Newsfilter article unlocker

I'm delighted to have with us Charlie Giancarlo, CEO of AeroPure, formerly Pure Storage. I guess before we get started, I'm just going to read a few statements on behalf of AeryPure. Statements made in these discussions, which are not statements of historical facts, are forward-looking statements based on current expectations. Actual results could differ materially from those projected due to a number of factors, including those referenced in AeroCure's most recent SEC filings on forms 10Q, 10K, and 8K. Thank you very much. Perfect. Anytime. Well, Charlie, thanks a lot for being here. I really appreciate your time. Maybe just even before we get kicked off, start off with all the questions. You folks reported earnings a couple of weeks ago, very strong numbers in Q1. There's a lot of questions around the guide and the back half expectations. Maybe just level set kind of some of the key points on the earnings call, how you folks thought about the back-half, and then we'll dig into the questions from there. Sure. Well, I mean, the key point about the first quarter is it was another billion-dollar quarter and the first time ever a billion dollar quarter in our first quarter, building on a billion dollar quarter. In in our fourth quarter of of last year and just as a reminder, our quarters end the end of roughly the end of January each year. We reported thirty five percent overall revenue growth, which is certainly, you know, a significant pick up in overall revenue growth, but I think even more importantly, 55% increase in our product revenue growth which by all accounts is a proof point of a significant gain in market share overall as a company. We also raised our guide for Q2 and for the year as a whole. So to give you a sense last year at the beginning of the year we had guided to 11% we delivered over 15 close to 16 percent for the year this year thank you so much this year we guided at the the year to 19% the guide at the end of q1 for the years 22% so already a 3% increase now what we did do is that 22 that extra 3% came about because of the beat we had in q1 as well as we increased for the full year by the amount that we're projecting for Q2, which is, again, a significant increase in terms of our guide as well as year over year. What we didn't do was raise the second half of the year. There's a lot going on in this market. Prices are increasing, which are creating all sorts of effects. Supply chains are tightening, which creates other types of effects, which we didn' We didn't think it was really worthwhile or that we were adding anything by trying to define the macro in the second half of the year. Also, we have a history of not changing guidance for a second half the year after first quarter because we just don't believe that there's enough information. Generally our, if you will, we guide on the basis of our pipeline and our, to some extent our backlog which is only one to two quarters in advance so guiding towards the for the end of the year is just we generally don't do that after a first quarter so there was some disappointment I think you know by by some people in the stock that we didn't guide even higher for the second half of the year but on the other hand we're not trying to be the best crystal ball viewers we're trying to trying to run a company you know by improving and gaining market share every each and every quarter which which we are doing perfect no it's only seems a lot more prudent of a guy that's perhaps more dearest I would say then maybe some of the other ones that out there um I'm sure we'll come back and talk about memory a bit but you know we had a very high level think about it right you see this massive investments in AI infrastructure more from hyperscalers correct heavily on compute but it's on memory storage as well as enterprises start to get ai ready right and some signs this is happening can you just talk about what role the storage play for enterprises as they get ai ready as they look to build out their infrastructure and where does ever purely play into that sure well for enterprises to get ready for ai it's very interesting a lot of the focus on the media is getting ready for AI on the compute side right building out compute farms super pods adding components for that whether it's networking or or or storage what's not what the media doesn't pay a lot of attention on is the fact that getting ready for AI is much more in the enterprise about getting their data ready for for for AI so what do I mean by that well for decades we've had an application centric application focused model for how we build infrastructure across the board doesn't matter enterprises hyperscalers it's all been application focused what that's done is it's fragmented the data across the enterprise you know we are you know where we hope to be soon actually part of the fortune 500 but we're not yet but despite that we've cut our application use in half just to be more efficient we're still well over 500 applications okay and that's the ones we know about most of the enterprises we sell to are thousands of applications every one of those applications has their own data set okay everyone has their on data set which is slightly different than another application and to do anything in an enterprise such as issue a quote or an invoice you have to pull information from many different applications well this is fret what's happened is the data is not consistent across these applications and when you have inconsistent data such as you know a customer means one thing the customer name means one thing in one application a different thing in a second a different thing in the third then coming up with the correct answers to an AI inquiry is it's a data problem it is not specifically a compute problem and so there's a lot of work going on if you would ask enterprises what are your biggest challenges for AI they also they will say it's their data and so there's a lot of that we are doing and helping customers get their data ready it's part of the reason why we acquired one touch it allows customers complete to create a