NetApp, Inc. (NTAP) Earnings Call Transcript & Summary

Welcome to Citi's 2024 Global Technology Conference. My name is Asiya Merchant. I cover the technology, hardware and tech supply chain companies here at Citi Research. Very thrilled to have NetApp's CEO, George Kurian; as well as members of the IR team, Kris Newton's here in the audience. This session is obviously for Citi clients only. I've been asked to read NetApp's safe harbor. I'm going to quickly go through that. Today's discussion may include forward-looking statements regarding NetApp's future performance, which are subject to risks 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 the website at www.netapp.com. We disclaim any obligation to update information in any forward-looking statements for any reason. So with that out of the way, thank you again, George. I'm going to make it back here. Appreciate you coming to Citi's tech conference. I'm going to start with a bunch of prepared Q&A. And if anybody has any questions, please do raise your hand, and we'll bring the mic to you.

So George, we've been just asking all our companies here. If you could give us like how you think about end demand, obviously, you guys just reported results last week. Given the market dynamics that are out there, the macro, if you could characterize the performance for NetApp.

Yes. Thank you for having me, and thank you for coming to this session. We have said that the macro continues to remain uncertain. It is certainly better than it was a year ago and certainly better than what it was at the start of the fiscal year. By the start of the calendar year, I think inflation is down, you're starting to see potentially signs of improvements in GDP outlook and hopefully, an interest rate environment that's a little better. I think what we see within clients is they are prioritizing the strategic projects. And I think we are the beneficiary of both data being an important part of the customers' landscape, particularly as they get ready for AI and our unique offerings. And so we've had despite an uncertain macro, we've posted strong results for three quarters in a row.

As you sit here in calendar 3Q, George, I mean, what is -- if you look across your maybe your key end markets, and you could go by key customers because obviously, you have enterprise, you have government, commercial. Since the start of the year, maybe how has demand shaped up, if you kind of sit back today to kind of reflect on how things have changed over the last, let's say, 6 to 8 months. And then as you look ahead, what are some puts and takes to the end demand?

Broadly speaking, we have seen improvements in demand over the course of the last 9 months. When we started the calendar year, I would point out that America's -- large customers in the Americas business were still challenged and they have started to be more positive in terms of buying. Some of that is, of course, us having a stronger product portfolio and a more focused go-to-market. But overall, we've seen positive trends there. Europe has continued to do well for us across the board. A lot of that is execution. I think that our U.K. business is starting to see signs of the new government being seen as a positive to spending in the stabilization in terms of kind of economic outlook and sentiment. I think the one place that has remained soft for a period of time for us in the geographies is Japan, where we've done well in terms of execution, but overall, the economic environment either because of currency conversion or because of just end demand has been more choppy there. I think with regard to our business overall, we've brought a stronger portfolio to market with a much more complete all-flash portfolio with sort of capacity flash, which has been the strongest performer ever in the history of the company as well as a new sort of market expanding block storage product. We've done well with both of those. Our overall results in Flash have been really strong for the last three quarters 21%, 17%, 21% growth year-on-year. So we feel bullish about the strength of our portfolio. In the cloud business, our hyperscaler first-party and marketplace offerings have been strong all year, and we just had a quarter where we were up 40% year-on-year. So Overall, I feel good about our portfolio.

And as you think about the outlook that you guys have provided, both more for the near term for fiscal '25, but then also kind of your 3-year outlook that you provided at your analyst event, what anecdotally, like what discussions are you having? Or what trends are you observing that enables you to give like the guide for the year? And then also as you kind of think about it longer term.

