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

Hey, everyone and thank you for joining us for the investor session at NetApp INSIGHT. Hopefully, you were able to attend the keynote this morning or watch it on the webcast. But if not, you can catch it on replay. Lots of exciting announcements for you. But before we get started, I'm going to read a safe harbor for that. Each of the 2025 INSIGHT Financial Analyst tech sessions may contain forward-looking statements and projections about our strategies, products, including unreleased offerings, future results, performance or achievements, financial and otherwise. These statements and projections reflect management's current expectations, estimates and assumptions based on the information currently available to us and are not guarantees of future performance or products, services or features. The development, release and timing of any feature or functionality for NetApp products and services remain at the sole discretion of NetApp and are subject to change without notice. Actual results may differ materially from our statements or projections for a variety of reasons, including macroeconomic and market conditions, global political conditions and matters specific to the company's business, such as changes in customer demand for storage and data management solutions and acceptance of our products and services. These and other equally important factors that may affect our future results are described in reports and documents we file from time to time with the SEC, including factors described under the section titled Risk Factors in our most recent filings on Form 10-K and 10-Q available at www.sec.gov. These forward-looking statements made in the presentations are being made as of the time and date of the live presentations. If the presentations are reviewed after the time and date of the live presentation, even if subsequently made available by us, on our website or otherwise, these presentations may not contain current or accurate information. We disclaim any obligation to update or revise any forward-looking statement based on new information, future events or otherwise. Okay. With that out of the way, we have an exciting agenda for you today. You'll hear first from our CEO, George Kurian, who'll give you a recap of the general session. Then Syam Nair will come and talk about some of the exciting AI innovations that we've made around the NetApp data platform, specifically AFX, which is the disaggregated ONTAP for exabyte scale and the AI data engine, a foundation for Gen and agentic AI. Gagan Gulati will come and talk about cyber resilience. And today, we announced enhanced Ransomware Resilience service. You'll hear from Sandeep Singh around data infrastructure and modernization, how we're helping customers streamline costs and operations. One of the announcements we made today is the Shift Toolkit, which provides near instantaneous conversion of VM reformatting across hyperscale -- across hypervisors, important in this day where everyone is trying to move around their current hypervisor. And then Pravjit Tiwana will come and talk about cloud transformation. And today, we announced a number of features that will expand our opportunity in the cloud, including block for Google Cloud NetApp Volumes and support for SnapMirror and FlexCache across all hyperscalers. Finally, we've got a really cool customer panel with customers from the 49ers and Levi's Stadium, the NFL and Aston Martin F1 team. With that, I'm happy to introduce our CEO, George Kurian.

Thank you, Kris. Welcome to all of you. Welcome to NetApp INSIGHT. We have a super exciting agenda over the course of the next few days, talking about how we are continuing the pursuit of our mission, which is to help our clients unlock the power of their data using the widest range of applications possible. We started that journey with network file storage, where work groups wanted to share data. We brought that to enterprise scale with the unified data storage platforms that we introduced many years ago. We brought it to hyperscale scale with our hybrid cloud solutions, our hybrid cloud data fabric. And what all of this was really driven by is the idea that the best return on investment on your data and the infrastructure that holds that data is that you can seamlessly connect it to all of the sources of innovation and services in the world. And the latest group of those services is really large language models and multimodal models, which is the AI landscape. AI itself relies on good quality data, right? I think you all know that. And the biggest challenge with using AI effectively is how do you actually manage the data, organize it, curate it and feed it into these AI models in a transformed manner, right? And this idea of going from your enterprise data, which is created out of the applications that run your business to AI-based AI-ready data is called a pipeline and a pipeline is essentially a series of steps that we talked about. The challenge with pipelines have been the classic way of building pipelines were created for structured data. This is data that is out of databases or data warehouses where the schema of the data is already well defined, meaning the structured data typically is in some table format. You've got a description of the data in the structure of the table and you've got access controls and governance rules built into the way the table operates, right? For unstructured data, you don't have a schema. You have to generate a schema and you have to generate that schema using technologies like LLMs. The second is the volume of unstructured data and the change rate of unstructured data is an order of magnitude larger than for structured data. A large database, for example, is a few terabytes. A single media file can be 100 terabytes. And so there's just literally no comparison between structured data and unstructured data for that. And so one of the challenges that clients have is, if they use the classic approach, they have to copy all this data into an application for annotation, then into another application for unification and then another application for transformation. It's insanely expensive and complex. It is extraordinarily hard, if not impossible, to carry forward data security, access controls, lineage, all of the things that you want, which makes traceability very, very hard to do. So let's say you run a model and you have drift in the model's results, you have no way to figure out what changed the source data. We envision what we call a data platform that is built to accommodate all the types of data in the world, right? And so we see 3 things in there. The first is the idea that you will have multiple data formats on which your applications want to operate. You'll have enterprise data formats. These are your classic file, block and object where traditional enterprise applications want to access data. You'll see GenAI and agentic applications that want to use what's called a vectorized embedding or a tokenized data format. And then the third is, you want to have metadata operations on what's called a canonical data format. This could be like an iceberg table or it could be a JSON representation of file data. And the data platform needs to support all of them. What we have done, which is unique in the industry, is we are saying, "Hey, you can keep your source data in one place with one kind of a copy. That's the original source data. You don't have to create multiple copies of it but you can present it in different ways to the different applications." So you can present it in a canonical way, for example, in an analytics application that wants to use Spark, you can present it in a vectorized manner for LLMs, you can present it in the classic file, block and object format to a traditional application. And a lot of our original IP in ONTAP allows us to do this. What this enables you to do is to massively simplify your pipeline, right? You can make the pipeline -- you don't have 6 copies, you have 1 copy, which is the original data. You can maintain security, access controls, lineage, everything is able to be built in as you transform the data. And importantly, you can keep the source data and all its transforms in the same volume so that if you delete the source data, you automatically delete all its transforms. Or if the source data changes, we have a data change detection engine in ONTAP that allows you to update the catalog of data and say, hey, these models need to be rerun because the data changed, right? So there's lots of intellectual property that we have built for a long time that helps us do that super, super efficiently, way more efficiently than any other solution in the market. Today, you've got a lot of dumb storage systems with a stupid parallel file system on top of it. They go fast but they can't do any of these transforms, right? They will say, oh, for transform, you got to feed it up into another pipeline. Now think about how much time it takes to extract, let's say, 20 terabytes of data from a storage system, copy it up into an annotation system and rewrite it back into the parallel file system, right? That's the most brain-dead idea I've ever seen. So you know what, we are like saying, "Hey, you want performance, we'll give you performance." And I'll talk about how you do that. But you need the data management because without the data management, you are basically doing batch data copies everywhere. The second element of what we announced was what we think is a new class of data infrastructure, right? Which is, you've seen newer technologies that are essentially what you call memory speed fabrics, where you've got memory speed connectivity across network fabrics. What this allows you to do is, build systems that are highly flexible. You can combine processing, memory persistence or storage in flexible ways. And so what we've done is we've architected a disaggregated system that combines data access nodes and data retrieval nodes, which are classic storage constructs with data processing and transformation nodes which are GPUs or CPUs within the same trust boundary of data access, right? So what -- how does that work? What that does is, it allows you to do the activities that you need on the data. For example, on unstructured data, we said you need to actually enrich the data so that you can get the metadata, right? It isn't created out of the gate so that you can actually now run AI models against it. You can run them on the GPUs but it is completely within -- it feels to the data that's resident on the storage like a trusted user is accessing it. So all of your security and access controls, your protection, your guardrails, all of that's carried forward. So that's really the 2 big things, right? We feel like, hey, a lot of the data platforms in the world, like data lakes and warehouses were built for structured data. They're not going to scale for unstructured data. That to really manage unstructured data and, in fact, all forms of data, you need to embed the intelligence right where the data is created. Like we've done that for security. We did that for storage efficiency. Now we're doing it for data transformation and enrichment. And to do that effectively, we've created a composable system that allows you to mix and match data access and data transformation nodes in one unified system architecture as well as a suite of software services that combines our tech with tech from NVIDIA that allow you to process the data and transform it in place without copies. We're super excited. I was out at the Expo show right after the conference. And I can just tell you how gratifying it is to have 2 or 3 clients walk up to me and say, "Hey, you were listening to my problem. You got the perfect solution." I'm super excited. One was a pharmaceutical company. The other was a manufacturer from Germany that we have worked with for many years. And they said, what's especially unique about NetApp is, you guys keep making ONTAP better and better so that the investment we've made in your tech, now you're making it available in so many new ways. So thank you for coming. Have an awesome conference.