complete catalog of their data it allows them to rationalize the different data that that exists in these different data sets that they have so getting enabling customers to get their data ready for AI in in my opinion in the enterprise is a huge market opportunity that is not being it's still very nascent in its infrastructure that being said let's go to the what the popular media talks about you know our flash blade product has been in in place for six to seven years in AI environments can easily handle up to several hundred GPUs which is what most enterprises, to the extent they build their own infrastructure, will be using. Our EXA, which now goes up to several hundred thousand GPUs, is available for sovereign clouds, for GPU clouds, tech titans who scale to that level. And then finally, Portworx is in tremendous use right now in data pipelines around ETL. So we have a lot of irons in the fire when it comes to AI. Got it. And maybe in the spirit of the irons of the fire with AI, you folks have this engagement with hyperscaler. I mean, Meta is the one you folks have talked about a fair amount. Clearly, I think investors have a lot of focus on this hyperscale engagement. But maybe we just stick with Meta, because you have publicly discussed this. You know, what are you helping Meta with? What is Everpeer offering them that they can't do organically, perhaps? I'm going to genericize this a little bit, because it's the same set of values that we can offer to any of the hyperscalers. So it's not specific to one. And we're developing this hyperscale solution to be roughly the same solution. I mean, each of them will have their little differences here or there. But it's the same solutions for any of hyperscales. And that's one of the things that makes it attractive to them. So what are the advantages? Well, first of all, we're the only company literally in the world that has what we call direct flash technology. So what is that? Most people think, as we would as consumers, that an SSD is flash. And that's what everybody uses today, with the exception of our customers. They use SSDs. But what is an SSD? Well, the truth is that an SD is flash masquerading as a hard drive. In order to get flash into things like laptops and everything else, instead of forcing all of the different operating system vendors to have to change their operating system, the flash manufacturers came up with a brilliant idea. just make it look like a hard drive and then anybody can use it no change in software necessary and that's great except that hard drives and flash work entirely differently so by developing software that works directly with flash we automatically have about a 30 to 40 percent price performance improvement we're not going to spend time on that now but we have lots of patents and papers in this area people can go reference that well that's valuable to a hyperscaler as well right because they you know storage is what is their second largest spend in infrastructure after after servers and if as more and more of their business goes to flash we can not only save them a lot of money but we have a number of other advantages as well so one is we make better use of the flash so better space power and cooling our lifetimes are at least twice as long as an SSD if not longer we are our failure rate is 1 tenth of SSDs 1 tenth so you know that there are those benefits and then finally if that weren't enough every SSD every every hard disk every different size every different vendor when if there is a flash change every new flash changes requires tweaks to the operating system because that's where disks are managed out of the operating our software goes into what's called user space in an operating system so no changes to the operating system and the hyperscaler doesn't have to worry about modifying the operating for the next generation or just for a different scale our software takes care of all of that so we're just a much easier infrastructure to be able to manage so a lot of advantages to the hyperscaler. Of course, it's a big change from what they do today. And hyperscale is only modify their environment on roughly a three year development cycle. So and there's a lot obviously of qualification. So it takes time and it's an engineering to engineering interface. So it takes a lot. If I just think of a lot of the way you described the way I've understood pure, right, it's it's all about, we can help you do this in a more efficient manner, right? Correct. that raw NAND, raw memory, right? And maybe hyperskillers could ignore it historically because it's like the price of NAND just kept going down. So your problem was getting smaller, not bigger in a way. Correct. If you're in a place where memory just keeps going up at this point for several years, isn't in a ways, like, I mean, it's not TAM, but the problem you solve is getting bigger. The dollar, even if you keep the percentage the same of our advantage, the dollar value of that percentage goes up. Right. And therefore, it is more attractive. So, Charlie, maybe two things on that. like a why wouldn't the hyperscalers accelerate qualification processes because you're gonna save them a lot of money I'd rather do it now than later right and then second really maybe you can just talk about how does this engagement work from a financial basis for you is it yours installed use how much data we manage and use how we get paid for it or is it all a car just talk about those two vectors of it yeah well what what the current environment has done is open the aperture or widen the funnel so we're getting more interest but I'd also say that the hyperscalers are more busy now and have got more to do so it hasn't shortened the qualification cycle. Secondly, we're only one part of their design, we are a relatively small part of their design process. Their design process is governed by the next-generation processor, the next generation of their own software, the the next generation of the switching systems that go in there. So we're not, the qualification could be speed up and not their design element. That doesn't get sped up at