Intel Corporation (INTC) Earnings Call Transcript & Summary

Good morning and good afternoon, everybody, and thank you for participating to this session. I am Pierre Ferragu. I manage the technology research team at New Street. I'm very, very happy and I feel very privileged to have with me today Dan Rodriguez, who is the Corporate Vice President at Intel in charge of the Network Platform Group. So we have about 40 minutes together. Dan will give us a quick presentation about Intel's role in the networking space and in the uprising of 5G. And then after that, I'll have a conversation with Dan. I have, of course, a lot of questions for him. And if you want to contribute to that conversation, please use your chat box and -- I mean your -- the platform to type in your questions. We'll do our best to handle them. And most importantly, we'll make sure we follow up individually with you guys if we don't find the time to address all the questions. Before I do that, I'll have to read a disclaimer for you. So hold your breath, here it comes. Today's presentations contain forward-looking statements. All statements made that are not historical facts are subject to a number of risks and uncertainties, and actual results may differ materially. Please refer to each of the participating companies' most recent respective earnings release, 10-Qs and 10-Ks for more information on the risk factors that could cause actual results to differ from those expressed or implied in the forward-looking statements in each companies' presentation. And with that, Dan, I'll hand it over to you for your introductory remarks.

Yes. Thank you so much, Pierre. It's truly great to be here at this New Street event and having the opportunity to discuss the build-out of 5G and edge. So with that, maybe let's share the presentation. All right. So again, I think I'll have a brief presentation, I believe, come in around 10 minutes, then we'll have plenty of time for the fireside chat. Next slide. So there are several really key technology inflection points that are really driving and changing the landscape of technology and business as well. And they are 5G network transformation, AI and the rise of the intelligent edge. Now 5G is really the first time in our history that we're seeing the convergence of both compute and communication technologies. And this will enable all sorts of rich new experiences, whether it's experiences delivered to a consumer or to many different businesses. We'll also see the use of AI and analytics becoming really pervasive, not only in the cloud, but also to the data center through the network edge and on various on-customer premise types of locations. And all of this, along with all sorts of new use cases that are being imagined, are really fueling the emergence of the edge. And with the edge, you're going to see more computes, more intelligent computes placed much closer to the user to deliver the right return for a specific business use case. And all of that, when you think about those 3 major inflection points, when you think about the build-out of both 5G and edge, it needs to be built with the same sort of benefits that are found in the cloud, meaning it needs to be built that's built in a way that's cloud native, deliver to the right automation across the network, the right scale and the right overall TCO and ability to deliver new services very, very quickly. Next slide. Now Intel, we've been on this journey really with the communication industry for quite some time, really for the last decade. And the goal of this journey that we call network transformation is really to move the market to more of a software-defined, agile and scalable infrastructure. Now to do this, we have a 3-pronged strategy. And the first part of the strategy is to really enable our customers to converge all communications or network workloads onto a general-purpose server or sometimes is referred to as a commercial off-the-shelf server or COTS. And to do so, we've innovated. We've innovated by enhancing our CPUs and making sure they're network optimized, adding different network accelerators. And in that journey, we've introduced processors such as our Atom-class processors for power and price and the right level of performance all the way up through our Xeon scalable processors to the right, performance and flexibility, alongside a host of platform-level ingredients, whether it's an FPGA, whether it's a finely tuned Ethernet or whether it's a specific network accelerator. And this is critical, having that breadth of road-map products because when you think about networking and you think about edge, there's all sorts of different locations to be able to run in many different diverse environments. But we all know, when you think about a transformation that we're talking about, it takes a lot more than just building great silicon products. It takes a company like Intel truly investing in software and in partnership with all sorts of open-sourced projects, and that brings us to the second part of our strategy. So we've invested in significant software development kits. And the reason for this, it really enables our customers to get the most out of our silicon to be able to scale across that wide and deep portfolio as well as it helps them just reduce their overall R&D and get to market very, very quickly. And a couple of our marquee software efforts include DPDK, the Data Plane Development Kit, which is now part of the Linux Foundation; FlexRAN, which supports the build-out of open RAN and virtual RAN; as well as a host of different software ingredients, including OpenNESS and Open Visual Cloud for edge computing. And then the final part of our strategy is to really build an unmatched ecosystem. And this is really critical because we definitely want a wide, open and robust ecosystem. And to do this, we've developed very strong partnerships with not only the comm service providers, but also the cloud service providers, the enterprises and partnerships really across the entire value chain. Next slide. So when we think about what lies ahead and we think about 5G, of course, 5G is really more -- is more than just broadband. It's more than just delivering consumer services. It is expected to deliver all sorts of