 Welcome back everyone to the next generation of storage. I'm John Furrier, your host of theCUBE. We're here in Palo Alto studio with AJ Singh, Chief Product Officer with Pure Storage. AJ, great to see you. Great to see you too, John. Thanks for coming in. Yes, I'm happy to be here. So you're the Chief Product Officer at Pure. You guys have had an incredible run. Congratulations to you and the company. Thank you. We've been following you guys from day one when Flash first came out. You know, really kind of like against the grain at that time, but now Flash has crossed over economics, sustainability, really obviously next-gen mechanism and ability to store, but now there's more to it. You get the keys to the kingdom over there at Pure Storage as the product lead. As you look at what's happening in the market, what's the storage market evolving to? We're seeing innovation happening in storage. Again, it never goes away because you've got to store the data somewhere, but it's evolving very quickly, certainly with GenervAi. Absolutely, John. I mean, if you think about storage, especially the unstructured data side of it, it's truly exploding. If you think of data, it's kind of exploding. You know, in 2023, roughly 120 zettabytes of data put out there and growing at roughly 25% a year. So just that's 30, 40 zettabytes added every year. And if you think about the budgets though, on the other hand, on the storage teams, the budgets quite are not keeping pace with. So what ends up happening is a lot of storage teams today end up in this kind of run the treadmill of just kind of keep the lights on, keep storage going. And so they're dealing with these issues of, what I would say, still mostly legacy mindset, legacy storage. Flash is traditionally penetrated in the high-end, high-performance side of the marketplace, but we are now here with Flash that we're truly penetrating every corner of storage, rather than tape. But all stages of disk and storage, Flash is here to stay. So what's the main difference between the legacy storage and Flash storage as you look at this next wave coming? Because there's a lot of architectural conversations happening, core cloud, on-premise edge, a lot of things are going on. What's the difference, true difference right now? Yeah, I mean, if you think about it, the two difference really comes down to sort of dealing with challenges like every five years with legacy storage, you've got to go do a forklift upgrade. Typically, that means you have to bring things down, do a data migration. So the reliability aspects are not there. A lot of times people are just used to that storage, but there's a better world out there. You can now run storage with 100% uptime for 10 years straight out with Flash. And even with new technology, you can upgrade the technology, no downtime, we call it non-disruptive upgrades. That's kind of new compared to legacy. Also, storage we feel like is becoming more reliable. You think about hard disk return rates, right? Hard disk return rates are typically two-ish percent. It's well documented, you've got to Google, you know, hard drive return rates, two-ish percent. SSD return rates, which is Flash, by the way, and I'll talk a little bit about how we differ from SSDs, it's about one percent. In our direct Flash module, return rates are roughly 0.5 to 2%. So you can imagine 10x more reliable compared to hard disk, 5x more reliable compared to Flash. And you can imagine, anytime you have a failure, you've got to run a hole, somebody coming in, pulling out the failed drive, rebuild a new drive, it's a lot of labor. So you don't need to put that much labor, you don't need to have the same level of failure rates, you can have much better failure rates, and overall, just a much better outcome for you as a consumer of enterprise storage. Hey, Jay, we were talking before we came on camera about some of the research we did at Wikibon now, it's called theCUBE Research, we just renamed it, because that really was the beginning of the Flash, we had made some forecasts. You have data now, you're seeing that kind of playing out. Can you share what's going on in the Flash, storage, price performance, cost, crossover, because it's clearly going to have some sustainability advantage, but talk about some of that, the trends that you're seeing now, and where it's come from and where it's going. Absolutely, that was a great piece of work done by Wikibon, and if you recall, what that data showed was, if you look at dollar per terabyte of hard disk versus Flash, think about it, right? Hard disk is on its seventh decade of innovation, done a great job, real workhorse of the industry, but the amount of optimization you can eke out of it is now limited, of course, the industry doing hammer and hammer and all that good stuff, and those curves are still reducing on a log scale relatively flat. Flash, on the other hand, is more of the third decade of innovation, as the Wikibon slide showed, a much steeper curve on the price drops of Flash, driven by consumer Flash, laptops, PCs, right? They already gone to Flash, so tons of volume happening, tons of innovation that the Flash vendors are doing, and so that curve is going much steeper, and SSD, what the Wikibon slide showed was, a chart