Why Storage Is the Strategic Control Point for AI with Jim O'Dorisio of HPE

(Tom Joyce at 00:00:00) So AI is super important. And to enable AI, you really need a trusted data foundation, and the trusted data foundation is where storage comes in. It has to be performant. It has to be resilient. It has to be secure.

(Tom Joyce at 00:00:12) We've actually built agents into some of our data resiliency products that watch other agents. Because what they're doing is they're paying attention to, like, is something happening in the environment that shouldn't be happening?

(Joel Beasley at 00:00:30) When I was given this opportunity to have this conversation about HPE storage and why storage is becoming important and strategic for the enterprises, I was like, okay, who is the guy on that? And I was told you are the storage guy. Is that correct?

(Tom Joyce at 00:00:44) I'm the storage guy. I've been in storage for thirty plus years, and it is different. Right? It's been in technology, but there's definitely a unique set of people running around in storage. And storage is in vogue again. Right? We've got a lot going on.

(Tom Joyce at 00:01:03) And with the advent of AI, it's becoming that much more relevant to the success of our enterprise customers.

(Joel Beasley at 00:01:12) Yeah. You know what? In my mind, being in this space for over twenty years, storage was kind of like the archive. Like, it was just like, ah, it's back there.

(Joel Beasley at 00:01:20) It's dusty, low light. But now it's sexy. People are talking about it. It's important. Why is it becoming so important?

(Tom Joyce at 00:01:26) I mean, storage is, you know, from my perspective, because I've been in it, it's been important for the longest time. But it's taken on a new level of importance within the strategy for AI. Because every customer out there is looking to use AI in one way or another. They're either looking to gain efficiencies within their operations to save money, or they're looking for ways to connect to their customers with new products and services that they can build from the insights they can extract from their data.

(Tom Joyce at 00:02:00) So AI is super important. And to enable AI, you really need a trusted data foundation, and the trusted data foundation is where storage comes in. It has to be performant. It has to be resilient.

(Tom Joyce at 00:02:12) It has to be secure. And if it's not, you know, if you don't build on a really solid data foundation, these AI projects tend to fail. Worst case, they fail. Best case, it slows down the velocity that customers can expect to get to value.

(Tom Joyce at 00:02:33) Right? And at the end of the day, our customers, especially when you think about the amount of money that's being spent on GPUs and GPUs are very expensive, you want to get to value quickly. You want to turn these projects into value and competitive advantage. And storage continues to be very important in that regard.

(Joel Beasley at 00:02:53) And so what's your role? Like, what's your day to day like?

(Tom Joyce at 00:02:55) You know, my day to day... I'm the GM of the storage business here at HPE. So I've got a large team of development people and go-to-market resources that we work to align. And we've got a very comprehensive portfolio here.

(Tom Joyce at 00:03:13) So, you know, look, it's not a part-time job by any stretch, but it's a lot of fun. We like solving problems, and we like taking care of our customers.

(Joel Beasley at 00:03:22) Yeah. And how did you get started? Were you always in storage? Did you come right out of school?

(Tom Joyce at 00:03:27) I came out of school, and I actually started in storage at Digital Equipment many, many, many years ago. And I've always found a niche there. Did a bunch of different things in storage, everything from developing storage arrays in the old days to running software businesses, to running a data protection business at EMC, and then onto building an ML/AI platform in a recent role at another company. And now I'm kind of back in my roots and running the core storage business.

(Tom Joyce at 00:04:05) And I have to tell you, it's a lot of fun. It's where a lot of action is. And we're in a great place to help customers solve some really complex problems.

(Joel Beasley at 00:04:16) When we were talking last week, you mentioned that storage was a strategic control point for AI. But what... That sounds cool, but what does that mean?

(Tom Joyce at 00:04:25) Yeah. It really is. Because at the end of the day, the data that customers are... Really, from an AI perspective, you're trying to extract value out of data. That data is typically spread across lots of different places within a customer's environment.

(Tom Joyce at 00:04:41) Right? So having a really solid data foundation that can deliver the data at the speeds that the GPUs require is super important. Right? So when you think about storage, it's not just the physical storing the bits and bytes on, and now it's on flash media, right?

