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

Alright. Hi, everyone, and welcome to INSIGHT 2024. We appreciate your time, especially for everyone who've made it in person. Before we kick off, I'm going to read a safe harbor. Each of the 2024 INSIGHT financial analyst tech sessions may contain forward-looking statements and projections about our strategies, products, 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. 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 find -- 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 Form 10-Q available at www.sec.gov. The forward-looking statements made in these presentations are being made as of the time and date of the live presentations. If the presentations are reviewed, there are no presentations. If the Q&A sessions are reviewed after the time and date of the live presentation even if subsequently made available by us on our website or otherwise, these Q&A sessions 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. So with that said, thank you. Let me welcome you again. As a reminder, today's sessions are all technically focused. We won't be covering any financial information. So this is your opportunity to ask our tech leaders questions about our products, our services, our value proposition, the competitive environment. I know those are always top-of-mind questions for all of you, so you get to hear it straight from the horse's mouth. After this session, will -- for those of you who are here, you have some free time. The keynotes start at 4:30. You're welcome to go to the keynote. There's a small section of reserved seats that you're welcome to sit at, but you're also equally welcome to go find a customer and sit next to them and get their live reaction. After the main session, the showroom floor is open and you can wander around. If you need anything, find one of the IR team, and we're happy to help you. With that, it's my pleasure to invite George Kurian to the stage. George?

Thanks, Kris. Good morning. Welcome to NetApp INSIGHT. Thank you for taking the time to be here. We are super excited. We have a lot of innovation payload that we will be sharing with our clients that our teams have been working hard on over the past year. We will talk about with our clients that we are at the start of the era of data and intelligence. What we've seen over the last several years that we've shared with our clients is that data-driven organizations are outpacing those that don't have their data well organized over many years. But now the tools that are being made available are even more powerful than they ever have been before on two dimensions. The first is you are able to analyze a set of data that is, frankly, for any organization the preponderant majority. 85% to 90% of an organization's data is unstructured data, meaning files and videos and documents of various sorts. It's conversations with customers and whiteboard sessions and so on. And that's about 85% to 90%. And we are the unquestioned leader in that part of customers' data landscape. The second is not only are the tools more powerful to be able to normalize and understand that data, but they have almost human-like intelligence to be able to understand the domain in which they operate without human involvement and be able to switch domains, so go from one domain to another, go from one modality of operation to another, right? So going from text to video or video to image and so on. And so we talk about the fact that we are at the junction of data and intelligence. And what's required for success is that you need to have your data and data strategy well organized. We said that to our clients last year, and we are reinforcing that to clients this year. I was just out on the road over the last few weeks with clients. And I'll draw 2 banks. One whose data is well organized, they're already using GenAI tools, they've got a hybrid architecture, and they're making good progress with those tools; another bank that had a classic siloed custom-built architecture with lots of different landscapes. They are a year away from buying an AI computer and then starting to put their data on it, right? So profound differences. You will see, for example, that in industries like life sciences, where there has been regulation and requirements to have high-quality data catalogued the right way with clinical data codes and with procedure codes, that they are making much more rapid progress than those that have not had their data well organized. And surprisingly, some of the more regulated industries are making more progress than the unregulated ones. So data and data strategy is super important. The second is when everybody has really powerful tools and all of the world's data supporting those tools, your domain knowledge and insight becomes even more important. And then the third is the ability to take your domain knowledge, your data tools, and then apply it in an iterative test, learn and adapt loop so that you can graduate some of these projects from proof-of-concept to production and then close the loop back to make your data science environment even more strong. And then finally, a data ecosystem that enriches your data just like your business ecosystem supports your business. Those are the 4 key things that we'll talk to clients about. And then we'll talk about the two challenges that organizations face in using their data effectively to support advanced use cases like AI. And there are 2 challenges there. One is the familiar challenge which we are skilled at helping clients with. It's a data management challenge. And so what does that mean? How do I find the data across my landscape that I might want to use for my AI project? How do I govern sensitive data so that I can bring forward my security and access control model into my AI landscape, I can ensure privacy and so on? And then the third is how do I actually keep my model environment and my data environment in sync as the data gets fresher and fresher or data progresses through the life cycle as it always does? So there's a data management challenge. The second challenge, which we have observed from interviews with around 800 clients over the last year and I have personally participated in probably 50 discussions with clients around the exact same problem, is that AI is being built as a silo. It's got custom networking, custom chips, and it's not integrated into your data landscape. So there are so many clients that we've met that said, "Hey, I've got my AI computer stood up, but I can't get data to it" or we had a semiconductor vendor that was copying 300 terabytes of data a week to try to keep their supercomputer moving. And I asked the gentleman who led that project, I said, "What's your life like?" And he said, "One word, hell." And he said, "It will not scale the way it is". Now if you think about that gap, the chasm between your AI environment and your data and operational environment, for us, that looks just like how cloud was many, many years ago. We stood up here in 2013 and said if cloud was to become useful and seamless, you needed to have a bridge between your on-premises environment and your cloud environment. And we call that bridge the Data Fabric. And we innovated over many years to make cloud and enterprise data work seamlessly together. And today, you will have many of those cloud partners talking about the AI journey that they are taking on with us. And so for us, this is a familiar problem, and we are innovating to bring capabilities to the world of AI that don't exist today, and those capabilities exist along three fronts. First is a set of tools and applications that make it easy to find the data, all of your data estate, so that you can quickly explore it and then choose what data you want for your AI landscape. The second is to have -- bring AI to your data, which is much easier than trying to bring your data to AI. You see data is the gravity part of that equation, and so we know how to bring AI to your data. And you will see 2 or 3 transformative capabilities that we bring there. The first is the best infrastructure for AI, and we will talk about the third architecture, the third-generation distributed system architecture. The first generation were shared-nothing architectures like Pure FlashBlade, or Isilon, or Qumulo. The second was distributed architectures with centralized transaction management, like what VAST Data has. And the third is actually a truly distributed architecture where the transactions and file operations are distributed across the system. And so that's the second area. And we combine that with capabilities that allow you to have model versioning and data versioning synchronized to have traceability of your data together with the models and to have highly efficient patented technologies for data retrieval. Retrieval is the first step of what you call RAG in your inferencing world. And then the last is a set of capabilities that we have built over many years that we've enhanced for the world of AI, where you can bring all of your security policies and privacy and controls across your AI life cycle and be able to detect changes in the data so that you can efficiently apply those changes to your models. And so super excited. We think that we have step function improvements in our capabilities for unstructured data reinforcing our leadership there. We also have really good announcements in Block Storage, Cloud Storage that Pravjit will talk about, the security capabilities in our portfolio, and all of that falls under the umbrella of intelligent data infrastructure. So I'm excited for you to hear from our technologists today, but equally from our clients and the partners that we co-innovate with. Thank you for coming. Look forward to a great conference.

Alright. Thank you, George. I appreciate it. So now we're going to start the first of our Q&A sessions. I'm happy to introduce a new face to you all, Pravjit Tiwana -- did I get that right? -- alright, who heads up our Cloud Storage Group. So one of the questions I get from you all a lot is what is the value proposition of your Cloud Storage services? Why do people choose to use NetApp in the cloud? So you can ask directly. Why don't you come on up? Thank you. And before we have you start asking him questions, Pravjit, why don't you introduce yourself, say a little bit about what you do, and then we'll open it up to the audience, and I'm going to switch places with you.

Sounds good.

Alright.

I assume this is live now, right?

Yes.

Alright. Hey, everyone. My name is Pravjit Tiwana. I lead the Cloud Storage Group. I'm the SVP and GM for the Cloud Storage, encompassing all 3 hyperscaler services from us in AWS, Microsoft and Google. I joined NetApp earlier this year in March, although I have been in cloud world for pretty much all my career. What really excited me was -- to be here is the world is seeing, right, really very interesting times with the growth of data which we are seeing. Zettabytes of data is being produced every week. And if you intersect with what is happening in cloud, right, like the migration in cloud, the momentum in cloud and overlay it with the advent of AI, it creates a perfect storm right now in terms of, right, like the intelligence, the capabilities which you can build on top of your data. So from that perspective, right, like -- since there is so much of data growth, there is so much of every organization is thinking of how to use AI to further accelerate their businesses. And when you mix cloud in this thing, the complexity and scale at which organizations have to deal with, it is becoming way more and more complex. And that's where, if we see, pretty much every business or organization have some cloud strategy. Either they are all in, or they are at least exploring it, or they have 1 or 2 mission-critical workloads running in the cloud. And that's where we play the role because we have our first-party cloud storage services available in all 3 hyperscalers today. And we have a deep -- thanks to deep partnership with all 3 of them, we continue to innovate on behalf of the feedback which we get from our customers. So I can talk about it or we can do a Q&A, whatever. Yes?

A question we often get from investors we talk to is sort of how do we think about your competitors playing catch-up to the first-party services that you already offer on the public cloud? They obviously go through the marketplace. But in terms of capabilities, how do you think about where the roadmap is to sort of stay ahead of the competition on that front? And do you at all see a risk of down the line sort of some of your competitors moving to becoming sort of the first-party services on those public clouds? How do you sort of think about that risk?

No. Yes, sure, right? Like if you see the landscape today, we are the only first-party cloud storage available in the cloud with all 3 hyperscalers, right? Yes, the marketplace offerings might be there, but marketplace and first-party is kind of like an apple-to-oranges comparison because the deep integration which we have with the entire stack there, the whole sales motion, to billing to especially support and operations, which you get in first-party is not the same thing which you get in marketplace kind of offerings. But that said, right, like having multiple players or having competition is [ usually ] not a bad idea for customers, right? Like we don't have any mechanism to say who will do it, and we don't need to speculate that either, right, like who else will be in the first-party cloud storage services down the road. But where we are focused on is that we are being relentless in terms of the innovation we can do on behalf of our customers. The number of capabilities which we are building working with our hyperscalers is phenomenal, and you will hear a lot of those capabilities either we have announced in the earlier part of the year and some of those we are announcing later this week. So you'll see a broad area of -- the one thing which I have seen running cloud services at a very large scale for a very long period of time, there is no compression algorithm when it comes to things like scale and building the capabilities over and over again, like iterating on those capabilities over the years. You cannot, like if somebody -- storage, compute, network, these are kind of [ primitives ], right, like where it takes time to understand how to build scale, how to build operational strength. If you see in my organization, we spend so much of time on things like security, reliability, availability, performance. Those are the things which are not just easily replicable from anyone because it takes years to build those kind of capabilities, and we're investing those -- in those a lot. But when it comes to marketplace offerings, we don't think that's an apple-to-apple comparison today. But we cannot also say that, hey, what's the future for our competitors down the road. But in short, customers having wide choices is not a bad idea. It raises the bar for everyone.

So many of the customers who end up buying the cloud storage are kind of new customers to NetApp. And so just how is the value proposition to those customers kind of different than those who would have traditionally been NetApp customers? And what's kind of the most effective way to get those customers on board?

Yes, you're right, 3 out of 5 customers which we get today in our cloud storage, they're net new to NetApp, and the remaining 2 out of those 5 are using us in hybrid mode. So from that perspective, yes, there is a -- the good thing about being available in the cloud is the -- is being available in front. And especially when you are integrated in the hyperscalers, it's being available to all the customers. And they can themselves play and figure out, right, like what's the value which we provide. We see lots and lots of that developer-based ecosystem also happening in now in our first-party cloud services where customers try, start with a small workload, learn that how much goodness we bring in terms of performance or about multi-protocol support or the data protection capabilities which we have or the security investments which we have done. Like this year, working with Microsoft, for almost 6, 8 months we have been working on the SFI initiative from Microsoft, so the customers are getting things like not just our security goodness, but they're also getting all the [ bars ] and the security controls which our hyperscaler partners are defining. So combination of those excite these new customers to start using. And we have been fairly focused on price performance optimization in our hyperscaler clouds, and that also resonates to the customers. And they -- but the final thing is, right, like the AI integration which we are able to provide. If you see, most of the science behind AI is being driven by also hyperscalers. Like all 3 hyperscalers are very, very heavily invested into building the AI stack, along with obviously the NVIDIAs and the Metas of the world. But being in our -- using our services in hyperscaler setups makes it almost seamless to use the AI stacks which hyperscalers have built. And without create -- need -- without the need to like replicate data or clone data or copy data and create silos, they can do it wherever the data resides and can use that AI. The combination of our security, our AI, our price performance, our operational strength and the capabilities which we have built in ONTAP for almost 2 decades, all those combinations resonate with the new customers as well as customers who are using us in a hybrid mode.

It's Tim Long at Barclays. You mentioned AI, maybe a 2-parter. Talk a little bit about kind of how your business is affected by a lot of these large language models that are going on now. I'm sure it's not major at this point. And then maybe walk us into when we get into inferencing, and there's a lot more on-prem and bursting to and from the cloud. So how do you see that dynamic impacting the cloud storage business for NetApp?

Thank you. This unstructured data and file like most of the AI is going to run on data, and data runs on NetApp, right? Like so it means it's a perfect combination for us. So we are really excited that hey, with AI, growth in AI and the improvements in AI, which we are seeing at least in the last 18 months, it has accelerated a lot. It is really good for customers to be able to -- our mission there is that we want to provide the AI capabilities, irrespective of whatever LLM models you want to choose, right there on your data, right? Like you don't have to copy, clone and those kind of things. That's our -- we don't want to create silos. So if you look into earlier this year, we launched Workload Factory for GenAI. That is basically a capability in which you can integrate LLM models from Amazon Bedrock into the data which you are storing in FSx for ONTAP, just with a few clicks, right? Like you don't have copy it to, say, S3 Object Store or any other storage system. You can do that. So we see that is -- and once you have integrated some capabilities like your Bedrock foundation models, then you can also start overlayering it with other capabilities which you might choose from marketplace or you choose from hyperscalers or you choose from us also, right? Like so that provides a very rich ecosystem for customers to use that AI. And there was a second part of your question, I forgot that part.

Inferencing. Hybrid cloud.

Oh, yes. Yes. So let's talk about the hybrid cloud, right? Like because we see very significant growth in our hybrid cloud setup year-over-year. So taking that same case of what we have done with Amazon Bedrock to use LLM models against the data which you have stored in FSx for ONTAP, you can also combine it with their on-prem storage. So things like with SnapMirror, you don't have to copy the whole data, but you -- whatever is the data set which you use to want to use for inferencing or your chatbot or whatever, you can combine that data into FSx for ONTAP. And you will see these kind of patterns also in other hyperscaler partners. So you will -- from that perspective, we want to provide, and working with our hyperscaler partners, we want to provide a mechanism in which you don't have to [ say ] here is AI for cloud, here is AI for your on-prem. We want to provide a seamless integration, and that's the path which we are on. On inferencing, we do support some of those capabilities today for your data and with our hyperscaler partners. We earlier this year launched GenAI Toolkit for Azure NetApp Files as well as Google Cloud NetApp Volumes, which provides very -- you can use it against your proprietary data. With a few clicks, you are able to build your RAG interfaces on top of that, and you can build applications like chatbot, knowledge base, all those kind of capabilities with a few clicks only. So inferencing part and the RAGs and GenAI toolkit, the Workload Factory part of ours also support building your whole RAG infrastructure with just a few clicks. I highly encourage you to look into -- we are doing these demos this week in different sessions and also on the floor. So please do -- if you have time, please do look into those, because these are really, really exciting and step-forward capabilities which are coming.

Steve Fox with Fox Advisors. Maybe just -- you mentioned a couple of things on the roadmap, but can you just sort of step back and talk big picture of how you envision the roadmap? And without, I guess, giving away the announcements you have this week, what are the big things we should think about that you're focused on, say, over the next 12 to 18 months for improving on?

In AI or in general?

Cloud storage.

Overall in cloud storage, right? I think it's a -- majority of our roadmap is controlled by our customers, right? Like whatever we are building is based on the feedback from customers. So if I have to look into what dimensions we are working on, right, like without going into, right, like what lands when and all, but the dimensions I can talk about, right? So that from a dimensions perspective, right, like we know that customers have trusted us to learn a lot of their mission-critical workloads, right? We read about their ERP systems or their databases like SQL Server or Oracle or high-performance computing or streaming content. The list goes on in terms of the kind of workloads they run with us. Our roadmap from that perspective is highly focused on making sure that if customers have to run their mission-critical workloads, we are the best destination for us -- them. So that's one dimension of what we are working on in terms of our focus. The second dimension for us is what we discussed about AI. Our mission is that, hey, we want to bring the best of AI stack to the data which you have trusted with us, be it on cloud or on-prem. And we provide those rich capabilities so that you can build inferences, RAG interfaces, whatever you want to build, with almost like in a frictionless manner. That's our second dimension. The third is the cost optimizations. We have -- like if you see in the recent past in this year, we have shipped a lot of capabilities. Our customers have told us that they love capabilities like auto tiering, which is now available in all 3 clouds for some good period of time, which basically moves the data from hot to cold tier or a different tier based on the data access patterns, without the customer have to do something about it, or the things which we -- or all the ONTAP richness like compression, compaction, deduping, thin provisioning, all those. So we'll continue to keep our focus on making price-performance optimizations. Our goal is that cost should never be the reason why you don't select our first-party cloud services. And the fourth dimension, which we will continue to focus on is on our hybrid. We understand that different customers are at a different stage of their cloud migration. But irrespective of whatever stage they are in, we want to provide the best capabilities so that not just migration but deployment and operations of those workloads and the data which they bring to us is highly, highly optimized. And that's why, right, like we'll continue to invest in things like disaster recovery, data mobility, and couple it with the elasticity which you get in cloud in terms of instantaneous cloud capacity or bursting or those kind of capabilities. So it's 4 things, right: like workloads, AI, price-performance optimization and hybrid.

Pravjit, not sure if this is sort of the focus, but cloud ops, the cloud ops part of the portfolio. Can you talk about sort of how do you think about the value add relative to the Cloud Storage part of the portfolio that you're sort of more focusing on? And then when you think about inferencing as well in terms of AI workloads, do you see the value proposition of cloud ops changing on that front eventually?

I'm pretty sure Haiyan is also talking today on cloud ops. I can give a little bit high level, but...

Yes. So we don't have Haiyan today, but if you could be very high level, and I can always follow up with you, [ Samik ], on any cloud ops you want.

We have a different leader who runs that cloud ops portfolio. I'm sure you know her. We have like a bunch of places where we have synergies in terms of, right, like baking it into the hyperscaler cloud, especially around things like cost optimization goodness, which comes from cloud ops portfolio, or the observability stack, which we can integrate with those. So we are looking into -- I would not say we are just looking into it, but many of our customers are looking at using it -- trying -- using us in that fashion, especially around the observability stack and the cost optimizations aspects and so on. And we will continue to work with cloud ops in that sense to build a comprehensive portfolio. That's where all the aspects of our unified storage, our data protection, our cloud ops portfolio, they all work together to provide a frictionless way to manage your workloads. So I think that is the part -- and I'm not super deep into the cloud ops portfolio, but this is what I know what we are doing.

Okay. We have a question online.

Yes. This is from Aaron Rakers at Wells Fargo. How has NetApp's share gains evolved for block storage? I'm not sure this is a cloud-specific question.

No. Sorry, what's the question?

So Aaron, I'm going to have you hold that until Sandeep comes on. But I do think you can talk about block storage in the cloud, because I think that's the thing that people don't expect to hear from us.

Yes. Actually, block storage is the part in our cloud portfolio or in our first-party services which is really -- which personally caught me by surprise when I joined. In last 1 year, we have seen 140% growth in our block storage and first-party cloud services. So it is pretty significant growth for us, and we are really, really proud and excited about. And the good thing is, right, like most of these customers, they go to, say, AWS console or some other console, and they self-provision their block capacity because they are familiar and they know that how much ONTAP richness we have built over the years into our block storage portfolio, and they can use that in the cloud also. So yes, Block is an integral part of our strategy as well as consumption from our customers. That was the question about block, right?

Yes.

Wamsi at Bank of America. I want to ask you a little bit about when you look at cloud sharing implementation with different cloud providers, it's kind of different the way the technology is implemented. So as customers are deploying, say, AI workloads, do you think that one architecture is more favored versus other architectures? And what are maybe some of the benefits of one versus the other, if you see any?

Yes. I think the way customers use us is from the interfaces, right, be it from the console or from the APIs or the SDKs, right? In general, right, like the nitty-gritties of the architecture are kind of like opaque from the consumption perspective from the customers, right? And then from that perspective, our goal is to bring uniformity. So we have uniformity in the same terms of, right, like the protocols which you use to access data, irrespective of what cloud you are, they are common, right, so. And then we have a BlueXP, our manageability which you can use to manage multi-cloud setup. So it gives you a single interface in all. So matter point being that, hey, we are trying -- and based on our customer feedback, we are trying to make it uniform so that you don't have to figure -- customer doesn't have to worry about, hey, if it is a one implementation behind the scenes or second, they get the same performance, same capabilities as they will get in one hyperscaler versus the other. I think the second part of your question, right, hey, like is it good to do [indiscernible], I don't think, like in data science, like there is a perfect answer for these things, right? Like I think both have their mechanism. I think GCNV and the ANF side, we pretty much do everything from the service delivery engine capability which we have built. So there's a lot of uniformity. [ FSxentials ] are a unique way to do it. And that is also very, very -- the kind of scale growth which we see with that is humongous. But customers are not bothered about that. We have a lot of customers who use us in multi-cloud setup. They're using FSxN or with ANF or something like that. And for that, we make it seamless for them.

