Good morning, everyone, and welcome to Oppenheimer's 20th Annual Industrial Growth Conference. I'm Noah Kaye, Managing Director at Oppenheimer Research, and we are very pleased to welcome in the management team of Vertiv, CEO Giordano Albertazzi, CFO David Fallon, Lynne Maxeiner. Thank you very much for being here today.
Thank you for having us.
We're going to, as a sort of point of procedure, go through a number of questions that we've prepared. For those listening on the webcast, you can submit questions via the Q&A function, or you can email me at noah.kaye@opco.com. We've got a lot to get through. Let's start, I think, with a 10,000-foot view of the industry. I think back to November's investor event, and the company had referenced 9-12% CAGR through 2029 for critical digital infrastructure with mid-teens growth in cloud and colo. You reiterated that during Q1 earnings. As we've talked with investors, I think a fundamental question they're grappling with is, when you consider the AI training breakthroughs, the proliferation of models, why do the growth assumptions from investor day still hold? How might the nature of demand be shifting versus what we previously thought?
You know, the question can be interpreted in two ways. Is that given all the investment, given all the shift to inference, enhanced to the real reason for the existence of AI, the existence of AI, not to have kind of self-referential models, but is to have a stock that can be sold a market. That is happening. That is happening fast. It's strange times in which people say, hey, your growth model to 2029 is too optimistic or is too pessimistic. Both ways. You can see both ways after the nervousness of the swings in the early part of this year or even July last year. For us, it's clear that demand is there. The demand is strong. This is a secular trend. Many people have compared this to electricity or the steam engine or whatever else. It's going to continue. It's going to be strong.
We always say that it could theoretically be even stronger than this were it not for some moderating factors like permitting and power availability all being addressed, of course, but all already factored in our demand plan that we've shared with all of you. By the way, that 9-12% range really becomes, if we focused on the colo and cloud, that more of risk at this stage part of the demand would be at 15-17%. If anything, as we look out, one could be more optimistic than that. We have always been measured in our view of things. The various inputs, the various analysis, the various opinions in the market certainly corroborated also by our pipeline pointed the direction of the model we shared with you is very sound. It's going to be better. We will all be happy.
Certainly, we have the capacity, the means, and the innovation to be a winner in an even stronger scenario. One thing I want to reflect on a second. We came out of a February and March period where everyone was nervous in the market about the demand, et cetera. We were very early in the earnings cycle this time around. We came out saying, hey, guys, the demand is there. It's going in the direction we told you it would be going. The market is good. Stay calm. We're focused on the long term. We are here to deliver our long-term promise as we explained that in November. One after the other, the hyperscalers, the competitors, et cetera, confirmed that. Look at Vertiv with confidence when we tell you where the market is going.
I guess to put some meat on the bone here, I'm not sure everyone realizes you just had effectively a record quarter for orders or close to it at least. You still had sequential growth in the pipeline. Maybe you can start by reminding us how you build up your pipeline, how you define it, and what you're seeing on conversion times from pipeline into revenue.
Sure. I will start saying that, as we always say, orders in this industry are lumpy. Let's keep that in mind. One should not read the long-term trajectory from just one quarter. Have you seen that? Of course, Q1, 13% up year on year, on a 60% up last year. It is very difficult. Calm. Something that we are very happy and very proud of, just like the strong Vertiv pipeline. For us, pipeline is everything where we are commercially engaged factory, not just we have, I do not know, a capacity agreement that goes out into future year long term. That typically we do not call pipeline.
Pipeline would be, in my example of a given capacity agreement, would be, okay, the next two, three projects that we have visibility for that we know are being built, that we are quoting specifically or working on from an engineering standpoint specifically, that would be pipeline. More generically, pipeline is everything from the identification of an opportunity from a commercial standpoint to the moment we close. It is a pipeline that's been growing, as we said, and sequentially growing as well. We see conversion rates that are encouraging, that we see a good stability there. The pipeline is, we believe, as it has been historically for us, including the last quarter, a very strong leading indicator of future performance.
You mentioned, Giordano, that there's the shift happening right from training to inference. Maybe you can help us try to understand how to think about the share of demand you're seeing from training versus inference. I don't know if that might involve smaller, more distributed projects in the pipeline, or perhaps you can push back against that. Maybe something that you saw in your order patterns. Just help us understand the mix today and how you see it trending.
Two things. One, I think we should not be too binary in the definition of what is training and what is inference from an infrastructure standpoint. Very often, we are here talking about people building an infrastructure, a data center that has to last 15, 20, 25 years. Defining today what type of loads you will have in the 10th year is very difficult, or even in the fifth year. What we see is more and more people are thinking dual-use type of infrastructure, so that the mix will be what it is. There will be a period in which it is model training. It will be kind of a specific application training. It will be inference. We see inference, of course, growing faster as time goes by because that's where the use case is. That's where the monetization is.
