Let's get started. Good afternoon, everybody. For those that don't know me, my name's Greg Williams. I cover the cable, telco, wireless, and fiber space at TD Cowen. I'm joined here, this is one out of two keynote lunch sessions. This first session, I'm joined by Alex Gellman, the chairman and co-founder of Vertical Bridge, and Steve Smith, CEO of Zayo. So gentlemen, thank you for joining us. So, I know, Steve, you and I spoke at this keynote lunch last year, and Alex, we spoke on private roundtable panels. But in the last year, maybe you can tell us, with the key developments in your businesses over the past year and where you're spending most of your time.
Well, now that Ron Bizick is here as our CEO, hi, Ron, I'm spending a lot less time in the office.
Right.
No, I think what, look, what we're focused on is, our swim lane as the largest private tower company in the U.S., macro tower focused. I heard somebody earlier say, there's no growth in U.S. towers. That's false.
Right.
There's a lot of growth. But we're really focused on, you know, our business is pretty simple. You raise capital, you acquire assets, we finance the assets into the ABS market, and we lease them up.
Right.
So that, that's pretty much where we're focused and deepening our relationships with the biggest customers.
Got it. Steve?
Yeah, I'd say at... Whoosh, that's loud, sorry. I'd say at Zayo, we're 4.5 years into the journey, and since you and I spoke last, Greg, we've added a couple more executives to the leadership team. I would tell you that the foundational investments that we've put in here the last 3.5 years are starting to pay off. It's, we've turned a 50-acquisition inorganic story into a 50-acquisition organic growth story, and so it's, it's a lot more fun these days. Adding lots of capability to the network. Leadership team is outstanding. It took a little longer than it probably should have done doing a transformation.
For folks that are, have done transformations in the room, you know, once you get the capable leaders that can dream about creating a future, and you're programmatic about fixing things that need to get fixed, and you get it done, it becomes a little bit more joy when you get to the other side of it.
Right.
But it's been a long 4.5 years, and we're in a very good position. We've had several quarters now of top and bottom line growth, customers settling. It's just a very, very different place here than it was a year ago.
Right, and I want to talk about one of those top-line growth drivers right now. It's Gen AI. I mean, this time last year, Mike Elias got all the love for the data centers, but finally, on the fiber side, we're getting some of that, that love. So, you know, last week or two weeks ago, I guess, Lumen signed a, like, a $5 billion opportunities with hyperscalers, another $7 billion funnel, the huge numbers here. You know, we'll speak to Lumen later today, but maybe just talk about understanding your fiber opportunity for Gen AI. Is it really connecting to these training data centers or, you know, bulking up the inference data centers? What's happening at Zayo in terms of the fiber demand from Gen AI?
You want me to start, Alex? Well, I don't think anybody in this room would disagree that it's probably the most significant trend that we've seen in several years. I think the market has taken notice. I think the data required is creating just an unimaginable amount of data center capacity in the world. That normally happens first. The connectivity and network density requirements follow that. You can talk to Lumen about those numbers. Those are total contract value numbers over a long term.
Right.
This industry typically lives in a monthly recurring revenue conversation.
Yeah.
I think those kinds of numbers exist at other places, too, when you start talking TCV. But we're seeing the front edge of it, certainly from hyperscalers and from data centers for the connectivity requirements.
Mm-hmm.
But our story right now is, we're very fortunate. We're in a place where the organic growth machine is working very well.
Right.
AI we view as just an additional benefit to us. Now, as we start converting the funnel and we start winning some of this, and we have won some of it in the last couple of quarters, these are big deals. These are big deals, big numbers.
Yeah.
It's probably not a lot. I think we're one of the few private companies that does earnings calls every quarter, 'cause our debt is traded publicly. So I don't want to go too far down that path. We have an earnings call on Thursday-
Right
... but there'll be more to come this Thursday with our debt holders.
Right. And Alex, I know it's a little more ethereal for the towers, the Gen AI opportunity. You know, do we need, like, a killer mobile app for AI to really take off? Help us think about Gen AI and the opportunity for the towers.
Yeah, I mean, obviously, we're watching with great interest the explosion last year in data center demand now hitting fiber. But ultimately, my point of view, and I've been around a long time, is everything goes mobile when it, if it can. So, you know, it was voice, then it was email, then it was e-commerce, then it was video. And whatever's Gen AI comes up with, when it goes to inference, it has to get to the customer.
Yeah.
The way we all get reached is right here. This is our first and last screen. All these growth opportunities will spike mobile network demand.
Mm-hmm.
I look at it the same way Steve does. We have healthy growth in our business. Data consumption per consumer is supposed to double next 4 or 5 years, depending on what source you look at, and that's without AI. So as AI, Gen AI moves to inference, it'll move to mobile.
Right.
One thing out of that, Greg, I think one of the things we're seeing is, I think you asked, is it training related?
Yeah.
I think most of the people in the room that are winning this stuff, it is a lot of training deployments. It's going outside of the major metros. It's going to places where there's power, where there's industrial buildings that are getting repurposed.
Yeah.
You know, cogen behind, you know, is happening. There's a lot of things happening now 'cause they're going where the power is. I do think as it shifts to inference over the coming years, you'll see companies like my old company, Equinix, that will start to get more active, because once users start using this and customers start dealing with each other, then, then you're gonna see the inference take off, and you're gonna see more interconnection, you're gonna see more latency-dependent applications.
Right.
So I think it's a natural evolution, but for me, I think this is following the cloud rollout, you know, 15 years ago.
Okay.
The dynamics for me, I kind of laugh when people ask about what's the durability of this thing? I'm sure this morning, most of the people in this room that talked about their data center capacity, people are signing deals out to 2027 now.
Yeah.
So the hyperscalers can't really see much further than 2 years-3 years today, so it's hard for them to project even further than that. But I personally think, on everything we're reading and hearing about with our customers and in between all of us, this is gonna be a very durable trend.
Got it.
I just kind of chuckle to myself when they say, "How durable is AI going to be?" I think it's gonna change the way we work, and it's gonna change our lives in very big ways over the coming decade.
Right.
People always ask, "What inning are we in?" It's like, what's the longest baseball game ever played?
Right. There's some Mets games back in the day that went 20 innings. But anyway, so does that mean, I guess there's no hype cycle, I guess, so is one what you're kind of suggesting here? Or-
Well, you know, I think we're in the hype cycle.
Yeah.
But if you follow the capital being raised by all these firms in the room, you follow the CapEx projections that everybody's doing, this is not 24 years ago when, when the dot-com blew up, when, you know, people were doing speculative building. All of us have anchors at the back of all these decisions. So I think it's a much different time, and I think we need to be careful, and I think it'll be spiky, but I, I do think it's gonna be durable. If you see the amount of money going into CapEx, just the three hyperscalers, I think, is over $110 billion in 2024 for capital. And just look at the land that they're buying, and you look at the enterprises that are consuming AI today. I, it's an irreversible trend.
