Okay. Hi, everybody. Thank you for joining. Tim Long here, IT hardware analyst at Barclays. Happy to have Pure Storage with us, Charlie and Kevan. Thank you for coming.
Thank you, Tim.
I really appreciate the time. I'll start with the safe harbor here. Statements made in these discussions, which are not statements of a historical fact, are forward-looking statements based upon current expectations. Actual results could differ materially from those projected due to a number of factors, including those referenced in Pure Storage's most recent SEC filings on Form 10-Q, 10-K, and 8-K. So, thanks again, guys.
Yeah.
Appreciate the time here. Maybe, Charlie, we'll start off kind of a high level, you know, talk to us a little bit about, you know, how Pure is differentiated. What are you excited about looking into the future? It's been a pretty successful journey, looking at revenue growth over the, you know, last 5-8 years.
Yeah.
What, what gets you excited, and what, what do you think differentiates you guys?
Absolutely. Well, we're really one of the newest, certainly, in the general purpose storage market, the newest player in the industry. We started out a little over 10 years ago, as a leader, in flash storage. We remain that way today in terms of all-flash, really second only to, you know, Dell EMC, who had been the historical leader in storage overall, and shown, you know, very good growth over that 10-year period. But really, what differentiates us going forward is partially based on where we've come from, which is the all-flash heritage. The all-flash heritage, we're the only vendor in the world, in any industry that has software that operates on DirectFlash, meaning we don't use SSDs.
The rest of the world uses an SSD, which has a translation layer to make flash look like a hard disk, which dramatically reduces the overall performance of that semiconductor. Our software manages flash on a system-level basis, giving us tremendous both cost efficiencies, but also performance enhancements. Having a very modern software stack, we also do a number of other things that no other vendor does. For example, we have our Evergreen program, which in its most simplistic form, means that our products never get old. It means that a customer, once they've bought our product, is consistently and constantly updated without any disruption to the customer's environment, meaning their applications keep working.
And the system is constantly brought up to a modern state, both hardware and software. So, you know, to give you an example, you know, you may have a Tesla or an iPhone, and every now and again, it gets a software update. In both cases, the phone and the Tesla stop working while the software update's happening. Not with us. While the software update is happening, the system keeps working for the customer. Furthermore, every three to five years, you may replace your Tesla. It gets, you know, it physically gets old. You bring it in, you buy a new one. Never happens with us. With us, the hardware of the system, not just the software, is constantly updated, again, without disruption. So the system that's...
Now we have systems that we sold 10 years ago, and without a single dollar of additional capital spend by the customer, they look like a system that we sold last year, both hardware and software at the modern level. That's the basis of our overall Evergreen program. Okay, so fast-forward, we've been doing this for approximately 10 years. Flash forward to today, we have a number of differentiating strategic advantages. One, of course, is that DirectFlash that I spoke about before, which gives us dramatically higher margins, better performance than competitive systems. It means we have, you know, between one-half and one-fifth the amount of power, space, and cooling than competitive systems do, so dramatically more environmentally friendly.
But also, you know, what if you're running out of power in your data center because of the increase in GPUs, we're able to dramatically reduce the power used by storage overall. Because of our modern software, because we started with all-flash, we introduced early this year, the first all-flash systems that can now compete at the cheap disk level. So right now, today, more than 60% of storage is still hard disk, believe it or not. The only mechanical system that still exists inside computers outside of the fan. We can now replace that with a semiconductor system, you know, a flash system, one-tenth the space, power, and cooling, five times the performance, same price as disk.
We just introduced that system, but we introduced it on the same software that we run all of our other systems on. So for the first time in storage now, we have one operating environment for the entire enterprise, whereas before, they'd have five, six, seven different systems, even if it's from the same vendor, to be able to run the same wide array of different workloads. We can now do that with one environment. And then finally, we operate that as a cloud of storage. As opposed to individual arrays, it looks like a cloud of storage. The customer manages it as a fleet or with our Evergreen//One program as a subscription to storage that we manage entirely.
So really, we're really at the forefront of changing the entire concept, if you will, of hybrid enterprise data storage, taking it from, you know, old school, mechanical, highly physical environment, really to a modern cloud operating model, where the customer is able to subscribe to their storage, to operate their storage the same way they would in the cloud.
