 Blanket, the HP Moonshot Announcement, a disruption in the data center, a revolution in servers, a game changer in this new modern era of software-led infrastructure, big data, mobile apps, cloud mobile social. I'm John Furrier, the founder of SiliconANGLE.com, and I'm joined by my co-host. I'm Dave Vellante of Wikibon.org, and Carl Freund is here. We've had Carl on before, and November, Carl, in 2011, you were on. Carl's the Vice President of Marketing at CalZeta. We were here at the original Moonshot. Actually, we were out in front of Bill and Dave's office, and that's a welcome back. And lots been going on since then. CalZeta's been out finding new markets, so give us an update. Yeah, it's been really exciting. I mean, that was a momentous event, sitting outside of Bill and Dave's office, and it was also a momentous event just because of what HP stood up to do, which was to really change the way servers are designed, the way the software is envisioned for these very efficient new processors that are just coming onto the market. Since then, we've been working very closely with HP in the Discovery Center and with customers to really understand what these new arm-based SOCs are good for in the data center. Real processors that can significantly reduce power, significantly save money and acquisition costs, and reduce the amount of footprint and space in CO2 emitted in the atmosphere to run the data centers needed for internet of things, for social media, and for the explosion of big data that everyone's experiencing. Bottom line is people are starting to realize those benefits. Still early days, with the HP Moonshot system being announced today, we look forward to getting these out in larger quantity and to more customers around the world who have both environmental power and space and budgetary constraints. So you mentioned SOC stands for System on a Chip. What makes it a system on a chip? That's a good question. So traditionally, when you design a server, any kind of computer, you start with a processor around which you add other chips, sort of a core electronics complex, as it's often referred to, and each one of those chips consumes power, each one takes space, each one costs money. And SOC takes those chips, both digital and analog, and puts them on a single chip, and that's really the uniqueness, is the combining of digital and analog circuitry of an entire server, which normally takes a big board like this and consents it down to a single chip. And what that allows us to do with the SOC is to literally put four of these on a board and be able to say, this is not a server, this is four servers on space that's less than you would normally have for a single server. Now, this doesn't have the same performance characteristics, but it has much better power characteristics. So for certain workloads, a processor like this could save a lot of money, save a lot of space and a lot of power. Carl, we were at HP's at labs, we were right in front of Bill and Dave's office last time when Moonshot was first launched, and we were blown away, because obviously we were covering the big data market and we saw the need for massive scale out industry standard or commodity servers, whatever people want to call it, but really proliferation of servers, but what that does is it brings up the issue of utilization. So when we talk to some of our friends and customers of you guys in HP, they don't have a utilization problem, you have massive scale out open source environments. So having this ARM-based architecture really makes a difference. So can you elaborate on that specific use case and then also the ecosystem around this massive sea change of servers? Because this is a significant change over when you're talking about server functionality at a high level of compute and this utilization issue where you can have all this massive scale, but then the utilization everything's not 100% utilized, so there's a lot of management involved. So talk about that and then the ecosystem. No, it really starts with software. I mean, the hardware gets all the headlines because there's shiny objects, but the fact of the matter is it starts with a software revolution. This has been going on for a long time, over 10 years, where software can be written to take advantage of a scale out infrastructure that's been enabled by better networking technology and so forth, but it is really the software that changes things. Whether you're talking about software-defined storage, software-defined networking, or software-defined servers, it's all about having software that can scale out very efficiently, avoiding the so-called Amdahl's law of being bottleneck in setting up the scale out, okay? And so this was all pioneered by the Googles and Facebooks of the world, but is finding its way into really data center. Let's drill down on that because this software is kind of an ambiguous term now because most people think of software engineering. You go to school before it's computer science, but also there's a lot of double E, electrical engineering software that goes on at the chip set level. So let's take us through some of the use cases that you see that's