Now, I think we're all gonna have to be smart about how we price and how we go after it-
Yeah
... but I, I think it's here to stay.
Right. And is it evolving, then, from maybe a lot of dark fiber opportunity, let's say, and then it would go towards the lit fiber? Is that the right way of thinking about it? 'Cause right now it's-
No, it's both.
Okay.
No, and I have some of our leaders here today. Our funnel is filled with both dark fiber and waves for backhaul and for just connectivity between metros, and-
Mm-hmm
... our long-haul route development has almost doubled over the last six, seven, eight months. So there's a lot of activity that's on the front edge of all this stuff we're talking about.
Got it. And Alex, how are you using AI internally? And I'll posit that same question to-
Probably not very well. No. I mean, for our business, we don't have a lot of people. So I think the opportunity's a bit limited. The beauty of our business is we operate steel and concrete, so you don't need a lot of AI to maintain that.
Right.
I do think, you know, what we've done is we've created a data analytics team, which is looking at how do we deploy it, and it's really around intelligence related to siting decisions, right?
Okay.
Underwriting sites.
Right.
What, what's the demand? We now have an RF engineering group, and we really look much, much deeper into big data around what are the carriers' results. You can buy this data, and you can look at what their quality of service is and, and underwrite it that way. That's probably the biggest area of, of development for AI. And, you know, the kids are running around drafting emails and letters and stuff with AI. I'm just like: "Please proof it and put something personal in it. Don't just send it.
Yeah.
We're seeing it. We're a little different business model than Alex, but-
Completely.
We have major automation upgrade programs going on, and all the systems integrators we're working with are bringing tools and techniques in. We work with all three of the hyperscalers to run our IT systems in our company. They're bringing tools and techniques in. So we're trying to push as fast as we can to have proof of concepts and explore, and we're seeing some very unique opportunity. I think it's gonna affect our entire business.
Right.
I personally don't think there'll be an IT organization in the world that will be able to operate without AI comb through their business. There's no way humans can manage the thousands of events per second that are in IT systems and companies today with this technology that can predict and analyze and report at a rate... So I just think it's game-changing for internally use also.
Yeah, it's great that you're saying that internally because that's what we all need to see and hear, is that if they're using it internally, that's the use case for AI. For you, Alex, so I'm curious, 'cause I always like to hear your thoughts. A couple of years ago, you talked about the blizzard of IoT in the mobile world. Like, what would be the killer app that you think on the mobile side to help proliferate, you know, AI at some point? If it's just more video, which is what American Tower said earlier...
If I knew what that was, I wouldn't be here.
Right.
No, I... Look, I don't know. I know it'll go mobile, whatever it is. You know, I don't. You couldn't predict Uber blowing up the way it did, or Waze, or any of the apps that LTE-
Right
... enabled. I don't know. I'm not sure that anybody really knows. That's why GenAI is churning away, learning models around businesses to race to have the first app. But I know it'll end up here.
Yeah.
Right? But what I will say is, a stop along the way to mobile networks is gonna be the mobile carriers have to be able to monetize it.
Sure.
More traffic with unlimited plans doesn't work.
Right. Right.
What happened this year that I don't know was noted that much is the three big carriers in the U.S. raised their prices by about $3 billion.
Yeah.
Verizon was $1.7 billion of that. Churn was not impacted.
Right. Yeah.
Churn was not impacted.
That's interesting. Yeah.
So I would say keep an eye first for more on the mobile side, keep an eye on them getting out of their unlimited plans more and more and more. That will enable them to generate free cash flow to then invest in blowing up the network to support these services.
Right.
That, that's essential, or it will go slower.
Right. And it just goes to show the inelasticity of the wireless world and the criticality of our phones, of course. And maybe we can talk about that, the non-AI, moving on, demand. Of course, in the tower world, we've just had a sort of a lull of spending, last year, and, you know, now flash forward to this year, we really haven't seen much of a pickup. I'm just curious your thoughts on, you know, any expectations of a pickup, in the tower space with your carriers? You know, what's gonna really push it?
Yeah, for sure it will. I mean, as was discussed in the tower panel, the, it's, it's very cyclical. It's predictable. So the carriers all got new spectrum, and they have a new technology, and they all spend to roll it out, add on to that DISH. So, like, 2021... 2022, perfect storm, everybody's spending, good times.
Yeah.
But very typically after that, they each move into their own cycle. So AT&T focuses on fiber, Verizon stays focused on RAN, T-Mobile pauses to monetize their investment in Sprint and the integration. That's very normal. What's going to happen now is infill.
Right.
which means sites may be two miles apart in a highway, now they'll be a mile apart.
Right.
In the city, they might be half a mile apart, now they'll be a quarter mile apart. It's just densification. It's always been like that, and so we're moving into that phase. It'll be exacerbated by the fact that the 3.5 GHz spectrum rolled out by AT&T and Verizon is far higher than what their networks operate today.
Right, more towers.
... which means it doesn't go as far-
Yeah
... which means more sites. And then lastly, there's no visibility really to, or limited visibility to new low and mid-band spectrum.
Sure.
So the spectrum that's available is very high, and it doesn't go very far.
Right.
The mid-band that everybody's focused on is owned by this thing called the military. Anybody know it? They use it for radar.
Right.
They'll give that up, right?
That's not an important use case.
They don't need that.
Yeah.
Who needs that?
Yeah.
So, in the absence of additional spectrum, densification becomes even more important.
Right, sure.
I think that's the phase we're looking at. That is on top of doubling usage-
Right
... and that's before any AI incremental impact.
Right.
The runway for wireless CapEx spend, which is what drives tower companies, is long and steep.
Interesting. Yeah, the spectrum scarcity could really drive more cell splitting then if we don't see any-
For sure
... sub-6 GHz come to market anytime soon.
I think it will. I mean, I don't know-
Yeah
... where it comes from, so I think it will.
Right. Interesting. And Steve, on your non-AI demand, you alluded to it earlier, that you don't even need AI necessarily for the organic growth engine to happen. So maybe talk about some of the other demand, you know, avenues you're seeing. It's enterprise, it's...
Yeah
... cloud on-ramping still.
I wouldn't want anybody to walk away from here thinking we don't need the AI opportunity.
Okay, sure. Sorry about that.
Andrés, Andrés will kill me if I walk off stage and don't say it. So but the good thing about what he and the rest of our leadership team have done here is they've put a growth, an organic growth investment machine, foundational investments over... You know, listen, we've been going for 4.5 years. This thing is operating as a full organic growth company, and it's doing great now. But it's, you know, it's in our core. It's waves, it's fiber, it's in our connectivity services. We have a little bit of managed services from a couple of the acquisitions we did, but we're seeing demand across all that. Because enterprises are still digitizing almost everything around them-
Mm-hmm
... the demand for cloud-related applications is still there.
Still there.
