Awesome. We're going to get started. Not a lot of turnaround time here. My name is Erik Woodring. I lead Morgan Stanley's U.S. IT Hardware coverage. Please be joined by George Kurian, CEO of NetApp. Before we start, quickly, before we begin, please see the Morgan Stanley Research Disclosure website at www.morganstanley.com/researchdisclosures. From the NetApp side, today's discussion may include forward-looking statements regarding NetApp's future performance, which are subject to risk and uncertainty. Actual results may differ materially from the statements made today for a variety of reasons described in our most recent 10-K and 10-Q, filed with the SEC, and available on our website at www.netapp.com. We disclaim any obligation to update any information in addition to forward-looking statements for any reason. I'm again delighted to be joined this afternoon by George Kurian, CEO of NetApp.
I think, George, maybe the best place to start. You know, I know last week at a separate conference, you kind of discussed the October quarter pretty thoroughly. Instead of doing a review, maybe just like, what are two or three highlights that you want to leave the crowd with here, key takeaways from the quarter or the January quarter guide, full year guide, everyone should kind of be aware of as kind of a starting point for our conversation?
Yeah, I think, first of all, thanks for having me. Thank you for being here. In Q2, except for Spot Business divestiture, we grew 4% year-on-year. We had continued strong momentum in our growth engines, first-party and cloud storage, which grew 32% year-on-year, all-flash, which grew 9% year-on-year. And we doubled the number of AI wins in the quarter. Product gross margins were above our expectations, driven by mix and continued prudent management of costs. Operating margins, gross profit margins, and EPS were all records for Q2. And we have strong expectations going forward where our business outside of U.S. public sector performed extremely well, showing our competitive position.
OK, perfect. And we'll kind of touch on a number of those points. I want to start just high level and really focus maybe first, like it's almost a little technical, but you and your peers all kind of had analyst days, product days over the last kind of month and a half. What we tend to hear is, you know, single pane of glass, we have the best software, ONTAP is world-renowned. And so I would love if you could maybe just help us understand where do you think NetApp is truly differentiated and how do you leverage that differentiation to try and drive sustainable share gains?
Yeah, I think we have pioneered many of the ideas that are coming to fruition in the customers for many, many years. From the early 2000s, we have been believers in the idea that you want to unify your data across all your departmental boundaries, across all of your different data types, so that you can actually have not only better management of your data, but the ability to extract insight. That is now de facto the norm that everybody in the industry is trying to get to. The second, when cloud happened, we always believed that the world would be hybrid and multi-cloud.
And so we built a hybrid multi-cloud architecture where our software is now deeply integrated into the major cloud providers, Amazon, Microsoft, and Google, in a way that, you know, it's a decade-long journey that we have undertaken and are well ahead of anybody else in the market. And then the third has been that in addition to having super high-performance scalable systems, which is sort of a basic requirement that, you know, many vendors can meet, we have a lot of data management value in terms of how people can exploit the value of data sitting in their systems.
This could mean, you know, advanced cybersecurity functions, the ease of management of data across hybrid cloud estates, and increasingly a suite of tools that allow you to discover where your data is and to build super efficient pipelines so that you can take data from your enterprise applications and apply AI to it.
OK, perfect. So that's a good start. Let's kind of go higher level and turn to kind of the market opportunity. Can you just help us maybe understand two things? One is kind of how you think about the market growth of your end markets, and maybe you can separate cloud and on-prem stuff. And then, you know, you've been clear that there is kind of a macro overhang on spending. Storage spend has lagged kind of server and compute spend. What needs to change to kind of get storage spend to kind of catch up to server a little bit?
I think broadly speaking, IT infrastructure spending is correlated to macro and within macro specific business outlook. And we have not seen a major refresh of the infrastructure deployed in customers since 2018, 2019, the first, you know, Trump administration's, you know, favorable tax structure and cash repatriation period. What that means is customers typically prioritize spending on infrastructure based on their application needs and their workload needs. And we are seeing a rotation of that workload mix into AI-specific applications, which we can talk about. I think with regard to the compute cycle, it has been driven by AI. And I think what you see in the AI world is you have a lot of work to get ready for AI, you know, all the model training, getting the models to a place where they can be applied to the data. That's what's been going on.
We did not expect storage to uptick at the same pace and timing as compute.
