Welcome to the NetApp fireside chat at the Goldman Sachs Communacopia + Technology Conference. I have the privilege of having CEO George Kurian here with us today. My name is Kat Murphy. I cover NetApp and IT Hardware here at Goldman Sachs. We have about 35 minutes for today's discussion, inclusive of audience Q&A. Before I begin, I will read a quick safe harbor. NetApp asked me to read their safe harbor disclosure. Today's discussion may include forward-looking statements regarding NetApp's future performance, which is subject to the risks and uncertainties. Actual results may materially differ from the statements made today for a variety of reasons described in NetApp's most recent 10-K and 10-Q filed with the SEC and available on their website at netapp.com. NetApp disclaims any obligation to update information in the forward-looking statements for any reason.
With that, thank you everyone for being here. George, thank you for your time. It is a privilege to have you here on stage. To kick it off, NetApp reported earnings last week, where you raised your fiscal 2027 guidance for 17% revenue growth and 22% EPS growth at midpoint. Before we dig into some more strategic questions here today, can you talk us through a brief recap of the quarter and what gave you confidence this early in the year to raise your full-year fiscal 2027 guidance and highlight anything that you think is important for this audience?
Thank you for having me. Thank you for joining us. It was a super strong start to the year. Pretty much every product line, every district, every industry, every country was well ahead of plan. We had record revenue, earnings per share. Billings was in record territory. All-flash array revenue was a record up 47% year-on-year. Cloud continues to grow in the high teens, adjusted for the extra week, and so we feel really, really positive about the momentum in the business.
Great. One of the central questions investors are looking to understand is how agentic AI adoption may create a TAM uplift for the enterprise storage opportunity, much in the same way we have seen in traditional compute and in networking.
What we are seeing is the broad-based pattern of demand that we're seeing, which gave us confidence to raise the full year, and the outlook provides real confidence in the durability, is this idea that to use AI models effectively, you need robust data infrastructure because the output that these models and agents generate are only as valuable as well-curated, high-quality data. We saw wins in our AI-specific configurations that were up materially year-on-year, but we also saw a broad-based upgrade of databases and lakehouses and streaming engines and all of the other associated applications that go alongside AI. Broad-based demand from a customer standpoint, and this is driven by the fact that as we have seen customers deploy these AI use cases, they need better and better storage and data infrastructure.
As a part of that, you've talked in the past about a significant share of AI projects being abandoned because of a lack of AI-ready data or data readiness issues. Can you talk about why a unified platform like ONTAP in particular helps address some of those data readiness challenges and how you're assisting your customers in preparing for that broader adoption trend?
Yeah. AI models are essentially probabilistic engines, and to get accurate results from those models, you need to apply them and use your enterprise data against those models. That enterprise data is typically smeared across cloud, across a variety of applications, and across a variety of different storage environments and customers. Our view has always been that it's important to unify all of these landscapes so that you can bring different data types, different data applications under one common rubric so that the model gets the best output. Our long-term thesis that hybrid multi-cloud needs to be unified across type, media, and applications is coming through.
Thank you. You recorded more than 1,100 AI deals in fiscal 2026. You talked about 350 more last quarter at increasing deal sizes as well. As this infrastructure spend evolves from pilots into more full-scale production, how should we think about measuring or quantifying the impact on NetApp's results specifically? If you could talk to how some of these AI-specific configurations may look from a margin contribution perspective relative to the rest of your product portfolio.
I think first of all, the size of the AI deals, which are really specifically tied to unique AI configurations that we can track, the number of deals has grown materially, as you noted, year-on-year, and the size of those deals has also grown. We have seen customers who deployed proof of concepts with us a year ago or six months ago come back and say, "Yep, the proof of concept worked. I want to expand the footprint." That being said, the impact of AI in our business is much broader. I think people are upgrading a lot of other storage and applications to work alongside the AI-specific configurations. We see the demand picture being really strong across a broad book of business.
I think with regard to the margin profile, there's no material difference between our broad-based enterprise business and the AI-specific environments.
To maybe dial into that AI-specific environment and the configurations there, I want to ask about the AFX platform, which you launched last fall. How should we think about the opportunity that the AFX platform goes after and how those types of customers, you've talked about those being more of a neocloud-based product, have different needs from your core enterprise customer?
