Intelligent Data Infrastructure for AI

Hello everybody, and welcome to Techstrong TV. We're having an interview today with Gaby Boko, who's the chief marketing officer for NetApp, and we're having a little chat about, well, AI and infrastructure because it's getting complicated. Gaby, welcome to the show.

Thank you. Really glad to be here. I think everybody's starting to understand the scope of the challenge with AI, and I think we're all trying to figure out how we're going to operationalize all this stuff at scale, because it just requires a massive amount of data.

But I don't think a lot of our existing infrastructure was designed for that. So what you're seeing out there, what is the scope of the challenge, and what should we be thinking about? Yeah, thanks for the question.

I think somebody said this to me a while ago, it's like AI is going to redefine how we think about plumbing, and I think that that's a really important statement to make. Because when you think about all of the things you just said about AI, because AI eats data and we're all super consumed with what we are doing with our data, which is the right thing, if we're not taking advantage of having that infrastructure conversation as well that is managing that data, then I think we're missing it. Especially when it comes to thinking about it as an afterthought.

We like to call that built in, not bolted on, right? Just like with anything, if you're thinking about your data infrastructure last, then you are probably thinking about it in a perspective that isn't going to work synergistically with how you're thinking about your data. Right?

The amounts of data have really forced us into rethinking not just data infrastructure and not just data storage, but what are we doing with that? What's the data management aspects? What's the security aspects?

So the amount of data that's coming from just what you create and then AI creating the data on top of that, and then doubling, it's doubling, tripling, and managing all that really does require you to think what to do with that. And how to think about it maybe more from the ground up, so have the conversation before you're just thinking about the workload or anything else, because then you're missing something. What is your sense of where are we on this journey?

Are people prepared to make that level of investment, or are they still coming to that and it's something of a surprise to them as they kind of work along here? Because I'm seeing people are building stuff in the cloud, and then they're starting to wrestle with the latency issues that go with that, the cost of the tokens and everything that goes around that. But it's not quite clear to me that they figured out just what is the budget requirement here.

Yeah. And I think that a lot of people think about it, but they think about it in maybe in a different bucket. So I like to think about it as a disruption, as everybody does, but it's an opportunity to disrupt even your model and your existing thinking.

So if you think about it, right, you're not just dragging your legacy technology stacks along. You're purposely reinventing it. And that's where I think people are really starting to wake up to the fact, especially our customers, right?

They're sitting there saying, "Hey, I've got security issues. I've got low latency issues. How do I think about that from the beginning?

" You're picking then priority workloads. You're picking how you're trying to transform, and you're really thinking about removing that friction in your thinking between data and how and where I'm trying to innovate. So those two things alone are saying, "I'm not going to just exist for what I've had before.

" Now, NetApp, obviously, we focus in on what we like to call intelligent data infrastructure, and we really are saying that that is the moment to reinvent, to say, how do we embrace AI and all the needs we have with AI, but continue to say, how do we solve for the scalability, the cost, the structure changes, and the security changes that are obviously creating challenges for us in what we're trying to do? Honestly, if we think about it, this is the truth. If you are the companies that have not just a great AI ecosystem in the cloud or the best kind of compute, but the ones who have really strong data infrastructure aligned with integrated intelligence, integrated security, those are going to be the people who have disrupted their own business to take care of the disruption coming.

To your point about intelligent data infrastructure, how smart will the data get? And I'm asking the question because every time I look at AI, it's basically a challenge to figure out, how do I get the right data in the right place at the right time? And that kind of means everything from the inference engine for the AI model all the way to the whatever prompt, and the data that I'm including in the context window to go with that prompt.

It seems like there's a lot of science in that. So how smart is smart? How smart is smart?

I love that. I don't think you can ever be too smart if that's really what we're asking here, right? Again, I go back to that built-in, not bolted on, right?

What we're saying is, is in that context, is that we already know that we are going to be overly smart. The workloads and the way AI is getting used today is already going to be different in six months or even less, right? So what you need to do is build in a foundation that actually is solving for that, again, that unified and intelligent and secure moment for your data.

The question is, who or what do you need to do, right? When you think about your data, is there parts of your data that are maybe legacy pieces of your data that you just need to keep safe and secure, and we're going to put that somewhere else? Do you really need to have and build legacy infrastructure to support all of your legacy needs?