AI is just another wave coming at us, so we'll watch all three. They'll, they'll continue to feed the beast.
Got it. And how about the pricing environment? You know, maybe we can talk waves first, and then we talk dark fiber. But in the waves business, or any of the lit services really, do you think GenAI is going to augment that? I'm trying to figure out if it, you know, increases the pricing just because of demand or, you know, there's going to be a lot more unit costs. What do you think?
That's a great question with your previous statement about one of our participants in the industry-
Right
... you know, what we're announcing this quarter. You know, listen, pricing is gonna... There's the hyperscalers are going so fast and are presenting a lot of NRR to all of us to help them go fast. I think we're all have to be disciplined about pricing, particularly on unique routes for us.
Right.
Data centers are going to be smart about, you know, the capacity they're going to give up here. But pricing's been pretty stable for us, and-
Okay
... there's an interesting mix here as the hype cycle's going through of NRR and MRR.
Mm-hmm.
I think we're going to have to watch that as we mature through these contracts, because, you know, all our businesses here love MRR. We want the monthly recurring revenue. The NRR is nice right now, where we're building the infrastructure to go support this.
Right.
We're going to have to balance that as we go through these cycles.
Right. Yeah, it's interesting. I mean, that's what we're seeing so far, I guess, is this high NRC cost and upfront. So it's nice for cash and liquidity, but then, you know, of course, it's economics, that means the MRR is probably going to be a little bit more much to be desired in the outer years, I guess. But Alex, how about the pricing environment in your world? You know, rent on the towers, you know, going from the LTE generation to 5G and then going forward, is it flat? Is it up?
Yeah, I mean, the U.S. market is very, very stable. You know, there are three public companies that are very disciplined operators. They don't do foolish things. And, you know, three giant customers basically dominate the market. So it's, it's super stable, overall, and the assets are protected, right? So it's, it... One, it's very expensive to move once you're present. So our churn with our big three customers is well under 50 basis points a year. They do not churn. They do not leave, right? They don't want to leave. It's expensive to move, but also, a majority of our sites are zoning protected, so you can't really move. But, it's a very long-term, stable business-
Right
... as a result.
Right. I want to switch gears a little bit about small cell business.
Yeah.
I feel like every year, it's like a year away where it picks up again.
Here it comes.
But this time might be a little different. You gave some good, compelling reasons of the scarcity of spectrum.
Yeah.
So you're going to need to sell split some more and get out further into the cities, and at the same time, maybe there's a GenAI use case someday that that's going to need more, you know, micro edge, low latency applications. So I'm curious to think, you know, what are you thinking about small cell business now?
Yeah, look, I think if you believe as I do, that AI will drive... Really, 5G hasn't been tapped either, but the potential 5G plus AI-
Right, which sets fixed wireless, really.
... will drive, will drive incremental mobile traffic, which I believe it will. And there's a dearth of spectrum and new spectrum. I think densification comes, moves further and further and further to the fore, and I do think that at some point, the carriers will shift more of their capital budget to small, what, what you small cells. to me, it's just lower to the ground. I mean, the industry, since I've gotten in it.
Mm-hmm
The first site I ever went to was the Seafirst Building in Seattle with McCaw Cellular, and there's no antennas on that building today. It's too high, but that was covering the whole downtown area. You couldn't do that now, too much interference. So everything's always come lower and lower and lower, but I think there will be a shift of budget to outdoor densification, low to the ground in urban areas, but I don't think it's real soon. I think it's a few years away.
Okay. And Steve, there's unsubstantiated media reports of small cell assets under strategic review, but nevertheless, I mean, just tell us how you small cells within your larger portfolio?
Yeah, I can't substantiate the unsubstantiated claims.
I can neither confirm nor deny.
But, I mean, what I've learned since I've been here is the previous regime accumulated a fair amount small cell business, just being in the industry. Pretty opportunistic in metros. You know, one of the things I've learned, and Alex knows way more about this than I do, but the macros were easy to deploy and cost-efficient. That put pressure small cells over the last few years.
Yes.
The carriers have been doing a bunch of their own deployments small cells. and then 5G, from what you just said, it has not accommodated small cell growth. Andrés and I have been in a couple of meetings where people are looking way ahead are thinking that as 6G starts to unfold in years to come, that level of spectrum will small cells to go faster. Who knows?
I don't think it's as much spectrum-driven as it is densification-
Densification
... driven. Now, if the new spectrum is higher, small cells will be more likely deployed 'cause they don't go very far if you put them too high in the air.
That's, that's what I've heard, yeah.
Yeah, they won't, they won't go anywhere.
But listen, we're always looking for value creative... I mean, we explore stuff all the time. I can't comment specifically on that-
Right
... but we've accumulated a little bit of that business, and it's not core, so we're gonna treat it that way.
I think small cell networks that were built were predominantly built where you can't build macros.
Right.
Right? In a lot of places, it's zoning-driven.
Okay.
So you can't build macros, so those networks are sustainable-
Right
... because it's not an option, but that market's more or less saturated. The real growth with, small cell would be where it's cheaper per megabit delivered.
Right.
And it hasn't been so far.
Got it. Got it. And then just moving on maybe to the capital structure of your companies, you know, there's a lot for the markets to process over the last few weeks. I mean, it was, like, good inflation numbers, not so good job numbers. Some are saying some recession, 35%, some are saying 0% now. There's a lot right now going on, and I'm just curious, you know, how... Talk about your capital structures and your leverage comfort levels, you know, in these elevated rate markets.
Yeah, what I'd say about that is we have two pretty sophisticated owners that spend a lot of time helping us think through that stuff, and we have a world-class CFO now that's made a big difference. We're super comfortable with our leverage and our capital structure and our balance sheet now. Near term, we're focused on de-leveraging and liquidity. Midterm, I think you'll see Zayo start to show up in the ABS market. Longer term, I think will probably be a combination of ABS and traditional debt and executing.
Okay. ABS on enterprise fiber then, in your securities?
Well, as everyone else has explored it, we're in the exploration phase now.
Yeah.
I think you'll see some activity as we look towards the back half of this year.
Got it. And Alex, how do you think about your cap structure in this environment?
Yeah, one thing we don't do is predict interest rates. I don't know if anybody familiar with the Hairy chart? You know what that is?
No.
Google it. It shows the actual interest rates versus what people forecast, the experts forecasted. Wow, they're wrong. Super wrong. We'd all be fired if we were that wrong. So our business is pretty straightforward. We use senior debt to buy and build towers.
Okay.
Then we flip it into an ABS market. We're very comfortable with the leverage available in that market. We don't want to go any higher. We could, there's opportunities to drive leverage higher than that. We don't want to do that. So we're very comfortable in that market, fixing our interest rates, five-to-seven-year timeframe, and moving on. The rates we, we issued in, February of 2022, like a week before the tanks rolled in Ukraine, and we issued this May, both very well received.
Yeah.
Were the rates as good as they were 3 years ago?