All right. Thank you, George. Appreciate that. Now I am happy to introduce to you our new Chief Product Officer, Syam Nair. He will tell you a little bit more about what we're doing in AI and give you guys an opportunity to ask questions.

First off, thank you. Thanks for coming. Hopefully, you had a good start today with the conference. Thank you. Look, I -- first, I'm new but I'm super excited. Super excited because 3 things. One, this is the time where customers are really looking at navigating 2 secular trends. Cloud, it still continues to be a journey for most enterprises. AI transformation, everybody talks AI. You hear about it everywhere. But most of the investments today is on the compute side of it. The real value for AI comes from data. Data is growing significantly. And not like the Hadoop days, now we are talking about LLM models, machine-generated data, growing unstructured data. This is actually an explosion where a lot of value sits in the data. It's really, really, really hard to get value out of the data. Okay. Anybody who has worked across the data industry for anything knows that the overall processing of data is where most of the complexity, cost, time spent is, right? We can change the game by bringing in this intelligence that George talked about to the platform. And intelligence into the platform does mean like think about this, in the future, every unstructured data set is like a database. If you know about iceberg, imagine everything is an iceberg table. You can just query, run search, semantics, data models, vertical models on top of it. Like it's just not a vision. It's something that we are executing towards now with our AI data engine. That's where we have actually built in a metadata engine on the ONTAP platform. We have a vectorization on the ONTAP platform. We have built in guardrails for AI security because a huge challenge for most of the people who are actually trying to get anything out of AI is how do you secure that data? How do you protect the data? These things are built into the platform. And I think it's a unique opportunity for us, unique opportunity to serve our customers who actually have this data set across industries, whether it is media and entertainment, pharmaceutical, manufacturing, across industries, this is the same problem that we get a chance to solve it. So most of the announcements today for us was actually moving on to the journey. Now one of the other key differentiators is the flexibility we give customers. [indiscernible] really proud to say that nobody else can do this because we have the same platform running across any of the hyperscalers and on-premises. And data grows in hyperscalers and AI factories and on-premises. They are not going to be moved from one place to the other. So keep the data where it is, like technologies like FlexCache that we have built into the ONTAP platform, SnapMirror, data for AI can move to the edge without actually copying the data, like data sprawls. I don't know if you talk to customers, I've talked to several customers who have said, sometimes, no exaggeration, sometimes 60 to 70 copies of data that they don't even know where they are because every departmental data is moved on. Some are moved on to the lakehouse for harmonization, modeling, et cetera. So like we can cut all of this and really bring the data to that. That's the power of AIDE that we are actually building. I've been part of the data outcome technology for some time. After my operating system career, I was mostly with databases, NoSQL, Big Data. Many of you may know it was a big thing. Hadoop was going to change the world. It did. It did actually create a new set of applications. Cloud changed the world. It did. It did create a new set of applications, new way of managing data. But some of the challenges in terms of getting value out of unstructured data continued. And now is the moment because there is compute. Now is the moment because there is storage capabilities that is intelligent where you can directly get intelligence and analytics out of it. So that's the excitement I have in terms of what we are embarking on. AIDE, AFX, true disaggregated storage. This is where we are actually building disaggregation on top of all the data management capabilities because you disaggregate compute and storage but you need all the data management capabilities. That's the most precious thing for most of the customers. It's the data. It's the semantics of what is in the data. Each file is just a file without the metadata. Once you look into the metadata, it has lots of precious information. How do you protect your IP? How do you protect the attributes that are in the data? How do you actually make sure that's your key value that you can continue to protect. These are all things that data management, AI security capabilities built into the platform provides. So AFX plus AIDE, the cyber resilience capabilities that we have had, right? Everybody world over, I came from a cybersecurity company before this. Most of the threats in cybersecurity comes from AI. AI-driven threats are growing. I don't know if many of you know this, the average time for a threat to break out is 2 minutes. And many of the people who say, "Oh, once it happens, I can actually figure it out." No, harm is done. You need to protect it before it happens, which means that protection needs to sit where your precious commodity is, where your precious asset is, which is data. So I think, super excited. The fact that we have AFX from a disaggregated storage standpoint. We have AIDE, which is a data engine built in, the intelligence built in, cybersecurity built in. I think it's a huge growth opportunity for us to build that capability and grow that capability. And then cloud across hyperscalers. Now, again, nobody else has this where all of the hyperscalers have all the functionality. What I was showing at the keynote, I don't know if you got a chance to watch that, where once I copy file, it can show up in every place where I can do read/write but I'm not really copying it across there. I can use FlexCache to bring data where the compute is. That's actually a phenomenal thing. It was actually built years ago for ONTAP. It was built years ago. It wasn't built for AI. But now is the time where everybody can leverage it because customers are adopting cloud. That's where most of the AI adoption is going. Customer data sits on on-premises because there is a lot of data over the years. This is their IP. So the opportunity for us is big. I look forward to continue to innovate in that space, create new business opportunities for us to grow. I've never been this excited in terms of what we can actually do to delight our customers, continue to keep the trust of our customers, make sure that customers are successful. So super excited to be here. Thank you and [indiscernible]

All right. Let's get some questions going. Ananda, in the back.

Yes, great keynote this morning as well. I guess, Ananda Baruah at Loop Capital. How -- could you describe to us how we should expect this to -- like the manifestation of this to begin to show up in the business? Where is that journey today? And is this -- is really what we're talking -- you're talking, you guys are describing, you and George, is it really a share gain story? Or is it a market share story, maybe more appropriately as this whole dynamic gets going?

Yes. I think it is on 4 fronts. One I would say is AI-ready storage, it's still a challenge for most customers. Everybody talks about there are new stacks but customers already have infrastructure that is -- they are leveraging. How do you make those storage AI ready? So that will be a share gain in the context of across blocks, files and objects, we can provide that one platform, reducing the complexity of managing multiple systems. So that is going to be a share gain story for us. The second one is a share gain as well as additional value is going to be cyber protection. I think, look, most of the money is being spent in terms of cyber protection outside of infrastructure. AI gets all the hype but cyber, all the way from zero trust to making sure data security. Data security is one of the biggest problems that is facing the industry today, especially when it comes to AI. When you talk to customers, you find that most customers tend to either open up or close down. There's no middle ground out. And most of the AI security value they get today are visibility, what is being used. Many of your enterprises would also have the same challenge. We can bring in AI security directly to the storage, directly to the data platform. That should give us an uplift both in terms of storage as well as value add that we are providing customers because customers can now take away additional tools and expenditures they have to rely on the platform. That's the second part of it. Third is going to be the AI data engine. AI data engine is going to reduce the complexity of what customers have to do to get their -- make their data AI-ready. Look, George talked about how lakehouses and the overall -- if you're familiar with the bronze, gold model of taking -- bronze, silver, gold model of creating data, harmonize it, model it, then create entities and find value on top of it. I think there's a lot of work that customers are spending there that they will see value in not having to do. I'm hoping that will bring us more additional value add in terms of what we're doing. So I would say storage is one but cyber and AI data engine would actually drive more.

Ananda has got a follow-up, then we'll get to Lou.

That's great. Just a quick clarification -- not clarification, just specification. So for cyber, do you think you capture cyber budget from other folks? Is that how you see it? And then, just in your experience, where do you think customers are right now with their proof of concepts and their journey to inference? That's it.

I think 2, both on the cyber and the AI front, I don't envision us being a cybersecurity company. It's more about bringing that value to so that our software, our hardware, our systems are more valuable. So it becomes a premium for what customers have to spend on us, reducing their budget somewhere else. I'm not looking to be a pure cyber play company because it's not the core competency that we are in.

