Good afternoon, everyone, and welcome to TD Cowen's 10th Annual Communications Infrastructure Summit. My name is Michael Elias, and I'm the Communications Infrastructure Analyst here at TD Cowen. For this meeting, we have Vertiv, and from Vertiv we have their CEO, Giordano Albertazzi, and Scott Armul, who is the VP of Global and Strategic Accounts. This session is structured as a fireside chat, and we have about 40 minutes for this. I have a bunch of questions prepared, but I will do my best to open it up for questions at the end. But with that, Gio, Scott, thank you very much for being here today. We really appreciate it. Thank you for having us. All right. Pleasure. Always a pleasure. So to kick things off, for the real estate investors who are in the room, you know, who are not as familiar with Vertiv, can you just give a brief overview of the company as well as anything else you think that may be relevant for the conversation? Certainly, yeah. To keep it short and sweet, for those that don't know us, we're Vertiv. We've been in kind of the data center and communication space for decades now, but we're primarily the guys that provide power, thermal infrastructure to support data centers and communication networks, as well as a comprehensive services portfolio and IT management and IT product and solutions for the data center space. Okay, perfect. Let's transition and talk about demand. So from your vantage point, how would you describe the demand environment for data center equipment? And as part of that, how does it compare to what you saw entering the year? Maybe Gio. Sure. The short answer is strong. Demand is strong. I mean, we've been here for the whole day, and I think that was pretty much the theme. So it's hard to add anything to that. But that strength is certainly visible in our first half orders, or I've probably doing the math, not exactly. Somewhere 58% on the quarter, up year on year, so certainly strong. But as I had the opportunity to share at earnings call a couple weeks back, we see that strength continuing in our pipeline. We certainly see it in our backlog, as we're saying. But, you know, in terms of how does our pipeline or funnel look relative to what we saw at the beginning of the year, well, it's stronger. It's stronger now. So the signs that we see are very encouraging. Okay, so you stronger pipeline versus what you saw entering the year. You know, maybe I could ask this: is that as you look at the pipeline, have you observed any changes in the ways that customers order equipment? You know, I'm thinking through beyond just the absolute magnitude of orders and thinking through maybe, you know, we were speaking with one of the data center operators today, getting a sense for their own changes in book-to-bill, right? Any other changes that you've observed in the way that customers order equipment? Well, first of all, we have to think, our business in terms of different layers. I mean, we have, a telecom business, we have a service business, we have, an enterprise business, all, kind of, having their own, type of dynamics. But I think, I believe, you want me to talk about the, the fastest growing part of our market that is everything to do with, with colo and, and hyperscale. What we have observed, an observation for the first quarter and also in second quarter, that our, what we call the, the velocity of our funnel, the velocity of our pipeline has increased. So the time it takes, between entering an opportunity and booking an order has shrunk. So, there is more desire, if you will, and more rapid cycle to turn an opportunity into an order. That doesn't mean that the opportunity is the first time we know something is there to happen. Mm-hmm. We have a lot of conversations with the customer early on, and we shape, you know, the technology with our customer that they will need, but that acceleration is there. The other aspect is an interesting aspect, when we book an order, the requested lead times have elongated a little bit. It's something that we started to observe, in the first quarter, and we saw also in the second quarter. If we think in terms of, let's say, 9 to 15 months, typically in this, the customer range in this part of the market, hyperscale and colo, 9, 15 being the historical requested lead time- Mm-hmm. Well, that is stretching a little bit to 12-18. Simply because we believe and we observe that our customers have better visibility to what's going on, are building data centers that are larger. So their project plans allow us to place orders earlier in the- yet having clear visibility of their demand and their need and specific site to which their order is for. I mean, well, that would make sense to me, just given as I think through, particularly the colo providers. One of the key themes at the conference here today is that the data center capacity that's being leased right now is scheduled for delivery in 2027, right? So your contracts are in today, in 2024, you know, you can put in your order 12-18 months in advance, because you know, at the end of the day, you're going to have- Absolutely. You know you're going to need it. Absolutely. You