 We're back here live at the Open Compute Summit in Silicon Valley in San Jose, heart of Silicon Valley. This is where the innovation is happening. This is the future of the data center, the future of the cloud. This is theCUBE, our flagship program. We go out to the events, extract a signal from the noise. I'm John Furrier, founder of SiliconANGLE. I'm with my co-host Dave Vellante, co-founder of Wikibon.org. And our next guest is Gary Ornstein, CMO of Fusion.io, CUBE alum, innovator. Eight times, zillion times, back again, to break it down. Challenging Pat Gelsinger's record. Break it down, Gary, because what's happening, we were on our crowd chat earlier, you're on Twitter, we're all talking, watching the keynotes. I was comparing this to be like, if the homebrew computer club is modern, this is what they would be doing. Trying to get the components together, get more performance, get that holy grail, which is essentially the future, which is cloud, right? The application tsunami we're seeing Facebook and what you guys call hyperscale, has set the standard. They built an infrastructure for themselves, and that's what Bill Atkins said about the Mac 30th birthday party. We built the Mac for ourselves, and that changed the entire world from that point forward. Obviously, it's legendary Steve Jobs and his team did that. Are we in that Mac moment right now where they're building the world for themselves? It's a great time to be participating at OCP, every year it keeps getting bigger. The fun part now, I think is the, as the ecosystem keeps growing, it's becoming easier and easier for people to participate. Participate by contributing ideas or designs or participate by acquiring OCP technologies that can be easily deployed in their data center. Frank Frankovsky spent several minutes this morning in his keynote talking about the solution providers who are taking the concepts of OCP and making them digestible and easy to consume for customers. Today at Fusion IO, we issued some news with two of these solutions providers that we're working with. Hive Solutions is one, Quanta Computer is another, and as an example, from Hive or Quanta, you can buy fully integrated OCP solutions that include Fusion IO flash memory. So this is just one example of the ways that it's, in my opinion, really working its way into the mainstream. Today we also saw new customers coming in to participate in OCP, so box.com. Now joining the roster, they talked about Riot Games, Goldman Sachs, Bloomberg, Orange, others. But why is all the excitement? I mean, I want to really get into this because I think at a root level, there's a big picture around, you have a lot of distinct parties here, diverse parties, motivated around the trend. You got OpenStack kind of on one side, you got the hardware geeks on the other. JR Rivers from Cumulus, which essentially kind of like VMware, Software Define, or Linux Networking. So the representative from Amax, Julia Shea, I think at one point in her talk, said anybody who has a few racks of servers is a great candidate for OCP. And what's the thing that keeps anybody who has a few racks of servers together is how do I squeeze efficiency out of that? You can talk about how many servers do I have or how many software licenses do I have or should I have more, should I have less? Everybody wants, maybe besides the real estate companies, everybody wants a smaller data center to be more efficient and to be more effective, serve more compute power with less equipment and less energy. And so I think what we're seeing now is that this mission of openness is really what has been defined as a great way to get to this level of efficiency. So to me, as the open compute mission expands, sure, open is great and is a wonderful vehicle for helping accelerate these things more quickly than they could ever happen before. But the payday, and we heard, I think it was Jay Parikh who said that Facebook over the duration estimates that they've saved $1.2 billion. It's the payday of the efficiency that is really, I think, gonna keep this flywheel turning more and more. Cole Crawford talked about the OCP's not pay to play. It's kind of reward to play. I wonder if you could talk about some of the contributions you guys, Fusion I.O., are making to the initiative. Absolutely, well last year at Open Compute, we announced our I.O. scale product, which is taking all of the core I.O. memory technology that Fusion I.O. has developed over the years with some of the world's largest customers and global data centers all across the planet. And we made that accessible to the world through the introduction of our I.O. scale product. We also contributed the design of that product to OCP, so that specification is available on the OCP website. And then between last year and this show, we also worked with OCP on a software project that we call OpenNVM, Open Non-Volatile Memory. And it's the first software project in Open Compute. It's also referenced from the Open Compute project page. And in that project are a collection of technologies designed to use Flash in a way that it can best be exploited from an efficiency point of view. How can we squeeze every last bit of efficiency and every last dollar of value out of the Flash memory? And so some examples of technologies that are in that OpenNVM software initiative include atomic rights for MySQL, which we've talked about and we're now working with a number of companies in the MySQL community to make that happen. Other technologies include a native key value store for Flash, which can be used as a plug-in for popular databases. And another piece of the technology equation in there is something that we call the advanced swap, basically the swap function for Linux, making that Flash aware. So all of these software initiatives as part of this OpenNVM umbrella share the same mission of we have this new media called Flash memory, it's a focus for FusionIO, and beyond