 So we're here at the ARM TechCon 2016, and so who are you? Hey, I'm Jan Fischer. I run our product marketing for ARM and other hyperscale architectures. And you are John Masters. That's right. Last time I checked. Yeah, and you're going to do a keynote. I am going to do a keynote. Hopefully it's going to be a good keynote. So I'm chief architect at Red Hat. And see, we have these fabulous banners this year, which Jan pays full with marketing dollars, which is amazing. And I plan to steal this banner right after we're done and take it home like I did with last year's banner. What did you do with last year's banner? It's happily sitting in my bedroom at home. And people wonder, why do you have a giant Red Hat banner? And I don't know, but I'll have two giant Red Hat banners pretty soon. We're talking about drinking our own Kool-Aid. There we go. Yeah, exactly. Yeah, exactly. Exactly. And it makes you dream even bigger at night? Maybe. Maybe, right? Yeah. So how long have you been promoting this vision that's happening right now with the ARM servers? So I've been working on ARM servers for about five years. And I was just in a different interview. And I was saying that actually when we started this, we realized that it was going to be a five to 10-year journey. So what we're seeing now, and I'm going to talk about in my keynote today to an extent, we're starting to see some traction. But to really get moving with a new architecture, it's a many-year journey. I would say at this point, where we are is we have a single operating system build in our development preview that runs on 10 different implementations of the architecture. So 10 different pieces of vendor silicon, completely differently implemented, all run the same unmodified OS. And then you can put those into dozens and dozens of different platforms. So we are now at the point. Part of me wishes we'd been here a while ago, but we are where we want to be in terms of the ability to run the same OS on many, many different vendors. And we are poised now, I would say, over the next 12 months to start to see a lot more adoption of ARM servers as some of the bigger players start to come to market, as we start to see the big OEMs taking a lot more interest in the enterprise space. So you envision this ARM servers market to be potentially very, very big, right? And you're playing a big role in there. We're absolutely trying to stay in alignment with what our customers are asking us to do. And we have had people asking for ARM by name. They have strategies for alternative architectures. They have been interested in exploring more of ARM designs, and we are supporting them in their desire to have heterogeneous data center infrastructure. As well as, you know, with our development preview, we keep adding features. We're enabling major technologies like virtualization. We're looking into providing newer tools with our operating system, while it's still staying in development preview. So ARM were saying last year, they were saying 2020. How many percent? 20%? They said 20%, right? Yeah, it could even be higher, yeah. Could be higher? Yeah. So what's the click that's gonna happen that suddenly everybody's using ARM servers? I would say there's three things happening. I would say the first thing is, if you look at where the growth is in the market, you've got a lot of emerging players in Asia right now in APAC that are looking for an alternative architecture. And some of it's driven by government mandates and so on. They're looking to say, I want to have an alternative. I'm very in favor of choice. This is their view. And so they are driving a lot of interest in the growth of, number two, you've got the hyperscalers in the U.S. and elsewhere, the big guys that say, you know what, if I can save 20 or 30% in my OPEX or my CAPEX, then I'll take a serious look at this. Yes, it's energy efficient, but I'm willing to have the benefits I can get from having alternative suppliers. And then when you look at kind of the way that the industry is going, the way that the different technologies are going, we've kind of run out of the cheap process innovations we've seen, right? So, we went from 90 nanometer down to 65 nanometer down to, the last couple of years we've been down at kind of the 28, 22, we're at the 14 FinFET kind of stage now, but it's getting much, much harder to pack those transistors more tightly. And there are certain inherent limits that are coming in the next couple of years. And so nobody has a magic wand that can magically make the laws of physics go away. And so what that means is there's an opportunity here for Arm to say, okay, well, the other guys are not just gonna get a faster chip every year, there's an opportunity here to come in with an alternative and be successful. And so the other thing that's happening in hand in hand with this is a transition towards using workload acceleration. So we've seen a lot of involvement within Red Hat in terms of working with the work groups that are building out coherent interconnects, new memory fabrics, new frameworks for interfacing with workload acceleration. So all these pieces are gonna come together. It's not one thing. It's gonna be the changing workloads, the ability for Arm to come in and disrupt the existing model because Moore's law is basically dead, let's be honest about it. You've got a lot of interest from the hyperscalers to see alternative choices. And you've got kind of a Chinese government mandate to go and do something different as well. And the chip makers in there, like Applied Micro and KVM, they had kind of first second generation chips where it's a long process to make a chip and now they're getting to, let's say, X-Gene 3 and Thunder X2 is pretty cool, right? That's right, that's right. It's one of these things where the first generation is, I don't like to call them pipe cleaners exactly, but you never kind of get your moon landing with the first generation because you go and do a first generation in order to get people's interest. They start looking at these things and start saying, okay, well, maybe this'll be credible when I get to the next one. But what they wanna see is they wanna see a pipeline. They wanna say there's a roadmap, there's a future. So I'm gonna pick this vendor and I'm gonna say, okay, this vendor has a roadmap. The reason that these guys announced Generation 2 when they're in Gen 1 is they want the others to say, oh, okay, if I choose this, there's a future for me, right? And the reason that we as a company have been doing these development preview releases, several of them now is we want people to see that we are interested in this for the long term. So we're not yet releasing a supported product and the reason for that is that we're waiting for, you know, some of the big tier ones to come along. We're waiting for some of the traditional kind of enterprise market pieces to come together before we can get into that space. But we're ready to go when the customers are asking for us to go there. They're starting to ask for this now and we're starting to get there. But, you know, what people want to hear from us and from other Linux vendors and non-Linux vendors is that they have a roadmap and a future and we do and we're working on this and the same thing for the Silicon Games. So as soon as, I guess, as American companies, let's say, when they see the mathematics and they can see they can save some money, they're just gonna do it, right? They're not gonna avoid saving millions of dollars every year on their server parts. I think money