Let me, I guess, tailgate a little bit off of what Anand just said. This is Lou Miscioscia at Daiwa Capital Markets. So on the last earnings call, George talked about 125 AI infrastructure wins. So just trying to understand, when customers have their proof of concepts, they're trying to do something, what's holding things back in the sense -- I mean we do hear about a lot of applications but there's an awful lot of money obviously being spent with this infrastructure. Really like to see and know if enterprises are really starting to move forward in size, which obviously would then justify all the investments that are being made or the justifications for the investments only really with the big cloud and the mega tech companies.

I'm not sure that there is holding back per se. I think the -- most of the spends have been more on the compute side of it, experimentation side of it. I think customers are seeking value in terms of trying to get this value out of data. I think it's more an opportunity for us that we'll continue to see. We are growing really well in object storage because object storage for lakehouses is a key ingredient for AI. And with AFX and AIDE, there's a keen interest from customers wanting the disaggregated storage and having -- being able to build AI engines on top of it. Like just a proof point in terms of just our own AI session, I don't know if it was overflowed but both the room full was overflowed and people are talking about, look, this is what I've been looking to. People have a lot of interest. Customers have a lot of interest to see driving AI value directly out of it. So I think it will grow for us significantly over the next few years.

Kat Campagna from Goldman Sachs. Shifting gears a little bit. I wanted to ask a question about today's announcement with [ GCP ] and the new block storage capabilities that you talked about. Why was this important for your customers? What does this really help? And how does this change the outlook that you have for public cloud growth into next year?

Just really quickly, we do have a cloud expert coming later today but...

Yes. [indiscernible] will talk about it. So a short summary of that is, look, customers, as they're migrating, they're leveraging cloud more and more today than before, right? And customers are seeing the flexibility and scalability in extending to the cloud. And every customer has a cloud choice or multiple cloud choices. The unified platform where customers see the simplicity of being able to leverage both NAS and SAN is important for them when they think about workloads. Virtualization is a good example of workloads that they're moving to the cloud, where having block capabilities actually help us. New AI projects, one of the things that is stopping AI projects is the capital-intensive nature of it. Many of the customers are leveraging new AI projects in the cloud. So having block capabilities actually help there.

All right. We have a question for Wamsi.

Wamsi Mohan, Bank of America. I was wondering if you could just talk about your announcement around the new DGX SuperPOD, like you're qualified for that. And I think back about 6 months ago, you had AFF A90 that was qualified for that. So how are customers looking at this versus the prior? Is it different -- is there any difference in the software offering? Is it purely the scalability across maybe compute and storage that's different? How are you positioning that for the market?

Yes. It is more on the capability of -- there are still workloads that don't need a disaggregated architecture. Like there are also workloads that are very focused on disaggregated architecture, especially where you have to checkpoint when you're doing model training, et cetera. So the AFX and the SuperPOD certification helps us actually get into a market that we weren't there much because AFF could actually help in terms of the inference on. So it's more an expansion. The other thing is it's also a proof point for us in the context of as compute and storage and compute and data -- data platforms are getting segregated, we can actually play a full fidelity role in the NVIDIA ecosystem. So we are working with what I would call as Neoclouds or AI factories that are coming up, being -- trying to be the partner there because we actually have the assets now to be with NVIDIA and go and win those games.

We have a question from the webcast.

This is on behalf of Aaron Rakers at Wells Fargo. Just given AFX is a new and incremental part of the NetApp portfolio, how should we think about sizing the TAM opportunity that new AFX systems and platform address? And I have -- there's a follow-up too.

I don't have -- know if I have an answer for that, Sandeep or...

Yes. I think we'll see if someone later can answer the TAM question. If not, I'll do some research and get back to you, Aaron.

And who are -- in terms of AFX, who do you see as the key competitive platforms? Is it Dell Project Lightning, VAST, WEKA, Pure? Could you kind of expand on that?

Look, I think from my perspective, it is an opportunity for us to gain. I'm not looking at this from a pure competitive standpoint, right? There are competitive products out there but our product because we are building on top of the ONTAP platform, all the data management capabilities that we have as well as the fact that whatever we are building as a platform is available in every cloud, that differentiates us. So it's -- to me, it's not a competitive play. Look, we will be the best platform for customers to solve this. That will actually help us grow.

Yes. Mehdi Hosseini, Susquehanna. Two follow-up. If I just step back and look at the announcement today and look at all the comments you made, would it be fair to say that the vast AI opportunity for NetApp is still focused on enterprises. I understand most of the investments so far has been on compute but many of these investments are also enabling native data, native workloads to benefit from AI. And I didn't hear anything that would give me confidence that you have actually expanded your exposure to hyperscalers and perhaps it is the enterprise given your ONTAP installed base and additional products you introduced today would actually help you with the incremental opportunity on the AI side. Would that be a fair way of summarizing this? And I have a follow-up.

I would say both. I think it is fair that we are strong and we will expand. But when Pravjit comes, he can talk about the number of new logos that we actually have on the cloud, these are not NetApp customers. We are bringing in new customers on first-party NetApp offering in the cloud. And many of those workloads are for high-performance and AI-related workloads. So I think there's a growing trend of using AI within the cloud that will also drive given some of the innovations that we have built.

Okay. I heard you talking about data management. Is this a new focus area, especially with enterprise opportunities related to AI? Are you trying to expand your installed base of storage and add another layer of value-add services?

Yes. It is the -- so if I talk data management in true context, just to be -- so the -- there is the storage-based data management capabilities that were built into ONTAP, that's part of the ONTAP. There are also functionalities in terms of -- where cyber protection is a good example of it where data has policies and that policies travel with data. Now when you think about AI workloads, that's going to be something that is going to actually accelerate adoption of the AI workloads. So when I talk data management, I'm talking about the core data management capabilities, but this data engine and some of the other capabilities that we are running, those are net new, enabling newer workloads, new scenarios for customers. So there's both sides.

Does that mean that you will go back and provide a mix of hardware and software within your product revenue?

I'm sure I'll update. Yes. Not a question for him. All right. Tim?

Tim Long at Barclays. Just wanted to get back to -- I think you've kind of touched on these a little bit, AFX and AIDE. Talk -- kind of newer solutions, so can you talk a little bit about sales force channel customer education to realize the benefits here? And as a result, does that mean a little bit longer path to revenues than -- or to deployment than some of the existing technologies? And just curious if in one or both of them, do you think there's a little bit more of a software maintenance, software services bent? Or is this more similar to some of the other product innovations?

Is César or Dallas [indiscernible]

No.

No. So I'll touch base on it. I may not be able to answer the whole question. Then, maybe Kris, you can, you may be able to follow up. The -- for us, both cloud and hybrid cloud in the context of it, what we are building as a product capability is going to serve both the customer bases. So that becomes added value for us, like we are working with ANF Azure, [ GCME ] from a Google standpoint as well as AWS to make sure that all of the platform capabilities are running there. As an example, AIDE that we have delivered is built on top of ONTAP. It's actually available as part of a system that we are selling AFX and with NVIDIA nodes. The same software can actually open up entire ONTAP estate, be it on the cloud for customers. So that's going to be additional value add. The exact go-to-market motion and business, I don't have an answer. Kris, you can. Yes.

Yes, we can follow up. And you'll probably should expect to hear more on the earnings call about how we're taking this to market. All right. Samik?

Samik, JPMorgan. Maybe if I can go back to AFX and sort of you described it in your keynote as like meant for exascale level storage at that point. Maybe more directly, does it really position you differently with some of the new clouds that have been looking at these opportunities? Does the scalability help position you differently? And then a follow-up on the cyber resilience or cybersecurity as well. Like mentally, I'm thinking you're just going to go up against like the companies like Rubrik, Cohesity. But why would the customer then sort of think about allocating some of the budgets that were addressed to those companies over to maybe paying more of a premium for your service?