just designing great performing and highly reliable hardware products, how can we use software all the way up and down the stack to make that more efficient? What's your take on the processor vendors? I mean, obviously you're in the middle of that. You have ARM, ARM is here, AMD is here, Intel is here. You have new players like IO here, just taking the world by storm. I mean, the keynote pretty much blew everybody away. And you guys sit right in the middle of all this action. Yeah. What's it like? I mean, what do you see around you? What, you know, we're trying to read the tea leaves here. What's next for the customer? Yeah, well, I think the great news for the customer is that as more people get involved with OpenCompute, it becomes easier for them to participate. So you don't necessarily have to be a server engineering expert in-house or have a server engineering expert in-house if you want to participate in all the goodness of OCP. You could call Hive Solutions, you could call Quanta and literally dial up a rack of, or multiple racks of servers with flash memory inside. On the processor equation, I do think it's fascinating that we're seeing the whole spectrum of solutions and of course the group hug initiative, you know, common socket interface for the motherboards, it's an OCP project. Regardless of the processor choice that customers want to use and they're all good and they all serve different reasons, the ability to use new processor technologies combined with new storage technologies, new flash memory technologies in a very efficient, easy to consume package being an OCP configuration. I think that's just goodness for all end users, whether they be hyperscale focused or enterprise focused. So I want to ask you about your quote on Twitter but first I want to get your comment to George Lechman's quote. He said, if people don't see open compute coming, they will be run over by it. As he was programming virtual machines running open stack on open compute on his iPhone on stage. Yeah, you know and the fun part about that comment is he goes through this whole online demo of spinning up virtual machine instances and then reminded everybody that there were no license fees in that process, which he was very proud of. Although JR River is from Cuma's point out that he's buying a boatload of networking boxes. This is the tax built in, not a license fee. I think most of us don't build our own house so it's good to get some people involved to help out. We're going to hold, we're going to just point that out to George and he's going to be like, oh, Farrier got me on that one. So that's interesting. So we do see this running over people if they don't wake up to this concept, this paradigm shift and then you made a comment on Twitter, you said that the hyperscale is bleeding over to the enterprise. Could you expand on those two points? What are people need to be looking for before they get run over to? What is hyperscale bleeding over to the enterprise mean? So I think the best analogy of hyperscale bleeding over to the enterprise is what we saw with Linux. When Linux came about, people looked at it with disdain, it's not mission critical, it'll never work in enterprise environments and lo and behold now, just about every major enterprise is multi-operating system environment and they have large windows installations, they have large Linux installations, they still have Unix installations, they still have mainframe installations, but Linux cemented its place as a significant deployment operating system for enterprises and that didn't happen overnight, it took a number of years. I think we're seeing similar things with hyperscale architectures, these scale out architectures where you add servers because that's the best way to do it for these environments. For example, when you're deploying MySQL, you're not necessarily paying a huge fee for an additional MySQL license, you might be paying to support that additional server but most of the MySQL deployments fit a scale out model and just like enterprises looked over their shoulder at some of the web companies deploying Linux and said, boy, that seems like a good idea, maybe not for everything that we do but certainly for some areas, I think we're seeing that now where enterprises are saying, gee, we're gonna keep a number of this or a fair portion of this compute power in-house, we'd like to lower our costs, we'd like to be more efficient, it seems like these folks know how to build large scale applications at a relative cost, let's get our feet wet in there and so I just think that we're gonna see more and more of that over time goes on as time goes on of enterprises adopting these sort of preconfigured OCP packages for certain workloads. So what mile marker word we both like to her choose but just take me through that because the inhibitors that Dave and I were talking about earlier on theCUBE here and also on our last event was, it's kind of complex for the enterprises, the regular enterprises, not the unique black swans or unicorns or whatever you want to call them, the ones that like Facebook that do their own thing, huge customers from themselves, the normal enterprise, guys who are, you know, rack and stack and storage drives, EMCs, NetApps, et cetera, you know, these guys have been dealing with the operating plan of the past 20 years. So what do they need to do? Is it automation? What's the, what do you guys? Well, I think the place for them to start is to call some of these solution providers that are focusing on building and selling OCP platforms. So Hive Solutions is a great one, Quant is a great one. There are many others that are mentioned today and these providers are very excited and very capable to work with customers and help them do the deployments for, you know, popular applications that suit the OCP architecture. So I don't think that was the case a couple of years ago. I think a couple of years ago, you did