Like starting on time. Welcome to the 2024 RBC Global Communications Infrastructure Conference. I've lost count of how many years we've done this, but pleased to have Mark Ganzi here, who was here at our inaugural event many, many years ago. Industry looks a lot different today, obviously. CEO of DigitalBridge, and welcome.
Thanks, Jonathan.
Yeah.
Good to be here. You're never laughing. I think the first one was us at a restaurant, and it was, like, 20 of us at a table in Chicago in 2010. So-
Yeah.
It's nice to see how this conference has really grown up and evolved, and you know, the success of a conference is really when you see your peers show up, you know, C-suite level peers show up, and you can have candid and frank discussions, and I think you guys have curated that, so kudos to you.
Thank you, and, pleased to have you kick off. I think you can speak to all the asset classes in all the regions, but we only have half an hour. So, why don't we kick off with some macro questions, and then maybe give us your view, overall for kind of the next six to twelve months as you think about, you know, elections, rate cuts, the overall transaction environment related to both new investments and monetizations. Any kind of general comments that you could, set the stage for us?
I think I look at investing kind of with sort of three different filters to think through. One, certainly the political environment is pretty interesting. I think, you and I have always talked about this, that digital infrastructure is purple. You know, it's not red, it's not blue. It kind of sits in the middle of the fairway.
I don't see a lot changing in a Harris or Trump regime from certainly from an FCC perspective, and having been around a couple of these things and first having sat for Obama's FCC Broadband Deployment Board, and then sitting for Trump's, there wasn't a lot of change, and I think today's FCC is probably gonna continue to change more, I think, but I don't see there being a huge tectonic shift from a regulatory perspective, particularly from an FCC perspective. But I do think you're gonna see, Jonathan, a much more aggressive DOE. I think everyone needs to start paying attention to that. I think the current sitting you know Secretary of Energy, she's very focused on ultimately how data centers disrupt the grid and how perhaps that affects consumers.
And they're trying to link, you know, power consumption from data center providers to ultimately that has the consumers paying for it, not the hyperscalers paying for it. So keep your eye on that. And we've had some direct meetings with her in the DOE, and I think that's going to be an issue. In a Harris White House, that will be a real issue. I don't think it's an issue in the Trump White House. So power and how power is ultimately regulated will largely dictate a lot of the success in this room. So we need to keep our eye on that, I think, a little bit. Rate cuts, you know, we telegraphed two rate cuts this year. I think that was correct.
I think we had 75 bps baked in our models and another 50 bps next year. But ultimately, rates will moderate, but I don't expect anything greater than 125-150 basis points over three rate cuts, maybe four. The first one was bigger than we thought, and probably the second one will be smaller than we thought. Again, you invest in digital infrastructure through a 5-, 10-, and 20-year prism, so interest rates can certainly change models and change outcomes, but, you know, we're building, you know, what we think are multigenerational platforms to serve customers over a long period of time, and so rate cuts aren't going to effectuate, to a large degree, our returns.
We've been able to invest through interest rate cycles over the last 30 years, and you've—you and I have been friends for over 25 years, so we've seen this, we've seen this play out. So I'm not terribly focused on that. But politics, rate cuts, and then just general macro setup. You know, customers are still spending money, customers are still signing leases, all sort of, you know, core four food groups, as we like to say, in digital infrastructure, are working, and certainly we can drill down to each of them, but CapEx remains robust, and we haven't seen any what I would call recalibrating or deacceleration of CapEx spending.
So you mentioned DOE. Any comments, given that you've got platforms that operate, you know, in Latin, Europe, parts of Asia, but any kind of tops of the waves you could hit on as it relates to, say, the European Commission, or some of the larger APAC jurisdictions that you operate your data center platforms in?