Yeah.
Of course not. Were they better than anything else out there in the marketplace? 100% they were.
Nice.
So we're just sort of like you see Verizon continually issue their phone ABS. I mean, it's a program. You can't try to pick rates. We don't time it, 'cause we're not going to get it right, 'cause the experts don't get it right.
Sure.
The good news is, you know, we've weathered the storm. Our rates have gone up. It hasn't been a problem. Our coverage ratios are good. We're continuing to grow and accelerate. And really, our customers have weathered the storm as well. When you look at AT&T, Verizon, T-Mobile, they've kind of figured it out, and they have to continue to compete with each other, so-
Right
... it's been probably better than most people thought.
Right. And their price hikes don't hurt either, with the helping out with the interest rates.
I think it's critical.
Yeah. And maybe we could shift to the M&A environment. In the tower space, you don't hear much M&A in U.S. towers for a while, but now, of course, Verizon has some towers for sale. UScellular has been broken up with T-Mobile, and so they have some towers possibly for sale. Just, I'm just curious, you know, what's the environment like for you? What are the multiples looking like in, you know, the U.S. M&A market?
Yeah, so I mean, private multiples have, and Ron said this at the earlier panel, always sustained a spread above the public multiples. Some of it is the age of the assets tend to be younger, but in the private market, the multiples have been higher. In the last since interest rates took off, the multiples have gapped. So if you look at the public companies, they've been cut in half.
Right.
They were trading over 30, they've been at 15, they've rebounded a little bit. The private multiples did not get cut in half. Now, interestingly, there's less bidders. There's no more deals where there's 30 bidders signing NDAs-
Right
... but there's enough bidders that keep the price high. A lot of assets have been on the sideline trying to weather the storm. If you think about it, over the last year and a half, everything experts have told us is, "Rates are coming down, rates are coming down, rates are coming down." They haven't come down.
Sure.
But people are holding on for that day because it definitely affects the pricing. The public's popped up something like 10%. Immediately, when there was a whisper of, "We're not gonna wait for inflation to drop to 2% to cut," look what happened to all the tower stocks.
Yeah, right.
So, you know, the market's still high. We're very active in it. We bid in it, we don't have to win it, and we're selective, and we lose a lot in anything competitive. So-
I mean, there's an issue that some are holding out for lower rates, and it's not happening, so now those balances are getting distressed, and at some point they, they roll?
Yeah, I mean, the nature of virtually all private tower companies is they don't get to scale, and they burn cash to keep the lights on.
Yeah.
Right? So at some point, that becomes a diminishing returns game. So do you hold out or do you sell? But we haven't seen what I would consider—like, in past cycles, the capitulation that we've seen in the past.
Okay.
You know, maybe... Supply is limited.
Right. Okay, and Steve, the M&A market in the fiber world, you know, I spoke to Kenny Gunderman at Uniti in saying that the reason they did one of those Windstream deals, there's gonna be a lot of fiber M&A, I think, in the next, you know, 12, 18 months. So curious to hear your thoughts on the M&A environment for Zayo. I mean, you're already so large and, but, you know, you can always use more-
Yeah
... anything on your wish list, and maybe talk to multiples there as well.
Yeah. I haven't actually caught up with Kenny, so I missed him so far, but I'm sure he's thrilled to-
Yeah
... not have to talk about the lease between Windstream and Uniti anymore, so I'm sure he's a happy camper.
I'm just thrilled, not covering it.
Listen, there's always gonna be activity and, you know, having running a company that's made up of 50 acquisitions, I don't... You know, we don't see any obvious scale issues where we are today or any obvious product portfolio mix issues. So we're not actively looking to fill any gaps in the company. But listen, we're always active in the market. We're watching things. We pretty much see everything that moves. And but since we've moved into an organic growth story right now, that's the core of the business now. 4.5 years into it, our owners are typically 6 year- 7 year holders.
You know, I think our focus now is to continue to grow this thing really fast, build more competence there, and, you know, if there's opportunities that make sense, that are accretive to the company and fit the strategy, of course-
Okay
... you'll see Zayo at the table.
Got it. And, my last question is just on DigitalBridge, the mothership. I mean, how do you guys, how has that relationship evolved, both with your companies, and is there cross-selling opportunities that you guys-
Yeah
... both work together on?
Yeah, I, look, I think, DigitalBridge was founded on the concept of convergence. That was the hot topic, you know, 10 years ago. You don't hear that word very often now. But I think it's in our, it's in our DNA to try to work together. So, for example, we have, you know, two executives in our respective companies that talk all the time. We're always looking for opportunities to work together. I think, one that has worked is middle mile, where we're working together to develop middle mile solutions for customers pursuant to government incentives. And then I believe, and you're gonna may think Steve might reach over and smack me, you know, we've talked for years with all the fiber companies about fiber to the tower and can we vertically integrate it? The customer wants it.
You know, you see what T-Mo's doing. You know, can we deliver a solution there? I think it's gonna become more in the frame because the volume of sites we see coming, new development sites in this next wave, and the opportunity to coordinate with a partner like Zayo to say, "Okay, which one of these work for you on -net?" And offer it together. I think will become real.
Yeah, there's probably a handful of companies in both portfolios that we interact with quite frequently. Actually, Alex, in our business, we bid on three middle mile programs under the NTIA banner, won almost $100 million of business-
Okay
... across 3 routes in 8 states. We did that together. We're also, I've worked on the LEO satellite deployments, so we're doing ground stations with Alex's team.
Okay.
So our two teams have done a lot of work together, and it's been, like you said, we have two executives that work for us that do a great job doing that. So we will continue to do that.
Okay.
The learning that you have when you have, for us, I mean, you have two owners that see as much as they see, there's plenty of opportunities to leverage and coordinate and try to win together.
Great. Well, with that, we're about out of time, so thank you, gentlemen, very much.
Thank you.
Stay tuned. Mike Elias is up with a keynote lunch with Marc Ganzi.
I'll help you up. Oh, oh, man. Wow! Great job, brother. Great job.
All right, guys, so I'm gonna jump right into it because whenever we have Marc, there's a ton of questions and a ton of interest, and hot takes always, hot takes always abound. So thank you, Marc, thank you very much for being here. Really appreciate it.
Yeah, thanks, Michael. Good to be here.
So, you know, one of the themes at this conference is, look, we have AI demand absolutely booming. We're seeing that in data centers. We're seeing it in fiber. We need a lot of capital in order to fund all this. You have a unique purview as an asset manager into all, into what's happening in the digital infrastructure ecosystem. Can you tell me, how do you think about raising the capital formation to support all this growth, to deliver AI to society? Like, how do you get the capital, essentially, to, to deliver all this?
You make it sound so profound and like all the weight's-
I mean-
... on my shoulders. I have to deliver AI to all mankind.
You know?
It's a big burden, right?
You know.
