or 2. And as you pointed out, Vivek, in a more competitive environment, we're continuing to deliver strong growth. You saw -- overall, for our data-centric businesses, not just data center, but overall, we saw 34% year-on-year growth in Q1. And despite the really, really strong demand and sometimes unforecasted demand, we've been meeting customer demand, which gives customers confidence. We feel good about our roadmap. We feel good about continuing to grow our share in the expanded market opportunity that we're pursuing. This is not just about microprocessors, but our ability to bring together a full portfolio across the data center at scale is super important across all segments of the way computing is evolving. So we have the cloud business. We have the build-out of computing at the edge. We have the networking business that's being accelerated by 5G. We have many new products that we're bringing out in the FPGA portfolio, in Optane in Ethernet, SmartNICs, silicon photonics. So across the board, we have a broad portfolio that we're bringing together. And as a result, we have confidence that as the world needs more computing network storage memory, we can capitalize on that. A few just data points. Cascade Lake, our 14 nm Xeon that you referred to, fastest-ramping Xeon in our history. We recently announced some enhanced SKUs there that deliver really outstanding performance on real-world workloads. We made a decision many years ago to embed AI acceleration into Xeon. We call it DL Boost or Deep Learning Boost. So it's the only data center microprocessor that does that. And that, we think, is giving us some advantage and uplift as more and more workloads move to embed AI. We have the ability to bring Optane persistent memory and combine it at the platform level with Xeon. That's helpful for a variety of different types of workloads. And so you're going to see us bring out our third-generation Xeon scalable processor very soon in Cooper Lake. That's another 14 nm Xeon microprocessor. We'll take sort of the AI features to the next level there with that product. And we remain on track. To your 10 nm question, we're on track to begin production of our Ice Lake server on 10 nm at the end of this year with a variety of new features, better performance, more cores, more networking capability built in. And then in 2021, we'll bring our next-generation 10 nm Xeon to market called Sapphire Rapids. And so we're on a cadence of bringing out more and more capability for our customers over the next 18 months.

Got it. Recently, we saw a pull-in of some 10 nm into Q2. That was reflected in the gross margins, as an example. What is the state of 10 nm? Is it the priority? Is client first and then server second? Or do you think this pull-in of 10 -- some 10 nm into Q2, at least from a cost perspective, that has implications on when you were planning to launch the server side as well?

Yes. 10 nm is making very good progress. And as a result of that, you're seeing us sort of just put the pedal to the metal, if I can use that terminology in terms of ramping 10. I would just say, first and foremost, our philosophy is to make sure that we're delivering leadership products, period. And at the end of the day, that's what customers care about. And having leading-edge process technologies is one element to make sure you have leadership products. But of course, we're bringing together many more ingredients than just the process node, architectural capabilities, packaging, novel memories, interconnect schemes. All those things on top of which we bring our software ecosystem to bear is the way in which we deliver product leadership. On 10, we have 2 fabs in volume production, and we have the third. The third one in Arizona, that's now ramping. We've been shipping Ice Lake to clients for several quarters now. We have our first 10 nm Agilex FPGAs that we're shipping, sampling to customers. We are ramping our first 5G base station SoC. That's on 10 nm, codenamed Snow Ridge. And we've got a number of additional products coming out in 2020. We have a variety of ASICs that we build for the 5G market. We have Tiger Lake for notebooks coming out in the middle part of this year. We have, and as I mentioned earlier, our first shipments on the unscalable Ice Lake at the end of this year. So I'd just say, in general, 10 nm is looking very good. We're ramping it very hard. The decisions about which products we introduce first are sort of a function of our yield profile, market windows, customer ability to absorb the overall roadmap. So we make those decisions fairly strategically based on a variety of different factors. And for, obviously, the 10 nm node, we went with client first in the second half of last year, and then we've been bringing out a variety of products across all segments, from data center to networking, from CPUs to FPGAs to ASICs as we go through 2020.

Got it. I wanted to touch on Intel's AI strategy. When we look at one of your competitors, so they have been investing in just the GPU market for a long time, right, very long experience with the software side, the developer side, range of products. Another competitor has both the CPU and the GPU and the interconnect aspects. When I'm asked about Intel's AI products, you -- the advantage you have is just a variety of products, right? You mentioned all the new features in Xeon. You have the programmable chips. You have the Habana Labs, and you have other accelerator options. But is that a positive or a negative? Like how would you contrast your strategy, right, versus your competitors? And just what has been your success so far in the AI market? What is the right way to measure and track that success?

