Mark Delaney: That’s helpful context. My other question was thinking about the addressable market opportunity. You talked about the full powertrain on Slide 5 of your deck and showed various — a variety of opportunities to address. Maybe you can help us understand how much of that addressable market Vertiv can serve today. And are there areas you can augment with either R&D or perhaps tuck-in M&A to improve the same you’re addressing?
Giordano Albertazzi: Two aspects. We’ve been vocal about what we think the TAM per megawatt is probably the easiest thing to say. We believe it will be available for us. I want to remind everyone, we believe we have the most complete portfolio of digital critical infrastructure. But again, a TAM that is on the traditional, let’s say, data center design at $2.5 million to $3 million per megawatt going up $0.5 million with AI and high density. That’s what we see.
And of course, as you know and as we said several times, we are continuing to invest in engineering and R&D, a CAGR of approximately 13%. And clearly, acquisitions — and acquisitions, certainly, as we have demonstrated with CoolTera, that enabled us to add technologies to our portfolio is something that is a part of our capital allocation strategy. But again, it’s never binary. It’s like technologies that we work on organically, and sometimes that organic then matches very well and inorganic opportunity. And again, instead of speculating about the future, I would refer back to what we did with CoolTera and liquid cooling by which we had both in indirect — sorry in inorganic and organic, let’s say, orchestrated that.
Operator: Our next question comes from Andrew Obin of Bank of America.
Andrew Obin: You guys put out a press release on the NVIDIA relationship. Can you comment what’s exactly the nature of the relationship? Are you being standardized from that product? And also, how are you guys getting ready for the Blackwell product? Do you need to do anything technologically to be ready for higher-power conduction there?
Giordano Albertazzi: Well, it’s a multifaceted relationship. And in a nutshell I would say that’s certainly been part of the network — partner network of NVIDIA, which is certainly essential from a commercial and go-to-market standpoint. There is an element of certification of our liquid cooling, but very, very importantly is the relationship that we have working together at engineering level. So still very, very important.
When it comes to the future technologies and specifically Blackwell technology, it’s clear. That’s a technology that requires liquid cooling. So that’s also the fact that we have now in the market a clear leader with something that is vocally liquid cooled as a chip. That will drive liquid cooling into band. And if you want to draw — connect the dots with what I said about the liquid cooling that, that is legitimate.
Now in terms of our portfolio, our portfolio is the right portfolio for the next generation. And next generation today — there will be more generations, and we talked in several occasions about certain level going to 2-phase instead of just single-phase liquid cooling. And again, as we been vocally and publicly explaining, we are on that part of the technology, though.
Andrew Obin: Got you. And as we think about the ramp of this new technology, right? Because I think as of last year, it was a fairly small percentage of your portfolio. And I completely understand how you are very well positioned to continue to have very strong position there from a technology standpoint. But is this ramp in technology? Do you need to invest? And the question we get from investors basically, given that it’s brand-new technology, should we expect it to be margin-dilutive, at least in the first stage as you ramp up to full production?
Giordano Albertazzi: So the fact it was small in the past in general for the industry is that the current generations can very well be air cooled. So the majority of deployments of GPU were air-cooled using the infrastructure. And that’s true across the industry. I’m not talking about Vertiv here. When it comes to the investment in CDU capacity, and I refer back to our February earnings call, well, that is pretty well explained, and we are progressing in that capacity growth. When it comes to margins, it’s a new technology. What we indicated already back then is the margin profile, I would say, is consistent with our thermal business, certainly consistent with the long-term trajectory to 20-plus percent that we shared with you as our long-term goal.
Operator: Our next question comes from Nigel Coe of Wolfe Research.
Nigel Coe: So there’s a lot of — quite a few investor questions out there on the backlog aging. How much of the backlog is set for 2025 and beyond? So maybe some color on that would be helpful. My question was really more about maybe some color on the on-prem. Gio, you mentioned, obviously, a lot of the focus right now is on the AI GPUs and the hyperscales. But what we’ve seen on the on-prem side? And then finally, the other thing that caught my attention was the capacity increase on switchgear. So I’m just wondering what that doubling in capacity sort of underwrites in terms of your outlook for E&I over the next couple of years.
Giordano Albertazzi: 3-in-1. Thank you, Nigel. So when it comes to the backlog agent, we will not go too much in the details about — the indication we have is we have an element of a strategy a little bit of — on the requested delivery dates and lead times that makes our coverage longer. On-prem versus hyperscale, certainly, we see, at this stage, the acceleration and dominantly in hyperscale and colo type of cloud for high density and AI. But I mentioned Data Center World, that’s predominantly and mainly an enterprise type of show. It was a very, very, very crowded. So I leave it to that. When it comes to switchgear and busways/busbars, we’re very happy. And again, that is a testament to the importance of the acquisition to make sure that we really have the entire portfolio and the entire powertrain from medium-voltage switchgear, all the way to distribution and inside-track PDUs.
