>> Announcer: Live from San Francisco, It's The Cube. Covering Pure Accelerate 2017. Brought to you by Pure Storage. >> Welcome back to Pier 70 in San Francisco, everybody. This is The Cube, the leader in live tech coverage. I'm Dave Vellante with Stu Miniman. Scott Dietzen is here, the CEO of Pure Storage, hot off the keynote. Scott, great to see you. >> Great to be back on The Cube. >> So I love the nickname. I grew up in a town where everybody had a nickname. We got Dietz, we got Hat, we got Danzig, we got Kicks, I dunno. You can call me V. He's, I guess, just S-tu. >> V works. >> I mean, that's it, you know. So, again, great show here, I love the venue. How'd you guys pick this place? >> So I can't say I was involved in the choice and this place has a really illustrious history. I mean, it goes back to the 1800's. And actually they manufactured steel here during World War II. I think they were turning out two battleships a week. But another piece of history that maybe isn't as nice is this is the last time this venue's going to be used. So it is scheduled to be brought down to make way for new condos I guess. So we really wanted to celebrate the venue and its history. It's just a great industrial feel to it. >> And they're tearing down a bunch, the new Warriors facility is going to be in Dogpatch, right? >> Yes, and so, yeah, we can't feel too bad about it because we are indeed celebrating the Warriors success. >> You needed a bigger house for all those trophies. (Scott laughs) >> I think they're poised to have a really good run. But I think Cleveland's going to be there contending with them for the next several years to come and it's really exciting. >> Well, hopefully my Celtics will get there in the next four or five years with some draft picks. So, I want to talk about sort of the ascendancy of Pure. When we first met you, you had a pretty simple message. It was like, look, we think we can deliver way better performance for lower cost. I mean, boom. It wasn't the same cost. I remember you were very forced. I said, "About the same, right?" You said, "No, no, lower. "We have the best data reduction technology "in the business." I remember talking to you at Oracle OpenWorld about that. >> Yep. >> And that's fundamentally what happened. And you attacked the legacy and stall base. And you won that game. But you're not resting on that, you've got to take it now to a next level. Talk about that next level. Well, talk about where you came from and then the next level of data and beyond just sort of public cloud. >> You guys have talked about this too, right. If you look at the curb of Moore's law. I mean, mechanical disk doesn't follow Moore's law. And so the cost reduction curbs, we did the math and we said, look, we're going to be able to drive down the cost of storage. We're going to be able to drive up the density and power cooling space. Simplicititly you can dramatically reduce the cost of storage. But Flash is going to help us, right? You know, we've gotten to the point where Flash is, you know, even with a tighter component market, it's cheaper to buy raw than fast disks. And way cheaper to deploy. World Bank talked about saving millions of dollars by deploying Pure Storage and getting a 5x performance boost at the same time. So if we can help customers pay for their storage both in terms of cost savings as well as new business value, that's a great outcome. >> Wikibon's been on the right side of that prediction since early on. >> That's very true, I've used your data. >> We're very aggressive about that. But the thing that excited us most was the second thing you said. Which was the business impact, the business value. So I want to come back a little bit and get a history. It used to be I would buy EMC for block and NetApp for file. You're sort of attacking that premise. Talk about that. >> Well, so we started in the performance end of the storage market, which is dominated by block. Because we knew that one was going to be the first to shift to all Flash. And we've already seen that play out. I mean, even the legacy vendors and their install base are inclined to use Flash. Cause it's actually cheaper than 15k disk to put in. That tech is about to hit a wall because as SSD's get bigger. You know, we've grown SSD's almost 400 fold since Pure got started. But we haven't changed the pipe, right? So if you make a vessel 400 times larger but you have the same pipe going in and out of it you're losing a lot of access to data. This is this new sea change to new protocols where we're shedding all of the disk. And I think the second big change is we're bringing the same wave to big data. Right, so we've been playing in the block market now we're playing in the file and object market. Because big data workloads, especially those that require deep learning, you just need massively parallel storage. And you're never going to be able to get that with, you know, 20-plus year old storage designs. >> So, Scott, when you talk to your customers, especially when you're talking to C-suite, how does storage fit into that discussion? I loved in the keynote, there's a lot of discussion of, you know, next generation applications. Everything from the, you know, buzzwords of the AI and ML type pieces out there. But, you know, what are the big challenges that your customer's facing? And how much is it a storage discussion? How much is it kind of a digital transformation? >> Yeah, I think we see all of it. We'll talk to customers that find that they can't innovate quickly, right? And they want to get so much more value from their data. One of the studies we cited in the keynote today was 80% of companies think they can make 20% more on the top line if they can just get insights out of their current data. I mean, that's a staggering statistic. 20% top line for every company if they could just get more out of their data. We want to make that possible. Their constrained with very expensive legacy technologies. That they simply can't give them the access to the data. They don't have the performance to mind those insights. And the infrastructure is so cumbersome, they just can't evolve and move their business forward. And so providing that recipe, you know, giving customers the ability to get dramatically more value out of their data and do it for lower cost is working. >> Yeah, and it's been interesting to watch kind of the data center to the cloud, and now cloud to the edge. And you've got solutions that are spanning across them. How do you see that maturing in really the vision to expand where Pure fits in the discussion. >> So, you know, from early on we targeted the cloud market. Because we knew that this is where the future lies, right? Even traditional enterprises still want all the benefits of the cloud inside of their own icy environments. >> And when you say cloud, you're meaning SaaS providers, service providers, as well as, you know? >> Yeah. We talk about the model that the big three are using. But, you know, this is very popular in many other clouds. The world is not moving to three data centers. Companies like Apple and Facebook are very committed to their own data center investment. And we seek to be a supplier to that consumer internet. The softwares of service and infrastructures service of providers. Because that's where the data center's going. But, you know, what we've seen recently with the proliferation of internet of things in sensor data is customers are just growing these huge data footprints that are just too big to move across public networks. So we talked about, in the keynote, in three years only one, out of every 20 bites that's generated, can fit on the internet that year. >> 2.5 out of 50, I think was the number. >> 2.5 out of 50 zettabytes. 