Probably not. AI is smart enough to be able to say, "You know what? You haven't used that in a while.

Let's keep it secure. " So again, that's coming back around to that model that says, if you're intelligent and built in along the way, then what you're doing is you are creating the kind of relevance to your data that almost, like me as a marketer, let's just use me as a marketer. I am saying I want AI to define where the relevance factor exists inside my data, not just give me more of the same.

So what we're doing, like what I do in marketing, is I sayIf personalization at scale is actually now a thing, then am I personalizing for my entire set of audience or my entire set of data, or am I personalizing based on something else entirely? That is me saying I want AI to define my dataset, to protect the rest of my dataset, but to really focus in on another dataset. And you can't do that unless you're really defining again and putting your smarts across the data landscape, across the data estate, and then giving and prompting AI to really understand and build that from the ground up.

So smart is smart, but it also is precise, right? It means that you're not boiling the ocean every single time. And that's why you need the kinds of tools that say, "I'm going to do this at every single layer versus just one.

" So honestly, you know what I think? I think that that's always what we wanted our data to be, and that's always what the promise was. I think now what AI does to it is it might actually just make it a reality.

So how do I navigate that? To your point, we clearly need a AI stack of some type that's optimized for running those applications, and yet I will have this massive amount of legacy data that I'm trying to move and expose to those AI models. So how do I balance between the need for an AI stack and my existing investments in IT?

I think that's a great question. I love that question. I think that this, again, back to disruption, what we're being given the opportunity is to disrupt existing models, right?

I think the real barrier to disruption is to say, "I am going to try to solve not my whole architecture, but I'm going to focus in on maybe real-time architectures that can focus in on the workloads that are going to maybe seamlessly connect my data or give me performance results," right? I think that what that does is says that the idea of what architectural shifts need to happen isn't I need to rebuild or get into a walled garden type of approach. You want to have AI and your agentic workflows determine what those architectural shifts are going to be, right?

So, I think that enterprises need, let's call it at the moment, on real time. We've been talking about real time ever since I've been in tech. " We would say that that's your security, your performance, and your mobility, right?

So that I can do whatever that is, right? I want to meet my customers where they are. I want to attract new customers.

I want to offer my best service. " So I don't know if that's a perfect answer for this, but what I do think is this is work that technology's been trying to do ever since I've been in it. I think AI makes it possible as long as you're looking at your system as something that is intelligent and secure and, quite honestly, something that is built to help you navigate the future.

And that doesn't necessarily always mean that you're rebuilding it. That just means you're focusing it. There are a lot of things that happen downstream from the data that actually trace back to how we manage that data.

In case of AI, that could be everything from our efforts to make sure that we're optimizing it for AI browser search, and then there's also context windows, and the better I am at managing that, the better and the answers are, plus less costly as it gets. Do you think people have enough appreciation for all those downstream things that they need to be paying attention to that are directly related to how you manage the data in the age of AI? I would say, I'm going to talk about it from a CMO perspective because that's what I do every single day, right?

I feel like a lot of times we talk about data as a strategy. We have our first-party data strategies. We talk about our data-driven campaigns, and all of those things are very downstream, right?

And I think that what that means is that sometimes that means we're missing from the conversation on the design of what we're trying to do with the data. I think that that's a really important concept for most people to get beyond. You have to be able to ask the hard questions, not just focus in on the downstream actions, right?

Yes. Am I looking at how I'm using SEO, GEO, and LLMs to drive greater reach? Absolutely, I am.

But if I'm also not asking how is my customer data being used or structured for AI ingestion to be able to deal with the SEO and GEO, then I'm missing the conversation. If I'm not asking the question on how does our data retention policy around our customers intersect with the AI training models, then I'm missing the conversation. So someone inside every organization is going to own that conversation.

And if it isn't you, then you need to invite yourself to the conversation. Otherwise, what you're going to be doing is you're going to be creating this grateful recipient of AI wonderfulness and not basically being a part of the shift or the disruption that needs to happen inside your own organization. So I think you have to be able to do both.

You have to understand what you want it to do, and then you have to understand what's going to drive it and make it successful. So that's how I manage it in terms of marketing. I am involved in that, not just because I'm in a company that does it, but because I think it's relevant to how my tactics perform.

