NetApp: Architecting Outcomes in the Era of Intelligence | Jeff Baxter, NetApp

Clips
More from NetApp: Architecting Outcomes in the Era of Intelligence

Jeff Baxter

Vice President, Product Marketing

NetApp

play_circle_outline NetApp's presence with hyperscalers like Azure, Amazon, and Google Cloud

play_circle_outline Value of NetApp's cloud transformation and AI focus

play_circle_outline Adoption of NetApp by customers for enterprise file storage, performance, and ease of use

play_circle_outline Expansion of AI infrastructure to smaller enterprises and edge computing

play_circle_outline Collaboration with NVIDIA to bring AI data platform capabilities into NetApp's infrastructure

Info
Transcript

Jeff Baxter, NetApp

Jeff Baxter

Vice President, Product Marketing NetApp

At NetApp’s “Architecting Outcomes in the Era of Intelligence” event, theCUBE Research’s Rob Strechay sits down with Jeff Baxter, vice president of product marketing at NetApp, to discuss how the company is shaping the future of cloud and AI. Baxter brings a wealth of perspective, reflecting on NetApp’s product evolution and customer-first innovation strategies.

Baxter highlights how NetApp’s tight integration with Microsoft Azure, Amazon FSx and Google Cloud sets the company apart as the only vendor offering native services across all three hyperscale... Read more

explore Keep Exploring

What recent technological advancements has NetApp focused on, and how has this helped them stand out in the industry? add

What are some reasons for NetApp's success in being natively available in all three of the world's largest clouds, and why are customers gravitating towards their simplicity and flexibility in data management? add

What are the reasons people choose to use NetApp on the cloud, and how do they typically discover or get guided towards it? add

What are some examples of how AI is being utilized at the edge of networks? add

What are NetApp's plans in relation to working with NVIDIA on the AI data platform? add

bolt Powered by CUBE AI

Jeff Baxter, NetApp

search

Rob Strechay

>> Hello and welcome back to NetApp Converge 2025. We're kicking off the afternoon getting into the groove here, and I'm joined by Jeff Baxter, who's the VP of product marketing for NetApp. Welcome back to the show.

Jeff Baxter

>> Thank you.

Rob Strechay

>> I always enjoy our conversations. I think, again, there's so much goodness going on here and I think, again, we were talking and it's rare when I go to these types of events where I see people just so overly energized coming out of keynotes and the message and all the alignment, and I think the growth pattern really is one of those things that people are big time behind with NetApp. One of those things that we haven't touched on at all really today has been kind of what's going on with NetApp and the hyperscalers and NetApp cloud. You've got a bunch of them. Why don't you give us an update on where things stand here?

Jeff Baxter

>> Yeah, so I think... And so again, thank you for having me, and this is always my favorite time of year. I joke, "I have my two holidays." I have NetApp Converge, our sales kickoff, and then NetApp Inside our user conference. And so this is my 17th sales kickoff or something, and it's a little bit like coming home. I started in the sales organization and now I'm in the product organization. But I think the cool thing is, even having been here that long, we're constantly reinventing ourselves. And so I think the most recent reinvention has been all about the value we're bringing to ai and right before that and continuing on was the value we're bringing to cloud and that cloud transformation, they're so inexorably linked these days, so it's rare in the tech industry and this where you achieve a true differentiator and have it last for five years longer. Typically, if you pick up on something that the market and your customers all enjoy, someone is going to grab that within a year or two and come along. We are still the only vendor out there that is natively in all three of the world's largest clouds. So when starting with Microsoft Azure over five years ago with Azure NetApp files, then Amazon came to us and we did Amazon FSx for NetApp ONTAP, and then after that moving forward with Google Cloud, NetApp volumes, it's a testament, I think to the strength of NetApp. It's a testament to how much our customers wanted to see our presence on those clouds and those clouds recognize us and how well we partner and how portable our technology is to allow us to partner.