Okay. I'll cover the product side of it, not exactly the financial aspect of it in terms of -- so the AFX one, yes, it does position us well with some of these -- you think about these AI factories, giga factories that are forming across the world, like sovereign clouds, it does actually position us much better to have a good footprint because now we are providing not just -- I think George used the term dumb storage, it's actually smart storage that can actually help customers drive value out of it. It does position us well. We are working with some of the partners in terms of how we become part of this ecosystem. It's in early stages of it but technologically, it positions us well. Cyber, I think it's a two-pronged thing. Look, we want to be the best cyber -- not just from a resilience and data protection standpoint but from an AI security standpoint built into the platform. which should make our platform much more easier to use, reducing the complexity for customers, much more secure, gaining trust. That itself is an added advantage to us from a platform. And then what we want to work with is, we want to work with the ecosystem vendors, like there are others out there to be able to leverage our APIs. So we want to work with open ecosystem, with the partners. So we're not looking at from a product positioning standpoint, this to be an alternative. How exactly the dollars move around, I think we'll have to figure it out. But I think technologically, it will make us much more advanced and ready to be the platform of choice for most of these workloads in the future.

All right. Steve.

Steve Fox with Fox Advisors. I guess I'm still a little bit confused on just the competitive advantages that you're laying out because you've had the advantage with ONTAP. You've had multicloud and on-premises. So like what is different or what came together in terms of these product announcements today that sort of you're bringing together and [indiscernible]

Thanks for that question. I'll clarify. The disaggregation allows our customers to scale performance and capacity independently, which was a challenge and many workloads, not just AI, many workloads need it. Media and entertainment is a really good example of it where there is lots of [indiscernible] and they want performance from a different standpoint. So that actually opens us for multiple new workloads where we have been playing but we become a major player in that space. That's #1. The AIDE in itself is, look, today, customers do get value out of it because it's complex. The pipeline that George showed, if you were at the keynote, that's real. That's actually from a particular customer and many customers and I've been in many of those shoes. It's really hard to move data, multiple data copies and transformations happen to really make it meaningful. AIDE takes that away and that is now available across the platform everywhere. So for customers, I think the differentiating factor is, if the data is in the cloud or on-premises, immediately, all of this data is accessible for AI without having to do that complex processing, complex set of pipeline that is needed. And that's a huge opportunity because it reduces cost for customers. We provide more value. So it reduces the complexity. And it also like actually helps them transform their AI projects much more faster. Like industry analysts have been talking about that, it depends on who you listen to, 40% to 60% or 87% of the AI projects in reality in enterprises are not succeeding. And most of them are not succeeding because the data isn't ready. The infrastructure is not connected for AI. These 2 innovations that we showcased today and that we are delivering actually connects customer storage and their data, especially their unstructured data estate to AI so that they can drive value. So I think it's a huge innovation from that standpoint for us.

All right, Frederick, in the back.

Frederick Gooding with William Blair. There was a nice slide in the keynote earlier talking about the combination of data management services, the metadata engine and then also unified data storage. I'm curious, do you see any of those specific segments driving either more interest, customer demand or as like a higher competitive advantage? Or do you think it's more of the fact that all these are integrated within a platform together and that is really -- sets you apart from anybody else?

Look, it's -- I could have, with all of the innovations, could have gone and claimed we are the new database. I'm not trying to do and claim something that it is not. What I really, really want to make sure that everybody, all our customers understand is the amount of work that it takes to get value out of this is simplified now because we are able to do it on the platform. It's additional compute, it's innovation, it's on the platform. So I would expect it to drive both, which is to make us the platform choice, also drive more value towards our platform because now you don't need to do all the other aspects of it. So it's not -- it is going to be a competitive advantage purely with the competitors we have. But it is also going to be a competitive advantage in terms of winning workloads because there are many other steps that can be eliminated for the customers. Does that answer the question?

All right. Ananda?

Okay. Quick follow-up. To your comments about starting to have conversations with the Neoclouds, do you know if there's any distinction to make within those conversations between training opportunity and inferencing opportunity?

Yes, there is. I think -- and this is something from a product standpoint, we are also looking at in terms of many of the major needs that is actually driving some of these data center and Neocloud investments are training opportunities. And in many of this training, the consistency of the data can be eventually consistent. If you think about the old database world, there is the atomic consistency and the eventual consistency. So how you checkpoint, if you go into the -- how you checkpoint, et cetera, can -- it's a little bit more relaxed. The ONTAP platform is actually built for full consistency. Like this is about consistent data. So we are also looking at where needed, tune it or provide that offering so that customers can leverage those eventual consistency, a different checkpoint need. Over time, I think as years go by, maybe quarters go by, I would -- my sense is and I think this is what the industry expects is, more and more of this will move towards inference. It's only so much you want to train models. And these models are changing on a -- every other day models are changing and there's a lot of money with the big players actually training models. So it's going to be more inference. We want to be and we are ready for that world where customers want to infer on the data that they have. So that's -- so it's a two-pronged one but I'm seeing more opportunity on the inference side of it rather than the training.

All right. Any more questions from anyone? So I have one question for you because it's a question I get from these guys all the time and I'm surprised it hasn't come up. But as you mentioned earlier, we see a lot of investment of AI on the compute side. And we all know that data is important to make AI valuable. Why haven't we seen a commensurate investment in storage so far? And what do you think would drive that?

I think it is a -- it's just timing. One, in the context of -- look, if you look at -- I don't know which analyst, AI-defined storage and AI is at the peak of the hype cycle, right? As more and more production workloads go into -- start executing and become live, you will see -- we'll see a lot more importance of data and storage. And it is not going to be purely a capacity game. It's going to be about that balance between performance and capacity. I would expect to start seeing that. And this is why the timing is right for us at NetApp in terms of what we are delivering. Look, we -- this is the right time to actually have AFX, AIDE. And we -- like Sandeep is here, my colleague, we are going to go after customers and talk to customers and showcase this value and get great wins. So I'm super excited. It is going to happen.

All right. Well, that's a great note to end on. Thank you very much for your time.

Thank you. Thank you, everybody.

All right. So thanks, everyone, for your questions. So our next presenter is new to you. He's not presented to the financial community before. He's -- but he's been with NetApp for several years now, 3, several, that's right. Okay. So Gagan Gulati, he heads up our value services and he's here to talk and focus primarily on cyber resilience. So with that, I'll introduce him and then let him say a few remarks and then you guys can open up with questions.