just put it, put it in earlier. I think that's a theme that we've seen in the data center space over the course of the last year, that further elongation- Exactly -of the pre-leasing. I don't know if the elongation will continue. I don't want to say that. We do not know that- Mm-hmm. But we have observed it right now. That's what we observed. Got it. You know, Scott, I would go to you for this question, you know, particularly related to the go-to-market, right? I'm no expert in building data centers, although I'm definitely curious. I think- You know, like- You know a little bit, a thing or two, but okay. So what I'm curious about is that, let's say one of these operators, they sign a 300 MW deal with a hyperscaler, right? When does Vertiv get involved? Is it, you know, right when they have the lease signed, and it's like, "Okay, we know we're going to need equipment," or do they come to you, you know, when they feel like, "Okay, we know we're gonna need it in X amount of time. Like, hey, can you line this up?" Like, how hand-in-hand are you with the customer? Yeah. Not to be coy, but I think it is, it is truly each customer is different, and each engagement is, is a little bit different in terms of the timing and what that, what that overall kind of technical and commercial engagement looks like. We pride ourselves on, on being there very early in the cycle- Mm-hmm. -especially in this environment where data center designs are changing very rapidly. Ultimately, like, what the buildings and the envelopes need to look like to take on kind of these newer AI and high-density loads. The kind of things we took for granted in data center design are up in the air a little bit, which is necessitating and driving a lot of very good early stage kind of engineering, technology, architecture type of discussions very early in the cycle- Mm-hmm -that informs our product selections, our product roadmaps, and things like that, that are going to be needed and used kind of 18-24 months down the road. We still have a fair amount of customer engagement where the design is kind of rock solid. They have a build plan, and 6 months out, 8 months out, that's when the infrastructure needs to be there according to their schedules. We have others that are looking at kind of much broader windows because of maybe they have a standardized design and kind of fungible product approach as part of their strategy for building for hyperscalers, for example, which allows them to extend out supply chain and de-risk some of those things from their perspective. So we truly see kind of the full gamut of flavors with our customer base, but I think the key message there is, there is an opening, and there's a lot of opportunity in market desire- Mm-hmm -for kind of the infrastructure-level design and engagement much earlier in the cycle. Got it. You know, I want to transition to talk a bit about orders, and the color that you gave, Gio, was really helpful, just a moment ago. Sure. You know, the 3Q order guidance that you gave was for a quarter-over-quarter deceleration in order growth. When we talked at the time of earnings, I got the sense that it was related to the timing of large orders. Perhaps, you know, these, you know, these lumpier orders that can move from 2Q to 3Q, 3Q to 4Q, right? Can you help us understand how, you know, the timing of orders has changed as scale increase? I think you talked about it's, like, 12 to 18, 12 to 18 months out is when, is, like, the time between when the order comes in and ultimately when it, when it converts to revenue. I think actually what I'm getting at more is: How large have these large orders become? Are they enough that, I mean, really could be a huge swing factor for you guys? We're talking $300 million, $400 million, $500 million orders. Can you give us a sense for the order of magnitude here? Yeah, maybe, maybe I take this one. So, I think the premise is, people can't exactly forecast when an order comes. It's not easy. Yeah. I mean, and so simply because you know that an order is coming, and the order still needs to go through procurement or, needs to be signed by the CEO of, of our customer. A CEO is not there, so you think an order comes, that day, and the order comes 15 days later- Mm-hmm -or sometimes, 15 days prior. Oh, that has always been there. So things have not changed. And in a world in which the average size of the order is, medium, whatever that means- Mm-hmm - you know, everything kind of evens out. We heard it several times today. We said kind of the average size of a data center is probably five, sometimes ten times what it was before. So simply, what statistically was kind of a compensated offset, this hard to tell exactly the day, the minute in which something lands. If that something is a spike in terms of size, then it swings, it can swing your quarter. So in a situation in which there is a reduction of opportunity to order, then all of a sudden forecasting orders becomes more difficult. So big orders can move back and forth and can change a quarter and without really changing the underlying business. That's why when we talk about trailing twelve