HPE’s Antonio Neri & Axios’ Courtenay Brown

Congrats.

We've been talking about AI for a year plus, even last year when we were here. First of all, thank you for inviting me and good afternoon everyone. I hope you will have a great week here at Davos. Looks like the weather will be benign all the way to Friday. So that's good. No, I mean obviously AI continue to be a topic of discussion now more from an idea to now being deployed and with the challenges we see in the world now stepping up into more the conversation about regulation, responsibility and obviously sovereignty.

No, I think we serve customers around 172 countries and we serve every segment of the market from companies that build models from generative AI to agentic AI to physical AI, all the way to governments who are deploying AI in the sovereign borders and then obviously the large enterprise. And what we see now is that the percentage of inferencing, because you talk about AI, you talk about training and you're talking about inferencing, the percentage of inferencing is growing very rapidly, which tell us that obviously it's being deployed for different reasons for production or for delivering specific outcomes. And so that's a very good sign which means customers are finding a return on investment and they're finding the value of it. But new challenges come all the time, you know, space, power, cooling, bias, all these things have to be dealt as you go and then regulation and compliance.

I think a year or two ago, people were still trying to figure out what AI was going to do. I think now they realized as well with some of the tensions we see around the world, through national security to geopolitical aspects that now you have to take control of your destiny. The potential for AI is there, no question about it. But there are other challenges you have to solve. Customers have been going through the digital journey and COVID was an accelerator of that. Automation now is at a new level but when you think about AI it's all about the data and how you use that data. And so in the past we talk about data sovereignty, we talk about here in Europe GDPR and the like, but now it's all about how I control what data is being used for these models and making sure the provenance that is protected. So I think it's definitely a topic as well as where the models are coming from, what data have been used to train them in the past in the public domain and therefore controlling the model also has an aspect of controlling your culture.

Yeah. If you think about some of the use cases in the past and thinking about controlling data, I actually think there is a way to share insights and knowledge without sharing the data. And so we have deployed in our company working with unique customers models that they are learning but they are not actually sharing that data, they are sharing the learning. That's what we call things like swarm learning. For example, that use case was deployed here in Europe with an entity called DZNE in Germany who is doing a long-term research to solve issues like Alzheimer and dementia. But you want to protect the data in some cases strict regulatory environments associated with that data. But you still want to share the knowledge so we all can proceed and progress much faster. So that's why that distributed AI learning technique is something that I believe is going to take hold. And if you think about the agentic processes, right? So you think about the ability for agents to learn without accessing data is very important.

Well, first of all, there is a scale problem. There is also a cost problem. There is an energy problem and then there is a sovereignty problem too obviously that you have to deal with it. So, to me it's not different than the cloud journey. We live in a hybrid world and I think over time we're going to live in a hybrid world also in AI. I argue that AI at the core is the typical definition of a hybrid workload and therefore be able to access either data in your own environments in a distributed environment because once you move data you are exposed, you waste energy, you waste cost, but at the same time sharing the learning with other industries is going to be an element if the right ecosystems and policies are established.

Well, you know, it started with the global economy, decoupling the global economy, right? But I do believe... but it was still operating in a global economy, believe it or not, right? So I do believe that at the core at the society and that's why the WEF is so important because it's this mission to make the world better through dialogue, to collaboration, to shaping the right policies. There is absolutely space to drive collaboration and learning. It just starts with the will and the intention.

Well, there are formal and informal environments, settings where you do that. Obviously in the United States through the government we have a lot of conversations between the industry and the government. But look, we have relationships with each other, we respect each other and we have a lot of discussions. But ultimately you're here to serve shareholders and your customers and be able to solve problems better than anyone else. But definitely make no mistake when we get together we actually share a lot. In fact here at EBC we'll be talking about some of these issues.

Well, I think it's going to be mostly around the geopolitical environment and then the use of technologies to solve some of these challenges.

I made a lot of mistakes in my life and I learned a lot.

Yeah. Well, first I will say the biggest mistake you make is not to get on board and wait. I think that's the... you know we have a saying in the company that the future belongs to the fast. And if one event that taught us a lot was COVID, people that were digitized and were automated were better prepared to deal with that challenge. So I think the first mistake is not to get aggressive in learning and understanding this technology and start using it where it makes sense. Obviously you have to have processes, governance and all of that but I think that's mistake number one I think many companies may be making today. Second is, you know, look, we as a company have unique set of culture and values and I will stay aligned and true to those. We developed an ethical code for AI which is public and we are sharing it for ourselves and with how we develop the solutions for our customers. But ultimately it comes down to really embracing it because this has the potential to change many things.