Yeah, I think we introduced the AFX platform for customers who want the flexibility to gang up different sizes of compute instances with storage in really, really flexible ways. Those are typically large enterprises and some of the neoclouds. We have seen good uptake. They go through a certification process, but we've disclosed on the calls momentum and progress with the AFX platform, which is good. We are going to be announcing even more advancements for the really high end of the neocloud for training and this frontier models use cases coming up at Insight. Super excited about that announcement.
Can you talk about what NetApp's right to win is with this neocloud opportunity? Obviously, have strength and leadership on the enterprise, but what about that translates to the neocloud as well?
Yeah, I think two or three things. I think one is, we have proven to have been able to deploy hyperscale level data management, data infrastructure, because we are embedded in all of the big hyperscale providers. We know how the technology from a scalability, from a provisioning, ease of provisioning, from a cloud-native operating paradigm is super proven. We also have go-to-market models that allows the neoclouds and our sales teams to work together on opportunities where we are more partner than vendor. Then, of course, as the neoclouds have gone or want to move from just training or what you call stateless, meaning workloads that are very temporal to more enduring workloads, data management, security, sovereignty, all of the controls we provide, as well as hybrid use cases become more important.
Our opportunity there is growing, and we see broader engagements from all of them.
Great. More on the sovereignty point. You have talked about your expanded relationship with Google Distributed Cloud to extend the company's reach into sovereign and some of these highly secure environments. As governments and enterprises increasingly prioritize this concept of data sovereignty, what is NetApp's right to win there? Is it different than the neocloud story? Or how can you frame NetApp's opportunity in that customer set?
We have been strong in sovereign for a long time. I think that, for example, six of the eight largest cloud providers in France have NetApp as their infrastructure, right? This is domestic cloud providers. We see the push on sovereign growing around the world because AI is seen as a national security priority in most countries. We have been working with both the national cloud providers as well as the hyperscalers to build sovereign environments. Those sovereign environments have our differentiation is that you can take advantage of all of the innovations, for example, in the hyperscalers, all of their advanced AI platforms, but use private data storage and private data management with our solutions.
For example, in the Google Distributed Cloud architecture, they are bringing the Google AI stack with NetApp storage to a private, disconnected environment in customers. We see more and more of those use cases as opportunities for us, not only with Google but in other cloud providers as well.
Shifting gears here, I want to touch on how NetApp is helping its customers navigate input cost increase. Specifically, how does the breadth of NetApp's portfolio across all-flash, hybrid, spinning disk, how does that help your customers navigate these input costs? How should we think about the demand in this environment of increasing ASPs for some of the non-mission critical storage projects that your customers are looking to undertake?
Yeah. We've always believed that you want to bring your data estate under one management framework and orchestration framework, whether that is media type or whether that's location, your data center, cloud provider's data center, hyperscaler data center, so that it gives you flexibility to navigate these landscapes. We are working with our customers on two fronts around the commodity situation. On one front is to give them the tools to optimize the use of high-cost flash for only the part of the data that really requires high-cost flash, and to automate movement of colder data out of those flash-based systems into cheaper disk-based configurations. We hinted at the fact that our disk-based systems have started to show growth, and we see strong uptake through the rest of the year as the cost of flash becomes 15x- 20x the cost of disk. That's one.
I would tell you what's interesting is that data in an all-flash system, a large part of that data is actually cold. It started out hot, but like a Hollywood movie star, faded over the life of the use of data. You want to move it off of that hot system. That's giving us competitive position. The second is for compute access. We are working with several clients on hybrid compute landscapes where they want to provision compute in the cloud as a risk mitigation strategy for not having access to supply chain components for their own data centers. We're seeing more and more pickup of that as well. Both of those are parts of our solutions.
Can you talk about how consumption models change as well, whether it is the capital purchase or consumption-based models like Keystone, how that has changed in the last nine months as prices on memory and hard disk have risen significantly?