months, we say, "Hey, 37, 37% trailing twelve on a year-to-date basis." What was it? At the end of June for Q2, and looking forward, we'll be in the 30-35% for third quarter or thereabout. So that's why, you know, it's false accuracy. Mm-hmm. So that's why we said, "Hey, guys, well, let's be prudent, and let's see how things will turn out. Got it. Yeah. Sounds like. So I would say it sounds like an element of conservatism, especially coupled with what you had just said, which is that the demand pipeline is stronger. Exactly. Right? Okay, that makes sense to me. You know, well, just going along the lines of the question that I was asking earlier, which is like, you know, how large are these orders becoming, particularly pre and post-AI? One of the things that I think through is that, look, in 2018, a large data center would have been 36, maybe 72 MW. Right before AI, I would have said, like, a 96 MW data center would have been a big data center. Now, I'm seeing, you know, 500, 600, a 1 GW campus- Mm -being signed at a single clip. So the scale has certainly increased. Is it kind of fair to say that the size of your large orders has increased uniformly or, or similarly to the size and scale of the underlying data center contracts that are being signed? I think that's a fair statement. Okay. I think that's a fair statement. I wouldn't ensure, let's say, the exact linear proportionality, but if we talk in terms of order of magnitude- Mm. Yes. Things have moved quite, quite significantly. You know, do you have internal kind of classifications where you say, "This is-" I'm making up, like, small order, medium order, large order. Do you have, like, points of demarcation that you- ways that you guys think about it? Well, you know, our spread can go from a few thousand dollars, some service or product orders or for our distributed IT business to hundreds of millions of dollars. So, anything in between is a little bit, you know, you know, it's, it's not a reference. It's, it, it's, it's just pure math. So I, I think we have an upper part of the large orders that has accelerated and increased. I wouldn't be too analytical here. When we look at our numbers internally, we're pretty analytical, but it serves a different purpose. Okay. I want to talk a little bit about customer mix. And you've talked about 75% of your revenue coming from data centers. If I close my eyes, I remember a pie chart that I think I saw in your Analyst Day presentation, which essentially showed that of that 75%, it was roughly half hyperscale, half colo providers. Am I remembering that correctly? Let me correct those here. Okay. It was. Of that 75%, half was colo and hyperscale, and half was direct enterprise and distributed IT. Really? So- Now, clearly, trends are such that one expects a different mix next time you see a pie chart. So when you say direct to enterprise, this would be like enterprise-built data centers themselves, like. Yep, yep. Exactly, on-prem. Okay. All right. Yes. But you would have all the services on that, so quite some stuff. Okay. So, let's say, again, the third-party data center operator that's leasing to the hyperscaler, that would be counted as part of that, you know, one, Yeah -one half, along with what you sell directly to the hyperscalers for their own self. Yeah. But also, the colos building for a real retail colo business would be in that colo hyperscale, 35%. Yes. So pretty much enterprise colo would also be captured in that? Yes, exactly. So pretty much, it's like self-build, plus the entire third-party colo market would be in that half, and then the other half would be direct enterprise and, you know- Distributed channel and whatnot. Got it. Okay. Okay. So then- But with very different growth rates- Yeah - obviously, obviously. Very different growth rates. Very. Okay. So along those lines, what I, what I wanted to ask is, I think you're in a unique vantage point here because you kind of, you sell direct to hyperscale, and you also sell to the third-party operators. You know, from your vantage point, you probably have a sense of the relative mix for hyperscale business between their own self-builds and their leasing amongst third parties. Like, have you noticed any shift in the amount of business that's, like, direct to hyperscale relative to what's going to the third-party colo market? Yeah, I can take that. Just anecdotally, and obviously don't have much in the way of numbers or hard data or hard facts- Mm. But from our view of the world, and a lot of what we've heard kind of discussed around this conference today, is there was a kind of a pendulum swing back to a lot more self-builds, kind of swung towards we're gonna try to lease and outsource as much as possible. It, it feels like it's, it's moving more towards the middle. Mm-hmm. And we've just got a lot in both camps, with a lot of emphasis both on kind of driving the demand in the lease space, as well as continuing down the trend of the hyperscalers kind of performing self-build, ensuring they're getting into the markets they need to be in, and still at least kind of driving their own approach to meeting the capacity and the