I think -- let's start with fiscal year '25, we completed a strong Q1 where we beat on top line and bottom line. And our major focus areas, flash, cloud, block and AI all had strong results. I think we took up the guide for the full year on both top and bottom line, and those are reflective of normal seasonal patterns in our business but also visibility into pipelines that our field has. And so we feel good about the momentum in the year. We said that our outlook for the year continues to be that the macro remains unchanged, right? And it's not because of anything else, and that's what we see today. The macro changes substantially positively, it's a tailwind to drive out top line. And if it changes conversely dramatically different, then we will update the guidance at that point. In regard to the 3-year model, our North Star is essentially a Rule of 40 company, right? And so we said top line should be in the mid- to high single digits, driven by two secular growth trends and two company-specific growth trends, the secular ones being we're in the early stages of a 10K disk-to-flash transition and the early stages of the build-outs of data links and data foundations for AI and the two company-specific transitions are the unique position we have in the cloud storage market, continuing acceleration of that as part of the overall growth of cloud and the second being a TAM expanding flash -- block storage product, where we're pleased with early momentum. So I think that's top line. We're balanced in terms of operating margin discipline. So we are going to continue to drive operating margin leverage. We are in the range of 27%. We think we can get it to 30% over a 3-year period without any unusual kind of changes and then I think from a capital returns perspective, listen, we've been balanced in terms of capital allocation, dividend the first call on capital at a minimum of offsetting dilution with share buybacks and then if we don't see an M&A pipeline returning up to 100% to shareholders.

The question about AI comes up a lot. Obviously, that's top of mind, specifically as it relates to general purpose infrastructure. So outside of GPU intense servers, I think people want to ask like why do you think AI would be an incremental driver? And when do we actually see that inflect? Obviously, we haven't seen external storage grow at the same rate of server GPU TAM has exploded last year. When -- do you think there is a catch up there? And when do you think we would see that inflection, if you think there's going to be a catch up with storage?

Yes. First of all, we don't disagree that storage lags the GPU build-out. And the GPU build-out is fundamentally, there's a new computing architecture required for LLMs, right? As we have pointed out before, Chat GPT did not create a second copy of all of the Internet's data. To train it, it just use GPUs and the publicly available Internet. So what we see are three areas of opportunity where we are investing and we are well positioned with. The first is clients are building out, and we talked about this on our earnings call, data lakes or some kind of data repository where they want to unify the data that they're using in AI use cases. What does that mean? Most of the data that they're using is in some transactional systems somewhere and to actually make it easy for a data scientist to actually manipulate the data and transform it to get insight. They want to unify it. And we had roughly 50 to 60 data lake build-outs this past quarter. These are large transactions and they used to be the prior generation of data lakes was what Hadoop and Big Data architectures had, those were a failed model. You're seeing a new architecture getting built out. We are super well positioned for data lakes because we have unified storage to underlie a unified data lake, block, file, object all of the different data types. You can unite data in your data center and in the cloud so that's one. That's where the action is today, right? There is -- the second in the enterprise AI landscape, is the beginnings of what we call inferencing. In predictive AI, they take the data lake models and then feed it into some kind of regression or mathematical algorithm and then we run it for computer vision or whatever it is. We are very successful in that market. We've got hundreds of customers. It's built into our kind of our flash growth rate. For Gen AI, early stages of inferencing and what is inferencing? It's taking a pretrained foundation model and then using your company-specific data to make that model understand your business better. And so we said, for example, one of the largest retail banks in the U.S. is using us as their platform for inferencing. We hold all of their mortgage and loan documents and checks and all of these things. And we are working with them, NVIDIA and a software vendor to summarize all those mortgage documents so that they can understand what's a good risk, what's a bad risk, what's the range of choices their agents make when they write that and then put it into production. Why does -- so what happens in inferencing? In Inferencing, you essentially create what's called a vector or a transformation of the data in your documents, so that the large language model can understand it as a numeric representation. And these vectors can get quite large. We are working to, sometimes, orders of magnitude bigger than the underlying storage of the documents. So we are working with some advanced technologies that we have developed to make the vectors more efficient but we see the vector database storage and some of the inferencing things is having an incremental growth rate for storage just because you're creating more variants of the same document. And then the last phase is production, right? This is when you take a model that has been trained and then deploy it into the operating landscape of your business. And there, you will see data growth coming because people will generate more copies of the data, right? Just think about the iPhone. Before the iPhone showed up, there were professional photographers with real cameras. And now everybody is out taking pictures just because it's easy. And so when you make it easy to generate versions of a document, I'm sure there will be 17 versions of every marketing brochure and 20 versions of every price book just because it's easy to generate. And I think that has proven true in cloud, where you saw when it was easy to consume cloud resources, people spend a lot on it, right? And so we see that over time. So as we've said, we see that the data foundations plus inferencing happens over the next 1.5 years, 2 years. And we think that in our model, AI will be more of a tailwind in the last part of the 3 years, right? So the second and third year more so than now. We are very well positioned for AI. We and Dell are the two incumbents with all the unstructured data in the world. And so if you want to do RAG, just do it in place on us. It's much easier to do than getting another system in place. And we have some really good innovations. The data management that we are sort of the software intelligence that we have has proven to be incredibly helpful around things like privacy and compliance and model traceability and all the concerns that customers have as they take these models into production so that there's no risk in their models. And then the last piece is our cloud solutions give us the opportunity to integrate the big cloud providers, AI tools with on-prem data. You'll see customers at our conference. For example, in health care and life sciences, where they have taken clinical data that sits on our systems, use the tools with AWS or Google to find new drugs and it's amazing transformative use case. And so I'm excited about what the future holds there.