(Tom Joyce at 00:05:02) It's also the ability to develop, you know, to have a data management layer built around it that gives you compliance and governance. It's having a really strong resiliency layer around data protection. You know, part of being, part of a data foundation is having that resiliency and data protection to protect from cyberattacks. I mean, the vectors, the attack vectors are just increasing from a cyber perspective. And now you're bringing all this data out to be analyzed, and it's kind of the crown jewels of a company.

(Tom Joyce at 00:05:42) It's where the value is. The data is where the value is for a company. And you have to protect it. You have to secure it. You have to enable it to be performant.

(Tom Joyce at 00:05:52) And so we've built some really strong cyber protections throughout the stack, throughout the storage stack, which is super important. Because as you embark on these AI projects, you want to have the confidence that you're leveraging a resilient and secure infrastructure. It becomes even more important because, to a certain extent, you might be looking at data that you haven't looked at for many, many years, and you need to ensure that that data is secure and protected and available.

(Joel Beasley at 00:06:20) What type of cyber protections? Can you explain that more?

(Tom Joyce at 00:06:23) Yeah. I mean, we built cyber protection into our arrays. We build cyber protection into some of our data protection products. So we have products like the Cyber Resiliency Vault. And these products can detect anomalies in the data stream.

(Tom Joyce at 00:06:41) So, you know, it's really down to fine-grained, millisecond kinds of checkpoints that allow a customer to identify what could be an issue, isolate the issue, and ensure that you're not propagating some attack vector that has impacted your organization. So it's super important. And I think customers... You know, if you think about the two sorts of initiatives that are at the board level and are being discussed at every board, frankly, it's AI. It's like, how do you maintain competitiveness, and how do you gain competitive advantage and differentiate your capabilities and become more efficient while you're doing it?

(Tom Joyce at 00:07:24) So what's your story? And then the other one is how do you protect from the increasing cyber threats out there? Because, you know, having a resilient infrastructure says that your workloads, your applications, the things that you're running your business on continue to stay up. And so not only do companies want to stay out of the news, they need to continue to operate.

(Tom Joyce at 00:07:48) And having a resilient infrastructure with appropriate cyber protections involved, absolutely critical to that. Right? So these are board-level conversations. So, you know, when we talk a little bit about why storage is kind of sexy, why it's in vogue, why it's so important, when you think about the threat environment out there and you think about AI, those are two of the primary reasons that folks are focusing in on storage to such a great extent.

(Joel Beasley at 00:08:17) I'm curious to understand the boundaries of what HPE does. So one of the questions I'm sure I'll get a lot because you're talking so much about data and readiness with data. Do you work with partners and do consulting in that regard? Like, if I'm a big company and I'm asking you about how to craft the data strategy, is that something you do?

(Tom Joyce at 00:08:39) We have a full boat of professional services and capabilities at HPE. So we not only build products from compute, storage, networking, and we're one of the few vendors out there that have the full stack from a compute, storage, and networking perspective, but we have a full breadth of professional services capabilities. And we build products that offer data management capabilities as well because, you know, one of the key things around data management is that to implement an AI project, you really need to understand your data. So do you have a data strategy? Do you understand... Do you have compliance and governance built around your data?

(Tom Joyce at 00:09:19) Because if you don't, you know, it kind of slows down the AI projects because customers are looking at it from the perspective, should I expose that data to this model? Is it appropriate? Does it have PII information in it? We don't want that exposed to a model. So you have to know what your data is.

(Tom Joyce at 00:09:37) And so the data strategy, where's your data? Do you have compliance and governance built around it? What data is appropriate to be in what dataset to be exposed to the appropriate project from a... You know, if you're exposing your data and you're always going to run your data through some model, it could be frontier models, it could be open models that you've modified, but your data strategy allows you to do that with some confidence because I know what data I'm exposing.

(Tom Joyce at 00:10:05) If you don't, you know, it tends to slow projects down because customers are reticent, right, to put the wrong data at risk potentially. Right? And that doesn't mean... You know, it's not like AI is going to destroy your data. It's the fact that you might have certain datasets or data that doesn't belong in a given dataset or part of a project.

