Welcome back everyone. I'm John Furry, your host of the cube. This is our AI factory series where we unpack all the technology that's driving the next generation infrastructure. AI infrastructure is the hottest sector of the market as innovation and also development is going to enable this next wave of tokens, agents and then physical AI as this all comes together. Of course, we're talking to all the leaders here. We got Giordano Albertazzi who's here, the CEO of Vertiv. Welcome to the cube.
Well, thank you for having me, John.
I was chatting with you before we came on camera about telecom and then the big data. You guys have been doing a ton of work across all major critical infrastructures, data centers. I should have said on the intro is the hottest area. Everyone's building $50 billion. So, Anthropic had news out. They're going to they say 50 billion, but whether that happens or not, we'll see.
Share your thoughts on what you guys are doing. First explain what business you're in because you guys are in the center of the action on where this money is being spent but also where the critical ingredient energy is involved. Talk about what you guys do.
Yeah, absolutely. Well, thank you for having me here. Vertiv is a global leader in digital critical infrastructure. So I think everything infrastructural in a data center or in a telecom, as we were talking about telecom network, is around power and cooling and all the physical infrastructure around the IT. It needs to be cooled. That heat is not just about extracting from the IT but needs to be then reused or anyway expelled from the building. And then you have to take power and bring the power in the various stages, probably medium voltage, and take it to the chip. That's what we do. And everything, you know, sometimes even system level integration and a lot of service to follow a customer during the lifecycle of their asset, global.
Talk about the business. You guys are NYSE listed company so we'd love to have you on the cube and our NYSE wired program and community. They've been there's been a lot of active discussions in this peer group around architecture. Talk about what you guys do on the business side. What's the core products? How's business going? What are some of the stats?
Business is going strong. Strong market, but we believe we're growing more than the market, so gaining market share. You know, last week or a few days ago we guided our end of the third quarter to a 27% year-on-year growth. So both in terms of top line, in terms of margin, we're taking numbers up. So very pleased with the direction of travel, but even more pleased about our role in an industry that is growing and our role is really enabling the industry. Without power, cooling, the physical infrastructure, it's hard to think about a data center, it's hard to think about an AI factory. So the centrality of what we do is absolutely undisputably important.
You know, one of the trends that I think is your friend is obviously this trend towards physical AI. And if you look at like Jensen Huang at Nvidia, he was just speaking at the they had a mini GTC in DC because it's getting very political. Everyone wants their GB300s. The buildout is not just getting started. There's a lot of headroom. So obviously the data centers are being kind of designed from ground zero to be basically supercomputers. Existing data centers that don't necessarily have the footprint design from with a clean sheet of paper, kind of a brownfield as they say, have to be retrofitted and reworked. Then you have other opportunities that are emerging on infrastructure.
What is your vision as AI factories get built? You got these centralized power sources. They have to talk to other distributed nodes in the network. I mean it's distributed computing. This is what we're talking about. So you got all kinds of physical digital things happening. How do you see that market? And as a CEO, but also with your partners, what's the formation? Is there an ecosystem forming? How do you see that market?
Well, the ecosystem is really between kind of a silicon and clearly Nvidia first and foremost, colo, hyperscale, and the likes of ourselves. We like to think of ourselves as the connecting tissue between all these players. And the buildout of course is happening and if anything it is accelerating. It's accelerating certainly in the large data center gigascale and sometimes beyond, but also in a more distributed, smaller footprint, smaller size. It is pretty ubiquitous. Power and cooling has had a comeback. It never really went anywhere, but if you look at go back 5 years ago there was some pretty much state-of-the-art power and cooling, liquid cooling, all kinds of new things. And now it's become such the center point of all the economics and the design side. What's changed in power and cooling as a category over the past say four years? What's the biggest change? Is it just more action, different designs? What's your view on that?