When people talk about AI and they want to know if there's some killer app. Like I know you talked about probably more of an influence in the latter part of your 3-year model. Are we expecting any particular killer apps that would develop that would drive that spending and grow the actual TAM by, let's say, more than just the normalized growth of storage which tends to be like mid-single digits? [indiscernible].

Listen, I think overall, the broad arc of what we are seeing is data-driven organizations are more equipped to succeed than those that don't have a good handle on their data, right? And we have seen that we come true even more so with the advent of AI. I think we look at AI as the continuation sort of generated AI as in one sense, the continuation of a long-term trend and then the other, a step function improvement in terms of the usability of the tool, right? And so what do I mean by that? The statistical analysis of large volumes of data started in the 1700s with the census for taxation reasons being the first large data sets collected in Europe and the tools that Gauss and Laplace created around mathematical algorithms, right? So it's been a long-term thing. It got accelerated in the middle part of the prior century with computation, enabling you to do more and more calculations at speed. And then probably 20 years ago with the really large data set analysis that Google, for example, pioneered, you could see good results. What Gen AI has brought to the table is the use of natural language so that a human can interact with an AI system, right? And so it broadens the reach and democratizes the access to it. I think with regard to AI itself, the range of tool sets, there are some that are super mature, right, like clustering and regression is the foundation of recommendation engines. It's the foundation of a huge number of algorithms, including financial trading algorithms. So those are mature. And I don't see those as going away. Predictive AI with machine learning and neural networks is very mature. We are part of those landscapes. I think with Gen AI, the big things that we watch for are significant improvements in productivity, which is always a spur for investment in cooperations and significant benefits to driving growth. I think today, what we see is productivity is certainly true in software development, it's certainly true in some of the back offices that support field organizations and customer interfaces. We are doing that, and we see we are close to the action because we're implementing some of those tools in our own landscapes and we see them being implemented in customers. I think with regard to revenue growth, there are some built on better customer experience, more integrated offers and things like that, but we've got to wait and see.

Talking about competition. People are often concerned like you talked about Dell, Dell has AI servers. Obviously, they have a broad storage portfolio like yourself. But they also have the -- maybe they have a better chance to attach their storage and gain share as they ramp up their AI server offerings in the enterprise. What would you describe as NetApp's competitive differentiation here versus somebody who has a full stack offering like Dell?