all, right? So, you know, unfortunately, while we're getting more interest, it hasn't sped up the design cycle. So, all right. So now on the economics. It's a different set of economics than our traditional economics in that we're not selling a system per se. We're providing technology these and it's not a large portion of it is not our supply chain so we allow in a sense the hyper scalars to they're responsible for the NAND they're also responsible for and they buy a completed product from their supply chain now our involvement is it's our design we work with their supply chain so that they can supply it but the NAND is sourced by the by the customer in this case the hyperscaler so what we get is a royalty on every what we call direct flash module that the hypersonal buys we have some cogs because while they while the hypersonic contracts if you will for the NAN directly there are other components that they don't want to be responsible for so we're responsible for that so our gross margins you know and there are mix issues that will will vary the gross margin but generally in the 75 to 85 percent range and given the mix you know a lot of you not not appears a lot of companies right now talking about hey we're seeing this uplift and general purpose x86 servers in a very profound manner actually yes historically I think there's been a decent relationship between hey if you want to spin up these new servers and create new workloads you've got to store your data somewhere ie storage will fall through at some point yeah just talk about like historically how's that lag being between compute versus storage and does that happen as you go into an AI world as well well they have somewhat different cycles sales cycles but on the other hand you know storage in a traditional data center so let's leave AI out of this for the moment in a traditional data set or you know compute will generally be on the order of 60 to 70 percent of spend storage 20 to 25 percent and what's left over is networking at 5 to 10 percent that's broadly speaking about about correct so storage is larger than a lot of people think you know in terms of percentage of spend on infrastructure in a data center I'm not counting power supplies and air conditioning and stuff like that the in in in a in a AI world because of the cost of both HBM and and GPUs storage is much smaller and in fact because AI uses the fastest you know highest performance networking network networking is actually number two storage is number three in a AI environment was there I'm blanking on the the bottom so much that yeah I would say that the purchase cycle around processors and the purchase cycle around storage tends to they're not tightly connected because a lot of times you can buy processors and you have spare capacity on your existing storage sometimes you start running out of storage because the data increases so it may not be tied to a directly so directly to a compute purchase cycle got it um you know memories being another topic that everyone's being focused on if you spend a lot of time on it from an input cost basis but you know if I think about what's happening with memory maybe in a bit more of a strategic basis versus just the prices historically when you've had this periods of component tightness it's reshaped a lot of competitive dynamics in the industry as well at the same time here's maybe talk about how doing the current memory environment influences your competitive positioning and you know how do customers value proposition change with us right so when we talk about memory I want to broaden it a little bit you know there's the true memory such as DRAM you know and now HBM but there's also storage so you know flash is storage memory right both of them are going through the same you know significant price rises the way it affects us is you know quite interesting we as you know in a up until recently we can talk about that you know our gross margin on the product side has been in in the upper 60s, right? Even past 70, which, whereas our competitive set, the storage industries as a whole generally has about 50% gross margins on product. So right away, the equivalent increase of cost in our BOM results in a lower requirement for us to raise prices. So we've been able to raise price less and less later than our competition because we don't use as much memory or storage okay for the same amount of for for us to be able to deliver the same capacity of storage secondly we've chosen to share some of the pain with our customers who are operating at the lower end of our guidance while these price rises are taking place for a variety of reasons but at the end of the day I can't go into a customer and say I'm so sorry that you have to pay more and we're going to take more profit you know at the same time so you know our first value is as a as a company is customer first and we live that it's not just a you know it's just a saying so you in the short term we have chosen to operate at the lower end of our 65 to 70 percent guidance but at the same time that also makes us more competitive I think you know part I think we're going to see enhanced growth from a from a competitive standpoint of market share standpoint because we don't have to raise as much and we're choosing to you know be in league with our customers and maybe just in the price increase I think it's been three price increases that I can track in the last since start of this year at least right the start of the year it is the best we can track you know your price increase of being less benign they're all crazy numbers for the less benigned then what others are do you actually had a letter out to your customers very well benign less less increase less increases more benign to your customers. You touched on this a bit already, but just talk about it. Why are you able to provide or have a lower price increases than your customers? And is it resulting in you winning more business over time, actually? So one of it is just the physics of the financials, which is, you know, if traditionally our COGS were 30 percent of our product sales versus 50 percent