But somebody's got to carry it. No, look, I think, you know, let's start off with some numbers, right? I think it's easy to get our minds wrapped around the CapEx, and I think we said it in our Q2 call, we were on a cadence of about $150 billion of CapEx on AI spend last year. We're on a cadence this year to go to $200 billion of CapEx. We see that, well, the market sees it in 2 years going to $250 billion. So I like the acceleration of the CapEx, but I also like the conviction that our customers have, and the conversations we're having with our customers is, they're not pulling back. You know, the hyperbole of AI is so funny.
Like, I go out and I fundraise, and everyone's like: "Well, tell us about the hype of AI, and the AI trade is on, and then the AI trade is off." Look, there's no hysteria to this. Let's all take a step back and understand that, you know, we spent over $1.3 trillion in CapEx getting cloud to where it is today. Right? Frame that. That was 2011 to today, and if we were playing a 9-inning baseball game, we'd all say that cloud infrastructure's probably somewhere in the middle of the 7th inning.
Because we're now proliferating edge workloads in public cloud and secondary and tertiary markets, and then the next place we go is near edge, which is really, co-located adjacent to the node, adjacent to the tower, as Alex said. Needs a lot of fiber, as Steve said, and so cloud is still... We're still investing in cloud. And we're 12 years, 13 years into that journey. We're 2 years into the AI journey, and framing the AI opportunity, we're probably somewhere in the second inning. And so this is a long progression, and I know people love to attach a label to things 'cause that's convenient, particularly, you know, for, for public traders and people who like to short and do other things, that. I get that. But the AI trade is neither on or off.
It's a once-in-a-lifetime opportunity to build infrastructure like I've never seen in my 31 years of doing this, that you've never seen, that Alex has never seen, that Steve's never seen. The enormity of the CapEx is accelerating, and so there's a big opportunity, and we have to stay close to our customers. We have to make sure that they have a fair proposition where they can make money. It is a fair proposition for Meta and Apple and Google and, you know, and Amazon and Microsoft. They've got to make money, and some of them are already making money.
It's just not perhaps at the velocity that the public markets want them, because everybody wants this AI trade to stay on forever. And I just never bought the AI trade 2 years ago. So we take a pretty stoic stance around this. Again, having built macro sites with Alex in 1994, when we were going from analog to digital, and having built, you know, public cloud with Sureel and Raul, I've gotten to see all the CapEx cycles.
You never get too excited, you never get too down, and you just try to stay level, and you keep trying to produce the results that are right for the customers. That's really been our core focus. It really hasn't changed in three decades.
So with the rerating of CapEx, 'cause to your point, we're still building cloud. Now, AI is additive on top of that. Like, CapEx broad, industry-wide, has gone up. More capital needed to fund it. What I find is interesting is that at the same time, given the demand and supply balance, particularly for data centers, you've seen the economics of data centers actually improve. So with that, what I would say to you, what I would ask you, is that when you go out and talk to LPs who are looking to put capital to work, you know, how has their view on the data center sector, the fiber sector, towers, evolved over the last, call it, year, right? And how has their willingness to put capital behind this sector evolved?
Well, I think it's evolved, particularly in the last 5 years. I think when we went out fundraising 5 years ago, we didn't bump into the hallway generalists that claimed to be infrastructure digital infrastructure experts. And so now everyone's an expert. It's awesome. And so we take that with a grain of salt, because, again, we're not selling hype. I'm selling 30 years of returns. That's a different proposition than perhaps another GP who's looking to pitch one of our LPs to steal our allocation. So we're pitching sobriety, discipline, and that this is a multi-generation opportunity. I think if you go in and you pitch an LP that this is a huge hype opportunity, you're gonna, you're not gonna get the next meeting. I think investors, particularly sovereign wealth funds and pension funds, they're extremely well advised today.
They have big staffs that understand digital, and they've already taken their first few swings with folks like us and a few others. So now I think the mantra has shifted in the private capital world to: "When are you going to return capital to me?
Mm.
That's the first thing we hear when we sit down with a real adult institutional limited partner is: "When are you returning DPI to me? Because I've already got $2 billion out with Blackstone, I got $4 billion out with KKR, and I've got $1 billion out with Brookfield, and I've got, you know, X amount of money out with you." And we say: "Look, that's fair." So we said it in our second quarter earnings call. We said it in the first quarter earnings call. You know, we've had multiple exits in the last 24 months.
We've returned billions of dollars of DPI back to our LPs in the last 24 months, which facilitates the discussion about, okay, now we can help you invest in AI infrastructure, because we returned money. The money you gave us in 2014, 2015, 2016, and 2017, we've now returned. That opens the door for future allocations. The key to this is investor appetite is there, but they do have acid indigestion, because everyone's coming to them with an AI data center idea.
And to be honest, none of them are terribly differentiated. And I'm sorry if I'm being... You asked me to come be controversial and be direct.
I love the honesty.
I'm being very direct, because-
Bring it on
... LPs talk to us, and they tell us. They say to us, "Oh, this, this guy came in pitching me data center today." They all call me and they say, "What do you think?" I say, "Well, that's a good operator. That's a bad operator. That's a really good GP. That's a so-so GP." And, and the reality is, there's 40 pitch books sitting in front of an LP right now for digital infrastructure.... The market has gone from a handful of players. If you came to this conference 6 years ago, there were less than 5 guys that did digital, digital infrastructure, and now today there's probably 30 or 40 GPs here pitching digital infrastructure.
And so the market's flooded, and by the way, it doesn't mean that we have good ideas or bad ideas, and I'm sure every GP here has a great idea. It's just the market is completely inundated with requests for digital infrastructure. So how do you stand out? How do you go get capital? And I think today, what you heard in our calls, we're being a lot more creative.
Yeah.
We've channeled our conversations in credit, private wealth, our flagship products, our continuation vehicles, our co-investments. We raised $14 billion of debt in the quarter, so we've done a good job of forming capital, but it's very competitive, it's very difficult, and you got to have great platforms. You got to have great companies that know how to execute for the customers. The best way to differentiate yourself is to go out and perform for a customer. That gives you the best opportunity to have a conversation with an investor to go get new equity, today, and return capital. If you're returning capital, that's really helpful.
Yeah. The precondition to raising incremental capital is to return capital that you have already raised. You know, beyond that... We'll transition, and we'll talk about the data center side of things, but, you know, you've talked about private wealth. $1 trillion pool is, I think is, as you've described it, you know, real estate capital, $2 trillion. So how do you think-
Three.
$3 trillion, okay.
Three trillion.
As we go and fund, try and fund this opportunity, you know, how important is it that we tap new pockets of capital that we historically, you know, haven't addressed as we look to build out this infrastructure?
Well, I think the asset class is maturing, and as the asset class matures, you have to pair the opportunity against the capital. And so, for example, in the real estate world, they look for a very different type of return. They're looking for yield, they're looking for downside protection, and most of that money is coming from insurance company balance sheets that has a minimum hurdle, and so the arb for that capital is about 100-250 basis points, and can you figure out how do you play in that swim lane and go in that direction? So we've been working on that for a while. When we did our merger in 2019, we were actually a REIT, so we have a pretty good understanding of that marketplace. Private wealth is great.