(Tom Joyce at 00:10:25) So that's why it's so important. So we not only have products that help with that, but we also have a full capability around professional services that can do that consulting with a customer.

(Joel Beasley at 00:10:37) You're fairly customer-facing in your role. Right? Like, you meet with customers who go to...

(Tom Joyce at 00:10:41) I am. You know, it's a dynamic role. You know, it's the GM. I visit customers. I work with our field resources, and I'm kind of laser-focused on the products that we build and the strategies that we're driving. I think, you know, it's frankly the thing that makes this job the most fun. It's the diversity of things that you have to really tackle in my role. So, yeah, it's a lot of fun.

(Tom Joyce at 00:11:07) But, you know... And I have to tell you, like, you mentioned it earlier, this not the emergence, but the rapid evolution of AI is just keeping everybody on their toes, but it creates a huge amount of opportunities for us to find ways to best take care of our customers and help our customers create value. So it's an exciting time to be sure.

(Joel Beasley at 00:11:32) It is so exciting. So I was a developer for twenty years where I built and sold software products.

(Tom Joyce at 00:11:37) Awesome.

(Joel Beasley at 00:11:38) And in real estate, fitness, and finance, like, three categories. And then I kind of stopped when the podcast got popular. Like, I managed one application, but the podcast has been going for almost ten years now. Yeah.

(Joel Beasley at 00:11:49) And so, I got a little dusty, Jim.

(Tom Joyce at 00:11:51) Yeah.

(Joel Beasley at 00:11:51) Not going to lie with my coding skill. I got out of the ecosystem, okay? And then about, you know, January of this year, I... Well, I took a poke like two years ago. I was like, what can it do?

(Joel Beasley at 00:12:04) Because I've been having all these conversations and I... It's like, okay. This is awesome. This is like having a junior programmer that you can do some stuff but you kind of correct them. And now...

(Joel Beasley at 00:12:14) And then I checked it out again in January and it's like senior brilliant, way smarter than me, an engineer that can do multiple things and organize with others. And so in January, February, I started building applications to improve my life, like, things for my family. And you can just do it. Like, what would take me six months and like $500,000 of my own capital can now happen in three hours on a Sunday afternoon. And...

(Joel Beasley at 00:12:42) I've just been going wild with it. Yeah.

(Tom Joyce at 00:12:44) No. It's fun stuff and just the... It's amazing. Every week, like, my teams, I'm never shocked by what they come up with. Right?

(Tom Joyce at 00:12:53) Because we're just making rapid transformation. And, you know, it's a little more complicated when you're dealing with, hey. How do I do AI-assisted development that's leveraging millions of lines of code? You know, because you have to train agents to understand the code that you're interfacing with. But, man, you could build things from scratch like, you know, just with a snap of the fingers.

(Tom Joyce at 00:13:16) Right? And so obviously you're into it. You know, you talk about your skills getting dusty. I was a developer many, many, many years ago, and I've been doing the management thing for a number of years. So I'm really dusty, and I'm even having fun with it.

(Tom Joyce at 00:13:31) So it's definitely an interesting time to be in tech right now. And, you know, our real job is to take and harness that massive set of capabilities and turn it into help or, you know, turn it into value for our customers. Right? And really understand how they're trying to transform their operations and how we can help them do that. Right?

(Tom Joyce at 00:13:53) And that's the part that I think becomes really rewarding to be in the space that we're in right now.

(Joel Beasley at 00:14:01) I fully get it. You get to meet all the different customers. Like, if I didn't do this or own my own business, I'd be a consultant because I like to go in, understand any problem, figure it out. When you're visiting customers, what is the... Take your recency bias of the past three, six months.

(Joel Beasley at 00:14:17) What is the conversation that they're coming to you with the most often?

(Tom Joyce at 00:14:22) Look. Customers are trying to solve a variety of different problems. You know, in some cases, and we're in storage, it could be that they're trying to solve a transactional workload. It could be that they're trying to solve unstructured workloads. Unstructured is the set of protocols around, you know, whether it be file or object that's...

(Tom Joyce at 00:14:42) You know, where the AI work is happening. Right? And so, you know, we have the consultative conversations, like, what problems are you trying to solve? But I think, you know, we have a resilient... We feel really good about the products and the innovation that we've built into the products and their ability to help solve these...