All right. Perfect. My name is Gagan Gulati. I'm the SVP and GM for our Data Services group. And as Kris mentioned, today, I'm going to talk mostly about -- today, I'm going to talk mostly -- oh, there is a mic outside. Okay. Today, I'm going to talk mostly about cyber resilience and the -- and what we're doing there. I want to talk about a few things that we are introducing and working on. First is Secure by design. NetApp, we believe, is the most secure storage on the planet. And we have been working extensively to keep it that way. In the Secure by design category, I wanted -- we announced a bunch of new capabilities around post-quantum cryptography. So that's pretty cool and our customers are loving it already. Second big thing, over the last couple of years, we have worked extensively on a piece of capability that's built into ONTAP called ARP or Autonomous Ransomware Protection with AI models built in. It provides our customers with real-time built-in capability for anomaly detection and it can, with 99% plus accuracy, detect ransomware attacks and then take snapshots as need be and then inform -- alert the security operations team about these alerts. It's been a functionality that's -- and capability that's growing rapidly. It's the fastest-growing feature that has been consumed by our customers today. And what we have delivered recently and we announced is that this ARP AI capability is now available for all of NetApp data estate, whether it's for the file workloads, block workloads, including cloud. So we have also announced this capability working with AWS for FSx and we're bringing it everywhere. So the ARP AI portfolio continues to grow and it is helping our customers secure their data estate against the biggest problem they have today, which is cyberattacks. That's #2. Third, ransomware resilience. So we are bringing to light a brand-new service, a value service on top of our ONTAP platform that we call as the Ransomware Resilience service. This Ransomware Resilience service, we are introducing 2 big capabilities today. Again, the first in the industry to deliver what we call as data breach detection capability. Most of you guys know that when it comes to cyberattacks, the world is moving or the attackers are moving towards what they call as double extortion attacks. They will first make a copy of your data, export it. And then they will encrypt the data and then charge you ransom for both of them, one for decrypting the key, right and one for the data that's exported and they'll say, oh, we'll give it back but they may never give it back, right and use it and harvest it later. So what we are announcing today is the ability for our Ransomware Resilience service to help our customers with this capability that we can actually go and detect data breach attacks. And we, of course, do it in real time, not after. We do it as it is happening. As we detect these attacks, we will generate alerts. We work with our own UEBA or user entity behavior analytics tools. We work with our partner companies who do network firewalls. And we, of course, integrate with the likes of Cisco Splunk, with which -- with whom we are announcing a pretty big capability, where we are working with Cisco Splunk SIEM and also their SOAR capabilities, which is security orchestration and response capabilities, so there is bidirectional work. So that's a amazing set of capabilities that we are announcing today with Ransomware Resilience. That's #1. The second big capability we are announcing that our customers have been asking us for is what we call as the isolated recovery environments. As you know, when these attacks happen, there is no guarantee that when you're using -- when you are recovering from them, the backup copy from which you are recovering is safe. Studies show that 75% of customers who have a ransomware attack end up getting attacked again. And 1/3 of them actually by the same attacker. Why? Because when you recover these -- recover from these attacks, the copy that you're using is already malware infected because the attackers have affected -- attack -- they have affected not only your primary copy but also your backup copy. So what we are delivering today is this ability to be able to recover from these attacks with confidence because we take the -- we run all kinds of malware detection, virus detection, all of our strength of data science on these backup copies and snapshots to make sure that when we recover them in an isolated environment, these snapshots and these backups are safe and these are not infected. So that's the second big capability we are delivering as part of the Ransomware Resilience service. Now shifting focus away from capabilities to where we are taking our premium value services. We are -- these capabilities are, of course, made part -- available as part of our ONTAP licensee, ONTAP One licensees but also something that we make available to our customers to purchase through marketplace. And that's one of the areas we are very focused on to make it easy for our customers, our partners, to purchase these directly from the marketplace, hyperscaler marketplaces like AWS, Azure and Google. And what we're also announcing today and tomorrow is this new Ransomware Resilience service. We want all our customers to use it and we are announcing a trial, 6 months trial of this service for free for all of our customers up to a particular limit. And we hope that all our customers utilize the power of Ransomware Resilience against their NetApp data estate with this capability. So with that, I'm going to stop and take questions from you. Thank you.

Frederick Gooding with William Blair. I'm curious how should we think about this ongoing convergence between storage, DSPM and like backup and recovery, you guys just announcing the isolated environments. I'm curious, like do you think it's more of, all right, we need to consolidate everything together like on top of the storage environment, we need the recovery, we need the backup, we need the data classification? Or is it more building interoperability with all those different types of capabilities?

It's a great question. So look, I mean, security is best done in depth, right? What I mean by that is every customer of ours, when they have to protect themselves against these cyberattacks, right, they start from the top, network security, perimeter security, so you have a bunch of firewall implementations that you have to have to make sure that you are protecting against network attacks. You also have to make sure that you have identity security so that compromised -- at end of the day, it's the compromised users and those identities that get used everywhere, right? And #3, then is about data security because end of the day, it's about those crown jewels. And this data sits on storage, right? And storage, therefore, becomes the last line of defense when everything goes wrong. But it doesn't mean that you only implement that at a storage level. You just can't have data protection done at storage level, security at storage level. You have to make sure that you are securing all different layers of the stack, if you want to call that. So that's #1. #2 is -- it -- and it takes a village, right? To the example you gave, data classification, super important to get right. So you know what's sensitive and what's not sensitive. What's sensitive is what you want to prioritize and go protect first. At NetApp, we offer a capability called NetApp classification that makes -- that we -- that allows our customers to classify data. But at the same time, we work with various different classification vendors, DSPM vendors today to make sure that they can efficiently run data classification and security from an example of a DSPM or similar other examples like DLP, on top of the data that is available on NetApp storage. Of course, they can do it in the most generic way like NFS protocols, just files or they can utilize the best of what NetApp offers in terms of our own data management capabilities and they can integrate directly with us. So DSPM, DLP vendors, that's just one part of the picture. The second part of the picture, of course, is data protection vendors. And the entire ecosystem, whether it's the likes of Rubrik, CommVault, Cohesity and others, right? So they -- at the same time, same story there. We want to make sure that our customers get the best data protection, the best cyber resilience. So we work very closely with data protection vendors and ISVs as well to make sure that the whole partnership utilizes the best of data management from NetApp so that the customers get the best defense against such attacks. So it's going to be -- it's a -- we, of course, want to make sure that we offer the best we can offer to our customers. But at the same time, we have an entire partner ecosystem that make -- together, we help our customers secure their data and keep them resilient.

So maybe just going back to -- and you referenced this a bit -- Samik, JPMorgan. If you can talk about, firstly, what's the current solution that enterprises are using? And when you sort of think about competitors in this field, who are you going to look to really displace on that front? When you're charging it as more of a premium service, how do you sort of envision that going? And then would you -- would this -- you said sort of you want to do it more in depth and probably the competition is where it's more going with the breadth rather than the depth on that front. But would you evaluate in the future like going and doing this on a different storage platform or a third-party vendor storage platform? Obviously, that won't give you the depth that you're looking for but is that something that gets you more into that pure-play competing in that area.

That's a great question. Look, #1, we -- our first and biggest job is to drive preference for NetApp storage. So it's -- for us, it's about making sure that our customers who have their most crown -- biggest crown jewels, their biggest workloads running on NetApp storage are safe. So we are basically, therefore, building both of the platform to allow for the -- our entire ecosystem to integrate with us. But at the same time, building vertical products, like you said, to make sure that we give our customers the ability to do it in the best possible way because we -- our products like Ransomware Resilience service is going to always use the latest and greatest of what we can offer from ONTAP layer and then, of course, go and deliver a new set of capabilities that we, of course, want our partners to integrate with over a period of time as well. End of the day, it's a dual job in that sense. So that's part one. Second, about the breadth play. We are very focused on helping our customers with where they want us to help them. We have certain products in our portfolio that go beyond just the data available in NetApp storage. For example, our observability solution, which is DII, that helps our customers with observing and monitoring their entire stack because that makes sense over there, right? And it's just not focused only on NetApp storage. We -- of course, as part of the road map, we depend on what our customers tell us and ask us to do and we will continue to evaluate. But again, our preference is making sure -- our first rule of the game is to help sure -- make sure that we keep our customers' data on NetApp safe and secure.

Tim Long at Barclays. Two as well, if I could. First, just curious if you could talk a little bit about like the solution you gave the example of working with Cisco and Splunk and some firewall. Just talk a little bit about kind of what's the NetApp IP in this solution and what's the partner IP in the solution? And then shifting to the -- you may have answered part of this, but shifting to the -- shifting over to the hyperscale storefront model. Is this kind of the first major add-on of security to that offering and was there kind of an ask for this? And similar to other cloud offerings, was there a level of co-development with the hyperscale partners.