showed was that in 2026, dollar per terabyte of SSD is going to get cheaper than hard disk. Now, we introduced in 2023, our first disks take our product, where we have a lower total cost of ownership with our direct-to-Flash versus SSD compared to SSD in 2023, so our curve compared to the Wikibon SSD curve, if you think of the direct-to-Flash module curve, is three years shifted to the left of that, and predominantly the biggest driver of the shift is if you think about SSD for a minute, when the Flash vendors first came out, they were like, hey, I got this great new technology, can I use it for storage, no moving media and all of that, and people said, yeah, but you know, it's got a proprietary API, all my software is in for disk, so if you can have a talk this to me, I'll use it. So the SSD vendor went back, I mean the Flash vendor went back and they said, oh, here's a new module called Solid State Disk, it talks disk northbound, it talks Flash southbound, but what it is, it's a complex translation engine in between, which has a CPU, uses a ton of memory, it's like, somebody wants to talk to me in English, but I'm first talking to them in Spanish, translating it into English, but direct-to-Flash, we said, let's eliminate this translational layer and just go straight direct-to-Flash, in some sense, someone like a Pure Story, we were born in NAND, and our engineers, they know NAND cold and they're looking at, how do you rehydrate a cell and how do you right-amplify, and in fact, all the NAND vendors come to us and say, hey, how's it going in the field with our NAND supply? The rest of the competition with SSDs, you have an arms length, they have an arms-length relationship to NAND, they're like, ah, NAND, I only want to consume it if it looks like disk, and so they really are not close to NAND, and so that's the reason our curve is kind of left-shifted. Yeah, you got better leverage there, better technology, you guys have always been innovative, I want to get to the market share in a second, but I want to stay on the innovation thread for a second, from day one, it's always been an innovative culture, even with all Flash at that time was controversial, it'll never work, it'll never get escape velocity, it kept getting there. As you look at the innovation now, what is the innovation strategy with Flash, because I see a lot of tactical benefits, I see speed, cost, scale, check, check, check, sustainability, very strong story on sustainability as powers are going to be the constraint. Yes, yes. As we see AI coming, only going to get more important. Yes, yes. You guys do well there. What's the innovation strategy for pure storage as you look at that next wave? Yeah, so the way we think of innovation at Pure is compared to the rest of the industry, we tend to invest about 20% into R&D. So we are really betting for the next decade. We think multi-cloud storage is the next horizontal that is going to become an industry of its own. And we find a lot of our competition tends to be single-digit percent investment in storage. I feel like for some of them, the cloud has been a little bit of a distraction, they're like, oh, all the storage is going to the cloud, maybe we should be focusing on some other things and not on storage and let's put it into private equity mode and all that stuff. We are convinced that storage, multi-cloud storage, flash is the media transition that's happening and we are leading that media transition and flash is needed not only in enterprises but even hyperscalers. Hyperscalers also have a ton of hard disk and they need to change that for all the sustainability, reliability, less labor required type of reasons. So there's a ton of room for innovation there but as you continue to innovate in this space, our view is that starting to of course have a simple portfolio. If you look at a lot of the industries with the new portfolio with flash, you have the opportunity to do that. A lot of the industry has traditionally had low end storage, medium end storage, high end storage, all disk storage, hybrid storage, all flash storage. That's a three by three complexity nightmare, right? Survival customer, I've got bits of pieces of all of these, silo data, it's not a pretty picture. But when you start fresh with flash, you have the opportunity to have a much, take a break from the past and so in our case we have just one purity operating system, one direct flash module that goes into many different areas, one scale up architecture under the hood, one scale out architecture under the hood for latency and throughput applications, one management system and this purity runs on Azure, AWS and on-prem, so it's purity cloud in some sense. And then we have one cloud native architecture with portworks and a cloud operating model with Evergreen and our Evergreen approach. The piece parts come together, in fact we had a research meeting just the other day with our research team, Rob Stretchy and our team, who's now with theCUBE Research, formerly Wikibon as you mentioned. He pointed out that obviously innovation strategy is strong but you guys have the price performance there, so speeds and feeds are coming back, that's a whole nother conversation, but there's a lot of TAM available, there's a lot of