you know a 15% you know what let's say or 50% rise in your cogs you know takes you to four would take you to 45 for the same price you only have to raise prices by 15% of you know roughly speaking right to make it up versus if it's 50% you have a higher you have to raise it higher to keep the same margin right so you're just the physics of the other financials make it possible to raise it less secondly we are choosing to share some of the pain with our customers, so that's even less. We bring it down to roughly 65. And if that doesn't raise market share, then I don't know what will. What I will say is that Q1 still, because of the lag between pricing, shipments, and backlog, Q1 was still mostly at Q4 pricing. Q2 will be much more in you one pricing which as you said you know has increased so so I think what you're going to see as we go forward or what I hope to see I should say going forward is increasing market share got it all right now I will say let me just go further yeah you know when we announced last quarter I mentioned you asked me about it right up front 35% growth right rough estimate we thought about you know about a third of that was on the basis of price increases in pull ins that would be about 10 to 12 percent we grew 35 which means 20 you know 20 percent growth if you look at our competitors they all grew less than 12 percent i i don't think we're special in terms of how many pull-ins are coming in no matter what anybody says so if their pull- ins are roughly the same because everybody's responding to price increases by trying to get in a bit early to avoid future price increases then we're growing quite substantially absolutely fair um you know beyond poland's which you know i think everyone's going to debate is it happening and how much is that happening but there's only this element of it when you talk to your customers right i mean i.t budgets at the end of the day are relatively finite correct um how are they navigating this surge in storage pricing very specifically are they saying i'll take a lower cure storage i'll maybe some hdds i've heard from people for for example just how are customers dealing with this right now the it is in fact from a number standpoint too early to say because the prices are just starting to take place and customers are attempting to get ahead of future price increases by ordering early so right now demand is you know very very high which is why we're trying to parse out what is true demand what is market share gains versus what is uh pull-ins right because what we care about is market-share gains those will last market share games will last you know short-term Activities by customers associated with with price rises or anticipated price rises, you know, that's that's temporary, right? So I think it's a bit early to say but historically, you Know our experience is that with price Rises or price declines, which is tradition the traditional You know environment that one lives in is that? demand of bits right does vary does change with price rises or decreases but not so much as to overcome the price itself so price trumps capacity but capacity does respond to price changes in both directions yep what about a desire to shift to slower storage or even HDD or hybrid arrays are you seeing customers actively make the decision i mean historically it's been a one-way train everything goes more flash right well yes that that has changed because the ratio and it's still it's still um in transition it's Still changing now because both flash and um hard disks on the source side are sold out through 2028 okay so you know there's no more supply of flash there is no more that hasn't been accounted for of hard disk until 2028. So prices are still demand. Sorry, supply has not kept up with demand or looked upon another way. Demand has not declined enough to match supply. And that's why we're seeing continuing price increases. What we can say right now is that the market that we were very active in about a year ago of flash being competitive on a total cost of ownership basis with hard disk, that doesn't exist right now. That's gone into stasis. Now, I'm a big believer in reversion to the mean. It'll come back, but it's just not with us right now. And then from your vantage point, you have to buy a lot of memory to meet your customer's demand as well. Just talk about what sort of visibility, what sort conviction do you have with your memory as an input cost? And you're talking about both Flash and RAM? More Flash, because it's a bigger input cost for you, I feel, than DRAM. But in broader, just talk about your LTAs, the visibility on bits and pricing or just bits. So we have very good visibility on both. And part of that is because of our hyperscale commitments that we currently have in place. So we Have plenty of flash. What has changed is that there is a separation now between being able to get long-term commitments on volume and getting long- term commitments on price. There are no long term commitments on price right now. It's an allocation game. And, you know, pricing is basically every 30 days. Right. So we have allocation on capacity. We price was a 30, which is part of the reason why, you know, prices are escalating is because you can no longer say, yeah, for six months, I can buy a head for six months. You can't you can get capacity, but not price. And does that change how you quote business, your customers at a 30 day increment as well? Yes. So we very, you know, for our entire life as a company, we have been quoting with a 90 day quote expiration. We're down to 30 day quote exploration. Now, by the way, that's longer than most of our competitors. Yeah. Yeah. I've heard stories about, you're right for cancellation, right to change pricing up on the shipments. It's yeah. Until shipment. It's all kinds of fun clauses are getting implemented and see how well they stick over time. But, you know, one of the dynamics that emerges like this, NeoClouds as a category, right, and CoreWeaves and others of the world, like they have a lot of spending dollars, they're doing a lot of stuff. How do you think of