And if I could just follow up. For the customers, I think, to Meta's question, there was a -- you've got 3 out of 5 customers are new to NetApp on the cloud. So for the ones that are not, is the AI deployment currently, whatever they're doing with that, is that more on-prem-centric versus cloud-centric? And are you seeing any signs of early migration of AI workloads on NetApp to the cloud?

We are seeing signs, but you also have to look into the landscape of the businesses, right? Like a lot of businesses today are in the learning mode, fine-tuning the models and those kind of things, right? So that's why I was talking about our GenAI Toolkit and those kind of things. So customers are showing a lot of interest, and I'll say a little more than like just POCs at this point in terms of that. But in fullness of time, I truly believe that, yes, it will be one of the core players in terms of, right, like especially using the AI seamlessly between cloud and on-prem.

Okay. Lou Miscioscia, Daiwa Capital Markets. Similar to Wamsi's question, I'm just wondering because this was all deployed at various different stages and some of the cloud you were -- customers were yours first. Where are you with all of them in the sense of, you just talked about trying to be uniform, but are some more advanced and you're trying to take those more advanced features and share them with the other ones? So I'm just sort of wondering if there's some that you're just doing a lot better with. And if so, how would you take that to the other ones?

You mean like from capabilities, say, in Azure versus capabilities in AWS versus Google? That -- in that sense?

Yes. And total revenue and growth and all the things that we care about.

So just a reminder, we're not doing any financial updates. So Pravjit will not talk about the financial ramifications, but he can talk about the high-level capabilities and partnerships across the cloud.

Yes. I think you have to look into a broad spectrum of things here, right? Like if you see on the AI side, like it's a relatively a new space, right? So the standardizations, the uniformities will take a little bit of time to land between not just hyperscalers, but in the industry in general. But when you come to the other things, right, like how you access your data be it by different protocols, right, we provide capabilities like, hey, you use the same protocol to access, we have Windows or we have Linux, without needing to duplicate the data or the unified storage capabilities. From that perspective, there is a lot of uniformity which we are bringing in terms of access as well as from our BlueXP framework or the capabilities, right? But in fullness of time, right, like we do expect our customers to start using more integrated services from these hyperscaler services. I think that's a natural progression. We don't want to control or be in the middle of it. Where they find value, they should absolutely use that. And yes, there will be a little bit, right, like maybe the object store in Amazon is different than Microsoft, or the security -- if they want to use something other capability from them, that might be different. But yes, there will be a little bit of that, but that is the part which we're trying to make it seamless through our BlueXP arrangement for multi-cloud setups.

Understanding from the customer standpoint they're most interested in kind of being able to not have to replicate all of this copying from various sources, I guess from a cloud customer standpoint, is there anything that they're asking you to kind of work on developing, other than just accelerating kind of the back and forth that they think can kind of help optimize their services?

I assume your question is in reference to AI, right?

Yes.

Yes. Okay. Yes, we are deeply integrated with our customers. That's -- there are a lot of capabilities, which we are building in terms of, right, like how to manage your RAG-based infrastructure. That's the part which I was talking about with our BlueXP Workload Factory. Now you can -- with a few clicks can enable that. So what customers are asking us, to simplify the complexity of AI, right? Like there is a lot of complexity involved in that today in how you set up your infrastructure, how do you bring your foundation models? How do you build inference on top of it? So what they are telling us is, right, like, hey, we are trusting to store our data with you, we are running our workloads on top of you, so make the whole AI ecosystem also simplify. So our goal is to make it frictionless and simple. That's what we are doing, and that's what our customers have told us to do.

I guess I mean the cloud. Like, I guess, I mean, what are Azure, Google, Amazon -- like are they asking you for anything different than your customers are asking for?

Yes. I wouldn't say it as an ask, but yes, we are deeply partnering on how to simplify it for our customers. So we do a lot of engineering and technical conversations with our hyperscaler partners to make it frictionless. They do -- like our hyperscaler partners fully understand that there is a lot of world's data which run on NetApp, right? And they want to utilize that with AI. So from the simplicity perspective, yes, we have a lot of integration discussions, a lot of tech discussions with our hyperscaler partners. And that defines the roadmap of the capabilities which we build in cloud.

Sorry, just a clarification. So when you do all this work to simplify the sort of, let's say, RAG for your customers, and the cloud customers want you to do that, do -- like does this come in as a feature that you offer to enterprises? Like how do you get monetized for it? Are you just more dependent on eventually more data finding its way to the public cloud? Or do you then have a feature that you can actually monetize at a premium with the customer for putting that work?

I think monetization part, we can discuss a little differently, but this is a core capability which customers need it today and especially in the future. Monetization part is I'm sure like we will -- if we bring goodness to what customers want, monetization part is separate. Right now, our main focus is to make sure that we make AI seamless to work with. Yes, there are some monetization aspects of that. We haven't fully built that model. I don't think anyone has built the full model yet on like, right, like the AI monetization part. But in fullness of time, we do expect it to grow our business. And I think Hoseb and [ Russell ] are going to talk more about AI. Is it in the next one? Yes.

Yes. And we have time for one final question on cloud storage.

Pravjit, can you talk about what is your understanding of how the hyperscalers -- and obviously, each one is different -- but their key considerations for build versus partners has changed over time. And -- because they do -- hyperscalers do source, right, a lot of the raw components. They could do a lot of it themselves, but you talked about complexity, about scale, and you hit on a lot of these topics. But what -- I guess if you really wanted to narrow it down, like are we -- the crunch of the question really is like are we at a tipping point in terms of like there's only so much that they can do themselves, that they'll lean more to third-party services and native first-party services like a NetApp, and therefore the floodgates will have opened -- or are about to open in terms of giving...

No, I get the sense of your question. I don't think I want to say on behalf of hyperscalers if they are on a tipping point or not. But what I can say is that, right, like one thing common across all 3 hyperscalers is they are very good at listening to the customers, right? Their customers are telling us that, hey, we want -- say, take the case of NetApp storage, right? Like they are telling them that they want those capabilities in the cloud. And they are responding to it in a like manner. Now there are 2 parts, right? Like we have built a lot of innovation, a lot of capabilities over the last 2 decades, as I was talking about earlier, right, like there is no compression algorithm to basically say that, hey, we can build the same capabilities in the next 10 months or something. That's part A. And the part B is, right, like we ourselves are, on behalf of our customers, are not also stopped doing innovation. So we are also accelerating that. So combination of what we have built the richness over the years and our laser focus on the dimensions which we've talked about. This is -- the combination of this will continue to keep all of us relevant, and it's the right thing for customers perspective also to get these capabilities. So our hyperscaler partners understand that. It's not like "we versus them" kind of a situation.

Alright. Well, thank you, Pravjit. I appreciate you coming up and taking all the questions and handling all the AI questions. I really appreciate it. So thank you for your time today.

Thank you. Thanks for having me here...

[indiscernible] to your next meeting.

Alright. Thank you so much.

Alright, thank you. And now because AI is such a hot topic, we have 2 presenters who are going to come and handle all your questions. So happy to invite Russell and Hoseb up. Some of you have heard from them before. They've done a lot of work for us. So before we kick it off with what I'm sure is an endless stream of questions from the audience, why don't you each introduce yourself and talk a little bit about what you do related to AI for NetApp?

Please go ahead.

Sure. Russell Fishman. So I lead the solutions product management for AI at NetApp. That is the -- one of the main growth -- is it on? I just turned it off. I think it was on. [Technical Difficulty]

Now you're on.

So yes, so Russell Fishman. So I lead the product management for solutions at NetApp, and AI solutions in particular. My responsibility, partnered very closely with Hoseb and our product and engineering teams, is on making AI real for our customers by taking products from our portfolio and combining them with third parties primarily to create complete use cases that help our customers adopt and accelerate their use of AI.

I'm Hoseb Dermanilian. Good morning, everyone. I run the AI sales and go-to-market for NetApp. I've been with NetApp for 10 years and been doing this for 6 years. So I -- I'm very fortunate to be on this journey for almost 6 years now with NetApp. So I'm happy to answer any questions you have about what customers are using NetApp for or how do we see the market growing from a customer standpoint.

Very big picture question. So George talked about the journey from cloud to the Data Fabric started in 2013. So now you're starting on a similar journey, any mistakes that you would warn us about that you're going to have to overcome as AI evolves? Because I think a lot of Wall Street is concerned about the hiccups, not the end point.

So you want me to cover the journey or...

Like what you're looking for and what are the potential problems or challenges?

Oh, potential problems and challenges. So I think what we are seeing is customers today are trying to use these large language models, but the first trial is not as good as they want to be, that they can deploy it in their own enterprises, right? And we're talking -- if I'm a customer, if I'm a large retail customer and I want to provide a robot that will answer support questions and that cannot be trained on whole data set that was available to the entire globe, and it might answer questions differently. Some of them, they are new, even going to these LLMs and they can't identify the name of their CEO. So I think that is the biggest challenge is them understanding that they need to bring that data to these models to make these models more specific. I think the overall expectation today that they can just go and use these large language models and it will apply to every problem that they have, what -- the biggest challenge we see is that expectation becoming a frustration, which then you -- people either stop using the technology or they adapt and, okay, if we want to use this, then we've got to do this, 1, 2, 3, 4, bring the data, fine-tune the model, RAG the model, et cetera, et cetera. So I think that's, from my perspective, is one of the challenges are. You could tell the other challenges, cooling, power, data center. But I think for the wider enterprise, the hyperscalers will be a big player in this. So even talking to customers, those who doesn't have the power and the cooling and data center capabilities, they will go to the hyperscalers, but then they will face this LLMs not being trained specific for their own models, right? So it's a mix of both. So either they fix the cooling and power for their data centers and train these models ground up in their data centers, or that data that exists today on-premises, bringing to these large language models and open source models at the hyperscalers, but then it become the data challenge. Do we move the data, the data is sovereign and all that. So that's what we see.

I might add to what Hoseb said. So if I look back at that transition, as you said, to the Data Fabric, one of the big challenges we had was we had the technology, but actually getting customers to adopt it was complex, right? It was difficult. It required knowledge and sophistication on behalf of those customers. It also required knowledge and sophistication on behalf of the partners that make it real for most of the customers. So I think one of the things that we're really focused on on this AI journey is to accelerate adoption. And so that really means having a fantastic set of partners that we work with. So obviously, Hoseb mentioned the hyperscalers, but our go-to-market partners, we already have our Partner Sphere Program that focuses on getting AI adopted by our partners with our technology portfolio, and also the fact that we are focused on integrations with third parties. So we already have this amazingly rich set of ecosystem partners, because we don't believe we can do it ourselves. We don't believe that we're the only people that people need to work with to deliver AI. We think we're an accelerator. But what we're doing is by focusing on that complete picture, it's helping people adopt. And that's really this kind of inflection point of democratization, where we've -- Hoseb's been brilliant at helping customers who want to get ahead, who are willing to go and build things themselves maybe have the data sets have that sophistication. What we're starting to see, of course, is many more customers who want to adopt without having to go through the development, almost like just buying it off the shelf. So that's the inflection point that we're starting to see, and all those integrations that we already do are the way that we're already helping that.

Maybe you could talk a little bit about the most common use cases, applications that you guys are seeing NetApp getting involved in, and how broad is that set. If you highlight 2 or 3, how broad is the set? And how do you see the overall set growing?

Yes, absolutely, Tim. I'll cover this. So I'll do it over the past 6 years, the journey, right? So we started actually 6 years ago when we certified our storage with NVIDIA DGX-1s. That's what we saw customers. At that time, it was deep learning and neural networks. So there was no GenAI, much of language models, et cetera. It was mostly customers who were developing their own in-house models to -- for example, we have one health care customer. This is 6 years ago, right, I'm not talking -- 6 years ago, we deployed NVIDIA DGXs with NetApp ONTAP to train models to detect anomalies in the x-ray images and et cetera, et cetera. So I think image processing and computer vision was a very big use case. And it's still a very big use case before and after the large language models. And then over the time, we started seeing customers build AI center of excellences. These are large customers. We have some of them here at the conference, they're going to speak later on, where it's not 5 users. It's basically they have multiple data scientists that's scattered across the nation as well as across the globe, where one of them will do a model training to detect fires in the forests. The other ones will detect something in the war zone, et cetera -- because I'll give you an example of a system integrator who does multiple different things. And they wanted to build AI center of excellences. Again, this is based on the GPUs they purchased in-house with the data that needed to feed those GPUs. Obviously, when large language models and the whole ChatGPT and GenAI boom, now we are starting seeing those customers who, as I mentioned, want to leverage the goodness of the cloud. But then their data has gravity, and their data is on-premises. Those use cases is mostly we're seeing AI agents. Customers want to build AI agents to either respond to support, to either respond to customer service. Now we're seeing people writing software through AI. So that's another thing that we're seeing as a use case. And this is now where the cloud and on-premise becomes together, and NetApp is actually the connecting the tissue from the data perspective. So that's the evolution of the life cycle we've seen. Now aside of this, we also have seen handful of customers you probably are witnessing, is those who are building their own large language models. It's not probably as big as the hyperscalers, but it is big enough to call it a supercluster. NVIDIA likes the term SuperPOD. They call it the SuperPODs. They are -- I wouldn't say it's the majority of the enterprise, because again, back to the cooling, energy, power, all that requirements. But we have seen customers actually build that large superclusters to specifically train foundational models. And this is because they either want to provide this as a service to some other folks, they want to compete with the hyperscalers or they just don't want to leverage someone else's model. So we have seen that type of customers as well. I hope that answered the question.

Well, I might add a couple of pieces. So yes, so you talked about personal productivity chatbots and copilots, enterprise knowledge management, obviously, pretty broad use cases that have, from a vertical perspective, broad applicability. But what we're starting to see absolutely is more interest in highly verticalized solutions. So we talked about LLMs and talking about SLMs and XLMs and those that have been specifically augmented with the knowledge that's necessary in each particular industry. So there's going to be an explosion of that. Absolutely, we're seeing that in the market. And that's really probably closing the gap between folks trying out these generative AI solutions and actually making them very useful in a more specific context beyond just general productivity. So what you see, for example, with Microsoft and their copilots, for example, but they're very broad. And then moving into industries like legal, finance, health care, in particular life sciences, we see these very specific versions of LLMs or SLMs coming into play. So I mean, that's going to be definitely where things are moving.

Can I just ask, when you talk about go-to-market on this front, how is that different from the traditional infrastructure go-to-market in the sense that if a customer today wants to sign up and sort of get more educated from you in terms of what that technology should look like, do they want a full understanding of the complete roadmap, including going up to inferencing before they even sign up in terms of what the first step looks like? And how would you compare those 2 aspects?

Yes. I'll answer that and, Russell, you can -- so it is different than a typical, hey, I want to tech-refresh my storage, right? I have 2 petabyte requirement, then they put an RFP out. Definitely it's a different DNA cycle. We're seeing cycles 6 to 9 months sometimes. We're seeing even faster ones, because if they know what they want to do. But definitely, the conversation starts from, let's look at the ROI. Let's test this and do a proof-of-concept, POC, and then move on into integrating this with the stack. Usually, as Russell said, it is not only a storage conversation. So it is a conversation where we are sitting in the room. Usually, we have NVIDIA. Usually we have our partners like Domino Data Labs and others because, again, this is not -- it's the cloud probably 9 years ago, right? They want to understand what's the value of them doing this, first of all. So that's where it starts from an ROI perspective. And then it comes on, okay, what's the stack? It's the ML ops provider. It's the GPU provider. It's the storage provider and then the people who are doing services on top of that. And that's the reason we at NetApp, we put a specialist team that are highly specialized AI sales personnel, including technical folks as well, who are really going to engage in these conversations in a more deep dive conversations rather than just talking about storage. Because as you said, it is a conversation that is going from an ROI -- or as I mentioned, from ROI to POC to a kind of purchase, if you would like, in the end.

There's an evolution here for sure, though. I mean I think that it has definitely been a highly specialized sale. It's generally been driven by the line of business and specific AI practitioners inside our customers. What we are seeing is, though, of course, as it becomes more mainstream, more of the components inside a company necessary for turning a POC into production are getting involved in that sales process. So as well as our -- these new buying centers that Hoseb talked about us focusing on, even our traditional buying centers are much more involved in those purchasing decisions. Again, mostly because these other folks that have been involved in making AI real are generally quite good at getting it to a POC. But then when it comes to productionizing and scaling it, that has been a massive challenge. So the more traditional folks that are involved in making those systems production-ready are much more involved now in -- much earlier on in the sales process because they know that eventually it's going to come.

And IT definitely getting involved more and more. I mean 6 years ago, one of our first sales was pure to cardiologists in one of the hospitals in U.K. And we didn't even talk to IT folks. It was -- the budget was staying in the cardiology department. They were buying everything, and they were just asking us a question how to manage this data. Now we're seeing IT being engaged more and more, especially that with the GenAI, what we are seeing is data being copied all over the place, data being replicated all over the place, the security of the governments, it all is going to come into the IT in the end. And people will ask, why is this data sitting somewhere that is not certified? Why is this? Well, that's why ITs started to gain control over this project more and more, because of all the -- now the challenges that comes with the data. In the end, listen, I think you can have the best models out there, but if you don't have your data grounded to those models, you're not going to get anything out of those models unless you're doing just basic stuff, which we all do it nowadays, say, write me a job description or write me -- like if you want to do those type of AI, you can. It's available out there. But if you want to now put this into a real ROI, that's where now IT is saying, okay, you need access to the data. This data is sitting there. This data is -- has privacy into it. I cannot move it to the cloud, et cetera, et cetera. So.

Thanks. Louis Miscioscia with Daiwa Capital Markets. Creating IT applications is very difficult and takes a lot of time, and you guys are sort of on the front row to that. Just curious as to where you think things are right now or sort of where maybe the quantity of proof-of-concepts and when would they actually possibly be deployed into real applications? Are we talking about months, quarters, years? We're just trying to understand that, even though AI is transformative, is it still 3 or 5 years out or something a lot sooner than that?

Do you want...

Well, I'd start by saying that we've been doing this for a number of years, and we have a lot of customers already very much in production. So there's nothing stopping people going to production. I think one of the things I was talking about earlier, though, is just how broad it's going to get in terms of customers that don't -- haven't necessarily invested in or the sophistication necessary to go build it themselves versus those that are more likely just to move to what I would call a value phase, just going straight past development straight into value. So I think as an industry, we've talked about an industry basis, we're definitely seeing a move towards more turnkey solutions. That will shorten those sales periods and those moves from POC directly into production significantly. But that's an industry-wide thing. And so that some of those mentions I had of sort of more industry-specific SLMs, for example, would be a good example of what would accelerate that.

No, you're right. And I think one of the things that needs to happen to accelerate this is, those who already have things in production, they need to start sharing it somehow and what the values they have achieved by doing this. Because a lot of customers right now is like, hey, who else has done it, how they have achieved it, if they got anything out of it. And we know a lot of customers who have done it, but they don't want to share it, right? So I think that's the biggest challenge as well in terms of -- I mean, we heard AWS CEO a couple of weeks ago. He mentioned, I think it was on the earnings, how AI kind of saved a lot of hours of work that it was used to be in the past. I think those will accelerate the adoption. But it depends who you're talking to. If you're talking to the small-to medium-sized businesses, it's still very much early on. If you're talking to Fortune 50, they are already advanced, I would say. But then you take that middle piece as they are in the POC stage at right now, if you would like to categorize it that way.

Great. Maybe building on the last question. I guess where do you see kind of the biggest bottleneck right now? Is it, okay, getting my arms around data governance, security? Is it what data do I even have that I could train it? Or is it kind of the productized turnkey -- like I just don't have these capabilities in-house and I need a product that I can just buy that can help me with that? Like where is the biggest bottleneck?

That's an interesting thing here. It depends who you ask. So typically, when we talk to customers that haven't gone on this journey yet -- and actually, we published a study with IDC a few months ago that talked about this particularly. But when we go to customers that haven't -- or organizations that haven't really gone down this path yet, they typically don't see it as a data problem. They think, oh, this is an infrastructure problem. It's a this, that. But all the ones that have already started down this path realized very quickly it's a data problem. Data is the #1 issue that's actually holding people back. And so that's, of course, where we can step in and really help. It's -- but it's been -- that journey has been super interesting to see. Customers will go ahead, think, oh, we can just move forward and try this out, and then they hit this data wall, right? And the issues that they have are, yes, where is the data? What type of data do I have? Do I have enough? Do I have the right sort of data to go do this in the first place? That sort of assessment often trails the decision to move forward because they haven't even thought about it, and they often don't have a good enough handle on their data estate to start with, to be absolutely frank with you. But the governance piece should not be overlooked. The reality is that AI, to a certain extent, we've seen it a little bit like the wild west. Money has started at the Board. It's been given to a line of business or a bunch of AI practitioners to go and just do stuff. And in fact, if anything, to go to break some plates and cause a mess on purpose. Because the feeling is traditional structures inside customers are set up to slow down innovation. So the idea is, hey, can I get ahead of my competition, this is going to -- as you say, this is transformative, so let's go and do that. Of course, the reality is that the regulations are starting to come into place. The AI EU Act obviously came to force beginning of last month, and it stipulates a bunch of things that customers need to be thinking about legally. And it puts them on the line for significant penalties, not unlike what happened to GDPR. Of course, they're not enforcing it yet. And that enforcement period will start to ramp up over the next 6 to 12 months. But this is really bringing into focus the need to understand and control data and manage data and ensure the right types of data are used in the right way. And so that wild-west mentality that was pervading this whole industry, that's going to move away very, very quickly. And listen, just like we saw with GDPR, we expect other regulatory environments to pick up similar rules as well. So that's just going to be the start of it.

