The 65 is on the road here with a view from Davos. We're at the World Economic Forum and as you would expect, a lot of discussions around policy, about technology, AI in particular, and we're going to have a lot of discussions about tariffs as well.
Yeah. Well, you know, it wouldn't be a great World Economic Forum if there wasn't some escalation unexpected right into it before President Trump comes here and speaks. But, you know, this event has become really one of the prime moments for having conversations about what's going on in these inflections in technology. Pat, and AI is on the rise. And we know that this is not only the year for we're kind of beyond the ChatGPTs and the LLMs moment. Now, we're really thinking about sovereign AI. We're thinking about enterprise AI. It's a really big year.
That's right. We had a great discussion with CEO and president of HPE Antonio Neri on that last year and I think this is a great time to get a follow-up. Great to see you.
Well, thank you for having me.
Excellent. How is this World Economic Forum different this year than last year?
Well, first it's clear that there is way more people than last year. The attendance is up quite significantly. As you both said, it's an important moment in time because a lot has happened around the world not just on the technology front from a geopolitical perspective. So I think there is a tremendous amount of expectations of what's going to be said when the governments show up between today and tomorrow and Thursday. So it's the right place to be to hear directly from the leaders how they think about the future not just from the technology perspective from the geopolitical standpoint.
For sure. Well, all that influences the business, the decisions, how you think about supply chains, manufacturing, how you think about making investments in different parts of the world because of course HPE's a global company with great scale. Speaking of great scale though, in terms of the AI and cloud space, we're seeing hyperscalers increasingly become kind of the kings of compute. All of the compute seems to be really moving that way, but we're increasingly, I know Pat and I are both increasingly hearing from enterprises that they want to possibly think about how they distribute risk, how they mitigate potential risk. I know you're having conversations because you're one of those suppliers that helps enterprises put AI on-prem or in more hybrid environments. Talk a little bit about how those conversations are going, a little bit about what the risks are of maybe overconcentration, what are enterprises doing to serve.
Well, I think for many enterprises it has moved from exploration to deployment. And as they went through that phase, we see the growth in inferencing growing very rapidly, which tells you that the models are being put to work, whether it's large language models or I believe agentic AI is going to be the future of the enterprise. Second is where they do the training really matters and obviously the concept of sovereignty and protecting the data is a core tenet of that discretion. But as we always said, AI is the true definition of a hybrid workload and we believe in the end they're going to use all aspects of infrastructure, some in a public cloud, some on-prem, and then I think there will be industries that are going to come together particularly here in Europe where they're going to use sovereign AI clouds because the investment to build this large AI infrastructure is quite high and the energy demands are so high as well. So I think there needs to be some sort of collaboration perhaps at the vertical level.
So I've heard some talk about having sovereign data strategy as being too introverted and potentially could lead to lack of access to a global market. I was thinking that you can actually have your cake and eat it too. You can do both. Can you talk a little bit about that, how your clients are looking at that, maybe also how HPE is enabling that?
Well, I think with the technologies we have been working on for a number of years now you can provide secure access to the data. At the same time, I don't think it's all just about the data. It's about sharing learnings. And with some of the latest technologies like MCP and so forth, you can actually extract the learning of that data much faster. You can share just the learning, not the actual data. So I think it's a combination of different techniques. But I think Europe right now is taking a very aggressive approach on regulation and protecting that data in the context of the geopolitical situation we live in and we see a tremendous growth in demand from the concept of thinking about it to now let's go plan and eventually run the traditional RFP process to build these AI clouds and we see it in Europe and we see it in the Middle East.
And so net-net for these customers who are doing that, it's not holding them back to go global?
Absolutely. No, because they're going to have a lot of access and choice where to do work. I think depending on the industries, you're going to do it closer to home and then when you go to that distributed learning, you're going to do it where the action is depending on where your footprint is.
Yeah, it makes sense. And sometimes I ask questions that I think I know the answer to, but the audience doesn't. So, thanks.
Pat's always a generous host here. Wants to make sure everybody gets a little extra education. But, speaking of Sovereign, right, one of the things that's going to drive data centers to be built in different locations. Another thing that's going to drive data centers is something that HPE's long been very focused on, and that's edge. You know, there was kind of the first IoT and edge era. You were certainly one of the leaders in that space. I think AI's actually brought that back to the top. It's actually more in focus physical AI, edge AI, how are you sort of thinking and talking not just to customers here, but really all of your enterprise customers about the placement of those workloads? What's driving enterprises to say, 'Okay, we're going to do this in the cloud, but we're going to build these edge data centers. How is that scaling? How are you seeing that being adopted?'
I think it's a combination of driven by demand and by physics. You know, in 2018 I stated that the enterprise of the future will be edge-centric, cloud-enabled, and data-driven. That edge started with the on-ramp to digital and the first step was to lay the right connectivity and it needed to be secure. We definitely focused on that. We built a world-class portfolio which now we extended through the acquisition of Juniper Networks from the edge to the cloud. And when I think about the edge, think about the use cases, manufacturing floors, hospitals, stadiums, venues. And that starts with connectivity. But now you can bring the right AI model and mostly it's going to be agentic AI to that edge to do the right inferencing because latency matters, cost matters, and ultimately delivering real-time decisions is going to be very essential. So that's why we see the growth in inferencing and that does not require huge amount of infrastructure. In fact, it could be just one server with GPUs with the right agentic models built around it and then you're going to decide whether you do RAG or fine-tuning of a model or actually build a model yourself which may be necessary for some of the use cases of specific industries and then you decide where to do it. You can do it in one of these locations that you have access to GPU per hour basis on a demand basis or you do it on-prem because you bring that model, you want to train it with your own data and trying to protect that. Either way is a hybrid approach and for enterprises we deliver a unique value proposition to our GreenLake cloud which ultimately allows you to do both in a very unified and AI-driven way because we also use AI inside the platform to lower the cost and give you a faster return to that investment.
Are you seeing really quickly here? Are you seeing this demand for edge and bringing these workloads on-prem? Is it growing?
Yeah. And we see that through a number of offerings that we introduce in the market. You are familiar with our Private Cloud AI which is a tightly coupled stack that we develop with NVIDIA and our GreenLake cloud and ultimately our infrastructure and we see that growth in number of adoption, right, number of logos and number of systems being deployed. But then also you see it in a single unit type of deployments with just our servers. But then also we see that growth also on the data side with our storage array platforms because ultimately you need a lot of unstructured data.