Rob Strechay

>> And I think to that point, you're right. When we look across and being on the, putting my little analyst hat on and looking across, it was one of those things. I was at AWS when FSx, NetApp ONTAP launched over there, and I think one of the things I saw was that everybody wanted to have their software-defined version of their array in the cloud, but really it was the simplicity that NetApp has always been known for. Are you seeing that really customers are gravitating to that because to me, and we'll get to the AI thing because I think they're definitely linked, and I think a part of the linkage is it's the same replication. I can do snaps, I can do all of what I need to do to move things back and forth between on-premise and cloud. And we were talking about it earlier, about 85% of the data that customers want to use in AI is still on-prem.

Jeff Baxter

>> Yes, absolutely.

Rob Strechay

>> Probably in NetApp.

Jeff Baxter

>> Absolutely.

Rob Strechay

>> Are you seeing this as a big driver for some of that as well?

Jeff Baxter

>> It's interesting because there's really... There's many reasons people choose to use NetApp on cloud, but you can kind of almost divide into two camps, right? There's the customers who know NetApp, who use NetApp on-prem, who are drawn to that and want to be able to use it across. And then literally there's the customers who are just looking for a fully-featured enterprise file storage, highly performant on the cloud, and they actually get guided that way by Microsoft or by Amazon saying, "Hey, you should try this out." Or they self-discover it, because we're actually a part of their service. So if you go on the AWS console or you go on the Azure portal or you go into Google Cloud, we're there just like any of their other services. And so they can test it, they can kick the tires, they can try it out, and we have a lot of customers come across it that way. Actually, I'll tell you a real quick story, if I could. My favorite story, I was at AWS Reinvent, not this last one, but the previous one. And I actually got to go as a regular attendee. I had my NetApp duties, but I got a regular attendee badge and I was going to sessions and I went to one that was about storage best practices for databases on AWS, knowing nothing about the session. I walked in, I sit down, it's a hands-on lab, and the person comes up to the front and says, "Okay, we're going to go ahead and set up this database, and to do that, we're going to use something called Amazon FSx for NetApp ONTAP as a backend storage." And the guy next to me turns to me and goes, "Have you ever heard of this NetApp company?" And I'm like, "Nope, never," have covered up my badge. And I didn't go there... It was not a NetApp session, it was an Amazon session. It wasn't advertised. It was about NetApp and for the person from Amazon talking about it was just another Amazon product because it is just another Amazon product. And that to me is the true differentiator is yes, you have the people come in wanting it because it's NetApp and then you have the people coming in just because it's amazing and just for the technical differentiation and the value proposition they get out of it.

Rob Strechay

>> And I think, again, when we talk to end users and organizations, a lot of what they also look at is the fact that from an ROI perspective, it helps them as well. Not only can they draw down on their cloud spend and get to their numbers that they need to with that, but also it's the fact that you're not, to your point skillset wise, you're not learning a new skillset and things like that as well. And all of the tooling that they've been using works there as well. We've been seeing that, and I think especially with some of the more cloud native types of workloads, we were just at KubeCon a couple months ago and it was like we were talking about how people are looking at this because of how do, I may want to want to POC my AI in a cloud, but then I want to run it and do inference out at the edge or something like that. And if I'm doing it out at the edge, I'm not necessarily running it on cloud at the edge, I'm running it on a system out at the edge. Are you seeing patterns start to develop around deployment models? And I know you had some announcements this week as well in that space.

Jeff Baxter

>> Yeah, we're definitely seeing... Data's everywhere. Data's at the edge. AI is moving towards where the data is, so definitely, we've had announcements in the AI space about taking our AI pod and moving it farther out into smaller enterprises, even farther out towards the edge. I don't know if I'd say it's all the way on cell phone towers, but it's starting to get... I think fog computing a little bit, people call it, right? Where it's not necessarily all the processing up the edge, but it's definitely happening closer to the edge. Not everything is coming back central. And so the idea that we can have these large-scale massive AI pods that can be hooked up to super pods, that can be hooked up to base pods for large language model training or more intensive and model training, but then we can go all the way out and do inferencing and rag and do things on much lower end sort of AI pod minis that we announced with Intel this week. It just kind of shows the extensibility of the NetApp infrastructure that we can do that across the board from massive installations down to just these sort of on the edge or very close to the enterprise installations.

