Kick it off here. Good morning, and thank you for joining us. My name is Luke Junk. I am Baird’s senior research analyst covering electrical and AI infrastructure. We are very pleased to have Vertiv with us this morning, a global leader in enabling the AI revolution across power, thermal, and prefabricated solutions, plus services. With us, we have Gio Albertazzi, CEO, and Craig Chamberlin, CFO. If you would like to take part, you can send an email up here, that address, session2@rwbaird.com. With that, let us jump right into Q&A. I want to start with the big picture, Gio. Over the past six months, 12 months, or maybe even the past week, can we just talk about what has most increased your confidence in the durability of AI-related demand or things that might be risks that could moderate the outlook from here?
Yeah. I will certainly go back to what we said a couple of weeks back during our earnings call. It is a combination of things, of course. First and foremost, always start with the pipeline, with the volume of business you see through your commercial, that we see through our commercial activity, but also more generally speaking with our customers. As I was saying, the speed of growth of our pipeline has increased, and the pipeline itself has increased. This is something that happened not only in North America, but globally, as we shared with all. Fundamentally and basically, it is the demand for AI compute and compute in general, not just AI, even the more traditional cloud compute. That demand is not stopping. That demand is also very clear when we look at the Google, when we look at the AWS earnings.
That part of the business is the part of the business that is growing the fastest, and the monetization is happening. If you look at it from a growth and adoption in AI, that certainly is at the very beginning and continue to expand. If you look at it from a pipeline, if you look at it the way our customers and the industry talks about the opportunities, the projects that are ongoing globally, that all converges. I think that trend is the underlying trend, and it is very much resilient. It is a very resilient trend. So we feel good about it.
Yeah. Coming out of the midterms, is there any particular policy or regulatory change that you are looking at that could be impactful or these pacing things or things that you guys can lean into as well? Is there any customer conversations that have already started post-midterms in terms of evolving to the environment?
Well, I go back to the underlying demand. The underlying demand for AI compute is very, very strong. We see it from enterprise, we see it from the hyperscalers, we see it also from the AI labs. Of course, there are a lot of news out there that are very, very encouraging in that respect. The environment is creating, as we have said in several occasions, pacing items. Two, three years ago, we were talking about power as the big ceiling. Some people will say, "Hey, the big ceiling that will stop the deployment of data center, of capacity, of compute." Well, know what? That didn't happen because the industry is very resourceful and finds way to move on. If there is demand and demand is out there, then the industry knows how to react. It's early to say, of course.
It's hardly a week into the midterms. But we think about whatever may happen, there will be pacing items, and the industry will be able to react to those pacing items. We feel good about the direction of the industry and the strength of the demand.
I want to probe a recent transaction that you just closed, the UtilityInnovation Group. How does UIG help getting you into those customer architecture discussions earlier, either translate into more content or just more visibility into where your pipeline is going to go?
I think all three, but let me start from another point, is that, again, we're talking about power as a pacing item.
Power also a pacing item that is very visible out there to the public. The industry is evolving. The industry is evolving with a lot of behind-the-meter power generation, behind-the-meter power generation that orchestrates with the grid in a way that makes also the power generation in the data center adding and supporting the grid for the larger public. It all happens behind the meter, and the orchestration between behind-the-meter power generation and the grid is an essential part of this equation. There is a lot of technology involved in this orchestration. It is not just the fuel cell, it is not just the gas turbine and the grid. It is how you orchestrate all these elements and how you then manage your load. If you think about this technology, that is the technology that UIG provides.
You can look at it, a big enabler for a more active role of data center in the grid in general, or a very important technology element to make a data center more independent from the grid, the entire spectrum of types of dynamics there. But then it adds hardware, controlled engineering, application engineering, all things that enable the next wave of evolution from a power standpoint of the industry. Yet it is true, that gives us very early visibility and opportunities because the first thing you do is when you plan a data center, land and power is where you start. Land and power is really understanding how you orchestrate the behind-the-meter and the grid, and that orchestration is uniquely engineered around the individual site. That gives us a lot of visibility.
It gives us, again, it helps us in terms of dollars per megawatt. Think about it as an extension of our power train. Every time we add a segment, and this segment is upstream from what we traditionally started.