enabling because if what I see happening here, what you're saying is this new innovation layer at the low level of the stack is going to enable software paradigms on top of it. We're seeing it already with software as a service, but in the cloud game, it's platform as a service and infrastructure as a service. You see open stack is getting a lot of traction. Can you explain that software dynamic and why it's different now? So I think what's different is that you're dealing with code that is built on interpretive languages like PHP that is extremely efficient and scales out very, very well. So it doesn't assume it's got a big vertical stack. It's not bound to an OS. It's not bound to an OS. It's not even bound to a processor. And so it's very portable. It's very efficient and more importantly, perhaps it makes the programmers more efficient, right? So if you look at PHP code, it just almost reads like English. And that's the whole concept is to make it simple for the programmers to deliver new innovation. Now it just so happens that those programming models are very efficient at scaling out. So the current model is you scale up with virtualization. So you take your virtualization layer. I'm gonna put lots and lots and lots of images of these PHP or Java or whatever you're running. And I'm gonna stack them up on a virtualized processor. Problem with that is several, full. One is that you're paying a, what's called a virtualization tax, right? Secondly, you've got a pretty big node there. And if it fails, your failure domain's quite large. So in terms of the user impact of failure of a processor or a storage device, it could impact hundreds of thousands of people. Whereas if I have a smaller processor, a finer-grained compute element, then the failure domain can be quite small. You can imagine the impact on this on something like cable on demand, movies on demand, or you can imagine this things like websites or analytics on Wall Street, where if one little element of the analysis fails, it's not a problem because it's a small impact. So failure domain as well as the efficiency of the programmer, I think what's really driving this attractiveness of the scale out infrastructure. And when you look at what the web scale folks have done, people call the hyperscale space, they really, they've tried to eliminate almost completely human involvement when something fails. They just sort of ignore it and keep going. And then eventually every component in the rack fails and they just throw the whole darn thing out, right? So that's a completely different mindset. But the specific question I have is, so originally we were thinking, okay, this class of system is gonna go into that hyperscale space. And it clearly is. I know you've got probably 100 plus systems out there, but what else have you found in terms of other potential markets for your product? You know, one of them caught us by surprise. And some people told us, you know, you really ought to look at storage. And we're like, eh, nah, I don't think so. We really focused on the web tier in terms of running applications and websites. And you didn't think so because storage is spinning disk and it's not this consuming power. Exactly. But it's not a power playing storage. Exactly, exactly. So you look at how much power the spindle consumes and you look at how much, you could take the power of the processor to zero and nobody's gonna care. Because you've got an entire rack of, you know, petabytes of spindles. Okay, so why the interest? And so what's happening is that the, by moving processing power out of the server and actually embedding it in the storage array, you're gonna be able to have a much more efficient platform than buying a traditional NAS or SAN. NAS or SAN is basically 10 and 100 time, respectively, more expensive than a spindle. So if it costs $500 for a spindle, it costs $5,000 for a SAN and it costs NAS and it costs, you know, $50,000 for a SAN. It's expensive. It's gonna roll a thumb there. And so building them yourself, sort of like do-it-yourself NAS, for these hyperscale guys can make a lot of sense. And what we're finding is that because our chip has five SATA ports, it's kind of geeky here for a minute, but it's got five channels to disks and it's got five 10 gigabit links to the network, it's basically a big, fat pipe to all that data you're storing on your disk drives, whether it's family photos, whether it's video on demand, whether it's imagery or seismic data, whatever it is, data you probably write once and read periodically. And so what you need is a big, fat pipe to write and that's what our chip provides. Okay, so it's not the power, it's the ability to get data on fast and the bit bucket, as they call it. It's really sort of a one-way trip to the bit bucket and if you're gonna need to get it back, you go get it back. Carl, talk about the folks out there who aren't in the weeds on the technology side who are just general observers of our program. What this really means to the marketplace, I mean, there's a lot of them, and I was just commenting last week about the Goldman Sachs analyst who downgraded HP, kind of called him out on Twitter, kind of