Okay. Sounds good. So let's start with the first part, which is the partnership we have with Cisco Splunk and what are we doing there. See, end of the day, when you look at the security tooling like what Cisco Splunk offers. It's about ensuring that the infrastructure players like us delivered the right set of alerts into the SIM and source solutions so that the security operations operator on the other side is able to take action, right? So the IP here is basically detecting these anomalies that there is an attack going on in real time. It is built into ONTAP and then making sure that we deliver extremely high-quality alerts into the scene, right? So that's the first of our IP. Just to kind of cover that conversation fully, ARP/AI, the capability we're talking about, has been independently tested by multiple different security labs across the globe. SE Labs that's based out of London announced us as the winner this year for enterprise data protection category for -- and gave us a AAA rating at first count -- at the first testing count. So it's absolutely amazing. This capability is top notch. So that's part of our IP. Of course, when we deliver these alerts into Cisco Splunk similarly in other similar solutions like Microsoft Sentinel and others. It's not just about the alert, it's about sending a lot more data along with the alert to make sure that the security operator on the other side is able to take action, right? And so that's basically all part of that IP. Of course, security operator on the other side can then come back and start working with the storage admins to see where the -- what needs to happen next. As part of the latest integration that we did with Cisco Splunk. We've actually went the next step with data breach detection, which is not only do we send the alert over to Cisco Splunk. We also now allow the security operator to block the user from going and making -- causing further damage, right? So we allow -- this capability is now built into Cisco Splunk. So that's part two of the security innovation and IP that we are sending, which is who is this actual user, who's the user, who's making -- who's exfiltrating data, right? And then working closely with the Splunk team to make sure that the security operator can block the user right away and therefore, not cause further damage. So that's an example of the IP that we create and then, of course, we work with the entire ecosecurity ecosystem wherever we can to deliver the capability that our customers demand of us actually at this point. Coming to working with the hyperscalers. Absolutely, right? So not only -- so all of these services like the ransomware resilience service or the NetApp backup and recovery service, which is well used by our customers. These services are SaaS services, right? And the control plane is hosted in hyperscalers, and they're available through the marketplace. So of course, we work very closely with our hyperscaler friends to make sure that these services run the most optimally. And at the same time, when our customers purchase these services through marketplace, whether go through the pay go model or a private offer model or whatever models their hyperscalers enable, we take part in that. We make sure that we make it available to our customers with all the innovations that those guys are driving a lot more from coating and pricing perspective and ensuring that the customers can, therefore, utilize their existing hyperscaler commits using these services. So that's a lot of the work we do with our hyperscaler marketplace teams there. And last not we -- of course, these services not only run against our storage that's in customers' data centers, but also our customer that -- the story is that we have natively built in all 3 hyperscaler clouds and our CVO offering. So these services basically -- of course, we innovate with our hyperscaler partners there as well.

Wamsi Mohan, Bank of America. So right up the top, you mentioned something about designing for Quantum -- and I was just curious, like, are you actually finding customers at this point worried about this? How true of a worry is it at the moment? And is this sort of a future proofing? And what exactly are you doing here to achieve that?

Fantastic question. Look the threat essentially or the risk that our customers want to mitigate today is basically harvest now and decrypt later, right? That's as simple as that, which is I have -- if I can go and steal your data now, even if it's encrypted with today's algorithms, it's all right. Later, I'm going to come and decrypt it when I have quantum computing because now you can actually go and decrypt all of this data and charge a lot more essentially. That's basically the threat. And we hear about quantum computing coming up and improving. It's not viable yet, as you all know, but it's going to happen. So what's happening right now, therefore, is multiple different government agencies, of course, are starting to put in new standards for computing and -- or in cryptography. For example, AES 256 is an example of an encryption technology that we just have built in now at rest, and we encourage our customers to start using that instead of a previous encryption algorithm, which is not quantum safe, right? And as more and more standards come into play, our job is to ensure that our operating system ONTAP is fully capable of encrypting customers' data with that algorithm of their choice, right, and continue to basically grow into that journey. So that's our job. Our job is to make sure that ONTAP is always at the pinnacle in terms of the standards, in terms of the encryption algorithms that we have to make available to our customers. And therefore, that's what we do. And our customers job is to make sure that they utilize those encryption algorithms on their journey. And to answer your other part, our customers demanding this? Absolutely yes. right? And that's why we -- all of these innovations coming in because some of the biggest customers, financial institutions, specifically and others, those are the ones who are demanding that we continue to improve, and we will always be, I believe, ahead of the game as the most secure storage.

Frederick Gooding, William Blair. I'm curious in terms of how important securing metadata is? And then also, if we look out over the next 3 to 5 years as AI becomes now more important, more integrated within the enterprise. And Kris might tell me to shut up here. But I guess where do you share the future gaps that are within the NetApp portfolio in terms of securing in what you guys are maybe looking at down the road?

Well, definitely not the future gaps, but maybe some opportunities to continue to enhance our platform.

I believe that -- we've already talked about AI data engine.

Yes. Okay. That got announced this morning.

Yes. We announced it earlier this morning, and we just ran a session before this about AI data engine. I mean, look, like you said, right, the AI journey is just starting. I mean customers are moving from AI pilots -- enterprise AI pilots, I mean, enterprise customers towards taking these projects to completion. And we all know, we've all been in the industry long enough to know that overall, the AI-ready data doesn't exist, and I think that's where most of our customers are and what they're trying to do right now is to start just cataloging data, which is basically starting to put the whole metadata together for all of their data so that it's easy for our -- they can make it easy for their data scientists and engineers to utilize it. Of course, Metadata, you can actually infer a lot from Metadata about the actual data, whether it's file name to the various attributes who's changing it, what have you done with it, the tags, et cetera. So there's a lot you can inform from metadata. And the metadata store, there for the catalog, needs to have the same level of security that the actual data has, right? So you have to make sure that your metadata store, your catalog has to be safe. It has to be -- it has to -- to make sure that only the right people can get access to it, not just -- not at the level of users, like which user can access what part of the metadata to find and access the data they want, but the actual metadata store that's on NetApp storage, we have to make sure that it's as secure or probably even more secure than the actual data is. So that's part of our job. And it's not like a gap. It's something that we actually do today. What we ship today is part of the functionality. So of course, we'll keep improving as customers tell us more. But for us, that metadata is actually end of the day nothing but data, right? It's customers' data that we have to just secure as well as we secure the actual data.

All right. Well, I'll ask one last question because I think while you've been focused on cyber resiliency here because we've had some cool announcements today, I think we have a number of premium value services available today through the marketplace. And this audience probably is less familiar with them. So it might be good to explain what they are, how they work in customer environments and how we deliver them.

Oh, that's perfect. So yes, so there's a lot -- so what -- so when you talk about premium value services, we are delivering end-to-end orchestrated SaaS-based services to our customers in 3 big categories. And all of these services are available to our customers through marketplace or a typical licensing model as well. The first one is around cyber resilience, that the one I talked about. And we delivered 3 big services, Ransomware resilient service that I mentioned already, with a lot of great capability that we are working towards. Number two is our unified backup and recovery service. This service is used by hundreds of our customers to go and back up their data and then recover that when need be. This service has been in existence for a few years, and it's available through marketplace, and it's a well -- very well used service. The third big service in the cyber resilience portfolio is our disaster recovery service, which helps our customers protect their VMware-based workloads, so they can have orchestrated disaster recovery from on-premises data centers to on-premises data centers. And we are the first ones who actually have made it available such that they can do a DR of their VMware-based workloads from on-premises to AWS. So you could have a workload running on premises. And if your DR strategy says that I want to actually have my second secondary site running in AWS, no problem. We actually work with our AWS team, and we were the first, ISV if you want to call it, to have -- make this functionality available to our customers. So that's the third big piece of the puzzle in cyber resiliency that orchestrated end-to-end services that we make available to our customers. So that's the first pillar. The second big pillar for us, of course, is AI. What we announced today with AI data engine. And that is also going to be available to our customers through marketplace. So that's second big pillar, and we are working towards that. The third big pillar for us is what we call as governance. And what we have available today there is a very well-used service called DII. And this service is also, again, available to our customers through marketplace so that they can use this SaaS-based services for observing their entire environment. And I think to the question that was asked earlier, this service goes beyond just NetApp storage and gives us -- our customers the monitoring and observability capabilities for their entire environment. So 3 big buckets of marketplace-based SaaS services. in cyber resilience, in AI and also, last but not the least, in the field of governance, and we're starting with storage governance and infrastructure governance and over a period of time, we'll do more. So those are the 3 big pillars that we have over there.

All right. Any final questions in the audience? Well, I appreciate your time very much today. Thank you so much for your debut voyage with us. All right. Okay. So our next speaker. You guys have all heard from before Sandeep Singh. He is the SVP of Enterprise Storage. So all of your flash block and probably AI questions. We'll send them his way.

