Speaker 1: 0:17

So you've been in the data center game a while. It's an exciting, it's probably the hottest space to be, I would say, out of any industry in the world right now, maybe outside of NVIDIA. Talk us through what you're seeing on your side and put that in context with how you've been in this industry for a while and what's changed.

Speaker 2: 0:36

Sure, sure, yeah, no, I've been in this space over 30 years Is that all.

Speaker 1: 0:40

Yeah, that's it no-transcript.

Speaker 2: 1:09

Same. Uh, you know, wind at the back type of feel. So, uh, no, it's great. Um, you can't complain if you're a data center operator these days. If if you're not doing well, then I'm not exactly sure what you're doing. And, uh, you know, these are really good times. I told my board in October. It's like, listen, this is the best market I've ever seen. And I only know one thing it's not going to last forever, right. But take advantage of it, while you can.

Speaker 1: 1:34

Yeah, so place big bets. Right now, it feels like the whole world wants to place bets in data centers, and I see that you've looked at the experienced operators who know how to place these bets, to be clear, who are in this game. I think you just raised a whole bunch of capital, correct? What's the plan there? What's that?

Speaker 2: 1:52

for? No, it's exactly what you said. We had amassed about 600 acres of land across a number of markets and in 2023, when we saw the demand that was occurring, you know we increased our development targets and, as a result of that, we needed more capital. So, for us, the money that we brought in the $2 billion, with Australian Super as our lead- investor yeah, which is also our investor. Yeah, I now have an education in superannuation funds, which is fantastic, but helping out all the Australian citizens.

Speaker 2: 2:24

We're trying to do our part, but in any event so that capital is earmarked for about 850 megawatts of new development that will start coming online in 2025 through 2028, predominantly in about 10 markets, but with the most of that in three core markets Dallas, atlanta and Northern Virginia.

Speaker 2: 2:46

So to your point yes, we're seeing, you know, an unprecedented amount of new development trying to come to the market, and a lot of that is in kind of markets that you know you would never have considered as a data center market right, and you kind of think about that in the big scheme of things and you know, certainly, you know, know people are doing that because you know there's available power in a particular market or, uh, the land is cheap.

Speaker 1: 3:10

Is it also lower cost power? Yeah, lower cost power, yeah you know leak, cheaper land.

Speaker 2: 3:14

But you know fundamentally, you know we think that, um, you know we'd rather be in kind of these core long term, as we call them. You know permanent data center markets as kind of technology this technology gets adopted. You know it's likely the geography around how it gets adopted changes.

Speaker 1: 3:31

Yeah, and one thing I love about data centers is just crazy, mind-boggling sort of metrics and most people don't understand what that is. But you just said 800 megawatts, which is, yeah, it's no small number. I mean mean there were data centers that were coming online that were like one megawatt or two or five or whatever, and yeah.

Speaker 2: 3:48

So you know the evolution's been dramatic. Right when I uh got the data bank, you know, we funded our largest data center. At the time, it was in 2017. It was, you know, a 10 megawatt data center and people were like, wow, that's really really big right, yeah. And now today, you know you're building 60 megawatt or 72 megawatt data center, 84 megawatt data center, so, um, and, and then the workloads, you know, and the customer requirements have increased, uh, significantly as well, where, again back in 2017, you know, a one megawatt opportunity was, you know, very rare and kind of at the top of the bell curve. Today, you know we're doing multiple, you multiple multi-megawatt enterprise deals and then, obviously, on the hyperscale side, you can get into the hundreds of megawatts on a certain transaction. So the scale of the leasing has increased dramatically as well.

Speaker 1: 4:39

So the one thing I wanted to ask everyone's jumping into this space. They're all like I'm going to build all these data centers, I'm finding this land, I've got this, I've got power, I've got access to power and land and that, oh, that's the problem. But this thing is not a simple process. It's also not a short process. So by the time you probably find the land, get development approvals to go there, plus the power. It's not as simple into the grid because you're talking about silly, silly amounts of power. So you, you've got to build what substations? You've got to help with the. I mean this is not.

Speaker 2: 5:08

It's uh, it's a multi-year process. Yeah, anything that you hear about today, you know, is probably three to four years out yeah, you know that's that's crazy. You've got to first, you know, find the land. You got to close on the land yeah you have to start your discussions with the utility at this point, the utilities. Have you know what's called the queue?