Yes. This has been the long history of the storage industry, you see now storage started as a peripheral of compute, right? And then there were several inexorable trends that needed separate from compute. The two most important ones were that storage was needed to connect multiple computer systems together so that users of these computer systems could share information. And so it became a network entity of its own rather than a peripheral of a single computer system. And the second was the value in storage is really the software, not the system, right, not the hardware versus in compute, it basically became a packaging of Intel chips with Microsoft software. And so if you look at the data, and this is all publicly available data, the server vendors have lost share every year in storage for the history of the industry. NetApp was created out of the failure of Sun, right? EMC was created out of the failure of some of the mainframe system vendors and NetApp and EMC took share from all the other storage vendors. Periodically, they would buy a company like HP did 3PAR and Dell bought EMC. And if you look at the scorecard of Dell over the last five quarters, there is yet to be a quarter where they have posted positive growth. So I think the thesis that an integrated stack has enormous benefits is probably a [ false ] thesis, respectfully. I think that what customers want is high performance which multiple vendors, including us have. I think what they will need more over time is data management because of all the challenges we talk about, how do you version models? How do you show traceability between a model and its data set? How do you create for example, almost a serverless data life cycle paradigm that we are able to do because of the intelligence in our system, Dell has to prove that it can do it and others have to prove that it can do it, right? And so that data management is important. And then, listen, what has been the most amazing part of this entire experience for me has been, we have all the cloud providers working with us now. And Microsoft was first with us five years ago. They are generally available, this is the fifth anniversary almost of the GA, of Azure NetApp Files, 3.5 years ago was Amazon and about 1.5 years ago was Google. And they are going to be dominant players in the AI revolution and no one can access any of their tech and we can.

Flash as it relates to -- I mean, obviously, you posted strong growth here. What is happening there? And why is flash sort of getting more pervasive in the enterprise external storage side? What's the driver there? And do you see that sustaining especially as inferencing and some of the enterprises start adopting AI?

So I think it's the long march of technology. I think that flash provides for specific use cases, benefits in terms of density, cost consistent performance and energy consumption. It is traded off against the price of that, right? And so if you look at the disk storage industry, there were 15K drives, 10K drives and 7.2K drives. 15K was about 10% of the market by capacity, 10K was about 35% and the remaining 55% to 60% was 7.2K. 15K basically got replaced by TLC flash and then 10K is in the second inning of a 9-inning ball game with QLC, quad level cell flash. And 7.2K, there's no technology today that is head-to-head cost competitive with 7.2K. So when the next click on flash comes or if the flash vendors are so hungry to take the 7.2K that they price flash down, maybe it will move there.

And then when it comes to flash, help us understand, okay, where is NetApp? What differentiates NetApp from the likes of Dell or storage from some of your other peers?

I think we are competitive on cost and performance. I don't think any vendor to be honest, has an advantage on cost performance. Everybody introduces a new platform every 18 months. And so somebody is in a cycle, but over the long arc, I don't think anybody is sustainably better on cost performance. I think those of us that grew up in the open systems world have a focus on driving cost performance advantages against, for example, a Dell or a Hitachi that came out of the mainframe world. So the high-end storage markets of Dell and Hitachi will collapse into the midrange because midrange is just better from a cost and performance. But all the midrange vendors are competitive there. We are differentiated on our software value, right? So when we bring our technology to a competitive cost performance point, like we did with QLC, you see the hockey stick.

And then you talked about block storage. And I think at your analyst event, you talked about a TAM there for block storage that is almost 2x that, if I remember -- on the file storage. So some of the questions that investors have is what's so unique about NetApp's block offerings? Why now? Why is this a big deal now? Why hasn't NetApp always been in block offerings?

It's a good question. I think there's probably two or three kind of important things to discuss there. I think one is, listen, we have 20,000 customers who run block workloads and environments on NetApp. So we're not a newbie in the market. We have always attacked the market as a single platform, meaning what we call unified storage where in a single system, you run file and block. And the benefit of that was just efficiency, right, for customers. What many customers have is, hey, I have a separate team for block and a separate team for file, I do budgets for block and budgets for file because I don't -- separately, because I don't want to refresh my entire landscape in one environment. And so we did not have a dedicated block optimized offering for the block customers, so that's one. The second is customers increasingly want a single vendor or fewer vendors, and there are many vendors in the market that are struggling to deploy -- that are struggling in the market. You look at most of the integrated system vendors they're losing share. They don't have compelling road maps and customers came to us and said, hey, you guys should step up to the plate. I'm happy to turn over my block environments to you. And then we saw evidence of that by actually having customers take our unified product and configure it as a block only solution. And when we started to see that, we said, man, these three data points should cause us to build a custom product, and we're excited about what that holds. It will not ramp at the rate of the C-Series, but it should be a strong growth driver for us. And it's all factored into our kind of outlook right? It's part of the outlook of why we have confidence in our outlook.