All right. Hello, everybody. As Kris mentioned, I'm the SVP and General Manager for enterprise storage. I've been part of NetApp for 3 years, and my background is predominantly in enterprise storage. I was part of a startup a long time ago called 3PAR and I led product there from pre-revenue to well over $1 billion post acquisition by Hewlett-Packard at the times. I was also part of Pure Storage for almost 5 years, helping Pure scale from less than $50 million in revenue to well over $1 billion. And then prior to joining NetApp, I was leading product at HPE Storage. Look, across the board, when we speak to customers and over the last 3 years, I've just had the tremendous opportunity to speak to hundreds and hundreds of customers globally. And they have struggled with having to support just a plethora of workloads. AI is now the newest latest greatest addition to that. But when you think about workloads in a typical modern enterprise, you're going to find high-performance files, whether that's AI or, for example, EDA workload or media and entertainment type of workloads. You're going to find virtualization, you're going to find databases, containers. You're going to find much more of the capacity flash more general purpose and test dev type of workloads. You're also going to find secondary workloads, whether it's backup or CyberVault those types of scenarios. And customers have to struggle with how do I ultimately provide the best infrastructure to support the plethora of these workloads. Their data is also spread across on-prem and public cloud. And they want that flexibility to be able to get the right balance of workloads across on-premises and cloud and be able to do that seamlessly. So when we hear about customers and their challenges, one of the critical challenges to that is just ever present is complexity. And as soon as you double-click into that, that complexity just exponentially expands with all of the infrastructure silos. That's part 1. Part 2 associated with that is that everyone in IT has talent shortages as well as talent and skilled gaps. The number 3 is they want to be able to seamlessly leverage the agility of public cloud and be able to have that flexibility of on-prem and cloud. Number four, what Gagan was just talking about in terms of cybersecurity. Ultimately, the last line of defense becomes storage as that last line of defense. And it becomes incumbent on the IT leaders to have the most secure storage. But that is a top C-suite priority across the board. And of course, what you've heard a lot about today, everyone is looking at how do they sees an AI advantage? And how does that become a game changer for them. Common thread across all of this is fundamentally data. The data fuels the overall workloads for our customers and data is the fuel for AI. So when we look at the opportunity and how we can be that strategic partner to customers, it begins with the storage infrastructure, but very quickly it is about data. And this is where fundamentally what we have done over the decades is invested in building a data platform for customers having that right foundation that data platform is so critical. They need to be thoughtful and mindful of building a unified data foundation to be able to get rid of the infrastructure silos. When they have infrastructure silos, complexity abounds. There's the inconsistent management, there's inconsistent automation. There's inconsistent data security models and the weakest link becomes the exposure window. There's inconsistent operational recovery workflows and then overall in consistent experience across the board. So first step really becomes building that unified data foundation, and this is where we have built this unified enterprise-grade data platform. So the customers can collapse silos. They get consistent management, consistent automation. This way, they don't have to worry about the talent gaps and having to reautomate. They get one consistent experience for the data security model. They get consistent operational recovery workflows, and they get consistent on-prem and cloud experience. That becomes a step 1. What we've also done is we have a fully refreshed comprehensive industry-leading end-to-end overall portfolio of our data storage products. It spans the high-performance flash and capacity flash and hybrid flash, so that customers can leverage that no matter what the use case, what the price point, what the performance levels and be able to leverage it for the breadth of these application workloads that are about empowering their internal innovators. We are a top leader in flash across the board. Our portfolio is also fully interoperable. So what that means is they're collapsing silos, they're also seamlessly able to get the lowest cost of data over the life cycle. In the data world, there's such a thing as hot application data and then cold data. And we give customers that complete flexibility with this automated granular tiering built in, where the cold data can be automatically tiered on-prem to on-prem as well as on-prem to cloud as well. And what -- everything that Gagan was just talking about, where we have the data management tied into the application workloads. So this is where -- whether you're running virtualization or database applications and you want to get application consistent backup copies and maintain those library point-in-time copies for recovery that is application consistent, so you can sleep better at night. That is built in. That's through our Snap center customers love that capability across the board. We also invest in full integration into the top workloads so that customers, including the administrators at the workload level are just able to seamlessly consume the underlying infrastructure end-to-end. And then with our announcements today on AI and how we are unlocking the value with the combination of NetApp AFX. It's enterprise-grade disaggregated storage that just delivers extreme performance, massive linear scale, and it is NVIDIA Superpod certified, including with DGX GB300. That enables customers for deploying their AI factories built on NetApp AFX. And then the AI data engine that pairs with AFX that enables customers to be able to deploy a full AI data pipeline that is secure and that is efficient. It comes with the integrated data discovery, data curation, data guardrails as well as the full vectorization and the vector embeddings for Gen AI applications, all built in. It makes it super simple for customers to be able to go and build an end-to-end AI data pipeline. And what we have also done is we full well recognize that enterprises are going to have AI at different levels of maturity within their organizations. Some are in POC stages, some will be in deployment stages. We're simplifying this end-to-end. All of this value of NetApp AFX, combined with NetApp AI data engine, we're also making this available as a service with NetApp Keystone. So whether the customers are in POC stage or production deployment stage, they get that complete flexibility of being able to adopt all of the net enterprise AI value that we're unveiling and they get to do that as a service and then be able to scale seamlessly as their AI initiatives grow. With that, I will open it up.

Ananda, in the back.

Yes, Ananda Baruah, Loop Capital. That was a lot of great detail. As folks begin to deploy AI applications or features, AI features inside of existing applications. So moving proof of concepts into production and actually, if they even have to do this for proof of concepts like let us know. Do you see them doing incremental spend, storage spend along that journey for those AI applications? Or do you see them phasing in the AI applications and making purchasing decisions along refresh cycle lines? And then I have a quick follow-up, too.

Yes. Look, in terms of the AI spend, it's fast evolving. The overall AI technology is fast evolving. Organizations on how they are deploying it is also fast evolving. A lot of the enterprises are forming AI centers of excellence, where they will formalize the best practices. They will also have shared infrastructure as part of that center of excellence. So for some, it's a matter of adding AI for others, it's a matter of building out net new AI initiatives. What our vision is and the -- what we are enabling for our customers is that AI should not be another silo because silos continue to propagate the complexity and customers need not only the performance and scale for AI. They need all of the enterprise-grade capabilities. They need all of that flexibility of hybrid and multi-cloud. And of course, security has to be just built in. So what we've enabled customers to do is basically be able to leverage it and be able to leverage it as just another workload along with everything else there. So they may start as part of building out dedicated infrastructure very quickly, it becomes part of the overall infrastructure.

And just as a quick follow-up, you actually begin to touch on it. So the point about increased complexity, I just want to ask and if the answer is no, please say no because -- but is there anything about AI in the complexity conversation that pushes organizational -- the organizational data management paradigm over any sort of tipping point such that the addition of the AI to the paradigm almost necessitates something like simplification. And just because you're here, I thought to ask the question, but I don't want to leave the witness and that's it.

