Local government departments, energy modeling, retail automation – real-world applications for hybrid AI are exploding.
Host Patrick Moorhead is at NVIDIA GTC 2025, joined by Fidelma Russo, EVP & GM, Hybrid Cloud and CTO at Hewlett Packard Enterprise, for a detailed look into HPE Private Cloud AI’s progress, and how enterprise AI adoption is progressing.
Key takeaways include:
🔹Hybrid AI Gains Momentum: Enterprises are increasingly adopting hybrid AI models, demanding seamless integration between on-premise and cloud environments.
🔹HPE’s Turnkey Private Cloud AI: HPE simplifies AI adoption with a cloud-managed, on-premise system, offering various configurations and focusing on software innovation.
🔹Real-World Use Cases Drive Demand: Practical applications in sectors like government, energy, and retail are fueling the adoption of HPE’s hybrid AI solutions.
🔹Data Management and Observability are Essential: HPE emphasizes data federation and observability tools to ensure effective AI deployment and management in complex IT environments.
Learn more at HPE Private Cloud AI.
Watch the full video at Six Five Media, and be sure to subscribe to our YouTube channel, so you never miss an episode.
Or listen to the audio here:
Disclaimer: Six Five On The Road is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.
Transcript:
Patrick Moorhead: The Six Five is On The Road here in San Jose at Nvidia GTC 2025. We are in the Hewlett Packard Enterprise booth here and it is buzzing. Why is it buzzing? Well, it’s all about hybrid AI. I have talked many times on this show and also in our research, probably for a decade talking about the power of the hybrid cloud. Now that we know all enterprises want hybrid clouds, guess what? They also want hybrid AI. And it just makes sense. I can’t imagine a better person to have this discussion than Fidelma Russo of HPE. Welcome.
Fidelma Russo: Thanks, Patrick. Good to be here.
Patrick Moorhead: You know, we finally got you on the show. We tried, we asked, but now you’re here. Thank you so much. So at HP Discover you announced your hybrid AI system and gosh Jensen was on stage. We were in the sphere in Las Vegas. It was incredible. Tell me how things are going. It’s been a few months. How’s the customer feedback? How are the rollouts going?
Fidelma Russo: Yeah, so no, it’s been an incredible couple, you know, maybe nine months. And private cloud AI is what we announced and basically a turnkey AI system on-prem. Cloud managed, but on-prem. And we’ll soon have an air gapped version as well and multiple T shirt sizes. But the real trick is the software.
Patrick Moorhead: Isn’t it always though? I mean, infrastructure is not easy. It’s hard, but software is really hard. Pulling all these things together.
Fidelma Russo: Right. And what we’ve tried to do is to really simplify it, you know, to do all of the hardware management and lifecycle management you do with a hardware appliance. But then working very closely with Nvidia on their NIM software and integrating that with our AI Essential software and their blueprints in order to make sure that people really get, you know, as I was listening today, it’s kind of their private AI factory on-prem. And so you asked me how it was going. So we have, you know, we, we’ve had a tremendous uptake and geographies, it’s quite interesting. A lot of European interest and a lot of European sales starting now in North America. And use cases are driven really by first of all, classification requirements around people’s data. Data they don’t want to go off prem and use cases in things like energy and modeling of, you know, energy usage. Another one is retail. Clearly you know, where AI is being used to automate everything and then things like state and local government and you know, interesting things. Tax departments. Tax departments love private cloud AI.
Patrick Moorhead: Yes. Yeah, yeah, I appreciate that. I’m really glad you brought up the use cases because a lot of times in this AI conversation it ends up being kind of technology for technology’s sake as opposed to solving problems in a unique way. So we are here at GTC25 and I’m curious, can you talk a little bit about some of the announcements that you’re making here?
