 Good afternoon, everybody. How are you doing? So my name is Rajesh Gadir. I lead architecture efforts. I'm CTO for Intel's networking business. And it's an absolute pleasure to be here in Amsterdam this week. So I'm going to talk about 5G and the innovations at the network edge. If I asked each one of you in the room here what your perspective is on 5G, I'm likely going to get many different answers, perhaps as many as the number of people in the room here. And perhaps for some of you, 5G is still a little bit further away, maybe a couple of years, maybe longer. But for us, especially for somebody like me who works on network infrastructure, 5G is here and now. And in fact, we've been actually hard at work over the last couple of years, transforming the network, virtualizing the network, trying to get it ready for the impending 5G transition. So thinking about some of the possibilities with 5G and all the innovation, one thing that all of you would agree that 5G does mean a lot of data. All that innovation, all that new services would lead to a lot of data. And so if you think about some of the research studies in the market today, I'll share a couple of data points. Over the next four to five years, 70% of world's population is going to have a smartphone. In fact, 90% of people over the age of six would have some kind of a mobile device. And 50% of the data that is produced by 2020 will be through the things phenomenon or by machines. So if you look at some of the comparison, by 2020, a human user would produce 1.5 gigabytes of data per day. Now, in comparison, autonomous vehicles are going to produce 4 terabytes of data per day. And connected planes would produce 5 terabytes of data per day. Smart factories, a petabyte of data per day, that's 1,000 terabytes. And you look at cloud video providers, that's 750 petabytes. Just imagine what that means, 750 petabytes. Now, so across the board, with all this innovation, new services, and that data deluge, from a network infrastructure perspective, what that means is we're going to need a scalable, programmable, and intelligent network. So when you think about 5G, and you look at it from an end to end perspective, from the access to the network edge, to the core, to the cloud data center, what you see here is a little data center, a little cloud everywhere, in the RAM, at the edge, in the network core, and of course, the cloud. Now, one of the happening places as a result of 5G, what I like to refer to as the epicenter of services innovation is going to be at the access and at the edge. Now, two things are happening here, from an edge perspective. One, we are seeing a lot of consolidation and convergence of network functions. So what's happening is, with all the work that we have done with NFV and SDN over the last few years, we are seeing a convergence of fixed network, of course, mobile network, and also the cable infrastructure, all coming together nicely and converging at the network edge. Now, the second part of it is actually much more interesting, which is actually all the new services that are going to be enabled as a result of the 5G technology, which has 10x lower latency and 10x better bandwidth as compared to LTE. So what you see here, in terms of new services, a lot of that is going to be led by video. As you guys know, 70% to 80% of the network traffic is video. And so we're going to actually see things like real-time video transcoding, immersive media, and video analytics, and AI everywhere. And then we'll also actually start, we're seeing a lot of interest in blockchain communities because a lot of what 5G enables is really service communities. And in those service communities, it becomes very important to be able to actually share data between the users in that service community. And blockchain as a technology becomes very interesting because of its distributed ledger and because of a secure way of sharing data within that service community. Now, the cloud itself is evolving as well. And we are seeing actually more and more innovations like the cloud native, service mesh architectures, new paradigms like function as a service, back end as a service, API gateways. We're seeing analytics everywhere, AI everywhere. And a lot of what is happening in cloud also applies in at the network edge. Now, one of the key tenets of NFV was going to be network automation. And this is an area that I strongly believe we have a lot more work to do, and this actually applies across the network. Now, we've actually made some good strides with Onap, but there's actually more to be done in terms of how we, instead of having to roll a truck every time we create and deploy a service, the ability to automate and be able to actually create a service from a central location. So what's the big hoopla about when it comes to edge? So when you think about some of these new services we are talking about from a 5G perspective, you think about things like autonomous driving, things like immersive media and video analytics, things like drones. One thing that is actually common between all of them is that they all desire ultra low latency. And so if you look at this diagram, the reason why edge becomes the epicenter of the service innovation is because the latency that is required by these applications is of the order of five milliseconds, or perhaps even lower. And