very interesting use cases that will deliver better business outcomes across a wide variety of industries. So with that in mind, there are really 3 focus areas that we're focused on and expect significant growth in the 5G area -- era. The first is the wireless core network. And the wireless core network, going into this year, we had about 40% of the network was virtualized or running on virtualized network servers, and that was in 2019. And this year, in 2020, that number is expected to go to 50%. And by 2024, really on the backs of 5G and network transformation, that number is expected to increase to well over 80%. Again, those are virtualized network servers running across the network. The next part of transformation and big growth area for the industry and for Intel is the radio access network. Now this portion of the network is also expected to go through a transformation, a transformation to utilize more standard-based platforms. And we, as a company, are incredibly excited about the traction we're starting to see in the marketplace with some customers starting to trial and some even starting to deploy virtual RAN and some even open RAN-based solutions based on Intel content, whether it's Intel Xeons, our FPGAs, our Ethernet or our purpose-built accelerators for this market. And these -- and this traction includes marquee announcements from the likes of Verizon, Rakuten and DISH. And then the final growth area that we see as a result of 5G, but also really, again, the result of the proliferation of AI, is truly the emergence of additional compute at the edge, whether it's at the network edge or on a vertical market customer's prem. Next slide, please. So in 5 years, humans and machines will create 10x more data than we did just this past year. And 75% of that data, we created outside of the data center, really in the physical world of factories, hospitals, retail stores and the many forms of video. And when you think about this, we expect that more than half of this data we processed, stored and analyzed much closer to the user to ensure that the use case has the right level of latency, bandwidth and security to be able to support the right overall business outcomes. So when you stop and you think about this and you think about the edge and you start imagining what this is, I want you to imagine that the edge is really a series of locations. It's not just one place. If you include various customer prems, it could include multiple places on the network edge. So with this in mind, services will be deployed at the location that really deliver the highest return to enable customers to deploy new revenue-generating services as well as lower their TCO. So there are a few different places on the network edge that we do think will deliver not only network services, but also new compute services in the future. And the first one is something that we call the next-generation CO. Now this is a fiber-rich location. And some -- and we believe it's a location that really transform into a mini-data center at the edge. It will support both fixed or wireline traffic as well as mobile network functions and be able to support traffic from enterprise customers, residential customers as well as mobile customers. And in the future, this will not only be able to support all sorts of network services, but it will also be able to scale out with those advanced computes and be able to support all sorts of new use cases, including the many flavors of media. And then the next critical location at the network edge is, of course, the radion access network. In this portion of the network, we'll see increasing demands from 5G. You'll see multiple different frequency types as well as the need to be able to support compute services at this particular location. And we've been investing for quite some time in this space, investing in incubating virtualized RAN with software reference solutions such as FlexRAN as well as hardware reference solutions for multiple years. And then the final place that we do expect really the build-out of new computes to occur is, of course, on the on-premise or a customer premise. And again, when you think about the customer premise, you can think about multiple different locations, whether it's retail store, whether it's a factory floor, whether it's a sporting venue, and I know someday soon, we'll go back to sports venues, as well as smart cities. So a tremendous amount of growth opportunity for us in the era of 5G. Next slide. So when you think about this overall opportunity, we are particularly excited about this opportunity. We do believe that 5G will be transformational and not only support all sorts of really fantastic consumer services, but it will deliver better business outcomes across a wide variety of industries. And to do this, Intel has been investing, investing significantly over the last decade to really lay a foundation for 5G, a foundation that's on the backs of network transformation and really enabling a network that's more software-defined, more agile and scalable to deal with not only these network services, but also all sorts of compute services. So when you think about 5G, I do want you to think about, again, the business of the public network, but also thinking about how they can scale out alongside AI, IoT, as IoT really build out edge computing and support a number of different services. And then the final thing that I'd like to comment on in this opening session is that Intel is truly committed, truly committed to supporting the build-out of an open and robust ecosystem. We truly believe this will fuel innovation and create all sorts of new and unique services for many different businesses for years to come. With that, I'd like to open it up to Q&A with Pierre.

Thanks, Dan. Thanks a lot for these prepared remarks.

So what strikes me, and it's not necessarily always obvious, is the point you made that how long Intel has been involved in network transformation. And I remember 10 years ago, how the company was already actively engaging with service providers and the industry. And so can you tell us a bit more about that journey? What have you built over the last 10 years that are going to be your competitive advantages for today and for tomorrow?