No, I think you hit on the right [ point ].

Go ahead, Wamsi.

Okay. I was wondering if you could maybe just contextualize for us at the high level, are you seeing signs of incremental either data movement from, let's say, tape or something like that where data's just been sitting over there, now we're going to train using this data. Are you seeing any movement like that? Is there any reason to think that the rate and pace of data growth is actually changing with AI, and how so, in -- at your customers?

Yes. As George mentioned on the last earnings call, we are definitely seeing data lake modernization projects happening more often than it was before. And we are subjecting that to the fact that these people are trying to build or make their data AI ready basically to start now moving to the next step of using these tools. Now how much that's going to change or -- I can't quantify that for you. But I think we are seeing that data lake modernization. We're also -- what's happening, Wamsi, is also people who have built their data lakes on old workloads, now that GenAI is requesting more access to that data. Because it used to be cold, right? Now they need to bring that back to life and re-access it. Some of the technologies that were available in the past or today are not really -- it's either creating more cost or it's not operating at the speed they need. So we believe that's why the data lake modernization projects are bubbling up more. Now I don't know if they're moving from tape or not. Historically, these data lakes have been sitting on servers with a bunch of drives in them. And then I think that was the tape to the servers, and now we're seeing that becoming more unified, because also a lot of these technologies didn't have much of a cloud connectivity, especially you're a heavy on-prem user. So that cloud piece is now coming back into like, hey, I need to use the tools in the cloud, but data is sitting cold. So we are seeing data lake modernization for sure.

When you get the lead AI person and the lead storage buyer in the room together, and they're evaluating the AFF A-Series, and I guess more recently the C-Series, are you -- can you confidently say at this point that -- because when you look at those 2 offerings, right, you offer -- it's built in tiered storage, you've got data replication on-prem to the cloud, right? I mean security governance capability is privacy, right, that the last few questions were asking about. At this point, can you say that this is the -- this checks all the boxes from...

A or the C or both, or..?

Or if you could talk about both and how do you feel like those compare -- like if the answer is yes, I think that we can confidently say that, could you prior -- could you say that prior to these 2 SKUs?

I would start by saying that the AI practitioners couldn't care less about A- or C-Series, right? I mean, sort of the storage guy is certainly interested to talk about it. But the way that we would actually engage with an AI practitioner has much more to do with understanding their AI data life cycle and how our portfolio of products are able to actually track that data throughout that life cycle, right, and make their lives easier. So that means all about things like productivity, but also creating an environment where the guardrails are automatically in place from a data governance perspective, and it gives them the freedom to go do what they want to do without putting their companies at risk from a regulatory perspective. So that would be more of the conversation we'd have with sort of an AI practitioner. And up to this point, I would say that the infrastructure people have kind of been on the back end of that conversation. So they've kind of had things thrown at them, and then they've had to pick it up. And then, of course, that realization that picking up stuff that was never built for production becomes very, very challenging when you try to scale and move it and move to the value phase of an AI project. So we are starting to see more of these folks come together into those conversations. I'm sure Hoseb will talk about it in a second. But the way that we talk to these different personas is different. I mean they -- again, an AI practitioner does not care about storage, but they absolutely care about the value we deliver. And we have this very broad ecosystem of both commercial partners, open source partners and cloud partners that enable us to take our value and expose it up in a way that's meaningful to those personas. And that's how we really engage with them. But I'll let Hoseb...

Yes. No. And back to the portfolio, I think the good news here is all of them run the same software, right? And it all comes down on what customer wants from a workload perspective. So if it is a data lake, it's different than if they are training a model, right? So -- and that comes down to a question of performance requirements, cooling, energy, power and all that. But again, I think one thing that differentiates us in this market and puts us up in the front is that we are -- our portfolio is well positioned to capture all these different requirements while at the same time keeping the same operating system running, whether it's in the cloud or on-premises. So I hope that answered the question.

Antoine Legault, Wedbush Securities. Just wanted to kind of expand a bit on your initial comments about converting a prospective customer into an existing customer, going through the proof-of-concept that might take about 6 to 9 months or even shorter in some cases. Can you kind of walk us through what an upsell might look like once a new customer has been on the platform, has been using a solution? How does that look like? And do they come to you and say, oh, we really like the value, do you have more to offer? Can we explore other products? Or do you kind of go to them? Can you kind of walk us through what that might look like?

Yes. So the reality is a lot of these customers, their heavy budgets are going to the GPUs. And then when it comes to storage, they're like, hey, let's start with something that works today because we all know where the expensive piece is. The upsell over there is like now that they start training these models, obviously, the more data, the good data, the model gets better, right? And that is the upsell where we will say, okay, now we want to train a bigger model. We want to engage more parameters into this model. We need to bring a bigger data set to this storage, right? And this is where we see it expanding. The other thing that we -- back to the data lake piece -- is, again, now that we have this system up and running, my data sits on either different storage, different platform, different architecture, how can we make it easier to feed - if I don't want to expand that compute cluster. Can we bring this now into the umbrella of ONTAP? So that's the upsell of now you can even modernize their data lakes to be on NetApp, right? So -- and sometimes we even start as large as they needed to. Not that every project starts small or big. There's no size difference here in terms of capacity. But even if we start small, let's say, there is an opportunity where the model is growing, they need more data to put in there or they need to modernize their data lake so that they engage the data feeding to these GPUs, if you would like. And that could be an opportunity on the Flash, on the Object, and different parts of the business.

There's another aspect I'll just add to what Hoseb said, which is when we win the centers of excellence, so when you win those deals, as more workloads come on, we just -- we get the capacity expansion. That's the reality, right? Customers are attracted to us because of our singular control plane that we have both on-prem and cloud. We believe that AI is probably one of the most hybrid workloads we -- the industry has ever seen. We expect customers to continue to consume resources in various places. GPU accessibility is one good example of that. But also the reality that data gravity exists in different places and the data sources are not always neatly in one particular area. So that's what attracts them to us, and so we're in a really good position to attract more of those workloads wherever they may end up being. So if a customer wants to, for example, have something on-prem, but then consume first-party hyperscaler PaaS services around AI, we're still the right partner to do that. So when we establish ourselves as the standard, we tend to get all the workloads, as opposed to some of our competition, where they get one area like on-prem, and they're having to fight again to go win at somewhere else.

And I'll give you an example, and I can't quantify how many of these will happen as well, where we were not present at a certain customer, large health care, where they purchased the DGXs with NetApp because of what we have showcased to them of capabilities in the AI space. Now because they like all the goodness of what they have seen from ONTAP perspective, and back to Russell's point, now we are kind of replacing some other competitors who have been running the old SAP, Oracle and the other workloads.

All right. We have time for one final question. I see now everyone wants to raise their hands.

5 more minutes.

Last question, Tim, to you.

I'll try not to mess it up. Just on the customer front, a lot of talk about the hyperscalers. You got the partners there. Obviously, there's a lot of attention in the AI industry to these cloud AI companies, cloud AI companies, CoreWeave, Lambda, et cetera. Can you talk a little bit about, just from a high level, what -- going forward, if those companies do better, is it better for NetApp? Is it worse for NetApp? If they go away, is it better, is it worse? Maybe just talk a little bit about how the interaction is different. Do they need more technical capabilities so maybe it's better? Anything you can parse out on just kind of that different customer base? Because they have gotten pretty meaningful, at least on the GPU side.

Do you want to cover that? Or do you want me to?

And you want to throw sovereign in there, another one, too.

I mean you've got us pausing, which means it's a good question, right? Look, I would firstly start off by saying that, of course, these are service providers. And the way that we tackle service providers is very different to the way we tackle customers. It becomes about joint service creation. So how do we help the service providers create differentiated value through their services? And that's where a lot of our value add comes into play, right? So if we are -- if it's commoditized storage services that the partner is offering, then it becomes more difficult for a company like NetApp to differentiate our higher-value services. So that -- so -- but we're really good at that, right? So I mean this is not a concern to us. And what we tend to see, interestingly, with some of these big hosted AI providers is that the customer base they're going after tends to be larger enterprise. So larger enterprises are the ones that actually really appreciate the value of the data manageability features that we bring to the table, right? So they're not just looking for scratch space. They're looking for rich data environments that protect their data, classify their data, et cetera, et cetera. So what I would say is I think we are very well positioned to go after those areas as they continue to mature. I think in that current state, they're very much just price-performance, raw horsepower GPU. But as they get consumed more by enterprises, they're going to want those enterprise features. I think we're in an extremely good position to take advantage of that.

Yes. And we're already in talks in a lot of them. We already have customers in APAC as well who are doing GPU service provider type as well as in North America, some of them. Now we are talking more because they are starting realizing that if they want to offer an enterprise-level services, it needs to have the multi-tenancy and the security features and all the sovereignty that you talked about. And this is where our value is kicking in, right? Us being in the 3 hyperscalers, it is -- it didn't come without no work. And I think that our partnership with the hyperscalers, first of all, puts us upfront. And then these new GPU cloud providers, we are in talks with them so that we are also building similar features from multi-tenancy and all the security features that they need. Don't also forget that us being in the industry for 30-plus years means there's a ton of customers out there who store their data on NetApp. And if these service providers would love their GPUs to be consumed, that data needs to come from somewhere. And if the same way we did do the hyperscalers where we provided that hybrid Data Fabric, we have a great value to add over there because no one is going to bring their data easily today. I think that is one of the biggest -- you asked about the roadblocks, right? It's -- everyone is holding very tight to their data that the goodness of AI is not showing up, and this is where we want to untie the gravity, we call it. We strip off data from gravity.

Alright. Well, Billy tells me we have time for one more question. So Billy's in charge of everything. Alright. So Wamsi, I'll give it to you.

Alright. Yes. I was just curious about the journey over the last 6 years, you said it was deep learning and machine learning which initially got really people very excited, but then the deployment challenges, everything kind of just didn't pan out maybe quite the way people wanted it to. And I'm just wondering, when you think about what the selling motion at that time was, who were you selling to? Who in the business was driving that versus now in GenAI? Who in the business is driving that and how is the sales motion today different?

Yes, good question. As I mentioned, 6 years ago, we were selling to cardiologists. So it was -- majority of it is line of business owner. The IT didn't have budget for AI at that time. As we moved forward, IT started getting more control of that because one of the customers just woke up and found a whole bunch of IT computers in storage, sitting somewhere in the room. I think that has started getting -- with the GenAI, especially cloud, playing a big role, Wamsi, IT is getting more engaged. Now the line of business owners are the ones who are pushing down the agenda, and the IT is trying to figure out how to implement it. Whereas in the past, it was line of business, line of business, and IT just watching and seeing what's happening. Over the time, that's why actually NVIDIA appreciates our partnership, is because of our huge experience in the data center. One of the things NVIDIA wants is to open up to the enterprise. And with our expertise in the data center and in the IT, I think we -- that is where the partnership becomes very important because of our expertise in that. And now the line of business, yes, they have the idea. They have the budget. The Board is pushing down, but it is still landing now on the IT to deliver the infrastructure. So that's the evolution happening. And cloud, actually, the GenAI and all the clouds made it that more real for the IT.

I mean also just the last thing I'll add is that we are going to see more customers that just want to consume AI as a value rather than from a development perspective. And that, again, has much to do with the fact they don't have the data necessary to do things from scratch anyway or the sophistication. So for those more turnkey solutions, there will be things that are kind of regularly things that our IT are going to be the ones responsible for delivering, like copilots, agents, enterprise knowledge management, et cetera, et cetera. And then it will be the lines of business that come in and want the highly verticalized solutions. But again, they're going to be more commercial off the shelf. I mean I think that's the mass market here.

I mean to give you an example, within NetApp we use our own ChatGPT version, right? I'm a line of business within NetApp, if you think about me. I didn't develop that tool myself. It came through our IT and the data science teams and said, hey, here's a tool you guys can go and use. So if you take our example, it is a great way of -- now if you're just yourself and you want to use AI to write a PDF paragraph for you, you don't need your IT. So it depends on the use case and what we're talking about here. That's why I said small to medium businesses are still like not that much. And then you come into the wider enterprise and then the top 50, that's where we see the curve growing, right?

Alright. Well, thank you, guys.

Thank you.

Thank you so much.

I really appreciate it. We're going to take an 18-minute break now, and we'll be back at 11:10. [Break]

All right. Welcome back, everyone. I'm glad you made it back from the break. So now we're going to switch gears and talk a little bit more about core storage and all things ONTAP. So I'm happy to introduce Sandeep, who you guys saw last year, if you were here. And so before we get started, why don't you introduce yourself, say a little bit about what you do at NetApp, and then we're going to open it up for questions.

Hi, everybody. I'm Sandeep Singh. I'm the Senior Vice President and General Manager for Enterprise Storage at NetApp. I look after our portfolio of on-premises products. I have had the pleasure of being here at NetApp for almost 2 years now. And over the course of this time period, I've probably spent at this point a little bit less than 50% of my time traveling, meeting with customers, meeting with partners, and also remaining focused and continuing to rapidly expand our portfolio to have this ability to have an unmatched simplicity at scale as well as transformational flexibility across our unified data storage portfolio. With that, I will open it up for questions.

All right. Well, I know we have a question from the webcast that came in earlier.

Yes, it was earlier, on block. So I'll read it out, I think I can give a little more color to it: how has NetApp's share gain views evolved for block storage as we've released new block products? And then can you talk about the installed base leverage, so the fact that we have an existing installed base area, has that improved our block traction? Or can you talk about success in block-only environments that resulted in net new customers for NetApp?

Yes. So first of all, we used to already offer block storage capabilities with our unified storage portfolio. Last year, we expanded our product offering with an all-flash SAN arrays, or ASA, series of products. When we initially looked at overall block storage, we roughly have about 20,000 customers who rely on NetApp and trust us in some shape or form with their block storage workloads. We introduced a series, we announced it with ASA A Series last year and then ASA C-Series. We've continued to see great adoption across the board from the use cases standpoint across our customers and across net new customers. The way to think about the adoption of ASA series for the SAN workloads is kind of threefold. First, in terms of our overall installed base accounts where we've got thousands of customers that trust NetApp, they love their ONTAP experience. With ASA, we're helping them bring that ONTAP experience to their block storage workloads. That represents then an expansion in terms of adoption of the block storage workload there. And the importance and the value for a customer becomes what I call the simplicity at scale. What does that mean? If you think about customers, they are dealing with complexity that gets compounded through bespoke infrastructure silos. Now when you think about the infrastructure silos across their overall NAS and file environments and when you start to then expand that into silos, VMware or database application workloads, that is where -- when it's a bespoke infrastructure silo, they're having to deal with separate inconsistent management, inconsistent automation, inconsistent data security models, inconsistent operational recovery workflows, inconsistent overall experience. We're enabling them to be able to bring that ONTAP experience to their block workloads and be able to then get consistent management and automation that overcomes the need for talent and skills gap shortages, right? It gives them one consistent data security model they can trust in. It gives them the comprehensive yet consistent operational recovery workflows, and it gives them an overall consistent experience. So that's for our installed base customers. That's an incredible value for them. Secondly, for a lot of the customers that are continuing to work through and figure out what's the right feature in terms of their overall VMware deployment, we're helping customers to be able to come over, offload data management, and through that get upfront savings in their VMware environment, all the way up to 25% savings there, and then be able to have this unmatched flexibility for the future in terms of the hypervisor or the container environment and/or the cloud storage options there. So that's the second key area where we see customers adopting NetApp in their block storage environment. The third key area is essentially as customers are looking at -- they've got all the hybrid disk-based block storage systems. There's lots and lots of them and being able to modernize them to all-flash and being able to do that affordably. This is where finance department CFOs are still requiring and asking IT to continue to lower the IT budgets. We're enabling them to go modernize to all-flash and be able to do that affordably. As part of this, this leverage and the flexibility that we enable customers with, they'll start talking to us, for example, in file environments and then recognize, well, there's this flexibility that exists of bringing a consistent experience to the block storage environment, so it gives us this optionality to expand in to these options. I was just learning about a recent win that was exactly that use case for customers. In other environments, once they recognize, well, I've got this standalone block environment, I can get simplicity at scale, that's helping us go out and win. And we now give them just a overall flexibility of having an end-to-end ASA portfolio that they can leverage us within their block storage environments.

So maybe on that front, when you talk to customers today in terms of refreshing their storage infrastructure, do you still feel like you need to convince some of these customers to move from disk drives to flash? Or is that a decision they've already taken, and really what you're doing in terms of your conversation is really going with the NetApp portfolio and convincing them relative to competition where you stand, and that's the conversation. And the second part to that, when you talk about block, how do we get comfort that most -- that your customers on the block side are not really just sort of moving over from the unified storage they were using previously, and these are actually incremental opportunities that you're capitalizing on?

Yes. Great set of questions. In terms of disk to flash, the -- larger scale, there is a shift that's happening where customers are optimizing for flash at a broader scale. When you look across our portfolio, we've just got a fantastic set of options available to customers in terms of the all-flash offerings. With that said, what I will say is that what matters to customers, especially with the unpredictable macro and tight IT budgets and IT budget scrutiny, is this notion of high-quality, lower-cost solutions. And when you look across the data life cycle, having the ability to have lowest cost of data over its life cycle is important for customers. So where we continue to see great adoption across the board is you look at primary storage, that's going to be typically all-flash, right? But when you look at the data, anywhere from 60% or more of that overall data is going to be cold data, and being able to automatically based on policy be able to tier that data to lower cost storage options gives us the advantage of enabling an end-to-end portfolio that is underpinned through ONTAP of being able to seamlessly enable customers to go and adopt us for their primary storage workloads yet also ensure that we're cost optimizing the cost of data with hybrid flash options as well as overall Object with storage grade options available to them.

How do we get comfort that the block customers are not unified customers moving over?

So in terms of the block customers, you will see the combination of the following. There are many customers who have their file and block environments in the same environment. They may have a large file environment and a smaller block environment over there or it could be vice versa. And this is where unified storage provides them with the best option of essentially being able to have file or block on that same shared underlying infrastructure. And we provide that. That's available to customers and lots of customers are leveraging that capability. On the other hand, if customers have a separate file environment and a separate block environment, this is where AFF for that file environment and then ASA for that block environment is the perfect fit for them. And this is where it becomes a net new expansion opportunity or a net new logo opportunity for us. And so unified doesn't necessarily substitute for the ASA use case, where it's that standalone overall block environment that they're going for.

Understanding that you've kind of been competing against folks who were more upstarts on -- for your C-Series product, you were competing against kind of storage start-ups that didn't have the full portfolio. But a number of your competitors have kind of announced that they intend to kind of offer QLC products. How do you envision that that kind of changes the landscape? Do you envision that that might change evaluation times? Or do you feel like that the argument has been made or the inroads have been made?

Yes. So in terms of looking at capacity flash underpinned through QLC technology, first of all, we have just continued to see fantastic adoption of C-Series across the board. Yes, some competitors have gone in and announced products. We have not seen that make a difference. Why? First of all, when you think about the customers' needs, when they're putting, let's say, their Tier 2 capacity-focused workloads onto C-Series, data management is still critical. We are bringing this notion of comprehensive data management and still making it available even in the capacity flash series. Secondly, and similarly in terms of our data management, when you think about ransomware and cybersecurity protection and detection and recovery, that is critical for customers across all of their data. And this is where we're also really changing the game in terms of enabling what's typically a post-process detection to that real-time detection in a matter of seconds to minutes. Being able to do that, which is designed for that 99%-plus accuracy, that minimizes the false positives as well as has the ability to be able to provide the accuracy of detection, right? We earlier this year became the first storage vendor with SE Labs validation to get that AAA rating with this ARP AI technology, to get that 99% -plus overall rating in that detection. That becomes critically important for customers as an overall priority and then couple that with the ability to rapidly recover from ransomware attacks. So that's another area. And then I talked about the complexity challenge and how can customers just simplify at scale. So rather than going and putting and continuing to propagate bespoke infrastructure silos, customers are increasingly looking at, well, geez, when I start to think end-to-end, not just a point solution, what matters most and continuing to simplify and get this notion of simplicity at scale that only NetApp is able to deliver across a fully interoperable portfolio that is then underpinned through the lens of ONTAP. So those become some of the prime motivations as customers are thinking through capacity flash options. And the last part I'll touch on that is in terms of the economics. Even being able to then cost effectively tier the data over the data life cycle and still get the lowest cost of data, this is another one where in the competitive products you end up getting bespoke silos versus a fully interoperable portfolio.

Wondering if you could discuss a little bit, just kind of in this tug-of-war over the years of movement to cloud and repatriation of some workloads, data sovereignty and whatnot, if you could just give your perspective on kind of where we are now, because we hear a lot of examples of both. And then related to that, as we move further along in AI, do you think the calculus around that tug of war changes at all?