Right. OK. Before we turn to AI, I just want to quickly touch on U.S. public. And the question really is, at least what our view is, is that, you know, the government will always find a way to spend its own money. We just need to get kind of that engine moving again, get appropriations moving through the right agencies. How do you think about government spending kind of coming back post the shutdown? And what are some of the signposts that we should be looking for as maybe like a precursor to that unlock?
Yeah, I think we look at U.S. Public Sector as a combination of U.S. Federal and U.S. State, Local, and Higher Education. For our business, U.S. Public Sector is low double-digit percentage of the total business, of which roughly 75% is U.S. Fed. Within U.S. Fed, there are probably three or four things that we are part of and that we track. The first is multi-year appropriations. These are program spend vehicles. It could be defense programs. It could be infrastructure programs or national security programs. We are embedded in a lot of them, and they are immune from the annual appropriation cycle. And so our goal there is to capture more and more of those program dollars. And we have pursuit teams that have been successful doing that.
Then within the, you know, when an administration changes, usually the first year of an administration change, we watch for their spending priority shifts between, let's say, civilian and defense or, you know, any other kinds of budget shifts. And we move our resources to where the targets are. We saw some of those shifts within this administration as well. And that's part of the natural cycle of, you know, federal spending. I think what we track now is the movement of, you know, appropriations to agency spending programs and the predictability of those spending programs instead of translating from awards, you know, to actually orders, right? And I think on the first one, in terms of appropriations being translated to spending, I think there's still work to be done. OBBB is still flowing down into the agencies. And some of that was disrupted because of the shutdown.
And then the second, you know, while there were some challenges, notably with DOGE earlier in the year, those challenges have, you know, kind of abated. And so we are hopeful that by the springtime, the U.S. government's back in business. I think that we have been cautious about our Q3 because of just how long does it take for the government to come back to a normal course and speed.
OK. OK, perfect. So that's kind of a good segue. Get that out of the way. Turn to AI. You know, your point, George, kind of resonates when you talk about, you know, we didn't necessarily expect that to come right now. Can you maybe help us understand what is the opportunity in AI? I think it's simplistic to say, well, there's just more data that will be generated, more data that will be stored. But, you know, is this something that you believe happens on premise, but also you kind of have this hybrid model? So how do we think NetApp benefits from that? So kind of how does AI permeate the opportunity? And then what is kind of the hybrid opportunity for NetApp with AI?
Yeah, we can capture value from data and data growth wherever it is because of our hybrid model. I think what we see is sort of three buckets of use cases within the enterprise, which is really our focus. You see sort of data prep and getting data ready for AI. You see people saying, hey, I want to take my models that are, you know, derivatives of big foundation models and kind of distill it or make it useful for my enterprise. And then you see people actually applying the two to actually run their business with AI, which is what is called Inferencing or RAG. I think for storage, data storage growth, the third bucket is the big bucket over time. And I think for enterprise productivity gains, it is also the big bucket, right?
If inferencing is not useful for the enterprise, AI is more fluff than wave, right? And so that's what I think everybody in the industry is tracking. We have seen the number of AI projects year-on-year double. So we said approximately 200 wins in the quarter. These are wins that have 45% of data lakes organizing my data to get ready for AI, 25% around training and fine-tuning, whether those are sovereign, you know, AI models or whether it's an enterprise model that's a distillation of a large foundation model, and 30% in inferencing. And our view is for AI to be as successful as it is, the inferencing part needs to grow steadily.
Right. So maybe you said differently, we're just early. And we might be seeing training happening in the cloud. We're at the early days of inferencing and kind of token growth. What do you think catalyzes that kind of within the enterprise on premise? Is it moving away from productivity gains to like creating new insights and business opportunities? Or can that kind of inflection on prem still happen with just kind of the productivity approach to AI?
I think whether it's on prem or cloud, the real value is the ability to get, you know, both unique insights as well as new ways of working that allow you to do more and bring more innovation, meaning more growth faster. I think in terms of new insights, we are seeing amazing examples in health care and life sciences where there's everything from drug discovery to disease prediction to absolutely high precision medicine to better, much better diagnostic and therapeutic care. And so I think there you are seeing transformative use cases now. I think with the others, there are, you know, several examples around, you know, ways to bring new capabilities to market, software development, for example, digital twins and immersive digital, you know, manufacturing. There is, of course, synthetic semiconductor designs where we are part of, you know, lots of large, you know, customers' deployment models.
But I think ultimately for it to really change how, you know, businesses operate, it either needs to drive growth or really transform the way work gets done.
