 Hi Nicholas. This is Kenyup from Cavium. Welcome to the Cavium booth at MWC 2016. We have a lot of exciting new products and basically a lot of new arm-based products as well. The one you see here in the middle is a standard 1U 19-inch server using arm technology. It's a 64-bit CPU for this standard server. It is from Cavium Thunder X arm processor. It supports a lot of different kind of server applications. For example, you can use it in web hosting. In fact, it's been deployed in run above web hosting where you can rent cycles and try different kind of workloads. If I want to host my website ARM devices on the ARM server, I can go there and I can get it hosted there on Thunder X. Absolutely. They will be very cool on ARM website on ARM server. And it's part of OVH, right? It's a very big company. Yeah, absolutely. That kind of space. And you have a bunch of other partners you can talk about? Yeah, so we actually target many different applications. So server is one. We focus on very optimized workload. Basically, it's not just a general-purpose server processor. We optimize it for also networking use cases. For example, network function visualization. We also optimize it for a lot of storage applications like CEP and big data analysis like Hadoop. Also, a lot of different web 2.0 hyperscale data centers use cases. For example, Nginx doing the web serving. For example, your website could be served by Nginx and managed by OpenStack. So thousands and thousands of these servers and can be managed by OpenStack as well as fully supported. And also, a lot of the distro support as well like Canonico, Suzei Red Hat, VPSD, we have all the support. So there's a lot of platform software support as well available on ARM 64-bit and on ThunderX server from Cavium. And you have a whole bunch of hardware partners here. There are also hardware partners. For example, NVIDIA graphics processor can be used along with our ThunderX ARM server to accelerate a lot of financial applications, financial modeling. So you would have an ARM server and then the same data center, it connects to a bunch of GPUs from NVIDIA? Yeah, absolutely. You can use that as just how NVIDIA is accelerating other financial models today with other server processors. We also have, for example, SSDs like SandDisk partners where you do a lot of storage for big data. You can get high-speed NVMe-based storage. So we are also demonstrating the complete SEV cluster today where it is used by many different kind of applications. You know, like in a data center where you manage a lot of both structured and structured data, whether you use object storage or file storage. SEV is running on ARM-based ThunderX as well. So you can use that for storage, for web hosting scenario, for networking, you know, like NFV scenario. And you can work with OpenStack where, you know, a lot of these data centers and infrastructure are using for managing their storage. So both storage and compute, you know, this ARM-based server can do a very good job and provide a very good total cost of ownership. And for the networking industry where the Mobile World Congress is a lot of telecoms? Absolutely. And they will use this? Oh, absolutely. They're testing it right now? Absolutely. This is Mobile World Congress. One of the most important things? Mobile, 5G. So a lot of new activities there is cloud. Moving, for example, base stations to the cloud, which some people call cloud-ran, some people call centralized-ran. So we actually have three different demos on cloud-ran today here. You know, one is a demo where we go through a user trying to download something from his home. You know, in this case, it goes through a complete mobile infrastructure which is all here. Starting with the user's tablet, you can connect through LTE wireless through a remote radio head, which is this reference box designed by Cavium. And over Ethernet, which is a very low-cost way to connect to a baseband in our cloud data center. We have, this is actually a development board for Thunder, our ARM server. So this server is actually running the rest of the mobile infrastructure like the baseband, the mobile core, all of that running on this. So we have structured this as basically a stack of servers at the data center in the cloud edge of the mobile network. And so through this, it completes the mobile infrastructure. And then it goes back to IoT Gateway, which is also based on Cavium Multicore processor Octane Line. It has connection to some sensors and cameras. So that information goes back all the way back through the mobile core, the baseband in the cloud data center, now goes back to the remote radio head, which is at a cell site called the user. Back to wireless leads through LTE to the user pad, you can see the video. So it completes the entire mobile infrastructure is running live on ARM-based server. Yeah, this is an entire... Everything on ARM, but oh, on MIPS also over there. Well, some makes architecture, but you know, they're all SOCs. The server is based on ARM. And what are you talking about here? So here is another proof of concept that we are working with some service providers like AT&T, Verizon, SK Telecom. And this is another way to look at next generation mobile infrastructure where you see a virtualized RAN, just like we demonstrated here. In addition, we are also doing some distributed EPC mobile core. So you get a lot more edge services. And all that could be running on ARM servers as well. So that's keeping a lot of people busy what's going on right now. And it's really important to use an ARM server for all this to be possible, right? Otherwise, there's too much power consumption. Absolutely. So ARM, you know, with our technology, we integrate a lot of CPU cores. In fact, each one of our processors integrates 48 64-bit ARM CPU cores, plus a lot of other networking acceleration and interfaces. So it's very high