(Tom Joyce at 00:15:00) The... be a foundational underpinning to solving these problems. But it always starts with, you know, what problems are you trying to solve? And, oh, by the way, have you thought about AI? Because every customer is thinking about it, but every... You know, you really kind of have to find out where they're at on their journey. Are they advanced? Are they at the point where they're trying to build some sort of a sandbox so they can validate, do proof of concepts and that sort of thing?

(Tom Joyce at 00:15:24) Are they on the verge of going into production? That sort of thing. And so you kind of... You know, let's understand where the customer's at and how we can best help them get to the next step. Right?

(Tom Joyce at 00:15:34) And I think increasingly, you know, what we're finding when we meet with customers is, you know, traditionally, storage maybe got bought project by project. Right? And, you know, customers might be buying capacity and, you know, performance and, you know, maybe not a lot of thought to the bigger picture. And I think increasingly, as customers are looking at AI, these businesses, these functions within a business that didn't typically talk, like, let's say, the storage infrastructure guy or the infrastructure guy called compute, storage, whatever. Maybe you weren't talking to the analytics and the governance guy, you know, or the data guy.

(Tom Joyce at 00:16:12) They... that was, like, you know, off somewhere else. And I think what's happening is companies are recognizing that you have to get these guys together so that you can understand where you're trying to take the infrastructure to support where you're going from a data perspective. Because data, again, is kind of where the value's at. And so it real... You know, when companies can start doing that, it definitely allows the subsequent projects that they build out to go much quicker. Right? Because the infrastructure guys, the data guys are kind of together. They've got a strategy. And then you're buying infrastructure that actually lines up to where you're going.

(Tom Joyce at 00:16:50) Right? So we do consult from that perspective and try to say, look, these are some of the best practices. And if you're not doing that, you know, here's some things you can do to kind of speed up your transformation. And it's all about time to value for these companies because they're out... You know, this stuff's not cheap.

(Tom Joyce at 00:17:10) Right? And, you know, we try to be as cost-effective as we possibly can, but given the rapid build out of AI, if you think about what the hyperscalers are doing, massive investment going on now. You know, we've all dealt with some constraints and supply challenges with material. We've done a really good job of overcoming that, but, you know, it has influenced the price of things.

(Tom Joyce at 00:17:35) And so it's even that much more important to help customers get to value quickly. Right? And that's really the center of what we try to do from a consulting perspective, but from a product perspective as well.

(Joel Beasley at 00:17:48) I like your point of the guys are talking to the other guys now. So people in the organization, they wouldn't. It's like you get to make new friends in this transition.

(Tom Joyce at 00:18:01) Hey. It forces, you know, forces new... You know, because if you don't, you've got friction that you don't need. And, you know, like you said, it fosters breaking down some of these stovepipes, if you will. And the most successful companies are doing that effectively.

(Tom Joyce at 00:18:17) Right? And even here at HPE, we're, you know, running lots of AI for all the reasons that I said earlier. You know, we're working to gain efficiencies, whether it's in the development process or whether it's in our operational process. And we're, you know, we're drinking our own champagne. Right?

(Tom Joyce at 00:18:35) We're leveraging the equipment and the products and the capabilities that we sell customers we're using internally, which helps us understand as we work with our own IT folks how to best engage with our customers. Right? So it's definitely something that, you know, you want to be in a position of experience here. And I think, you know, we're very fortunate because we're a customer of our own stuff. And then it puts us in a better position to be better suppliers to our customers.

(Tom Joyce at 00:19:05) Yeah.

(Joel Beasley at 00:19:06) I prefer that phrase "drinking our own champagne" because before it was "eating our own dog food."

(Tom Joyce at 00:19:12) Yeah. I've used that. I've used that. I was trying to be a little more elegant in this setting. You know, I...

(Joel Beasley at 00:19:17) We want to be classy.

(Tom Joyce at 00:19:18) Yeah. Yeah. We'll try to move this thing up and be classy, so I purposely used the "drinking the champagne."

