 Okay, welcome back everyone, we're here live in Las Vegas, this is SiliconANGLE's theCUBE live coverage, exclusive coverage of IBM, Paul's IBM's premiere cloud conference. This is theCUBE, we extract a signal from the noise. I'm John Furrier, the founder of SiliconANGLE. I'm joined by co-host Dave Vellante, co-founder of Wikibon.org, and our next guest is Doug Baylog, GM of Power Systems, Systems and Technology Group. Welcome to theCUBE. Thank you, appreciate being here. So we had Steve Mills on earlier, we're going to tell you about the perspective. You've been a 30 year vet at IBM, you've seen many movies being played, transformations. IBM has maintained their DNA over all those transformations. First question is, how does this rank relative to inflection points within IBM? You get the cloud, you get the big date, everything's happening on premise in the cloud, full disruption, a lot of opportunities. Yeah, as a 30 year veteran, I mean this is, I have to admit, sort of one of the most profound set of changes occurring at a single point in time that I've seen in my 30 years. And we've gone through plenty of, almost sort of single cycle changes, obviously some economics up and downs, but this is pretty profound what's happening, right? As you said, the whole role around big data and how clients are just drowning in data, looking for solutions to that, trying to get business insights from their data, the role of mobile feeding that data, creating that data, the social aspect of how that plays in and then the on premise, off premise, and aspect of hybrid cloud. I mean it's just moving at lightning speed right now. And the winners will be those who respond to it, right, at lightning speed. So you've been in the, managed a lot of the business from the mainframe, UNIX, and then now Power Systems. Computing's changing, you got, you know, potentially infinite available compute in the cloud, you got virtualization, you got a lot of disruption and innovation, even on prem in the data center, software, kind of at the center of the value proposition. So what is the state of the systems these days? As people are designing the hardware and the systems, you got virtualization, you got software on a chip, new compilers, with all this tech, tell us what's your vision there and how you see that playing out? Yeah, so a couple aspects about the systems. We still see there's a strong client need for highly innovative systems to solve real purpose build issues, right? The issue around data and the different types of data are not one that can be addressed by simply taking commodities off the shelf and strapping them together. It really takes good innovation, good systems design, I believe certainly in the power business, a strong processor design even focused on big data and then how you build the system around it. So then the other thing is openness. I mean, I think it's foolish for anybody to think one company can figure it all out. So that's from a system design standpoint, why it's so critical in my view, not only to do a lot of the great design yourself, but then open it up to a set of ecosystem partners that can sort of add on extra value and take it up a whole nother level. What's your experience with the ecosystem? Obviously some ecosystem formulas are different, but now you mentioned that the entire landscape from channel partners to developers is different now. It's more open, it's more dynamic, it's more transparent. You know, which all in democratization within he earlier, democratization, people are being liberated with software and big data. So is it a new formula? What's your approach? How are you going to win the ecosystem? Yeah, so in the past, we would have kind of approached the ecosystem almost one vendor at a time saying, hey, would you come work on power, right? One vendor at a time, you'd do it through economic models, maybe it's investments of people in capital upfront, maybe it's investments at the backend. That just doesn't scale though in the end. So you've got to find a new way to attract the ecosystem. Obviously the approach we're taking is this open approach. So we're attracting the open ecosystem in our view through a couple of things. First off is open power. We fundamentally believe open power changes the game for the power architecture. It creates a consortium based approach. Some describe it as an arm-like model for the data center that says, you know what, anybody can have access with intellectual property to the power processor. They can do derivatives off it to add value to it. In fact, at the end of the day, and it says surprise some, we'll even let eventually clients or users take it and fab it at their own fabs. So it creates a whole new access point for technology. We were never going to sell power servers from IBM to some of the large internet data centers. They've got a model that says they're going to build it themselves. We're seeing that validation. We just covered the open compute summit and San Jose and Dave and I were joking because I was just at the Apple 30th anniversary of the Mac and seeing all the old timers they were talking about. I had 6K to work with and had to take it from the finder from MacPaint. But it was kind of like the analogy was you're seeing kind of a homebrew, tinkerer engineering culture coming back as developers in hardware for things like this to innovate. So you're