Good afternoon, everyone. Thank you again for joining the UBS Tech Conference. I'm David Vogt. I'm one of the tech analysts here at UBS, and we're excited to have George Kurian from NetApp here, Chief Executive Officer. Before we kick off, Kris has a disclaimer she needs to read for all of us.
Very important information. 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 statements made today for a variety of reasons, as described in our most recent 10-K and 10-Q filed with the SEC and available on our website at netapp.com. We disclaim any obligation to update information in any forward-looking statement for any reason. Back to you, David.
Thanks, Kris. Thanks again, George, for coming.
Thank you for having me, and thank you for coming.
So we just came off of earnings last week, so I think this is incredibly timely. What I thought we should do is maybe start with kind of just a quick overview of what you saw last quarter. Results were stronger and gross margins strong, despite, I think, people's fear about commodity prices, other competitive issues. So maybe we can start there and talk about kind of what you saw in the quarter and how we're thinking about the January quarter and beyond through your fiscal year.
Yeah, I think when we entered the quarter and guided the quarter, we said we were concerned about softness in EMEA, driven by unsettled government structures in some of the larger countries, overall GDP outlook, and then U.S. public sector. As part of the results on the top line, the European team, while GDP continues to be choppy, we have out-executed really, w e've done a super job in Europe. I think you can see that in the market share data, where not only are we number one in many of those countries, we have opened up share gaps to our competitors quite nicely in some of the bigger markets. So at the halfway point in the year, Europe, we feel better, and our outlook in the second half reflects that. U.S. public sector, conversely, has performed materially worse than we expected going into the quarter.
The guide for the quarter was cautious, but it reflected the public data that U.S. government spending through the fiscal year was super back and loaded. And so Q4 of the government calendar, which overlapped with our Q2, was expected to be a bulge spending quarter. And then with the shutdown, that kind of has pushed out. It cannot be pushed out forever, right? So it'll come back. I think we are cautious about when it'll come back. So Q3, we're cautious about Q3. We're hopeful it's better than Q2, but still pretty cautious. I think with regard to the competitive landscape, listen, our flash business has performed well. Our cloud business continues to run unabated. Our first-party cloud storage grew 32% year- on- year, and gross margins are super strong in the cloud segment at 83%.
I think if you look at hybrid cloud, the product gross margin beat was driven by mix better than we expected. The cost structure through our structured pricing agreements with NAND providers were part of our outlook.
Got it. So I want to come back to margin in a second and maybe just touch on demand drivers for a second. So obviously, state and local was a little bit better, I think. U.S. federal worse. Is that a fair characterization?
I think overall it was really tough.
Overall, it's really tough. But to your point, you noted that the dispersion, the money's there. People have to invest.
Correct.
So what is sort of the kind of mechanical process to really see that money start to flow into programs and projects that would benefit you? Is it just, hey, the person needs to be in the seat, the correct person needs to be in the right seat? The assets have been swept so long that they no longer can defer or delay, push out. Is there anything that you can point to to help us understand how we should think about? Obviously, we don't think it's going to come this current quarter, as you pointed out, but as we move through the subsequent quarters next year.
Yeah, I think we are speaking to every kind of connected party in Washington, D.C. And so far, the thinking is the springtime is when all of the pitchers and catchers get aligned so that you can get back to some degree of normalcy in spending. I think the two big buckets are the appropriations need to flow down to the agencies. I think that's at various stages of kind of the receipts of appropriations being available to the agencies. And then within the agencies themselves, the right priorities in terms of what programs are they going to spend, what IT priorities are they going to spend on, I think both of those need to fall into place.
And does that business coming back? Your all-flash business was very strong again. Prior quarter it was strong, ex-U.S. federal public sector. Are they heavy consumers of flash at this point in the cycle, or are they still sort of a mixed customer in terms of buying legacy drive systems, or is it mostly an all-flash solution that they're looking for at this point?
They look like other customer segments.
They're similar. Okay.
Two-thirds of our business in hybrid cloud is flash. If you do the math on if U.S. public sector were just flat, not up, right, you can see the acceleration of what I've given to our all-flash business.