So in that tug of war, I think across the spectrum of thousands and thousands of customers, we will see all of the above. We continue to see customers who are adopting the public cloud storage options. We've seen a lot of customers who are adopting the hybrid cloud storage options. We're also hearing about customers who, as they have scaled workloads, and increasingly with the overall cost pressures, where we're hearing also about some customers who are repatriating their workloads. What is important and what we're focused on is ensuring that our customers have that complete flexibility to be able to have market-leading offerings for the on-premises options available to them, whether in a CapEx form factor or storage as a service side of it, have the leading -- the public cloud storage offerings, the first-party native cloud offerings that are underpinned through ONTAP, where NetApp is the only one with the first-party native cloud offerings across each of the hyperscalers, have the necessary technologies that are available to them for the hybrid cloud use cases, which then is inclusive of secure and efficient data mobility that becomes critical for customers across the board. So we are ensuring that we have the necessary choice points as and when, whether they are going to go and be able to leverage the agility of public cloud for being able to seamlessly move their workloads and be able to still get the enterprise resiliency, the data management capabilities that they value be available to them. Or if they need the hybrid use cases, we're enabling those, as well as if they're looking at repatriating, we're providing the right options available to them. We feel incredibly privileged to be in a position of providing this unmatched flexibility for customers and being that partner no matter what their use case there. You asked a follow-on question, which was tied to AI of how does that change the calculus there? When we think about AI -- and hopefully this is building on the prior session there -- when we think about AI, we're going to see the customers leverage, first of all, a lot of their enterprise data that sits on premises and the enterprise AI use cases. Ultimately, customers are looking at how can they truly unleash the power of AI and GenAI with the context of their enterprise data, right? And so we're going to see a lot of that happening. We are enabling customers to have their AI or their data to be AI-ready as well as bringing AI to their data to what I call this notion of an AI data gap. So that's kind of the first area. Secondly, you will see, obviously, there's lots and lots of investment in terms of AI, GenAI tool sets that are available in the public cloud. And we want to ensure that we are providing the flexibility of enabling a kind of a seamless experience for customers to be able to leverage their data and making it available to their AI, GenAI tools in the public cloud. That's building on some of the announcements that we have made with the, for example, the AWS workload or BlueXP workload factory and various integrations that we've demonstrated in terms of the AI, GenAI tool sets in the cloud.

Well, Sandeep, since the audience is reticent, one question I get from investors all the time is, do disk drives still have a place in a modern data center? Maybe you could talk a little bit about that.

Yes. It's a great question. Look, in terms of overall media underlying it, it becomes important when customers are thinking about what is that cost of storage. And as you think through the life cycle of data in terms of the disk drive, that hybrid flash storage, what you're going to fundamentally find is it is still that lowest cost of storage that you can get from an on-prem standpoint. So the short answer is yes. Disk drives still have a specific use case that they serve in the customers' data centers. When you look at basically from a raw overall cost standpoint, flash is still not at a point where it can substitute for that disk-based storage perspective. And so what becomes important is, when you start to think about the optimizing cost data through tiering, a backup target use case, say, when you think increasingly about cyber vault use cases as well, ultimately you need the economics to be there. And that's where the disk-based options are still important.

Can you talk a little bit maybe about where you think we are in the cycle of customers sweating their assets, on-prem assets? It feels like storage spending has been relatively muted now for a fairly long period of time. Are you seeing any green shoots about things kind of starting to pick up in that regard? Or how much longer, given that you have a pretty good view into where your customers' utilization rates are and the capabilities are, how much longer could that extend out to?

I would say, look, overall, when we look at the -- just the overall adoption of systems, whether it's on the AFF A-Series side or C-Series side, we're continuing to see just overall strong adoption. And utilization rates, when you think about sweating out the assets, one would assume the utilization rates would either increase significantly or the customers are sweating out the assets. I would say generically we're seeing that across the board; that's pointing to customers have a need. Their data fundamentally continues to grow, and they have a need to continue to go and purchase more underlying storage capacity and systems tied to it.

You mentioned as-a-service in consumption models. Can you talk a little bit about -- some of your competitors harp on this a lot, but it doesn't seem like at scale, it's really happening. So what's your sense of demand or appetite for the different consumption models, as-a-service versus CapEx across -- more broadly, I'm sure there's some of each in your broader customer base. But if you could just give us a little sense on how we should think about that potential transition over time?

Yes, absolutely. Look, storage as a service, various customers are transitioning or looking at storage as a service across their spectrum, especially when, for example, refresh or net new purchases come up. The Keystone is our storage as a service offering. We're continuing to see just an overall fantastic adoption across Keystone. I know some of the financial information, we release that as part of the earnings announcement there. But we're focused on the enterprise segment. In the enterprise segment, our position is we want to provide those leading options to customers, whether they are looking at CapEx, whether they are looking at storage as a service on premises or whether they are looking at essentially cloud storage as a service options for them. We're unique in that way of being able to provide that complete flexibility with the leading options to customers. We want to ensure that we're not pushing customers one way or the other. We're providing them the flexibility to adopt as they are ready for the right options for them, right? With that said, we are seeing overall great momentum with Keystone. And it's important to not only have that momentum to also provide that overall service level that they're continuing to then go and build out the adoption with Keystone. So we're incredibly happy of what we're continuing to see there. The other thing I'll say is the customers who are looking at essentially the -- they want to get that agility of cloud, and they're on a journey, we also see a cohort of customers who then are leveraging and transitioning to storage as a service on premises as well. And this is another one of those unique capabilities that we're bringing and making available to our customers.

Lou Miscioscia, Daiwa. We just had the AI session, which was very good. But if we could go back to that a second, if you look broadly at your customer base, could you just break it down? How many are in proof of concepts right now, how many are maybe not doing anything, and how many have actually gone in size from proof of concepts to deploying? And if you want to throw time frames into that over -- looking forward, that would be great, too.

That would have been a great question actually for the last session.

Well, we did. But I figure you're here, so...

The way I'll talk about it more generally, right, is that we see an incredible opportunity forward-looking in terms of enterprise AI, right? The enterprises across the board have been looking at how do they go and take AI, GenAI and use that for driving a productivity for delivering net new customer experiences and/or enabling net new areas of innovation. Many have been looking at holistically and in concepts and as well as the specific use cases that are going to be incredibly important within their context. What becomes important is essentially this notion of an AI data gap that I touched upon, where customers are challenged with how do I get my -- I've got multiple data science teams. And how do I ensure my data is AI-ready? How do I ensure the right data sets are available to the right data science teams? How do I ensure that the right levels of privacy and security are assured there? And then how do I ensure that overall efficiency is available and overall model versioning and the data sets associated with those are available? Those become incredibly important for customers. This is where this notion of helping customers bridge that gap in being able to discover their data and prep the data and make that data AI-ready becomes important. Secondly, in terms of essentially being able to bring AI to their data also becomes important in bringing overall efficiency to that end-to-end AI life cycle. So when you think about the spectrum of how are we helping customers, when you think about the -- one set of customers has been the overall AI as a service or GPUs as a service, and a lot more of that use case is the overall foundational large language model training, this is where -- and if you think about it in terms of our portfolio, we're discovering where a lot of it -- those set of customers, they want the overall scalable levels of very high levels of performance. This is where we have a SuperPOD with the NVIDIA SuperPOD with the BGFS plus our E-Series solutions that customers are adopting. That's one use case. When you think about it in terms of the overall enterprise AI use cases, if they are not -- not a lot of enterprise AI is going to focus on as much of the large language model training, yet there are still customers that are going to go do large language model training or small language model training, all right? So we've got, whether the overall -- the SuperPOD that's there or the overall BasePOD solutions that are in place for them to be able to leverage it. And then when you -- we also see when customers are on that beginning of that journey in that data prep phase, they're modernizing their data lakes. That usually takes them to Object, and that's where our StorageGRID offering is seeing great adoption. And then when you start to shift to the retrieval-augmented generation, RAG, use case or inferencing or fine-tuning with their enterprise data, this is where our overall data management, a lot of the data management challenges that I was touching upon, becomes critical. We've also put a overall converged solution that we call the AIPod with Lenovo, with the NVIDIA L40S GPUs with OVX, that we announced back in May. That also provides them this converged infrastructure stack for them to be able to use for their RAG or inferencing type of use cases. And then our FlexPod solution is also being used in those instances.

Actually, we're up on time. So thank you, Sandeep. I really appreciate you coming here. Sorry, I just know he has somewhere else to go, and I don't want to be the inconsiderate one to make him late for his next meeting. So I really appreciate it. Thank you. But I think whatever questions you didn't get to for Sandeep and still have outstanding, Jeff can probably handle. So I'd like to introduce Jeff Baxter. He's our VP of Product Marketing. He can kind of handle a broad swath of all your questions. I know every session ended a little bit early for your taste, so you'll be able to pose all the questions to him. But before we open it up, Jeff, why don't you introduce yourself, say a little bit about what you do and your time at NetApp, which I think helps provide the context for why you can handle such a broad set of questions.

So thanks, everyone. I'll just probably take a seat, if you don't mind. So my name is Jeff Baxter. I run Product Marketing here at NetApp. Before that, I ran Product Management for ONTAP, which is the operating system that powers a lot of our on-prem systems as well as our in-cloud systems. I've been with NetApp now for 16 years. So gone through several transitions at NetApp from -- when I started at NetApp, I was actually in sales. I was an SE. So I was out selling our actual individual storage systems when we were a storage company. And the evolution since then in terms of going out into the cloud, and now the evolution that we're embarking on with AI, it's just -- the company has transformed 2 or 3 times over, as I think all of you have seen, and it's been just a remarkable ride. So I've been privileged to be a part of it. I've been in all different parts of our business, gone from sales to product management, now running our product marketing as we continue to reinvent ourselves as the intelligent data infrastructure company. So that's a little about me.

I don't know if Kris is going to like this question, but...

If she tackles me off the stage, we'll know.

It's about pricing. So there's been a lot of -- I mean, obviously, you guys don't price per gigabyte or terabyte or anything like that. But there's now a lot of storage as a service. There's a lot of new applications. There's more value being placed on ROI for AI. So how -- as we enter sort of this new phase with your customers, how does pricing change for the better? And I'm sure there's some reasons it changes for the worse. Like how is your pricing power look over the next few years?

[ Can you go on that one? ]

So I think you can talk about big trends, right? What you see in the industry. ONTAP One would be a great thing to talk about.

Yes. So I think in general, we're trying to be cost competitive in the industry, right? We definitely are an option where we add a tremendous amount of value-added services directly into the operating system. So ONTAP One is our ability to add all the data protection you would need directly into our storage systems, more recently adding anti-ransomware capabilities directly into our enterprise system. So I think we see customers terrifically valuing what we're seeing. I don't know if any of the sort of secular trends, like AI or others, I'm not going to comment on what the pricing will do there, right? I do think we continue to be cost competitive in the market, right? We continue to see wins against competitors. A lot of those wins are based on value and based on cost. I think for our customers, when they look at our price tag, the important discussion for them is really around TCO, right? So as you said, it's not about dollars per gig. When we have storage as a service customers, they're actually buying a given service level through our Keystone program. When we have customers still buying through a traditional CapEx model or a leasing model, they're really buying more based on outcomes rather than just raw dollar per gig. And when you build in best-in-class storage efficiency, build in the data protection, build in the simplicity operating system that I assume Sandeep talked about in terms of making it simple at any scale, all of that tends to really drive down TCO. So one of our best sales tools with customers is we'll sit down and walk with them through a TCO calculator and put in all of their assumptions, all their power, right, being far more power-efficient in their data centers, for example, right, can drive cost savings. So when we get to the end of that, we almost always find that NetApp is a more efficient solution over the long run from a TCO perspective.

Great. I also have a pricing and packaging question that I actually wanted to ask Sandeep, but I think, as Kris said, this might be a better question for you. So if you think about the storage market in terms of P x Q, I mean we're all investors or the investment community here, if you think about [ P x Q ], it's been pretty fascinating because the P, or the unit price on a per gig or per terabyte -- it doesn't matter, right -- the P has come down significantly, but it seems like -- and I'm talking actually on the -- specifically on the all-flash side, right? -- and -- but it seems like it's kind of hitting -- it's like -- there's only so much more to go perhaps in the near term. Meanwhile, the Q has increased dramatically, right? And that's opened up -- obviously off the smaller base. And so if you think along those lines, how does the -- so if you were to apply it to -- the question for you is, like, if you would apply it to the enterprise side and then also the multi-cloud hyperscaler side, how does this play out? That's kind of one part. And then the second part is when you -- if you're like just about there in terms of the P, then are you thinking about programs that you can work with customers? So maybe -- because their wallets are fixed or they're not growing nearly as fast as the volume growth is. So are there programs considering where you can maybe help them bridge the gap maybe for near term so that you can -- once you get past that hurdle, it unlocks this massive opportunity?

Yes. So that's a multipart question. I think you're right, the data growth is outstanding, right, especially going into the AI [ area ] with the massive growth of data there. So I won't comment too much on the future price or where we see that affecting revenues because Kris will tackle me off the stage. But what I will say is, I think for a lot of those customers where they're concerned about, that's really where they're looking at storage as a service offerings, right, and being able to go out and apply certain service levels, right, which tend to insulate them a little bit from variations in pricing and other things like that, right? So they're buying a service level. And the other thing for a lot of our customers is as they're not sure about what the rate of data growth will be as they're entering sort of an AI space, pre-purchasing CapEx multi years out, an alternative option can always be going with storage as a service. And that lets them basically grow as they need to and basically match their cost curve more carefully to what their actual consumption is. And so I think they're addressing some of the variability through the continued growth of our storage as a service. I think that's the main thing I would answer out of that.

Over the past couple of years, there's been a lot of changes in go-to-market, maybe different incentives for kind of cloud sales. And I just wonder how that's changed kind of product marketing? Or has the marketing to the customer changed? Or has it really been kind of more go-to-market side?

Yes, that's a good question. No, I think the marketing has definitely changed over the last 10 years. We actually went through a shift probably 6 years ago. Actually, I remember being at INSIGHT, I want to say 2014, when we first talked about putting ONTAP on the cloud, right, and we first did a demonstration of ONTAP running on the cloud. So we're now literally at a decade, and it was 5 years ago when we unveiled Azure NetApp Files, and we were [ the first pioneer of ] cloud service. And I think at the time everyone saw us as a storage company, right? And so from a marketing perspective, we went incredibly hard at being a cloud-centric company, right, and really getting that message across and getting that ramp started with cloud. Where we started to balance over the last couple of years is bringing it back to talking about that balance that Sandeep talked about that basically, all of our customers are in some way hybrid multi-cloud, right? They're either starting their journey, well along their journey. And so we've been able to really embrace that, and you've seen that change towards where we talk about ourselves as the intelligent data infrastructure company, right? There's nothing in there about cloud or on-prem storage, right? It's about data infrastructure regardless of where it lives and about applying intelligence to it. So that really is the message we've taken out to market. And I can honestly say all of our -- like even within my team, for example, our solutions team, our launch team, all that other stuff is integrated across cloud and on-prem. It's not separated anymore, right? So we certainly have subject matter experts. Like I'll have a marketing person on Azure NetApp Files, I'll have another marketing person on Sandeep's AFF systems or ASA systems. But when it comes to how we bring those solutions to market for a customer in terms of a VMware solution or Kubernetes solution or database solution, all of that is horizontal, right? All of that is what's the right solution for you. And so when we go talk to customers now, it's not about we've got a great on-prem solution for you, and let me bring in a sales specialist to talk to you about a cloud solution. It's -- we've got a great data infrastructure solution for you. Let's work with you on where your choice of data needs to belong. And we can help guide you along that path and use AI ops to look at your workloads. We actually have cloud advisers that will look at workloads and say: this one is cloud ready, this one you could put in the cloud but it might be more of a lift. And we can actually guide customers in taking that journey. So I think that's the short version of how it's changed marketing is it's really talking to customers about data infrastructure as opposed to either being a storage company or being a cloud company. We've elevated really the entire message.

More on-prem question. Just in terms of how you think about mix within flash changing with your customers, particularly when customers sort of used to use more high-end flash products now moving more towards mid-end products like CCvs, et cetera. How do you think about how material that mix shift could be either because you didn't have those products earlier in the portfolio or for a variety of reasons, that they were more performance-conscious, they were using more high-end products. How do you think about that mix shift? And how material would that be? And what would be the counterbalancing forces there?

Yes. So I can't talk directly about how material that shift would be. I think that would be more, unfortunately, a Sandeep question or others. I can say I think we're achieving a healthy balance in customers, right? So customers are adopting the whole breadth of our portfolio, performance flash, capacity flash, even hybrid flash. And so we're seeing, I think -- I guess I'll just say, I think we're seeing a good balance there. And I can't really go too much farther than that as [indiscernible] shift.

Lou Miscioscia with Daiwa again. If we think of AI before the NVIDIA moment last May and afterwards, maybe if you could talk about your product marketing budget, however you want to define it, people or whatever, I guess before what it was, which I assume wasn't very high, to just where it is now in proportion to whatever else you're doing?

Yes. I think, like everyone else, we've seen the tremendous opportunity for AI, both in our customers and in our own business. There's been significant shift and interest and investment in resources and marketing AI. For example, in my own team, I won't get into specific numbers, but it's been a significant burst in resources moving towards that as an investment engine, adding additional specialists. I think one key thing with AI is we have to be able to market to multiple different segments, including the infrastructure buyer, the data scientist buyers, some of whom don't know each other within the same customer, right, and being able to build messaging out to the market that embraces all of those different personas. So that's definitely required additional investment. I think without previewing it at all, I think from the keynotes today and tomorrow, you will see just how much focus we've placed on AI and how much -- in a lot of ways over the next day or 2, we'll have announcements that will obviously, in some way, concern AI. And you'll see a lot of the marketing messages that will come with those and how we unveil those on stage. And that's going to be a key focus, I think, of our selling and our campaigns over the rest of this fiscal year and well beyond.

Jeff, in the beginning when you guys were working with the hyperscalers, there were some fits and starts. It was a different motion. And so can you talk a little bit about the learnings, and are all 3 now where you want them to be? Maybe talk a little bit about specific programs and how you've been able to get through the friction and kind of jointly sell these solutions to the end customers.

Yes. So I'll give you what I know. Unfortunately, I'm not the subject matter expert. I think you had Pravjit up a little bit earlier today who maybe would be a little bit more inclined on those sort of questions. I think it took a while for us to learn how to sell in a motion where it was not a direct SKU on a NetApp price sheet, right? I think we saw initial tremendous growth from NetApp customers who just wanted to run ONTAP in the cloud, right? And that was the very early versions. But when we moved to this first-party native service and co-selling with Microsoft and Amazon and Google more recently, that took a lot of work on the go-to-market side to figure out how to do that as a co-selling motion, where we could add appropriate effort. And I think we've resolved most all those issues. I think the team that the go-to-market team, between Pravjit working with Ashish and working with Dallas and Caesar and really the whole go-to-market team there, have really figured out how to make that moving forward. So I won't talk about the future, but I think we're really satisfied with the progress to date of where the cloud business has gone.

All right. Well, I have a question for you, and I'm going to build a little bit on Samik's question, and I get this a lot from investors is, right, we have the A-Series, the C-Series and FAS products, they're all ONTAP, all unified. How do customers think and choose between them? How is the product positioning work across those different families?

Yes. So at a high level, the product positioning is all around mission-critical, incredibly performance-sensitive applications. So from a technical perspective, it's sub-millisecond, typically sub-500 microsecond latency on those A-Series high-performance systems. That happy middle ground with the C-Series, the capacity flash systems, where you're getting low single-digit milliseconds of latency, so 2 to 5 milliseconds of latency, is really where a lot of customers are finding the vast majority of their workloads. And I think that is one learning, I think, over the last decade of the move to flash, is we went from hard drives and we went straight to incredibly performant flash. And in some ways that was, as an industry, I'll almost say gold plated, right? You almost overshot the performance market some ways because that was the only alternative. There were no middle grounds along the gradiation. What capacity flash has allowed us -- and certain parts of the industry, I think there's a lot of our competitors who have not caught up at all in capacity flash -- but it's allowed us to go in and capture that middle ground where customers are actually being overserved on performance, right? They were being overserved on latency. And in doing so, that's one of the ways we've been able to achieve cost savings for them if they were literally overprovisioned in terms of their latency. So that's where the C-Series stands. Then FAS and Hybrid Flash is increasingly -- and not -- it hasn't 100% moved off of primary workloads, but it's pretty much moving in that direction. And we tell customers in general, with only a few exceptions, you're not going to want to put primary workloads on disk any longer, but they are a fabulous place to replicate to. There is a fabulous place for disaster recovery. I think one place we're putting a lot of focus on is cyber vault solutions. So we announced one of those earlier this year. I think you'll continue to see things at INSIGHT about them. You'll continue to see us talk about it, because we increasingly have customers ask us about how can we get logically air-gapped solutions that just keep the data cold and completely isolated from attack, and that's a perfect application of disk. And that's something we can do within our architecture because we have our disk running on the exact same architecture as we have all of our other products. The final point I'll make to Kris' question is, well, 2 points. One is we have a whole bunch of tools, some that the customers have access to from an AI apps perspective, if they're an existing installed base customer, that will guide them towards what sort of system would be right for a refresh. And then we have a wide variety of tools available for our partners and for our internal NetApp sellers who will specifically say, workload by workload, performance by performance, this is the exact right platform to position for this one. And the good news is the capabilities are all the same. So they can still sell the value proposition of intelligent data infrastructure, and the idea of which platform to host it on on the back end or which system to host it on on the back end is more of a speeds and feeds and architecture discussion as opposed to a front -- start of the conversation discussion. And finally, we have customers who have all of them, all of the above. Most of our large enterprise customers are going to have a mix of A-Series, C-Series and FAS, sometimes mixing together in the same cluster even, seamlessly and using AI ops tiering data between all of them. And so the idea is the data can flow to wherever it needs to at a moment's notice. So data can start incredibly hot on the A-Series, it can sit there for a user to find cooling periods, a week, a month, whatever, automatically flow down to capacity flash. It could sit there for another quarter, wait till you're past quarterly earnings or whatever reporting period you're in, and then tier that data off on to FAS and keep it there for the long run. And then if that data ever needs to be accessed again, it can seamlessly flow right back up that river. And so by building that sort of combined infrastructure, that's where customers are really able to optimize their service level directly to their cost.

