Oh. Ready to rock? All right, let's do this. Well, at this point, is it morning, afternoon? It's afternoon, right? Yeah. It's afternoon. Yeah. Pretty officially. So we'll go with afternoon. So good afternoon, everyone, and welcome to TD Cowen's 12th Annual Communications Infrastructure Summit. Again, Michael Elias, cover comms for TD Cowen. For this session, we have Equinix, and from Equinix, we have their EVP of Global Operations, Raouf Abdel, who's back here with us for the second straight year, which makes me very happy. This is structured as a fireside chat. We have just under 30 minutes because I've gotten a little carried away with the prior sessions. Got questions prepared, but I will do my best, if you're okay with it, to open it up to the audience, and I see some people in the audience who may have questions. So I'll just- I'll throw that out there. There might even be some hecklers out there. Yeah. I understand. I think you could say there may be some hecklers. No tomatoes, none of that stuff. But Raouf- You behave. It's a pleasure having you. Thank you so much for coming. Thank you for having me again. It's always a privilege and an honor to be at the conference and to spend some time with you and this group. Awesome. Thank you. Well, let's kick things off. Do you mind if I do my- Oh, yeah. I forgot. obligatory disclosure statement? You have to do the obligatory, yes. That's right. You go ahead. Well, for those of you that are not in a public company setting, you don't have to deal with this, but at Equinix, we do. Some of what I'll talk about today contains forward-looking statements. Please read our SEC filings for more information about factors that could affect these statements. Awesome. Other than that, let's let it rip. All right, let's do it. Let's kick things off. For those of you who weren't here with us last year, can you give us a sense for, one, your role, but more specifically, given the evolution in the industry, how is your role evolving over the last year? I am sure it has evolved a bunch. Over to you. Yeah. Think about my role as sort of the development end-to-end cycle of building our data centers from real estate to energy to design build, the procurement aspects of that, and then ultimately, when we do build the data center, it is the operate side of it. So that whole life cycle of the data center, and that I have taken on a couple additional pieces of that to really try and bring it together and weave it together as one continuous sort of responsibility. The short answer to your question, but I am happy to elaborate, is that our world has gotten a lot harder. Yeah. Every dimension around what I like to use the term resources. Yeah. Those resources come in multiple dimensions. You have got people resources, you have got manufacturing, you have got real estate, and you have got energy, which is the one that gets a lot of attention. But where to go, where to build community sentiment, the backyard sort of dynamic that we see playing out, definitely putting some stresses and strains on the whole ecosystem of building data centers these days. Nothing I'm sure most of the crowd doesn't already fully experience or appreciate if you're out building data centers. That's right. One of the things that I think, building on your point about complexity, we had the second quarter earnings call come out with new CapEx guidance. Yep multi-year plan for building. Yep. At a time where things are already complex and challenging to build, the rate at which you're going to build is going to increase further. Awesome. As we think about you delivering on this plan, a few things. One is, can you just help translate for us, what does this mean in terms of $5 billion-$7 billion? What does that mean in terms of number of projects that you got to manage? Megawatts or gigawatts that you're going to bring online. As part of that, how the supply chain itself really needs to expand in order to accommodate Equinix delivering on that plan. Yeah, as you can imagine, Michael, we didn't from a standing start plan to increase the CapEx. We've been, for a couple years now, sort of gliding into a very different throughput in terms of the amount of capacity and the volume of projects that we were going to manage. We did, I know we talked about this last year. Yeah. We've also increased the size of every single project. But as you know, and maybe some in the crowd do, we've announced 52 some odd projects that are in flight that have been publicly announced across 33 markets across the globe. Behind those is another 50 that are in planning, development stage, securing either the land or the power or the supply chain that's required in order to hit these kinds of numbers. We've been planning this for a couple years, so the output is a result of those years of planning. Back to your earlier question the duration to plan is actually extending now. Yeah. So we're planning today what we're going to deliver in 2028. That's the way to think about it, right? 2028 and 2029. And so all of our land banking, power banking, pre-buy on the manufacturing side is all for future delivery, right? And what we're delivering this year, we put on a path two, three years ago. One thing that I'm curious about is, I think you've talked about a 3 GW land bank, I believe. Is that the right number? Maybe a little overstated, but Okay we are in the multi-gigawatt land bank. Okay. The reason I bring this up is, Tag, on an earlier session, from QTS, he made the point about what the historical motion for delivering a data center looked like. You buy the land, you sit on it for a few years, and then you go to the utility and say, "Hey, can I have the power?" There is a ramp, and then from there. I see you shaking your head already. No. What I want to get a sense of is, I appreciate the data centers that you are building now are bigger. I think 60 MW, somewhere around there is going to be the That's our template. is the standard