have to have some of the server experts and server automation experts in-house. You know, one of the items that we heard today, helping with the automation is Microsoft, including their contributions, not only some of the designs that they've worked on from a hardware perspective, but some of the software tools that they've developed to manage their multi-billion people that they service with a variety of cloud computing architectures. So they're still, I'm sure, is work to be done at that management layer to make the consumption easier. I think we've certainly crossed a threshold of making the acquisition of open compute hardware and software solutions. That part is done and now it'll only get better for customers. I saw Insook Ray earlier, he's a VC at Rembrandt Ventures. Also was one of the guys at Loud Cloud with Andreessen and Horowitz, big cloud guy. He's probably going to be pissed at me for calling him out like that because now the VC doesn't know he's farming for good deals, hunting and farming for deals. But I saw him walk with a big-time financial services company. Won't say their name. I want to ask you the same kind of question as the adoption. I mean, obviously the financial services guys are in that same boat as Facebook where they got to do their own thing. I mean, they have to innovate. The innovation strategies for these companies are significantly peaked on IO. They're peaked on latency. They're peaked on cost structures as well. I mean, we saw Facebook power. I mean, Facebook just produced the data that says they saved $1.2 billion in savings over three years. I mean, that's not chump change. That's significant bank, right? That's big go, right? So these financial services, what are you seeing in those early adopter profile customers? So we are seeing continued interest in enterprise customers to how we can help them do more with less infrastructure. Either capture more data, process more data, query more data, present more analytics. We've heard a lot of talk today about the new applications. It was the gentleman, I believe, from Merck, Mr. Lyons. Yeah, who said one of the things that they're spending a lot of time looking at is what are the new analytical models that we'll be able to do to move drug discovery and other things in healthcare initiative forward and recognizing that that will take a significant amount of compute power. Now we need to know, okay, if we're going to do that with a significant amount of compute power, how do we keep our costs down? How do we keep our energy consumption down? An open compute is on the leading edge of that. Jay from Facebook mentioned earlier that at times they're seeing a 1.04 power efficiency rating on their data center, which really is unheard of. The white paper that Facebook released last year on efficiency details that, but it also covers what they're doing to shrink the workload on the, how much data the apps consume, but also shrink the amount of infrastructure. So when it gets back to what's next for these enterprises, they're going to look and see, wow, if a company like Facebook can get a 1.04 PUE number and the industry average is significantly higher than that, well, they can't sit on the sidelines for long, otherwise they'll be run over as we learn here today. There are headwinds. I mean, the average enterprise app is almost two decades old, but it seemed like it might take away from the panel that we did this morning in Jim Lyons. It was like, you know, we're not going to worry about those apps. We're going to focus on these new workloads like analytics and that's going to go there and eventually it's just going to tip the scales. Exactly. Do you see it that way? No, I think certainly enterprises will look to, if they're going to deploy Open Compute, I don't think it's a matter of ripping out a whole bunch of prior infrastructure that was running classic apps like Oracle and SQL Server and porting those over to the new Open Compute platform, but they might take a look and say, hey, we're going to use some of these new data stores like MongoDB or Cassandra or Hadoop and run some analytics there and since we're starting that one as a green-filled opportunity, let's look at these new hardware platforms and maybe that those initial applications start more of in an offline capacity for reporting and analytics as opposed to a real-time capacity and they can work their way there and there's a whole bunch of options that I think work well for enterprise customers that they can work Open Compute hardware and software combinations into their critical path step-by-step. Well, it's interesting because your application space is pretty broad. I mean, you're extending the life of a lot of Oracle apps. At the same time, you've always been a player in the hyperscale business and then hyperscale as we talked about bleeds into the enterprise. So my question is, are we going to sort of hit an equilibrium? Will there be that tipping point? What do you think? Well, if you look at it in terms of the number of servers deployed, obviously the customer concentration count, if you include some of the big players who are up on this, there's a smaller number of players playing a huge volume of servers. So I think now what will happen is we'll see more and more customers that can get started because there are options available from service providers. Going back to Julia's comment, if you have a rack or two or three of servers, you should consider OCP. Doesn't mean you have to flip overnight and do it, but that's about the scale that we're talking about now where it could make sense for people given that you can dial up a rack of OCP servers from one of these trusted solution providers. What do you think about the Microsoft news, Gary? I want to get your perspective on that. There was some commentary going on Twitter. Obviously they're donating some, some spack and it's just licensing it. So real honest effort by Microsoft