Yes, absolutely. So when you think about the last decade, it really started off with partnerships with communication service providers. And they came to us about 10 years ago with a very simple but a bold challenge. They asked Intel to drive the same sort of transformation that we drove in the data center and the cloud into the world of networking, meaning that we could move the network to benefit from server-based economics and help them reduce the overall TCO of their network as well as the ability to have a more agile network to roll out services much more quickly. So you heard in my opening remarks about our 3-pronged strategy. So of course, we invested in silicon to really tune it to be network optimized. We invested in software. We invested in standards as well as creating a robust ecosystem. And there were a couple, let's say, key events that really stood out as really [ sentinel ] or really foundational to making network transformation really happen. The first of which is in 2013, Intel partnered with the communications service providers and ETSI to create the network functions virtualization project. And this really set out to deliver virtualization, which is, of course, a cloud technology in a very standard way in the world of networking. And since that time, they kicked off a series of talks and trials. And then in 2015, we really saw the first kind of virtualized network server deployment on the network. And then you heard those stats that I've mentioned earlier in the opening remarks, it's definitely taken off. And then the second really critical thing that I think Intel did is that we invested early in supporting probably the most critical workload in the world of networking, and that is packet processing. And we innovated and recreated the Data Plane Development Kit, which is an open-source software package that allows customers to run packets very, very efficiently on general-purpose servers. And once that Data Plane Development Kit got a lot of momentum on the market and a lot of adoption, we made the bold move of transitioning that into an open-source project with the Linux Foundation. And that happened in about 2017. And since that time, DPDK has really become the de facto standard to running packet processing on a general-purpose server. And obviously, those are some of the things we've done in the past. And now when you think about where we're headed in the future, we are continuing to invest, of course, in DPDK and virtualization, but we're also laser-focused on the cloud-native foundation and software packages such as Kubernetes as well as innovating our new robust software packages such as FlexRAN to support open RAN as well as vRAN. And when you add this all up, this has made our business very successful. Our business in 2019 was roughly a $5 billion franchise, and it was on the backs of several double-digit growth years. And we do expect this business to be strong for years to come.

Thanks, Dan. In the way you've described your journey, I think things have always been very server-centric, CPU-centric. So am I right thinking like the -- if I think about it from a product portfolio perspective, your Intel server chips are the core of your role in telecom. But can you put that into perspective? First, tell us about how much you think -- how important are your CPUs in what you do in telecom compared to the rest and maybe go over your portfolio strategy? You have CPUs, you have SoCs, you have ASICs, you have FPGAs. You might have GPUs. So tell us about that from that perspective.

Yes, absolutely. So you're right, so we have an incredibly broad product portfolio that really meets the needs of networking customers. And it does definitely start with our CPUs. Our CPUs are incredibly important in this journey. They provide the right flexibility as well as performance to be able to support multiple different types of network workloads as well as compute services across really a wide range of network locations. Again, everywhere from back in a core network, it could be a couple of hundred miles from a user all the way into that next-generation central office, which could be 1 kilometer or up to 10 kilometers from the user. And then, of course, last-mile networks, which would be the wireless access network as well as wireline networks. So we need to have a diverse set of products that will support all of those different environments. And Intel has invested in, obviously, our Xeon processors. Xeon processor is an incredibly important part of this journey, supporting both core networks as well as even RAN deployments. And even you can see the use of Xeon on all sorts of customer prems as well. But in addition to this, we do know that, that deal to support multiple network environments is really critical. Thus, we're investing in our Xeon D product line, which is an SoC version or a power-reduced version of Xeon as well as Atom-class products to be able to fit across those, again, those diverse requirements but also have architectural consistency. So the CPUs, you can say, is really the heart of Intel's strategy and really makes up a bulk of our revenue that I mentioned earlier. But in addition to this, we are also investing in FPGAs because FPGAs are -- they are a critical component because sometimes customers want to put their own unique algorithms, and an FPGA, of course, is an extremely flexible platform to be able to do that. And then we also do offer custom ASICs for select engagements where a customer wants a truly a purpose-built device to be able to meet their specific use case.

And one maybe last quick follow-up on the portfolio front. In the long run, do you see everything getting integrated into a single platform, a single SoC? Or do you think we will continue to have this intelligent edge portfolio around telecom?