(Joel Beasley at 00:19:25) Wonderful. Okay. I'm going to get a little nerdy. We don't have to get, like, too detailed technically. We try to keep it high level, but I'm just curious. I'm just a curious person so we can... You could just direct me wherever we go. Is there some... With the rise of the agents in relation to storage on a technical level, is... because the agents are always, like, thinking to itself and running back and forth, is that a workload that is the same as the rest of the workloads that we've been doing? Is there anything unique about the way that the agent is consuming data and interacting with the storage interface or no?

(Tom Joyce at 00:20:05) I mean, it certainly depends on the agent, right, and what the use case is. For example, when we're using agents to help us in the development process or in our test processes, you know, they're operating... You know, we... you do a lot of testing. You know, you build an agent. You test the results. You're kind of... You always have... Typically, you have a man in the middle until you really feel comfortable. And then you build guardrails around what the agents can and can't do.

(Tom Joyce at 00:20:31) But they're, you know, very... You know, they'll consume a lot of resources. Right? And one of the things that we've done is, you know, build out our own infrastructure so that we can scale quickly as the agents become more powerful. Right?

(Tom Joyce at 00:20:48) And, look, we have... We all have a long way to go in terms of, you know, taking full advantage of what the capabilities are, but we're on that journey. We know the journey that our customers are on. Frankly, some of the best AI we've written actually runs in some of our products because they allow us, for example, agents that can look at a storage system and proactively detect what's about to happen so that we can improve our resiliency and our uptime significantly. And when we've done that, we're still striving to do more, but we've had really good results.

(Tom Joyce at 00:21:23) I like to tell customers that, hey, you know, we're... Some of the best AI we've written, like I said, runs in our array, and it's doing very practical value-added sorts of things. Like, how can I detect performance bottlenecks? How can I detect potential capacity challenges that could lead to an outage long before they happen? And we're seeing the benefits of that, and then our customers are realizing the benefits because the systems stay up longer, the availability is

(Joel Beasley at 00:21:54) better,

(Tom Joyce at 00:21:54) and, you know, and they can continue to, you know, focus less on taking care of the infrastructure because we're doing a better job of taking care of it for them. So there's a lot of practical use cases and, I think that's the thing with the customer too. Right? The, you know, part of having an effective AI project and I think, you know, is basic... You have to have a data strategy.

(Tom Joyce at 00:22:14) We talked about that. Having a strong data foundation that's, you know, underpinning your data strategy, you know, resilient infrastructure, performance infrastructure, intelligent infrastructure. You know, these are the things that we at HPE build. But, you know, a customer has to have a notion of what they're trying to do. What are the use cases?

(Tom Joyce at 00:22:35) What problems are they trying to solve so that they can be much more targeted, you know, with the projects and get to the value that we talked about earlier much more quickly. So, you know, I think all of us, when we look at the power of this technology, it's, you know, it's like, you know, what problems do you need to solve? You know, what are the particular things that you need to drive your business forward? And then, you know, then go after designing your projects around that.

(Joel Beasley at 00:23:05) Isn't that most of just being an executive? I... It sounds so simple, but after doing this for so long, I'm like, wow. Really, the hardest thing is to keep the outcome at the center and keep it... And keep focused on that.

(Tom Joyce at 00:23:18) It really is, but it's even more important now because, you know, people could dabble. You know? But it's expensive to dabble. You know what I mean? The sooner that you have a well-formed set of ideas, you know, and then I think, you know, there's definitely learning. Right? You're going to experiment. You're going to say, look, did I get the result that I expected to get? But you start with, you know, it's too expensive to just go, hey. Yeah.

(Tom Joyce at 00:23:41) We're going to go spend a lot of money, and we're not going to expect any results or any timeline on any of this. That's just not the way it works, especially now. That's a

(Joel Beasley at 00:23:50) lot of churn. You'll see... You'll see a lot of churn.

(Tom Joyce at 00:23:52) Large churn. Lot of churn.

(Joel Beasley at 00:23:54) You'll have to look in the mirror.

(Tom Joyce at 00:23:56) Exactly. What do... What did we accomplish this week kind of a thing. And it's no different for me and my team than it is for the customers that we serve. They're... They have the same expectations of their teams. And what we want to do and what we try to do is be part of their solution, right, to help get them there quickly.