Can you maybe, let's say it this way, I'd love to dig into those 200 wins, the 200 projects, AI projects that you talked about. Obviously, you just helped bucket them for us to understand kind of what you're doing within those. Can you maybe give a few granular examples of like what I would call keystone wins, not the keystone products, but flagship wins and exactly what the value add is that NetApp is doing underlying that project?
Yeah, I think, you know, let's talk about data lakes, right? I think there are what a data lake is, is a customer says, I want to bring data from multiple application landscapes into a single environment so that I can then organize my data and apply large language models into that controlled environment. We have really sophisticated tools for, you know, unifying unstructured data, structured data on prem and cloud. And so we have examples of in life sciences of customers that say, I want to bring research data, clinical data, lab data all together across research teams in multiple geographies. We can do that in a way that nobody else can. With regard to training, we have examples of large, you know, country-level language models where customers say, hey, I want to, you know, train this model. I want a trusted enterprise, you know, vendor.
I've got, you know, good experience with NetApp technology. It's got the performance and scale and predictability in terms of resilience to support my environment. And then in inferencing, listen, we hold an enormous amount of the world's unstructured data. And so when somebody else tries to start to talk about inferencing, they say, oh, let's copy all of the data into another environment to do inferencing. We said, just leave the data where it is. We'll help you do inferencing right where the data is. And a lot of the innovations we announced at our customer conference was to just say, hey, keep your original data where it is. We'll allow you to create a, you know, derivative of it for inferencing in place with all of the security permissions lineage tracked, which makes it much more a much lower bar for customers to adopt.
I'd be remiss if I didn't kind of dig in with you or at least ask you about the all-flash transition because I believe today two-thirds of your hybrid cloud business is already all-flash. You've obviously been able to benefit from this with growth and share gains. Does that two-thirds go to 100? Or like how do we think about all-flash mix going forward over, call it, one to three years? What does that do to your growth rate, do you suspect?
Yeah, I think two things there. First is, you know, two data points. One is as a percentage of hybrid cloud revenue, all-flash was two-thirds. As a percentage of our install base, it's 46%. And it's been ticking up roughly 1% a quarter, which is an indication of the size and scale of our install base. And both numerator and denominator are growing in the second, right? I think that, you know, flash will continue to grow as a percentage of the total out of the factory shipments steadily over a period of time. Disk, with the data that we have, we don't see all-flash becoming 100% of the use cases. We see that there's a steady place for hard drive-based storage, especially for workloads that are either sequential workloads or workloads that just need a lot of capacity.
I think so it'll tick up as a percentage out of the factory. I think in terms of the install base, you can do the math. It's going to be many, many years of refreshes of hard drives-based systems out of the install base before we even reach 60%.
I should have asked you this earlier, but your point on the installed base refresh last time being over half a decade ago, what do you think catalyzes that? Is it kind of a greater prevalence of AI or is there something else? Is it just the macro getting a little bit more certain to where CIOs can say, you know, I'm ready to partake in this type of refresh activity? What is it, do you think, that catalyzes that refresh?
We are seeing workload-based refreshes going on right now. These are project-driven, you know, spend where someone says, hey, I've got this data set and this infrastructure. I want to bring it, you know, into my AI landscape. I probably need more modern systems. I was here yesterday. We were actually talking to a large, you know, financial services client about doing exactly that for fraud analysis and, you know, risk mitigation. I think for a broad-based infrastructure refresh like we saw in 2018 and 2019 where the customer said, listen, I want to modernize my entire estate, that's going to require a more constructive macro environment and/or within specific customer circumstances, a more constructive outlook for their business, multi-quarter, multi-year outlook.
OK. OK, perfect. If we kind of summarize all of this and take into account refresh opportunity, all-flash mix, early days of AI, you just I would love for you to take your fake magic ball and look out and just say like, what do we think this should and/or could mean for NetApp when it comes to growth and margins and free cash? Obviously, we'll touch on public cloud shortly, but just high level, what does all of this do for business fundamentals at NetApp?
Yeah, I think our long-term model was, you know, mid to upper single-digit growth on the top line. We should, if you look at the year-on-year compares for the second half of the year, excluding the Spot Business we divested, you're starting to be in that range, at the bottom end of that range. And so we feel good about the acceleration of our business in the second half versus first half. We hope that public sector becomes a more constructive environment, you know, next fiscal year. And so that's on the top line. Our goal is to drive mid to high single-digit growth. We are, you know, we've got many secular drivers that allow us to participate in that, you know, flash, cloud, AI. You know, we're taking share in flash from, you know, our competitors, both structured and unstructured data workloads.