Okay, great, great. Good, good stuff. Covered a lot,
Sorry. Okay.
No, no problem. You know, one of the, one of the things I'd love to get both your perspective on is kind of macro impacts on the business. We're seeing it across different product categories and enterprises, but bit of a challenging time now. Could you just talk a little bit about kind of macro impacts, visibility? Obviously, you've got some share gain dynamics that's a little different than many of the other companies playing in the area, but just curious how you, how you see the backdrop now.
Yeah. You know, the macro certainly hasn't made anything easier this year, you know, in the and certainly in our environment in storage. On average, you know, the storage market is a bit down this year. Not unusual in the early phases of a tough economy, where customers will generally sweat their assets in the storage space first, but nobody ever throws out their data, so eventually, those arrays fill up, need to be replenished. So that eventually catches up. In the early phases of a poor economy, you do see the storage market tend to suffer a little bit more than others. Now, on the positive side, it's accentuated our sales of Evergreen// One. Evergreen// One, again, is our subscription model.
So whereas, you know, traditionally in a CapEx world of storage, customers would buy an array at a scale that would last them about 5 years. With our Evergreen model, they just subscribe to a service level agreement. It's a purely consumption model. They only pay us as they use as they start to consume storage. By the way, whether it's on-prem or in the cloud, we operate in both environments, and it's a pay as you go. But of course, since it's a consumption model, the revenue is realized over the term of the contract, which tends to be about 3.5 years.
And the only thing I'd add on the macro that's been helpful for us as well is the coming out with our FlashArray//E.
Yep.
It's, you know, it's price performance solution, so now we... You know, for customers who are a little bit more sensitive, from a price standpoint, as they're navigating more challenging spending environment, we've seen some really strong traction on our family, including now FlashArray//E, which has just been recently GA'd.
Okay. Pretty great. Maybe digging into the kind of as-a-service offerings. Obviously last quarter, the last few quarters have been impacting the top line. As you said, Charlie, it's instead of a CapEx up front, it's amortized over like, you know, I think you said 300 basis points impact this year. Could you talk a little bit about how you see it progressing? So a lot of the storage world and hardware world is CapEx.
Right.
But do you think this is the beginning of a really major trend? Maybe you could take that, and then, Kevan, walk through your view of the dynamics on how this is going to impact the model in the next few years.
We do see it as the beginning of a trend in the following sense. We've had a variation of this, we call it Evergreen//One. We've had actually several names over the last 5 years. But this as a service model, we started about 5 years ago, and we've been consistently investing in it, improving it, making it more cloud-like as we went along. It started out primarily as a financial model, that is, for customers to be able to pay over time rather than right up front. But increasingly, it's a cloud service or operated as a SaaS service for our customers. They no longer either buy, rent, or lease a product. They subscribe to a service level agreement. The products belong to us.
We are in complete control of the way they are configured, of the way they operate. The customer manages it entirely through a web interface. So it is now a storage as a service model. I will say that for the first several years, it was very new for most customers inside their own data center to think about doing anything as a consumption model, because they were buying everything else CapEx. They're buying processors, CapEx. They're buying networking CapEx, power supplies, everything else.
Really, we were the first into this market saying, "Well, you should consider doing this on a consumption basis." And frankly, in many cases, it was the finance department that said, "You know, we're not really set up to do that, so we'll go to the CapEx." Again, five years later, we're seeing... And especially as a tough economy comes in, and the finance organization is saying: Well, you know, we don't want to spend the cash, or we don't want to commit to so much, you know, depreciation, you know. And so, and we saw this in the last, like, at the beginning of COVID, with the downturn of the economy, we saw a progression towards Evergreen//One.
And then, as you know, when the whole supply chain, the economy picked up, the supply chain with all the stimulus, the supply chain challenges came into place, and we saw the customers go back to CapEx. So, you know, as we went into this year, our year starts in February. As we went into this year, we saw or we imagined, because the economy had taken a downturn, we imagined that Evergreen//One would grow faster than we'd seen the year before. We were just surprised, you know, later in the year. Let's just put it another way. We underestimated significantly the take-up on Evergreen//One.