That's where AI becomes, it's AI at work, not AI training. When it comes to the mix of type of data center, we continue to see large data centers being built, designed, or planned. That has not abated. Size matters for a data center. We also see a number of more distributed, more edge data centers, data center ideas. When we were talking about iGenius, the project that we shared publicly, that's typically a combination of a southern and edge application in many respects. It's a combination. We see in the medium and also long term, really a combination of very large and edge. The other side, the other aspect that is interesting, is that we always talk about enterprise. Enterprise in and of itself, on-prem, is a smaller data center model. It will not be one way or another.
It is a large industry and growing. That will have multiple ways about itself.
Yes. I mean, Giordano, as you alluded to earlier, I mean, I think that there was some concern from investors in months past around these headlines of specific hyperscalers maybe pulling back in areas on lease negotiations. I think when we listen to industry experts, for example, at the Data Center World, I mean, they were characterizing this more as really a reallocation of demand to account for scarcity of resources like water and fiber and power and land. I think the question for us is, are there areas of any significance where you are seeing some pause around long-term planning for some of the larger AI factories right now? If so, how much of that speaks to these resource challenges? What are the implications for the company?
We do not see anything meaningful long-term change of market dynamics in terms of there is a long-term need for data center capacity for cloud or AI or proprietary application in the enterprise development and whatnot. That is not changing. Now, as you say, there could be different short-term movements, adjustments of decision on what to prioritize in terms of investment. That is the normal course of running a business, whether the business is a cloud business or if it is an enterprise business or a government business. That is more about the mix than the long-term trajectory. We are very optimistic about the long-term trajectory. The optimism is grounded on the things that we see happening in our pipeline, but also in the industry more broadly speaking longer term.
You mentioned some of the new entrants, Sovereign and NeoCloud, and even some growth in enterprise, new colos. Maybe talk a little bit about how you approach those new entrants versus perhaps more established blue chip, if you will, hyperscaler customers. How do you see kind of counterparty credit risk? How far out would you be willing to book orders for customers like that?
I would say certainly the industry is expanding. It is not so much the IT infrastructure or digital critical infrastructure. That has been always extremely broad. That has been extremely broad for the last 20, 30 years. If you think about pre-AI, let's go back four years. I mean, AI has been there for quite some time. Let's say before ChatGPT, November, when was it? November 2022. The industry was very broad from government, Sovereign, AI, et cetera, et cetera. Sorry, Sovereign, IT, enterprise. Always very broad. Indeed, we were very strong in terms of our go-to-market and reach also the smaller player. Overlap on that, the AI acceleration. Certainly the AI acceleration at the beginning has been and still is strongly and kind of a large cloud player activity. Now we see that that is broadening.
It is broadening because the market is very attractive because there is space for, for example, new clouds. It is almost inevitable that other IT players, the enterprise and again, historically, that do step up their game to AI. This is really what is happening. This expansion of the market should not be viewed as, oh, it is an expansion of the risk base. No, because those actors have always been in play. Maybe not the new clouds, but the other actors have always been in play. They are now upping their game to AI. When it comes to the credit risk, of course, we are very thorough. We are very thorough. We make sure that not only do all the checks necessary, but also more often than not, we work with advanced payments.
Make sure that there's little to no fluff in the type of commercial contracts that we engage ourselves in. The last part of your question, if I remember correctly, was how far out into the future are we committing from a backlog standpoint? Not differently than what happened a year ago, say, time or historically. We do not see kind of a runaway, people that just talk the big talk and transform orders into 2020, whatever. We're just very rigorous. It's credit-worthy, yes. Is this a real project? Yes. Do we have a purchase order against which there is a construction site that is alive and there? I think that I'm pretty sure, actually, I am sure the risk profile of our business has not changed as much.
In a bit, and I'm looking forward to it, I want to take it into the area of technology differentiation and portfolio evolution, which is quite exciting. We would, of course, be remiss not to address tariffs. Perhaps you could talk to the impact, if any, that you see tariffs having on the demand environment. Has there been or could there be any pull forward in demand to address reciprocal tariffs? Any change in pipeline or order conversion you've seen thus far in the Americas versus EMEA and APAC?
When it comes to the impact on general demand, we have not seen it in any material fashion. Again, the demand is there. The demand is there for our customers. The opportunity is there. Nothing material to report in that respect. When it comes to the aspect of demand pull-ins, it is really not possible in our space. A, because what we build is, in the majority of cases, always, but in the majority of cases is destined to a location. People are not just ordering and having stuff delivered ahead of a certain set of tariff rates coming into play. Nor from an order standpoint do we see big movements there tariff-related. I would say no real meaningful material impact.