Yeah, well, I think about been in the industry myself for decades. So you move from a period in which things were not dramatically changing from a type of IT loads to a period now, the last few years, as you were saying, where density, power density has increased dramatically. And that is the quantum change in terms of challenges and opportunities. From a cooling standpoint, you have to move to liquid cooling, something that a few years ago didn't even exist, well existed but it's a very niche, niche scale. And the same is happening on the power side. So the whole conversation about high voltages is really to enable this continuous ongoing densification of loads, which is a phenomenal opportunity because with the densification of loads you have way much higher performance.
Performance of the super computing exactly. Okay. So that the whole industry is changing radically. It's super interesting for us and of course we are at the center of that and we are leading together with Nvidia and some of their customers, most of our customers really, this transformation. This transformation is not about the technology you have today.
Today the technology you have today is very important, but the technology you're working on two, three years out is even more important long term. So being there at that table and being an actor —
Super impressive.
Yeah, you guys are in a good position as they say in your middle of the fairway, use golf analogy, or pole position if you're on the racetrack. I want to ask you about the system side of it. About four years ago we were reporting, Dave Volante and our research team and I were saying large scale clustered systems or we were using I think you word supercluster. It was not yet clear that the supercomputing would kind of come out of, you know, Nvidia going, 'Hey, let's connect InfiniBand to Blackwell and look what happened.' And then all that software that they wrapped around to really kind of change the game. That densification is now a system architecture. You guys have this system thinking. You have to think like a system architect because you are the connective tissue. You're a control point. You have to write code and technology to know when to do something with power and cooling. You're in the critical machine layer. Absolutely. You're like right there where everyone wants to be.
Absolutely. It's this multi-layer thing. I mean, of course, in a critical infrastructure, be it, we simplify talking about a power train and thermal chain if you will. Every element of this train chain needs to be optimized from a technology standpoint. But then it is how the system works together that makes a difference because you need to really think about the system, optimize the system, and deploy rapidly the system. And then there is a software layer that controls and optimizes the way the various parts of this complex system work together: the power, the thermal, the IT loads. Now Vertiv Unifi is what we use for that and it's again another layer of competitive advantage if you will.
Yeah. Not it's very nuanced but I do want to highlight that value you guys have because if you look at all the buzz going on and some of the core conversations where the real money is being made, the word interconnect is used. Yeah. Yeah,
You know, interconnects, co-packaged optics. So, as the hardware starts getting smarter and closer and intelligent, you're essentially an interconnecting things.
Very much. So, that's —
Not in the classic definition, but you're in a network. But the ability to have perfect interoperability of all the elements of the physical infrastructure is so important for the efficiency and ability to scale the data center.
I know you get a couple acquisitions looking at the news here, but I want to talk about your relationship with Nvidia. Really bullish on Omniverse. I think the digital twins work that they're doing is going to usher in way more efficiencies. So I want to get your thoughts on Nvidia Omniverse and how you guys are looking at that and partnering with them. And then also throw a little futuristic question out at you: not to date myself, but grid computing was a concept in the '90s. Okay.
You know, use some cycles from one PC. It's kind of a, you know, but this idea of grid energy —
And the way you guys are looking at this, I could see potentially scenarios where energy could be managed from these large systems —
In a very effective way. And it sounds like you guys are doing that now at many levels, are kind of in that world. Are you in that world of gridding out or connecting energy?
Yeah. Two big topics. Exactly. Let me use the next two hours to address. So for the first, clearly Omniverse is extremely important. It's a vision of where gigascale the industry is going. And the fact that we are prominently visible in everything Nvidia material, discourse, presentation is testament to our role, not just the role in the industry, but role in terms of partnership. And again, the partnership is not just about having the technology or the ability to scale today, super important, but working together with them, you guys co-designing —
Co-designing exactly what do we want to achieve two years from now. Don't forget that the type of infrastructure that we make available needs to be there ahead of when the new GPU versions or models are deployed. Because the physical infrastructure needs to be there. So being ahead is extremely important.
Talk about extreme co-design. That's a big topic of Nvidia. They've actually pioneered with you guys this notion of actually leaning in, not just doing some sort of business deal, 'hey we're partnering, get this API or you this code.' Design is an intentional NVIDIA strategy. You guys are in that. Explain what that means.