Speaker 2: 5:24

yeah and they put you in the queue and transmission becomes one of the biggest issues in the transmission required, which means there's right away required which is public hearings that are required so yes these things are, you know, really measured in in multi-year, especially at the quantities that but you're building transmission, or you.

Speaker 1: 5:43

How does that work?

Speaker 2: 5:44

no, we what? What ends up happening is the utility, depending on the situation, depending on is there an adjacent substation, is there a new one? There might be some kind of short amount of transmission that need to be built to, let's say, the substation that is serving your property right from another substation and you know, look the utilities. This is what they do all day long.

Speaker 1: 6:06

Yeah, right, but everyone's turning up, I'm sure.

Speaker 2: 6:08

But everyone's turning up and there's a lot of requests and you know these utilities are, you know, getting. They don't know what to do. Yeah, because they've never experienced anything like this.

Speaker 1: 6:18

And then you've got to get all the parts, like a diesel generator or something. I'm sure you can't just call them up and say, hey, I need a whole heap more diesel.

Speaker 2: 6:25

No, all that long. We call it long lead time equipment. The time intervals are not, anywhere from 52 weeks to 110 weeks.

Speaker 1: 6:35

Yes, which means that how are they keeping up with the production?

Speaker 2: 6:39

They have a long backlog right.

Speaker 1: 6:41

And that's again you have to, which is why you can't just turn up and say I'm going to start putting it to use.

Speaker 2: 6:46

I think you know people are envisioning. It's pretty easy.

Speaker 1: 6:49

Yeah.

Speaker 2: 6:50

There's a lot of moving pieces and to do it on time, on cost right, you have to make the commitments you know to the utility. You know deposits on their, you know their transformers and things like that. It's I'm reminded of a funny saying right, how do you make a small fortune in data centers? And I'll start with a big one? I think a lot of people are going to find that out.

Speaker 1: 7:12

That's exactly right. And the last one is you've got customers. So the other piece to this is I think there's a lot of risk in terms of how they look at this space. We look at it in terms of imagine being the company that buys all the chips and runs them and then they just keep replacing them there, company that buys all the chips and runs them and then they just keep replacing them. There's a lot of capital to go into that. It's kind of like the picks and pans and shovels. I feel like the data center is the space that you can almost guarantee almost that this will always be required. It doesn't matter what chip it is, it's always going to be burning a whole heap of something we're agnostic to technology.

Speaker 2: 7:51

It doesn't matter. Yeah, and even as the tech changes you can swap that out. We know it's always going to need to be in a daisy. We're never going to go back to like a nice cold chip that just runs and everything you know in the server closet. Yeah, exactly in the early 90s. Yeah, no, I don't think so. The data you know. Ultimately, you know the world lives on kind of this digital platform yeah, it was created over the last 30 years. It operates on this digital platform.

Speaker 1: 8:04

It's kind of working, working up to it.

Speaker 2: 8:05

There's no substitute. You can't put this physical stuff somewhere else. It has to go into a secure facility, which obviously is a data center.

Speaker 1: 8:14

Protected power. Redundant power cooling.

Speaker 2: 8:17

All that.

Speaker 1: 8:18

Yes, exactly, and it doesn't feel like it's slowing down. The piece is I mean, it's a few years now since ChatGP2 sort of exploded and then all of a sudden, sudden, everyone's just been rushing to this space. The hyperscalers are the biggest investors, that's correct, and I assume you're building a bunch of hyperscale infrastructure.

Speaker 2: 8:37

They're part of our customer base.

Speaker 1: 8:39

But you also have a huge number of data centers that we work on. That's correct. For what is traditional sort of colo for enterprise customers. Give us some of the metrics on that. I know you've got a whole heap.

Speaker 2: 8:49

Yeah, we have a great diversified base of customers over 2,500 customers. Obviously, the majority of them kind of enterprise or large technology. Our geography we have the largest geographic footprint of data centers in the US, with 70 data centers in 26 markets.

Speaker 1: 9:06

Say that again, the largest.

Speaker 2: 9:09

Geographic footprint of data centers in the US, with 70 data centers in 26 markets right, Say that again the largest Geographic footprint of data centers in the US. What does that mean? We're in more markets than any other data center.

Speaker 1: 9:13

Is that?