Can I follow up to your question there? Because you have interesting perspective. On the -- so like I think there's a perception in the investment community that our all-flash growth is the transition of business from our FAS hybrid or HED-centric business to the C-Series. And so the all-flash is just a transition in revenue. Could you speak to that C-Series business, whether or not -- you mentioned at the start of your answer that you thought it was a hole in terms of the latency that that product would have kind of addressed. Do you think that that is then bringing net new customers to NetApp? Or do you think that it kind of smooths out that kind of product kind of offering from low end to high end, and so that's like really what's going on. And so we're just kind of transitioning people that [ went ] in a product that we didn't originally have, I guess?

Yes. So I think it's more than just a transition. I think that we do have customers from competitors who are sitting on legacy disk arrays who the C-Series opens up new opportunities to transition, right? They weren't able to transition to the A-Series at a price point. They didn't need that performance point. And so we've been very successful going out and penetrating sort of the mid-range of some of our competitors on their refresh and being able to move over. So it's not just a refresh of our own internal disk. The other thing I'll say is, based on my previous answer, it's not like disk is going away entirely, right? So it's not a one-for-one replacement. A lot of customers are -- and a gentleman asked earlier about the data growth -- some of that data growth is in primary data. A huge part of that data growth is also in secondary data, right? Because secondary data now can be used for building data lakes, for building data analysis, you need a third or even a fourth copy of data for some of the regulations and to have it locked in place in a cyber vault. So all of that growth in data can be served by the FAS market. So to the extent that even there's some of the FAS market moving up to capacity flash, which there certainly is, there are plenty of workflows flowing in the back end to keep the FAS market moving as well. So I wouldn't say -- without going too far into the future, I don't think it's a 1-for-1 refresh. I think that there's legs on both of those for many years to come.

So you guys have talked about the siloed people that you compete with. Is there a way to think about like -- and you mentioned like the ability to supply customers so they can go up and down depending on data use. Is there a way to think about like where that evolution is? Like how many customers out there, whether large or small, are still very much siloed and have to overcome that, especially in AI, versus how many have already made that transition to being more flexible like your vision fits.

I mean, there's a joke: just look at our market share. But that's probably not what you're asking. That's a good question. And I don't think we fully know the answer. I do honestly think, all joking aside, None of our competitors have a unified stack, right? That has continued to be the NetApp differentiation. There are a couple of competitors that are, without going into names, they just focus on one niche, so they just have one offering, and that's fine, right? I wouldn't call it unified in terms of being able to do structured, unstructured data, object, in the cloud, on-prem. And so almost by definition, any customer we deal with who is not a NetApp customer does have some degree of fragmentation within them. There's NetApp customers who have fragmentation, right, that we're working with. That's one of the reasons we have our BlueXP unified control plane and are bringing all of it into our common orchestration set. But just about every customer we run into who isn't a NetApp customer has some degree of fragmentation, unless they're a very small or sort of niche-focused customer. It's very rare for me to run into a customer these days that doesn't have multiple storage arrays from multiple competitors lying around in some state or the other that they just have not been able to unify because each one is addressing a very specific need, that they haven't been able to find anyone until they met NetApp that can meet all of those needs out of a single operating system, out of a single set of appliances.

Kris may get mad at this one as well. I just want to go back a little bit before you had the QLC, you characterized it as if someone was refreshing a hybrid or disk-based system and they wanted flash, they would be overserved more or less on latency. So -- but my guess is they wouldn't want to pay for that. So what was the NetApp strategy then before you had QLC? Was it more discounting, was it more share losses? And then if we take it to now, is there a risk that we start to see more workloads that are more on the hot side go to QLC because of macro or just deal with the latency or the shortcomings? Is there a chance that there is a little bit of a price change or whatever because it might be good enough for a little bit more of the workloads than what we're seeing today?

So I don't know about price changes, and I won't go into those. I will say, going back several years, I don't think there was significant discounting in regards to being overserved, because people didn't realize necessarily that there was any middle ground. And so the market kind of found the price point that was right, as it is wont to do. I do think that the fact that NetApp had hybrid flash offerings that were still incredibly performant, and we had that FAS business that continued to perform meant that there were options for customers, right? Whereas if they went to some of the all-flash competitors, they really didn't have an option. So those all-flash competitors, they either had to pay the higher price tag or those all-flash competitors were pricing, I'll just say, significantly below market at that point, right? So no, I think you saw the success story with our ramp to all-flash over several years and how fast we grew in all-flash. And I think we did it -- without going into the detail, I mean, you can go back and look at the earnings statements from there, I think we did so quite successfully. And so I don't think it depressed earnings there. To the second part of your question, will there be, as we introduce and as we continue to rapidly grow C-Series, will there be customers that refresh their A-Series over to C-Series? Probably, but that would have been probably a natural transition regardless. And I think we're continuing to see them growing more and more high-performance workloads. I mean if you look at these high-performance LLM training models, if you look at high-performance inferencing, different things like that, there's a whole new generation of workloads that are no longer underserved by A-Series. So as much as there's the potential for some moderate shift of those overserved workloads down to capacity flash, right, there's as much potential, if not more potential, for new workloads coming in that are incredibly latency-sensitive and want incredibly high throughput. You -- if you're hooked up to an NVIDIA DGX system, for example, the cost of keeping it waiting is far in excess of the price differential between a C-Series and A-Series system. So customers who are spending millions of dollars to build out that sort of training environment for NVIDIA are more than happy to pay that premium between the types of flash to not have any cycles of their GPUs sitting there in I/O state -- in an I/O wait state.

I guess it's back to me. So you mentioned BlueXP. And I think that's probably an under-understood offering from NetApp, at least amongst this community. So maybe you could talk a little bit about what it is and how customers leverage it and the capabilities it brings.

Yes. So BlueXP is our unified multi-cloud, hybrid control plane. And in some ways, I think we get less credit for it because it's been this very gradual evolution, right? It started back in 2015, 2016 as like cloud manager. It was just a way to deploy some instances of ONTAP into the cloud. Over the last 3 or 4 years, it's really evolved into having complete management of basically the entire NetApp estate. So customer's entire data estate sitting there, not just on-prem, but in all the major clouds. So we have customers that have chosen NetApp because the whole is more than the sum of its parts, right? They're able to literally see their entire data estate at a glance. And then the nice thing we've started to do over the last couple of years is be able to add intelligence into it, right? So the whole intelligent data infrastructure. Yes, it's a marketing slogan, but each word actually means something, right? So the data infrastructure is what we built up, not just as infrastructures on-prem, but it's data across all of it. So we actually see your data, we can see metadata, we can see what's changed about the data. And then adding the intelligence into it. So for example, we added a ransomware protection service that can actually understand not just individual [ LUNs ] or individual files, but actually understand workloads. And so it will know an entire workload, it will know where that workload is scattered, some on cloud, some on-prem, and it will be able to monitor it in real time for ransomware attacks and be able to respond, allow you to protect an entire workload in real time and recover an entire workload in real time regardless of where that infrastructure is. So it allows customers to not necessarily have to have an Azure specialist, an Amazon specialist, a Google specialist, an on-prem SAN specialist, an on-prem NAS specialist and then a security team to come in and do the restores. You can literally have an IT generalist perform the job of so many of those specialists by being able to orchestrate entire restores directly out of that operating system. You can have -- you can drag and drop to tier from on-prem to the cloud. You can drag and drop to tier from 1 type of on-prem to another type of on-prem, another data center. You can set up your disaster recovery in a couple of clicks directly within that interface. And it's all included essentially at no additional charge to customers directly within that interface. So once they get in there, they're able to see the entire state of their fleet. We've added AI operations directly in there, so there's a digital adviser that will tell them if they have any risks. It will tell them about things they could ameliorate or tell them about the health of their entire environment. And it puts it all there at the tip of their fingertips. And that has been, I say honestly, one of the more hidden, but one of the more sticky features of NetApp, because once you can get a complete unified view of your data estate, it gets very hard to want to fragment it out after that.

All right. time for one last question. And if it's not from the audience, you guys are going to have to take it for me.

I was wondering, there was the point in time when NetApp was talking a fair amount about hyperconverged, and we don't really hear that all that much anymore. But if we just step back and say there is a role for hyper-convergent transition to cloud maybe for a lot of customers, how do you think about that? Where does it fit in your discussions any more or your messaging?

So I think in my own personal discussions, hyperconverged has almost disappeared, which is fascinating, right, given how much prominence it had just a few years ago. There's a couple of reasons. I think a lot of the people who were initially looking at hyper convergence realized that they could go to cloud and do it that way as a service, because the move towards hyperconvergence was really about trying to simplify the stack. And if you're trying to simplify the stack, it's a good sort of slippery slope to saying, "I just want someone to provide it for me as a service," right? So doing that as a service on the cloud, [ simply has ] replaced hyperconvergence in a lot of places. The other part is I think customers started to move down the hyperconvergence path and all of a sudden became very afraid of lock-in, right? So if you move down the hyperconvergence path, for example, and you were on a hyperconvergence platform that use VSAN or used VMware, you got an interesting surprise over the last couple of years, right? And so customers have been, even if they're not on a VMware-based platform, they've been very conscious of the fact that if they lock themselves into a hyperconverged platform, they're very dependent upon a single vendor for all aspects of their pricing, all aspects of their stack. And as they're looking at how to adopt the cloud, how to adopt AI, I think they've all become very hesitant, for good reason, about lock in. And so we want to continue to partner and be best of breed, right? And we build these converged infrastructure stacks. So we have AIPod with NVIDIA. We have FlexPod with Cisco, right? Sandeep talked about the OVX partnership we built out with Lenovo, right? So we build these converged infrastructure stacks that are just about as easy for a customer to consume. But if they decide they don't like NetApp tomorrow, it's all open standards based. If they decide they want to switch to another server vendor, they can do that. And I think that flexibility, especially with uncertainty about workloads continuing to move to cloud, uncertainty about how AI is going to change everything, is for the most part, I think, scaring people away from hyperconvergence with maybe the exception of the very, very low end, sitting in some small shops somewhere. I think that's really where HCI is getting relegated these days.

All right. Well, that wraps up our session with Jeff. Thank you so much. I really appreciate you as always.

Thank you very much, Kris.

Thank you. All right. Okay. So now we actually have a couple of customers that you can ask questions of. So I'm going to invite Scott and Casey up to the stage. Hi, guys. Thanks again so much for doing this. We really appreciate it. This community doesn't always get to hear from actual customers, so I think it's really important to be able to get some real-world practitioner input in there. So I'm going to just start by asking you guys to introduce yourselves, your company, and what your experience with NetApp has been. So why don't we start with you, Casey?

I'm Casey Shenberger. I'm a cloud platform architect at Hyland Software. I've been there a long time. And our company, we use NetApp to host our products as well as internally. I work in the hosting side where people host our products in our own private cloud. We also do hosting in public cloud. So we use NetApp products. We use all-flash products for certain workloads. We have started using the C-Series NetApp products for other workloads that don't need as reduced latency. We use FAS systems in data centers where we don't have the performance requirements of A- or C-Series currently. We also use StorageGRID products to tier our data off to. And we are moving some of our workloads into the cloud with AWS. So there, we use NetApp's products called Cloud Volumes ONTAP and FSx for NetApp ONTAP as well. So we kind of use all of NetApp's products across the board.

All right. And Scott?

Wow, little volume. All right. So you can hear me but a little bit too much. I'll try to whisper. So Scott Brindamour, I work for Lumen. I'm the Vice President of Product Management. I have 4 different areas I cover. So our edge compute and cloud business that we provide, I can explain some of the solutions that we do there, NetApp is part of that business; I also have our data center strategy from connectivity as well as solutions on top of data centers -- AI is a big piece of that, we mentioned earlier; our hyperscaler strategy, so how we're going to market with co-innovating and co-selling with the big cloud providers; and they also have a wholesale business. So the legacy large pipes all over serving and working with joint network customers and partners going forward. So I've been with Lumen for 18 years. So if you've heard about Savvis Communications, that's where I started. Got acquired by CenturyLink, and then when Level 3 and CenturyLink merged together, we became Lumen. So NetApp has been an amazing partner. So you heard Jeff earlier talking about hybrid. Lumen is all about hybrid. Really, what we're really trying to work with NetApp on is we are being coined through our CEO and our executive team as the network for AI. There's been a lot of buzz around our stock, who's gone a little bit crazy lately with a number of the custom private fabric deals that have been sold, so to a lot of the cloud providers, social media platforms, et cetera. So in the billions, with billions coming, that they're looking for. So this real rush for AI is real. We want to be that network, that central nervous system that connects the people, data and applications together across wherever customers want to execute. One of the reasons why we're doing it with NetApp, we have a platform that's built on our network, which we call Lumen Network Storage, that combines -- the solution combines the power of the Lumen Network to connect the data to its destination that it needs to go to or where that data is originated. So great use case for AI. We're in the midst of creating a solution with NetApp around how we go to market together to connect the network and the data, which I think are the 2 most important parts to build AI platforms, to continue to invest in AI platforms and how you actually harness the data to use it and get customers on that journey. So -- that's why I'm here today. Happy to answer questions and talk a little bit more on our relationship with NetApp.

Yes, I'd be curious to hear from you sort of, as you did your evaluation work and you looked at across vendors, what were some of the key points that primarily led you to choose NetApp?

So I think flexibility. So we standardize on their ONTAP platform as well as Object Storage, StorageGRID. So the combination of those products to be able to meet -- so we wanted to create a solution by which customers don't have to think about what performance storage they need to match to the application and have a platform that's versatile that can support whatever hybrid infrastructure and applications they want to put on it, as well as connecting to the cloud on the premise and integrate with that seamlessly. So regardless of what -- we're [ Switzerland ]. We work with every provider and platform provider out there. We're trying to make our product so it supports whatever the customer journey wants to be. And that's it. So we've done from the highest financial trading applications, real-time data access in oil and gas, to a general file system customer down in mid-market. The platform scales to the size and the capabilities and the ability to automate the delivery of storage to a customer and even down to we can provide, for our larger enterprise customers, a go-to-market by which we do dedicated storage with them as well. So I think the flexibility of the hybrid approach to work with everybody, it was talked about earlier when I walked in, that's the message that we had. So very good alignment of how we serve the market, how we work with the cloud providers, how we work with the data center providers together. So there's a very symbiotic relationship there. They focus on what they know, which is storage and data and apps. We focus on the network, and we combine together to create solutions together, and that's really what we've been successful with NetApp as well.

We have a similar situation, except for I obviously work on the technical side. And it was a similar thing. We chose NetApp because of the unified interfaces that Jeff was talking about earlier. We have a very small staff to manage all the equipment that we have. So by using NetApp, we had high-performance tiers, mid-performance tiers, archival tiers, all of that was in the same platform, the same operating system, the same expertise is required. So we looked at other platforms and it meant we had to have one vendor for high performance, and we might have to have a different vendor for an archive tier. Well, that means that you have to have people who have knowledge in both or separate people with knowledge in one platform versus the other. NetApp allowed us to just have one group and one set of knowledge, so that was a big help for us.

Yes. And I would say I'm always having a storage vendor knock on my door trying to get a piece of our estate and our customers, and it's usually an easy conversation that we've got a platform that can pretty much do what we need to. And the other thing I'd add to, in listening, is the innovation, the ability to go to market and innovate and try new things and go to market together. We've done a lot of proof of concepts, a lot of prototypes, some of them do well. I used to run the innovation team before I moved into this role. They're very willing to jump in with us and learn together and understand where the market is going. So that's been big for us as well as we're trying to create the next big thing in the market. AI is a good example of that as well.

Maybe a bit more specific product level question. In terms of NetApp's recent launch of cloud block storage products, does it materially change your buying pattern with the company itself? And then maybe a second one, just in terms of utilizing the public cloud more over time, how do you see that engagement sort of changing in terms of are you using NetApp just because that's sort of the on-prem versus evaluating other sort of opportunities as well there?

Well, we've been using NetApp's Object Storage, StorageGRID for quite some time. And as we moved to the cloud, we still needed a way to do fabric pooling or some way to offload that. So we do use S3 there. But we chose to use NetApp Cloud Volumes ONTAP or FSx for NetApp ONTAP in AWS both, because we still got the deduplication, we still get compression. We still get the performance, and we have the same knowledge set, like I talked about. So we stuck with NetApp because that knowledge transferred maybe a little bit at the back end was different, but -- but ultimately the storage knowledge required to function there, and we still got all of our performance, we still got all of our compression and compaction and those things that help us to reduce our data footprint. So that's why we chose NetApp, right? They're leading that field, too. So we stuck with them in the cloud as well.

Any -- do either of you use the ASA products? Or do you use ONTAP in a block, like a unified block?

Yes. We use a unified block today. So, yes, I don't see any -- I don't think any changes. I mean we try to wrap it up as a solution to abstract the technology. It's flexible enough and hybrid enough, right, combined with Object and StorageGRID. I actually see the -- with now with -- we see all the data that's being created for not just AI, but IoT and data analytics that's being created. We're hearing a lot from customers that distributed is more what customers really looking for. Moving -- obviously, the cloud providers want you to move all of that into the cloud providers. More and more enterprises are taking a more distributed approach, so the ability to have a platform that can work with the cloud provider and the data and the apps you have there, but also connect to on-prem, dedicated and anywhere in between, especially with an AI model, where the model is going to be distributed, the data that you're feeding the model is going to be distributed. That's really where we think -- and NetApp is a perfect opportunity for us to scale, to deliver a footprint where we need it and scale up from there without having to change the platform, add capabilities as we go. XPBlue and that whole control plane and portal or ransomware and data classification has been gigantic for our customers as well that have access that at. We're working with them to productize that as a solution that we can give and provide to kind of as a multi-tenant access to customers as well. So they continue to add value on top of it, right? I see that they have the same vision that we do, that distributed storage as a service with solution capabilities, where we're not talking about the storage and the technology underlying -- obviously, the techno-geeks are speeds and feeds and are thinking about that -- but ultimately enable customers to solve a business problem with a combined solution on top of it, right? That's where I think NetApp and Lumen are really working very well together. Their vision is similar that we need to add and talk about the data and the value of the data and the outcomes you're trying to get from data, reducing the size, reducing your risk, your security profile, the ransomware features are great capabilities there, and accelerating what you're trying to do with your data to get value out of it for time to market. AI is just the newest version of that, right? So.

And just to clarify, when you say distributed, you're talking about like what NetApp would call hybrid cloud, right? Like on-prem, through the cloud.

Hybrid cloud, and we bring our -- what we consider to be edge is the network edge or the [ metro ] edge, which is the edge of our network where NetApp lives. It's in the network. It's part of the network. So any customer that's using our network can use NetApp as part of that as well as to serve -- we talk about it all the time, if you're a retail giant with 3,000 locations, to put storage on all those locations in a footprint to do automated checkout using computer vision cameras and image recognition can get pretty onerous and expensive to actually put all that gear all that storage on every premise location. And then by the time you finish doing it, which is usually a 3-year exercise, you have to upgrade the other ones, right? To be able to offer an alternative to that. Some of it on-prem makes sense, but there's OSHA regulations. If you have a small footprint for sound, there's complexity of supporting that across all those locations. So we believe that the [ metro ] location is what we call the third execution venue. Everybody talks about cloud and premise, but somewhere in the middle, that's the [ metro ] edge capability that NetApp lives, and we enable all those abilities to do distributed applications and data workloads with that as well. So very aligned to your hybrid cloud, and I put the and and the metro edge to add to that as well.

Maybe for both of you, both different types of businesses, but when you think about the importance of storage in your budgeting or your CapEx or however you want to look at it, can you talk a little bit about where that sits. I imagine it's a little bit more for Casey than for Scott. And then just -- there's obviously a lot of demand for data. How do you see your spend on data solutions over time tracking with the data, increase in data usage? You don't have infinite budget. So any insights you can offer on those.

It's key to our budget, right? We stay -- I mean, by hosting enterprise content, we ultimately store people's data, and we have to manage it, maintain it. So it's a key in our budgeting. The key there being these features we get from NetApp help us reduce that, right, compaction, compression, deduplication. They reduced that footprint to help us get down to a more budget-friendly answer to how that works. But it's always front of mind for me. Obviously, I manage the data in general, but it's very -- we sit down and we work very closely with our NetApp team to make sure that we're taking advantage of all the features that allow us to reduce that footprint but still maintain resiliency, reliability, availability. That's the key to us, don't reduce anything like that, don't reduce reliability or resiliency, allow to have ransomware protection, things like that, but continue to drive the cost down per gigabyte of storage or per terabyte of storage, however we're calculating it.