to be involved. I mean, as Dave and I were talking privately before we came on, it's like, you know, you either want to be at the table rather than on the menu. You know what I'm saying? As the expression goes. So that's really what it's all about. I mean, Microsoft has a seat at the table at something big and that's not their Forte hardware, but they've spent money on CapEx, OpEx for their infrastructure. And Amazon's clearly not sharing it's server to server. Yeah, there are a few companies that are absent. Good move from Microsoft, a seat at the table. What's your take on this? I think it's a great move from Microsoft and they do some hardware very, very well. Xbox as an example. I don't happen to have one, but. I do. Two. There you go. So you know how well they do hardware. You know, I saw, I just saw that news today and I sat through the Microsoft President. I think it's phenomenal. I think Microsoft brings a set of capabilities to this group that didn't exist prior, which is how do we scale the adoption of technologies to Main Street USA? I think we could safely say that Microsoft has made as an example, being able to run a fairly sophisticated database, an accessible technology through Microsoft SQL Server. They're doing very cool things now with SQL Server that allow you to do always on configurations and do multiple instances of SQL Server on a single machine and high availability tools built in where you don't need a separate storage area network. In fact, we have customers who are doing exactly that. They're taking Microsoft SQL Server and running many instances on a single machine and using the Microsoft HA that's built in. Now there's no reason you couldn't do that on an open compute platform and it might be exactly the right thing for a certain customer set to marry the highly efficient, scalable hardware designs with the very familiar and battle tested Microsoft solutions. And then I think we mentioned briefly the point of the Microsoft announcement that I found particularly interesting is helping with some of the automation. So if we expect the average administrator around the world to be able to deploy these technologies, hardware and software included, we have to give them the tools to be able to manage dozens or potentially hundreds, maybe even at some scale, thousands of servers at a time. And if you look at what Microsoft does today in terms of standing up services with Azure, with Bing, with Skype, with the rest of them, they know how to do that. There's no skin off their Apple, so to speak, unintended to give them, give the hardware up because that's not their core competency. And but they have that core competency from an op standpoint. And so what's their software angle? Is it going to be just classic enterprise software? Do you have any insight into that? Well, one of the things that they talk about just generally in the public is, I pay attention to their messaging around Azure and that they want to deliver a set of technologies that customers could deploy on premise should they want to do that. And they also want to have a similar set of tools available from a cloud services perspective with Azure. They want to be able to enable service providers to fill a gap there as well. So one group of technology, I think they use the Microsoft One term to talk about technologies available for customer deployment, technologies available for service provider deployment and technologies that Microsoft might make available on its own as a SaaS service. And I just think that's a winning combination. You could carry it even further to say, wouldn't it be great if I as a customer was deploying a specified OCP configuration, Microsoft Azure was deploying the same specified OCP configuration. And I could know exactly what's happening with my SQL server instance locally and should I decide that I want to move that because the equipment is the same and the specs are the same, I know exactly what I'm going to get on the other side of the wire. Whether or not people do that is their choice, but I think that confidence level that can be delivered to the world through those kind of discussions is impressive in something that we haven't had previously. So you've seen some, Gary, you've seen some interesting and I wonder if I get your opinion on this interesting shifts in the server business, which for many years was kind of boring. Now you have the rise of the ODMs I mentioned today, the Matt Eastwood data that I think 7% of the revenue was coming from ODMs, 12% of the volume. You saw IBM get out of the x86 business, which a lot of people said, okay, yeah, that's nice. It'll get IBM out of the low margin business, but it represents quite a sea change in the dynamics of the server space. So what do you make of that shift that's going on? Help us squint through that. Well, I find it very exciting that we're seeing new server architectures come to market that are getting tremendous adoption. Those are some of them coming from ODMs overseas, directly some of them coming through the Open Compute Foundation. It's certainly causing a shakeup in the industry. Now we know that shakeups don't happen overnight and I don't anticipate some immediate overnight flip where everybody in every Fortune 5000 company is going to be changing their server acquisition strategies. But I do think it represents an opportunity for customers to have more choice for how they want to pursue the scale out of compute power. And I think it was Andrew Feldman who said today that all the interesting stuff Andrew from AMD was, the founder of C-Micro, which AMD acquired, he said all the cool stuff happening these days with technology has a data center component. And so I'd say that for most U.S. businesses around the world is not a U.S. specific comment, but for most global businesses, they're going to have to figure out to what degree they want technology to be a part of that strategy. And if so, maybe