And then just the side question to that is -- or the next question to that is just as you talked about margins, right? I mean, typically, files are considered more superior, if you may, type of storage block. Like how should we think about margins here, if your customers are previously using file, reconfiguring it for block and now you have a more competitive block offering? How should we think about the impact to margins?

I think, first of all, block will drive the shift from disk to flash in the mix because our block offerings are only flash-based offerings. And so it should have a higher margin than our disk-based file storage business. If I look within flash at the different offerings, unified will be the highest value, files will be the next, and then block will be the lowest, but it should be still higher overall than the disk-based portfolio. So we think this drives the mix towards flash.

I'm just going to quickly ask the audience if they have any questions, to raise your hand. Okay. We have one.

I guess a question, you mentioned in your comments, which was literally really helpful just the breakdown of the disc market historically by 15K, 10K and 7.2K. And you made a comment that 7.2K, which I believe is just synonymous with nearline, but you can correct me if that's wrong. But there's no flash product that can be competitive unless a vendor wants to go out there and reduce cost. I guess, what do you think the risk is that some competitor just tries to get super aggressive on flash pricing in the near future just to drive adoption of flash in a nearline disc. And if that happens, do the other competitors in the storage market have to respond or not really, they can just continue to go after their own 10K basis for a very long time and just sort of leave that low-margin opportunity for somebody else?

I think the rule that we think about that is, first, you got to be at a competitive cost position yourself and so we have done many things to bring into our portfolio competitive silicon as well as competitive procurement approaches, right? I think on the silicon side, we have broadened our access to hyperscaler style flash, so that it's high volume, low cost and part of our QLC offerings include that as well as from a procurement standpoint, you've seen us being aggressive in the market, doing prebuys to protect ourselves. I think over the next [Audio Gap] right, and so I think the other vendors are going to be all disciplined around pricing. I think you may see an anecdotal transaction here or there, but no one is going to be systematic across the board to drive pricing. I think as supply becomes more available in the next calendar year, second half of next calendar year, we will certainly look at, hey, should we position ourselves to drive our transition, right? And can we negotiate deals. It is much easier for us to drive a transition from our disk-based system to our flash-based system because the software stays exactly the same. And so we will take a look at that as the silicon trends ease over time. I hope that answered your questions.

Any other questions? If we can switch a little bit to the cloud side of the story, George, the public cloud portion of your revenues. You're starting to see that the growth there, especially on the first-party side of your business, I know there are some headwinds as it relates to the other part of the cloud business. So just help investors understand what's the sustainable growth trajectory here? And when does it become more meaningful for your financials?

Yes, two things there. I think the first was we have seen changes in the customers and in the way the hyperscalers created offerings for customers where they were -- customers wanted to move from a subscription-style model to more of a consumption- style model, right? I think that as the hyperscalers drove that transition for better fidelity between usage and expend we were caught up in that, too. And so we've needed to transition our business to subscription -- from subscription to consumption. And so consumption today is 80% of our business, it was a much smaller portion a while back, right? So it has been the inexorable driver. Some of that is we've exited some of the subscription offerings. And in others, we have migrated customers from subscription to consumption, which will cause a dip for a period of time before it causes [ a problem ], so that's one. I think the second has been completion of our cloud portfolio from a technology standpoint, where we started with the high price points in some of the cloud providers like Microsoft and Google, and we started with a low price point in Amazon. And we've now done a really good job bringing our solution up to addressing a much broader range of price points in Amazon. We are doing that in Google. We introduced a new lower-end offering in Google a 1.5 quarters ago, and we're working with Microsoft to bring up more full-fledged offerings. So that brings us, by the end of this fiscal year, I feel like we would -- the goal is to make us really complete across price points and feature sets as a same member of the cloud, right? I think in -- the third part is scale and go-to-market. And so we've done a good job there aligning with the cloud providers' go to market. I would say we are in the third inning of a 9-inning ball game. They got 40,000 sellers. We got 1,000, right? So even if all of my guys spend time, you still are dealing with 1 to 40. So we've got a lot of leverage should we be able to scale that. We're working with them diligently on how to do that. I feel really good about the differentiation and the value that we have. I mean we were up 40% year-on-year in cloud -- first-party cloud, we think that, that drives, as subscription has become smaller through the course of the year, you should see continued acceleration of cloud storage revenue. And what we said was we expect it to grow in the mid-teens in the model over time.