You know, it's a copycat league, so we copied quickly, and we became a fast follower. Very encouraged, as I said on the call, about how fast we were able to pull that amount of capital together in 30 days. It really, it really blew me away. But I think it's also an awakening of sorts because, you know, private wealth investors that have, you know, a net worth between $10 million and $50 million can write $500,000-$2 million tickets in AI data centers. And so when we launched that sidecar vehicle and had three funded products in there that were performing really well and a sleeve to go do discretionary, that's pretty cool. There's not many funds you can get into where 75% of the assets are funded with best-in-class companies like we have.
So that was a good product at the right time. I'd love to do something like that in fiber. I'd love to do something in cell towers. Big thumbs up from Alex. He likes that. But that channel is big, and I think, you know, we look at what Blackstone has done, we look at what Starwood has done, and they've done a really good job in their broker-dealer networks, creating those swim lanes, and there's more capital there to be found. I think credit remains interesting. I think that's a vertical, and again, when you're sitting with a big allocator, like a sovereign wealth fund, you know, there's all these different slivers that you can go see.
Mm-hmm.
And so when you're dealing with a, you know, a $200 billion sovereign wealth fund, there's all these channels that we're now—we've figured out how to multi-product sell across those channels. We weren't doing that a year ago. And so having a, a, a 38-person marketing team, which I never had before, in Asia and in Europe, in the Gulf, in North America, pitching credit, pitching private wealth, pitching our flagship products, picking, pitching our, our continuation funds or, or co-investments, we now have the right soldiers on the field to go execute that, and it's something—it's, it's honestly, it's, it's something we didn't have a year ago. So it gives us a real leg up, and it gives us a chance to go compete, because now, at, at where we are, we're being measured against Blackstone.
We're being measured against KKR, Brookfield, in terms of fundraising. And so we moved up in weight class, so we had to have our capital formation team move up, you know, in a commensurate way.
That makes sense. So that's the, that's the capital formation side. You raise that capital against an opportunity to deploy it, and as we think of the opportunity to deploy it, particularly within data centers, you know, we heard from Alex, we heard from Steve. When you were on the stage last year, you were saying that the generative AI opportunity was a 38 GW opportunity. Now that we're a year after that, you know, how has your view on the size of that opportunity evolved?
I whiffed. Complete whiff. Look, the opportunity is probably somewhere between 50 GW and 80 GW. I mean, Elon puts a number out there like 80 GW, and Sam puts out a number like 50 GW, and then they flip-flop, and I don't know if they like, they like, call each other and be like, "Okay, I'm putting this up on X, bro. Are you ready?
Yeah.
Yeah, okay." But somewhere in between is probably the truth because we are in a leasing cadence right now, where the market is leasing, not absorbing. We'll get to absorption in a second. The market is leasing about 1.5 GW per quarter right now. So we're kind of on this, you know, possible 6 GW trajectory for this year. We'll see where the back half comes in. That takes the sector... Last year, you know, we finished the year somewhere around 14 GW, 15 GW. This year, we'll probably finish at about 20 GW, but there is a pretty clear roadmap on how you get to 80 GW. Now, now comes the: how do you activate it? How do you get there? That's, to me, leasing's not the problem. The desire and the appetite of our customers is there....
The question is, you know, how do we as an industry responsibly build that infrastructure? To me, that's actually the bigger challenge.
Yeah. You know, we'll get to the obstacles to building out the data center capacity. But as you think of that, let's call it 50 GW. What do we want? We wanna take the midpoint and call it, you know, let's call it 60 GW-
65 GW.
Why, why not?
Right.
All right, so 65 GW-
We'll better be here next year.
That sounds good.
All right.
So 65 GW. How do you think of the long-term split between inference and training as part, as part of that?
Hard to say, 'cause right now we're pretty much 90% inference and 10%, you know, generative AI workloads. We're still training AI. We're not there yet. I mean, there's a lot of, like, people have inflection points where they say they reach inference, and those generative AI applications are really happening in a very controlled environment. That's not in the sort of public cloud way, but in a private cloud. We're doing that at Articul8 with Intel. That's one of the platforms we have, where we're building generative AI models and systems for small, medium-sized businesses. But when we build those, those models for those companies, it's really important, Michael, that that data's coming from inside.
Yeah.
We're not going out to public cloud to grab that data, where the model can really spin up and move faster. So it depends on your datasets, right? You gotta keep your swim lanes clean in datasets. And so when people talk about generative AI today, they're using private data. They're not using true sort of public cloud domain data, or there's a mixture of a controlled set of public, public cloud data with a mixture of your private siloed cloud data, which is a little dangerous, by the way. So we're still 90/10, I think. I think probably next year we go 80/20 or 70/30 as we start moving into inference, but it's still very early innings, and I think the word "generative AI" is wildly misused right now.
I think people use it and use it pretty carelessly, but to truly get to where we think this goes, we're still probably 2 years, 3 years away.
Yeah, it's still, it's still early innings here. You know, from the conversations that I've had, you know, it appears that as you talk to the hyperscalers, there's still a preference for the amongst them to have the capacity for training be in and around the major markets, just fungibility of capacity. You know, are you seeing this dynamic change as power becomes increasingly constrained in some of these major metros?
Well, I think it's a latency issue, and I think the acceptable sort of distance is somewhere between, you know, 2 ms and 0.2 ms. More leaning towards that 0.2 ms is an acceptable, you know, distance in terms of speed. And so that, that really, that, that extra, you know, 0.1 ms- 0.2 ms really starts precluding a lot of these areas where you do have a lot of power.
Yeah.
So people are like: "Oh, let's go to Wyoming, let's go to North Dakota, let's go here." And I'm like: "Great, the customer doesn't want to go there, because if they wanted to go there, they'd already be there." And North Dakota is not a great example 'cause there have been some big data centers have been built in North Dakota because there's very cheap and a lot of power where you could build those models in a private cloud environment and then ship them out. So I think interesting markets that are gonna really play out here in the U.S. is, you know, Reno is a big battleground right now because it's inside 0.2 of a millisecond to Santa Clara. Atlanta's become a really interesting market because it's from a latency perspective, very close to Northern Virginia.
Certainly Columbus or Dublin, Ohio, anywhere around there, that's become an interesting market. And so you see these pockets of AI site development where today there's power. We don't know how much more power exists in Atlanta. We don't know how much more power AEP will allow us to do in Dublin, Ohio. And certainly, Nevada Power and Light is reaching a threshold issue because you've got Switch there, you've got developers coming in. There's two or three developers that already have will-serve letters, which is the most important thing. And then you've got Google with a mega campus next to our Switch facility that's building a huge amount of capacity, and then you've got Tesla there as well. So there's... The question is, when we sort of lease through this next 18 months, where do we go?