Yes. I would say that the data storage and what we're continuing to deploy and purchase is not slowing down anytime soon, it's escalating. But at the same time, I think the opportunity to optimize what you have in working with our customers, there's a huge optimization opportunity, lots of customers are keeping hold of lots of data that they should be archiving out. So having a solution like NetApp with Object Store to move it to a lower cost option, to optimize what they have on their dedicated arrays with NetApp as part of the solution, but also be able to have that tiered storage option that you can use what you need, rather than having an all -- an expensive high-performance all-flash array that is going to cost you a lot more per terabyte than a tiered solution by which you can lower your cost footprint as well. So us internally, I mean, with all the data assets and AI that we're doing in-house to understand inventory and customers and where we're deployed, the amount of data is going to continue to grow tremendously. We see it. We see the investments. We're moving a lot of applications and services to the cloud. We see it's exactly what our customers are doing as well. So tremendous opportunity to reduce so they can keep their budgets while they're expanding their storage as they go forward. You need to do a little bit of both. As you start to -- I think you're going to see that as customers start to adopt AI, they're going to try to make it take advantage of more of their data and more of their assets that they don't -- they're not collecting and they're not using today, it's going to continue to increase that growth as you go forward. So the optimization piece to optimize as much as you can and modernize that infrastructure going forward is huge as well as protecting that data as it grows and as you start to move, that key data that gives you those insights in an AI model around -- you need to make sure it's protected and not available to the bad guys, right? So.

Yes. All right. Well, you go. Okay, Wamsi.

Kris might not like this question. Maybe she would. But I guess if you were to give feedback on NetApp, like what would be things that from a product perspective you would say are things that maybe you're running into pushing the limits, so to speak, where maybe they could be more helpful, or maybe they already are and you're in discussions, but what would you say are some of the key things that you -- you would give feedback to them on?

Yes, for me is, as a solution guy and a product guy, I mean, but I've spent most of my career on the sales side, too. So I've kind of got multiple perspectives. But the biggest thing, and what I mentioned before, is less speeds and feeds in technology about storage arrays and their capabilities and adding and incrementing. A lot of what I used to hear was an update on the next array, the next technology, the next capability, which is great, I think, for my engineering team, but from a product and solution perspective and sales perspective, I have a bunch of network sellers that trying to understand everything that goes on in that space and then layer on top of the storage conversation. And the hardest part in the go-to-market has been melding those 2 expertises together to go after a customer jointly, where you both get a value prop and a benefit back to each other. So changing the conversation to a solution-focused conversation where you're focused on joint customers' problems that we both see in the market and then building solutions together to go after that problem, where we're both adding a piece of the solution and then integrating it together as a conversation where we're changing the conversation with customers. So now you're not just selling to the architect who is managing and developing the architecture and keeping it up to speed, but you're selling to the business owner who wants to take advantage of how we get value out of my data, right, going forward, and what do I need to enable that, what do you need to experiment with AI and GenAI models. I don't have the infrastructure. Can you partner with Lumen to give them the network and the infrastructure on top of NetApp that you can deliver a solution to. So -- that would be the big thing. It's -- the company has definitely changed its positioning and it's approach and is becoming more solution-oriented. And they may have done it to the market, but didn't go as quickly with the partners, as you would see. So as a customer of NetApp and a partner that we're going after the same customers, I've really seen that shift and like to see it more. As a person who's looking to create the next innovative solution in the AI space, they're leaning in, but it took a while before they got there. So that would be the big thing that I see.

From the technical side, because that's where I am, right, I think the biggest thing that we struggle with or that we continue to provide feedback to NetApp is, they need to continue to provide better ways for automation and scale. So, right, as a hosting provider we get lots of customers coming on. We're doing more and more, and we want to we want to have all that performance, reliability, resilience I've been talking about. But also we don't want to -- we want to maintain our staffing levels. We don't want to have more staff to do that work. So we rely on automation. And as that -- as NetApp makes these shifts and they start to transition, we need them to continue to give us the ability to do that in an automated way. There are some things there, here and there, that are tough to do that for. So we continue to work directly with NetApp and ask them, let's make it so this is much more scalable and much more automatable. That's our biggest feedback to NetApp, I think.

Well, let me ask the flip side of that question, which is, right, you've both used NetApp for a long time. What capabilities did we bring that surprised you or that you learned through actually using the product?

I think probably the one thing that was the most -- I guess that really caught us, and it wasn't necessarily a surprise, but the ARP capabilities, right, we were...

ARP is automated ransomware protection.

So that functionality kind of came around -- we had been talking about needing ransomware protection and how we were going to do that. There were some -- all kinds of different methods for that. And then NetApp started adding that directly on-box. And now you can, we can do that on-box automatically. It's all GenAI based, so it's very accurate. So I think that was one thing that kind of -- although not really a surprise, it was a big help to take away. We didn't have to go now look for a ransomware vendor that we could partner with, right? Our vendor just brought it along.

Not AARP, right? Yes, I would say the ransomware protection. So back to what I was saying that the shift to a solution mindset, security and ransomware is huge. The ability to actually sell that as an add-on service to our customer and security market, which we do a lot in the security market ourselves, was huge. Like that is an added value that I can monetize with my customers going forward. I think Cloud Blue in general, the data classification capabilities that's built in, the control plane to see all of your storage across your hybrid infrastructure. All of that, the [ want ] that was mentioned earlier, automation and visibility and control is gigantic. So that's -- as I said, that solution shift has happened recently, and I've been surprised at how quickly they've adopted new capabilities as well that are benefiting not just Lumen to another capability that we can sell and add value to our customers, but for our customers themselves that they made the right choice in NetApp, they continue to get value out of the platform as they go forward. So that's been gigantic.

So I know you both talked about the value of the unified storage approach. But the fact that we have kind of ONTAP underpinning everything is what enables us to bring those incremental features and have them broadly scaled. Audience questions. Otherwise, you're going to keep hearing from me. Lou?

So the AI session from earlier said a lot of small, medium businesses aren't really yet using AI internally, so I'm not sure how big your companies are, but are you all using AI internally yet? And if actually you think that your firms are just aren't big enough yet, where do you see possibly that happening in the future?

Yes. We're pretty big. Yes, I'd say, 50,000, 53,000 employees as a telecommunciations company that's trying to be a technology company, we're adopting. We're huge users on the Microsoft CoPilot side. So we're probably the poster child for Microsoft around Copilot, our CEO, Kate Johnson, came from Microsoft. So there's a connection there. But it's tremendously valuable in having data at your fingertips transcribing and summarizing meetings for me where I'm triple booked, like I'm here at a conference for the next few days, all the meetings that I'm missing, I can get access to it. But just the data deluge of data that's available, trying to find that content across multiple systems and e-mail and all the things that go on, is huge. So that's one example that we put in that we started small and then it's pretty much available to anybody. We thought the price tag originally was pretty high, and we limited it, but the value that we got out of that [ broader ] productivity from that solution is gigantic, living in a world of tremendous meetings, that's valuable to us of how you're going to optimize people's time. And there's a whole regular course internally of enablement and adoption of that, which has been huge. We've been actually trying, as a company who has infrastructure everywhere, every data center, every cloud, millions of buildings across the U.S. and in Asia-Pac as well -- we sold off our European assets recently -- the ability to understand those assets and using AI to not just trust the tech that installs something or turns up a circuit, the ability to have to use cameras to understand what's there, what the capacity is, what was installed, was installed the right place? So I got the nice blinky lights on, like inventory management and actually putting that against demand and bringing that data against sales force forecasts and demands on a particular location that may say, a tech that's installing something for one particular opportunity or customer can look at the demand and say, I'm not just going to bring 1 [ chassis ], I'm going to bring 2 to install that. So the ability to use the data for real time to train people of how to do particular jobs and capabilities, that's all automated, and to check their work after the fact has been gigantic. So I think throughout the process of understanding data and how to use it has been tremendous across the business.

And did you just develop that application, the last one, yourselves internally with possibly a partner?

Partner and you name it, we're working with pretty much everybody in NVIDIA and Intel as well as Microsoft and Oracle and some of the SIs as well are helping us adopt it. So it's been pretty much across the board. I mean, it's pretty incestuous relationship. We sell to them, they sell to us, balance of trade is huge. So there's always someone looking for us to do more consulting or do other things with them as well. But the opposite of it is partners that are willing to invest a lot with us to use this as a use case as well going forward, and we do the same with them. So it's been really amazing, I think, for a big company that has the capital to do it, but also capital is tight in these times. It's -- we think we've got a tremendous return, but we're learning what we need to do and how we need to go about it. I think it's very similar to what enterprises are. They're getting their feet wet and they're learning how hard it is and how they need to focus and where they're going to maximize the value.

Casey, any comments from you?

We -- I mean we have AI, like you said, about meetings and some Copilot stuff. But for -- really, for me, I'm not super involved because I'm hosting customer data. I'm managing that underlying infrastructure. So it would be more on like the software that we write and that we host. And that's -- Hyland is -- we actually have our CommunityLIVE this week. So we're going to have some new announcements about what we're doing with AI and where we're going there for -- to allow our customers to have like more seamless integration with AI to get access to the data that I store and manage. So today, I don't really use it a lot, but it's coming in our products, and it's coming in our cloud, and we will definitely embrace that.

Maybe we could get both a technical and a business perspective on this question a little bit. We do hear from some of NetApp's peers about how power has become an increasingly important consideration around storage. And so as you think about it, could you help us think through, maybe -- from you specifically, how much of your budget is actually power budget is being consumed by storage as a practitioner from a tech perspective? And where do you see that going? Is it different between all-flash versus disk? And from a business perspective, how important is it? I know you guys are, I think, bought, committed 10% of the fiber from Corning or something recently like that, right? Like that's a lot of fiber capacity building a ton of capacity out there. So as you're building these large data centers, is your storage strategy going to need to change? And also as power in that context, how do you think about that?

Yes. Power -- I mean power is a big part of our storage budget. And then like you said, it's directly related to the type of storage that we're using. So as we do more performance, and SSD is huge, right, that gives us more, more I/O and less power. Additionally, we get it in less space. It's why we've embraced using C-Series. We have plenty of workloads we don't need sub-millisecond latency for, but we do need still high I/O access and their latency tolerance needs to be low. That allows us to get a lot of capacity with a lot less power. And so -- and then because we're able to do that, and we use that with tiering, that same unified approach we've been talking about all day, that allows us to -- that archive tier maybe uses some more power, but it uses less because it's disk that's kind of just there and not accessed near as much. So in all these new things, right? We choose -- sometimes we choose the type of platform, with NetApp we put our workloads on based on power consumption or what kind of space they're going to use in a data center. So that's how we handle that piece.

Yes. Power is always a big cost in consideration, especially when you're distributing infrastructure everywhere. So I think I mentioned earlier about having a platform that's scalable and modular that can meet all the use cases without putting it -- the "build it and they will come" kind of days at Lumen are gone. So how we optimize what we think we need now and then how we can easily scale up has been a big piece of what we do, and optimizing the same footprint everywhere to make sure we have a kind of scalable, supportable power infrastructure, cooling infrastructure as well. Now we're getting into GPUs, things of that nature, they're at order of magnitudes over what the storage system, which is unified and supporting across all workloads. So I think from the storage perspective, we're pretty comfortable with the model we have. Dedicating it where customers need more power for customers. We've had big financial services enterprise customers that have deployed dedicated capacity in areas. And making sure we're starting small and building up as well, so that we're not consuming a bunch of power and not using has been kind of what we think about. But then going forward, as you distribute network capacity in compute and storage and security capabilities across the platform, it's becoming harder and harder that a lot of these destinations are old central offices where we had all the telephone gear, which is now in the network, which is now virtualized. So thinking about how you put it into low power. We're always looking to optimize the power of every device that we put in as well, as well as how can we contain -- the power goes straight with the heat as well, right? And the heat in some of those facilities to be able to cool them and do liquid cooling and some of those innovations as well that a lot of vendors are bringing that. So we've been talking to NetApp about how we optimize that as much as possible as well. But the modular systems approach helps with that tremendously. So we don't have to deploy something gigantic and then be consuming a lot of power, it's a cost we can't get back, right, because we're not optimizing that. So it's been the big piece of it going forward.

All right. Well, since the questions seem to have died down, I'll ask one final question, not necessarily specific to NetApp, but more around how you're thinking about the future of storage and your data infrastructures. What are your -- when you look on the horizon, what do you think the biggest things coming for both of you are in terms of opportunities or challenges?

I think opportunities that we've been really thinking about, composability of systems and composability of data and storage that I don't have to buy, I can assess, basically build and have a customer compose a system on demand that they need, whether it be compute, whether it be memory, whether it be storage or network backplane. So we've been -- in my innovation time, we've even looked at a number of vendors for around down that route as well. So how can I actually get integrated systems that have all of that together where I could build that modular on the fly. So we've talked to NetApp a little bit about that as well, not giving up any research and R&D that we've done, but I think that's a piece, is how do you deliver just enough storage and just enough compute, enough memory, to support the requirements of an application without prebuilding and building it and we'll wait until they come. But the capability, I think that's already started is the AI capability on top of that to understand the data, categorizing that data, understanding that data and making you aware or even automating some of the maintenance capabilities around that data, segregating the data off. Trying to work across the infrastructure layers and the stack, including storage, that our vision or my team's vision on the edge side is to create an environment where the application dictates the infrastructure that you need. So if you have composable components in a solution that you're delivering, including the storage, how can the application understand and predetermine how much storage, what type of storage, where it needs it, and understands the network and understands the compute and can build the system that it needs to support what it's doing now and then adjust as it goes forward. So kind of that autonomous system that's composable that you can build up. So how do we support that in a business model going forward with NetApp as our primary storage partner as well. So that's one of the things that we think about that we've been trying to do, we're probably early on in the technology, but that would really optimize where we're not putting a tremendous amount of infrastructure in one place or building a lot of racks in the data center, that we can really compose it. And just have just-in-time inventory, where I can swap out the latest, greatest performance of all-flash arrays with something else going forward as the technology leapfrogs. It's big with GPUs that I don't want to -- a GPU that cost me $60,000 sitting there not being used if I can afford it. It's locked into a server and a system. Same thing with storage. How do I unlock that and create systems that are modular and snap together as you need it, like Lego blocks? So that's what I think about.

I think for the future for us, it's probably very similar, right? With storage, storage is growing at an unbelievable rate, right? It's not getting any smaller. We need more analysis. We need more ability to know what that data is and where it's going. So for the future for us, it's just going to be to continue to optimize as we're storing that data, how do we optimize the space, the power, its resiliency, how do we optimize the intelligence that we have about the storage so that we can classify it properly and move it to the right location. And those are all things that we've continued -- we work with NetApp on and we work with other vendors, even for the compute side, the same thing, right? We need to optimize all the capacity that we're using. So that's the future for us. We just continue optimizing that as much as possible to reduce those footprints.

All right. Meta.

Just maybe on that, I know probably over the past couple of years, you went through a bunch of cloud optimization type of initiatives. Just is how you look at that optimization different kind of -- or is there any way in which it's different pre or post kind of some of that evaluation you might have gone over in the past couple of years?

I don't know if it's any -- I mean, it is somewhat different because the technology changed, right? Like before, in the past, it wasn't necessarily -- like, can I put it on SSD versus spinning disk? That was the only 2 options, right? But now we have single-level cell, we have multi-level cell, we have quad-level cell, right? We have lots of different SSD options. So that optimization now, and NetApp has grown that way too, right? They originally only had all-flash, and that was on single-level cell. Now they have the C-Series, right? So as we continue to work down that optimization as vendors bring multiple options, then that gives us the ability to kind of see -- so it's changed in that way. But it really hasn't changed from a -- we're still always looking a way to optimize that and put workloads where they belong.

Yes. I think more from a solution and product perspective that the cloud providers have almost abstracted a lot of the technology and gone through services, like what's the application approach, what's the service you actually need, and kind of abstracted the infrastructure away. So -- but it's also they made it easy. But at the same time, a lot of your eggs are in that basket, right? So I think that's where NetApp has helped us tremendously to kind of be that neutral platform. But I don't see it any different optimizing cost. I just always see the pendulum of people moving a lot to the cloud, trying to move it away from the cloud. I don't think that's ever going to change. It depends on where you are and your maturity as a company that you go through those patterns as you go forward. But again, simplifying and the ability to offer storage as a service, as a solution. That's kind of our vision of what we try to do. NetApp has been helping drive in that direction as well, but also be hybrid that they can participate with the cloud, they can participate on-prem, they can participate with us on the edge, wherever it may be, going forward and just abstracting that capability away in supporting the application and what it needs going forward. So that's the thing I have seen change a little bit with the cloud providers that have been leading in that generation. But there's another angle that you need as well is private and close to where your users are, et cetera, that the cloud is never going to get to. Maybe it will be one day, but we're not there yet. It's always going to be enough data and apps and performance that you need locally is -- but having that hybrid approach that supports whatever you need going forward is a huge benefit, right? But making it easy, like you don't have to be a storage expert to buy it and use it and automate it, right, so it can run itself, is going to be gigantic going forward.

All right. Last opportunity for a question, otherwise, I'll release Casey and Scott back out to talk with other customers. Going once, going twice. Okay. Well, Scott and Casey, thank you so much for coming. I cannot thank you enough. I will happily take that for you.

Thanks. Good job.

All right. Well, we're done. So thank you to everyone on the webcast. If anything piqued your interest here, please don't hesitate to reach out to the IR team. We're happy to get to any follow-on questions or connect after this event. For those of you who are here, lunch is outside, so thanks for staying for a late lunch with us. Again, there will be the keynotes later today at 4:30. And then the show floor is open after that. As Jeff mentioned, there are some announcements that you could expect to see in the coming days once we kick off Insight officially with the keynote today. So thank you again all for coming, and always don't hesitate to reach out to the IR team if you have any follow-up questions. Thanks. [Break]

Las Vegas, headliner stage. Welcome. DJ Graffiti, thank you, man, you sound amazing.

Thank you so much, Mario.

That's incredible. I'm Mario Armstrong back here with you again to start our final countdown to day 1. And I can feel the energy in this building and in this room right now. I know you can feel it too. And shout out to our live stream audience. We have live stream members that are watching us. They just joined the party. I bet you guys can feel this energy online as well, right? Everyone is getting powered up for what you're about to see on this very stage. In just a few moments, NetApp's CEO, George Kurian, will take the stage and lay it all out for you: the evolution of data and its growing role in our lives, the massive potential of generative AI and how NetApp, the intelligent data infrastructure company, will give you the confidence to win in this new era. And he's not doing it alone. Of course, he's going to bring some friends. He's got some amazing insightful guests to tell you more as well. In short, it's an N-powered hour that you simply cannot miss. But first, I've got a little bit of a different kind of power-up in store for you. I'm about to do something a little different from the headliner stage. I'm going off script, I think we should do VIP right now. Let's do VIP giveaway right this moment. We're about to upgrade two lucky members to VIP status at tomorrow's party, Encore at AREA15. That's right. You and a friend will get a chance to hang out with the Royal Machines, the world's finest rock supergroup, and it's up for grabs right now. This is what we're going to do. We're going to play a round of Powerball. Who's ready for some NetApp Powerball? You're ready? Let's do it. We're going to do it. Okay, here's the deal. We are going to randomly draw the names of two INSIGHT attendees, and each of these lucky individuals get VIP passes for themselves and a friend. It's that simple. So let's do this. Okay. And our winners, drum roll, please. [ Jacob Dickson and Rebecca Elliot ], congratulations. Big kudos to both of these folks. You are the two VIP winners from the Powerball from NetApp. Look for an e-mail with instructions on how you will get to meet the band. Enjoy the meet and greet and have a blast at the party while you're there. Make sure you take some selfies, so everybody knows it actually happened. Before we start our show, there's one more important note that I need to bring to your attention. Now the slide above says it all, but the point is this: please remember that some of what you're about to see on the headliner stage is vision content, and that means it's for informational purposes only, and the details are subject to change. Okay. Now that we got that out of the way, please take a moment. Silence your devices, so we can get the festival energy going on in here because we're all here. You came from all across the globe. You're now at the eye-popping venue. We've got the towering stage. We have the killer lineup. All the pieces are in place. NetApp INSIGHT, I've got to ask you something, though. Before I do, DJ Graffiti, drop that beat. NetApp INSIGHT, where does your power come from? What does it allow you to do? We know because we see its impact every single day. It fuels us, takes us to another level. It gives you the confidence to rise to every moment. Now is the time. We plug in and power up. Las Vegas INSIGHT 2024 is on, and it's N-powered. Yes, clap it up. Yes, absolutely, you can clap it up. Yes, fantastic. Let's kick this thing off. If there is anyone here who does not have a seat, please make your way to the Boulevard Ballroom. There is overflow seating for you there, just make your way to the Boulevard Ballroom, and you'll be able to see all the action as well. Las Vegas, NetApp INSIGHT. Are you all ready? Make some noise. Let me feel you. There we go. Look at all these beautiful people. You all look amazing. Look at this whole crew right here. I love what I'm seeing here. Welcome, one and all, to NetApp INSIGHT 2024, the premier tech festival of the year, happening right here. My name is Mario Armstrong, and I'm so excited to be back with you once again at the MGM Grand as we kick off INSIGHT 2024. Now for those of you who I have not met yet, by the way, let's fix that. Let's make sure that we meet. I'm an NBC technology correspondent, Emmy Award winner, columnist for Entrepreneur magazine, and I host a podcast called Wake Up and Level Up. Super excited to be here, and I've really been fascinated by the people that I've been meeting along the way. The first person I met was in the elevator from Austria, which was incredible. So all of you that have come from across the globe, let me hear you make some noise. And greetings to all the folks that are watching us on our live stream right now. Yes, that's right. Even those that can't be here with us in person will be able to still hear and enjoy all of these keynote presentations because they all are available to watch online. Along with that, you heard from Nick earlier, Snapshot, which is our live post-game show that's hosted by Nick Howell, with recaps and commentary on all the day's content. We'll talk more about that a little bit later. Right now, we're kicking off NetApp INSIGHT 2024 with a blockbuster keynote, a deep dive into winning in the era of data and intelligence. And there's no better person to take us on that journey than the CEO of NetApp himself. People of INSIGHT, please make a big welcome to the headliner stage for George Kurian.