Speaker 2: 9:13

right, yep, any, any. Yeah, equinix is in about 14. Wow Digital is in about 17.

Speaker 1: 9:19

You know we're in 26, right, that's amazing, I did not know that.

Speaker 2: 9:22

Yeah, it spans across, you know, as we call them tier one markets and tier two markets. Yeah, and obviously you guys have, you know, come into a lot of our multi-tenant data centers because they're ideal from you know, the customers the clients needing connectivity solutions. Yeah, and a lot of these, as you know we've done. You know you guys are in our Salt Lake data center or you're in our Kansas City data center.

Speaker 1: 9:44

We want to be everywhere.

Speaker 2: 9:45

It makes sense, you're in our, I think, pittsburgh data center right. So you've extended your network into more of these tier two locations where there's less connectivity options and it's certainly kind of you know, the on-demand connectivity options that you guys offer. So I think it's been a great partnership. It really enhances the value of our facility.

Speaker 1: 10:04

Yeah Well, we appreciate it and the goal is to give that connectivity to all of your customers to make it easier for them to land. There's always a lot of discussion around like, hey, what right now? It's like find power, find land, find land that can get access to that power and then build. The other piece, which I think a lot of people either sort of think as an afterthought, is connectivity.

Speaker 2: 10:23

Yeah, and is connectivity.

Speaker 1: 10:24

Yeah, and it's like probably the last sort of leg stool. What was it last chair in the in the stool last leg? It's the last leg in the stool we'll get there last leg in the stool. Um, connectivity is this piece that often gets forget, forgotten. Yeah, but if you look at it, um, this is where we sort of come into play and enable that access for all of these different customers that you've got in there. How do you think about connectivity when you're looking at the?

Speaker 2: 10:55

build it's core. And I'll give you a perfect example A couple of years ago, data center development in Dallas was for the most part on the north part of the downtown. Development in dallas was for the most part, on the north part of the of the downtown, yeah, um, but you know that that place has grown so much so far north.

Speaker 2: 11:11

it's gotten too far away, so people started to think about south dallas, where there hadn't been historically a lot of the development and, um, you know, obviously it was a it was really a dark fiber kind of connectivity company that you know built a network down there. Okay, to enable a single hyperscale deployment. By doing that. It kind of unlocked that geography so that it is now becoming a massive data center cluster. We're, as an example, building a 480-megawatt campus down there, 860-megawatt buildings, and we have a lot of neighbors that are doing the same thing.

Speaker 1: 11:46

So that's funny. It's like it became the fiber that they ran, and probably some multiple ducks or a couple of hyperscalers. One then opened up the market for a whole range of others so really, you know, I call it follow the fiber follow the fiber.

Speaker 2: 11:58

Yeah, the fiber, then you can figure out.

Speaker 1: 12:00

You know good locations, uh to develop, uh you know new capacity and what do you think about everyone else that sort of decided all of a sudden to be a data center company right now, at this moment in time?

Speaker 2: 12:11

You know it's a free country, right, there's no restrictions. Look, it's obviously. You know there's a lot of excitement about AI. You know, to state the obvious, you know the data centers, the enabling infrastructure like we talked about. You know when you have, you know the president-elect introducing someone on TV that you know some unknown, you know Middle Eastern investor that wants to invest $20 billion in data centers. You kind of know that there's a lot of interest in the sector, right. But look, ultimately, you know we have, you know, a business with. You know that's been around for a long time. We have a track record with these very demanding customers. We have a development pipeline that aligns, we believe, with their needs. So, yes, of course there's going to be more players in the space, but we think that our customers are going to choose carefully.

Speaker 2: 13:00

And a lot of these new entrants really aren't there to build a business. They're there to create an opportunistic monetization event around some powered land. So, like everything, the water will find its level.

Speaker 1: 13:12

Yes, I suspect it will, and you'll be floating on top.

Speaker 2: 13:17

We hope so. We'll keep the boat watertight.

Speaker 1: 13:21

So is there any other things you want to sort of get out, anything you wanted to share that I can ask you about?

Speaker 2: 13:26

No, just, it was a great conference. I hope you guys are having a good one for a while. And you know it's also indicative of how much interest is there in the space. A lot of new faces here, a lot of new pools of capital trying to figure out. You know how to invest in this sector.