Keystone has done well. Our performance for Keystone, which is our Storage-as-a-Service business, in Q1 was roughly similar to our results in Q4, and we are above our internal targets. I think that we want to give customers the broadest range of procurement choices so that they can use the right tools for the right use case. For several examples like, "Hey, I want to stand up a pilot environment. I am not sure it is going to be a long-standing commitment, but I want to experiment," we suggest Keystone or cloud as a better alternative than paying an upfront CapEx purchase.
I will have one explicit memory question, but in terms of fiscal 2027 guidance, how should we think about how much of that outlook is supported by NAND and hard disk that NetApp already has under contract through LTAs, versus volumes that may need to be procured at more floating prices in the future? Anything you can share just on pricing assumptions on these components embedded into your full year outlook.
I think obviously when we provide guidance, especially the guidance that had a significant uplift, you want to make sure you have good visibility into demand, which we do, as well as to work with suppliers a priori to ensure that you have access to supply. We feel good about both sides of that equation. With regard to the work we do with suppliers, we work with them on a broad range of commercial vehicles, including buying in the open market as well as through long-term agreements to assure access to available supply at competitive cost structures, and they prefer a range of tools that we work with.
I think if you look at our balance sheet for the quarter, inventory was up because we want to make sure we have adequate inventory on hand to be able to meet demand.
Product gross margins performed better than investors were expecting and better than guidance in the most recent quarter. As we think about the broadening demand that you've talked about across flash, across hybrid, across disk, how should investors think about this balance of mix, pricing actions, and then some of your cost recovery efforts in the impact on the margin profile for your product margins going forward?
Yeah, I think ultimately, product gross margins are a balance of having raised prices to meet the increases in costs. We can't match them always one for one, and the ability to recover those price increases through discipline and the value that we offer customers. We feel good about our execution in Q1. Obviously, every quarter you got to go execute and match price with cost. I think if you look at the full year outlook on product gross margins, it's incrementally up every quarter relative to what we had laid out when we guided the full year at the start of the year, which is a reflection both of better visibility into cost as well as more confidence in our ability to recover pricing in the market.
I think with regard to our view of the full year, product gross margins are a component of overall company gross margin. Company gross margin is really determined by the mix of product and service. I think if you look at the outlook for the year, product is growing much faster than service, which affects the corporate gross margin rate, even though the underlying rates for all of the components are the same as what we thought, and then in the case of product gross margins, better than we thought 90 days ago.
I want to shift gears to the public cloud segment. You posted 28% growth in the quarter, 19% when you exclude the extra week, operating at very attractive gross margins, as you mentioned. What is driving that momentum, and why are you confident that the public cloud business can continue to grow at this elevated rate as we look into the back half of fiscal 2027?
Yeah, I think public cloud has grown in the high teens, low 20s for a long period of time as we have scaled that business. We feel good about it. Would I like it to grow a little bit faster? Sure. I think our sales teams know that I'm pushing for faster growth. I think the picture is driven by three trends. I think first is sort of secular growth in the hyperscalers. They're growing at very high rates, and we are a small part of the overall storage market, so we can grow more without any real encumbrance. The second is broader product offerings that allow us to deliver more value to customers, as well as the confidence driven by strong net customer retention and expansion numbers. We see really good numbers there.
The third is better execution distribution. We see good returns from investments we're making around go-to-market to expand the range of account types we can penetrate, the range of channel partners we can engage. Super bullish about our cloud business. We are extraordinarily well-positioned, not only in public cloud, but as they build private environments. In sovereign, we are part of several of the build-outs in Europe. In distributed and kind of disconnected cloud, you saw the work that we did with Google, and we're working on others like that. Really been a good strategic bet for the company that gives us a really strong competitive position and growth engine going forward.
On the more offerings point that you mentioned as part of those three drivers of public cloud, can you talk about how AI scaling or enterprises increasing their adoption of AI could drive incremental demand in this category in particular? How do you envision an enterprise using the public cloud and NetApp services through the public cloud to kind of go after this opportunity?
Yeah, I think we see that we have made investments and continue to do more innovation work to connect our offerings in the public cloud with the hyperscalers' AI platforms, like Bedrock from Amazon or Vertex from Google or Azure, Studio from Microsoft. What we are seeing is AI is truly a hybrid use case. Most clients have data in one place and want to use tools in another place. Because we make it so frictionless for them to use those tools, they kind of are putting more and more use cases on NetApp storage. We saw, for example, in the quarter, innovations that we introduced a couple of quarters ago with Amazon see super strong uptake. Life sciences companies using some of the tool chains in the cloud for molecular analysis or for diagnostics, use cases that we would not have won on-prem.