big markets still in storage, so tons of headroom, so your comment about private equity is a little bit of a tongue in cheek but the companies that don't have innovation, they don't have a lot of prospects, and so hence the private equity, you meaning they have to kind of sell basically? Exactly, I mean I would always look at any company and if they have anything associated with private equity means I'm kind of giving up on innovation, I mean I'm not going to throw the private equity price under the bus. No, they're arbitrage. Yeah, they're optimizing it for EBITDA, right? On the other hand, I mean, Code words, trade people off and sell and get some dollars back, it's a financial transaction, it's not an innovation transaction. It's not an innovation transaction, right? On the other hand, we see a ton of opportunity in TAM, I mean just if you look at the number of exabytes consumed in the industry, it's literally hundreds and hundreds of exabytes that are still on disk between hyperscalers and enterprises, it's a huge opportunity. We see it as a 50, $60 billion TAM for us, for flash to consume and we feel like we are leading the disruption. And actually when I am talking about disruption for a minute, there are other parallels in the industry, right? Media based disruption, if you think about in the audio industry, right? We had LPs, CDs, Walkmen, then the iPod, streaming media and with every media shift, a new leader emerged, right? We had the Sony Walkmen, then the Apple iPod and then it was Pandora and Spotify on streaming media, media transition, right? Same thing in the video side, you saw that happen as well, right? Hey, it was VHS, DVD, streaming media, like in the VHS DVD days, so we had blockbuster. And the functionality one got better too, everything got better. Yeah, and now streaming media is so much better, right? So we feel like the same transition is happening in the storage industry as it goes from hard disk to flash and we are kind of with a three-year lead leading the charge there. Let's get into some of the market share data. We see IDC seems to suggest you guys are gaining share over the last decade. What's driving that? I'll say it's an innovation strategy. Is it the fact that it's going to outlast out with the competition and you guys have the better product? What's the reason for the share increase? Yeah, and actually the biggest driver of the share increase is that customers start to see that we deliver really compelling outcomes. The industry, the customers care about outcome then I'll elaborate on some of those outcomes. Typically we tend to be two to five X less power. So from a sustainability standpoint, less power compared to other all flash systems, 10 X less power compared to disk systems and space. So power and space, we are much better because of the direct to flash. We don't have that inefficient translation layer sitting in the middle. But we are also 10 X more reliable. We talked about return rates, right? All that extra translation layer that fails all the time, right? So SSD return rates versus our return rates are five X better, right? We tend to be, as a result, fewer failures, easier to operate, less labor to operate. We'll typically be five to 10 X less labor to operate. You talk to a customer, they say, oh, that pure storage array, it kind of just runs. We actually forgot about it. I mean those are some bold numbers. So 10 X, what were the numbers? 10 X, what was the numbers? Yeah, two to five X less power, 10 X more reliable, five to 10 X less labor to operate. And a really consistent, simple product line. Not this three by three matrix that I talked about, right? That's kind of another reason. Well, operation simplicity is one of the cloud ops we're seeing with the cloud, next wave of cloud, it's not your yesterday's IT environment, it's more platform engineering, a lot of important words you mentioned there in there too. So how do you guys deliver those claims? I mean, that's like a pretty significant performance. I think sustainability may seem two X is still damn good. I mean, convinced of the 10 X, big numbers. Yeah, so it comes. How do you deliver those? Yeah, so there are four, really, what we call core sustainable advantages. What I mean by sustainable is it's not like you can get 10 engineers for six months and do it. It's a 200 engineer or four year kind of effort. The first one that we talked about, the direct to flash, we don't have this inefficient translation layer that burns a ton of power, takes a lot of space, so we don't have that, so that's why we do two to five X better power. The same translation layer, by the way, causes a bunch of outages, and it's not as reliable. And if you look at the, from an outcome standpoint, it's public data on 1% return rate for SSDs, we're roughly 0.15 to 0.2%. So five X better on that. Again, the same translation layer. And the labor to operate are evergreen model. So we have engineered into a product that you can upgrade, it's a thing of beauty to watch. You walk to the array, you can just pull out the controller. The array is running, put in the new controller, and the thing just keeps running, there's no migration. So an array that somebody bought 10 years ago, 10 years later, is the latest and greatest