Yeah. Want to get into some more specific things here, starting with your prefabrication business and execution, certainly a focus area going into the third quarter. Can we just talk about what went right operationally in the third quarter to drive the beat and where you are seeing incremental areas of operational progress in that business right now?
Well, as we said in the second quarter, and certainly in our earnings a couple of weeks back, first of all, we do not like to call it prefabrication because prefabrication means that there is a fabrication on site. No, we really think about a manufacturing of data center modules or manufacture of an entire data center. When you do that, you remove from the field, from the construction site, a lot of complexity. But of course, you have to build a much more multidimensional system and structure, internal processes to really deliver all that value. Modularity is essential, is a very strong trend to accelerate things on site, as we all know.
So, that kind of bringing that complexity inside the four walls of, or maybe more than four walls of our factories and systems, liberates the field from complexity and enables us to industrialize a process that had never been industrialized before. That has a learning curve. That is the learning curve that we described. That is the learning curve that we described at the end of the second quarter. So, we are progressing in that learning curve, and we are pretty pleased with the direction of travel. So, that is really the evolution.
Yeah. Craig, I want to probe this a little bit more because it is probably the number one thing I have gotten questions on in the last three months. Just bringing that standardization, bringing industrialization to this business, can you just bring to life what that means? Is it product design? Is it the design of your manufacturing, your internal supply chain, commissioning? Just what are some of the particulars or maybe an example of something that you have really honed on incrementally in the last three or four months here?
No, I think whenever you talk about what Gio was saying is bringing everything into one factory, what you got to understand is the pace at which that happens, and you create a project-level business. It is different than your product-level businesses. Your product-level businesses are build, output, build, output, build, output. Your project-level businesses have a lot of complexities in terms of what they do to drive your product businesses into there, ensure that you have the right system-level outcome, and then you can create the output for your customer. So what we have done in the last, and we talked about this in the first quarter and throughout second quarter, was ensuring that we had the right redundancies and the right capacity to, what I would say, buffer that ultimate complexities that you would have in a normal product business.
If you think of a product business that has standard items that happen during a manufacturing process, you can normally pick that back up pretty quickly. But when you start to see that happen over and over again into a factory that is a system-level factory, you might end up with multiple different items that are impacting that project outcome. What you have to do is work from there backwards and create the right level of redundancies, the right level of what we would call safety stock in your inventory process, and the right communication between the sites. This is where Gio's saying of getting up the learning curve, it is learning curve on communication and capability as well. If there is an engineering change, how do you get that communication back and forth?
If there is a product change, how do you get that information back and forth quickly so that the product, what I am going to say, the product flow into the project business stays standard and stays ongoing? Then you create an output that looks more like a product output. It is regular, it is rhythmic, it is on time, it does not have complexities that are influencing it, and you have the right resiliency and redundancy behind it. That is where we are saying we are continuing to work up that learning curve, because it is not something that is, I would say, ultimately new to our business, but it is a different level of complexity. We have done power mods before, but at a much smaller scale. We have done converged infrastructure at a much smaller scale.
These are much larger with a lot more inputs, so you have to learn as you go and build the resiliencies and redundancies in, and that is what we have done. I feel like that is what you saw in the second quarter, was us catching back up on the converged infrastructure, but the additional capacity allowing us to even serve more customers on the product portion, which is where the beat came from.
Yeah. I want to pivot the conversation to power, and of course, 800 VDC is part of that conversation. If we click back six months ago to Investor Day, you are discussing and outlining some different customer approaches, including solid-state transformer-based approaches, as well as some other ideas in terms of using, say, medium voltage UPS. Just how has thinking evolved since Investor Day, and as we move closer to commercialization, kind of how do you see the mix of that opportunity right now?
Well, the architectures that we shared with you in May are the way we see the market continue to evolve. We will see, to make the story extremely simple, we have on the one end of the spectrum the traditional low voltage AC infrastructure, all the way to the other end of the spectrum, an entire 800-Volt DC native end-to-end type of infrastructure. The same, let's say, rules of the game or who will apply what, who will adopt what, that we shared with you are still there. Indeed, that what we shared has been validated with the many conversations with our customers and with, of course, the silicon provider. That is where the industry is going. That spectrum is very important.
