Vertiv Holdings Co (VRT) Earnings Call Transcript & Summary

We are very excited to have Vertiv Corporation with us. We've got Gio Albertazzi, who is the CEO; and we've got Dave Fallon, who is the CFO.

And Gio, as I walk over to you, maybe just start with -- there's obviously a lot of noise out there after you reported earnings, which were very good, by the way. But if we could start off with why do you still seem so confident that you're well positioned to take advantage of the AI-driven CapEx cycle? Maybe to sum it up, you've talked about being a one-stop shop. So sort of what does that mean between your experience in thermal power, service, infrastructure solution? What does that mean?

Well, thanks, Andy, for the question. And as I said, Q4 was a very strong quarter. So very proud of our Q4, very proud of all the progress and acceleration in the fourth quarter. AI clearly is making the entire infrastructure more dynamic and more challenging in every possible sense, in terms of the loads that are being serviced today, but even more importantly for the long term, in terms of the intrinsic dynamics of the industry. We know that the type of IT loads and infrastructure today, the GPUs, the servers, the clusters will only become denser and more challenging, more performance, if you will, more efficient over time, but more challenging from an infrastructure standpoint. So our ability that comes from decades in the industry, that comes from most complete portfolio in the industry, our technical prowess makes us uniquely strong in being the privileged [indiscernible] for many of the hyperscalers and large colos, but as well as a lot of the silicon providers these days. Why is that very important? Because we know that -- those loans as the infrastructure -- as the IT infrastructure evolves, everyone want to make sure that today, their decision, their capital is allocated in something that is future-proof. And you can have that conversation with them if you know what the future will look like, and you know that only if you have a very strong relationship with the silicon providers. We have a very strong relationship with NVIDIA, and we work the technology road maps together, but the same we do with many of the hyperscalers. So this forward-looking and ability to think the entire system is something that positions us uniquely strongly in the market.

And Gio, it's safe to say that with all the changes out there, do you think you're at least holding or gaining share? Is that the right way to think about it?

Yes. Yes. We -- not only we believe, but we gave an indication of the market, the space in which we operate, kind of blending all the various components, something that grows 9% to 12% over the next few years. If you look at our trailing 12 orders at 30% at the end of December or if you think about our 18% growth in '24, well, that shows much again.

So you got about 6 questions on orders on the call, maybe about half the call. So I'm going to try and sum it...

It felt like...

I understand. So I'm going to try to sum it into one question, right? So you said your orders were up about 30% year-over-year trailing 12 months, right? But much of the investor [indiscernible] that we've seen will be -- is revolved around the recent flattening in the actual level of quarterly orders and backlog. However, to hit, you've got a $14.7 billion revenue target out there for '29. It's 12% to 14% organic revenue growth. Backlog and actual quarterly orders would have to trend up again to hit those numbers, I think. So would you ascribe the Q4 European order delays, you said just lumpiness, are you seeing any significant delays with hyperscale orders in the U.S.? And is it fair to say that your confidence level is still high that your order backlog upcycle is in your [indiscernible]? I tried to do the one question.

Yes, exactly. So that's 24 in one. Well, I think when we look out in the future, we have to absolutely think in terms of 2 things. One is the backlog creation that we have delivered in 2024 with 1.2 book-to-bill. Certainly, there is a backlog creation that goes in the right direction. The trailing 12 is important. We always say that our industry is and especially as the individual projects become larger and larger, the lumpiness, if you will, exaggerates. And that's why we moved away from talking necessarily and obsessively about orders in the quarter, and we think that kind of a 12-month representation is more interesting and more relevant to define what future looks like. So it's a combination of things. There is a pipeline that is strong and growing. There is an industry that is very convincing across the board. Yes, there are some geographies that might be a little bit slower than others, but it's still in an overall kind of a global landscape that we see as very favorable and certainly very consistent with the model that we shared with all of you in November in Atlanta.

So I'll ask you one follow-up to that, Gio. So like you mentioned at the Investor Day, you were answering a question around, you said 100 gigawatts of cumulative power added from '23 to '29. And -- but you -- you were conservative around your organic rate guidance, 12% to 14% because there could be headwinds that could slow down versus the power needed, what your growth would be. So would you say EU regulation is a headwind? And are there any other headwinds that have developed over the last few months with the understanding that the DeepSeek news came out between then and now?