Got it. Outside of public sector, obviously, we've had reasonably good demand signals from other vendors. I think we had one earlier today. One of your competitors talked about their first-party IP starting to really accelerate. Do you think the market overall is healthy besides just what you're seeing, or are you taking material enough share that you're not seeing some of the, well, let me rephrase that. You're not seeing some of the growth dynamics that they're talking about acceleration, but they're growing far slower than you. So is that share-driven? Is it product mix driven? How do you think about where you're positioned? Because you've taken a lot of share in the last couple of years.
Broad-based, I think that macro drives business outlook and IT spending. That's been the consistent theme. I think macro is still choppy. When we look at what that means is we have not had a broad-based storage infrastructure refresh since 2019, right? It's not like compute, which saw the new buildouts of GPUs. Storage has been muted for a long time. So what you see in that dynamic is two things. One is, hey, the projects that customers are spending on, those get storage spending as part of those projects. And then so you got to win your share of those projects and be in the swim lanes that have spending. And the second is you got to take share.
In the swim lanes that we are focused on that are seeing good traction, I think one is data prep and data infrastructure modernization for AI is showing good strength, and our leadership in unstructured data gives us a strong entry point into that sales motion. The second is our technology leadership in building cyber resilience directly integrated at the storage layer, unlike our competitors that all rely on third-party applications, is giving us the ability to go back and have conversations with our installed base and net new customers that says, "hey, do you not want to have your production data sitting on a cyber resilience system." Cloud has been sort of immune from this. So our cloud business is really about completing more certifications for more workload. That's on fire. It's been rolling for a good, strong period of time. Those are the key buckets.
Can I go back to the comment you made about compute seeing a little bit of a cycle recovery? If I look at compute historically, pre-AI, storage and compute were generally attached at the hip, if you will. We haven't really seen that. Why do you think that's the case, and would that be additive to your outlook or your expectations, not just next year and the year beyond, if we do see a healthy recovery in traditional storage attached to traditional CPU-based server?
Yeah, I think two things there. I think the compute buildout has primarily been GPU, right? And GPUs have been built for training LLM. That's been the majority of the investment in GPUs. The training use case, as we have consistently said, does not generate or need a lot of storage. It just uses storage like a memory buffer so that if the language model runs into trouble, it can recover its original position. We have solutions for part of that market, but it's not a giant market.
Right, but I guess maybe what I was trying to ask is we've seen some customers look for more power-efficient compute outside of GPUs, but we haven't seen that translate into traditional storage demand.
I think storage is starting to tick up, right? So just following the earlier point, I think what we're seeing now is the AI market moves from the training part of the LLMs to actually the use of the LLMs to do business activity. You will see data infrastructure getting built out. We said that in the quarter we had 200 AI wins. These are GPU-connected storage landscapes. There's three types of them. One's a data prep environment, which is essentially where customers want to put all of their data so that the data science team can actually analyze it. The second is training and fine-tuning. This is where either you're training a large language model or you're taking a pre-trained model and then optimizing it with your data or synthetic data. And the third is, hey, inferencing broadly defined. It could be reinforcement learning. It could be RAG.
It could be another form of inferencing. We said during the call that it's pretty similar. It's 45%, 25%, 30% in terms of the mix: 45% data prep, 25% training, and 30% inferencing. So if you look at it, the storage buildouts are up from a year ago, and they are like the compute focused on the get ready for using AI part.
Right. So it's a good point. So when you think about in the past, I think you've said you want to bring AI to where the data is, not data to the AI, right? It's expensive. So in that framework, that three-prong framework that you laid out, so when we think about bringing fine-tuning models to the data, where do you think we are in that cycle when you talk to enterprise customers? And maybe can you talk to where you are internally? I would imagine that you have multiple projects and test cases and are exploring how it benefits NetApp.
Is this a calendar 2026 kind of dynamic where we're going to see more use cases and customers really going out in that direction, or is it more 2027, 2028 at this point where everything is still training-centric and we're not really focused on fine-tuning these models and bringing it to where the data is on the enterprise side?