demands they need. I think the prevailing thing we've heard across the board in a lot of today's sessions as well as out in the market is we're just, there's just not enough capacity, and they're leveraging both of those tools. Got it. Okay. As you consider your product portfolio, are there particular products for which you could call out standout strength within the direct to the direct hyperscale business, perhaps PDUs, switchgear? Where are you seeing, like, the strength on the direct side? It's really hard to pin one specific area. It's actually been a nice blend and a nice mix for, for our participation, both along, the major kind of large cloud and colo providers, as well as direct to hyperscale. It's been a pretty good mix of our product portfolio. Okay. Certainly, our engagement and our kind of quest into liquid cooling, as well as, kind of the powertrain- Mm-hmm - architecture that we've discussed along the way of switchgear and power distribution has certainly, it has certainly allowed us maybe some different levels of conversations, especially kind of building on some of the recent momentum we have with the integration of the E&I acquisition over the past couple of years. Okay. Now, I'll say this to you just from my, from my perspective. There's one hyperscaler's data center, whenever I, I'm in it, it's wall to wall with Vertiv equipment. So I'll just. I will say that. It seems like they're, at least one of them, is taking the full suite. I was just curious, as you think about some of the other ones, if there's relative strengths or it's, like, kind of a similar mix, but that's helpful. That's helpful color. You know, one of the questions that I would have is that, look, there are, you know, only so many hyperscalers who kind of like build their own, their own data centers. Have you given any kind of stats or color on the, your top customer exposure? i.e., is one, you know, do you have pretty much uniform revenue exposure with all the hyperscalers, or does one really stand out relative to another? - Well, maybe, maybe I can take this. We wouldn't go exactly into these details, but in general, we participate in the whole market. Clearly, there might be different share wallets for different customers, and these may even change over time. So nothing that stands out, or, you know, certainly not the case to go into the individual share of wallet with individual hyperscalers or even the large colos. But a good visibility of all the opportunity across, like, all the opportunities across the board. Okay. We're gonna transition to talk a little bit about lead times. As you consider your product portfolio, for which products have you seen the elongations or large elongations in lead time, or maybe conversely, contractions where, you know, look, you've ramped up production capacity, and now, you know, relative to what it was before, the lead times are shorter? So, first of all, you know, two things. When we talk about lead time, we talk about what we can provide- Yeah - to the market. And in general, if we look back a year, the lead times that we can deliver are way shorter than we could before. That might not be true exactly for every product lines, but in general, kind of an average shrinkage. And that's a very strategic decision, because there are markets that we serve, where lead time wins. We said there is an elongation on the requested lead time. That's true for the bigger projects. That's true for the hyperscalers and the large colos, and that, of course, we serve according to the requested lead time. Yeah. What we can provide when asked and what we are normally asked are diverging a little bit. Mm-hmm. And we like it because that gives us, on the one hand, the possibility to win with the shorter lead times. Again, not every time, not for all products, but we have examples of very meaningful projects that we have won recently because of a particularly short lead times. But in general, you know, our lead times are more than adequate to support the request of our customers, and the fact that they elongate their request give us more visibility on the capacity that we may have to make available- Yes - on the supply chain that we have to provide. So it's almost the two things in parallel, but very, very different point of view: what we can provide and what we are asked to provide. That's a great point, and let's double-click on that, which is the difference between the your equipment lead time and the customer lead time. When we talked about earlier, 9-15 months, going to 12-18 months, you're talking about customer lead time. Customer request. Right. It's like how long between when they put in the order, and they expect you to have- Exactly - equipment at their door. Exactly, exactly. The equipment lead time is the cycle time of production from when you would start to when you can essentially get it off the lot. It's you call Vertiv today and say, "I need a number of UPSes or something else," we can give that to you in a number of weeks. Yes. Clearly, if you ordered 20,000 UPS, it's a different story. We have to have a deeper conversation. But if a more kind of a normal size