We are seeing them use serverless functions, LLM tools with the data infrastructure solutions we have. Just as I walked in this morning, we actually moved a large competitive customer to the public cloud for their AI data lake. It is just one of many. We have been working on it for a while, and they just kind of went live with us. So gives us more wallet, more value, the ability to not only compete in cloud, but also on-prem for the full estate.
Maybe sticking on that point of competitive wins, how is NetApp seeing the competitive environment or some market share dynamics evolve as you compete with vendors who may be taking a more full solution approach, where they are bundling storage with the sale of compute or networking? How does NetApp's kind of best-in-breed independent storage vendor status compete against some of those more bundled offerings?
Yeah, we have always had to compete with people who have tried to build full stack solutions, bundling all of the offerings together, and I think we have held our own. We have gained share over many years against those providers. I think two or three things there. I think one is we have innovation advantages over full stack vendors. Today, when we talk about full stack to clients, we say, "Hey, your stack is not just on-prem compute network storage. You want to think about hyperscale compute network storage. You want to think about neocloud compute network storage." The real kind of full stack is an integrated hybrid cloud stack. We feel good about our position.
No fundamental changes from six months ago. I think if you look at the overall share metrics, there are several players in the market who are losing share, and it is probably us, Pure and Dell that are the kind of focus players, either gaining share or holding share.
Very helpful. I want to return to a comment you made earlier around ONTAP and talk about how the positioning of ONTAP as a software layer across on-prem, cloud, and AI environments is a competitive advantage or a competitive moat for NetApp. Can you talk about how that's helping your competitive positioning in the increasingly complex configurations of your clients? Maybe for the purposes of this audience, why is this something that NetApp is able to do and your peers aren't or your enterprise customers aren't able to build in-house?
Yeah, I think the core value proposition, ONTAP is our operating system. We have been the pioneer in the industry that said it's better to unify your data estate under one operating framework rather than to have custom siloed environments for every use case. The benefits of that accrue over time, and now you can see that becoming even more apparent to clients, right? I think the benefits accrue from, hey, much simpler operations. I don't have to patch a thousand different environments. I don't have a huge amount of technical training that I need to have for my staff. I can learn one thing. I can use it in a much broader set of things.
The second benefit has been, hey, we have taken our clients to places that no other vendor has taken their clients. We've taken them into public cloud environments. We've taken them into national cloud environments that gives them strategic flexibility and return on that investment in a way that nobody else has. I think with regard to the benefit to NetApp is we make one operating system investment, and we can recover that across a much wider range of customer types and use cases than anybody else in the industry. There's benefits to customers that accrue at an advancing rate over time, and there's benefits to NetApp from having one platform investment that we accrue across a huge amount of use cases.
Maybe I'll ask a few more, and then I'll turn it over to the audience to see if they have any questions. I do want to ask about the U.S. public sector. There's been a lot of disruption and budget uncertainty in the U.S. public sector, but we saw some nice growth in the most recent quarter. What are you seeing in the federal demand environment today, and how should we think about AI as factoring into growth and recovery in that part of your business?
Yeah. I think first and most importantly, the disruption from a procurement side has mitigated. The government was operating under a continuing resolution. There was a lot of uncertainty from DOGE and other disruptions in the public sector, in the federal government a year ago. I think those have largely abated. There's been a funding bill that's been passed, the 'One Big Beautiful Bill,' so we are optimistic. I think with regard to our position in the public sector market, listen, it's a two-horse race between us and Dell. We feel strong about our position. I think with regard to AI, certainly there's national security use cases, there's broad-based public interest use cases in public health or weather forecasting or resource management use cases, and of course, in war fighter and national defense.
We feel excited about the opportunity to serve our government in ways that would protect American and advance American interests.
Okay. We've talked a lot about how you're helping your enterprise customers deploy AI, but you are a large enterprise yourself. Are there any examples you can share about how internally you're adopting agentic AI for operating efficiencies or other ways in which it's added value within NetApp today?