Rob Strechay

>> Yeah we see it in the AI Pod Mini, I think hits a nice sweet spot because what we're seeing a lot is that people are going towards and leaning towards small language models or SLMs instead of, and we talked about this, that the long tail, you have the curve that we looked at the distribution of, you have a lot of stuff in those big pods, and then there's a long tail where there's just a massive amount of pent-up demand for the smaller language models and things of that, and that the size of the models doesn't necessarily have to be that great when you're using it for special purpose. It's very focused and things like that.

Jeff Baxter

>> If you're trying to answer any question that could be posed to you, those trillions of parameter models are great for general purpose usage. And so if it's ChatGPT or anything like that, yes, you want very large language models, but if you're, especially in the era where we're moving more and more towards agentic AI and agentic AI, that's not just to answer questions, but to actually do things. It's like taking a factory worker who their entire job is to ratchet one bolt and training them on the Encyclopedia Britannica. You don't need that to do this, right? Training them on the basics of a small language model just so that they can interface and then giving them more routine instruction. That's really what we're seeing, these sort of smaller training instantiations in enterprise starting to be. They still will use large language models, but they'll use it more as a service and they'll use things from one of the cloud providers, and then they'll add onto and use things that aren't retrieval augmented generation there. Otherwise, they'll build their own small language models

Rob Strechay

>> With the AI Pod Mini. Where do you see the sweet spot for that from a use case perspective or a customer perspective in that?

Jeff Baxter

>> It's really about those customers that they're not necessarily looking at things as having hundreds of GPUs that they need to keep satisfied. In fact, we often look at it and say, there's a lot of things now especially sort of in the inferencing and different things like that where they're not necessarily GPU bound workloads. And that's one of the reasons the AI pod mini, we did it with Intel, that idea that you can satisfy those out of still a set of beefy sort of reference architecture servers, but you can do that at really sort of the mid-sized enterprise, maybe some of the larger enterprises, the people who are not going to build up an entire center of excellence around building out an AI infrastructure of their own.

Rob Strechay

>> So almost like, "Hey, I'm going to do a POC or I'm going to get to production, but maybe it's departmental or something of that nature." Or like you said, even in the smaller, and we totally agree, we think that much of inference is going to be done on CPUs, not necessarily on GPUs from that power perspective, what else do you see happening in that AI realm and what's going on from a NetApp perspective there?

Jeff Baxter

>> So I think you're going to continue to see us work hand in glove with NVIDIA. We do a lot of work with them. You saw a lot of announcements from us at NVIDIA GTC a few months ago. You're going to continue to see us doubling down with them on working on the NVIDIA AI data platform, and we're obviously one of their lead partners to do that. And it aligns very well with the vision that we laid out at our user conference at NetApp Insight last year of building this capability, this AI data platform sort of capability into actually the NetApp intelligent data infrastructure. This idea of bringing AI to the data. We always talk about data having gravity, the idea of, "I'm going to take a petabyte of training data and just move it somewhere just to train it," is relatively absurd, but the idea of can I move the vector databases? Can I move the metadata catalog? Can I move all of that to be next to and closer to the data and part of the storage infrastructure and make it truly into a data infrastructure? And it's amazing because we really see eye to eye with NVIDIA on that. That's the whole idea of not to speak for them, but the NVIDIA AI data platform is you can take the NEMO microservices, you can take everything they're building in their stack and run it directly attached into the data infrastructure to just achieve unimaginable levels of performance, unimaginable levels of integration between the data and the AI platform itself.

Rob Strechay

>> And I think anybody who's looking at it sees that if you're spending all this money on GPUs, you definitely want to keep them busy. And I think that seems to be a big place that NetApp is really innovating as well.