Thank you. Thank you, Mario, and a warm welcome to all of you. Welcome to NetApp INSIGHT. I want to thank you for the great work that you do with us, the trust and commitment you place in us, and especially for the time you spend with us here and back where we do the work that's important to you. If you're here for the first time, I hope you have an amazing experience. We have so much innovation and customer experience to share with you, customers teaching you about how they use our technology and all of our technologists demonstrating the hard work that we've done over the past year. And if you're here again, having been here before, thank you for coming back. This is going to be the best INSIGHT ever. Before we discuss our strategy and innovation portfolio, I want to go back to last year and share a bit about what we told you then. You see, last year we told you that we were the data infrastructure company and that we were integrating intelligence into our data infrastructure. And what we told you about the world that we operated in was that there were three keys to success, that digital and data-driven leaders were expanding their competitive advantage relative to those that weren't. We said that to be a data-driven leader, you needed to have a cohesive data strategy and organization; operationally, you needed to treat data as a product; and technologically, you needed the foundations of a modern data architecture that was built on an intelligent data infrastructure, which allowed you the flexibility to transform the parts of your architecture that you needed to transform while having the benefits of integration and evolution in the parts of your architecture that needed continuity. And we told you that NetApp was delivering the intelligent data infrastructure, combining unified data storage for any type of data anywhere you choose to store it and access it by any method, block, file and object for any workload, truly unified data storage. We talked about the fact that we were investing in intelligent services, intelligence being built into the infrastructure and intelligence about the data that resides on that infrastructure, and you will hear enormous amounts at this conference about our progress in that dimension. And we talked about solutions that integrate these two fundamental underpinnings with the world's most important applications and platforms, giving you transformational flexibility and simplicity at scale for all of your data, for all of your infrastructure and all of the workloads that run on that infrastructure. So before I talk about the year ahead, let's talk about what has changed during the past year, because this has been a momentous year. Over the past year, we have seen enormous progress in the capability to understand and analyze data. Artificial intelligence made progress over many years step-wise, understanding handwriting recognition and then speech recognition and then image and object recognition. But over the past few years, we have seen a breathtaking acceleration in its capabilities. It is almost capable of being able to understand without human involvement the domains in which it operates, as well as to independently switch domains, and the pace of improvement is absolutely profound. Today, you are able to build rich, immersive experiences combining multimodal capabilities as well as, importantly, be able to understand all of the data that resides in your enterprise, not just the transactional data but all of the unstructured data, which is typically 85% to 90% of your enterprise. The conversations across your teams, the conversations with your employees, the design blueprints that are typically locked away in documents of various types, the underwriting and risk policies that are actually implemented in the way you serve your customers, all of that data can now be structured and analyzed using the tools at your disposal. We are now in the era of data and intelligence. So now let's talk about what has not changed since the last year. What has not changed is the importance of data. It's actually become even more important, as we will talk about, how it's managed, protected, governed and optimized, because data is the foundation for intelligence. Without data, there's no intelligence. And we are in the third stage of the modern era of data collection and analysis. The first stage was to simply digitize data production so that you could actually take the outputs of business processes and make them computerized records. The second era was to keep a longitudinal history of that data. For a single business process, you could track trends and history across timelines. The third era, matching the capabilities of the tools that are available, is about unifying your data, analyzing a wide scope of data so that you can get differential insights and to bring, for example, all of the data about your customers into one unified landscape, to bring all of the information about your suppliers and partners into one unified landscape. We talked about the importance of data as a product. It is even more true today. What has also not changed is NetApp's continued innovation at our fastest pace ever to help you build the intelligent data infrastructures that you need to thrive in the era of data and intelligence. This year, we broadened our all-flash storage portfolio with a new A-Series lineup, expanded C-Series and ASA products. We have significantly expanded the range of capabilities in cloud storage. We have woven intelligence for security and privacy and governance of your data into our infrastructure. And we have made it easy to consume all of this innovation the way you want. You see, at NetApp, data is at the heart of everything we do, because we are the intelligent data infrastructure company. So now let's talk about the work that we've done and continue to do and how it will help you in the days ahead. Let's talk a bit about how we got here in the first place, to this intersection point of data and intelligence. It began with the human interest in understanding our behavior and the behavior of populations around the world as well as the need to codify those human behaviors. You see the first large data set that allowed humans to understand how we behaved with each other was actually the census. And the first census was recorded in prehistoric times in Babylonia, well before the birth of Christ. The first modern census was implemented in Europe in the early 1700s to understand where populations were and to be able to tax them. Another important capability was the use of mathematical tools, now aided by computation, so that you could normalize and standardize all of this data, so that you could first understand human behavior, draw insight from it and then be able to predict the future. You see that the era of data and intelligence had its roots many, many years ago. And this has been a part of an enduring trend of progress in the amount of data that you can capture, in the tools that can be used to analyze that data and the ability to draw insight from them. And so what I will tell you is that we stand at the juncture of two profoundly important capabilities that allows us to turn disruption into opportunity. And where we stand today is that we need this technology more than ever before. You see the Nobel Prize-winning economist [ Michael ]'s productivity gains that we have been able to all experience across the globe because of the emerging market economies providing ample labor pools is going to be dissipated over the next decade. We also have aging populations in the advanced economies that form 75% of the worldwide GDP. So we desperately need a tool to drive productivity gains, a general purpose tool like artificial intelligence. And we are blessed that supporting that tool are the underlying capabilities, enormous growth of data and extraordinary progress in the pace of capability improvements of these tools. So let's hear about the opportunities that AI creates. The AI opportunity is recorded in enormous numbers because of the nature of the tools being powerful, being general purpose and being accessible by everybody using natural language interfaces. It is expected to transform -- deliver 50% automation and efficiency improvements for large classes of the work that we do. It is expected to transform the way we interact with each other. And more and more people are rapidly using these tools to derive data analysis and insights. A simple number will put it in perspective: $7 trillion of annual labor productivity improvements over the next decade. To put that in context, the worldwide GDP is about $100 trillion a year. The U.S. GDP is about $25 trillion a year. A 7% productivity gain is astoundingly large. And to do it, not only the largest corporations should be able to take advantage of AI and data, but also every school and every hospital and every human being in every part of the world needs to have fair and easy access to these capabilities. And what you will hear from us and others in the industry over the next few days is how we are making that possible, how we are working with the industry to remove the challenges to the adoption of this technology. And so I want to invite first, through video, we'll have one of the industry leaders who was responsible to help catalyze the AI revolution across so many markets, so many countries and so many parts of the global population. We have worked with Microsoft for many years to build joint solutions combining our technologies and the powerful tools that Microsoft has to offer. And I want to have you watch a video from Satya Nadella, CEO of Microsoft. Let's roll the video.

Thank you so much, George. It's great to join you at NetApp INSIGHT. We are in the midst of a massive AI platform shift. And it's the companies that are data-driven and AI-ready that will be best positioned to capture the immense opportunity ahead. That's why our long-standing partnership with NetApp is so important. And together, we have delivered industry-leading data services and helped thousands of customers move their most mission-critical workloads to Azure. And it's great to see how we are bringing NetApp solutions together with our AI platforms and tools, including Azure AI and Microsoft 365 Copilot, to help our mutual customers drive their AI transformation. Thank you very much for the partnership, and enjoy the rest of the event.

The work that we do to help you succeed could never be done by us alone. And it is with that humility and the recognition that we can bring unique capabilities and partner deeply with the world's leaders that allows us to stand here and guarantee you the benefits of your data with the world's best intelligence. Azure NetApp Files, the unrivaled work of co-engineering between NetApp and Microsoft, was generally available in May 2019. And now to mark the 5-year anniversary of Azure NetApp Files general availability, please welcome the President of Azure Core, Girish Bablani. Hi, Girish, welcome to NetApp INSIGHT. Thanks for joining us.

Thank you, and thanks to your team for the incredible partnership that we have had over the years. It's really a privilege to be here with you and your team, and it's a privilege to be here talking to our joint customers. I'm super excited about everything that we are doing together as we go forward in this era of AI.

We've worked with your teams for a long time, and we're always interested, what do you at Microsoft see as the benefits of the partnership with NetApp? And what do our mutual customers tell you are the benefits of the collaboration?

Yes. I'm super proud of Azure NetApp Files. Not only was it -- its kind of -- only kind of service when it was launched, but even today, it ends up being the largest and the most well-established NetApp service in the public cloud. And what our customers really have been using and continue to use Azure NetApp Files for is moving their mission-critical workloads into the cloud. In particular, when the workload is super demanding, like Oracle or SAP, Azure NetApp Files really, really kind of shines. Let me talk to you about one of our customers, joint customers, Coca-Cola Bottlers of Japan. So they have one of the largest SAP Db2 estates in the world, and they wanted to move it to the cloud. And they chose Azure NetApp Files. And they were able to do this migration in 5 months, just super quick, with 0 operational downtime. And 1 month into the migration, they saw 30% improvement in their TCO, 40% improvement in their application performance and a whopping 90% improvement in their DR time, so a huge success. And this is a good example of what all our customers tell us. When they want to see scalability, cost effectiveness, performance and a secure data regime, they use Azure NetApp Files.

Thank you. Awesome story. Let's pivot a little bit to the future. We're integrating our technologies with the range of Microsoft data and AI applications. How can our partnership help companies tap into the power of AI in this era of data and intelligence?

Yes. Like Satya said, AI is going to fundamentally change the way companies and we all do business. And so it's super important and urgent for companies to be able to exploit the power of AI. And what they need is an easy way to do it. So to be able to do this effectively, you need a performant, secure AI platform, a way to access your unstructured data of all types, all temperatures in a cost-effective way, and an application platform that can help you gain the insights from the data and realize the value to the customers. So the work that our companies are doing together actually brings all of this together. So we are integrating the whole Azure AI stack and the whole Azure application stack with Azure NetApp Files, ONTAP Cloud Volumes. And that basically lets our customers be able to use the leading models, have APIs to do training, inferencing and, of course, access to all that data that really powers the AI. Again, let me share a story of another customer. An automotive company is wanting to or has moved their autonomous driving workload onto Azure. Now they are doing simulation, inferencing and training. So when they were moving the workload for autonomous driving, it's a huge, huge data set. On top of that huge data set, they wanted to improve their deployment and also increase their development agility. And they chose Azure NetApp Files, and were able to move onto the Azure cloud by using the large volumes, which let them compress their data sets into the smallest number of volumes possible and achieving their goals of improving the agility as well as deployment time. So a great story, and I think all the customers here can also take advantage of the same thing.

Awesome story, combining data and intelligence, right? And so I'll just ask one final question, which is what part of our work together are you most looking forward to?

Yes. Just staying on the theme of AI, I'm super excited by what we can do together to bring AI services, copilots and data together. Personally, I'm a huge user and fan of the Microsoft Teams Copilot, which automatically takes notes, summarizes the meeting, action items, and for me personally is a huge productivity enhancer. So by the work that we do together, we can bring this capability to all our customers so that they can build intelligent AI assistants and copilots for their applications.

Awesome. Super excited. Girish, thank you for the partnership. Looking forward to another amazing year together.

Thank you.

Girish Bablani. We are super excited about the ability to bring the most sophisticated AI tools together with your data and to be able to help you transform disruption into opportunity. I want to take a minute and just share a little bit about what are the risks with AI. We talked about the opportunities. Now let's talk about the risks, right? Any powerful tool that's available to all of mankind has risks. Some of that's inherent in the tools and some of that is inherent in the way we humans use those tools. And AI has its risks. One of the risks is that when you group data and try to draw insight from it, there is the tendency to have bias or to have the wrongful conclusions. In fact, some of the underlying mathematical capabilities that forms the foundation of many of the advanced analytic tools, clustering and regression analysis was invented by a gentleman named Francis Galton, who is also, unfortunately, the creator of eugenics, the foundation of most modern racist ideologies. And so there is, inherent in the nature of the work that we do, some risks. Second, because data is important as the foundation for which AI operates, there are more people going after your data than ever before. If you look at the growth rates of data-focused attacks like ransomware, it is up 70% year-on-year, and over $1 billion of ransom have been paid out over the last year alone according to independent and analysts. And there are concerns about privacy and upcoming regulations. You see it's important that you are ready to deal with these risks, to manage them and to be ready to allow your customers to have confidence in the AI and data implementations that you run your business upon. And so we want to tell you a little bit about what it takes to win in this era. We talked about how we got here. We talked about the opportunities and the profound need that we have as a global population to embrace these technologies. We talked about some of the risks. So now let's talk about what it takes to win. When everybody has these enormously powerful tools available for use, using natural language, easy-to-use interfaces, what do you think are the keys to success? The first is your data and your data strategy and the way you provide access to it, your unique enterprise data, wide in scope, unified across time and types, high quality, well governed, first and most important. Second, the ability to use the AI tools with that well-managed data to draw insight. And the way that you are able to calibrate that is your deep understanding of your industry and the domains in which you operate. When everybody has the same level, base level of capability, your deep understanding of your industry is never more important. The third is an agile operating model to test, learn and adapt, so that you can pick a couple of domains and move these projects from the data science lab to production and to learn in return. And finally, a data ecosystem that complements your business ecosystem. We've always talked about the fact that you need complementary capabilities and value delivery systems in your business ecosystem. What you now need is complementary data capabilities and complementary data sources that enrich your data set so that you can apply the intelligence capabilities that are available for you, a data ecosystem to complement your business ecosystem. And so I will tell you more about what we do to help you in these areas of trying to win. NetApp is focused on the first area, your data and your data strategy and the tools that we can give you to make it easy to manage your data. But before I get into that, I want to tell you a little bit about how we at NetApp have made progress by unifying our data and how that has helped us to accelerate our business impact, our decision-making and the progress in our business. And I'm going to roll a video by Mike Berry, our Chief Financial Officer, telling you about NetApp's own data-driven transformation. Let's roll the video.

Hello, everyone. I'm Mike Berry, NetApp's Chief Financial Officer. It's great to see so many of you out in Vegas for our INSIGHT conference, and I hope you're having an awesome time. This week, you've heard about being data-driven and AI-ready can have an immense impact on your business. The productivity, the analysis and the innovation available to your team can all increase exponentially. I'm here today to tell you that I believe that's right on target. Like many of you, we have undergone a significant transformation at NetApp, and I want to tell you about what we did and the outcomes. When I joined NetApp several years ago, we were operating as three different silos in terms of our data reporting: our storage business, our cloud business and our Keystone business. This greatly impacted our operational efficiencies in a few ways. Number one, our unmanaged data stores meant people were getting different answers to the same questions. Number two, individual analysis of data meant we couldn't scale insights to the entire enterprise. And lastly, our legacy platforms couldn't process the increasing size of our data fast enough or support the new AI workloads. This clearly was not going to be sustainable as we continued to scale NetApp. So George tasked us to help drive better alignment across our organization and enforce standardization and integration. Today, we have one data source that informs our entire business. Everyone is making decisions from the same set of data. As I like to say, we have one version of the truth. And the impact has been huge. Latency has been halved. We get to the right insights faster. Data hops have been reduced by 70%. We have better quality and more compliant data across every aspect of our business. Redundant data duplication has been reduced by 65%. This really helps with having a single source of truth across the entire enterprise. And perhaps most importantly, we are more agile and can innovate with new technologies like AI much more quickly and bring you, our customers, what you need from us to drive your business forward. So from everyone at NetApp, we absolutely believe that helping you unify your data will be a game changer for your business. We hope this gives you some useful information as you leverage intelligent data infrastructure to chart your own data-driven and AI initiatives. Thank you, everyone, for your time and your support of NetApp, and I hope you have a great time in Vegas. Enjoy the rest of INSIGHT and best of luck on your data journey.

I can tell you the benefits that we've seen under Mike's leadership around our data-driven business transformation have been profound. We can make better decisions. We can understand our business more intimately than we've ever before. We talked about how NetApp is focused on helping you deal with your data strategy and change the challenges that you face with AI. So let's talk a little bit about what are the challenges that you might face as you embrace AI. You see there's been a lot of discussion in the industry about the capabilities of the algorithms that form the foundation of large language models, for example. And there are brilliant scientists and engineers working to make those algorithms much, much, much better. What we have observed conversely is that there's little attention being paid to the readiness of your data to be used with AI. And what we've also seen is that those that have well governed and managed data, like regulated industries and life sciences and parts of enterprises where the data has been, for historical reasons, better managed, are able to derive benefits and use large language models transformationally much, much quickly. So let's talk about the data challenges, because, to summarize, a big part of the AI challenge is a data challenge. How do I find, select and unify the data I need to build an application? How do I govern sensitive data so that an app that uses AI cannot provide the data in a way that's inappropriate for a given user or purpose? How do I ensure that my data is fresh and accurate so that my models don't drift because of stale data and so that my customers are always given the correct information using the latest-breaking changes? Data challenge is a big part of the AI challenge. The second is that AI is complicated, costly and time-consuming. And so let's talk a bit about that. Today, AI is being implemented by only some of the largest organizations, and it is being done as a silo, with specialized infrastructure and completely disconnected from the data systems and operational systems where all of the data and intelligence that forms the foundation of intelligence in your enterprise exists. And that is a familiar problem, because when we stood up here 10 years ago, cloud and on-premises data centers had the same chasm that exists today between your AI systems and your data systems. And what we told you then was that we at NetApp were going to work with the leaders in the industry to build a data fabric that makes the use of your data across all the public clouds possible and easy and seamless, and that we would integrate the public cloud into your IT architecture and operations so that you could take advantage of the flexibility, the agility, the innovation possibilities that exists on the leading public clouds while keeping your data secure and well managed. We told you that we would deliver the intelligent data fabric for the hybrid cloud era. And today, we are making you that same commitment, that very same commitment for the age of data and intelligence: that we will build you the intelligent data infrastructure that builds a bridge across the chasm that exists between your AI systems and your data systems that allows you to manage your data in the way that it is supposed to be managed so that you can leverage it for insight and competitive advantage. Today we are making that same promise to you, that over the next few years you will have with NetApp the best intelligent data infrastructure for AI, bar none. I promise you that. So you might ask us, so how do we plan to do that? And let's begin by talking about how a modern AI stack works. This is a picture of a canonical analytics and AI stack. Whether it's generative AI or advanced analytics or predictive AI, it all follows the same steps, which is you collect data from a variety of sources, then you spend time conditioning it, unifying it, normalizing it, building structure on top of unstructured data, and then feed it into algorithms. And you take the output of those algorithms and you either put a human in the loop check or another check, machine-based check, before you feed it into users that deliver on the missions of your organization. 80% of the time is spent in that bucket called data conditioning. And you might say, wow, why is it that you spend 80% of your time on data conditioning? And it is because if you double-click on that bucket, it is a very complex picture with multiple steps. And NetApp's vision is to help you simplify all of that dramatically by bringing AI to your data, wherever and however you want, in a way that is intelligent, agile, attainable and secure. And to do this, NetApp is delivering 3 transformative data innovations for AI. We will help you understand and manage your data for AI, bring AI to your data wherever and however and deliver the power of AI efficiently and securely for your business and your customers. And you might ask, so how are we doing that? We do this by eliminating data silos and giving you a unified structured view of your data assets so that you can easily explore, understand and prepare your data for AI, all while leveraging ONTAP's proven resiliency, multi-tenancy, security and robust data management features. Second, seamlessly deploy AI on data in place, on premises, on the public clouds or anywhere in between. We will enable you to confidently scale your infrastructure with cutting-edge performance, effortless scaling when you need it, and most effective cost and allow you to utilize your existing AI ecosystem tooling while leveraging NetApp AI features like integrated data versioning, model traceability and highly efficient data retrieval for training and inferencing. Then the third, leverage policy-based classification, governance and security that follow your data through the AI life cycle, not just once, but all the time. We will enable you to have automated AI data change detection and updates so that your models will always have the freshest data and the most accurate data anytime, anywhere. Super proud of the work that our team has done to build the best underlying infrastructure for AI. You will see a robust data engine for AI and intelligent data services for AI. And to just see that, let me introduce to you Krish Vitaldevara, Senior Vice President, responsible for NetApp's ONTAP Software Group and our AI technology portfolio. Please welcome Krish to the stage.