technology. And over the 10 years, they've had no downtime. So it's truly and always on running storage. That checks on the performance side, checks on the operation side, also on the power side too. Yes, five X less power, it consumes. So that's a big, talk about the, this is one of those things that I won't say is overlooked because people always talk about it, but it's kind of like falls into the save the planet category, people kind of roll their eyes. But it's not so much save the planet, it's much more, we don't have the space in the data center or the power envelope to do it. Absolutely, if you think about power, IT consumes two to 4% of global power, just a macro perspective. And typically storage is 25% of that. So storage could be 25% of 4%, that's a 1% of global power. A lot of that is on inefficient spinning media, hard disk. With Flash, we can cut it by a factor of 10. So imagine that 1% becomes 0.1%. And you can imagine the hyperscalers, places like Ireland, they said, no more data centers, we just don't have the power to give you. So they're like, where do I put my next data center? How do I get the power for it? Iceland. Iceland, yeah. Built-in cooling. Built-in cooling. No, but this is much power and cooling, it's power and cooling. No, but this, I heard, I just heard the other day there was a customer that said, I either got to pay my power bill or my IT person to run it. So if I don't pay the power bill, I don't have IT to run it. If I don't pay the power bill, I can't pay for my IT guy. So you're in this, this is real economic hardship. Absolutely. With the economics of power. Yes, and actually, especially in Europe, it's becoming really big. And in the US too, in California, it's starting to become big and it's starting to pick up here as well. Okay, I got to ask you, you're the product lead. You've got the keys to the kingdom at the company, as I mentioned, because the product is a fun position because you get to see the customer and you see the engineering side of it. How would you describe why Pure's product is so good? In the elevator or at a party, what's the bumper sticker? What's the elevator pitch on why the product's so good? Why is Pure? So innovative, why should I invest? What's in it for me? Why is Pure a great product? So I mean, Pure ultimately is really trying to, what I say, to store, manage and protect the world's data. And the reason the product is so amazing is that we've got the best flash technology in the marketplace, three years ahead of everybody else that takes substantially less power, substantially less space, is much more reliable. And we have the Evergreen model built into it. It's built to just run forever. Think of it as self-driving storage. I mean, imagine from the days of, where you're constantly, so think of it in that sense. We are like self-driving. Why the Evergreen model for the folks that don't know what that is? What does that mean when you say we have the Evergreen model? What does that specifically mean? Yes, what the Evergreen model is that, basically say that once you buy storage from us, the first time, you put your data on the storage, 10 years later, the storage will still be running, and you will have the latest and greatest technology, which means you get software updates, all the latest great from snapshots for ransomware recovery or, you know, snapshots to ransomware snapshots, a bunch of software upgrades, plus the hardware controller, you know, went from PCI Gen2 to Gen3 to Gen4, controller upgrades, all happening on the fly with no downtime. And it's the best in terms of recovery because any of the old parts we recover, be recycled, so it's very friendly from a climate standpoint, but the benefit you get is non-destructive upgrades. You know, it's always on, always running. Awesome. Where is the product portfolio going next? Obviously, you got the roadmap, you got the leading edge in, you've always been innovative, love the innovation strategy, plus you got product leadership, but best of breed, that's table stakes now. You know, it's platform now, you got cloud, it's a platform, market, storage is going to be horizontal. You mentioned that earlier. I would say multi-cloud, or we call it super cloud layers, control planes, starting to see companies build these data planes around their distributed environments, whether it's multiple clouds or environments. I'm sure power at the edge is going to be an important part for this. I can see great things there. Am I kind of on track? I'm like, I'm wondering if you know anything. Absolutely, I think you've kind of said it pretty well. Essentially, we've got this really simple portfolio that runs on-prem with these amazing numbers, runs on the cloud. On the cloud, by the way, we can save customers storage costs, 50% of the storage cost, running on the cloud itself by coming in between. So think of us, we've taken this simple portfolio and we're elevating it into a platform. And the platform is, think of it as a storage cloud that effectively runs in the public cloud on the edge, on-premise, with a control plane that is all API-driven, where you can track your data. It has metadata in the control plane. You can track your data. You're creating snapshots. So essentially, think of