the only way to do that, in spite of the 10x better latency characteristics with 5G, is to be able to actually host these applications and these services at the network edge. So when you look at the network edge, now edge, where is this mythical edge? Now edge can mean a few different things for different people, but for me actually, and actually most of the com service providers or the telco operators, the edge, as in the central office edge, is of a lot of interest. And the reason for that is because this is such a precious real estate that is available very close to customers, both enterprise and end users. And as a result, the ability to drive innovations by hosting applications and services at that network edge is something that would transform not just the business, but it has the potential of transforming our lives. So here, what I thought I would do is show you a few examples of what these central offices look like. And they actually come in different shapes and sizes. And as you can see here, there's a small hut at the bottom, which is like an aggregation office or aggregation center to maybe a mid-sized central office to something that you see at the top, which is a multi-story building and a few floors, sort of like housing central offices. And I actually tried to plot what this looks like in a place like US. There are 20,000 such central offices. So you can just imagine the magnitude of what this transformation could mean if you actually put a little data center in each of these central offices. And what kind of innovation, what kind of low-latency services can that drive? Now, that's just in US. In comparison, if you look at China, there are 70,000 of these. So I think when you look at what's possible, this transformation and the potential to actually drive these new services enabled by 5G at the network edge at the central offices, transforming the central offices, it's a huge opportunity. So now, of course, one of the fundamental foundational enablers for 5G and these edge services is really the network transformation, the work that we have together been driving over the last few years in the form of NFV and SDN. And if anything, 5G and edge services are actually going to put a lot more pressure. And we need to actually continue to do the network transformation journey, continue to virtualize the network with more urgency. Let me actually give you a couple of examples that will illustrate how the flexibility that is introduced by NFV is critical for the next wave of services at the network edge. So one of the sample applications, one of the applications that we've all been working towards virtualizing is the packet core, or the EPC, or the wall packet core. And a key function there is the user plane, or the packet gateway. And so you can imagine an application such as an IoT gateway for farming. The packet gateway associated with that can be located in the cloud. Because for that application, there really is not a big sensitivity with respect to latency or there's not a huge amount of bandwidth. And so we are perfectly OK taking that disaggregated packet gateway and putting it in the cloud somewhere. Now another example, if we are actually trying to enhance the user experience in a stadium with a 360 video, maybe with some AR, then that same packet gateway needs to be located as close to the user as possible. Perhaps within the stadium itself. And yet another example of a content delivery network, CDN, maybe that packet gateway or that user plane is OK in a central office like I was showing earlier. So all of this flexibility is foundational to how we are going to actually be able to deploy 5G and the edge services. Now from an Intel perspective, here's the journey that we've been on. And you guys have heard us talk about this. We've actually spent a lot of time on the workload convergence and have made great progress. So this is where we do application control plane, packet processing of the user plane or data plane, and signal processing, such as media workloads or baseband processing on base stations, all running efficiently on a general purpose server architecture. Now, we have to continue to make progress on this. And some of the things that we are trying to do as a next step here, preparing for 5G is driving further innovations in DPDK, which is actually a foundational enabler for efficient packet processing. So what you're doing there is looking at evolving DPDK with a hardware abstraction or a device abstraction layer so we can accelerate the data planes with for technologies such as crypto, with FPGA, smartnecks, and so on. We are also looking at evolving DPDK for cloud native applications because networking is no longer going to be about layer to layer 3 networking. I think in a world three, four, five years from now, networking is going to be increasingly about layer 4 to layer 7 networking and service mesh architectures. And so we are looking at what that means in terms of DPDK packet processing on general purpose server architecture and signal processing. So here, we are looking at how do we improve video processing, video transcoding on standard server architecture. The second part of our strategy was network transformation. Of course, we've made great progress there. I think you look at examples such as virtual CPE and SDVAN in enterprise networks or EPC