Thank you, George, and hello, everyone. How do you take this amazing vision George talked about and make it real? We do that by anchoring that vision in the deep understanding of the customer needs, and getting those customer insights baked into the vision. We talked to 1,000-plus customers. Many of you are in the room today. We talked and spent a lot of time deep diving with 100-plus customers. We talked to tens of AI analysts who spend all their time thinking about and covering advances in AI. We used all of that to build a picture of common customer AI environment. And believe me when I say it, it's a complex picture. In one training cycle, the data gets copied over 7-plus times. There are thousands of workflows, all of them copying, tracking the changes to the data, and 13-plus tools used across the board. The vector databases explode 8x to 10x as you go through this AI data and model life cycle. You need to maintain privacy and compliance and security throughout this life cycle. Our goal is to help you all drastically simplify this day-to-day complexity and significantly accelerate the desired outcomes. We want you all to walk out of this conference fully trusting NetApp to deliver and remove that complexity for you in those AI projects. So when we look across verticals, media and entertainment, financials, health care, EDA and more, the customers' asks of NetApp intelligent data infrastructure have been pretty consistent. They want an infrastructure that scales economically while letting them fully leverage the AI-accelerated compute. They want an integrated data engine that can make their data AI-ready and bring AI closer to the data, wherever and however they choose. The ability to access, move, process and manage this data in a secure, private and compliant way. They want this AI infrastructure to be inherently hybrid multi-cloud and is well integrated into the ecosystem so that they have the freedom to choose on their terms, right? It's all of you asking us that you want the freedom to choose on your terms. You have also asked us to provide you with an end-to-end solution that's so simple all of us can feel like we are AI specialists, right? Okay. Let's see how we can make that happen. So George just talked about how we plan to innovate in AI by solving the data problem that requires you to bust the data silos, that requires you to make your enterprise data AI-ready and that requires you to bring AI closer to the data, which means we have to innovate across the entire stack: infrastructure, data engine and intelligent data services. Today, I will give you a sneak peek into our vision on how that innovation across intelligent data services and the AI engine work hand in hand to take that complexity away. Tomorrow, I will go into detail on the best infrastructure for AI and the interplay of the data engine and the intelligent data services. Okay. So let's see that in action. First, we start with your data anywhere and everywhere it resides. You already know the value of ONTAP empowering your data centers around the world, on-premises or in any of our 1P native offerings in Microsoft or Amazon or Google. So we are building that unified global metadata name space. This will provide you with an easily synthesized, single unified global view of the data. You'll be able to put structure over this data, you'll be able to synthesize your data across all your data assets and have all of that metadata seamlessly brought together to create a complete view of your entire NetApp data assets. File or object, structured or unstructured, ONTAP or storage grid. So let me demonstrate what you can do when you bring that power to your fingertips. So most of you will recognize this as BlueXP, our hybrid multi-cloud control plane. We have pre-populated this with various untapped data sources. And we built the new Data Explorer capability directly within BlueXP so that you get to explore all your data using the simplicity and familiarity of BlueXP. Within the Data Explorer, you are able to see at a high level what working environments are being scanned. You get to see how many files, how many object streams, how much storage has been synthesized and what's your global metadata name space made of. you can go directly and start exploring and searching your data using natural language queries. For today's demo, let's put ourselves in the shoes of a data scientist in the health care industry. Here, you will see all the files related to the AI project that they are working on. Because we are using AI, we can actually go in and quickly find all the files that actually have content that is about or related to diabetes. Of course, you can look up by file names and strings, but the power of Explorer goes way beyond that. You can also set a variety of additional filters based on the metadata. In this case, we have set the filter to only include the files that have been accessed in the last 3 years. Let's double-click on an individual file and see what are the specifics of this metadata we are talking about. You will see common file information or file attributes like location and size. You will also see user attributes and system attributes. You can also use extended attributes. You will see information we have synthesized based on our data classifiers, like analyzing, identifying any PII information within the file. You can drill directly into the permissions of the file so that if there is any PII information, you get to choose based on your policy posture whether to include the file or not, right? This is incredibly important in determining which models and which inferencing workflows have access to what data. So you're taking your investments in security policy permissions all the way from bottom of the stack to the top and to the prompts. With the right permissions, you can see the preview of the file right in the Data Explorer. You will notice that there are some text fields in here that have PII data, and via a configurable policy you can automatically anonymize multiple of these fields that are displayed. Okay. So we have built a unified global name space with a powerful natural language Data Explorer on top. What's the tie to AI? Okay, let me show you a new trick. I'll go ahead and select the files that we have found through our natural language query and filters. Let's add these files into an existing or new collection. The data collection in the enterprise data are the data that's used by the models for inferencing prompts. This is how you make enterprise data available for retrieval augmented generation, or RAG, pipelines. This new data collection can be static, as in you're given a set of files, make that a collection, or even better dynamic, by using the query and filters to create a constantly updated collection of data. Given all the changes are tracked in our global metadata name space through technologies like SnapDiff, we can instantly update the RAG enterprise data collection as things change. Data and AI operations are expensive, compute, power and just the raw storage. So the ability to operate only on the differences is incredibly important. You want to do expensive data and AI operations only on the changes and not on all the files all the time. Okay, now that we have created our dynamic data collection, let's put it to good use. We are going to deploy this data collection directly to a NetApp AI data node or NetApp AIPod or your choice of any inferencing system. So let's choose one of NetApp AI data engine nodes to deploy this data collection to. Once deployed, we take care of everything. Remember, we are removing complexity here. You can hop right in and begin using it for data experiments. And if you're an API person like me, don't worry, we have you covered. We provide you with a rich set of APIs so that you can point this collection to any data science tool of your choice to access on demand. So with this kind of powerful natural language driven data discovery tools with the right APIs, we are transforming data discovery of our largest customers, including some of you in the room. Your enterprise data is now AI-ready and is integrated into your AI data pipelines with ease. Not only that, you can ensure that the data you're using for your AI applications is always accurate. Remember, George talked about the changes and keeping the data fresh and accurate. It's an incredibly big part of the value prop here. The security and policy posture of your enterprise data that you have spent decades building can be inherited and expanded as you go through your AI and model life cycle. It includes model training and tuning, right? So we will not only make your data enterprise AI-ready, but we'll also bring AI to your data by integrating all of these data smarts in place wherever the data is. Okay. So let me quickly recap. As George told you, we are delivering 3 transformative data innovations for AI. The first one is helping you effectively understand your enterprise data and letting you manage it by providing structure over unstructured data. The second one, we are bringing AI to your data by making your data AI-ready in place through inferencing. We are also delivering the power of AI efficiently and securely through ONTAP's proven data life cycle features. So tomorrow morning at the keynote, we'll take an even deeper dive into the product innovations directly related to our AI infrastructure vision. We also have deep dives, meet the expert sessions, customer testimonials and solution demos. So please take advantage of those. With our intelligent data infrastructure, NetApp is ready to help you on this transformative AI journey. So let's bring back George. Thank you.

Super excited all of the work that we've been doing around building the best data infrastructure for AI, integrating intelligence about your data and intelligence being built into your infrastructure. And tomorrow, you will hear how we're building the best distributed systems architecture that ever existed. So go to those technical sessions tomorrow. I want to thank Krish for his leadership of our AI journey so far, and you will see him more tomorrow. I think one of the things that Krish talked about is that you could apply these capabilities not just for the AI landscape, but for all of your unstructured data, whether it's for EDA workloads or whether it's for media and entertainment use cases or where you want to be ready for the era of data and intelligence by unifying all of your data, structured and unstructured, wide in scope, unified in type and time. One of the things that Krish talked about is that having intelligence about your data allows you to be ready for the era of AI, and one of the things that I shared with you is that we have observed that industries and teams that are particularly good at structuring their data and managing it effectively are well positioned for the use of the transformational power of generative AI and AI in general. One of those industries that we have been profoundly humbled to be a part of is the pharmaceutical and life sciences industry, where the history of building disciplined data management allows them to apply the tools of artificial intelligence today to drive differentiated insight and to make the world a better place. Johnson & Johnson is an organization redefining health care, connecting the best of health and care for every provider, every patient, everyone. They're leveraging state-of-the-art rigorous science, data and technology to address the most complex diseases of our time, including oncology, immunology, neuroscience and others. We're going to talk to one of their leading practitioners, one of those industry transformers herself. But before we do that, let's roll a quick video. [Presentation]

Please welcome Director of Data Science at Johnson & Johnson, one of those real innovators, a practitioner and a transformer who has experience with machine learning and generative AI, large language models and graph databases and any other fancy term you might use. She combines these capabilities with data and clinical trials to accelerate drug discovery and make lives better. Please welcome Monica Jain.

Monica, welcome to NetApp INSIGHT. Thank you for the work that you've done with us and for the work that you do to make the human condition better. Tell us a bit about your role at Johnson & Johnson and how J&J has been using AI? And what's your future vision for AI?

Sure. So as the Director of R&D Data Science team at J&J, I lead our data strategy and GenAI initiatives. My role specifically involves driving the integration of advanced AI and data science methodologies to transform how we manage and utilize the data across the organization. You asked about how at J&J we have been leveraging AI. So it's to enhance our decision-making, optimizing R&D processes and accelerate data discovery and development, right? So how AI is at the core of our efforts to bring innovative solutions to the patient faster is where we are looking forward to at J&J.

We are super excited at the possibilities for personalized medicine and the ability to build really effective clinical trial processes and methodologies because we know we live in a world that needs that. What challenges do you face when optimizing your data for AI, and how have you tried to solve them?

Yes. So optimizing our data for AI specifically presents several challenges, right from -- particularly around the data quality, integration of those different data sources, accessibility and especially governance. Being in the healthcare industry, it is very important, right? At J&J, we deal with diverse data set coming from different -- multiple different sources, including clinical trials, research studies and real-world data, right? So our primary challenge, how to ensure that the data is clean, standardized and ready for AI consumption, it's an essential piece when we are talking about AI. Poor data quality can lead into significantly impact the performance of all of our AI models, right, leading to inaccurate outcomes, which can impact our patients' lives, right? So it is very important to have the standard and clean data. Another key challenge certainly is the data silos. So we have data coming from different departments, systems. Integrating it to the unified, accessible format of AI applications can be very complex, especially in the R&D world, right? Additionally, what we are also facing the challenge around privacy, data privacy and security, especially in a highly regulated industry like any industry, health care industry, that adds another layer of complexity. So which is like to address these challenges we are trying to implement hybrid solutions, which is hybrid cloud strategy that allows us to leverage the strength of both, on-prem and cloud environments.

It's amazing for an industry where the data quality is so important and the privacy and security is so important, I'm super impressed with the fact that you'll have thought about a hybrid architecture so that you can accelerate innovation. How does hybrid cloud play a role in your AI infrastructure strategy?

Our strategy for AI infrastructure at Johnson & Johnson is centered around flexibility, scalability and especially security, which are crucial for supporting the diverse and evolving data needs for the data scientists and AI applications. Cost is another area where we are focusing on keeping our infrastructure divided, which basically we have adopted a hybrid cloud approach as a backbone of our AI infrastructure, allowing us to harness the benefits of both, right, private and public clouds. This strategy also is enabling us to optimize resources, like manage cost, as what I'm saying, manage cost effectively, because in cloud it can go uncontrolled, and ensure data compliance while maintaining the agility to how we can scale AI initiative very quickly, right? This -- the whole hybrid cloud model plays a very pivotal role in our strategy by providing the flexibility to deploy our workloads in the more suitable environment as different use cases may require.

We've talked about the world being hybrid for a long time, and AI seems to be the quintessential use case that drives hybrid adoption. Because of the pace of innovation of applications, you don't want to fall behind, right? And so how do you see NetApp helping you with these decisions as you think about what do you want to keep on-prem, what do you want to keep in the cloud, the flexibility that we enable you to have.

Yes. I mean before that, where we can actually reach out to NetApp, our -- the hybrid cloud strategy at Johnson & Johnson, we are also trying to evaluate by ourselves that what we should keep on-prem, what are the data and workloads, to determine the optimal placement between on-prem and the cloud environment. That's where I think we are taking help of NetApp for sure to identify what -- how we can secure our data. For example, a highly sensitive data such as clinical trial information, patient records, for example, and any other critical data subject to stringent regulatory requirements, we prefer to keep it on-premise, right? And this approach allow us to maintain very tight control over security. Compliance and data governance is the key piece, ensuring that we meet all our industry standards and protect our patients' privacy. On the other hand, that we are moving less sensitive, high volume and compute-intensive workloads to the cloud. So this includes our data used in large-scale AI to model training and analytics and machine learning experiments where the scalability, flexibility and our advanced computational power, the cloud provides significant advantage. NetApp, we are utilizing NetApp to help certainly to manage cost to have divided loads over here and there. Data mobility is the key piece is what we are, but there are multiple such solutions we are trying to look into.

Awesome. It's so gratifying to see the work that we started a long time ago with this belief that you want to unify and manage your data and make your data infrastructure hybrid cloud and well integrated is coming to play out for all of you. So thank you for your insights. Thanks for the great work you do and for the help you've given us to learn more about your business and to be able to innovate on your behalf. Thank you very much.

Thank you.

We are so humbled by the work that we are able to help organizations like Johnson & Johnson do. And we would never be able to do that without the humility to learn from them and the willingness to partner with others that help build a better solution for the customer. Monica mentioned the use of NetApp services on the cloud to help her accomplish her data-driven AI outcomes. And so let me introduce another guest who knows a thing or two about cloud and AI. Please welcome Matt Wood, Vice President of Artificial Intelligence products, Amazon Web Services. Welcome to NetApp INSIGHT. I remember the first NetApp presence at AWS re:Invent in 2012 where I was at the booth, and we were so excited to see how the partnership between us and AWS could help really solve customer problems. We've spent a lot of time working with you all around a variety of capabilities, and 3 years ago we announced the general availability of Amazon's FSx for NetApp ONTAP service. AWS has a comprehensive AI strategy all the way from chips to applications. Could you please describe for our clients what are the most important AI use cases you see people taking on and your strategy for AI?

Yes, for sure. Thank you so much for having me. I really appreciate it. I was there in 2012 as well. I thought I was -- I actually walked past your booth, so thank you. It's a pleasure to see you all. Good evening, everybody. It's my opinion and I think I probably shared with a lot of you that generative AI is probably the single largest shift in how we're going to interact with data and information and each other, probably since the advent of the very, very earliest Internet. And organizations that invested in the early Internet like Amazon.com 30 years ago, I went on to experience pretty tremendous growth in the intervening 30 years. And it's my guess that all of you that are investing in generative AI today are going to experience multiple Amazons of growth in the next 30 years, maybe even sooner. The big use cases that we see kind of fit into 3 main buckets. The first is those that really want to get into the weeds of building and optimizing their own large language models, machine learning models and artificial intelligence models. And there, we make available a pretty broad set of accelerated chips, some from our partners like NVIDIA, some that we built ourselves at Amazon called Trainium and Inferentia, designed to accelerate the inner loop of machine learning for training and for predictions and to deliver the by far the lowest cost per dollar and the lowest cost per watt for machine learning training on AWS. And that's really where part of FSx comes in. You really want to be able to get the data onto those chips, onto those accelerators as quickly as possible to be able to build these remarkable models and to be able to use them with your own data when you're running predictions against them. So that's deep in the weeds for the data science teams. We also make available a service that we call Amazon Bedrock, which makes available a range of different pretrained models such as anybody can pick up a model and start building it inside their own applications going forward. So we believe that there is no one model to rule them all. So we have a range of 32 different foundation models. We make it super easy to evaluate them and pick your best -- pick the one that works well for your organization and for your use case. And the goal there is to really be able to fit your use case against the model. And there you can lean into the model's advantage, whether it is price performance or latency or intelligence, and find a mix and match the right model for your mission. And then finally, we make available a set of applications for those who don't want to futz around with models, they just want to be able to use the application with their knowledge workers, with their support teams, whatever it might be. And there you can just turn them on and start using them with your own data very quickly and very, very easily. So those are the 3 main use cases that we see.

We launched Amazon FSx for NetApp ONTAP in September '21, and we are seeing accelerating joint innovation, service growth and records for the service each quarter. What do customers tell you about why they choose our joint solutions?

Yes, it's a great question. I think the number one thing we hear time and again is just security. There is a lot of customers, when we're working with them, when they start along the generative AI journey, they have this schism in their mind, and I can understand where it comes from, that in order to be successful with generative AI, they have to make some negative trade-offs as it pertains to the confidentiality and security and privacy of their data. And what we've found is they're working with FSx and ONTAP and the way that we're doing it at AWS, that just isn't the case. We do not use any of the data that flows through our paid AI services to improve the underlying models and you get full visibility over what data gets used where. And so it's just a very different model from some other folks. And so security is job zero for us at Amazon. It's job zero for all of you. And so that is really where it kind of begins and ends for a lot of folks. We also see just the raw performance being incredibly important, both for model training, when you want to be able to get very, very large data sets onto the accelerators to train the models, but increasingly for inference. As these workloads start to move into -- from prototypes into production, the majority of the workloads actually start to move into inference. You may train your models once a month, but you're going to run tens of thousands or even millions of predictions against those models every single day. And so having the right way of getting the right information with the right security with the right performance characteristics into those models to start working with them, super, super important. And then building on that, the final piece is just resilience. These workloads, as they start to move from early experiments and toys that people get very excited about to workloads that are proving to be very, very useful, the move to the very center of a lot of organizations, like we were hearing it from J&J, resiliency of those workloads is paramount because they're going to sit at the very core of the business. So those 3 pieces are what we...

Awesome. Couldn't have a better endorsement of our technology. Earlier this year, AWS published a framework for integrating FSx for NetApp ONTAP with Bedrock, Amazon Bedrock. Can you tell us how these new capabilities they could take advantage of that?

Yes, for sure. I think there's -- most customers that we speak to. They're moving very deliberately down their generative AI journey. They're starting with early experiments, they're moving to prototypes, moving into production. There are some customers that I've spoken to that have gone all in on AI. They've retasked thousands of their engineers over a weekend to focus exclusively on generative AI, moving very, very quickly. But without having to go all in, there is a group that is moving kind of faster than the average, and counterintuitively, it's actually the regulated industries -- financial services, insurance, health care, life sciences, manufacturing, those sorts of areas -- where the compliance with those regulations that maybe felt like a bit of a headwind at times for those organizations have actually driven the right set of behaviors to be successful with generative AI quite quickly. It's all around the unsexy stuff of governance -- I think it's pretty sexy -- but the governance, privacy, the policies, all of those parts where you know what quality data you have, who can use it, what services it can be used on and what it can be used for. Putting all that in place has already been done by a lot of these organizations. And as a result, the generative AI piece is actually a relatively small incremental lift and they're able to move much more quickly as a result. In addition to that, as I'm sure many of you in this room, these organizations have absolutely huge volumes, petabytes, if not exabytes, of privately held text data. This is information which the models have never seen before and which is clinical trial reports, market research, financial reports, all those sort of things. And generative AI today is really great at looking inside that information and starting to find correlation, find areas which are similar, create summaries and, almost as important, find areas where there's disagreement. And so being able to run those workloads quickly and easily means you can move very, very quickly. And as a result, most of these organizations that have had to sit on the sidelines a little bit with their digital transformation journey as they've been looking at other industries like media streaming and hospitality and travel that have been completely revolutionized with the Internet, they're looking at generative AI to not just catch up but to actually leapfrog ahead.

You've had a couple of really important points there, Matt. Is that your call to action, get your data ready? We've got people that in our audience that are responsible for the unstructured data assets of most of the world's large organizations. And if I -- if you had one call to action for them to get ready for this transformative set of possibilities, what would you say to them?

Yes, for sure. I think number one is to just get your data in order. The way in which these models operate mean that you can take all of the data that you already have and start extracting some of the latent value from that information. There's low-hanging fruit across all of that data, and to be able to very, very quickly and safely and responsibly start to inspect that information in new ways is going to open up just myriad new opportunities going forwards.

Awesome. Thank you so much. Thanks for the work that we do. Thanks for coming to NetApp INSIGHT. Thank you, Matt Wood. You heard from us about the era of data and intelligence. We said that this has been hundreds of years in the making, but we stand today with unprecedented possibility because of the enormous progress in technology and data to help improve the human condition, to satisfy our ever yearning for understanding, for insight and the possibility of predicting the future. We said that the keys to success in this era are your data and data strategy, your unique enterprise data, wide in scope, unified across time and types, high-quality, well organized, well governed, deep domain insights, knowledge of your industry, your functional domain, so that you can combine the possibilities of machine intelligence and your data in the most relevant way to make your business and organization successful. We talked about the operating model of being able to treat data as a product and be able to test, learn and adapt quickly and the importance of building a data ecosystem to complement your business ecosystem. We said that at its core, the main challenge with AI is a data challenge. And the fact that there is a deep chasm between siloed AI systems and enterprise data systems. And you heard that not only from us, but from all of our guests today. And we talked about the fact that, just like we did in cloud, NetApp will bridge the chasm that exists between your AI systems and your enterprise data by bringing AI to your data. We are delivering the intelligent data infrastructure for the age of data and intelligence. This helps you bring to your data, wherever and however you want in a way that is agile, attainable and secure. And to do this, we are delivering 3 transformative data innovations for AI: effectively understand and manage your data for AI; bring AI to your data, wherever and however you want; and deliver the power of AI efficiently and securely for your business and outcomes. These innovations have been the work that we have pursued for many years because we have always believed in the importance of unifying your data for competitive advantage. Rather than keeping it in silos, we have always believed in the possibility of connecting your data with the innovation capabilities wherever and whenever they may exist, and you should have that expectation of any of your data infrastructure providers, that they might build the roads and bridges to take your data to the places that you want them to be taken. About innovation, the awesome innovations we are delivering across every single aspect of our intelligent data infrastructure portfolio, from unified data storage to intelligent services to solutions for enterprise workloads and AI workloads on-premises and in the cloud or anywhere in between. And on day 3, Cesar Cernuda, our President, together with leaders from our go-to-market organization and partners, will tell you about how we are bringing enablement, capability and skills to enable you, our customers, to be successful. Don't miss it. I hope you have a wonderful NetApp INSIGHT. Together, we commence the journey into the era of data and intelligence. Thank you for being here. God bless.