Of course, it is very important to be very strong on the leading edge of the new technologies, as we are extremely strong on the traditional, and that there is a lot of focus there. We are pleased with the direction of travel. We are pleased with the product roadmaps and deployments and the technology that we are testing now or validating with our customers. When it comes to the way it evolves, again, going back to May, do not think about a future in which everything is 800- Volt DC. That will really depend on the mission of the asset. If the asset is a colo that has to have 20 years of flexible resilience, then it will be probably more similar to a traditional infrastructure and power architecture.
If it is a very focused AI factory on Vera Rubin and above, that will probably be designed as a native end-to-end 800- Volt DC. When it comes to the timelines, we continue to believe, and we continue to see it even more so now, that 2027, second half of 2027, is when we will start to have sidecars. That hybrid infrastructure that converts to 800- Volt DC in proximity of the IT load, and probably it will be the second half of 2028 when you will start to see end-to-end full 800- Volt DC. The key is the load.
Yeah.
The load will start to require and demand at rack and row level, 800- Volt DC.
Craig, what about investment in this business? Is there potential that M&A could strengthen the power portfolio as we sort of build out the breadth of product for 800 volt, and how do you just weigh acquisitions versus internal investments in the power business more broadly?
Yeah, I mean, I think what you've seen us do in the past, and what I think we'll continue to do, is we like dual paths. We don't want to be, I'd say, all inorganic or all organic. We like to look at both scenarios. What we've done both in this area where we're continuing to build out our strength in 800 volt, and along with other spots, even like the microgrid, is we'll start down a path of an organic development, and we'll also keep our eye on the inorganic path and see if they can merge together or if there's one that's going to a more accelerated path or a better outcome for us. I mean, you saw that with UIG. I mean, we had a microgrid business we were building.
We saw an opportunity to go out and get something that would accelerate us and marry those together, and we went and made that choice. I think that was a great choice for us because we had some technology that we can go and marry up with what UIG can do, and UIG had a very strong foothold on what they were able to do and their architectures behind that. I think it's a very similar story when you think of something like 800 volt, where we're looking at the inorganic plays that could make us stronger there around all the different infrastructures that you would need, and we're also developing the arm and the strength internally through Scott and his team. So it's a strategy that I believe has played out well for us in the past. I'd say we continue to look at it that way.
It's always evaluating how fast can we get there and how fast can we meet our customers' needs, and you look at that as a trade-off in terms of the value you can add versus an inorganic play versus an organic play, and you marry those two together. To wrap that one up, I would say that's the way that we will do it going forward. I don't think it'll change our strategy because we like to be able to look at both of those kind of scenarios. That's not just power. That's also the way we look at it on thermal or on converged infrastructure. All of our businesses, we evaluate that different paths the same way.
Yeah. Sticking with power, clearly a lot of growth not only in power, but the company overall.
Yeah.
A lot of innovation and new product introduction, quality and moving things towards something that is more steady state production, also in focus. Can we just talk about quality control measures that you are putting in place incrementally to allow you to scale as rapidly as you are right now, Craig?
No, I would think in terms of just manufacturing controls in and of itself and quality controls, we talk about capability and building additional strengths there. We have brought on, I would say, lots of levels of capability in terms of our quality control and our manufacturing. I mean, you saw Mike being brought on in terms of the role that he's coming on. I think he brings a different perspective in what I would say globalization and looking at a global supply chain and what good looks like from a need to induce quality through a, what I'd say, an efficient manufacturing processes. He gets that a lot from the automotive industry. Frieda's in a similar situation when we brought her on. Again, it gives us a different view of what good looks like.
We've continued that down through what I would say is our entire bench on our quality and manufacturing processes. Again, it comes from my own background in terms of understanding global supply chains, what quality looks like differently, what quality looks like, I would say, on a scale basis. If you look at where this business has come, it's come from a $4 billion business back in 2020. Now we're at almost a $14.5 billion business this year, and that requires a different level of skill and thought and overall, I'd say evaluation in terms of how you're looking at your manufacturing processes, your quality processes. I think we had some of that internally, but we needed to add to that and add capability to that and add a different perspective to that.