So if we go back -- let's go back 2 years -- sorry, a year is a little bit more when we were our first Investor Day. We said, hey, there's certainly probably characterized as insatiable appetite for infrastructure bill. And said, we would caution everyone, hey, this insatiable appetite is moderated by factors like power availability or the fact simply that there is permitting. It's -- there could be areas where regulation is a moderating factor. But also, the fact that the industry is -- this is a construction business in many respects. So there is a natural, let's say, limiting factor against that insatiable appetite. So when we were talking about our -- when we gave our growth indications, market indications, if we see -- focus on hyperscale and colo, we were saying, we see them growing at 15%, 17% range, that factors that in. So -- but clearly, we're talking about kind of the medium term. We're not talking specifically one quarter or the other. So the fact that there might be, as we said, some of the projects in EMEA moved to the right, I wouldn't say is changing the assumptions of the model. It's just -- it's part of the lumpiness, it's part of the things, that if you look at things on kind of a more longer period than 12 months, well, then we see that, for example, Europe is clearly lagging the U.S. in general by 12, 18 months. We thought it could have been a little bit shorter, but it's lagging at this space. Is that totally new? No. We believe that our model going forward in terms of the indication of growth, margin growth, et cetera, is still measured relative to the market. So we feel good about the market relative to our indications but also our model relative to what could be the hiccups of the market around that trajectory.

Yes, Gio. That's very helpful. So Dave Cote on the earnings call talked about Vertiv seeing benefits from seed planting and R&D, your customer relationships and your CapEx, where I think your CapEx this year is up 50%. So can you give me more color regarding what specifically you're doing in these areas and why they're helping you?

Yes. I will let Dave talk about CapEx, I'll gladly talk about innovation. I was mentioning about the very dynamic phase in which the industry is and will be probably for the foreseeable future. In this industry, it's not just about being -- having the best point product. It's about having the best system and an ability to be ahead of competition for the entire powertrain, thermal chain or more generally speaking, the entire infrastructure. So we continue to invest in innovation. We have given an indication of 15% CAGR spend increase in innovation. And we're probably at 24%, a little bit ahead of that. And if being ahead of that is something that we think is the right thing to do, we'll do that, no hesitation. We like that a lot, but we believe that being able to deploy technologies that enable our customers' success and business model is key -- is one of the keys to pricing power and market share gain.

Just quickly on CapEx. So I think traditionally, we described our business as being relatively capital light. I would say, in general, that is still the case. But the exception is when you see or anticipate periods of significant growth. So historically, our CapEx as a percentage of sales has been south of 2%. I think that was the case in 2023, we crept a little bit in '25. So a significant increase in '24 versus '23, but as you mentioned, there's going to be a step change, another 50% increase to $275 million in 2025, which is 3% of sales, and that's at the upper range of the directional guidance we gave in November. And the only thing I will say about that, if you're looking for breadcrumbs, if you're looking for signals, there's no more significant signal in our confidence in the growth we're seeing then with the checkbook and that's with the $275 million investment. Some of that inevitably will support demand we see in '25, but it's more of a forward-looking investment in what we see beyond the current year.

Yes. It's very helpful, Dave. And maybe to that point, it's interesting like, [indiscernible] seems to follow different sort of trends. So right now, it's orders, but like last summer, it was liquid cooling, right? And so to that sort of capacity additions that you have, like maybe an update on where you are in your liquid cooling build-out. I think by now, you might be 40x higher than you were when you started or at least that's what you had talked about at the time and how much that is contributing to your sort of full solution capability?

Yes. The -- about a year ago, indeed, we were indicating a 45% capacity growth kind of annualizing the beginning of '24 to the end of '24, we are certainly there. We're very happy about the capacity that we have deployed and the geographical, let's say, spread of this capacity. So we are happy there. We think we have the capacity needed to serve the industry and to have a very relevant participation in this specific technology. It's absolutely, absolutely important. And I think that is a demonstration of the technical -- not only technical progress, but the ability to read the market and the technology and to combine organic integration and inorganic innovation because we made an acquisition at the end of 2023 that exactly fueled our strength in liquid cooling. And nowadays, you have to have the entire -- speaking about cooling, you have to have really the entire range of technologies, the liquid cooling, the air cooling and everything, heat rejection, chillers or other more advanced technology like -- we like to talk about the CoolPhase Flex. So it's the entire system that you optimize. It's that entire system ability of value proposition that takes you to the meaningful tables and influence the design cycles of our customers.

[indiscernible] your own expectations, anything in terms of the competitive landscape from the cooling side, you ramped up you feel good about where you are, again a lot...