block size. But when you think about the land parcels that you have in your bank, what's the right way to think about securing power? Is it that, hey, when we bought the land, we had an ESA with a long-term ramp that gives us visibility into the power? Or, because 60 MW isn't hundreds of megawatt scale, you can still go to the utility and be able to get the power that you need in a reasonable timeframe. Just help me understand how much visibility and certainty you have into the power that sits within your land bank, if you will. Yeah. So one of the changes, again, that we made two plus years ago is the flipping of the You don't go buy land and then secure power. You go look for where there's land, and then you look for power. Or, excuse me, land to support that power. We won't take down land if there isn't some line of sight to that power. Because you can't just tell the utility, "Bring me 200 MW, 300 MW of power to this location anymore. Yeah. You have to be much more thoughtful and planful around, okay, we see there's an opportunity to connect at a high transmission line over here to build a substation. You have spare capacity. You're having the utility conversation as the first step before you're thinking about the land, and we do that in combination now. It used to be, you're absolutely right, as mentioned earlier, it used to be you could take down land and just put a request in the utility, and they would show up at some point. If you apply that methodology now, it could be eight years before you get that power, because it requires upgrades, it requires infrastructure, it requires a connection. We're trying to pick locations that simplify to the extent possible. None of it's easy, but to the extent possible, where that transmission connection happens, where we can build a substation. For the most part, generation is the tertiary issue. It's really distribution and transmission that's the bottlenecks. Trying to simplify those two dimensions with where you pick land has been our strategy. When you think about it, here's something I think about. I think you said that, what, 80% of the CapEx is going to go into the top 25 markets That's correct, yep. or something to that effect? Yep. When I think of the top 25 markets, right, they're big data center markets for a reason. There's critical mass and density. There are also a bunch of operators. There also tends to be a correlation between the largest market and the biggest amounts of power constraints, right? Northern Virginia, huge market, very constrained in terms of power. You can say the same thing for Silicon Valley, and so on, right? That's kind of the direction I'm taking this is, I want to understand, since so much of this is going into the newer market, or into these established markets where you have a deep ecosystem, do you feel confident that, hey, power is not the thing that is going to gate our ability to deliver this capacity, and as such, the constraint sits somewhere else? Or is power something that we need to keep in the back of our heads, and then after that we can kind of tick it away? Absolutely. There's not a day I'm going to lose sight of the power complexities and the fact that we have to manage around it. You're right, those are some of the most constrained markets, but again, remember, this is a multi-year planning horizon for us. You have planned for this. One of the differences maybe from an Equinix versus others is some of our growth and expansion is phases and add-ons to properties and sites where we had a power bank or a power plan that was in motion. Yeah. We are adding the third building or the fourth building. Take Ashburn as an example. We are building in Ashburn today because those are projects that have been on the docket for years. Yeah. Look, power complexity, power challenge is not going away for anybody that is in the data center space anytime soon. You have to think about all dimensions. I would say, honestly, the key is planning. You have got to have a long-term horizon. One of the things that is going to enable us to continue to deliver capacity is our ability to move projects in and out, and pick which ones are on a trajectory to actually deliver, because the energy is going to arrive. 1 00 projects in the portfolio to work with. I talked about the 50 that we are very confident in delivering. The 50 behind them, which ones we pick Yeah is going to be a function of which ones are going to get energy. That is a fair point. I want to build on this. When we were on stage last year, you said something, or we were talking about this, how it is hard to accelerate the build cycle for capacity, right? Then I get on the earnings call and I hear, what did I hear? 7,000 cabinets got pulled forward? The first thing I thought is, "How did he pull that off?" Right? When I think about the mechanics of accelerating that kind of capacity Yeah how does one pull it off? I am looking at you for this. The short answer is really hard work and maniacal focus. But peeling that back a little bit so it doesn't sound patronizing, one of the things we really learned is that because of our dependency on capacity, it's a much tighter linkage now between capacity, delivery, and bookings. Yeah. In fact, I'm sure you noted on the earnings report. Our pre-sales are higher than they've ever been. Yeah. Not only once we have it built, do we sell it quickly, before we even have it completed, we're pre-selling it. So that relationship between when we build and develop and when we sell it is tightening, and demand- in many markets is insatiable. So we got less latitude for slippage, and so there's just a higher focus there. One of the things that we've really learned is that traditional project management would be a little bit more relaxed about managing float in the early part of the project. You would sort of let it build up to the back end, and if you run into a problem late in the project, then you're going to be late. Yeah. What