Yeah, and I'll, I'll go into a little bit more detail on that, Charlie. I think the other thing to highlight, though, is the focus we've had on service level agreements.
Yeah.
Do you wanna-
Yeah.
Provide some color on that, and then I'll-
I do
... dig a little bit more on the economics.
So, you know, like most SaaS environments with the hyperscalers, customers do not, you know, buy, lease, or otherwise rent infrastructure. What they do is they, they negotiate a service level agreement that is how much performance, how much capacity, over what time frame, and what are the guarantees, you know, in terms of those items that the vendor is providing? We do the same thing, and we've upped those service level agreements. We guarantee no application downtime because of, you know, any change management that we make, might take on. We guarantee power, space, and cooling. In fact, we pay for power, space, and cooling now, even if the product is on the customer premise, just like any SaaS service pays for their own infrastructure and labor.
We guarantee reductions in labor of the customer, because when they buy CapEx, they typically manage the entire environment. Well, we guarantee them we manage that environment. We guarantee, you know, performance in terms of resilience, no downtime. So there are a lot of guarantees now that we provide, standard.
Yeah, I appreciate that, Charlie. And then, you know, when we think about it from a transition period, look, I mean, our foundation, again, is built on the Evergreen architecture that Charlie talked about. And so when you think about a traditional CapEx, it's always been attached to an Evergreen subscription. So a traditional CapEx arrangement, it's got about 30% of its value attached to a subscription, and that's why, you know, nearly 40% of our total revenue is subscription today, because it's really been built on the Evergreen subscription and the Evergreen architecture. What we've done is evolved now to a full service level agreement with Evergreen//One, which is complete services, and there's no ownership of the asset. We own the asset.
And that's where we're seeing this inflection point in terms of a preference, especially this year, on choosing a complete service level agreement to consume our technology versus a traditional CapEx. And that's why we've talked about it, and this is you know, obviously, we've had the storage as a service offering for some time, several years now. This is the first time we've really seen inflection point. I think to Charlie's comments, piece of it's macro, but also a piece of it is really around the technology, the service level offerings, the guarantees we're making, the idea that a customer just does not need to worry about operating the storage environment. We do take care of that for them.
We have a ton of experience doing that as a result of our Evergreen offering that's been in place for over a decade. So then the question is: Well, what is that bias, and do you prefer Evergreen//One storage as a service or traditional CapEx? Absolutely. It's offering for Pure, it's a great offering for our customer. We've always been looking to make the Evergreen offering a preference to our customers. We entered into this year, we revised our incentives to ensure sellers get paid a little bit more money to sell an Evergreen//One offering versus the CapEx. But to Charlie's point, the growth just completely outpaced our expectations this year.
As I look at it next year, in terms of what we're looking to do, we're going to continue to emphasize our storage as a service offering with Evergreen//One. We'll again step up the incentives. We'll work with our channel partners to have them get paid a little bit more money to preference the Evergreen//One over traditional CapEx, given the value we see on both sides, from our perspective and from our customer's perspective. And as long as that transition, when we really think about our sales today, the majority of our sales are traditional CapEx versus Evergreen//One. So there'll be some time as we're working through that transition, but it will never be 100% in terms of moving completely to storage as a service.
We have a lot of customers who will always consume our technology through CapEx style customers, SaaS customers, hyperscalers. You know, these are the types of customers that will always consume the technology generally via traditional CapEx. So what we'll do next year, as we've seen this inflection point, is give you a view not only on what we're expecting for revenue growth next year, but also what we're expecting our Evergreen//One, as well as our Evergreen//Flex sales to be next year, and then translate that to a traditional CapEx. You've got a nice view of what our company growth rate would be expected to be next year, if it had been all traditional CapEx. And that just gives you a nice feel in terms of the health of our business and the momentum of our business.
Okay, great. Maybe we've got to talk AI anytime we're doing a tech conference. So talk to us a little bit about it. I know you've talked about wins in the past. What's the differentiation there? How do you see, you know, Flash working its way into AI data centers, and particularly branded, non-white box Flash, so-
Right.