Yes. So I think actually, it's a really interesting question. I do think there's really a heritage in the industry. And particularly, Intel is about innovating and then integrating. So essentially, you're going to see Intel continue to innovate, and we're going to continue to innovate within our CPU. It means we're going to add new instructions to continue to meet the needs of networking as well as tightly coupled accelerators. But we're also going to be innovating at a platform level, meaning we're going to be innovating on network accelerators that could be discrete in nature at the beginning. And over time, if they gain in upscale economies, we could, of course, consider putting them tightly coupled next to the CPU. But in addition to this, we also will continue to innovate with Ethernet. For example, one of the key -- when you think about Ethernet, you obviously you think of the connectivity first. But then you also should think about how the Ethernet controller can help pass packets very efficiently to the Xeon processor. So in one case, with our new Columbiaville Ethernet adapter, we've innovated on something called the dynamic device personality or a DDP for short. And essentially, it's able to run comms protocols very, very efficiently. And just to give you one really quick example of that platform-level feature is that we partnered with SK Telecom with Xeon with this Ethernet controller to help them really drive and innovate as they build out their 5G core network. And in a lab setting with Xeon, with this Ethernet adapter, again, with that additional accelerator block in there, we were able to reduce general latency by up to 88%. And when you think about a 5G network, obviously, you want to reduce the overall latency, reduce the overall jitter to able to support all sorts of new services, whether it's cloud gaming, video streaming or perhaps all sorts of immersive media services that could support a ton of different business use cases in the future. So I do, again, believe that the breadth of portfolio plus the innovations we're doing, either at a platform level or at a CPU level, has really given us a strong differentiated product line in the marketplace.

Okay. That's very clear. And your last comment is a great segue into my next question. I wanted to just to dig into the core network. So you've given us some stats about how core networks have already -- are already well into the journey of moving onto virtual platforms. Now with 5G coming in, so what are like the big changes to the core network 5G triggers? And beyond what you just mentioned, that's improving latency with better integration, what are the associated challenges for Intel?

Yes, absolutely. So when you think about the core network, again, when you go back to even that ETSI story that I told just a few moments ago, with the network functions virtualization project, again, we saw a lot of talks and trials in that 2013-2014 time horizon. And then 2015 is when we really started seeing significant deployments, which you'll know that, that was roughly halfway through the LTE deployment cycle because LTE roughly started in 2010. And then, obviously, with the pace of adoption continuing to pick up, again, 50% of core network deployments will be virtualized this year. And then when you think about 2024, we're expecting well over 80% to be virtualized solutions. And the reason for this is that when you think about the core network and you think about those consumer services and then all sorts of business services, whether it will be targeting an industrial setting, a retail store, a smart venue or perhaps smart cities, that core network needs to be incredibly flexible and service-oriented. So service providers are really counting on having a core network that is really cloud-native in nature, so it can scale all sorts of new network services as well as compute services. And to give you one just really quick example about how 5G and edge may come together and how the core network needs to be flexible and dynamic, when you think about the needs of the edge and you think about a lot of high-bandwidth and low-latency use cases, there's going to be times where you're going to need to pull the packet functions of that core network forward into a location that's much closer to the user. So when you think about that and you think about Intel's road map and our architectural consistency, it's very possible for our customers to write, let's say, on a Xeon scalable processor for deep in the core network, but then also put a smaller version of that core network onto a Xeon D platform that is architecturally consistent but has a low power -- lower power profile to either sit on a customer prem or either at a network edge location. So again, when you think about the 5G and you think about network transformation, it screams of having a very flexible core network, and we're really excited about our partnerships in the industry. And of course, I just mentioned the SK Telecom one.

That's an interesting thing you just mentioned. So while we are talking about virtualization as consolidating the radio access network, you're telling us actually virtualization need to expand the core back to the edge. So it's a good illustration of how flexible the underlying infrastructure is becoming. Now maybe -- so that's for the core. Tell us about the radio access network. So with 5G, how does the RAN have to evolve? And what role do you guys play in that?

Yes, absolutely. And the RAN is obviously a very hot topic among telecommunications service providers as well as all sorts of different industry analysts. And we are seeing a transformation at RAN. And the RAN is, over time, moving to more standard-based platforms to be able to support all sorts of new use cases across a wide variety of industry verticals. But we do believe, when you think about 5G, that both a traditional RAN will be deployed as well as virtualized RAN. And we do believe that they will kind of seamlessly coexist for many, many years to deliver this next generation of really 5G-based radio access networks as both agile as well as cost effective. And one aspect that we're very excited about, we're starting to see traction really in the marketplace for virtualized RAN solutions. So these are RAN solutions that are based on general-purpose servers that use cloud-native-based technologies to be able to support the needs of the radio access network. And in this journey, we're seeing a significant traction with our Intel Xeon processors, our Intel Xeon D as well as FPGA's Ethernet. And we've also implemented, which I haven't mentioned yet, a purpose-built accelerator for the radio access network, and it's called our ACC100 accelerator. And this supports forward error correction, which is a very, let's say, compute-intensive task as part of the baseband unit. So we're excited about the progress we're seeing here. And as I mentioned, we're starting to see different customers make announcements, whether it's trials or whether it's even some early deployments.