we've really tried to do is make sure that the project doesn't consume float in the early part of the project. If you do that, and you don't run into issues late in the project-t hen you deliver early, if that sort of linkage makes sense. No, it does. Yeah. It makes sense. Again, we didn't deliver every project early b ecause of the portfolio we have of 50 + projects, and then some percentage we were able to deliver early because of tighter management of risk and float and supply chain management. Every risk dimension of project management. Yeah. I want to build on something you said, but I also want to be careful not to put words in your mouth. So one of the things that I think through is that we saw for Equinix very big bookings in the fourth quarter of last year. It also corresponded, from my vantage point, with 12,000 cabinets delivering incrementally. To your point, we know where vacancy rates are in the market, and we know that we need new supply. So obviously there's going to be a big focus on adding incremental capacity. You're pre-selling the capacity that's coming online later this year and then into next year, which will ultimately convert to bookings. Right. What I'm curious about is now that you've pulled these cabinets forward, is there, from the 2028 perspective, the ability to pull forward so that you can kind of create that smooth acceleration of where we can get to the point where we're delivering 14,000, 15,000 cabinets a quarter and putting up big bookings? From your seat as the one delivering the cabinets, do you see that as possible, or would you caution me, "Hey, Mike, there are some things in the supply chain you need to be aware of, don't get too carried away, and don't get over your skis? Yeah. No. Again, I think we feel confident that we're going to glide into those levels because we have the plans in place, we have the land banks in place, we have reasonable certainty around energy. There's always risk and there's potential delays, but we don't foresee those to be so exaggerated that we're not going to be able to deliver those. Our goal, my goal, my organization's goal is to continue to look at every opportunity to move up as long as we play within the capital envelope that we have Yeah available to us and that we've committed to the market. So within those levers, those parameters. Honestly, if we see demand continuing to increase even further, we'll look at whether further acceleration makes sense from a planning and a capital utilization standpoint. But of course, we'd guide to that. Of course. Okay. Sticking along this topic, one thing I think about is as I think about Equinix and its position in the market, first mover advantage in terms of the carrier-neutral data center model. But we've obviously seen data center designs evolve over time. We don't build 2N anymore. We build N+1, and for some of these AI lab deployments, we're building N. When you think of the broader portfolio, what do you think, or how do you think about the requisite investment that is required to future-proof some of the data centers that have been in your portfolio for 10, 20 years? Or is it just, hey, we're going to run them for cash. There's going to be a new build. Somebody is going to use them, and it doesn't need as much investment as you would think. How should I think about that from a portfolio standpoint? Yeah. I would say there's two points to tease out here. First is we look to put the workload in a place where it belongs, and not everything is super high density. If it is, we put that in our newer builds. Again, we have 280 data centers today b efore all the capacity we are talking about. The second point I would make is, keep in mind the draw in our existing sites isn't at 100%. No. If we have a churn-out event at one of our sites, our ability to add somebody new at a higher density is very viable Yeah because we have headroom of unused cooling, electrical or energy supply, and the space. As we churn out, invariably, we are putting somebody in at a higher density, and we are able to do that within some bounds and some limitation. Look, we are not putting 40, 50 kW a cab in a site that's 20 years old. That's not realistic, right? Can we put 10, 15? Absolutely. We've also retrofit a bunch of our older data centers with liquid cooling. Okay. In that scenario I'm describing, not only can we go higher density, we could apply liquid cooling as well. That can be direct-to-chip or that can be liquid-to-air. So we're managing the fleet for what's appropriate for every site's capability. We put workload where it makes the most sense. Of course, as we plan into the future, everything's liquid cool ready. We're building to higher density specs. I think the thing that is maybe a little misunderstood is that all the numbers we throw around tend to be individual cabinets and peaks, not averages. There's plenty of cabinets inside a data center that actually consume very little power because it's the cabling connections and the various other parts that make a data center work. So yes, there are 40, 50, 70 kW cabinets, but there are also four and five, right? Or zero. So, one of our latest builds is 18 kW average. Okay. Which means we can do 50, 60, 70, or we can do 10, 15 as well. What's different about modern age data centers is that the cooling is much more homogenous. The concern around hotspots, because we don't use raised flooring anymore, and the concern about, you can't put concentrated heat sources in one location, that's where data centers have really evolved. You basically have an envelope, call it a room or a pod, that can absorb so much heat. Put the heat anywhere you want. Yeah. That room and the way we design the infrastructure can take it any which way you put it. You can put it all together, you can distribute it. Whereas before, you had to be super careful. I'm going to have very distributed heat load. That's definitely one of the advantages of modern-day data