take a hard look at what they've done historically for server acquisition and does that need to change if they're going to need to scale their infrastructure to a certain point. Gary, I want to talk about culture. This show here represents, to me, a flash point of cultural intersection between hardware geeks, which have not been getting a lot of mainstream love lately from the press, but us industry nerds know that they make it all happen. And couldn't even highlight even better than the 30th anniversary of the Apple Macintosh this past Saturday to DevOps, right? So you having this, and we even had cumulus on JR rivers, talking about bare metals back. We never left the building, so to speak, as we said. You have it all kind of coming together and you got IO shipping containers of data centers, IO cloud, you guys are innovating inside everything. So this culture of DevOps is truly taking hold and it's going mainstream. So the question is, talk about the culture and is DevOps the same as, say, CloudOps? Because DevOps guys are eating glass, spitting out nails and they're almost like Navy SEALs in a way. All special forces of developers. Can that go mainstream? Or as it gets more neutralized? Well, I don't know if it gets neutralized, but I think there'll be different levels of DevOps and how you manage servers at scale and scale being number of servers and quantity of servers. So there's plenty of solutions that have been around for a long time. CF Engine, of course we have Puppet and Chef, which are wonderful automation tools. Then there are tools, sort of cloud in a box tools that do that all for you and you could carry that all the way to what may come from Microsoft is, you know, what is the simplified interface for managing 100 servers at a time. All of those things are complimentary and don't necessarily negate one another. And I do believe that once we, we're now at a point where we've crossed the barrier in terms of making the OCP platforms easy to understand and easy to acquire, the next big step will be, can we make them equally easy to manage? Gary, how about, I want to forget to get a new title, CMO. So we'd like to ask the brand question. The evolution of the Fusion IO brand has been pretty interesting. I remember 2009, John, we're at one of these old, I think it was storage networking world or something, thinking that's pretty cool. And right around that time, EMC had announced the flash in the box kind of approach and wow, it would come quite a long way. But you guys pioneered that memory extension, obviously drove a lot of the hyperscale innovation bleeding into the enterprise. Where do you see taking the Fusion IO brand going forward? What do you want it to stand for? So at the big picture, we want to help customers scale their applications and application acceleration. Our mission is to allow customers to run their applications faster with less infrastructure. And we do that across a variety of applications, could be database applications like Oracle, SQL Server, MySQL, SAP HANA. We do it in virtualization environments like VMware and Hyper-V. We do it with VDI. And we do it with big data applications like MongoDB and Cassandra. All of these applications are mainstream in different areas of the market. And because flash is a disruptive technology, we felt it very important that we take the time to educate our customers on the options that are available to them to deploy flash and then work with our server OEM partners to make the deployment easy. So again, at a high level, it's about how do we make customers infrastructure more efficient, faster applications, less infrastructure, a variety of applications and then do it in conjunction with server partners who can deliver the technology in a one-stop shop to our customers. The other thing that's expanded over the last couple of years with Fusion IO is that customers want to deploy different flash in different places and that's great. So some customers might want to deploy flash in the server and that's a place that Fusion IO pioneered with our PCIe-based flash memory products. Other customers might want a storage-centric approach and we have those options through our IO control, hybrid array and our ion data accelerator, all flash array. So those things make it easy now for customers to say, where should I deploy flash? Work with Fusion IO to say, how should I deploy it? And the options are all theirs. The bottom line for customers is the ability to increase the application performance and significantly reduce the cost. Okay, so that's a ambitious sort of mission statement if you will, not that that was a mission statement, but you laid out a lot of things for a lot of people. So what gives you confidence that you can sort of hit those marks? How are you feeling about where you guys are at today and going forward, how are you positioned? So this is a market that Fusion IO really jump-started in 2007, we launched the technology in 2008, we started shipping products. It was a rocket ride over the last few years and we're now participating in some of the largest deployments of flash globally in both hyperscale companies and enterprise companies. But if you look at the spectrum of how these technologies are adopted, I still believe that we're in the early stages. When you go around to these kind of shows, we, maybe OCP is probably a little bit on the leading edge of some of these, but at your average trade show, we still find far more people who have not deployed flash in a significant way than those who have deployed flash in a significant way. And it's one of the things that gets me so excited about participating in both the OCP community, but also the larger data center infrastructure community and the application acceleration community is the opportunity we have to work with customers to make their applications run faster on less infrastructure. And so that is a multi-year effort. There are still