Can I just follow up on that and ask conceptually why that's the case? Because I think we had, for example, one hyperscaler increase their CapEx estimate for the year and really say this was because of tariffs. Of course, they go ahead with their plans. For critical digital infrastructure, how should we think or should we think differently about elasticity of demand if we have tariffs increasing the cost to build? I recognize that you are a relatively small percentage of that build versus the actual IT, the chips that will ultimately go in, particularly for these large IT factories. Help us understand why that's the case.
I would not have kind of an exact answer, if you will. What we heard also from many of the hyperscalers in their recent earning releases and calls is they have a lot of demand for AI. More often than not, they have insufficient capacity to satisfy that demand. They believe that this is going to continue over the years. You are right. Of the total cost of a data center, let's say the long-term CapEx, as someone has characterized that, one of the hyperscalers has characterized that, is a relatively small portion. Make no mistake, these hyperscalers are very savvy from a commercial and cost management standpoint.
The fact that there isn't a stock price elasticity such that an increase that is driven by inflation or is driven by tariffs directly impacts demand is that the demand, their demand, not the demand for us, their demand is stronger than that. The equation still works very well. Again, I should be truly in their shoes to give you the answer. I think this is probably very close to what's in their head.
Yeah. Yeah, that makes sense. Thank you, Giordano. You have commented in the past that your lead times are very competitive, perhaps even running ahead of some peers in terms of your own ability to deliver. What could that mean, if anything, in terms of opportunistic share gains? It sounds like there has been no pull forward of demand. Perhaps your agility to provide amidst all this tariff uncertainty is helping you win business that you would not have previously gotten.
Yeah. I would decouple the tariff uncertainty from lead time. Tariff uncertainty, if you will, is a very temporary situation. Everything, the equilibrium, rebalance under the new trade rules. I would not look at it as something that answers or explains dynamics. We believe that short lead times are a competitive advantage. Now, clearly, not everyone will need a short lead time. Very often, we've been vocal about that, but very often, the requested lead time for our customers is between 9-15 months, and sometimes 18 months. We are talking about lead times well below that. There are areas in the market, typically the enterprise or sometimes we talked about the rapid shift of priority for some of the big players. That shift of priority may require a course correction also in what you build and what you do not build.
Maybe you have a construction site that has initiated and you accelerate. The short lead time is more opportunistic than everything else. It is additional market share while the bulk continues to stay in that 9-15 months, if it makes sense in order.
It does. It does. I appreciate that. I think just on managing the cost side of this picture with tariffs, you led the company, Gio, out of a very challenging supply chain environment as CEO. I'm sure, and perhaps you can comment on it, you've adjusted your approach and taken some lessons during this new period of uncertainty. I think more concretely, the company quantified net tariff cost impacts for the year, stated expectations to exit the year at tariff neutrality, roughly 50/50, I think, between price and supply chain mitigating that. Maybe just help us get a little more clarity on how much of the backlog actually has contractual repricing or protection for tariffs versus more ad hoc discussions and how you're kind of framing any incremental pricing here. Is it truly incremental price, or is it more of a surcharge for tariffs?
Okay. Let's say clearly very different conditions than three years ago and very different in many respects, also different company relative. Certainly very different in terms of execution capabilities that have been the obsessive center of focus for everyone and continues to be forever. Certainly a level of supply chain resilience. As I said, build the resilience for China plus one for nearshoring and region for region. You build tools, muscle, competencies, ability to work with people, and supply a network that can adjust relatively quickly to a changed perimeter that is a partially changed perimeter. We will see what the longer term will be with tariffs that we are facing now. Hence, the company's fit, the company can react, and is reacting quickly to this changed environment. The other is we built a price muscle that we did not have before and we demonstrated.
When it comes to what is the structure of our contracts, we do not go in those details. Certainly, we are in a much stronger position than we were three years ago in terms of an ability to act on inflation increases, tariff increases. We will not go into the details whether it is a price increase for inflation, if it is pricing increase for tariff on a case-by-case basis. I think that is one thing that I want to be sure I send a strong message about is we do this cooperatively with our customers on a customer-by-customer basis. There is no way that we would just harness a contractual agreement and just go and steamroll because in the end, it is about our reputation with the customer. It is about our partnership with customers, about designing the common future with the customers.
This is a short-term, temporary thing. As the entire industry reconfigures, it's the long term that matters. When it comes to a price increase that is more general, not backlog related, then it's more you can have a price increase if the conditions change, if there is an inflationary, let's say, inflationary dynamics in the market, regardless of what is causing that inflationary demand. As we said already in our call, we have taken price actions in that respect.