centers. With the architecture that you're building for the newer data centers, keep me honest here, I remember the dynamic about hotspots and you having to manage to that. Now it is that, "Hey, with the cooling architecture, we can configure this however you want. The gating constraint is going to be power within the data center." Like you said, we'll have an envelope. You want to draw 300 kW a rack? Great. Awesome. You want to do 50? Fine. At the end of the day, it's how much power can I get you? That's really the constraint. The cooling has got to match. It is a one-for-one match. Yeah. If you are delivering 5 MW to a pod, well, you have 5 MW of cooling. Yeah, of course. That has to match, and your heat discharge has to be proportional. But yes, generally speaking, what you said is right. Okay. Now, one of the things, so just for context, I come from industrial engineering background, and one of the things they beat into you- I didn't know that. Yeah. That's it. And you know what's funny is they really beat into you is that you can't remove the bottleneck from the system. All you can do is shift it. Yeah. It's the weakest link in the chain concept. That's exactly it. Yeah, absolutely. Or if you're a process guy, Herbie, right? Where's the Herbie in the supply chain? Now, when you think about your process to deliver, right, where are you seeing the central bottleneck? Is there a central bottleneck or is it different across different markets? What is the thing that you find is the gating factor to you being able to move faster from a supply chain standpoint? We've talked about some of it. Yeah. The two things I would draw out are energy, but I feel like we've talked about that enough. Yep. Mm-hmm. The second I would say is trades labor. On the ground trades labor e lectricians in many locations are very constrained. There's not enough of them. When you look at what's happening in the greater Chicago metro area as an example, there is 15 GW. We'll see if it all goes through or not. Yes. Because every other day, we hear about a project being pulled, but there's 15 GW of planned on the docket expansions. Yep. With another 15 that's at least marketed to be planned. Yes. There is just not enough electricians. Again, one of the advantages we have is we have very longstanding relationships with both GCs as well as electrical contractors t hat trust our ability to move forward with projects, build projects, and they know that they can count on the relationship with Equinix. That is definitely a challenge. The industry is saturating. We can get to operators here in a minute. Yeah In terms of the build side, is saturating the key trades that are required to build a data center, electricians and plumbers probably being the most acute. Yep, that is correct. Along those lines, one of the things I find interesting when I have conversations with different operators is the differential in build cost. You remember there was a time where hyperscale was cheaper per megawatt to build relative to retail. Now it feels like that has flipped, right? I would be curious your thoughts on that, because I see that smile. From your vantage point, how do you think about, for the incremental capacity you are going to be delivering, how do you think about the build cost on a per megawatt basis? As part of that, how has that evolved in the last year? Has it stayed pretty constant, given what you are trying to deliver? I laugh because I have been dealing with this question and this dynamic for my entire time at Equinix. Yep. I do not know if you know this, but I have a construction background. That is the roots I came through. If you are in the construction world, you know that your cost is a function of what you build and where you build. To articulate that a little further, building a high-rise building in Japan or in Tokyo is different than building a warehouse-looking building in a field in Atlanta or the suburbs of Atlanta. There is actually a really wide range of what cost to build is, and I think it is one of the most misunderstood dimensions around data centers. Because it is easy to conflate, confuse, what do you include, what do you not include? Do you include all the supporting infrastructure? Do you include land? Do you include getting the energy now, which costs you big money now, Yeah that it did not use to. Costs us tens of millions of dollars to deliver power to a site now. It used to cost us nothing. Yep. Is that included? Do you include fiber? Do you include any internal capital? There is a lot of things that go into it. Yes, all of those. Yeah. The range is actually really broad. Just to throw out indicative range, it's anywhere from $10,000-$20,000, depending on where you are and what project you're talking about. That's backdrop and context. What I would say is that in these highly competitive and in locations where there's a lot of build going on, we're definitely seeing some inflation. Cost is going up to some degree. We haven't seen it as much, and it's definitely moderated on the manufacturing side of the equation. Okay. I would say back to whenever you have supply-demand imbalance, electricians are making $150 an hour these days. Wow. I'm not kidding you. Yep. It's like doctor wages now. That's driving some inflation on the cost, which is at the same time is forcing people to think through what work do you do on site, how much do you prefabricate, Yeah how much do you do offsite so that you minimize onsite labor to offset that a little bit. That's the continuous challenge that all of us have been dealing with for a decade plus, but it's at its heightened peak right now. All right. With that, the clock hits zero. Oh, wow. It's always a pleasure. We rambled that long? Yep. We did. We didn't take any potshots here. No, I thought it would be best if we didn't. No? But thank you so much, Raouf, for being here. Really appreciate it. Pleasure. Thanks for joining. Absolutely, pleasure. Thank you.