(Joel Beasley at 00:24:13) I like the predictive failure technology you were talking about.

(Tom Joyce at 00:24:16) Yeah.

(Joel Beasley at 00:24:16) It's not discussed as much as I think. I think it's underrated because what I have seen is I have seen it arise over the past five years personally. I'm sure it happened before that. But with radio towers, like cell phone towers, because it's expensive to get the parts to them, so then it created this need to predict when they're going to fail. Then I saw it happen in the, like, the trucking space. And I... So I see this, like, one technology kind of peak up everywhere, and now you're telling me it's in the storage and actually down in the...

(Tom Joyce at 00:24:46) Yeah. No. I mean, I think the thing is, like, we've always done proactive kind of support things, but the way we're doing it is the difference. You know, the agents can very quickly diagnose things that they can tie it to. Hey. Was that a problem that we've ever seen before? Can we align it to a fix that we already have? Things that might have taken us weeks to say triage, like, in the background. You know, when... If we have a problem with the customer, you know, we're going to get them relief ASAP.

(Tom Joyce at 00:25:15) Right? But did we solve the underlying problem, or did we kind of give them a Band-Aid in the short term? It might take us a few weeks to triage the issue, fully understand it, code a fix. Now we have agents that we've trained that can identify that in an hour because they're not only looking at what's happening in the customer environment from a failure signature, they're looking at it across what we're doing in our own QA process. Oh, we've seen that. We already have a fix. We can marry that up and provide the fix back to the customer almost instantaneously. And that and so, you know, taking things that we've always done and leveraging, you know, the technology, the AI technology that we have today and the sophistication is just improving the customer experience.

(Tom Joyce at 00:25:59) I mean, that's just one use case. And then, you know, obviously, you know, we're leveraging the same sorts of development techniques and models and infrastructure that our customers are using maybe for a completely different outcome. Right? So, you know, it's, you know, and so I think, you know, when I look around my team, the massive, you know, innovation that's coming and just the transformation that's happening, like, month after month, it's just truly amazing. So I think, you know, we engage with our customers in that spirit. Right? And, because we know what, you know, what's possible, you know, with the technology that's out there. You know, we... And again, you have to use it, you know, to know it.

(Tom Joyce at 00:26:44) And I... It was great to hear that you're out there dabbling because you're doing stuff that's very practical, but you're getting to the crux of just how powerful these tools can be. And, you know, and storage underpinning that is super critical at an enterprise level because, you know, is the data available when you need it? Is it secure when you need it?

(Tom Joyce at 00:27:03) Are you going to... Can it feed and keep the GPUs busy? Because, you know, the compute, the networking, the GPUs are super expensive. We've done things within our infrastructure with cache management, with, you know, direct memory access technologies that allow GPUs to be more efficient. Right?

(Tom Joyce at 00:27:26) And so that... And again, we talked a little bit about why storage is such an important layer because if you can't, you know, if you can't be up and running, if you're not resilient, if you're not performant and really help GPUs become more efficient. The whole thing slows down, and it's much more expensive. And customers are having to, you know, obviously, budgets being what they are. You know, they have to decide where they're going to spend their money.

(Tom Joyce at 00:27:50) So, you know, we want to give them, you know, really good value for their investment.

(Joel Beasley at 00:27:56) I've been hearing this term "trusted data" pop up quite a bit. Yeah. You mentioned it earlier too. And so, like, what does that mean, like, in practical infrastructure terms?

(Tom Joyce at 00:28:06) Yeah. I mean, you know, the data has to be there. So it starts with having a resilient infrastructure. It has to be free. You know, we... You have to have that layered data protection strategy from a resiliency perspective, and that starts with a really resilient storage infrastructure. It starts with a resilient network, etcetera. But equally so, it's really important to have a layered data protection strategy that includes cyber protection. We've actually built agents into some of our data resiliency products that watch other agents. Because what they're doing is they're paying attention to, like, is something happening in the environment that shouldn't be happening?

(Tom Joyce at 00:28:46) Right? And allowing the customer to kind of define, okay, if you see that, if the agent sees that, what should it do? Just tell me. Okay? I'll alert you.