[indiscernible] '24 have seen how, certainly, this is an area that is very dynamic in terms of portfolio expansion, very dynamic. Relative to engineering capacity that we had of '24 to now, we're not specific, but it's much larger, even post acquisition, much larger now -- so there's a lot of development. It's not unusual actually, absolutely expected in any situation. This is life cycle -- the phase of the life cycle, there will be some entrants people that come with some good ideas, but they may not [wherewithal] the service capabilities, the presence or the ability to scale or there will be a consolidation in the industry. Some of these will be acquired just like we acquired CoolTera back then. So I would say, are we surprised? No, we're no surprised. I think that the -- this part of the industry is behaving pretty much as you would behave a relatively new technology, and we're proud of the position we have.

Yes. That's helpful, Gio. And then just one question back on Europe. Was there one particular country that you do that was more of an issue? Like you had talked for a while, I think, about Europe being 6 to 12 months behind, right? So like now 12 to 18, is it like literally a few projects in one country move 6 months? Or like how do we think about it versus...

Yes. I think, yes, we -- no, anyway, it doesn't work. So well, thank you -- so we're saying it's not one specific. We talk about the bulky nature of large projects and sometimes these projects have their own cycle. But it's true that we thought that Europe is sticker currently than we expected. It doesn't mean that things are not happening, but not at the speed. But go back one week at least, a lot of kind of key players, European players who were in Paris kind of looking at themselves and saying, hey, what can we do to unleash a little bit more of a vigor in the AI industry and more. So I think it's blatantly obvious.

That's helpful. And then Gio, I feel like I asked you this before literally same question. How concerned are you regarding next-generation chip power efficiencies, right? So let me just leave it at that because I know how you've answered me before. But now with this news, how do you answer me now?

Yes. I think 2 aspects. One is -- before I say -- we talked about the fact that anyway, the sheer amount of capacity increase overwhelms and is way over the efficiency gain that the silicon is providing. I think that holds absolutely true today. And indeed, when we model the future, we always assume that there is efficiency, just like there has been efficiency in this industry for decades. So this is no different. The amount of volume, the amount of traffic, the amount of compute -- raw compute power needed is going to increase. I mean, if one thing DeepSeek has done is clearly start to accelerate the transition from training to inference. And I always like to think about training as something that is a pure cost for those who do the training and inference being the revenue-generating part of the industry. So talking with many of the players, they see the transition to inference is accelerating. Inference is not less power intensity or computer power intensity. So our position that anyway, net demand is way in excess of the efficiency game is still valid.

So from your conversations, you don't think hyperscaler CapEx will change materially versus before DeepSeek?

Look, we have seen post-DeepSeek declarations of CapEx expansion from quite -- some were saying, hey, Q4 was our peak CapEx '24 and by the way, think about the future as we planned out, which means a year-on-year increase or some we're blatantly out with higher numbers than historical numbers. So I think it's going in the right direction.

So I'm going to open up to the audience in a second. Let me ask you just about enterprise because it doesn't get a lot of attention, but it used to be a big portion of the business. So maybe you can talk about enterprise customers, what they're saying, is growth picking up there at all as AI starts to take hold?

Yes. Clearly, it's a slower part of the market. We see an increasing number of conversations and attention. And the enterprise market is moving and it's moving on traditional compute or AI compute. It's gradual, it's slower. Sometimes it's hard to really separate AI and traditional, but we see an improving landscape. If inference moves more to the edge still in the hands of the usual players or the new cloud people, which is certainly happening or even enterprise, we are very confident that we have the right go-to-market for -- definitely for both.

Got it. Any questions from the audience?

On the last call, you noted that despite the normalizing TTM orders growth number quarter-over-quarter that the internal pipeline that you track grew sequentially. Can you just flesh out what the process is for that and what sort of rigor and analytics go into that number, even though you don't disclose it externally?

Yes. Because I will be only that -- let's say, I will not be giving numbers, of course, here, let me be straightforward. But we're pretty rigorous in pipeline. So I'll start to be clear about what is not pipeline for us. When we sit with the customer, and that customer shows us their CapEx plan for '27, '28 and '29, and we know what the implications are likely to be, we know that, that's in our heads. It's not in our pipeline. In our pipeline, we put the portion that is quoted, be it quoted at, let's say, budgetary level or it is quoted at the various stages of the sales cycle. So it's something that is quite quantitatively well defined, but also something that has certain stability in method and indeed, we are very rigorous in this because we run our entire selling motion across every level of the organization, across every region of the world. So we feel pretty good about our ability to compare pipeline in different -- pipeline evolution over time. So we're rigorous in that respect. So we know that there is much more out there. Sometimes you have a lot longer visibility, but we want to make sure that the rule is what is in the pipeline, what we call pipeline is something that has an actual dollar value attached to it. It doesn't mean that we sell everything. Of course, we measure and analyze pipeline statistically, win rates, the pipeline velocity, what time it takes from an opportunity to be created to becoming an order or sometimes we don't win 100% unless that makes sense or anyway for the opportunity to close. So that process is quite mature at Vertiv.