Well, I think, you know, a lot of the infrastructure in Asia. Let's start in Asia. A lot of the infrastructure in Asia is relatively new, so they don't have the same grid transmission issues that we have. I think you do hit into certain markets, whether it's Singapore, Johor, Hong Kong, Sydney, Melbourne, and Tokyo. Those are all markets where you have grid sensitivities, and will-serve letters are taking anywhere from 12-24 months, so Asia- Pac is a really interesting opportunity. It's a laggard market. It's trailing the U.S., and it's trailing Europe in terms of, you know, workload consumption, and those AZs are developing, and they're developing rapidly.
And I think what you're seeing is some of these governments, particularly Australia and Japan, are very calibrated about how much power goes out to the data center community. So we'll keep our eye on that. So far, we haven't been blocked. Johor was a tough environment for us, but we got a really good outcome there with the will-serve letter. Singapore is still shut down, unless you're building, you know, 5 MW, maybe you get that done, but anything at scale, you can't get done in Singapore. Tokyo, same thing, so Osaka has become the tethered market to Tokyo, just like Johor has become a tethered market to Singapore, and like Melbourne's become a tethered market to Sydney.
So these secondary AZs are really opening up in Asia, and that's where we're seeing some of the big demand points. I think in Europe, you know, the grids are challenged. The flat markets continue to be a problem other than Paris. Paris, we see that's been an easy market to obtain power. Amsterdam still has its challenges. Offenbach, we still need a grid upgrade in Frankfurt, probably a 2028 event. Slough, you find little, what I call, Tetris pockets of power, like 6 MW here or 10 MW there, and then you gotta go to Reading, or you gotta tether out to Cardiff. So there's, you know, London continues to be a challenge, and so if you look at the, you know. And then Dublin has no power as well.
The Dublin Municipal Power Authority has shut all power down, so you know, the power that we're deploying in Dublin is an alternative energy solution that's off-grid and grid independent, so we actually find that Europe is a slightly bigger challenge than the U.S. right now.
Got it. Maybe hitting back on the macro themes, how are traditional institutional LPs adapting their allocations in private markets? And how do you envision the competitive landscape evolving with major asset managers like Brookfield and Blackstone focusing more on infrastructure and digital assets?
Yeah. Well, look, we saw last week a you know formation of a $30 billion fund, you know, that's backed by the UAE and Microsoft and a few of our frenemies or competitors. And my philosophy on this is you know pretty much kind of the background of where my family started 95 years ago in the restaurant business. And somebody always asked my dad, "If another steakhouse opened up on the street, how did you feel about it?" And he would always say, "I feel great about it because it drives more traffic to our restaurant." So I adapt that same philosophy in fundraising and digital infrastructure today. So if GIP and BlackRock are raising capital or Blackstone's raising capital in digital infrastructure, we're okay with that.
You know, our customers are spending $250 billion a year in CapEx to deploy AI infrastructure right now. That's a pretty big bar bill. That thirty billion that they may or may not raise, we don't know how long that'll take. Right now, we're focused on what we're doing, which is deploying out of our third fund. We've been able to raise, you know, a lot of capital for our third fund, and we're continuing to form capital. In this quarter, we've had a couple of good logos that came in, and we've been continually, you know, adding capital. I think so far this year, Jonathan, we've raised, like, $16+ billion of capital. We anticipate raising more capital through the third and the fourth quarter, so...
But I would say on the fundraising trail, you do hear the same thing, which investors want to know that you're returning capital. The most important metric you hear on the road today is DPI. And DPI, if you're in the fund business, is, you know, if you wanna go ask for a new check, you better start returning some capital from some of the old checks you got from somebody, because if you're not returning that capital and you're not showing the returns, and I don't mean the synthetic returns that GPs are putting out and saying, "We're marking our assets to here, and this is the return." That's not a return. A return is when you actually return the money. So we've had eight exits in the last 24 months. We're working on exits nine and ten. We're returning capital.
We returned capital in the first quarter, we returned capital in the second quarter, and we have a plan to return capital in the third and the fourth quarter. So as long as we consistently deliver the right returns in terms of DPI, our fundraising cadence will accelerate. And we're seeing that already in the back half of this year. LPs want to invest in AI infrastructure. Now, I would say, generally speaking, when you're sitting with an LP today, there are only three things that matter. I hit the first one, which is DPI. The second one is, you have to have track record. If you're a first-time fund, you're probably gonna have a really hard time raising money. So what LPs are doing is they're going with brand names, people that they know and they respect.