I think the number of kind of AI centers of excellence plus, hey, the number of AI projects has grown, right? You can see we said a year ago it was 25, then 50, then 100, 150. Now we're at 200, right? So it's going up. So you're seeing more enterprises engaged in the activity. With regard to the maturity of the use cases, listen, there's public data around it. We see the same thing. So in healthcare and life sciences, there are truly transformative things happening, right? I mean, man, things that you never saw in the history of mankind: drug discovery, genomic analysis, large population studies, hybrid infrastructure use cases. I mean, it's just unreal, all the things that are going on.
And it's because the data was well organized, you are able to, we are part of many of those because we can help bring together data sets across a diverse range of environments. I think if you look in sort of from a horizontal use case, wherever there's software development, yep, that's ticking up. We have it. It's ticking up nicely within our own company. There are fairly standardized back office use cases. Hey, document analysis. It could be check analysis in retail banking. It could be contracts in procurement organizations. It could be legal documents in law firms. It's kind of like the same thing. And then there's customer success where people that use our infrastructure, they want to correlate data from multiple parts of their customer success pipeline and do better chatbots or better calls.
Does that suggest a longer tail for legacy systems and storage? As companies keep storage longer, keep data longer, so maybe asymptotically your hybrid system business, that's what, a third of your storage solutions maybe stops declining effectively or the growth meaningfully moderates. I'm just trying to think of the longer-term ramifications for all of this incremental use case for data. We're going to create a lot of data. People will probably want to preserve said data for unknown potential incremental use cases. I mean, should we think about the storage market then being sort of a healthier market over the long-term or intermediate term than what we've seen over the last five to 10 years? Is that a reasonable way to think about it?
If the data is seen as the way to get better value out of AI, which is a hypothesis that everybody seems to agree with, but you got to see that put into deployment at customers, that is true. We are seeing that evidence where people who historically, like I was with a large weather forecasting agency that collects an insane amount of satellite imagery and then analyzes it to provide better weather prediction, various kinds of storm analysis. They used to archive a lot more data onto tape. Now they're keeping it on online storage. And it's for the reasons you mentioned, David. So it's early, but if that is the hypothesis that plays out, then yes, it should follow the trajectory that you had.
Got it. So we have to talk about margins. So storage historically has been a market where you've been able to increase prices to offset rising inputs, maybe albeit with a little bit of a lag issue. Do you think we're going to see that same type of cycle from customer behavior where you're going to be able to navigate maybe the higher NAND pricing, higher DRAM prices, and other components by raising prices? I mean, I think you talked about it on the last call where customers buy dollar amount of storage. I mean, can you kind of walk through how we should think about that?
Yeah. Customers budget in dollars. I think the way it works is IT as a whole gets a pool of funding from the CFO. They divide it up into projects. It could be, "Hey, I've got a cloud modernization project. I have a higher cyber risk management," or, "Hey, I want to build data infrastructure for AI," right? And then within that, there are buckets for software infrastructure, and storage is a part of the infrastructure bucket. So that's kind of how it broadly gets laid out. Typically, customers budget in dollars, right? They'll say, "Hey, I got $250,000 or $1 million for my data infrastructure." And what they will ask us to give them is the best solution for that use case. Sometimes it's a flash-based solution. Sometimes it's a disk-based solution. Sometimes it's a combo.
And they might get more footprint out of that if it's a year when NAND or commodities are cheaper and less when the prices are higher. With regard to our gross margin outlook, we run the business to drive corporate gross profit dollars. That's sort of the piece of the P&L that drives earnings leverage. When you look at the overall P&L, four buckets in there. There is support, which is a stable, low-growth but growing, very high-margin business, 92% margins. Cloud enters next year at a much higher number than it started this year because both it has grown in revenue, but also margins are up at 83%. Our Keystone business, which reports as part of the professional services line, is growing. It's a smaller number, but growing at 80% in the first half of the year from a revenue year-on-year perspective.