Any other questions?

So I'm just curious if there's any metrics you look at, the conversation about training LLMs versus inference. How do you think about the demand for cooling for inference versus training, if there's any industry guidelines?

Well, that's true for cooling and it's true for power, it's true across our portfolio. At the beginning, I mean, probably a year ago and even before that, and even in New York at our November '23 Investor Day, we were giving some indication what is AI, what is not AI and then one could say, hey, maybe we can talk about what is inference, what is training, we stopped doing all that, even internally, if you will. Of course, we know from a product technology standpoint what is certainly going to serve a GPU type of server or not because you have liquid cooling, you have other things. But what we see, and I mentioned that last week, but what we see increasingly so is that the lines between AI and non-AI are blurring. But even -- and I'll explain why. But even within AI, the lines between training and inference are blurring. I think more and more our customers understand that dedicating an asset only to training, that means that you make an investment where you have, let's say, in the useful life cycle of 20 years, to a portion of your compute needs for the next 20 years. So more and more, we see our customers, be them hypers or colo then say, hey, I need to build an infrastructure that is capable of handling the various type of lows. So going back to non-AI to AI, they say, well, I know today that whatever the AI non-AI mix will shift over time. The medium density racks, high density, very high density will change over the life cycle of the asset. Some overprovision cooling, for example, a 100-megawatt data center will have 80-megawatt of liquid and 60-megawatts of air. So 40% overprovision because you built in flexibility. By the same token, between -- this was non-AI, AI, loads are going up, data centers are being designed for longevity. And over time, the mix between training and the same facility will be used for both. Now some of the players are very explicit about the fact that started to build more and more close to the central gravity of the use of the population because we know that during the course of the life cycle of the asset, we want the asset to serve more for inference than training.

Any other question?

So Gio, I wanted to ask you, like maybe getting into the margin topic because you guys have executed quite well. And maybe this is for Dave too. Like -- so I'd just actually start with APAC because you have seen some sales recovery there. There is a pretty big discrepancy in margin between APAC and the other segments. So is there anything structurally that keeps those margins lower? Or is there more opportunity there? How do you think about that versus the other segments?

Thank you for asking about margins, by the way, so...

No worry, I got a few. Not 24 though.

Exactly. So we posted this quarter with 21.5% operating margin, and there were 0 questions on the call about it. So just to put that in reference, you go back to 2022, our operating margin was around 8%. We grew it to 15.3% in 2023 and then this past year at 19.4%. So a big driver there. Certainly, we can get into the mechanics, but it's Gio's relentless nature. I think you look at our culture today versus what it was 3 years ago, and it's night and day difference. And you can get into that debate, does winning drive culture? Does culture drive winning? There's probably a codependence there. But we talk a lot about the Vertiv operating system. There's 3 pillars to that. It's the rigorous operating cadence. It's sharing best practices and lean continuous improvement. And all of those things play a part. And what we talk about the drivers of margin expansion even going forward, of course, the operational leverage, everyone thinks that's a fixed cost focus. It's really a productivity focus driven by the Vertiv operating system. Of course, we have operational execution and commercial excellence. But to your question, APAC margins. So we did see a pretty nice step-up in margins in APAC in the fourth quarter, I think, 12.5% and we were pleased by that. Some of that certainly driven by the 25% growth that we saw in the quarter. But all of those things that I talked about, driving margin improvement, they're pertinent across all of the regions. So we expect to see margin expansion in '25 and going out to 2029 in all of the regions. But to your point, there is some structural difference as it relates to the margin profile we see in China and in India and Southeast Asia. So we definitely see opportunity. But when we're talking here in 2029 or 2030, talking about 2029, we're not going to be talking about APAC at par with those other 2 regions.

So I definitely think that we take your margin performance for granted. You guys have done a good job there. You answered kind of my next question to some extent, Dave. So let me ask you in terms of balancing price versus cost, Gio, so over the last few years, you've come a long way on your pricing, as you know. And you've also built up your supply chain pretty significantly in terms of resuming. So those are the 2 things that were tougher a few years ago, right? So maybe talk about how that prepares you for the future because there are the questions on tariffs, Mexico, what have you. So talk about how you can react pretty quickly on the price versus cost side, if need be and the ability to price in the current market?