that as an emerging customer base for EverPure? And you, I think, have a relationship with CoreWeave on a couple of different levels. Just talk about like NeoCloud, what do they want? How are you tackling that kind of customer base? So the NeoCloud or GPU Clouds, or for that matter, the sovereign clouds, was a market that we did not have a product to engage with up until the last year or so. And to put that in perspective, most of the large-scale LLM, large-skill AI environment grew out of the HPC market, the high-performance computing market. that for storage was typically somewhat of a cul-de-sac it was a very specialized market very scientific in nature very specs oriented versus reliability oriented very tuned you were selling to scientists and they like to tune things I like a lot of nerd what we call nerd knobs in the business and and And it wasn't a, think of it sort of like, oh, what was the name of the company that HP bought? What's it? No. The supercomputer company. Cray, thank you so much. Think of it like Cray. Cray didn't expand to do other things in the standard commercial market, right? And that was a lot like the companies that satisfied the HPC market. Well, AI grew out of that. And so it turns out that it was the people that were in HPC that now people all of the large scale GPU environments, whether it's sovereign clouds or the tech titans or the GPU clouds. And so we didn't have a product in the race. We didn't speak the same language. We didn' go to the same schools. We didn't have the secret decoder ring. And we certainly have certain beliefs as a company, in particular, that all of our systems should operate on the same software and have the same abilities, if you will, reliability, upgradeability, the evergreen capabilities that we have. And so it took us a while to modify or add a new set of capabilities to our FlashBlade product. But now, as I said, we introduced FlashBlade EXA about a year ago, same purity operating system, different configuration, where we are now publishing the best performance specs in the business, bar none. And while we are still a little bit behind in features, we're catching up and we're starting to to we're starting not only are we replacing some of the traditional competitors in that market, but we're we're starting to play a bigger role in these GPU environments. So it's still early days, but were very we feel and we feel that as these GPU clouds start to have to perform you know in a more um you have to be able to provide a higher quality more reliable consistent service they're going to appreciate a lot of our abilities that come with our product you know versus some of the the uh the ones that were designed specifically for the scientific community i was wondering where you would stack up i mean one of these i hear is like hey these new clouds the gpus need a parallel file system and that's not what you need to deploy but then then that, to your point, doesn't come with all the redundancies and services that you want. That's right. Is that kind of the delta that, hey, as they get more mature, purity will work? Purity will work. I mean, these are all just features. You know, I personally, having been in the business a long time, things like reliability, sustainability, diagnostics, that's really hard stuff that you only get over a lot of experience over a long period of time. Things like individual features, like a parallel file system, or KV cash or quality of service guarantees. These are all just features that... One of the other things that distinguishes us that is at a much more foundational business level is that we are a dyed-in-the-wool R&D company. One of things that 70% gross margins allows us to do is we spend 20 percent of our of our revenue on um on r d by the way in storage that makes us the largest r d spender in storage bar none okay we invest more than anybody else now i think that accrues to our growth uh it accrued to the quality of the product that we can put out but it also allows us to expand in other areas and um you know unless you believe that we're not terribly efficient at that uh but it's something we work on a lot to make sure we are efficient uh in the way we we invest our r d um and i think our growth uh shows that um you know these i don't worry about a feature gap if we're focused on the market segment perfect um maybe if you shift gears a bit uh evergreen one right uh it's sort of your subscription services offering um i would imagine at a time where memory's inflationary it would resonate more with customers versus not i'm sure it is uh but just talk about a bit more of the evergreen one offering and are you seeing a better uptake of it in an environment like this yeah so just to put some numbers behind that uh evergreen 1 grew i think it was 50 year-over-year uh in q4 and roughly 70 in q1 that would that indicates that it is growing faster than our capex product our historical experience with this is that as prices go up it becomes more competitive and by the way our price increases on evergreen one is far less than our price increases on the capex we can go into why but there are reasons why it is not as directly affected by price increases or input cost increases and so what that would say as we go forward and we only have one quarter of price increases really so far to measure this off of but I'd be surprised if it didn't mean that that our purse that the relative amount of bookings that come in via evergreen one will be higher than than the capex maybe there's two things on evergreen one what are the gating factors like well wouldn't why aren't why wouldn't customers opted faster than already have they would limits them in that maybe I let's see the same yeah well first of all it's a different way for the you know even though they're used to it's it's so funny even though there used to buying on subscription you know in the cloud right more for SAS or for the hyperscalers they're very unused to doing it inside their own data center it's at capex model inside their data center and honestly sometimes we spend two to three months you know with the