Jeff Baxter

>> I think there's two pieces. I think there's the traditional speeds and feeds aspect. So there's the most expensive Silicon Air data center is an unused GPU, right? Or a GPU that's throttled because of not enough storage or not enough storage bandwidth. So that's to me, almost lowest common denominator. And so we look at things like BasePOD or SuperPOD, and certainly NetApp has qualified SuperPOD configurations. I'll say this, maybe I'm not supposed to say it, our competitors do too. So it's not like from a speeds and feeds perspective, that always becomes a race to the bottom so we can meet any sort of performance requirement for any AI workload, and we've got that. But then the next part about not only are you keeping the GPUs busy, but are you feeding them with the highest quality data? Because probably the second most expensive thing in the data center is the GPU being trained on the wrong data, right? Because yeah, you're keeping it busy, but you're doing useless work and you're enabling hallucinations, you're enabling all sorts of bad outcomes If you're training it on data that is non-compliant, data isn't secure data, all sorts of bad things can happen. So it's kind of keeping it busy, but keeping it busy with the right stuff.

Rob Strechay

>> Yeah, I was going to say, let's kind of dig in on that a little bit because I think that that is also where you guys have been going for the past year. We were here last year and we kind of started to get a peek inside that, and then it was more at Insight. And now I think you're again leaning into that and being able to not only do that, Sandeep was on and he talked about bringing that to block for the ransomware detection and things of that nature. How do you see these features really, because there was a lot of talk about data lakes and other things earlier today, how do you see that kind of transforming from a data platform perspective, because that's where people are looking to figure out?

Jeff Baxter

>> Well, so I'll start this by saying, and this may sound a little bit orthogonal, but we have an amazing data science team, our lead data scientist, he's a brilliant guy. I catch maybe half of what he's saying. I think I need to go back and get a PhD in statistics to understand much more than that. But we've built them as a central function so that we can build AI into everything we do. And by that I mean, can we use AI to enhance the anti-ransomware protection that Sandeep was talking about? Can we extend it to file, can we extend it to block later on? Okay, can we now take that data science and bake it into an AI data platform where we can use it to intelligently allow our customers to classify data, to categorize data, to determine what data you should train on, what data you shouldn't train on, determine what data is PII, what data isn't PII. So it's really building that intelligence when NetApp says intelligent data infrastructure over and over again. And yes, it's a tagline or whatever you want to call it, but it's one of the most honest ones I've ever heard because it's literally what we're trying to do. We're trying to build infrastructure that is used for data that we bake intelligence into. And so from that perspective, that's everything our data science team is doing. That's everything our product team is doing is how can we make this more than... People throw around the term dumb storage, probably unfairly, but how can we make this about more than speeds and feeds? How can we build the intelligence into the data infrastructure so that it can truly serve data to you, not storage to you? And I think that's what you're seeing our partnership with NVIDIA, our partnership with Intel. That's really what we've been building to, that's what we've started to do with the AI pod, and I think that's what we've teased out. You're going to see us do more and more over the course of the year. We kind of teased out our vision starting last year at Insight. I won't say a lot about it, but this is going to be a super fun year for innovation in the AI space from NetApp.

Rob Strechay

>> Yeah, I hope so, because I think there's a lot of help needed to these teams. And we talked about it a little bit earlier around how do you get from and with Sandeep again, but around getting from POC to production and how are you going to do that? And there's a lot of moving parts. It's still... And I think some of the opinionated stacks that you're coming out with, with the AI Pod Mini and some of the AI Pod and like you said, the DGX cloud stuff as well. And pods, one of the things I'm curious about, and we've talked to a few of the sales leadership this week about this is kind of the changing personas that you're selling to. How do you see the, because you have to market to them, so how do you see the personas that are actually buying the technology changing?