didn't, kind of said he didn't really know what he was talking about, because he really downgraded him because of the shipments of the PCs and the sort of Windows 8, the normal stuff that, you know, the one level deep, but this announcement really speaks to an arc that goes out five, 10 years, maybe decades in terms of a seed change. It's kind of an inflection point, as Dave Donatelli pointed out. So can you explain to the folks that are out there that you might see at a cocktail party or general people who aren't in the weeds technically, what this means? You bet. I mean, think about how many of you had a basically a mini supercomputer in your purse or your wallet five years ago? Nobody did, but if you look at what's available now from a Samsung, HTC, or Apple today, those are amazing computing devices and they're generating a lot of data. Now take that to the next level and look at your automobile and all the information that is being collected by the sensors around your automobile. Now, connect everything together, including sensors on the highway, sensors in your car, sensors in your phone, all of that data is incredibly useful and can really help the world be a better place to live, quite frankly. But also the power requirement too. I mean, think about like, you don't want to have, you know, timeless power being generated. Well, and what's happening is all that data has to reside someplace. All the analysis of data has to be done someplace and so the demand, while the revolution appears to be happening in the mobile space and in the internet of things, it's causing a revolution in the data center because you've got to deal with a lot of data. Where are you going to put it? How are you going to, how are you going to have to get in the car? Which is storage, right? You got to rack and stack. The old way was just build some, you know, new racks. They'll do racks. Well, guess what? You run out of space. You run out of power. You run out of money. So the revolution on the mobile side is causing a revolution, has to cause a revolution in the data center. There's no choice. It's got to change. And that's what this is really all about. So when you think about Calzada and it's sort of mission, a lot of people just immediately go to, okay, it's going to completely change the existing x86 market. And now we've been talking about these new markets. So how do you see that shaking out? Is it the case where the new markets are sort of interesting and growing faster and then eventually these systems on a chip become more capable? Or do you see this as more just incremental? No, I think it's not incremental. It's pretty radical. And so as a result, it's probably not for your run of the mill application stack. You're not going to put Oracle Rack or IBM WebSphere on an ARM chip anytime soon. What will happen instead is you have these areas of the market that are growing extremely fast. Big data, media serving, web serving. And they are kind of green field and the code moves over pretty quick and easy. So that's where we'll see traction first. Meanwhile, as you say, the technology will mature and it will intercept the point of sufficient performance and software ecosystem maturity that it can become a viable platform for increasing number of applications. So it'll start out kind of nichey, right? And over time, those niches get big and then there's adjacent niches as we deliver more performance, as we deliver more memory, as we deliver more bandwidth. But when that happens, what you'll see happening is the software around it matures at the same pace. And so I think most industry analysts predict that by, let's say, 2016, this could be 20% of the server market, roughly. And there are plenty of ARM vendors out there that will be vying forward competing for that market each with their own unique approach. The ARM business model enables a lot of this innovation because you start with readily available intellectual property that allows you to build a specialized product, okay? So if you look at what TI's doing with their DSPs on ARM, if you look at what we're doing with our fabric switch on ARM, they're very different solutions targeting very different markets. And yet they're all built on the same common code base which is standard Linux, thanks to the compatibility of the ARM instruction set. Yeah, so part of your job is figuring out customer requirements. And as you said, some of the stuff is in Nitchie and there are a lot of different Nitches we were just talking about. I mean, obviously power is one theme, but as you just gave in the storage example, it really wasn't, power wasn't the driver. So what patterns are you seeing? And maybe it doesn't matter because you've got that common code base. I wonder if you could talk about that a little bit. Well, the main patterns right now are just experimentation, right? So universities around the world get access to our hardware and they find new things and potentially open new markets. So a good example is in the life sciences field where the genomic codes are exploding because genomic sequencing has become relatively inexpensive. You can go to your doctor now or a local clinic and