talks. I think, though, that it's not an instantaneous saving. It takes time to deploy a new architecture and that's why it hasn't happened instantaneously. That's why we've been doing this for five years but because we've been doing it for five years, we are now at the point where if somebody comes along with a credible offering and they have and they are and people do the numbers and say, okay, I'm gonna save millions of dollars by using that, then it's gonna drive competition in a good way. And by the way, we don't prefer any one architecture vendor. Now, I personally am a big fanboy. I think people know that but we also have people that are big fans of other architectures inside Red Hat and Red Hat is a very neutral company. Yeah, since we're rooted in open source, we technically don't prefer not just any specific chipmaker or any specific architecture. We want to build from a common core that is upstream and we want to apply it across different architectures across the board, across the variety of the systems that exist and the more systems that we can support with a single, singular operating system, single distribution, if you will, the better it is for the market, the better it is for customers, the better it is for upstream. So Red Hat is one of the, let's say, big successful Linux companies, right? We are the biggest successful Linux company, if you will. Yes, we have to admit to that. Yeah, I think that's a true statement. So the way that it works at Red Hat is companies come and say, hey, can you help us? And then you help them to make it real? How does it work? So we started a partner early access program about three years ago. That's what actually was the foundation of us building the Red Hat Enterprise Linux server form and the one that we have in the development preview right now. We are cooperating, we have been working with the Silicon vendors, was actually was armed directly and also our OEMs and ODMs in designing the operating system that runs across variety of their designs and their implementations of the RMV8 architecture. And at Linaro, lots of stuff's been happening over the last years. That's right. It's a cool bunch of guys, right? It is. It's fun to work with. It is, we actually just renewed our membership in the Linaro Enterprise Group, which is a collaboration between all these different companies to accelerate arm in the enterprise. We also doubled down and went into Linaro Enterprise IoT Group light. Because did you see the keynote? When Masayoshi-san said one trillion ARM CPUs, one trillion IoT devices, right? They need servers. They do, they need gateways. So the thing that we're very, very bullish on is in terms of infrastructure, in terms of telco infrastructure. So NFV, I was speaking at TechCon here about OP NFV, the open platform for network function virtualization. It's a tongue twister. There's a lot of acronyms and buzzwords in NFV. And I tried to explain what they all are in kind of how we're working with ARM and Linaro and other groups like OP NFV project, as part of Linux Foundation, to accelerate those pieces. So we need the telco story, the CSP, the communication service provider story, that they can deploy NFV infrastructure, which also expands to the edge, right? To the gateways that are near the customer. Because in 5G networks, there's so much data moving around. You can't just centralize it all kind of in a back office. The next wave of IoT will most likely be running right over that 5G networks. And therefore there is a direct connection between what's happening in the telco world and the entire re-haul of the infrastructure and what's happening in the IoT world. And that's why there is a very strong connection. And Massasan definitely sees it in the right way. It is an explosion of sorts. I don't know if it's Cambrian explosion, but it certainly is an explosion. And we also there, I mean, the unique position that we are in, I think from our perspective, from Red Hat, is that we can bring some order to, the existing ecosystems that may have not been standardized, may have not been looked upon as in need of standardization. We're doing it very well in the enterprise world. Even using the Linaur example, we brought some order into leg, so to speak. We brought our designs together, helped with the specification and standards. And I think we can have the same influence in IoT. And that's our goal. We would probably look at get-ways and up in terms of that ecosystem, but it certainly is in need of having some standards set. And let me use an example, right? To help everyone. Yeah, let me use an example from last week. So last Friday, we had, in the US, we had a particular, we had a major outage of a lot of different companies that provide services to people, right? It was a distributed denial of service attack, right, DDoS attack. And it was done against DNS infrastructure that was being used by these services to resolve their names on the internet. But the way it happened was by exploiting insecure and openly accessible IoT devices and embedded devices that people had in their homes. Which there are billions of. And billions now. And now I mentioned, every light bulb, if every light bulb has an IP address, right? And every light bulb becomes a. Every light bulb has a opinion about giving light. Yeah. We have a problem with it. We have a problem, right? So what we need to do, the reason why, one reason that I think we can really help as a company is that we believe in standardization and we believe in having industry standard platforms so that if you're building a gateway that's going to provide the secure interface to a lot of devices over time, what you want to do is, you don't want to build, this is kind of a hacked up embedded thing that has no future and no security updates and some custom board support package running on it and all this stuff. What you want is to have a standard platform on which you can run standard software. If you don't like our software, our software is great but if you don't like our software then run somebody else's software. But let's have a standard platform so you can get security updates and you can handle this over time because you're going to deploy this infrastructure and it's going to live for decades. And that does not scale if you are saying here's my kind of hacked up custom embedded solution that I built last week in my garage. That isn't going to scale for 10 years or 20 years of support. And do you have a role to play to help these IoT guys build more secure IoT devices? Absolutely. Well, if we apply same good practices that we have done for years and in terms of sanitizing, cleansing the code and actually applying best practices of security kind of, you know, approaches, so secure approaches to code handling, I think they would benefit from it. I also want to lean on the example from today's keynote about hacking the GYPS and how Chrysler is unable to fix the problem without having to recall all these vehicles physically. So having a call home or call for help line is very important. That's part of what embedded system may not, ecosystem may not know or understand. And we're trying to say, give yourself a way out of it. By sticking to standards, you guarantee that it's not going to be hackable. You have a future and you have secure updates through provisioning. You have all the things that we know how to do in enterprise. It's different, but there is, there's a lot of commonality in terms of the concepts. Cool. So looking forward to your keynote. Thank you very much. Good talking with you.