order, then we can deliver in relatively short amount of time. Would it be fair to say the majority of the product portfolio is inside of the customer lead time, like the lead time for the majority, the equipment lead time for the majority of the portfolio is inside the customer lead time? When the majority is the average size of order and the central gravity of the market, we are there or thereabout. I think it's a fair statement. Good. My gut would tell me that probably the only exception to that is, would be the medium and low voltage switchgear. Would that be a fair statement? I will not comment necessarily there. My comment will be, we are expanding capacity with switchgear, like, crazy. We have doubled capacity since we acquired E&I two and a half years ago, and since the beginning of this year to 2025, we're doubling, doubling it. So there is quite some demand there. Yeah, gotcha. If I'm investing in capacity, you know, there's a lot of demand there. So okay. You know, but, what I wanted to talk about is that the implications of that divergence between the equipment lead time and the customer lead time. Because to me, what that says is that the hyperscaler, the operator, is essentially saying, "Look, I know I need it. I'm gonna give you- I'm gonna give you this order. You figure out your production slots, whatever you need to do, get it to me on this day." It's almost like pushing over the management of the timelines to you, and they say, "Deliver it here," and you deliver it there. To me, that speaks positively for the relationship with the customer. I'm curious how you think about the evolution. Certainly, there is an element of strength of the relationship, but there is also an element, if I'm building a large data center, I have a very clear project plan, and there comes a moment in the project plan where it's clear what type of design and what type of equipment I need. They normally place the order there or thereabout, so that gives us a little bit of a wiggle room to the supply. Not always the case, but that is typically what happens. Said differently, when the basis of design for the data center is so forth- Exactly - which can happen after leasing, which makes sense in the context- Exactly - of the conversations that I've had today. Got it. Okay. Interesting. Very helpful. Thank you. So I want to transition and talk a little bit about input costs and a bit about cost management. As we think through the raw material inputs to your product portfolio, I imagine copper is, like, something that you guys will use a bunch of. Could you give us a sense of, like, what the largest inputs to the Vertiv process are? I'd be very curious. Yeah, I'd say very large portfolio. I mean, we go from, we're talking about switchgear- Mm-hmm. We're talking about liquid cooling, chillers, racks, PDUs, whatever. So the range is very vast. So there's not one, let's say, commodity or one raw material that stands out. Clearly it's a good, it's a good mix. It could be the metals, all the metals that, steel, copper, aluminum, it could be semiconductor, fans, a number, a number of raw materials. A lot of raw material is then embedded in the components that we, that we purchase. So we, we never think about, "Ooh, there is that commodity- Mm. - they're gonna change, change the profile of our cost structure. Yeah, it's a bunch of- It's pretty, it's pretty, it's pretty balanced. - different inputs, which completely makes sense to me. I guess what I'm, what I'm getting at is that if I rewind the clock and put, again, putting on the data center hat, not the equipment side of the hat, one of the things that I think the industry went through is that, we went through an inflationary period. There's a lead time between when the they sign a lease with their their customer and when they order the equipment from you. They underwrite certain equipment costs. That time passes, inflation hits, and all of a sudden, you're saying to the to the data center operator, "it's gonna be X more expensive," and they're like, "I did not underwrite for that." Right? So I guess what I'm trying to get at now is, you know, if we were to go through a period of inflationary shock on the input cost side, you know, what's the ability to pass through higher input costs to the customer? Like, is there a price indexation kinda clause where you can push it through? I'm just trying to get a sense for your insulation from really abnormal changes in input costs. Yeah, from our perspective, as we engage commercially, I think our approach mirrors a lot of our end customers' approach with some of our larger customers. We collectively, as an industry, I think, learned a lot of lessons when we went through our inflationary period and our supply challenges. Mm. At the end of the day, it's not a one-size-fits-all, but there's a lot of tools in the toolbox we use on our commercial agreements and the way we engage and discuss with customers- Mm - on price protection mechanisms, escalation clauses, certain abilities to potentially renegotiate backlog or, in certain instances, time phasing to pricing, that we have leveraged. We have bits and pieces