have your DNA analyzed. And now the question is, now that I know what your sequence is, what does that tell me about your health or likelihood of health situations down the road? That analysis takes a lot of compute power and it turns out that compute power scales out really well. It's primarily integer based, not floating point, doesn't need a lot of memory. So it's a perfect example of a new kind of application that is a good fit for these small fine grain processing elements with ARM. How about the big data crowd? I mean, everybody wants to super glue their marketing to big data, but what makes your system, your philosophy perfect for big data? Or is it? It is. I think our second generation platform called Midway, which will be out later this year, that's probably the target for big data. Our current product, it's great for web serving, it's great for media serving, it's great for storage. It doesn't really have enough memory for big data in the large. So we'll be getting that second generation product out and that should be a really good big data platform. Why? Because it's low power, it's low cost, but it's also got a lot of IO. Big data is all about moving a lot of data around. And if you can do that more cost effectively, you're gonna save a lot of money. Yeah, okay, and maybe I said before you got 100 plus systems out there and sort of rough estimates, but can you give us an update on where CalZeta is at as a company, what kind of progress you're making, what you're shipping and when you're shipping? Yeah, we're Austin based. We do have offices now in China and in Silicon Valley. We're at about 115 people right now, mostly hardware and software engineers. We've had this product in market, it's been shipping in volume now for about six months and our second generation product is in the engineering labs now and we'll be shipping that out in the second half of the year. That'll basically double performance, increase memory capacity, bigger, better, faster, but in this case not bigger and not more power hungry. And then we have a roadmap that goes beyond there to our 64 bit product will be available in the first half of next year. So we're pretty much in the pack, in the middle of the pack of the 64 bit players. We do have the ARM A57 license. Our strategy is to use standard ARM cores, not create new cores, which may or may not be compatible with the ARM standard. And therefore we focus on everything but the core. We think the core is a commodity and so we focus on things like our fabric and our management, which helps reduce costs, not only of the processor, but really of the entire electronic complex and even the top of the rack. And so your products at the second half of this year and the 64 bit will ultimately find their way into moonshot cartridges, is that right? We're working closely with HP to make sure that the HP customers have access to the latest ARM technology from Calzada. Carl, my final question, we're coming up on the time limit here, but it really has to do with kind of the future of the business side of this, which is obviously success and penetration and actually putting it. I think it's going to be a big game changer. Both just on the energy side alone is a winner to the point you mentioned about telematics and what's going on around the world with data and obviously the footprint issues and performance. But in terms of the ecosystem, always when you've seen these seed changes from the early days, you had an ecosystem that's floating at all the boats underneath the technology. What's your vision and how do you see the ecosystem develop and what needs to happen for this to really go crazy and be successful? No, the ecosystem is primarily an open source. This is really about open systems. You look at things like open compute sponsored by Facebook. You look at the kind of software people are running on our platform today. It's all open source and people that can build on top of that open source platform are the ones that are changing the environment. This is not about replacing your current SAP stack. This is about hosting what's going to replace your SAP stack in the future. And so I think that's what has to happen here is that as people's applications move to the cloud, as the Internet of Thing is hosted, basically its data sits in the cloud, then it reduces the complexity of the overall and breadth of the overall software ecosystem to something that's more manageable and actually can be lifted up, put over on a new platform, in this case ARM. And the software paradigms are changing to support this new kind of revolutionist modern era that we're living in. Amazing announcement, I'm really bullish on it. I think it's going to be a boon for entrepreneurs out there as well as the ecosystem. I think it's early, not even the first inning, it's warm up. It's like spring training for this market. So Carl, thanks for coming on theCUBE. My pleasure. We'll be right back to SiliconANGLE Wikibon's exclusive coverage of HP's moonshot announcement on the ground, getting all the action. Here in New York City, Times Square, I'm John Furrier with Dave Vellante. We'll be right back with our next guest.