I think just to put a last piece of this together for listeners, I think at the beginning of the year, you guided to a substantial increase in CapEx, which we know that you typically try to maintain 25-30% capacity headroom versus demand. We took that as a very clear signal of your expectations for demand growth. You can help us understand your primary uses of CapEx. I will ask on a related point, did in any way the tariffs that were announced perhaps change your view on CapEx, where you were going to spend it? Did it impact your strategy?
Let's start from the end. It's clear that we operate, let's say, in the geopolitical and environment at hand and what we believe the future is going to be. Only a fool would not do that, if you will. Clearly, the current situation has implications on how we allocate CapEx. I say this is not totally surprising. We've been moving in certain directions, as I said, kind of region for region for quite some time already. If anything, what we see happening corroborates a certain direction. By all means, in general, we operate in the market environment or in the trade environment in which we are finding ourselves now and believe we will be in the future.
The answer is yes, but there has been no kind of a dramatic shift because I believe the direction in which we are heading is certainly supporting the situation. Now, CapEx increase. Correctly, you read into that optimism in growth, as we said, as we said explicitly, if you will. This connects also the CapEx dot with the narrative, the higher level narrative, and more of that will shape with you. That 25-30% wiggle room is something that we like to retain at all times in general. It cannot be always for all product lines, et cetera, because again, we know that no matter how well we know the future, the future will surprise us. We want to make sure that we are there to capture opportunities.
Yeah, growth in CapEx is not a growth that is dramatically different in terms of CapEx to revenue ratio, but certainly is headed in the right direction. Again, it's not just CapEx that defines our capacity. We do a lot of lean optimization. Our output per sq ft is constantly increasing and will continue to increase. We talk often about operational leverage. This is an element. There's an important element. CapEx is one of the, let's say, levers for capacity expansion, not the sole.
I want to shift to the portfolio. The company had a blitz of new product announcements the week of GTC, new chillers, modular solutions, rack UPS software. I'd like to understand how those connect to the growth in the market share opportunities you're trying to unlock.
I think there are two aspects. One is, oops, the screen went blank for a second.
We're still hearing you.
You mentioned chillers. That's a very important one. In the sense that it's a space that we participate probably not, let's say, we're punching below our weight in that space. That is clearly something that we are strongly intentioned to change. In general, clearly, we use innovation and new product launches to not only create market share, but to bring to market things that the others are not bringing. We do that cooperating with our customers and other partners, with silicon providers. We do that very, very intentionally. You talked about the slew of GTC announcements. If we go back to Supercomputing 24 , there too, one product that I like extremely is our SmartRow to accelerate, dramatically accelerate the white space build-out times. It's a space that is new for us.
is new for everyone simply because we are creating opportunities and creating solutions. Yes, this is so central to what we do. We believe that in knowing this space so well, we have an opportunity not just to have the product that suits the need today, but also to see what the future needs are and how we can innovate in ways that really challenge the way things are done in our own very industry. Very pleased with the progress. Rest assured that we will continue to progress in that growth of R&D span and our top line growth.
Gio, I think I'll ask one more before we have to close here. It is related to this. You already have leading share in thermal management in the industry. With an expanded portfolio of chillers and now liquid cooling, how should we think about your growth opportunity? Maybe within that, you can also update us on the ramp for the liquid cooling business.
Okay. Again, certainly we have a good market share in cooling thermal management. We can't have a better market share in cooling and thermal management, quite honestly. It's not that we say, hey, whatever the number, two more points and we're happy. There's no such thing as being happy at Vertiv. We're not happy. I mean, we are a nice bunch of people, but truly, we're never satisfied. That applies absolutely to the cooling market share. Just like the same is true for thermal, for everything, for power, that's true for everything we do. We are particularly happy with the acceleration that we see in everything liquid cooling. At the beginning of the year last year, we said 43 times the initial capacity by the time we are out of the year on an annualized base. That's exactly where we landed at the end of 2024.
We certainly enter 2025 with a lot of liquid cooling capacity. We'll continue and continue to expand capacity because what we hear is that everything is going liquid. Now, that liquid- air mix 70, 20, 70, 30, or sometimes 70, 50. It's not because I can't do the math, but because there is provisioning. That will continue. Liquid is becoming ubiquitous. We'll continue to expand that capacity. Probably not at the breakneck speed that we had last year, but that was kind of a totally new technology for us, for the market, and we needed a lot of capacity to command the market share that I think we're entitled to.
All right. We continue to look forward to your journey down that path. We thank you all for the time and the thoughtful discussion, Gio, David, Lynn. Hoping everyone has a great day at the conference. With that, we will speak with you also.