I think that's what Gio and I have invested in, not just the capacitization of it or the lean capacity that we've added in the manufacturing shops, but also the capabilities that we've added with different people.
Yeah. Adding a couple of points. Clearly, scale requires and has required a continuing evolution of everything execution.
On the manufacturing, the planning, material management, quality, and whatnot. I've been on this journey for quite some time. We are vocal and we continue to be vocal about the fact that quality and reliability is a competitive advantage for us, but that means also that you continue to evolve as the size and the complexity of the things you do evolves. One thing that I always like to draw a lot of attention, this is an industry that brings a lot of new products, sometimes new products, net new products to the market. When you have net new products in the market, one thing is very, very important, not only the quality that you deliver at every station in manufacturing or in assembly or at the end-of-line test, but what happens in field. We have a metric that we follow obsessively. It is the field first-time yield.
Especially when a product is new, we have an extremely strong service organization, and we pretty much commission every single product. We have obsessive loop closing going on, especially on new products, because it's that commissioning when you capture if there's something on a new product that you might not have captured along the way, and you close that loop rapidly with a factory before that hits a customer. That kind of a reinforcing loop and improvement loop is extremely important. It's, for example, a metric that I personally follow quite obsessively.
I like it a lot.
I want to move the conversation to your thermal portfolio. Maybe a bigger picture question to start, Gio, in terms of incremental thermal opportunities that you see right now being unlocked by, of course, what is rapidly rising compute and power density on rack, in the row, in facilities more broadly. Then maybe we could dovetail that with the competitive environment, because certainly there is a lot of investment coming into the thermal space right now, especially, it seems like a lot of CDUs from traditional HVAC players, and just how you look at those incremental opportunities versus maybe the competitive temperature rising a little bit as well.
Yeah. There is a lot packed in your question. Let me see, and if I may miss one or two—
Why don't we start with—
—of the, and the—
—incremental opportunity and—
Yeah
—and yeah, complexity.
Well, certainly we see a market that has now adopted quite broadly liquid cooling. The penetration is increasing. We are very happy. We are a leader in liquid cooling, and the direction of travel of the technology certainly continue to reinforce. It doesn't mean that liquid cooling is at the expense of air cooling. It is a mix change, but there is still a lot of air cooling. What matters is the liquid and air and heat rejection as a system. We've been talking about how crowded CDU space has become for the last three and a half years, probably before now.
Yeah.
Fact is that it is not as easy as it looks, and it doesn't look easy to start with. Again, it's about the technology, it's about the scale, it's about the services that you can deploy, because these are extremely critical elements in the system. It's like your blood vessels of your AI compute. You can't afford downtime. If something goes wrong, you have to have the best services teams available to fix. It takes scale, it takes a lot of technology, it takes a lot of service capability. These are not easy to replicate. The CDU business is consolidating. Some of the larger players have made acquisitions, as we have seen. Well, you know what? We always welcome rational and reputable competitors. That's the type of competition we respect and we like. This is true across the spectrum. Densification will continue. Densification plays in our favor.
I think just to add on what Gio said in terms of the CDU quality kind of conversation, another thing that I think we pride ourselves on our think-through as we think through this space is the fact that we are getting a lot of reps. We do have a lot of product that's out in the field, going through the install base and understanding what it looks like, and having the services arm behind that. If there is anything that we see, we're quick to a response, we're quick to understanding, we're quick to a redesign, we're quick to a resolution. I think that process in and of itself, because it is a new technology and it is a new outcome, it gives us a competitive advantage.
It gives us a competitive advantage to understand what the install base is doing right now, understand how to ensure that we can make it as most effective and efficient as possible, and to introduce any new changes that need to happen. That knowledge that's coming back to us on a regular basis. I think some people that are new in the industry don't have that knowledge yet. They haven't seen how it's reacted. They may not have the services capability to go out and fix as fast as we would be able to. There is a unique situation there when you think of an infant product that is brand new to an industry, what it's like to get that information first and understand what it's doing and being able to fix before there would ever be an issue.