and examples of all of those things. I would say, in general and at a very high level, we feel much better about the overall resiliency of the pricing tools and mechanisms we have in the backlog compared to where we were. And I would probably suggest that our customers and their customers probably feel the same way, that we've gotten more open in our engagement around some of those contracts. Got it. So you'd feel comfortable relative to maybe prior iterations and your ability to maintain your margins to the extent that there were increases in prices for- To be clear, I would never be comfortable in that environment, but just directionally, yes. Yeah, the other element is, of course, in general, a very, very keen attention on what happens on the market, raw material cost, et cetera. So we are—we have learned our lesson very, very dearly and very keenly, I would say. So I feel good in terms of the way also we price accounting for what can happen. So we plan for what can happen, and for what is unlikely, we have mechanisms, let's say, so very, very important to us. Okay. I want to talk about production capacity. Sure. So let's say, at the conference over the last 48 hours, you know, no surprise, Gio, you've been in the same conversations I've liked. Demand has been very strong. There may be a world in which demand accelerates further, right? Based on certain dynamics that, you know, I'm seeing. I'm curious, relative to the maximum output of your production plants, and where do you run from a production standpoint? And what I'm trying to get at is, to the extent that there was another acceleration in demand, you know, do you have the ability to keep pace with the growing demand of the industry without needing to invest in incremental production capacity? So a couple of things. We are investing constantly in additional capacity. So the acceleration is ongoing, so it's not that we have experienced an acceleration and then plateaued. The acceleration is ongoing. If we think about our capacity now, I don't know the audience that side, our capacity is constantly growing. We have conversations like this one. We have the conversations that we have with our clients and all our clients. We have a very, very clear funnel pipeline we relied upon, and we have a well, kind of a beefy backlog, let's say, that covers us well. All elements that we use to- Define what capacity needs we have in the next 3, 6, 12, 18 months out, and we invest for that. We talk about capacity, just like our own capacity. As important is, or more importantly, often forgotten, is the capacity of the supply chain. And orchestrating the supply chain is not trivial, because, you know, it's a lot of signals, and it's a lot more rigid. So we're more sophisticated than we were in the past, in terms of the way we pass our forecast to our customers. Having said that, no matter how good our forecast of capacity is, we know that it will be somewhat wrong because the timing cannot exactly be predicted, et cetera. So that's, that's why you heard me in several occasions talk about a 25-30% wiggle room- Mm-hmm. At any given time, at any point in this growth patterns, because we know that, you know, maybe capacity will be required a bit earlier than we need. Or again, I was making an example of a very short time, big order that we got to say, "Hey, we have a spike, and we have to grab the opportunity." Well, then we can use that extra wiggle room that would be a third shift, a full second shift. Mm-hmm. We can, in a way that we can add the capacity and then go back to the normal. And then, you know, go back to the normal growth of capacity that we have designed once we are through that. So to us, it's very, very important. That's helpful. As you think of where your production capacity is, would it be fair to say that your production capacity is geographically dispersed in accordance with your revenue? Let me rephrase. The way we design capacity is consistent with our demand, in the sense that, you know, we like manufacturing in region for region. Mm. That might not be exactly true for everything, but that's generally the philosophy. At the same time, though, we have become more capable of leveraging a global capacity, because, again, it is a matter of an ability to absorb demand peaks or to optimize the leveling of factory with which then translate into efficiencies and efficiencies that we all like when we see them in the P&L. So, it's certainly a two-dimensional, region for region, but also an ability to leverage a global footprint when it's needed. Okay, that makes sense. And then you've talked about investing consistently in capacity. I'm curious, from when the decision is made internally to invest in incremental capacity for, pick a product, I won't even name it, but for that capacity, what is the time between that decision and the first unit coming off the lot? It really depends on the product line sometimes, and what that further capacity step. Sometimes it could be a couple or three months. Mm. Sometimes can be, you know, 6-9 months, or a year, 18 months, really depending on what does that further step mean. One thing that