seeing that trend back. Does that kind of, does that tie into that? It absolutely does. I mean from an ecosystem, we're not just stopping though at the hardware layer in terms of innovators. Certainly we love the fact that a company like Mellanox is a part of open power and bringing IO innovation. We love the fact that NVIDIA is bringing GPU accelerators to the party. We love the fact that some of the ODMs like Tion and others are bringing that innovation. But at the same time, I want to continue to extend up the stack into the software layers and even the end users, right? So that we get more of the likes of Google participating in open power. It's a big opportunity. Talk a little bit more about open power and the initiative. You mentioned Mellanox. So essentially you're saying you're opening up the IP, people can build derivative products. Who other than Mellanox is sort of involved? Yeah, so as I said, David, the first five founding members, and this started back in August of last year, we announced together the five of us. That was IBM, Google, Mellanox, and Tion and NVIDIA announced sort of this intent to form open power. So that was sort of the five folks who came together saw a common need in the market for openness around servers. Obviously each of those brought a different perspective from the server ecosystem and they got together and by December we actually announced the official formation of the company. Since then we've sort of thrown the gates open and said anybody can join. We now have 12 members. Recently we added Samsung two weeks ago and we announced that out at IBM Partner World. Samsung very interested from a memory perspective of bringing sort of their memory innovation to be part of the open power foundation. We have 70 other companies now that are sort of working through the process of filling out the paperwork, applying to be part of it, and becoming part of the open power foundation. So it's a rich looking ecosystem right now. So where is IBM applying power? I think we understand that pretty well but I'd love to understand that more. And where do you see the ecosystem applying power? I think in use cases, workloads, applications. Yeah, so open power, we have had other power-based ecosystems in the past. They've been focused at times on the embedded space. They've been focused in the past in the PC space. Open power is focused on the server space. So we're not being distracted by other, we're not trying to be mobile. We're not trying to put a power in your pocket. This is about four solving problems for the data center. So around the ecosystem how I see it is because I get asked this question all the time. So what's the end goal look like? How do you make this successful? There's three ways in which we capture value from this, right? One is we sell our intellectual property just like an arm-like model for those who want to take it and use it within their market. A great example of that is a couple weeks back we announced Suzhou Power Core in China. China very interested in moving as a country from a manufacturing country to a one building their own domestic IT infrastructure, right? Very excited it looks like in terms of having power be part of that domestic IT agenda. So that's a licensed model. That's a licensed model around the processor and what they do with it obviously that's up to them and I can't disclose all that, but very set about taking that and creating derivative rights around it, right? That's a processor piece around intellectual property licensing. The other is, as I said, we'll sell power eight chips in the future to companies who want to buy motherboards from Tian who's a motherboard company put power eight chips on, maybe do something different with the server and deploy it within their internet data center. So again, that's a chip sell then from IBM, right? And then the third really is the amount of innovation these partnerships are driving. As we look at some of the innovative features coming to market here later this year around power eight, the ability for these companies to innovate for the market, but then I get to capture that innovation and bring in an IBM servers, allows me to bring phenomenal innovation to my clients. So in particular number two is, so if you go back 10 years, you looked at big giant internet data centers, you know, the common wisdom was, oh, they're just going to layer software on top of commodity hardware and that's it game over. What's actually happened is the whole flip, 180 flip, highly customized, right? I mean, you look at OCP, as John said, we were there a couple weeks ago, you talk to Amazon, they say, well, OCP not enough customization for us, right? So of course you come from the world of customization. So first of all, did that surprise you at all? Is it something that you expected? And how does particularly number two, sort of power eight ships going into the ecosystem, play into that world of high customization? Yeah, one of the things we found through engaging with a lot of these internet data centers, as you said, is they're each doing innovative things, but they're all slightly different. That layer of software clearly exists, but how that layer works with the underlying infrastructure as well as the middleware and applications on top, varies dramatically between each of those companies. So they all have a strong point of view