You mentioned that. So, just to provide some context, we've been in the AI market for about five years. We have a consistent business in the AI environment. Now, of course, ChatGPT, GenAI is a relatively new phenomenon, and that's all just starting to play out now. You know, our involvement in AI really spans, and this is, I think, what is most surprising, I think, to audiences. Our involvement in AI spans the price performance range. It's not just the highest performance systems. And we have our FlashArray, sorry, FlashBlade//S 500 series, which is one of the highest performance file systems in the world. It is what feeds the massive number of GPUs at Meta's AI Research SuperCluster.
and continues to do well in, you know, in training environments. But even in that Meta example that I just gave, only about 15% of the roughly 0.5 exabytes that we've sold into that is that high performance system.
... 85% of what we of that 0.5 EB is actually what you'd call a warm data system. It appears to be our FlashArray//C, could easily be our, our E Series, except that we did this some years ago before the E Series. And this, I think, really is a, an example of how I think AI is going to affect the storage market. And that is, yes, there's gonna be an exciting piece at the highest performance level, but there's gonna be a much larger piece at upgrading everyday cold storage to warm storage. You know, you were talking about a vast majority of storage, you know, still being not at the speed that is required for the new GPUs.
But since you wanna get access to your data across the board for inference and for, you know, answers to your prompts, you need that data to be available. A disk-based storage environment, barely able to keep up with what it's connected to, you know, within a flash environment, all that data becoming ready. So we see AI as the opportunity to upgrade storage across the board, and you're talking about a $50 billion market, at, you know, a lot of which, majority of which, has to be upgraded from hard disk, to, to Flash. So I think it opens up really, in a sense, two opportunities for us.
One is to continue to grow in the training environment, the high-performance systems, but in some ways, you know, from a pure revenue standpoint, even more exciting, upleveling the capabilities of the general purpose storage market.
Okay. And how, how broad you mentioned Meta, but how broad are your wins that you would, you are, qualify as AI? How, how big is that business or customer base, and is it, is it kind of inflecting now, or is it still early?
I would say it's a little bit early, but really see inflection. It's a good, steady business for us. You know, we haven't really given metrics on it, but it's, you know, it's a good segment, and I wouldn't say it's one of the larger segments.
Okay.
Yeah.
Maybe if we could talk telco a little. It was one of the other, you know, hiccups last quarter, but it sounds like your second very large deal. Maybe just talk a little bit about what you're seeing in the service provider environment and how you, you know, how you're identifying the trend in that end market.
Service provider is a really good segment for us. It's one of our, like, four or five largest, if you will, verticals. And more recently, I think last two years, we've been making really good progress in the 5G space. So typically, we had sold it, their IT environment. Now we're actually getting engaged in their network environment, specifically with 5G. And a number of the advantages I mentioned early on, really play well for us in that environment. I mentioned one-fifth of power savings and cooling. That works out really well in a highly distributed 5G environment, where they don't really want to be sending people, you know, to the on location to be making upgrades. The fact that we can manage it all remotely, again, very, very good in that environment.
I didn't mention the fact that literally, over 10 times more reliable than our competition, that's very good. One-tenth the amount of labor required to manage the same amount of data. Again, very good in a highly distributed 5G environment. All of it to our natural advantages.
Okay. Great, great. You mentioned FlashBlade//E and FlashArray//E. Just give us a little, you know, update on how you know initial feedback on FlashArray//E and, you know, how much of a ramp we see FlashBlade?
Yeah. So we're gonna just characterize it now as the E family.
Mm-hmm.
We only just introduced FlashArray//E a few weeks ago, this last month. So that's just getting started, although pipeline looks very good. We just talked about the E family. You know, it has the highest growth of any new product we've introduced so far. Only two quarters of sales. You know, we only have two quarters of sales really in place, but still very fast growth. It's a very new motion for our customers. They're used to buying disk-based systems in this environment. So, you know, convincing them that Flash is ready for their environment. Actually, if you don't mind a small anecdote, two quarters ago, I did something I never, ever thought I would do on an earnings call, and that is, I announced the price of a product.
The reason for selling price of a product, not even the list price, but the selling price. You never want to do that on an earnings call, because what you've just done is identify your ceiling, right? You've got to negotiate below that. Why did we do that? We determined that nobody would believe that we could sell Flash at that price in the marketplace, and we had to really shout from the rooftops that we can now have Flash at disk pricing. And that's why we made that announcement. So we're seeing very good traction, although early traction. I think, you know, next year. And again, we introduced it after budgets being set for the year. So I really think next year is we're gonna see continued good growth in that area.