Fidelma Russo: Yeah. So we’re, I mean this is just a fantastic conference. It’s a little bit back to my roots where I used to design chips at some stage. But you know, our announcements are all really about bringing the technology that’s relevant to our customers. So where do we start? We start with our GPUs, our GPU driven servers, both liquid cooled and air cooled and completely all in on the Blackwell transition. That’s number one. Number two is actually associated with that because we all know with AI some of the limiting factors are how do we power data centers, how do we get enough power into data centers, where do we get the energy sources from? And one of the announcements we made was this mod pod which is really a data center. You know, your data center is all kitted out, supports liquid cooling and you can put it in your parking lots, you know, where your employees are still working from home. So a great starter kit. Then on private cloud AI we’ve announced support for the new blueprints, we’ve announced a strategic relationship with Deloitte. But we announced that last year. But now we have a proof point there. Zora AI, a financial AI package for CFOs and the CFO office is now qualified on private cloud AI. We’ve announced, I can’t remember at this stage, we’ve announced a number of ISVs landing through our Unleash AI program and then storage. On storage, we have our new X10000 which is object based. We are supporting the new AI data platform from Nvidia both in our storage but also within our private cloud AI stack. Because if you look at that stack, what you start to see is that the whole end to end is what we’ve really packaged in private cloud AI and then RDMA support with objects supporting the Nvidia in the Nvidia transports. So a lot, a long list.
Patrick Moorhead: I think you just rattled off about 10 of these. Great job. I’m the one who has the notes, not you. But it just came right out. So I like to categorize what you’re doing on your hybrid cloud platform. We’ll get to hybrid AI. It’s the easy button. And with hybrid cloud AI, you took it one step further because there are some customers who want you to do the driving, make this as easy for us as we possibly can. Others which you allow, you kind of mix and match different parts of the layer cake and ecosystem. I think that’s important as well. You’re kind of letting your customers choose how they want to buy from you based on where they are with their sophistication level.
Fidelma Russo: Well, I mean it’s interesting. I just came from a customer meeting. Okay. A particular government customer. They have multiple use cases and some of it is more suited to where they want to mix and match the software stack on top. And they want the ability to change out some of the MLOps software over time because not all MLOps software is applicable for their use case. And then the other use case they have is all around private cloud AI. And that one, you know, when you probe on the customer, why are you picking something different here and why would that be a choice? He said, I have a limited number of data scientists and ML engineers and I want to make sure that in the cases where I really have to get the most out of it, I focus my people there. What I don’t want my people doing on everything else is spending their time getting a stack to work that you can get to work and then I’ll put my value on top of it. So I thought it was very interesting. This is just the way this big customer is thinking about it.
Patrick Moorhead: It is for sure. So I’m curious, I mean as an analyst, I like to sometimes say, you know, it’s kind of like the clown show. I know from show show to show, I go from circus to circus. I think I was on the road 42 weeks last year going to different conferences and meeting with customers and things like that. And sometimes I’m hearing a lot of the same words and I’m hoping. Can you turn up the contrast ratio for our viewers on how your private AI cloud is different from your competitors?
Fidelma Russo: So, you know, I pity you having to go to shows 42 weeks of the year.
Patrick Moorhead: You have the real job here. I had a real job for over 20 years. This one is easier.
Fidelma Russo: Oh, I don’t know.
Patrick Moorhead: I used to do not exactly what you did, but I ran product groups. Right. I mean, what you do is hard.
Fidelma Russo: Well, thank you, but I. I wouldn’t swap it for a moment. So what’s different? Okay, so if you think about, you know, first of all, what do I need to do in order to, you know, to get, to actually even run an AI proof of concept, the amount of work you have to do to think about the sizing of your infrastructure, compute, networking and storage, and tuning all of those pieces to work together. You know, you have too much compute, not enough storage. We all know that that doesn’t work. We have too much storage, not enough compute or GPUs. That’s not going to work. So first of all is the upfront, how do I size it for what I’m thinking about doing? And the next piece is, how do I keep up with taking all of the software from Nvidia, how do I qualify it, how do I make sure it interoperates with the other layers of the stack? And that’s what makes us different in private cloud AI than the other alternatives on the marketplace. Because there and the other alternatives you have to do, the hardware may be sized, but it may not have been done in a way that guarantees the number of tokens you want to have. And then the second piece is the software. Integrating the software is not easy. And then qualifying other people’s stacks on top of it, which is also what we’re doing, isn’t easy either. So our job is to do the easy button and you access this on the Greenlake platform, which also is a differentiator from the rest of the industry.