Sure. Good question. And I think it's interesting. We're at the beginning, still in the early innings of this trend towards this workload that we call AI. And we still sort of think about it as a market. Over time, we won't really do that because AI, it's kind of like saying, well, what's the Internet market? Everything is the Internet market because the Internet defines everything we've done for the last 20 years. The same will be true of AI. AI is going to be embedded in everything that we do because workloads of all varieties and all types will leverage artificial intelligence in some way. And why is that? Because it's extraordinarily efficient and extraordinarily beneficial in terms of delivering ROI on massive amounts of data, which is the -- what we've seen over the last several years. So that being said, if you were to use sort of conventional definitions of what does an AI market look like, our view is that the market is roughly $25 billion silicon market by 2024. We estimate that we had about $3.8 billion of AI sort of driven revenue in 2019. And yes, to your point, we believe in offering choice to customers on AI. In the data center, we've embedded AI into Xeon, first and foremost. We're building the dedicated accelerators, the first of which will come to market in 2020. So we've been at that for several years now, accelerated with the acquisition of Habana Labs. You'll see the first products really come into the market in 2020 on that domain. And at the edge, we think we have really unparalleled AI-oriented solutions. We've got -- in ADAS, we have Mobileye, and they're doing a tremendous job. They just recently posted, if you haven't seen it, I'd encourage investors to go check it out, an unedited video with a drone over a Mobileye autonomous vehicle driving through the streets of Jerusalem, doing things that I don't think any other autonomous car has been able to demonstrate. We have edge AI solutions through our Movidius portfolio in the IoT segment for video and edge inferencing that are super exciting. We've built out a very strong developer attraction and momentum around OpenVINO, which is our video inferencing AI software stack. And on Habana, we've rationalized our dedicated AI accelerator solution for the data center market, and we're betting on Habana. And so the key to having that breadth and portfolio is to make sure that we can develop and build a software ecosystem that can leverage that capability. And so what you've seen us do is invest in software solutions to somewhat abstract away the underlying hardware and give developers a consistent programming environment. We call that oneAPI. It's an ambitious undertaking. We're in beta now. We have a bunch of milestones that we're going to cross over the course of 2020. But it's very promising. It gives developers an ability to sort of not really have to worry about the underlying hardware. And then it gives data center operators and enterprises and networking companies the ability to choose the right silicon for the job at hand. They're power constrained. They can choose lower-power silicon. Maybe they choose a CPU as an example, a client-oriented CPU. If they have high change in their algorithmic development, like you see in some of the cloud service providers, they might choose FPGAs because they can continually tune and change hardware as their algorithmics change. If they have hardened algorithms and they know what they need to do, they might choose an ASIC. So there's a variety of different solutions that we think will evolve. And of course, I think the key to the AI market evolution is adaptation. The winners and losers will be defined by the companies that adapt the best because this field is fast changing.

Got it. Maybe let's touch on the 5G market. What is Intel's strategy in 5G? Because it sometimes seems like -- I think your communications business sometimes is underappreciated as the focus has really been on the cloud aspect, but the communication business has been growing at a very nice pace. And here we are at the start of the 5G cycle. And I think you previously alluded to some of the ASICs and other 10 nm-based products that you're making for base stations. What is Intel's overall 5G strategy? And as part of that, how important is the edge computing market?

Sure. Look, I think 5G is one of those very interesting transitions. It's as much about computing as it is about communications. And the intersection of computing and communications is where Intel can play a very significant role. And it's a great example of where we've invested to win and where we've tried to get ahead of transitions and where I think -- we've been at this now for a decade. And the investments we've made, you're really going to start to see pay off here in 2020 and 2021. Our -- it's about a $5-billion business, okay? You don't see it in the way we report, but it's about a $5-billion business if you cut across cloud, enterprise and communication service providers. The networking business as a horizontal is a roughly $5-billion business. And that business to date has largely been driven by the virtualization of core networks. So an AT&T, a Verizon, a Vodafone, large communication service provider companies around the world, they've historically built their network infrastructure on proprietary infrastructure or vertical solutions. And what's happened over the last 3, 4, 5 years is they've recognized that the cloud economics, the ability to horizontalize, to use standard high-volume servers, to program, to make the network more programmable is the way they get agility. And instead of buying a new piece of equipment every time they want to add a workload, they just write software, right? So the economics lead to the virtualization, lead to choosing an Intel architecture-based server. And so that has been the predominant -- and by the way, I would say, I would estimate that maybe 15% to 20% of the core networks around the world are virtualized today. So there is a long way to go, and there's a lot of track ahead for us to continue to grow in just the virtualization of the core network. On top of that, you're seeing us increasingly move towards the edge. And the move towards the edge is a combination of bringing an Intel architecture SoC to the base station market for the first time. That's Snow Ridge. We introduced that product a few months ago, and it's now starting to ramp in networks around the world associated with the 5G build-out. On top of which, the acquisition of Axxia that we did several years ago gave us a portfolio of ASICs for a variety of the functions -- signal processing functions that happened in the network. And so that product line is now ramping on 10 nm for 5G. And we have our FPGA portfolio and structured ASIC portfolio also being used and leveraged in 5G. So this is an opportunity where we cannot only grow because the market is growing, but we can also gain significant share as we deliver products, which address the core of the network, and that part of the business shifts to virtualization in servers. And we can increasingly address the RAN, the radio access network, which, historically, we haven't really addressed with Intel architecture. So I think demand here is going to be strong through the second half with 5G as the big driver. And I also would just reiterate that we've stated previously that we expect to be the leading base station silicon supplier. Initially, we had predicted by 2022. We changed that and pulled that prediction into 2021, a year earlier than we previously announced because of the momentum we're seeing with our large customers in 5G. So I agree with you. I think it's a super attractive opportunity and one that's not fully appreciated yet.

Excellent. I think with that, we are at the end of our time. I really want to thank you, Navin and Trey, for participating in our conference and sharing your insights, and I want to thank the audience for tuning in. If you have any follow-up questions, please feel free to call me or e-mail me. But let's close the session here, and thank you so much again, Navin and Trey.

Thank you.

Thanks, Vivek.

Appreciate it. Take care.