and virtual IMS in the wireless core. There are numerous examples where increasingly, the network functions are being deployed with NFV and SDN. Now, with 5G and edge computing, increasingly, people are asking the question, NFV, yes, but how about cloud ready and cloud native applications? And so that's going to be the additional focus now over the next couple of years as we actually begin to roll out 5G and edge services. And then the last part here, Intel takes great pride in how we collaborate and work with the open source community. And I think there's numerous examples. DPDK, of course, is something that we created, and it's in the open source now. But we also actually have an excellent participation collaboration in OWNAP, OPNFE, and many other open source projects. And of course, we actually collaborate with many service providers and OEMs, like you see on this page here. So next, actually, one thing I wanted to talk about, a question that I get asked a lot as the CTO for the networking business, I get asked, hey, how do you look at the key tenants of network architecture? What are some of the key vectors that we need to actually push harder on? And so I was thinking about that as I was actually preparing for my keynote here. And I wanted to capture it on a single foil. So what you see here on this page are the three vectors that we are pushing really hard in terms of innovation. First is what I like to call scale up. And so this is where we want to actually improve the performance per watt per dollar. These are innovations such as new instructions for packet processing, for baseband processing, new kinds of accelerators, and fundamentally improving the general purpose server architecture to convert it into best in class network platform. Scale up. Second is scale out. This is disaggregation. And the vision of creating pools of compute, pools of networking, pools of storage, pools of accelerators. And my vision really is, if you look at four or five years from now, you should be able to compose a service by pulling these resources out of these pools. So dynamically create an application, a service, no matter where the resources lie in the cloud. And so that's what I like to call scale out. And then, of course, the middle vector here is automation. And we've actually done a lot of work here in terms of getting the platform telemetry and to be able to do build resilience and fault tolerance. But there's more work to be done. And in fact, one of the areas that we are going to push hard here is the ability to apply machine learning to all this platform telemetry data so we can really close the loop and drive network automation faster. So the three key tenants then, again, scale up, scale out, and network automation. So just a quick word on the portfolio products Intel offers for 5G and for Edge services. Of course, I have to do a little advertising. Blue blood runs thick in my arteries and veins. So here's sort of like the Intel's portfolio of products. What you see here is a nice scalable set of products, entry-level SOCs that integrate ethernet, even switching capabilities, and accelerators for crypto and other workloads, to the Xeon D processors. So one of the things that we're going to need at the Edge is efficient and high performance within the power and thermal constraints of the network Edge. And so we have Xeon D processors here that provide up to 200 gigabits per second of application throughput. And then, of course, Intel's flagship Xeon processors that scale all the way up to 600 gigabits per second today in terms of what a single dual socket Xeon server can do in terms of packet processing. Now, when you actually combine all that with the platform technologies, such as our 3D cross-point technology, memory technology, our next generation of 3D cross-point SSDs, FPGAs, and the Silicon Photonics technology, and you have everything that you need to build scalable, efficient, and intelligent platforms for the next generation network infrastructure. So bringing it all together, I just want to conclude with a few things here. First, 5G is here and now. So it's not something that is very far. I think the work that we are doing today is to really transform the network and prepare the network infrastructure with not just the hardware technologies, but also key elements of software that are going to actually drive towards a more cloud native and an efficient infrastructure that can support 5G and edge services. Second, there is a lot of innovation that's actually going to happen. There's lots of different applications, new services. And we at Intel are investing end-to-end. I showed you the product portfolio. But even when you look at across the board, from modern technology to radio access network to Edge to the network core to cloud, Intel's looking at this from an end-to-end optimization standpoint. And so my ask to you is, I would love to actually collaborate with you, spend the time here in Amsterdam at the ONS, talk a little bit more about some of the things I talked about in terms of the evolution of network infrastructure with more of a cloud native transformation, and some of the key things that we can work together to realize the full potential of 5G. So let's collaborate, let's work together, and drive the next set of innovation cycle for 5G and network edge. Thank you.