And the customers that are on your public cloud that are coming to you through their first-party cloud versus the ones that have on-prem, is there any difference in the amount they're buying, what they're buying, aside from obviously the business model where once you pay for it on a consumption basis. But just is there any difference in the workloads why some are coming to NetApp through the first-party cloud, some are coming to you directly.

I would say that what has been particularly pleasing has been the fact that we have been able to win a lot of new customers in cloud. It has been about 50% of our cloud names were customers that we never talked to, which shows that it's a market-expanding opportunity. Originally, when we brought our cloud, the big concern that investors had was it would just move our installed base to the cloud. We are actually seeing the complete opposite of that. And so that's a good sign. I think the second has been after the phase of optimization has past and based on some of the investments we've also done in customer success, the cloud storage expansions have started to come through. Cloud ops is still a work in process, but the cloud storage expansions are on a healthy trend. And then I think with regard to the workload mix, listen, we believe that the world will be hybrid, that what customers are doing now is they are -- when there's a new workload or when they are replatforming, meaning rewriting the code in a legacy workload, they think about cloud as a good place. They don't always end up with that, but they think about that as -- so it's not like there's a massive lift and shift moving in one direction or the other. And so it's -- we see the world as hybrid. We are a small part of the cloud business in the hyperscalers, right? And so there's -- regardless of what their growth rates are, we have a huge opportunity to grow.

And then margins. Everybody is obviously concerned, you talked about flash pricing. How do you guys think about your margins as you think about flash and component costs getting inflationary and at least Citi's view is, our house view is calendar '25 will see further inflection in flash pricing. So how do you guys think about your overall margins on the product side of things and on the hybrid cloud offerings that you have?

Yes. So listen, we have been through almost 30 years of company history in this commodity environment. What we said was we see flash pricing going up through the course of this fiscal year. We are not yet guiding '26, so we will guide it when it comes. Disk prices, meaning hard drive prices are relatively stable. We have, with the information that's available to us at this time said that we feel good about the 58% to 60% product gross margin for the full year. It will start higher in the first half and be lower in the second half, but for the full year, it should be in the range of 58% to 60%. We are seeing, as is common in the industry, that vendors in the market have raised prices. And so you saw the announcements from Dell, who are the largest player in the market to fairly significantly raise prices to reflect the reality of the commodity environment. We are contemplating what our response should be. But historically, I will just say every vendor in the market raises prices in some way, whether it is controlling discounts or raising list prices or a combination of those. And so yes, we see the commodity environment has gone up. We have procured most of our supply for the fiscal year. If we dramatically outperform on the top line, that will be a different discussion, a good discussion. But from what we see here to date, we've done the work to put ourselves in a position to deliver 50% to 60% for the full year.

And is there any -- does the concept of elasticity come into play here? So for external storage as it does -- with higher flash pricing, does it affect the mix or what people are buying? And does that in some way have an indirect impact on margins?

It will affect -- first of all, CIOs are not super strategic. So they don't plan ahead for timing the market. Most of these purchases are done on an economic cycle and a depreciation cycle of equipment. So it's not timing flash market pricing to say buy now, buy in the future. They budget in dollars and so they don't buy more because this cheaper, they just say, hey, I got x number of thousands or hundreds of thousands of millions to spend on storage refresh. I think on the margin, you see a shift in the mix to different configurations depending on what the vendors choose to do.

All right. We're out of time here. So I just wanted to thank George and Kris for coming to Citi's Technology Conference. Hope the rest of your meetings go well. Thanks again, take care.

Yes, thank you so much for having us.