Patrick Moorhead: I appreciate that. So when I talk with CIOs, this has been very consistent. Aside from the, hey, how do we get ROI, what use cases do we pick? Culture issues. There’s inherently this data conversation, which is, my gosh, I’m having a really hard time getting my data ready for generative AI, but also for agentic AI, where the stakes and the value are higher. What kind of counsel are you giving companies around getting their data management in order? And I’m talking, you know, beyond. I have seven instances of SAP. Okay. There’s even some people that are, hey, I have one instance of SAP. Just an example. I’m having a hard time activating this, so I’ll leave it.
Fidelma Russo: You know, and you’re perfectly correct, okay. We all talk about the stage by which the data is ready. Okay. Because that’s an easier discussion. We’ve all lived through the: I’m going to take all of my data puddles and I’m going to consolidate them into a, you know, one data pool. It’s never worked.
Patrick Moorhead: Never.
Fidelma Russo: It’s never worked. Okay. And, and people have spent a fortune, they’ve spent years, and all of those projects have gone by the wayside. And, even if you could do it, then a company does an acquisition and guess what? More of it comes. And so what our strategy is, it’s federating the data pools. Okay. So, and one of our acquisitions we did a number of years ago, you know, timing is everything in this industry is Ezmeral data fabric. We have tremendous installations. You’ll see in one of our press releases, Vodafone Ziggo, they basically used it to, you know, give a global namespace to all of their data. And now they’re performing AI on top of that. And so, and it supports all sorts of different formats and with private cloud AI, it’s built in. And also the other piece of it is also you can use it on the rest of your data. And the second piece is, we didn’t talk about, is a little bit around observability because if you can’t see into what you’re doing, then you’re running blind. And so, and there’s where, you know, with Op Ramp, we’ve now expanded it to, to basically provide observability on your GPUs. Are they busy? Are they getting fed? Which one? Where should I move the workload? And so all of this has to do with how do you really operationalize AI within the IT organization? How do I get my data, how do I run my agentic AI, but then how do I observe and manage it on day two when everybody has left.
Patrick Moorhead: No, I appreciate that. Oh, I appreciate the words of wisdom here. And sometimes people forget how much software HPE does, but you literally built an entire enterprise stack over years. You’ve created some of the software on your own, you’ve made a lot of acquisitions, you even have a virtualization and a container stack. Again, back to the easy button stack. We can offer it. If you want slices of that to connect, we have that for you.
Patrick Moorhead: So Fidelma, thank you for your time. I really appreciate this and I’m hoping we can get an update with you at Next Discover. But I’ll talk to your team about that. That would be great.
Fidelma Russo: Yeah, no, let’s do that. That’ll be my next show. But it might be your next show.
Patrick Moorhead: Oh, I will be there.
Fidelma Russo: You will be there. All right, we’ll meet then.
Patrick Moorhead: Yeah. Thanks again.
Fidelma Russo: All right, thank you.
Patrick Moorhead: This is Patrick Moorhead here in HPE Nvidia GTC 2025. We are talking about private cloud AI. And by the way, I don’t always do victory laps on my research, but I pretty much said this is going to be a hybrid world, and it’s a hybrid world for clouds and AI clouds. So here we are! Tune into all of our GTC and HPE content and we have HP research papers too, on Moorinsightsandstrategy.com. Take care. Bye bye.
Author Information
Six Five Media is a joint venture of two top-ranked analyst firms, The Futurum Group and Moor Insights & Strategy. Six Five provides high-quality, insightful, and credible analyses of the tech landscape in video format. Our team of analysts sit with the world’s most respected leaders and professionals to discuss all things technology with a focus on digital transformation and innovation.