it as a, you had a CMDB for configuration management database. Think of it as a data management data, DMDB, right? Tracking your global data. Now think of the AI use case. The biggest challenge in AI today, if you think for a minute, is that the data is locked up in silos all over the place called the old model. And so if I have to suddenly train a model, I got to go copy the data everywhere, then train the model. And now I need to drive inferences from the model. And then I'm going to figure out where that data out there, I have to cleanse it. But with this approach, your data is always on. And it's always on with a truly unique technology in flash that is highly sustainable, very friendly, very reliable so that your data is available and you can start to drive inferences from that data now, right? And get those applications that'll help you win in the digital transformation. I think the Evergreen model also points to the fact that as the Gen AI comes in, you'll need software updates to manage this inference. When to make data available, governance, a lot more intelligence in how to manage the data is going to come down and be programmable, not so much human driven. API driven, both for traditional and cloud native. You know, we haven't talked much about Portworx, but Portworx is effectively our solution for the cloud native side. It's all integrating into the platform. So the platform will be truly for both traditional and cloud native applications and being able to... We've been covering Portworx, I got to say that's a secret weapon you guys have with Portworx. It's doing great. Doing great, it fits into the platform. I think that's a nice tie into where I see Kubernetes going, because as we pointed out in theCUBE, Kubernetes is getting boring, which means it's working. Like Linux, no one talks about Linux anymore other than it's a standard, everyone runs it. There's no Linux conference anymore. So we see KubeCon evolving into the conversation of platform engineering and end to end platform data engineering. Absolutely. It's coming very fast. And we got great research on that. We've been reporting on it. Data engineering is the next persona for IT. And this is where the data platform, right for both cloud native and traditional that enables you to, now imagine back to the AI example, you know, of course you need the data to be able to train models and all of that, but then to write the training apps or the inference apps, a lot of times they're written for the first time and they're written on containers, using Kubernetes. So, you know, we've got many cases where Portworx is on training apps and inference apps and our core platform is on the training side. AJ, it's always great to chat with you on theCUBE. It's like we're having a little RIF session here. It's like we need a whiteboard, a great leadership. Last question for you as we, as you lay all this out, it's very clear that you're enabling a positive disruption for your customers or transformation. Yes. I call it disruptive enabler. It's disrupting the status quo, but you're also enabling a transformation for the future. What are the customers seeing for benefits? What can they expect for the next gen storage to enable them to do? And what are they doing today? Take me through customer anecdotal use case or a slice of life of a customer of today and then what this will enable for them. Yeah, so you think of the slice of life, a life of a customer today, right? It's kind of, I've got limited budget. I'm kind of running on this constant treadmill of upgrading my storage, a forklift upgrades. I'm running constant, hey, I've got to flip a failed drive, put in a new drive. So it's a constant kind of treadmill, right? And it's quite fragmented. A lot of grinding. A lot of grinding, different use case, fragmentation by use case. Now, if you think of the model of this storage cloud, a consumption-centric model, so we have the Evergreen One, which is focused on that. It's a true horizontal cloud where now the failure rates are dramatically less. Two, you're not running this treadmill of upgrades all the time. It just seamlessly upgrades. So you're always on the latest technology. And you're leveraging public cloud storage resources, edge storage resources, and on-prem storage resources with data that is annotated and tracked. So imagine this data, and with data services, that you can now truly work with this data to get the insights that you need to drive to gain a competitive edge in the digital transformation. Data is gold, and we are truly unlocking this gold for our customers. Yeah, and I think the idea of having storage is always there, available for developers, available for the app developers, and then platform engineers. Yes. Key to have that platform. And again, the horizontal speaks to the availability of data. Absolutely, absolutely. AJ, great to have you on. Thanks for coming on for the next generation storage session. Appreciate it. Thanks, John. Appreciate it. Okay, in a moment, Prakash will be on, who's the general manager of digital experience for business unit that's pure storage. We'll be here in studio. You're watching theCUBE. The leader in high tech enterprise coverage, extracting the signal from the noise. We'll be right back.