Yeah. What about innovation in liquid cooling? A decent number of our conversations look towards what might be coming in terms of two-phase liquid cooling or things that could be disruptive. How do you think about your investment roadmap and—
Yeah
—mapping to those evolutions?
I think people tend to apply the liquid, move from air to liquid categories to the one phase, two phase categories. We are not that, how can I say, binary.
Yeah.
It's liquid cooling, it's liquid cooling. You can go with evaporate, kind of a phase change, or you can stay in the same phase. You can go from use water and glycol, you can go to refrigerants. We know refrigerants very, very well. Again, it's another area where we parallel organic, inorganic, many things. But don't look at it as one day will change. There will be pockets or maybe big pockets of application over time when single phase reaches a certain threshold. There will be segments where two phase is coming. We don't see it as a disruption. We see it as an evolution. Liquid cooling was a disruption for the industry. Single phase, two phase is not a disruption, is an evolution, and it will have a roadmap. It's yet not clear exactly where and when it will happen.
Yeah.
Yeah. In the time we got left, I want to steer back a little bigger picture. First, just a simple question, Craig. What is the single biggest constraint on growing faster today? I know it's probably a lot that we could bring into this response. Maybe if you were to focus on one particular item that you're looking at.
No, I think scaling into what I would say this environment is ensuring that you scale with quality and efficiency at the same time. Ensuring that you bring on capacity, you bring on capability to deliver to your customer on time, deliver to your customer a quality product, and ensure that you do that efficiently. I think all three of those things, you have to dance across at all times. Ensuring that you meet your customer commitments. Everybody wants everything faster, but as you do that, you have to be sure that your manufacturing processes and capabilities are of the right level that you can do that at a quality level, a right quality level, and then ensuring that you can do that at an efficient level so that you are not burning margins.
What you have seen us kind of dance through in the last couple quarters is that process. We look at that on a regular basis. We look at that almost maniacally, because any of those things that slip down on you can cause your customer to have a bad experience, or you can cause your investor to have a bad experience by overspending funds or being inefficient. I think with Gio and I, it is a constant evaluation of those three levers to make sure that we can scale the right way as this is going to levels that are not normal for a manufacturing process. The levels at which you are being asked to scale are not what normally happens in a manufacturing process.
It requires a level of what I would say focus that is just different, and you have to make sure those trade-offs are not going to affect your ultimate outcome. I know it is a generic way to say it, Luke, but for us, it is about being laser focused on all three of those, because they all influence each other a little bit, and making sure that the customer ultimately walks away feeling like Vertiv delivered a very strong product, it has got great quality, and they delivered it on time. Then we walk away saying our investors see that in the margins being an improved margin, because we were able to price for it, and we were not inefficient in the way we delivered that. It is, again, a very standard way of saying it, but it is what we need to do every day as we wake up.
Probably the last question we have time for, but I want to touch on inflation. Clearly, there is continued pricing pressures out there. Can we maybe just talk about some of the mechanics of managing price cost—
Yep
—near term, and are you counting any pricing fatigue out there right now in terms of your customers?
I think what we've done is structurally put in a process where we look at ongoing, what I would say, commodities and pricing items that come to us on a regular to basis embed those in our quotes and ensure that our customers see in our quotes what our prices look like with those embedded in. Anything that we think is going to change go forward, we embed it in. Any of the pricing that we know is going or inflation that we believe is going to happen is embedded in our quoting process. That's a pretty strong muscle for us in ensuring that we get that out there. Now, we would see that in a fatigue if we were losing market share, and we don't believe that we're losing market share.
The process that we're utilizing ensures that we cover what we believe is going to be inflation, we protect our margins, and that we go out and we compete. If we saw a pullback on any one of those areas, then we would believe there was some fatigue in there. We're not necessarily seeing the pullback, but we are letting our customers evaluate us on all those things, our price that we have out there, the quality we're able to provide, the service we're able to provide. I think for us, that muscle is something that we continuously strengthen because it's the key to us being successful.
Okay. Well, I'm going to leave it there. That's all the time we have on stage, but Gio and Craig will be available for a breakout that's down the hall in Salon A. Thank you for the presentation.