Where does the absorption go from there? And what are the next markets? And I think there is this balance as you move to inference between what's an acceptable amount of latency and what's an acceptable amount of, power loss you're gonna take. And this gets us sort of the next leg of the conversation.
Yes, yes. Before we go to that next leg, I wanted to ask you, I wanna ask you this: So I think of Northern Virginia. Northern Virginia is not only the largest data center market in the world, it's like the deepest network in the world. As we go to these new locations, let's call it Tier II markets, we're going to need to bring network with us. We've sat on this stage for years, and we've talked about convergence. How do you think about the opportunity to deliver a bundled solution to the customer, where you're bringing the data center, you're bringing the power, coupled with the network together, to essentially serve the needs of the customer in one shot?
Well, I think that is nirvana, right? That's kind of the sort of next phase of data center development is: Can you really do turnkey? And can you deliver without intermittence a high degree of power density with renewable power, with your own microgrid, with your own data center, with your own backup batteries, and most importantly, with your own interconnection? I did some canvassing of some of the panels the last two days and just listening and being an active listener. Let's be very clear: We do not have a generation issue in this country. We're generating plenty of power. Canada is generating even more power. Europe has a little bit of a generation issue, but the biggest problem right now is we have a transmission issue.
Yeah.
And remember, transmission infrastructure is completely promulgated by the PUC. And how many PUCs do we have in this country? Many. And so the challenge for us as an industry is: how do we get to grid independence? Can you get there? And I think that's really the conversation piece we're having with our big customers, is we've now picked two locations where we're going to try to do this concept. Won't name where they are, but we have enough solar in both locations, we have enough land, we can build the data center, we can deliver between 200 MW-400 MW of capacity and do it with renewable power adjacent, with our own microgrid, with our own batteries, and our own interconnection. And honestly, the key ingredient in that formula is interconnection.
By the way, just like it is in fiber, interconnection is super important in power, because your ability to interconnect and in peak hours, sell... You know, in peak, you want to be an offloader. And when you're, you know, off-peak, you want to be an intaker. And so as good luck would have it, when your batteries run out at 3:00 A.M., you're buying power when the grid is at its, its best point in time, and then when the sun comes up between 6:00 and 8:00 A.M., you're transitioning back on to your, to your generation capabilities. So there's a big opportunity to create these environments, create your own transmission infrastructure where you're in control. But at the end of the day, you still have to be connected to the grid.
Not only connected to one power provider, I would offer to you, you probably need to be connected to two, three, or four power providers, because then you can trade the power and you create an active market for your offload, and then you're creating an active buy side for your, for your intake. We've spent 2 years on this. We spent a lot of time on it. We've got partnerships with all the major utilities in the U.S., powering our existing 190 data centers. And so we've had a really unique chance to get deep with the utility companies and create what we hope are win-win solutions. But it's now getting complicated because the power guys think, you know, they've got all the say now.
So there's a new player in the space, which is the regulated utilities, and they think they're going to make a lot of money off of all of us in the room. So we're at this really unique stare down inflection point with the public utilities.
Well, you mentioned, you know, on the renewable front, you know, you talked about, one way that that could be used. Data centers require consistent baseload, right? And renewables are obviously have, you know, intermittency challenges. Considering the renewables, the renewables that, you know, you have within your, your portfolio, how do I square away the baseload data center needs, and kind of this expansion into, renewables that not only DigitalBridge, but then also the hyperscalers are kind of- are, are pushing for?
Well, I think the one thing we've learned in 2 years in working particularly, we have two platforms that are 100% renewable, Scala and Switch. And so we've learned a lot about microgrids, we've learned a lot about interconnection and intermittency. The reality is, even both of those companies that are 100% technically renewable-
Yes
... they're interconnected to the grid, and we're buying power every day from the grid, but we're putting power back into the grid, and we're creating that equilibrium so that we can maintain our net zero status. And so really, at the end of the day, as much as we're data center operators, you become an energy trader, whether you like it or not. And if you're truly going to sort of walk the walk on net zero, you really got to walk through that door and understand what you're getting into. Because remember, critical load, you just said critical load is really important. So if you've got 800 MW of solar, that does not equate to 800 MW of critical load.
That can be as low as 1.20, and it can be as high as, like, 2.70-2.80, depending on the conversion and how strong your solar is, because there is strength in solar. Again, this is a road that's really new. Everyone's talking about it, few are executing it, and it's going to be hard. We are executing it, but it's not without our bumps and bruises. And again, we've been at it for a couple of years. We didn't just start six months ago or three months ago because we thought it was kind of the cool thing to do. We did it because we knew it was table stakes, and it's survival for us as a firm.
Yeah. So, you know, on the first quarter earnings call, I think you had made note of something like there's less than 7 GW of grid power left, and in Europe, I think you said 2.8 GW-3.0 GW of grid power left. You know, you mentioned that there's a transmission issue, right? Or, it's primarily a transmission issue. As you think of circumventing the transmission issue, does it make sense increasingly for the data center operator or the sponsor behind the data center operator to become involved in power generation and co-locate the data center next to the source of generation, thereby circumventing the transmission issues?
Yeah. Well, look, as I said, I want to be really clear: you don't circumvent the transmission issues. You have to be interconnected, and you have to be actively a participant in the grid.
Right.
That's really critical.
Right.
It's not just a one-way street where take, take, take, take, take. You've got to put back in. We learned that the hard way with Switch. So there's a balance there, and at the same time, because of what happened with Amazon and Talen, we've woken up the DOE, which, make no mistake, that's a huge problem. That is a massive problem. If you're thinking about nukes or you're thinking about small generator nukes, you're like 8 years-10 years away.
And I know some of the public utility companies, which I will not mention, have put on their earnings calls that they're going to go build data centers next to nukes. Well, I can tell you that is not what the DOE is interested in, because if we take 800 megawatts out of the grid, you imbalance the grid, and then who suffers? The consumer. This current sitting chairman-chairwoman of the DOE, she has no interest in punishing consumers for the benefit of the hyperscalers. She's been very clear about that.
Yeah.
She couldn't be clearer. So Talen got it done through Susquehanna, unique location, had a lot of excess power, and a lot of people think, well, there's a replicable model there, and I will tell you that I don't agree with that until the sitting DOE says, "We think that nukes are great, and we don't think it's going to cause a disruption to the grid." So we haven't put a ton of our eggs in the nuke basket. You know, where we've spent our time is solar, wind, hydro, and LNG. Those are kind of the four things that we really like, and to be honest, we've spent more time on microgrids. So there's a lot of power out there. Again, I'll come back to where we started. There's a ton of generation capabilities.
There's very limited transmission capabilities because we built that transmission infrastructure in the 1950s and 1960s, which never contemplated, FDR never contemplated that we were going to be building AI data centers where you need 2 GW of power inside of 50 acres. That was never contemplated. So we have an antiquated grid. It's got to be modernized, it's got to be updated, and this will be the constraining factor for data centers over the next decade.