and they all think they're right, by the way. And for their, and they are. And they are for their business, right? So to think we were going to go in with sort of vanilla for everybody, wasn't going to happen. That's where sort of those three different aspects that I laid out, everything from doing derivative rights around processors for those who want high customization to selling ships for end users to build motherboards that are unique for their environment, to IBM capturing that innovation, me selling servers, allows me to address a whole set of clients I was never going to get to before. It really, as Ginny likes to say, it's about a new era, trying to get to new buyers. That's really what I'm going after, new buyers here. It is, as the general measure, it changes your mental model a little bit, right? You're talking to a bank one day or somebody in HPC or somebody wants to apply Watson as an end customer and bringing in services and the other end, you're talking ecosystem, huh? How does Doug split his arms, legs, toes? Talk about that a little bit. A lot of that does come together though, as we're seeing it, a lot of the cloud deployments are looking for ways to do rapid pace innovation in the cloud, but they're also trying to connect the core business systems out of clients own infrastructure. So there's sort of a whole notion of a dynamic cloud, hybrid cloud. I see it really playing out and certainly in the power business. My existing on-premise clients today, they're not looking to throw their infrastructure out. They're looking at how do they extend it, get access to the cloud development, set a population out there. How do they mobile enablement? How do they create that next level of web-based perhaps engagement models? And I think that sort of notion of hybrid, power in the core, running as traditional workloads, power in the front end, we're running some of those systems of engagement at a whole different economic model is really where we're going to see this thing play out. So let's line up the horses on the track. I mean, Z differentiates itself, right? I mean, it's a different animal used to run that business, you know it well. So thinking about, obviously you got x86, you got power, we got spark hanging around. I've heard that. I've heard about that. Getting injection of capital from Japanese government, okay, so Mills predicted it was dead, but the Japanese government changed that a little bit. But so you got the horses there, the big players. What's the differentiation in power? I wonder if you could talk about that. Yeah, one of the things we see, in terms of how does power differentiate in the market? And it's a great question. And the good news is we look at the history of power which is power was really born to do data from the day it arrived, right? But that was sort of the data we knew of 20 some years ago. Very much structured, relational database-based deployment. Now if you look at the different types of data coming to market, no SQL, key value stores, right? Hadoop-based deployments, all the rest of them, they're all different types of data. But once again, the value prop of power for data-centric applications remains really key. And so as part of the strategy for power of changing the dialogue from speeds and feeds to really addressing clients' data problems on-premise and off-premise, this data-centric role is where I'm targeting the power platform. I'm not trying to get too distracted or not trying to be all things Intel tries to be. I'm really focused on that rich set of data-centric applications. Now, given open power, you got to be more open about sort of the roadmap. Can you talk a little bit about power eight, the roadmap, what observers should expect? Yeah, absolutely. So what we've talked about publicly already, so publicly available knowledge, we recently, a couple months back, did a disclosure of the power eight processor at the Hot Chips Conference. And what we said there from a timeline is, very exciting technology. Clearly, I think it seems great articles on that. But at the same time from a timeline, we said clients should start to expect power eight in the middle of this year. So that's when the first systems of power eight will come to the market. We'll be bringing power eight for what I call scale-off deployments, one and two socket systems first, followed later in the year by the scale-off platforms that are four sockets and above. Okay, what else should we know about power eight? Well, I think it was really designed for big data from the way in which the processor was designed. So again, if you sort of break down the big data needs, it starts with running the calculations. You need lots of processors, lots of cores, lots of threads. Exploiting that. Yeah, I mean, you got to have the bandwidth of the process of the compute to run the calculations, right? We're in the equation. At the same time, big data needs lots of memory. And so as we look at the memory footprint for power eight, it's a pretty phenomenal improvement from where we've been, as well as versus the marketplace today. And then the same as on IO. Data's never sort of always at rest. Data's in motion. And so you need an IO bandwidth. You need that ability to move data in and out of that working set in memory so that you're dealing with this sort of live flow of data. And then you still need that