(Tom Joyce at 00:28:55) Or should you stop the operation until I come and approve it? But what, you know, again, you know, we've all heard about what, you know, agents that get a little bit aggressive and maybe aren't under the kind of control they should be. I think we've all heard about that. You know, whether an agent, you know, is maliciously being applied somewhere because I think the next round of cyber challenges are going to come from maliciously crafted agents.

(Tom Joyce at 00:29:25) You know what I mean? And so having that layered protection is super important. And having a strategy to detect, you know, to isolate, to recover from those kinds of events is super important. And then you have the other scenario where you just have an agent maybe, you know, that didn't have tight enough guardrails. It wasn't malicious, but it was just doing something it shouldn't be doing. And so we're building in those protections not only to detect and isolate, but, you know, for agents to watch other agents because, you know, they're...

(Tom Joyce at 00:29:55) You know, look at this as enterprise infrastructure that's running your business, that's running the infrastructure that serves your customers. Right? So super important to have that all kind of articulated and contemplated as part of your strategy.

(Joel Beasley at 00:30:12) And there's been so much interesting research in the agent observability space about how to incentivize, control, contain, organize the different agents. Yeah. It's fascinating.

(Tom Joyce at 00:30:25) Yeah. I mean, you know, it's... I think... And we're still learning and everybody's still learning.

(Joel Beasley at 00:30:30) Here's an area that I'm curious. So HPE storage, data AI consulting, all that. Who are, like, the counterparts that you interact with the most in those other lines of business? Like, who are the... I'm trying to just figure out personally, like, who are the people? You're the person in storage. Who's the person in AI?

(Tom Joyce at 00:30:48) Well, listen. I mean, so we... My storage group works as part of the hybrid cloud organization, which is... You know? And we've built packaged kind of AI factories in a box to help customers. So, you know, we have a product in the private cloud space that delivers AI in a converged stack, which, you know, is really a turnkey solution for customers. So we're building software that helps customers quickly get to a pipeline, right, to support their dataset and get to, you know, get to tokens. You know, time to first token is a huge metric. Tokenomics is becoming a bigger metric, obviously, because token consumption is so large and so high, it can be very expensive. So doing that efficiently, understanding what your burn rate is is super important. So we have tailored products there that can... That also include our storage as part of the total stack. And then at HPE, we literally deliver AI factories that leverage our compute, networking, you know, GPUs from our partners, and we... And storage that literally deliver an AI factory that could be a data center-wide AI factory. And in that context, you might have 10,000 GPUs that you're trying to feed and keep busy at scale. Right? And so, you know, to do that, you know, we can bring that full breadth of solution to the market, and we do.

(Tom Joyce at 00:32:26) But, you know, obviously, you know, and I think, as I said earlier, we're one of the few vendors out there that has... That can deliver compute storage and networking. But we work closely with our partners in the x86 space, whether it's AMD, whether it's Intel, whether, you know, Nvidia from a GPU perspective as well, you know, are key partners in helping... You know, all those guys are key partners in building out the ecosystem. So, you know, and you have to understand who the players are in the ecosystem. The customers are looking to ensure that you are leveraging the best technology, that you're pulling together the best solutions, that you're doing it in a secure and repeatable way. And, uh, so, yeah, we think really hard about who our partners are and, you know, and what, you know, how do we bring that whole solution?

(Joel Beasley at 00:33:22) Side question here for you. The HPE Altra... How do you say that? A-L-T-R-A. Alletra.

(Joel Beasley at 00:33:27) Alright. I'm gonna ask you a question about that. What I want to do because we're coming up on time is I'm gonna ask you like your favorite thing about four or five different products. Just the first thing that comes to mind is fine. That way we can make sure we get all those. What is... You've got several different products. What is your favorite thing about HPE Alletra Storage MP?