Jeff Baxter

>> Yeah, it's interesting. So I started at NetApp as an SE, 17 years ago. And at the time, yes, we talked about selling to CIOs, but that was always kind of seen as the apex. And most of the time you sold to storage admins or VPs of infrastructure. And don't get me wrong, those are still some of our most loyal and trusted customers, and we work with them on an ongoing basis. But you're right, it's expanding. So the data scientist is interesting. We always talk about whether data scientists are a target persona. And I say that just because in a lot of cases, the first thing you hear back from a data scientist is, I don't care about the storage. Don't talk to me about the storage. And so there is this new classification really of sort of data engineers. It's almost like if the DevOps practitioner arose to sort of bridge the gap between developers and IT staff, it's almost like data engineers are coming to bridge the gap between data scientists and infrastructure. And I think those are going to be... If I had to look into the next decade, I would say those are going to probably be the most influential persona for us. It's not that we won't talk to storage engineers. We will on a daily basis. It's not that we won't talk to data scientists and we'll bring our own data science team in to talk to them, but it's really about who are the data engineers that are bridging that gap. That's one cohort. And then almost the other direction, we're having far more conversations with CISOs and security teams than I think we ever were before. Because the CISO a decade ago was perimeter security. It was like, "Why am I talking to the storage guy if the threat actors are touching my storage, I've already lost." Now the assumption is the threat actors are always touching your storage. The zero trust model means you assume that they're always calling from inside the house. So now we're dealing more and more with security teams about how can we help them secure, classify and protect their data over and over again. So it's kind of this weird branching out where it doesn't mean we stop talking to that core persona, but we get to talk to far more branches within the enterprise, and it's fabulous.

Rob Strechay

>> And I think we see it as well as that data engineer living within platform engineering, which is the new name for IT in most cases. So kind of as we get to the end here, go towards the vision for us. What do you see coming down the road? Some of the stuff get people excited to stay tuned towards Insight coming this fall. What's going on?

Jeff Baxter

>> I think registration for Insight is open. If it's not... I think it is, so go do it now. If you haven't had a chance, it's coming up in late October. It's fabulous, right? It is the intelligent data infrastructure event of the year. It's a great time to learn about not just NetApp, but everything going on in ai, cloud data infrastructure, modernization, cybersecurity. So cheap plug, got that out of the way. But then what are we going to be really doing there? Well, here's what I'll tell you. We really have said that we're focused on for our customer imperatives this year, right? We're focused on data infrastructure modernization. We're focused on cloud transformation. We're focused on cyber resilience, and we're focused on AI and the transformation of ai. So we are developing in the product group, I report to the chief product officer. We are focused every day on innovating on that. And our roadmap leads squarely to a pretty large, pretty large drop, a pretty large explosion that'll be happening at Insight. So we've talked about the vision. I think Sandeep talked with you a little bit about what we're going with disaggregated, ONTAP. I'll say something about disaggregated ONTAP. The thing that I love most about disaggregated, ONTAP is not the disaggregated, it's the ONTAP, right? Because unlike everyone else, we're not reinventing the wheel, right? If a customer today, tomorrow wants to buy into all the data management capabilities, they don't have to wait until Insight, right? It's all there because it's all ONTAP. So all that we're doing is we're continuing to expand on what ONTAP can do. ONTAP can be your file solution, it can be your object solution, it can be your block solution. It could be your dedicated block solution. And over time, now, we're expanding it to even more use cases by bringing forward, at least this is our plan and our vision to talk more about disaggregated ONTAP. So there'll be a lot of that. We'll talk about even more we're doing in the cloud and more services will roll out there. I can't say a lot about what we're doing in cybersecurity, but there'll be some real cool cybersecurity announcements coming. So just across that entire swath, you'll have a lot of excitement coming.

Rob Strechay

>> It always is good. You guys always do bring it. And Insight, I've been to many of them when I was here and ever since. So it's a good event to go to and also be with your peers. I think to your point on the personas, it's funny, I saw one posting for a storage engineer the other day, and it almost blew my mind. I'm like, "They still are out there, and they're still getting hired," so that's good.

Jeff Baxter

>> Oh, that's great.

Rob Strechay

>> But I agree with you, the data engineering, how that ties into AI and how people are looking for the right platform and the right capabilities to be where they need to be because the AI is coming to the data. I totally agree with that. Well, hey, thanks for coming on, Jeff.

Jeff Baxter

>> Thanks for having me.

Rob Strechay

>> Really appreciate it. And thank you for watching this episode. We'll be back pretty soon from NetApp Converge 2025. Stay tuned.