the finance organization in the customer trying to train them on how they do it and obviously it's growing faster than our capex product but it's not unusual for them finally say you know it's just gonna be easier if we do capex so we're just gonna do that now pricing does make a difference so we'll see but you know the financial that is customer financial models sometimes are more difficult to change than their technical model so but your book but why isn't it just automatic it's generally because for those reasons. And then why is the investment from your side lower for Evergreen One versus the traditional products? Yeah. Well, first of all, obviously our terms are longer, but it's more than that. Our Evergreen one is a true subscription. So it's not a lease. It is not a financing. And what do we mean by this? What we sell to the customer is purely a service level agreement. It's a service low agreement on performance. It says service level agreement on capacity. It's a service level agreement on reliability. That's it. There's no serial number involved. We own the equipment. We get to choose. There is no guarantee of new equipment. If you go to Amazon, do they tell you we're delivering new equipment to you? No. You're just getting a service. We provide a service. Our Evergreen Forever subscription, which is for customers that do purchase, they are guaranteed hardware upgrades every few years. That hardware that comes back has more life associated with it. We're able to use that in our Evergreen One service. And what does that mean? That means we're blending costs over a long period of time. We refurbish it, of course. All the data is erased. we're we're using refurbished equipment in addition to new equipment and it's all again and we get to sell it you know at prices that are competitive with as-a-service pricing so so our costs don't go up automatically and we happen to be in a great cycle because we just went through a refresh on our capex product which means we have a lot of equipment coming back for refresh that was priced at three years ago so perfect that is helpful you know try you folks recently change your name from pure storage to ever pure you've certainly done a few deals like one touch port works a couple years back you are trying to become more than just hey use the storage layer right just talk about like maybe this was this should be my question to you at the start not then what is the vision with this is it more like we're gonna be the storage bottleneck solution provider because yeah it's more than traditional storage media no we we fundamentally believe and i i alluded to this early on that the nature of of applications and data is fundamentally changing and it's fundamentally danger changing because the app world has fragmented the customer's data environment and ai if left to the current way of doing things is going to fragment it even further everything wants another copy of your data every sas company wants every sask company's strategy right now is give me all your data and I'll give you the AI answers you know the data bricks of the world and the snowflakes of the they all say give us all your data and we will provide you all the AI answer and how many different organizations is a customer going to give all their data to I don't think it's scalable I don'T think it'S a viable solution our belief is that if if if these applications are no longer systems of record which they can't be because you know in order to do anything you have to have information from all of these different systems and if it's if that information is not consistent the world's in big trouble so what can be consistent is no longer the application but the data has to be consistent so if we're gonna if the world is going to migrate from app centric to data centric data primacy and what we're calling data primancy then we're in a very good position to help customers get there so what does one touch do one touch for example allows customers to get a catalog of all their data all of it not just what's on our stuff what's on competitive stuff what in SAS what's in the cloud what's you know what is with partners and allows them first of all to get not only get a the catalog but get a map of how this data set relates to that data set you know that has the same name of the customer in two different places but it has different semantics for what that customer is doing it has different information about that customer and one touch like the name sounds brings that all together so you can have a common semantic understanding of an entity like like a customer or like an asset or like a like it like a quote all right or an order and so what where we're going is we're going from pure data storage pun intended to data management where we can help customers manage their data and have a more have fewer copies have more comprehensive understanding of their data as we go into this world but the world is in it's not just about AI there is a data crisis going on if you ask any customer what is your biggest challenge to implementing AI they're gonna say right away that it's their data their data is a mess and and that's the that's so you know perhaps we're taking a what is not a well-known or untraditional approach to helping customers in this world of AI but we believe it's fundamentally a data problem and that's what we're going after let me see I suddenly have an affinity to the name ever pure but maybe it's because I work at ever core yeah there's a link there but I still think it's more than just a data storage problem so it does make a lot of sense I think the time's almost up so maybe I'll pause my questions I try that turn it back to you anything we did not touch on that we did not cover the info should be aware about well you were you were you're very kind and not spending half the time on the the hyperscale challenge, so I appreciate that. But that, you know, we are very pleased with the progress we're making there overall. But outside of that, no, it was a great conversation. Thank you so much. Thank you very much, appreciate your time, thank you. Thank you.