Our cloud storage business, which is now the preponderant majority of our cloud business, is growing, has been growing north of 30% year-on-year. The mix of our business is shifting into higher quality business. Both margins are structurally much higher than they were in, you know, in the prior model, as well as the mix of kind of subscription and consumption-based offerings is higher. And so that's top line and margins. I think in terms of operating profile, listen, we've been disciplined operators of the business. Our last reported quarter was north of 31% operating margins and pretty disciplined, you know, allocators of capital where a majority of our capital dividends the first call on capital. And we've returned, you know, in the range of 100% of free cash flow to shareholders through dividends and buybacks over the last several years.
You kind of touched on it, but I want to make sure we hit on the public cloud business. Two questions is, what do you believe? I think the business in totality grew 18% in the last quarter, adjusted for the spot divestiture. Like what is how sustainable is that growth? What is driving that? And then obviously margins are very strong at 83%. Where do those go?
Yeah, we said that. Let me get margins and then we'll talk about growth. I think we said that, you know, margins would be between 80% and 85%. We raised the range and we are ticking along nicely as the mix of, you know, offerings shifts more and more into software, and as the depreciation from the original capital investment, particularly in Microsoft Azure, comes off the expense line, and so we see, you know, strong trajectory on margins. You know, could it go over 85%? It could, right? We're not sort of saying that we will adjust the range upward, but is there something structural that prevents it from ticking upward? No. I think with regard to what's driving growth is, you know, both the overall growth rate of cloud, which has been strong, but also the differentiation and the continued addition of new technology capabilities to our infrastructure.
So we started out with a narrower swim lane. We are now in a much, much bigger set of swim lanes and opportunities with clients and with the hyperscalers. We started out with their public cloud environments. We are now part of sovereign cloud, disconnected cloud, distributed cloud. So we are in a huge range of use cases of those hyperscalers. We have done well in enterprise workloads where it's purely now scaling the go-to-market routes. And you have started to see us. You will see us do a lot more with their AI workloads over the next 12- 18 months. You should look at the Amazon re:Invent announcement that Matt Garman made about the integrations of our technologies with a huge range of AWS's AI apps. And there'll be much more to come. So stay tuned for the conferences this coming year.
Okay. Cool, cool. We have a handful of minutes. I just want to quickly, you know, it took me whatever, 10, 11 questions before we hit on memory, but I'd be remiss if we didn't bring it up. Obviously, you know the backdrop better than we do. You know, maybe the first question is historically, actually in storage, you've been able to pass through these higher input costs. Is there anything that would not allow that to happen this time again? And again, the question is, you know, how does that impact kind of demand elasticity as you're thinking about it in the environment today?
There's no reason why customers will not be asked to pick up higher input costs because we have passed that through to them in the past when it goes up, and we've passed the benefits of lower cost to them as well when it went down. With regard to, you know, elasticity, customers budget in dollars and there historically has been very little elasticity, positively or negatively due to input costs.
Right. OK. So maybe before we end, I just kind of want to give you the dance floor because we've talked about a lot. We've talked about the market, AI, share gains, technology, memory, cloud. You know, just maybe what's the final message you'd leave for everyone if we think about where NetApp is in your journey, but also where you are from a stock perspective? Kind of what's the message that you want to send, either what's underappreciated or where are you leaning into and what's the opportunity for these guys that are sitting in the room with us?
Yeah, I think first of all, data is an ever-growing asset to the enterprise. Without data, there's no AI. And so especially unstructured data, which has historically had limited insight derived from it, is becoming now a much bigger wellspring of, you know, value to clients. We hold a huge amount of the enterprise's data in the world. And we are now, because of our unique capability to do so across multiple cloud environments, very much seen as a data platform provider that can unify all of these estates. We are adding a lot more software functionality that allows us to differentiate in cloud, in cyber, in flash-based storage that gives us, you know, confidence that we should be able to grow above the market.
The growing parts of our business are an increasing share of the total, while the declining parts of our business are not only smaller parts of our total business, but also stabilizing in terms of the rate of decrease. And then I think below that, listen, higher quality revenue, higher margin components of our business, good ability to manage, you know, sort of gross profit dollars, disciplined operators of the business, which should give us earnings leverage going forward.
Awesome. We are just out of time. George, thank you very much.
Thank you.
Thank you, guys.
Thank you for coming.
Thank you.