That pushes up the professional services margin line. And then comes product gross margins. Within product gross margins, we manage that in multiple ways, right? One, the first thing we focus on is availability and access to innovation. So we don't want to fall behind competition on either the latest types of silicon or assured access to supply. The second, because we and so we work with pretty much everybody in the market share charts. The second is we have structured pricing agreement with them. I think they appreciate the fact that both we are number one in flash market share, but also we have been a stable, we don't hockey stick our business up and down. We've been a stable buyer of theirs. We have, in the second half of this year, a good line of sight to our kind of supply needs for the rest of the year.
And we are in active discussions with them about what next year looks like. And we'll take the appropriate actions. In general, the storage industry is, as you said, David, we pass through higher prices when commodity prices go up. We pass through lower prices when it's structurally adjusted downward.
Maybe just go back to your point about customers have a fixed budget of dollars, and they may get more GB in one year, less in the next year. Do customers trade down? Let's say you were thinking about a higher-end, more performative flash solution. Maybe they trade down into maybe a more modest cold storage-focused solution if it's adequate to meet that need, or is it really the demand for that particular use case or workload supersedes that decision, and they just buy less of said product for the same dollars?
There's no one consistent answer for most customers, for all customers. Most customers say, "Hey, I got to serve this use case. Give me what can actually solve the use case." And then they might say, "Hey, I'll just buy more next year," right, to offset the fact that they have less systems. Others might trade down, and so it just depends.
It depends. So the reason I ask this, so is there, I know you don't break this out, but is there a margin differential across product lines based on performance within your portfolio? So very high-end flash performance, I would imagine, has more robust gross margin, more solution, more software versus maybe a lower-end solution. Is that a good way to frame it?
Yeah. I mean, generally, the highest performance systems in any product family have the highest margins. And as you get lower, the capacity overruns the value of the software. And then so margins get lower. I think in our portfolio, as we said, flash has higher margins than disk-based solutions. It's interesting. This quarter, actually, the mix shifted towards higher performance systems, which is what drove the peak.
Gross margin frame.
So it's really hard to tell. Some clients might say, "Hey, I want to downshift," but it's a big market. You're serving thousands of clients a quarter, and so it's hard for me to give you one answer.
Got it. Maybe on pivoting to public, obviously, gross margins have been working their way higher, 83%. Congratulations on the quarter. How do we think now that we're anniversarying spot coming up, how do we think about the growth trajectory in those, I guess, discrete items within your public cloud? So first-party storage, etc. Is the holistic solution for public cloud kind of a mid-teens growth driver? And the reason why I'm asking is if I look at the profit dollar contribution from public cloud, it's kind of incrementally on par with your traditional business.
Correct.
And so if it's growing 2x, 3x your traditional business, obviously, then we're going to mix higher to mid-80s gross margin solutions. Is that a fair way to frame it?
Yeah. I think, first of all, ex Spot, the cloud business performed at 18% for the quarter. The print was 2%, but that was because of the headwind from Spot. I think if you look at first-party cloud storage, which is the natively integrated storage, that is growing much faster. It's at 32% and is an increasingly large share. It's well more than the majority. It is an increasingly large share of the cloud business. So both of them should provide for, if we can sustain the performance of the first-party cloud business, which is the anchor tenet of our cloud strategy, which it should help drive the cloud numbers up because it's a bigger share. It's growing faster than the rest of cloud. And listen, from a margin perspective, we just gave the guide of 80% to 85%.
I feel like it would be inappropriate to come and re-rate the guide.
No, no, of course. But mix-wise, it's amazingly still meaningful to everyone.
Correct. Correct. That's exactly right.
So in this quarter, what was the strength in first-party? Was there anything you can call out in first-party storage? We're definitely stronger than we expected.
I think it was right on what we wanted to do. So we see good strength in the business. All of the hyperscalers now have mature offerings in first-party. We are bringing, we brought block to Google in addition to Amazon. So we have truly multi-protocol first-party. And we've got a broader range of enterprise workloads that we are part of. And we are bringing more price points, more packaging offers so that customers can buy through more vehicles from the other cloud providers. We also have work going on, and there's more to come to integrate into the migration motions of the cloud providers. So Amazon, for example, uses our tools to migrate enterprise workloads to the cloud so that you can go faster in terms of the ramp of those businesses. But we feel really, really good.