Speaker 1: 13:46

I know they never used to turn up or didn't turn up in years gone by, so I think there's a lot of folks really interested in this space. It's a mixture of either probably trying to find places to deploy capital, place bets, or just to figure out what the next phase of this is. Yeah, I think they're super excited, but they're also trying to figure out what plays out. It's actually tricky. Like I think you're in the game, it's very hard for you to probably see around the corner. The one thing that I feel is like it's just it's still a long way to go, like. If I look at unlike, I think, dot com, where it's all sort of land well, look, who knows we, you know nostradamus, looking back at all seen, but just the amount of training that needs to continually be done in that race.

Speaker 1: 14:31

It's unlike just rolling out fiber to get connectivity, to get to the internet yep, this seems like almost like an insurmountable problem, because you can train more and more and more and faster, and faster and faster, and then you're not going to stop training. So it's sort of like, unlike you're trying it once and just leave it, you're gonna have to keep training these things.

Speaker 2: 14:51

So yeah, I mean, I totally agree with you. It's very early innings, right? If you think about the technology number one, you got to believe that, hey, this technology is impactful right now. This kind of at large language models, transformer based large language models are going to make real business.

Speaker 2: 15:06

every day we see a new path, so I think you know the answer to that is it is, and now it's really a question of adoption and it's a question of how do you incorporate that technology into you know what we do today. You know the way we operate as you know humans and as a business world, and I think that takes a little time but you know, certainly there's enough evidence already to show. For example, we're seeing companies use this technology quite a bit for software development. It's a legitimate way to accelerate that activity, which is a very core activity. So they're already finding kind of verticals where the LOM is a really good solution and I think we're only at the beginning.

Speaker 2: 15:45

of you know, verticals, where the LOM is, you know, a really good solution and I think you know we're only at the beginning of as well. I like to call it, you know, enterprise AI adoption right. People are still figuring it out, and I think over the next couple of years we'll see more and more use cases that will then drive more demand.

Speaker 1: 16:00

Agreed. I had an interesting conversation last night. I mean a few things One last night. I mean a few things one jensen. What's so amazing is jensen.

Speaker 1: 16:13

It's like he's innovating 10 times faster than he was before, like so by the end of the year they seem to come out with a newer chip, a faster chip, and not just like a little bit faster, significantly faster. It's putting I think one of the guys was telling me that they're running these things. They're burning out pretty rapidly because they're running like 150 kilowatts into this rack. It's just like insanely hot and they're burning out in a year. But even then by the time you finish the year, anyway the new it's moved on to the new chips, or the new chip set, which is, I don't know, 100 million times faster than the previous one. So it's like I'm not sure how anyone keeps up to this. It's like you just buy the chips and all of a sudden, by the time you've deployed them, it's yeah, I mean it is an extremely aggressive release cycle, especially for semiconductor which is usually measured in years right.

Speaker 2: 17:00

And these people have, like you know, three generations over 10 years Unbelievable, you know. Look, most of the stuff that's in our data center is still. You know, h100s right, the Grace Hopper platform. You know people spend a lot of money on those. You know, in the last 18 months, you know our customers tell us, you know they have every intention of you know, sweating that cap out and getting a return on it, right, the GB200s? They're just starting to show up.

Speaker 1: 17:26

Yes.

Speaker 2: 17:26

Those are the ones that some of the racks need up to 130 kW per rack.

Speaker 1: 17:31

And that's where you've got to cool these things. That's correct. Are you changing the design of the data center or are you retrofitting with some sort of liquid cooling in some way? How's that play out?

Speaker 2: 17:41

Yeah, so a couple of years ago, about four years ago, we decided, because we had been starting to see more of these high-performance computing workloads that were higher density, that you know we needed a solution that could kind of pivot between water-cooled and air-cooled. So we came up with an MEP design that you know utilizes a closed chilled water loop as its core cooling medium.

Speaker 1: 18:03

And is that immersion or that's to the chip?

Speaker 2: 18:06

Yeah, it's underneath the raise floor or above the slab. Yes, and from there you end up in a. You know you distribute it via CDU.

Speaker 1: 18:14

Okay, and then?

Speaker 2: 18:15

it goes to either water, to the rack, like a rear door heat exchange or water to the chip.

Speaker 1: 18:20

Okay.