Do you think organizations will finally revisit data management in general? Because I might argue that few organizations that I know would get a really good housekeeping seal of approval, shall we say, for the way they manage data. And in the age of AI, are we finally going to come around and have that conversation because, well, after all, it's all about the data.

I think yes. I think that data management becomes far more likely. However, I think that if you just start from a data management layer in terms of, oh, I need to, and you're not actually thinking about why you need to, then it still will feel like a boil the ocean moment, right?

I think what you want is something, again, do I have fragmented systems or do I have fragmented data definitions? Or have I not cleaned any of this data and does it just need to live over here? I think you have to ask the questions that allow you to be prescriptive.

Again, AI eats all of your data. Do you really want to deal with data management across everything, or do you want to make it super prescriptive? " Yes, you do, but focus where it's going to make the most meaningful value for you first.

" Yeah. Thanks for the question. I really like some of our customers, again, in their hybrid cloud environments, really looking at moving their data and their relevant data between cloud and on-prem, and really taking a look at that movement as part of what intelligent data infrastructure can bring to them.

When you're thinking about not just where your data's living, that's kind of an isolated view, but when I see my customers saying, "I'm going to put high impact data over here because I'm going to use it more. I'm going to put low impact data over here, and I'm going to revisit it in X number of months," I really view that as a smart motion, especially when you think about unified and real-time data that's reflecting where their customer journeys are. I also really like the ones who are focused in on what they're doing with security.

Again, the threshold and the map of where you would have risk with your data can be everywhere. " So a security mindset is something that I think is helping get to that precision of how they want to deal with data. Again, back to that built-in, not bolted on, right?

" So I really love both of those aspects between the hybrid cloud and the security that our customers are working on. That's some of the most successful ones that I've seen, and they're really fun stories, quite frankly. To your point, haven't we come full circle?

And I would argue we spent the last decade or more trying to push as much data as we could into the cloud. But when I was younger, the prevailing wisdom was bring the compute to the data. And now have we come back to that, where we're now making intelligent decisions about where the compute and storage and networking resources need to be based on, well, data gravity.

I think we absolutely have. I think that's a really good assumption and observation. We've actually said that on a variety of perspectives when we talk about AI, stop bringing everything to it.

But bring your data to this, to AI. I think that that is absolutely a full circle moment, and you're actually seeing that that's how people are using the cloud and AI moments to say, "This is really actionable. " So yeah.

Again, this data conversation is not new. I think AI has exacerbated and accelerated certain conversations that might have been maybe ignored or just maybe put aside, or maybe we can't solve it the same way. So I'm excited about the power that AI brings to the data conversation, the data estate in any customer.

I'm excited for what it does for the industry. And most of all, I'm excited for what it does to, again, my area, data infrastructure, because it finally makes the conversation relevant because people are thinking again about it at every single level, not just at the moment of use. So, yeah, full circle, 100%.

Mm-hmm. And aren't people going to come up with an actual strategy for managing their data? And I'm asking the question because we've heard for a long time now, data's the new oil, but I always observed that the problem with that whole analogy was we didn't have any way to process the oil and turn it into something useful.

So are we finally going to get to the point now where we not only have data as the oil, but we also have the mechanisms in place to process it and then pump it to where it needs to be? Yeah, I think so. I agree with that.

Obviously, what that means is that leaders in any company need to recognize that your outcomes on AI or anything with your data are determined by how well your data is getting accessed or governed or activated. And that shift from simply maybe just managing the systems or having a technology architecture conversation is really creating more, I don't know, maybe the data is becoming more meaningful. Maybe it's becoming more human-centered because you're focusing on the experience.

That means your data is continuously in motion. Hopefully, it means you're driving decisions at scale. I think that obviously, if you're committed to your data and you're committed to making data the oil, as you say, then you're committing to say that we want to operationalize it to accelerate our innovation.

I personally believe at NetApp that that means that AI is part of that, intelligent data infrastructure is part of that. Hopefully NetApp is part of that. But ultimately, your data strategy inside a company is part of creating that unified foundation, part of creating that oil that you referred to, without forgetting that it's the human experience and the performance level metrics that are going to magnify those experiences that matter.

Hey, folks, you heard it here. No matter what era we're in in IT, it always comes back to the data at the end of the day. Hey, Gaby, thanks for being on the show.

You're so welcome. All right. And back to you guys in the studio.