integration. As a result, you can pack more compute in a much higher density way with our ARM server. And it consumes less power. It costs less money. Total cost of ownership is much better. And also the important thing is that the ecosystem support is there. So the software for running a data center, for managing these machines are all supporting ARM and Thunder X. So you can get the software and same experience, but at a much lower TCO. So lower TCO makes it affordable for the consumer at the end to get a better service. It's getting faster and faster. It's going to be gigabits per second at some point. It's going to be crazy fast. And we cannot just continue the same way we're doing things today. It's interesting. You mentioned gigabit. I mean, each one of these CPUs includes 8, 10 gig interface. And you can use it for 40 gig also. So 2, 40 gigabit internet as well. So these CPUs are already built, you know, integrating very high speed internet connection. That's totally awesome. And you have on the slide here, you're showing... So you have all these hardware partners. Yeah, exactly. So interestingly, there are different form factors. And people use it different ways. You know, for example, we saw this 1U server. We should close the box here. Is this 1U? Yeah, this is 1U 19. It's basically the standard rack master. How many SOCs in here? So this one has one, but you can have one with two CPUs inside. Any one of those? Yeah, you could have that. Yes. And then it looks like this inside, which is... You know, basically a lot of the space inside is just hard disk and power supply. So the board is actually very clean because we integrate everything on one chip. So the board is much cleaner. As a result, you can actually build a lot more denser kind of server designs. And in fact, there are other designs where you see dense form factors. So, you know, for example, this one you see a four sled. And each one can have two CPUs in there. So each CPU is 48 cores. You get a lot of thousands of cores in this one chassis. And then we have also OCP, open compute platform designs. So these designs are actually made by OEM and ODM partners. These guys? Yeah, some of these guys. You know, for example, we see Kegabyte, which is a major ODM for servers. Asus, Widewind, these are all major server design ODDMs. Pecatron? Pecatron. You see also like Penguin and Stack Velocity. They focus a lot on storage type of platforms. So there are also form factors similar to this, but we pack a lot more hard disks. So as a result, the total cost of ownership again is much lower. Because, you know, directly all these hard disks up to 16 can be connected to one CPU. I can imagine all these hardware looks awesome. It's too bad you didn't bring more of the different crazy server designs. Right, yeah. There's a lot of different variability happening. There's a lot of designs happening right now. A lot, yeah. So a lot of different use cases, as I mentioned. And then each use case could have different form factor. There's a lot of new innovation in server design. So basically the industry trend is that servers are very different nowadays. So historically servers are very general purpose, but nowadays you have a lot of Web 2.0 companies. They focus on one application. They demand all kinds of optimization, acceleration, because they only need to run one application. So instead of just general purpose, they want workload optimized. And we provide a lot of relevant hardware acceleration for that and integration of IO for that. So there are also different form factors to squeeze out more cost, make it lower power, make it higher density. I can imagine if somebody was a designer of servers five years ago and they slept for five years and they wake up now, it's a totally different world, right? It's very different. It's very different what's going on. Before it was just boring, Intel-only stuff and nothing so customized, right? Yeah, we provide alternative. And you can have them run around in circles and try to imagine the future and how to design the VCB and how to best use your... Absolutely. I think the future is that there's a much bigger ecosystem and that's happening in terms of hyperscale data center, Web 2.0 companies. Nowadays also service providers. They want a much bigger ecosystem. They want alternatives. So we're providing ARM-based alternatives with better performance and better total cost. I hope you can have some very big cloud company guys happy to announce that they have huge data centers with your... Yeah, absolutely. This could happen. That's up to them, right? Absolutely. So hopefully they'll announce it very soon that there will be millions of them... Yeah, there was announcement of a web-based web hosting company, one above, which is subsidiary of OVH. So there was announced last year. Also, you know, this ARM servers are also very popular with high-performance computing. You know, about one or two years ago, Cray computer actually announced that they will use this processor, ThunderX ARM processor, or their high-performance compute. So we are building supercomputers based on our ARM... So Cray is a supercomputing... Yeah, Cray is a supercomputing company. They build supercomputers. That's cool. And Acer is doing a bunch of servers, and then Mitac... Yeah, so these are all ODMs. And there's more? There's even more than that, right? Yeah, there's even more than that. We're just showing some examples here. There wasn't enough space on the slide. You have more partners. Yeah, we do have more partners. Absolutely. More partners and customers. So it's going to be a busy year this year. Absolutely. And this is in production. So... This is in mass production. Yeah. So... And there's several of these designs in production. Absolutely. Absolutely, yeah. And as you see, as some examples of ODMs, there are more that we have not shown here. And ODMs also.