Listen, we're just in full-on mode. It's just, how do you scale a business that's growing fast? And then over the next 12 months to 18 months, you'll see a lot more co-innovation around AI. We talked a little bit about it on the call.
Within public cloud.
Correct. Within public cloud around their AI motions, so we talked a little bit of it. I'll just tell you, watch the conferences coming up: Reinvent, Pi Day, Google Next. There was a little bit of that at Microsoft Insight, so there's just a lot more coming in that area.
So outside of the expansion of the relationship on AI with the public cloud partners, is there anything that you need? I mean, you mentioned you added block to your Google solution. Is there anything else that's sort of missing in the portfolio that you look at and you're like, "We need this," and then we have the portfolio or the complete portfolio that we want at this point?
I think on cloud storage, there's really nothing, right? It's just scaling go-to-market. I think when we look longer term, we want to build on our lead with cyber. I think we have been just like when we introduced deduplication on primary storage and everybody said, "Hey, that was on for secondary storage." We've got leadership, thought leadership, technology leadership. We want to expand that part of our capability. And then what we are seeing in AI is people want to do more of the AI work where the data is created rather than copy the data into another environment and do all the work and then recopy it back. And so there's more things we could do there. And so we look at that, building some of that organically. And if there are small tuck-ins to fill that out, we'll look at those opportunistically.
Got it. I have to ask the final point. With all of this tailwind potentially for public cloud, for traditional, for disk, for flash, you talked about margins on supply chain. How do you think about the prospects for incremental capacity coming online? I know you're close to all the memory vendors. In your experience, does higher prices mean more supply effectively? What we always hear is the cure for higher prices is higher prices, right? So then supply comes online. But it doesn't feel like that's the case this cycle when you talk to participants in the supply chain responsible for that capacity.
Yeah. I think it's dependent on who is the person suggesting there's more demand, right? I think there are people who do large volume purchases of infrastructure and then stop. And so it's very hard for a technology provider to them to kind of plan for their peak demand. On the other hand, we have been a very steady, growing procurer of these memory technologies. And so when we talk to them, they're like, "Hey, you guys see a higher uplift? Okay." And we also are seen because we use merchant silicon. We have been a good partner to many of these vendors to bring their new innovation to market. And so they do work with us on denser silicon, more cost-effective silicon, new types of use cases, and so on. And so track record matters.
Got it. So we've been asking everyone today, what do you think the market's missing or a little bit unclear on with regards to your strategy going forward? We've had challenging U.S. federal. I think everybody understands that. I mean, it has been a little bit uneven, to be fair. But parts of your business are performing exceptionally well given the circumstances. So what do you think the market is missing about the storage industry, the NetApp sort of strategy going forward for the intermediate term? Not next quarter, not the following quarter, but longer term.
Yeah. I think the sentiment this year is clouded by the fact that we have on our prints the headwind from the divestiture of spot that masks some of the strength in the underlying business, as well as the exposure to U.S. Federal and U.S. public sector, which should correct itself, right? It's not a permanent structural reduction in a part of the customer segment. I think that masks the underlying strength of our franchise, which is we are growing share in the all-flash market. We have cloud growing strongly, and the declining parts of our business are both smaller, and the pace of reductions are less over time. We are a software business. You can see that in the cloud gross margins. You can see that in the overall margin profile of our business.
And so gross margins, as well as the quality of revenue, in addition to the fact that, hey, these headwinds should ameliorate and the core franchises are strong, the quality of revenue in those franchises is very good. And then I think we have been disciplined operators in terms of operating expense, which says that the incremental dollar of revenue and gross profit converts at a very high clip to earnings. And then we have been disciplined stewards of capital. And so I think that those are the big buckets of things that we talk to our investors.
Great. I think we're out of time, so we're going to end it here, George.
Thank you.
Thank you for your time. Thank you, everyone.
Thank you very much.
Have a great afternoon.