You talk to some of the big asset allocators in Asia and in the Gulf, and they'll tell you, "We're generally shrinking our GP list." So that's interesting. You know, so if your performance is in top, top quartile and you haven't been around for a decade, there's a pretty good chance you're not gonna get a check. Investors today are flying to safety. So if it's private equity, you know, they wanna be with, they wanna be with Blackstone, they wanna be with TPG. If it's, you know, renewable energy, they wanna be with Copenhagen Infrastructure Partners. If it's digital, they wanna be with DigitalBridge. They wanna be with the leader in that chosen vertical, provided the returns are there, and they're not really interested in taking a lot of risk, and they're paring back their GP list. So that's really important.
The third thing I would say is, if you come with proprietary pipeline and you come with great co-invest ideas, you're gonna be sequencing yourself to the top of the list really fast. So in our first two funds, we delivered $25 billion of co-invest against $12 billion of fund commitments. So that 2-to-1 ratio of delivering really interesting and proprietary co-invest is why we hit our hard cap in both fund one and two, and why we have confidence we'll hit our hard cap in fund three. Having really good, unique ideas that investors can only get access to those ideas by being on our platform, is what really differentiates ourselves. We saw that with Vantage North America. We raised $2.2 billion in co-invest in 30 days. We're seeing that with some of the other ideas that we're executing right now.
Our LPs really like some of the new ideas that we're generating out of the third fund, and if they want access to those deals, they have to get into the fund. So that's a good way to really push fundraising is have good co-invest.
Absolutely. So maybe talk a little bit about your investment approach in the infrastructure space as an asset manager versus how REITs would look at it, advantages of your asset management strategy, and then maybe more broadly, any potential misunderstandings on the DigitalBridge investment case that you find yourself confronting?
First, let's start with investment philosophy. I think we maintain the philosophy that you wanna back great businesses and back good platforms. A great case study that we use is Switch. That was a story you covered. You covered Rob and Tom and that management team for a long time. As a public story, Switch really wasn't working. And why wasn't it working? It wasn't working because the types of at-bats and the types of situations that customers were presenting them with were CapEx decisions that were measured in hundreds of millions of dollars. And so as a public company, and to be direct, Equinix and DLR struggle with this.
And so if you're gonna go compete for those big workloads, and you wanna commit $1 billion of CapEx to a project, you have to have what's called ready-aim-fire capital. If you don't have that ability to take the shot, and you don't have that capital at your discretion, you can't go compete for that business. So Switch was always constrained by the fact that they couldn't go do the big deals that they wanted to do. Take that business private, double, triple the EBITDA in 24 months, feed the business with the appropriate CapEx. Unfortunately, you know, steal someone from RBC to be the CFO. Sorry, Nick. But, you know, ultimately, that ability to form capital around Rob and his team, and also to make decisions quickly, has been massive.
We've already achieved, in less than two years, six years of our underwriting leasing, and they're sitting on a backlog that's over 3 GW . So Switch is one of those great examples where if you can deploy third-party capital from LPs, you can really accelerate the business. And if Switch were still public today, it would have grown EBITDA by 10%-11%. Instead, we've doubled and almost tripled the EBITDA in that in a very short time period. So we like to give businesses, good management teams, a lot of capital, give them our customer relationships, you know, teach them how to engage in what I'd call, you know, risk-adjusted behavior. And so we've been rewarded with Switch because putting it into a private format and putting it under DigitalBridge in that asset-light format has allowed it to flourish.
Look, you've been doing this a long time, Jonathan. Our business is complicated. It's a CapEx business. If you don't have capital and you can't move quickly, you won't succeed. I mean, we know how fast the industry moves. Our customers want us to react immediately. And so when you're public today, and you can't form capital, and you can't go out and do secondaries, you end up... Good businesses, by the way, nothing wrong with American , Crown, Digital, and Equinix. They're all friends of ours, and I think they're fantastic businesses. But in this environment, where we're building AI infrastructure, which is a $5 trillion opportunity, you have to have a lot of capital, and you have to be able to go fast.