redundancy, that security, because nobody wants to be the one who has to send the letter saying, sorry, I lost your data. And power's really been designed around those principles. So I've got two other questions, and I'll flip it over to John. So power in cloud, you've got some announcements with software. Talk about that. And the other is, what's the play in integrated systems? So start with cloud. Yeah, so start with cloud. So just today, we announced here at Pulse that we were moving power into the software cloud. We were starting with Watson as IBM announced the Watson Group here about a month ago. One of the key plays was moving Watson to the cloud. And so as part of the move of Watson to the cloud, since Watson runs on power, that's how it delivers its cognitive capability. It just made perfect sense, of course then, to move Watson on power to the software cloud. That'll be the first deployment in 2Q this year, next quarter. And that'll then sort of scale out to as we move BlueMix, which we've talked about here as well. Some of the DB2, DB2 Blue capability, Cogniz capability, those set of middleware services. Once again, there's great value running that data-centric applications on power in the software cloud. And eventually, we'll get to later this year, bare metal virtualization. And okay, so that means a swipe the credit card self-service type of model to get access to power, or is it a workload sort of paradigm that I'm looking at? The first couple, it'll be more workload-centric. Clients will see, obviously, with Watson, they may not even know it's running on power because you're requiring Watson as a service, right? Same thing when you get to the API marketplace model with information management products. Once again, better economic answer of choosing those to run on power in the software cloud, followed by bare metal, which obviously is what it is, it's bare metal. And what about the play in integrated systems? Yeah, so integrated systems around pure flex, pure scale. Obviously, listen, I've got clients who buy pure flex today with an all-power deployment. I just brought the market in, I think it was third quarter last year, a whole new set of power seven plus ITEs for that. So integrated systems remains important to me. It's clearly a market need. Clients are looking for that speed of deployment of infrastructure and the ability to acquire in a simple purchasing model, switching networking with the switching, compute, storage, all well-managed. That's an important market and we're going to continue to serve that market. So I was going to ask about the software announcements that you drilled into it, but I would like to drill down on Watson because obviously Watson, we're big fans of, it humanizes the big data piece. Talk about the Watson discovery. What is Watson evolving into with your group and how does that fit into the cloud? Just do a little deeper dive on that. Yeah, and in all fairness, John, I am not the Watson expert. So, you know, but if you look at what Watson does from a cognitive era, and you know, we've seen this sort of evolve over the last couple of years, right? It really is about deep query analytics, you know, it translated into, that's lots of really complicated algorithms to as you build up the data set and the knowledge and you have to teach Watson, right? At the same time, it's about natural language, the same too, to be able to interact in a human-like style based on speech. Again, the engine of innovation underneath that from the time it started in research to the step we saw it in jeopardy to now where we see it today, it's continuing to elaborate the power technology road map. So we've seen that evolve from power seven to power seven plus and now as soon as we get into the software to power eight. So talk about the cognitive piece because one of the things Dave and I always, we love talking about some of the new big data stuff, especially when it starts to get into machine learning, you know, reasoning, learning machines, all that great stuff. Cognitive computing. Right, right. Right in your wheelhouse, right? So, that's the pressure with mobile first is there's a lot of cognitive overload. You got to simplify. It's really hard not to crack and some people can't do it. So how does that trend? Does that help you? Does that trend your friend? Is that going to help you guys? It's absolutely my friend. In fact, we have a, I think a great demo here at Pulse that the one could go see and it takes an application from a company called MD Byline, which is about sort of the way in which one acquires medical necessary devices. But at the same time, because it's running on Watson on power, while you're sort of going through that purchasing sequence, looking for what you think might solve your health needs, you can then also ask Watson questions. Let's face it. I am not a medical expert. And so trying to figure out this medical language and how I understand it as a person, it's a great intersect from a human level as well as from providing the right economics for purchasing of health care equipment. So great. It's a great demo. I hope people go see it. Doug, thanks for coming on theCUBE. Really appreciate it. Got a break. This is theCUBE. We're live at the IBM Premier Cloud Conference, IBM Pulse. This is exclusive coverage from SiliconANG. It's theCUBE. We'll be right back after this short break.