(Tom Joyce at 00:33:47) Yeah. I tell you, it's unique in the industry because we have a composable disaggregated storage architecture because... And that... What that says is compute can scale independently of capacity. Right? And that... And that's... That really gives the customer, you know, that kind of return on investment that they need. The other thing is we built intelligence into our product lines. The X10K product is built to inline data intelligence. So what's really great about that is the data gets stored. We're literally tagging it with metadata. We're calculating vectors. We're storing vectors. And what that does... And I talked earlier about really feeding GPUs and making them more efficient. If you don't... If you do this as an offline process, you're copying data from storage up to the GPUs. You're copying it back down. You're storing it. We're literally applying the intelligence as the data lands on the media, which is a huge advantage. So we, you know, we felt like, you know, when we went after that market, we built intelligence into the product, and we built the capabilities directly into the architecture. And, you know, I think that that gives us a significant competitive advantage, but it really gives the customer value, which is what's... You know, what we're really out to do. You know? So really, really important. Yeah. Yeah. Really cool.

(Joel Beasley at 00:35:09) What's your favorite thing about HPE data fabric?

(Tom Joyce at 00:35:13) Yeah. So from a data fabric perspective and, you know, obviously, we talked about some of these concepts, you know, and I wasn't, you know, trying to, you know, sell all our products. But, you know, we... Implementing a data strategy is super important, and what's happening in a customer environment, one of the things that's most challenging for a customer is that their data is spread between the data center, on the edge, and even in the public cloud. Right? And then it's spread across disparate, you know, sources of data. It could be, you know, an object store or a file system on vendor X. Or it could be the same thing on vendor Y. And now all those data, all that data has to come together. So data fabric gives us a unified namespace and a kind of a... A global data plane that allows you to see all that data. And it allows you to move the data where you need to apply the processing to it. The other thing it does is it... We've built in MCP servers on top of it. So now agents that can apply compliance and governance can talk directly to the system. So it... It's really that data layer gives you that unified view of a very complex environment and everybody's environments are hybrid. You know, it's... You know, we think and it... We embrace this notion that, you know, IT, the predominant model in IT is hybrid because data sits, you know, like I said, in the data center, on the edge, and in the cloud. By definition, that's hybrid to us. At the same time, AI is the mode... Is the predominant hybrid workload because you're trying to apply this intelligence to your data, and it exists everywhere. So data fabric really helps solve that problem, both from an access, mobility, and from a governance and compliance perspective. So it's, uh, it's a very significant... Plays a very significant role in the in the stack.

(Joel Beasley at 00:37:12) And what's your favorite thing about the HPE Private Cloud AI?

(Tom Joyce at 00:37:17) You know, like I said, it is a turnkey solution that integrates compute, GPUs, networking, and our storage products that we talked about, along with some software, software from our partners and software that we've built that really speeds time to value for customers to get to the time to first token, accelerates time to first token. And it really... It's... It helps customers avoid having to build out some of this infrastructure that for some... Depending on... I have said earlier, you have to really, everybody's on a different point in their journey with an AI perspective. It helps meet customers that maybe haven't built out that infrastructure in such an advanced way, and it gives them this very turnkey solution that gets them to value quickly. And, um, so so really, uh, powerful set of capabilities, and it's doing quite well in the market. Yeah.

(Joel Beasley at 00:38:08) I've got a leadership question, then we'll wrap up. Is that cool with you? Okay. Sure. I get to meet with all of these brilliant minds like yourself and all these other people. And I like to ask them when they have great experience, what is something that you learned when you were younger? Let's say career, maybe thirties, and you picked up on something, you might have seen it happen. Someone might have shared it. It might have been in a book, but it's some piece of insight that you're like, okay. I'm gonna try that. And then you did, and you were like, oh, man. That worked. And then you've kept it with you your entire career. I want that piece of advice.

(Tom Joyce at 00:38:45) Interesting. You know, one of the things that I... You know, you just kind of said it, like, you're not sure where some of these things come from or at what point it kind of, like, clicked for you or became something that you kind of made sense to you. But I... You know, one of the things I like to make the complex thing seem simple. You know, we, you know, we had a lot of complexity out there. How do you connect things in ways, whether it's in your organization or your products, that reduce complexity, because that speeds value. It allows our teams to work more effectively. So I like to make the, you know, complex things simple and easy to use. And that's kind of one of my management mantras as well as I work with my team.

(Joel Beasley at 00:39:27) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you would like to hear discussed on the podcast, either add me on LinkedIn or send me an email [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.