Okay. Maybe talk a little margins for a little bit, both gross and, and operating. There's been some good leverage in the gross margin, you know, particularly last quarter is really, really strong. So how do you see the gross margin line working with commodity costs, et cetera, and the move to as-a-service? And then what about the operating line? There's obviously a lot of investment still in go-to-market and R&D with all the, you know, products you've got off. So update that.
Yeah, I'll take that. I mean, look, when we think about, your gross margins, obviously, we need to talk separately about product gross margins versus our subscription services gross margins. Product gross margins are very strong, and that's really a testament to the architecture, leveraging our Purity with native Flash and our DirectFlash modules. This marriage that we've created between our software and our hardware really creates that advantage, and we see the benefits of that. Now, obviously, pricing of flash helps as well, including the fact that we're negotiating directly with flash vendors, multiple flash vendors. You know, obviously, the majority of our capacity shipped is QLC, but we also have TLC, so we have the ability to arbitrage to the extent we need to to maximize cost.
But one area that we do wanna accelerate, obviously, is the penetration and take out of disk with our E family. And we, we do plan to be aggressive from a pricing perspective to accelerate that trend. And that's an area where that might have a little bit of an impact, which is good from our view on product growth margins. So I would see that ticking down slightly would be our ideal view on product growth margin. Now, I've said that for the last few quarters, and Charlie keeps asking me what's going on, so I've, I've got some work still to do on that front.
But again, I really want to see that acceleration, because to Charlie's point, our E family, and primarily so far, FlashBlade, has been our best-selling product, both from a bookings and selling perspective, but also from a pipeline perspective.
Yeah, from a growth.
From a growth perspective. Yeah, from a growth perspective. Okay, so then if I take subscription services, that's why I think I've got some upside over time in terms of modest expansion. We have terrific margins that we've developed on our Evergreen subscription attached to CapEx. That's generally our Evergreen//Forever service offering. We do expect that we'll continue, as we continue to scale Evergreen//One, those margins will be, in fact, better than what we can actually do with our Evergreen//Forever, that's attached to the CapEx. That'll be over time as we scale, but generally, we're taking the leverage that we're getting on the product side, because it's our infrastructure, so we're taking full advantage of that, as well as scaling and automating the service capabilities around Evergreen//One.
I look at operating margins in totality. You know, Charlie and I have been very aligned, in terms of our goals, which is, hey, we want to prioritize top line growth. That's our priority, and will continue to be our priority, but with modest expansion in terms of our operating margins. With modest, we're looking at 1-2, is how we're thinking about it.
I'll just add, you know, I think if you go through, you know, our historical quarters, you find that, and you try to correlate to NAND's pricing, that it doesn't affect our gross margins that much. What it does affect is top line. And so our top line growth, I should say. And so we don't expect, you know, the NAND price changes, which have been, you know, fairly substantial over the last couple of years. We really don't expect it to affect our gross margin very much.
Okay. We have, like, a minute or two left. Maybe you mentioned Meta SuperCluster. I think in the past, you've talked about other kind of large-scale, hyperscale opportunities. Where are we in, you know, adding some other others to that?
It's always a little difficult to talk about these things. You know, it's like talking about a potential acquisition.
Yes.
You don't know until you know-
Right
... everything is dry, you know, whether something like that is actually going to happen. So I want to make that as a clear caveat here. But the goal I've set in front of the team, and I think and I'm pleased with the progress we're making, is to see some type of design win within the next year. Yeah.
Okay.
Yeah.
Is the expectation that would be as a service or probably a CapEx or maybe-
It's probably going to be a very different model from what we have today, hard to, hard to qualify at the moment.
Well, and that's what's really important to understand, is our business model can be very flexible. We can go to a full solution, we can do all software, we can do a combination of both. And however that business model gets shaped with a hyperscaler, we have that flexibility based on the,
To be really clear, we're not going to do a deal that's not operating margin accretive.
Right.
Right? So, but that, and that's really the test. However we structure it, operating margin accretive.
Okay. All right, great. Thank you, guys. Really appreciate it.
Thank you.
Thanks.