many, many untouched areas, untouched industries. We have a few industries where I think we can claim they've been totally invented or reinvented. I think hyperscale and social media wouldn't exist without flash memory. I think that online advertising is constantly being pointed to as a leading edge for technology adoption and flash, streaming music. We recently last year announced Spotify as a Fusion IO customer, but there are so many industries that haven't deployed or don't have a company in that industry that has taken a leap to say I'm going to an all-flash infrastructure. I just think there's a lot of room ahead. So Gary, I want to ask you the final question. Your take on open compute, where do you see it going? I'll say a lot of groping for market position at the same time, it still cleans you to paper if you think about it. You see some vendors here, unlike last year, you got booths here, you have some folks here. What's your take on that? I mean, how do you see the community evolve when we heard from Colin earlier? He's put it right up front. We're transparent, we're lean, we're mean, we're not going to be driven by the dollar, we're driven by contribution, which is great to hear. How do you see all this evolving? You know, in the span of just a couple of years, this organization has gone from an idea to a movement and you can sense it in the comments from the speakers and the reaction from the crowd that there is a collective sense of purpose and once you have people coming together with a common sense of purpose, I think it does lead to some pretty spectacular things. Couple of years ago, you really had to be a server expert and know every last bit of detail if you wanted to participate in OCP. Now you can call a one-stop shop like Hive Solutions or Quanta and get a full or multiple OCP racks delivered to your doorstep. I think what we'll see now is more participants, both on the supplier side and the customer side, more designs. The Frank mentioned today, the certification program where they want to certify specific configurations. I didn't get all the details there, but wouldn't it be interesting if those actually specify the application workload as well? So you could know that out of the box I'm going to get this well orchestrated and tuned MySQL configuration across 16 or 32 nodes and I can spin it up and it'll have the management capabilities right there to just be off and running with a high-powered MySQL database. So those things are not far away. It's just going to take the collective energy of this already well-fueled group to make that happen. And I think if I were to make a prediction about next year, watch the OCP configurations get more, they'll always remain open, but be more integrated and more tied and specific to application workloads that ultimately make it easier for a customer to say, I want to deploy that. That's that notion, John, of that single managed entity that Floyer was talking about, the SME and then Gary, you're talking about solutions that also address workload specificity, which I think is compulsory for success here. We always like to ask executives that come on theCUBE, especially alumni, secret questions at the end that are unprepared for, not that we prepare any questions in advance, the teleprompter is broken again for the fourth year here in theCUBE. Smashed. No teleprompters, that's an inside joke. Gary, I got to ask you your opinion on Bitcoin. Oh, I'm trying to get up to speed on it, but I've been so busy preparing for all this stuff with OCP. I think it's fascinating and I still don't understand all the intricacies of it, but to me, anytime you can take friction out of commerce and friction out of payments, it generally leads to more activity and hopefully positive activity. The thing that's easier for me to understand is Square, which really has made payments frictionless. You get home, take a taxi late at night, they send you the receipt, the whole thing's done, you don't have to do anything. And I think we're gonna see Bitcoin become a platform on which people can build some of these very seamless services for payments where the payments process and the payments transaction becomes invisible, which will only lead to more payments. They're building on OCP, right? Well, there's a lot of inside baseball. The early Bitcoin miners were guys who had these big exchange hosters who were mining early and often made a boatload of cash. So Bitcoin is not a foreign entity to a lot of these big OCP players. You talk to any of the guys who know about infrastructure and networking, they can know they probably arbitrage a little bit about Bitcoin. But Mark Andreessen wrote a great article in the New York Times, which essentially mirrored my editorial from weeks earlier. Oh, I did a video post on Silicon Angle. And it was on the same thread. This is about a future paradigm, not so much the nuances of Bitcoin. I talked to some folks in some big banks. It's politically incorrect to answer, no, we will not accept Bitcoin. Everybody, I mean, everyone is saying we will, we're looking at Bitcoin. It's kind of like a venture capitalist. They'll never say no, but they don't always say yes. So Bitcoin is still yet to be judged, but more and more evidence is coming down. I always said it's all about the clearing house. Who's going to be holding the, are you sitting in a chair when the music stops? Right. That's what everyone wants to know about Bitcoin. Is it a bubble? Who's going to be holding the hot potato when the music stops? That's always, always case Gary. Great to have you back. Thanks for having us. Thanks for having us. Great to be here. CMO Fusion IO continue on a toward pace. Guys, great success story in the industry here in the center of all the action for Open Compute Summit. This is a technology revolution, changing the game for the next generation. This is theCUBE. We'll be right back with our next guest after this short break.