Speaker 2: 18:21

And the water to the chip stuff is pretty new. We only have a couple of those deployments because the equipment for that is really what's the difference between the chip and the back of the rack? So the back of the rack, it's kind of like a radiator.

Speaker 1: 18:33

Okay, yeah.

Speaker 2: 18:34

It fits on the back of the rack and obviously the computer oh yeah, it's rolling through Pulling that old air right and then it's getting hot, obviously on the other side, while water to the chip, the water goes to some manifold which actually sits on top.

Speaker 1: 18:48

Yes, yes, and there's a heat transfer, yeah, yeah, so it's okay to run inside the existing data so you can retrofit enough. It's not like we have to throw it out. No.

Speaker 2: 18:57

I mean, you know, obviously you know. We have data centers of a lot of vintage Some of them, though, wouldn't be able to do that, but the vast majority can and then, obviously, all the new data centers, where these new workloads are going, is exactly where we're deploying this technology awesome, it's crazy, it's so hot so it's uh, yeah, it's good time to be in the space you get out of bed and punch the air and then you just go and that's, it's beautiful uh, let's see how long you can keep going yeah, so.

Speaker 1: 19:23

So how big is the team now?

Speaker 2: 19:25

So we're about 1,000 employees now. Oh wow, obviously, we're spread out. By definition, our employee base is across our data center assets, and then, obviously, we have our centralized functions.

Speaker 1: 19:37

And how much innovation is there in the build itself, in terms of how you build the data center. I mean, a lot of people just think that's easy. Yeah, we're a real estate company, we're just going to put a slab and that's why I think everyone jumps into it. But there's a huge amount of innovation in terms of the design for the cooling, how you power, how you have a gun and power like. How much of that is just a standard off the shelf versus what you bring in terms of innovation inside.

Speaker 2: 20:00

Yeah, I mean we, you know, I think us, like a lot of other data center operators, we have a preferred kind of design basis when we look at these data centers. Ultimately, obviously, when we build we think about cost on a per megawatt basis. We work with our construction partner Optimizing that.

Speaker 2: 20:17

Yeah, exactly, Our construction partner, our architect, our engineering firm, our internal team firm, our internal team. We're constantly cost engineering and figuring out how we can drive that cost lower, Because obviously if we can deliver a redundant design for cheaper, then that is, we can be more competitive in the marketplace. So that is an ongoing effort across the sector and over time you're able to engineer some cost out, but you're also having inflation at the same time.

Speaker 1: 20:48

Yeah, yeah, you've got costs go up, yeah, but I guess the benefit of it, ie the incentive to reduce cost, is a huge power game. Ie if you can improve even small percentages of power, you reduce costs. So there's like a real incentive to solve for that and the other incentives it solves. It's good from an environment perspective, correct? I know you guys are focused on that as well. We are, yeah, yeah.

Speaker 2: 21:14

We, as you know, we're part of the Digital Bridge group of companies, which has a corporate company Incredible company, by the way Thank you. It's been great partners of ours since 2016. And that mandate is to be 100% renewable on Type 1, type 2 by 2030. Today we've really made a lot of progress along that way. So today, with renewable power, with RECs, with kind of grid mix, we're over 60% of the.

Speaker 1: 21:42

Which is impressive Because I think roughly the think the roughly US is what? 20%? So you're actually getting an additional 40%, which is not easy. You got some countries like Brazil where they go we're 90% sustainable power inside the data center and you're like, yeah, but that's because you're 90% sustainable power in Brazil you basically plug it in like.

Speaker 2: 22:04

That's why it's a very popular data center 90% sustainable power in Brazil. That's right. You basically plug it in my grid. That's why it's a very popular data center, or one of the reasons.

Speaker 1: 22:09

Yeah, yeah, so it's harder to do that in the US, exactly.

Speaker 2: 22:13

We're limited by. Ultimately, as I talked about, we're in so many different markets. There's so many different utilities. We're really dependent on the utilities.

Speaker 1: 22:21

What's providing them in that region To kind to give us a product that is renewable, or they're increasing their grid? Yeah, and as they increase you just take that feed yeah exactly Brilliant Well, we appreciate the partnership. We're excited about your growth. We're also going to keep building out into the data centers you've got, service the customers that you keep bringing on, and world domination.

Speaker 2: 22:44

I love it. Yeah, it's been a great partnership.

Speaker 1: 22:47

Thank you very.