50 zettabytes will be produced that year but only 2.5 is going to be transferred across the internet for the entire year. So we've got to get better as an industry at helping customers capture that data where it's generated, right? We call that edge. Sometimes it'll be on the devices, or it'll be in data centers that are close to the edge. And they've got to mine insights from it right there. >> Dave: Absolutely. >> One of the exciting demos we're showing here is actually AI co-processing with the public cloud. So we've got an edge data center that we're running deep learning in. But then we're selecting particular data sets through the deep learning to transfer it up to the public cloud for more machine learning. >> Those key nuggets, the needles maybe you transfer. Cause otherwise it's too expensive to transfer all the data. >> You can't transfer all of it. So if it's a self-driving car, you know, if I'm just routinely driving along, no big deal, you keep the data. But if I slam on the breaks because a dog's in the crosswalk that's the thing you want to do the training on. >> That can't be an asynchronous operation, right? So, okay, you're already getting the hook, I can't believe it, he just got here. (Scott laughs) Cube is a comfortable place but we got to throw some hard questions at you. So >> Please. >> Stu asked me the other day, or, actually, today, "Who's going to reach a billion dollars first?" And you don't have to predict, you can leave that to us. "Nutanix or Pure?" Okay, so talk about HCI. You made some comments up on stage about hyper-converged. Said that, you know, it's good for its own specific use cases. What's your point of view on that? >> So first of all, Nutanix has built a great business. >> Dave: Awesome, yeah, sure. >> We're absolutely fans. I will say, in the markets, those two new markets that we're playing in, in the cloud market and in the next gen applications and deep learning, we don't see hyper-converged infrastructure. We do see hyper-converged in business and enterprises. But it's usually the smaller scale deployments. The reason is, at scale, you don't want to collocate applications, data, and storage all in a single tier. It limits the ability to easily scale independently. You know, if you need more capacity you need more application compute versus data compute. You want to be able to flex those independently. Which is why all the big clouds and enterprise data centers run converged rather than hyper-converged. But the change that's coming is fast networks are changing this even more. So what I believes going to turn hyper-converged inside out is it's now more efficient to access remote storage than it is the same storage on your local chassis. And that's because we're offloading compute to the server net cards on there. So these new protocols NVMe over fabrics are actually making the network finally really the computer. There's no longer a chassis that's even meaningful. >> Big fan of that infrastructure and NVMe over fabric. Okay, next tough question is the narrative, from the big guy, EMC in particular, Pure is small, they're losing money. And your return narrative is tell EMC they're large, they're slow, they're outdated and confused. Okay, we love that, you know, it gets a little juices flowing. But here's my question. A lot of customers are large and slow and outdated and confused. So how do you get that fat middle to move faster and become a tailwind for you guys? >> So I think it's happening. I mean, customers just want technology to be made easily. I mean, one of the disrupters that's really helped is the AWS user experience, right? AWS has reset the bar for IT everywhere because people are like, why am I paying for consultants to visit my data center and take care of this mainframe or client server error technology that used to be so expensive. You know, consultants coming along with it. And permanently staying with it was okay. That's not okay, right? The world needs to move to self-driving infrastructure and they need radically better performance if they're going to use these new techniques. And so I think the key motivation is customers need to get more value from their data and they need to drive down costs. And we're in the sweet spot of being able to provide it. And these 20-plus year old designs can't. There's no way. >> So it's inevitable is really what I'm taking away from that. And you've got a lead that you can sustain in your view. >> You know, it's been very interesting to watch our competitors talk about the new FlashArray//X. With all NVMe and the new FlashBlade. They've said these are science projects that won't be real for three years. And, yet, we've won one of the biggest AI platforms in the world. You know, 25% or more of our business is coming from cloud customers. So, you know, from where we sit, things are going exactly as we'd hoped. >> Love it, we're talking about the edge, you're pushing the envelope at the edge. Alright, Scott, we'll give you the last word. I know you're super busy, but give us the wrap up. The bumper sticker on Accelerate 2017. >> Oh, it's such a phenomenal group coming together to talk about innovation. We've already shipped the hardware form factors this year, with our new FlashArray and the new FlashBlade. But the thing that I'm so excited about is we've got more than two years of software innovation teed up that we've been very quite about. So when you can bring two years of innovation and pack it into six months like we have this year, it makes things really exciting. >> Well congratulations on getting to this point. We're really excited about the future. Scott Dietz Dietzen, thanks for coming on The Cube. Great to see you again. >> Thank you, always good to be on the cube. >> Alright, keep it right there, buddy. We'll be back with our next guest. This is Pure Accelerate, live from San Fancisco. We'll be right back. (soft electronic music)