And I think what constrains the traditional digital REITs is that they are constrained in that respect, that they can't take the big shots on goal. And when they do take big shots on goal, they're having to do it with other people's capital. So CoreSite with Stonepeak, right? That was what American Tower had to do to go do that deal. Equinix has now done multiple joint ventures in xScale, to go do hyperscale data centers. And I think DLR has done an artful job of creating... They have their JV with Blackstone, but they've also sold assets, and they've recycled capital, and Andy's done a nice job getting that business to the right place.
But if, make no mistake, if a 600 MW workload comes in front of me and Andy, there's a pretty good chance I'm gonna win. And it's not because Andy's not good, and it's not because Digital Realty is not good. It's because I can go fast, and I have the money at my discretion. And so getting to the investment thesis around DigitalBridge and why we converted, why we went from a REIT to an alternative asset manager, is what we saw is that the world was changing, and we saw that our customers needed, you know, a lot of capital, and we saw that the workloads were changing. We saw that leases were moving from 10 to 20 MW to 100 to 200 MW. And we saw that customers weren't purchasing two pairs of fiber. They were purchasing 48, 24 pairs of fiber.
That CapEx cadence has completely changed, and as a consequence, we had to change. That's why we made that decision, you know, four years ago to de-REIT, and it's also part of the decision why we became an alternative asset manager in our current format. This first quarter of this year, we had our first clean, you know, reporting cycle as an asset manager.
So there's four empty seats in the front row and three empty seats in the second row if people wanna come in. You mentioned Tetris earlier, so kind of good visual for what we're looking at here, the seats. You know, you can get a lot of megawatts covered in terms of meeting demand with the powered shell, and then you can maybe get more share of wallet if you do turnkey. But any broad comments as you look upon your portfolio companies as to how the preference and trend may be changing between powered shell and turnkey?
I think it's a function of the AZ, and I think it's a function of what the customer is trying to achieve. It kinda actually reminds me of small cells sometimes because some of our customers say: Look, we want you to use our fiber plant, so we want node and pole only. And some people say: Look, I really don't have the fiber plant. I need you to do complete turnkey for me. So I think we've over the last, you know, twenty-plus years, been pretty adaptable in listening to customers and sort of going where they wanna go, not where we wanna go.... And I think also having the right products that can match and pair to a customer is really important.
So if it's an edge compute workload, and it's that kind of, you know, 5 MW to 20 MW workload, Raul is gonna be DataBank's gonna be really responsive and competitive. We're in 30 markets, 70+ data centers, and I would say that that edge compute space is probably the fastest-growing space that we see. And when we show up and we're competing for a, you know, 10-MW workload in the suburbs of Cleveland, there's nobody competing against us, and I like that. That's a good, that's a good setup. If it's, you know, 200 MW in Ashburn, I'm in a knife fight with, you know, seven other people, and I've got to demonstrate I have power, and I got to move quickly, and that's a race to the bottom.
So I think the location, the workload, sort of sets the requirement, and then you sit with the customer, and you try to understand and unpack, you know, their needs from a power configuration, power density perspective, what type of connectivity do they need, you know, and then we decide whether we're just gonna go do a BTS or whether we're gonna build a campus. Because traditionally today, Jonathan, what's interesting is if one of our customers is trying to build a location with a certain power profile to serve a certain workload, it's interesting to me that usually there's a second and third tenant that actually have a similar requirement in the same AZ. So there, I'm not interested in powered shell, I'm interested in building a campus.
I'm looking to have a multi-tenanted approach to that, and provided we don't hold the customer up, and we can still go fast, we can achieve the customer's needs, and we can achieve our needs. We've kind of stayed out of powered shell. It's not what we do. It's not what Switch does in private cloud, it's not what Vantage does in public cloud, and it's certainly not what DataBank does, you know, from an edge perspective. So having three products and three different platforms that can address private cloud, public cloud, and edge, and then, of course, you know, ZenFi doing kind of the near edge with our RAN hubbing business.