I think start from the price side and 2 things. One is a pricing muscle that we have developed got proven as well. We're reporting price specifically and kind of exactly -- sorry, specifically, up until 2023 and that clearly showed some improvement. So pricing muscle is like having the right processes, having the right decision-making, having the right delegation of authority, having an ability to really translate price or the value creation into price. That is a muscle that we didn't have was a little bit atrophied and certainly now it's there and it's strong. But to me, price is really and pricing power is really about -- I'll go back to what I was saying. If we are not able to augment the value of our customers' business, if we are not enabling our customers to be more successful with us than they would be with someone else, then anything else is worth 0. So it is relentlessly being focused on that, that creates pricing power. The rest on the cost side of the equation is about relentlessness. So yes, it's a resilience in the supply chain, and we know that we have learned that lesson dearly in the past, but never take anything for granted. Continue to work your supply chain from a performance cost resilience. That activity goes on. With that is the overall efficiency of the execution machine. And the last piece is the operational leverage that we've several times explained or elaborated upon. You're talking about tariffs, clearly, we are in little bit of a limbo to really understand what they will look like. We have we believe developed strong playbooks depending on the various scenarios that we have developed. So we need to understand exactly what will happen to deploy the right playbook that we have ready at hand.

Let me ask you 2 follow-ups on that. So one, these are big complex projects that you're signing a lot of them. Do they have escalators in there? So -- because tariffs might get thrown at us from different angles, God knows. So they have escalators in there. And then the other is as the projects get more complex, I think early in the AI evolution, revolution, you said, I think AI projects could have better margin than traditional projects, is that still the case?

Let's see what the tariffs -- but again, very importantly in a cooperative manner. What was the other side...

Pricing of the newer complex project versus the traditional margin?

As you said, there is intrinsic complexity in the technology. And that increasing complexity plays very well to our capabilities. That typically translates into pricing power, but again it translates into pricing power only to the extent, and it's quite an extent that we are really creating value for our customers.

Got it. So just talking about capital deployment optionality, your balance sheet is in great shape, right? You mentioned CoolTera a couple of times. It was a great acquisition. You did BSE in China recently. Like -- so it seems like you've talked about going after this kind of technology innovation type acquisitions. Are there other "CoolTeras out there," that you can buy for reasonable valuations? And if there aren't, do you lean into things like repurchases or what's your reaction that?

I think we have laid out our capital deployment strategy quite extensively in a plant in November. So -- but pretty much going down the same type of architecture, of course, repurchase is always for us out there. We can do it. We have authorization from the Board for doing that. We will -- we told everyone we will be opportunistic and that's our strategy. When it comes to acquisitions, let's say, we like those kind of focused technology acquisitions because they reinforce our organic innovation road maps, but also sometimes we accelerate technology that we can then deploy globally and scale globally. We like them a lot. So do not be surprised if there will be more in the future. But clearly, the strength of the balance sheet allows us also to do things of a different size. So our M&A pipeline is live, is very well. But then what will happen, of course, remains to be seen. But I think that other types of kind of guiding principles in the acquisitions like, is there kind of a technology that will be developing in the future or in the industry that we might benefit from kind of a larger acquisition for or are there kind of a market share or route to markets or capacity plays. If anything like that, it should be the right thing for us to do, we would have the means for doing that. So I wouldn't exclude that at all. The thing that I always like to say is that we are not going to completely change the profile, the characteristics and the DNA of the company in doing this.

So just quickly last question. Just what are the top 2 or 3 innovations and structural changes affecting your company over the next 5 years? And are there any emerging industry trends that are perhaps being overlooked in the current discourse?

No, I think pretty much the changes -- the most important thing, I believe, is the fact that the technology will continue to evolve. Technology will continue to evolve. The -- this evolution will be the driver for everything. And we see that as a favorable driver out there. There are other aspects that we like and we will increasingly consider is the fact that our portfolio is very, very, well suited for commercial industrial application data, critical infrastructure in their nature. So that's an area for us for further grow, another access of growth, another avenue of growth that we -- that might develop over time. And critical infrastructure is something that we can do very, very well. We do it very well in the data center space and why not elsewhere.

Thank you, Gio. Thank you, Dave. Appreciate you coming.

Thanks a lot.