I think is important now to remind everyone, for us, a big source of additional capacity is all the lean efforts that we are doing in the factory. So lean manufacturing, very big for us. Mm-hmm. And we have seen lean increasing the output per square foot quite significantly, and it will continue to happen. So it's not just the sheer CapEx dollars, it's also utilization and optimization and the factors. And it's not a small percent. It can be all the way to a double-digit capacity expansion. Can I, can I say- Gain, not expansion. Gain, that's it. That makes sense. I want to switch gears and talk about liquid cooling in a bit in the time that we have remaining. But my question here is, and I'll put it this to you, is I don't- As I look at the market, I don't see too many providers who are able to provide liquid cooling at scale, at least at the scale in the United States, that, you know, you hear we're talking about 100 MW, right? Like, and stamp that out over and over and over again. How do you, how do you think about your position in liquid cooling? And, you know, are there any other competitors that really stand out to you in terms of, not so much the technology, but the ability to produce at scale? Put that to you. Yeah, I think that's a very good question. I think we've been pretty public and open about the investment we've made in scaling the technology we have in the portfolio- Mm-hmm - with liquid cooling, trying to stay ahead of the demand and kind of trying to climb that steep curve. That's a global discussion for us. Really and truly, from a liquid cooling perspective, I think it's very important for us to take a step back and kind of hone in on the part of the portfolio we're talking about- Yeah - and then bring it back to more or less a system-level discussion at the data center. Because standalone CDU production capacity is interesting- Mm-hmm - but it's ultimately not all that relevant without a lot of the, I'll call it, the other pieces of the solution puzzle, including- Mm, mm - the ability to actually deliver it, the ability to scale it, like you've talked about, as well as the service content that we kind of include and bundle around it. Mm-hmm. Which is critical, especially in this, I'll say, the nascent ramp period of liquid cooling adoption. There's a lot of open questions around: How do you actually deploy it in a production facility? What are the right, I'll say, technical metrics on flow rates and temperature and things like that? Flow rate, temperature for the- Yeah. How do I need, if I'm a new data center operator, how do I go through the commissioning process to ensure that it's tested properly and handed over appropriately, and the SLAs match what's in my contract? There's a lot of open and developing questions now that, that I think your scale point-- we often think of just pure technical capacity and ability to, to build in a factory and deliver. Mm-hmm. But there's a scale to the overall business, and the ability to kind of talk solution, and the overall technical portfolio for a liquid cooling implementation in a data center that is very, very important as well. So you can scale and implement. So there's no one in your view that has a similar capability to Vertiv at this moment in the liquid cooling market? Is that- Would that be a fair assessment? I'm not sure I'd comment specifically on competitors' capabilities and where we see it. We're very happy with the position we have relative to our technical competence, and I would say the trajectory we've been on, and the investment we've made in capacity, and where we're at with that relative to the industry. Do you ever white label any of your solutions? From a white label perspective, I think where your question is going is, especially within liquid cooling- Yeah - the sale is not always direct. We do have partners, and we do have channel engagements, and channel partners, and different routes to market for the liquid cooling portfolio. So you will see, we will definitely see multiple routes to market to end destinations for liquid cooling implementation. You know, last, 'cause I see we're running out of time, have you given any updated metrics in terms of what liquid cooling as a percentage of revenue has become for you guys? Either as a percentage of revenue or maybe more applicable as a percentage of backlog, actually. No, that, that's not something that we have disclosed, and it's not something that we are, we are disclosing. What we, what we've been qualitatively saying is that liquid cooling has been, in the second quarter, the fastest growing product line in our portfolio. Though, overall, it is a quite, quite a balanced order intake that we, that we have taken. So when it comes to the long term, you know, I think the best approximation for now is to think in terms of, you know, total thermal market for critical infrastructure. Probably long term, 30% of that will be liquid cooling. So we still think that's a, that's a fair assessment. Perfect. Well, Gio, Scott, we're just about out of time. Thank you so much for being here. Thanks a lot, Michael. Really appreciate it. Thank you. All right.