We think we attack all the verticals very surgically, and we attack the workloads surgically as data gravity takes shape from, you know, from a training model to inference, to ultimately back into the public cloud, back to edge, and then back to, you know, near edge, and then, you know, micro edge. Data is gonna keep evolving. The shape of data changes, the workloads change, but it is still very much a geographic business. It hasn't changed from when you and I got started in 1994 looking at towers. It's about the location. It's about, can you get there faster? Can you get permits? Can you get power? Can you get fiber? And can you do it better and at a better cost than your competitor? By the way, it hasn't changed in thirty years. That's how you win business in digital infrastructure.
You've just got to be first, you've got to be fast, and you've got to be prepared. I think those are kind of the key, key tenets that we believe in across all 46 companies we own around the world today.
So you talked about a knife fight when it comes to doing some of the bigger deals in heavily high-interest markets like Ashburn, but there's also energy transmission constraints a lot of the time. So putting that all in the mixer, how do you view development returns now versus what you might expect in a year versus, say, a year or two ago on lever development returns for hyperscale? Any movement that you would want to comment on?
I would say from two years ago, it's up about 100- 300 basis points. Again, if you have the land, if you have the permits, and you have the power, you have the pricing power, right? If you're gonna take permitting risk and you're gonna take power risk, you're gonna find that your returns aren't as healthy as the situations where we've already taken the risk and we're well ahead of our competitors. And I also think, again, it gets back to the market. Like, you know, now Ashburn is interesting. Pricing has moved up because, largely because very few people have a lot of contiguous, significant power loads left. So the capacity that we have in Vantage in Ashburn five, six, and seven is really valuable.
The capacity that DataBank has in Ashburn four and five is really valuable because it's ready, and so therefore, we're able to price that higher because it's ready to go. I think in some of these new markets, what I call these frontier markets, like Reno and Atlanta, Georgia, again, you know, those who do have power and those that have been able to demonstrate some grid independence, you can price accordingly, and what I mean by that is, if you've got the ability to have your own microgrid transmission infrastructure and you can ultimately accelerate the customer faster, the rents go up. I do think that rents sort of probably plateau in the next 12-18 months, but all of this, Jonathan, is contingent on power. Power is the big gate.
You know, if power becomes readily available, and we have a, you know, a red White House that sort of clears the path on that, you know, pricing power probably stabilizes, if not, maybe even diminishes. I think if you have a very hawkish White House and you ultimately have power that's gonna be constrained, I do see pricing moving up. I see our existing locations being more valuable. And I think in locations like, you know, for example, in Brazil, we've got a really interesting business model there where we are not on the grid. We have our own grid infrastructure, we have our own microgrid, we pull our power from the north, from Minas Gerais and Brazilian, we're 100% hydro there, and we're supplemented by the grid.
We connect to the grid because we sell power into the grid. We're a trader of power during the day, and we're an offtaker at night. But that ability to have your own transmission infrastructure and your own ability to trade power is such a competitive advantage. And, you know, in Campinas, we're happy to sell power to our competitors because we have enough power that we can collaborate and work with Ascenty and DLR, and we can work with Equinix. And, you know, we have an Equinix location right in the middle of our big campus in Campinas, and we would sell them power. We would sell them our renewable power.
We're trying to create competitive advantages where we can be, you know, grid independent, is what we call it, and we've been able to demonstrate that in Reno, we've been able to demonstrate that in Campinas, and I think we're gonna continue to be pushing very hard on that. You'll see over the next year or two, that's gonna be something that we think is a definite comparative advantage for us.
So Brazil has something like 92% renewable mix, large population, large land mass. Do you see large GPU clusters about to be deployed, already deployed? What's the demand profile look like in that jurisdiction?