Yeah. I think first of all, we believe that whenever there's a disruption, you got to lean into it so that you can learn it and you can really get the texture of, is it really disruptive and valuable in X use cases of Y use case? We've been using AI in our business aggressively. I think from a broad basis, we had about 400 bottoms-up projects. We narrowed that down to 140 real proof of concepts to 40 projects that have been pipelined to 13 that are contributing to positive returns in the P&L. In those, the big ones are really in product development and in customer life cycle and customer support.
In product development, it's classic, "Hey, can you deliver more innovation to market faster? Can you improve the security of your products? Can you make your products more self-healing?" In customer support, it's, "Hey, can you make your knowledge bases, which our customers love, searchable using conversational AI?" It's reducing the number of inbound customer calls, which nobody likes to make, right? They want to self-service themselves and to make the user experience much, much richer. We have also been able to look at forecasting, for example, on demand or supply at a much more granular level using AI tools in ways that humans just could not do, right? You can forecast now at a sub-component level. You can run really complex simulations across the scenarios.
I think the three or four things I would say is, one is you want a center of excellence and control of your AI platform. We are using more open source and open weight models to manage costs. We are building our own infrastructure to manage token costs. That AI center of excellence gives you reusable patterns, and we match that with business unit sponsorship, because the business units have to fund and to take responsibility for the AI use cases. Ultimately, it's a people process change as well. You can't get AI adopted widely if you don't embrace the human change element of it.
Very helpful. Let's see if there's any questions in the audience. If you could get a microphone over here.
Thanks for the presentation. Can you just explain to us, if you look in between cloud adoption, the public cloud, and enterprise adoption, and training and inference, what is sort of the sweet spot for you in terms of where these vectors are headed towards? In other words, is there a point in which you might see sort of a greater inflection point in your business as we have a greater mix of inference or as enterprises adopt more AI? Just sort of give us a sense of how that plays out.
Yeah. I think from our exposure and for the longer term, inference really is the important element. I think it's true not only for our business but for the industry at large. Inference has to deliver value for continued investments in AI technology. For data storage and data infrastructure, inference is the much bigger part of the overall business than training. I think we have seen more projects becoming successful in inference. You can see that from the engagements we have with customers, where a year ago they were in pilot mode in a lot of them, now they're putting it into production.
I think with regard to the growth in our public cloud business, these are production use cases. They're not training or they're not proof of concepts. They've run the proof of concept a year ago, now they are really doing scale production.
As inference grows, who's likely to be your biggest customers? Is it the enterprise themselves? Is it cloud customers, content aggregators? Who's likely to emerge as sort of the largest incremental buyer for your solutions?
I think we do not sell to the hyperscalers. We sell through them to the enterprise. That has always been our business model. We think that there's a set of digital natives and AI-native companies that are not the hyperscalers, but people that are actually building applications using AI on behalf of others maybe, or software-as-a-service companies, as well as enterprises. We are making some product announcements at Insight for super high-scale frontier lab kind of training environments. If those innovations are successful, which we expect them to be, they will also bring us a new class of customers.
As the price of components and memory has increased, storage as a percentage of the overall solution as a cost has also increased. Are you seeing some kind of a pushback from end users? What is sort of the conversations as you have to raise your prices to maintain margins?
I think we have had to be responsible about making sure that you optimize the customer's estate. On the things that we can control, we make sure that they're using high-priced flash, for example, only for the data that deserves to need that level of performance. We are actively engaged in helping customers reduce or reuse equipment in more sort of to help them reduce the impact of the price increases. I think with regard to what we have seen, which is counter-cyclical, is usually when prices go up, the amount of new equipment purchases are constrained because customers budget in dollars. We went into the year, we said, "Hey, typical pattern says X amount of new use cases, Y decrease in the amount of refresh of infrastructure upgrades."
We are seeing infrastructure upgrades way above normal pattern. We are seeing new use cases also way above normal pattern, which says that there is a secular spending shift towards our category. That is why we are so far ahead.
I do think we're out of time, unfortunately. I want to keep everyone on schedule this early in the day. George, it's been a privilege to have you here. Everyone, I hope you have a successful and fruitful conference. Thank you.
Thank you for having me.