Look, in the AI world, first of all, you have to recognize that within an organization, you have multiple different personas that are part of that journey. You've got the data engineer, you've got the data scientists, you've got the AI developers. You have the IT teams that are beginning to be part of that conversation. And then clearly, there's AI frameworks and tools and infrastructure that is even outside the enterprise. Obviously, there is a ton in the public cloud there, and you now have newer AI factories that are emerging as well. So ultimately, for customers, what they have to think about is, firstly, how do I actually get the data AI ready. That's really that first step of the journey because data becomes the fuel for AI and for enterprises, unlike the consumer AI for enterprises really AI needs to be informed with the context of their data. So that's kind of the important first step for getting their data AI ready. The next question really becomes in terms of how do I get my data from where it is to where the GPUs live and be able to do that while preserving all of the security permissions without propagating a plethora of copies because as soon as you make copies, you lose the data lineage and you've also lost the context of security there. So we have technologies, for example, our FlexCache technology and/or our SnapMirror technology that enables customers to just seamlessly be able to make their data accessible to AI and do that in the context of preserving all of their security permissions without making copies there. Then when you think about the overall data pipeline that customers ultimately at the data scientist level are having to go and stitch together. That becomes essentially this notion of there are multiple fragmented tool sets and along those tools, there are multiple copies that are being generated. Often, we hear about this challenge that I articulate as data bloat where customers are complaining about my data is multiplying 10x or 20x especially during that overall vectorization process. What that means fundamentally is if that problem isn't solved for them it becomes incredibly costly for them to go and deploy AI at scale. What we have done is simplify this end-to-end AI data pipeline with that AI data engine. And we have built in our own technology to go and build a super efficient overall vector embeddings to help customers avoid the data bloat challenge. We've also partnered closely with NVIDIA in integrating in their overall NIMs technology into this AI data engine. So -- and one last step, we're also investing in the integration with our public cloud partners. And so that the customers' data, not only can we make it seamlessly accessible into the cloud, but we can seamlessly integrate it and stitch it in to all of the AI frameworks and tools that are being invested in the public cloud. So that's how we're looking at this in terms of just simplifying this end-to-end. You also heard George on the keynote stage, talk about ultimately this whole notion of metadata fabric and a knowledge graph because when you fast-forward AI. Ultimately, when you think about enterprise data, today, you've got basically a lot more of the LLM powered use cases, but tomorrow, the evolution is taking us to overall Agentic AI. And this whole notion of an enterprise's data set, curated, classified, protected and then made accessible through a knowledge graph become so critical for customers to then truly unlocking the power of agentic AI.

Lou?

Louis Miscioscia, Daiwa Capital Markets. Well, I haven't heard the words 3 part David's got quite a while. So there we go. But I see though, you seem like you've been with some great storage companies, so you have probably insight that many others might not. So what could NetApp do better given obviously that we always hear about the strength of the uniform operating system. But what could NetApp do better in the sense of why isn't it up, maybe not gain share fast enough in comparison to the other competitors being HPE, Pure or some of the other players out there?

Look, first of all, it starts with building a unified data foundation, and that experience is unmatched by a bar none in the industry. ONTAP is a gift that keeps giving. ONTAP has been matured over decades. And the level of power of unifying application workloads and serving that with a unified data plane coupled with a unified control plane is an unmatched experience across the industry. No one is able to deliver that. When others talk about unifying, they still have infrastructure silos. And that infrastructure silo means you might get some value for a given application workload, but you're still siloed within the boundaries of that given application workload. As you go from either blocks to file to object, you end up segregated and that complete flexibility of a unified data foundation is that step #1. Step #2, Look, nobody in the industry has had the foresight and/or the level of integration that NetApp has done with our hyperscaler partners. This is giving customers this flexibility of I can rightsize, I can shift the workloads with that having to go and re-architect their application workloads. That's amazing for our customers and having that complete flexibility. And thirdly, when it comes down to when you think forward-looking, even present now, top of mind is cybersecurity and security being built in, right? No one is able to match that game-changing technologies that Gagan was describing, it begins with that real-time ransom or detection where we can detect a ransomware attack within seconds to minutes unlike otherwise where it would typically happen in backup or secondary workloads, which is hours to days later. This means basically is very little amount of data is impacted before that attack is detected. We notify it in the scene. But then more importantly, we're able to go and create these temper-proof snapshots for rapid recovery, right? And everything that we were just talking about AI. So when you think about, basically, look, we have a fully refreshed portfolio. We have a comprehensive unified enterprise-grade data platform. And we're evolving AI from another silo into just another application workload and giving customers that complete flexibility of not only just performance and scale, right? You've heard a lot from others in the industry about performance and scale. This is about delivering AI with performance and scale with enterprise-grade capabilities with overall the most secure storage and having that full hybrid multi-cloud.

Just a little bit of a follow-up. And if you want, you could brag a little bit. Which competitors would you think are the easiest ones to compete with and which are the ones that are a little more difficult?

Look, I won't go into specific competitors here, right? Fundamentally, it comes down to the IT leaders once they have this recognition of, I don't need to just refresh, I need to modernize because I need to be cyber resilient, I need to have my data AI-ready, I need to be able to enable the outcomes that are being demanded internal by internal customers across the board. As soon as that realization happens, very quickly, it comes back down to what is that right data foundation. And we're right there for our customers to be able to help them see and showcase how we can be that strategic partners to customers to overall modernize that their end-to-end infrastructure. So we look for how do we solve the customers burning pain points and how we can address them. And so long as we are doing them incredibly well and differentiated manner. That is what we're looking for.

All right. Then we'll get to you, Mehdi.

Steve Fox of Fox Advisors. I guess there's been a lot of talk from the company in the last few quarters about just having enterprises doing a lot of testing on new workloads, et cetera. You've laid out a path for how these workloads could be monetized a lot better. But I'm trying to understand the bottleneck here. Like when is NetApp going to start winning. Like is it one workload at an enterprise customer 5? Is it that they see that they try to pipeline it and they can't, like how is this going to play out so that you guys are ultimately successful?

Look, I would say, overall, we are a top leader in Flash. You've seen our continued growth in overall flash storage. That's just an overall trend as customers are continuing to modernize across the board when we look at the deployments that customers have on NetApp. You will find virtualization, database workloads, high-performance file workloads, secondary workloads. Those are so prevalent across our customer base. And our customer base spans all the way from the top most strategic, the largest of the largest enterprises, enterprises overall as well as a lot of the corporate and commercial accounts, and we span the gamut across the geos. And so we see the overall customers continuing to go and consolidate and unify their application workloads. And that's not only across on-prem, it's across on-prem and cloud.

I think your question is like how do we get into a customer? Like is it a single workload. There's a specific pain point that we address and then expand from there?

Yes. Look, from that perspective, there's multiple ways of addressing and how we land in customer account. But when you think about the journey for getting the data, AI ready and accelerating their AI initiatives, it begins with helping customers build a unified data foundation. The way they take advantage of that, that can be by landing a file workload, that can be with landing a block workload, that can be with landing an object workload, many of those application workloads that I talked about. That begins that journey for them. It evolves into customers then seamlessly extending to cloud or extending on-prem to as a service with our Keystone Storage as a Service offering. And then that evolves into essentially becoming cyber resilient all of the capabilities that are built in as well as the ransomware resilient service that Gagan was talking about. And then the final step turns into getting their data AI ready. It's not necessarily a sequential journey. But these are different paths, ultimately for onboarding and onto the NetApp data platform.

Mehdi Hosseini, Susquehanna. If I just as a follow-up to that and rephrasing the question. You're working on unified data lake, a unified data -- enterprise-grade storage, unified data foundation, unified data lake. There's a lot of stuff that you're doing and you're working with customers and the fact that FY '25 was a strong year for all-flash array gives you a tough compare. So perhaps all of these unification and new approaches and problem solving would manifest itself to some traction in FY '27 without asking you a specific financial question. I just -- we've been hearing of all these problem solving. We just -- we're kind of -- not desperately. We're trying to figure out how this puzzle is coming together. And it seems to me that it was more of a '27, so we're in the sixth inning.

Look, I won't comment on the financial side of it. Sam is here. Kris is here. They are much better equipped than I. What I can say is that we are laser focused on making sure: one, we fundamentally understand what are the burning pain points for our customers across the board; two, that we have and continue to enhance, but we have the best differentiated unified enterprise-grade data platform as that foundation for our customers; and thirdly, then just giving them all of the necessary capabilities, whether it's as a service offering on-prem or in the cloud or it is the necessary software capabilities or workload type capabilities to not only just accelerate and continue to simplify their existing workloads, but ultimately go and deploy AI.

I don't think competitors are doing any -- it's not like competitors are ahead of you. We're all in it together. But I think if I just like trying to think about what has happened over the past 12 months. And all the efforts you put in, we're in the sixth or maybe seventh inning in that journey, all the good things that you have done is now like -- it's not like we're still in the dark room trying to figure out where the door to the AI stardom is. We're getting close. Would you agree with that assessment?

I would say, look, I don't know about the innings, but I would say, look, the AI journey, especially enterprise AI. That journey is just getting started, right? We see forward-looking just tremendous opportunity of working with customers on their enterprise AI journey, and we're super excited about the innovations that we're bringing to market on that front.

All right. We have one final question from Samik.