Brazil's been spectacular. I think that's become essentially the Ashburn of Latin America. Campinas has become that, and there is more land to grow. We've got more power to put to work there. You know, and the customers like the fact that it is, what we're offering at scale is a renewable solution, so, we win there. I think, the question is: Is that model replicable in Santiago, Bogotá, Mexico City, and to a lesser degree, you know, I would say to a lesser degree, you know, Buenos Aires, which is a volatile place from a power perspective. But I think we'll test that in Santiago, we'll test that in Bogotá, and we'll see how it goes. But we really have a big competitive advantage in São Paulo and in Campinas.
So with a lot of the demand currently around AI being around large foundation models, how do you see inference evolving? You mentioned Cleveland, so does inference exist in markets like that? And then pivoting maybe to U.S. towers, and around just leasing around 5G. Maybe we can kind of finish the last four minutes talking a little bit about what you expect around rental revenue growth in leasing for towers.
Yeah, I think on sort of generative AI workloads, you know, moving out of training models into inference, we don't see those demand characteristics existing today. We do see it coming quickly and fast. And again, our model is we're looking at the deployment of public cloud, right? So you go back to, you know, 2011, when the public cloud was first sort of conceived from an infrastructure perspective, and we're sitting here today in 2024. We're kind of 13 years into that journey, and most of the hyperscalers would tell you they're sort of about 70% deployed across, you know, their existing public cloud workloads. So that's a lot of that stuff that we're talking about in Pittsburgh or Cleveland or Bluffdale, Utah, those are those sort of 5, 10, 20 MW workloads that we really like a lot.
And that, again, is still, you know, building out public cloud. I think from an inference perspective, you're probably, you know, two to five years off, and today, our customers are still deploying these massive, you know, language-based models. And so, maybe that accelerates a little bit, Jonathan, but right now we don't see it. We don't see it as pronounced as some of the public cloud workloads. But I think secondary and tertiary markets in AI kind of have their day three to five years from now, following that same cadence that public cloud has followed, which we started doing a lot of the big edge hyper edge workloads started to, you know, sort of pop up in 2011, accelerating in 2012, 2013, and now 2021, 2022, 2023, and now 2024.
Got it. Now, so U.S., we, we've kind of had a halftime pause in, say, middle of 2023.
Yeah
With the, with U.S. leasing. Is that, is that changing?
We do see it coming back. I think that, you know, 5G will play out kind of in three sort of CapEx cycles. I think you'll see the first cycle's over, which was the one-for-one overlay against existing LTE sites. So that's really a macro overlay. Most of the customers will tell you they're kinda 80%-90% overlaid on top of their existing infrastructure, and there's some more amendment traffic coming. I think what's next up is we're starting to see densification.
And what's interesting about 5G versus sort of that 4G CapEx cycle is, the customers are coming back now and densifying, saying, "We're just gonna densify in this band at this location." So it's a bit more surgical than I think we're used to seeing, where in LTE and LTE plus, Jonathan, when we were densifying, it was two bands or three bands being introduced in a fill-in site. It's not unusual today for a customer to come to us and say, "Okay, we're gonna infill here. We're gonna put this spectrum here. It'll be three antennas, you know, one- you know, three RUs, and then fiber to the base band." And so that's a different kind of... That's not a colo as you and I would define it.
I think we have to get our minds wrapped around that customers are being a lot more surgical about how they're deploying spectrum and how they're going to deploy resources. And so I think there'll be a little bit of that, and then I think we predict that small cell infrastructure will be really important starting, you know, next year, is what our customers are telling us. And so that ability to densify and infill will be heavily fiber-fed mini macros, or they'll be just traditional de novo small cells. And depending on who the customer is, you know, if it's Verizon or AT&T, trying to use their fiber plan as much as possible, and if it's T-Mo, it's turnkey.
But I would sort of say, you know, from an aggression perspective, we definitely feel like the CapEx will swing from macros into small cells next year, and that cycle starts, and it goes 2025, probably through 2030.
Excellent. We are just out of time. I wanna thank you for the very comprehensive comments. We could have lasted all day, but.
Thank you, Jonathan. Appreciate it. Good to be here.
Thanks.
Thank you, all.