How do you see the data centers having a role in upgrading transmission infrastructure? I mean, we were talking about it on the hyperscale panel earlier. It's like, you know, the utility asks the data center operator, "Hey, like, we can get you this, but you need to make investments in this for the surrounding area," right? Like, where... Is that something that we can do as an industry-
Well-
If we want to make sure that we have the power?
That's I had the privilege of sitting with our chairwoman, the chairwoman who runs the DOE, and she was super clear about this. She's like: Look, I have no problem with the hyperscalers building their businesses and building AI, and the President's made it clear that AI is, we need to be, you know, on the front foot on AI. But she's like: The hyperscalers will pay for it. They make enough money. The consumer should not pay for this. So there's a plan at DOE to put in, you know, 10 to 12, 14 nuke sites to power AI, but that seems a bit misplaced because my experience in working with customers for 30 years is they're going to tell you the AZ.
They're going to tell you where they want to go, and no matter how big of a nuke you build, you can't tell Amazon, "Okay, you're going there. Microsoft, you're going to go there. Meta, you're going to go there." They'll look at you and go, "No. No, thank you. I'll pass. I'll find the private market solution." And so all of this is kind of converging, Michael, at the same time. It, it could work out. It could be a magnificent train wreck sometime late next year. But we are headed for a crisis-like event in data center construction in the respects of will the will-serve letters match up against the amount of capacity that we're leasing? And that's, that's the conundrum that sits in this room right now.
So, building large-scale nukes, okay, could see that being part of the solution, but SMRs seem to provide more, what's the right way to say it? Flexibility in where you can, and you can place load. Is that something that you'd be willing to bet on long term as part of the solution?
If you think zoning boards-
Are going to be comfortable with nukes?
like NIMBY, they don't want a tower in their backyard.
Yeah.
I'm pretty sure they don't want a small nuclear reactor in their backyard. I think SMRs are probably commercially, 8 years away in data center speak. And I do think it's incredibly viable. I think it's one of the more faster evolving technologies that we like. But there's other things we can be doing. There's CO2 recapture, there's, you know, how do we clean water and retain water in the West, where that's going to be a big issue. We're wasting a lot of water. So there's a lot of big threshold issues ahead for data centers. It feels really good right now. We're all in the feel-good version of SportsCenter, but that can change really fast. And it will change.
I can sit here today with a high degree of conviction and tell you, you and I are going to be sitting here next year talking about these issues, and we're going to have four or five case studies where data centers have been turned down.
Yeah.
Because why? No one likes the water profile. Nobody likes the CO2 profile. Nobody likes the power profile. Nobody has the right renewable solution. Nobody could get the interconnection agreement. There are five or six major threshold issues facing the industry, and if you weren't thinking about them 2 years ago, you're late.
Yeah. That's right. All right, let's shift a bit to investment focus. Given DigitalBridge as the asset manager, you know-
Get away from the world is ending?
Yeah, like-
Okay.
... yeah, maybe just... Yes, we have power problems. How are you prioritizing investment across each of the subsegments of digital infrastructure between small cells, enterprise data centers, hyperscale, fiber? How are you thinking about the relative balance of where you'd like to allocate capital? And as, maybe as part of that, how has that evolved over the last year?
Well, I think, I think naturally, just through gravitational pull, greenfield CapEx today for us is probably 80% data center, probably 10%-15% fiber and 5% towers. And just towers is not out of favor at all. It's just the sheer quantum of investing in AI infrastructure just dwarfs what we're doing in terms of build-to-suit towers, particularly in LatAm, the U.S., and Asia, and Europe. But I'm not sleeping on towers. I agree with Alex.
I think there's a next leg of growth coming into towers. We're seeing our BTS backlogs get bigger quarter -over -quarter, so some of that greenfield CapEx will shift into towers, and we're seeing more shift into fiber. I think you heard Steve's optimism here, but we're bullish. I mean, we've got great fiber businesses in Latin America, got a great business in Europe. We've got Everstream and Zayo here in the U.S. that are performing well, and so, you know, I do think there's an uptick in fiber CapEx, but I also think that we've changed our model, where we're asking the customer to front more of the CapEx. And I think the rest of this room has responded. Everyone's becoming more hawkish around NRC against MRC, and so that burden has to be a shared burden.
This is shared infrastructure, so every big route we're building today here domestically at Zayo or at Everstream, there's a shared CapEx proposition with our customers, and it's a partnership. And so I think that formula works well, and there's gonna be more CapEx coming. And I continue to see, as I said before, CapEx spend will continue to accelerate in the data center space. Now, brownfield's very different. So when we're buying platforms, we tend to actually be a little bit more differentiated. I mean, we did recap Vantage earlier this year, so that was-that's kind of skewing the numbers so far in 2024. But we've been very active in fiber. We've been around two situations in fiber. We're around a couple of situations in towers.
We're pretty symmetrical in our approach to how we're investing in brownfield and new platform development. I would say, you know, probably from a brownfield perspective, we're 30%-35% in data centers, which is hyperscale, private cloud, and edge. You know, we're probably 25% in towers. Add another 5%-10% small cell RAN infrastructure. I really like where RAN infrastructure is going. I think near edge, you know, for data centers will be RAN-oriented. Ultimately, you're gonna see cloud and generative AI racks and servers adjacent to Nokia and Ericsson's C-RAN and O-RAN deployments. It's just natural. It's a natural progression for mobile networks. When we talk about RAN hubbing, and we talk about building hubs, those really become the edge data centers of the future. That's an area that we're pretty excited about.
Look, fiber will be 25%-30% of our allocation this year. As I said, we've already done one deal, we're working on another, and there's a couple in the pipeline. So I think for us, greenfield is very skewed towards where the yields and the returns are, which is in data centers, and then on brownfield, we're being very opportunistic and balancing value against, you know, high-quality assets.
But I think as we invest out of this current strategy, our third strategy, we've been very clear with investors that that mixture of, like, 35-40% data center, 25% towers, you know, 25-30 in data centers, and the rest in what I would small cell RAN and emerging technologies, areas that we think are interesting that are coming up on the horizon.
Now, that's the split as we look across the verticals of digital infrastructure, but if we look at it on a geographical basis around the world, are there any areas, North America, APAC, where you're seeing, like, "Look, this is an opportunity area. We're going to put an incremental share of our capital to work in this geography?
We're generally risk on in the U.S. We're generally risk off in Europe. We're generally risk on in Asia. We like Asia.
Mm-hmm.
out of our third strategy, it'll be our second sort of basket of investments in Asia. We don't see the supply chain issues, and we don't see the power issues in Asia that we see in U.S. and Europe, so that bodes quite well for data centers. Now, there are pockets like Johor and Singapore that there are sort of flare-ups, and there's no more power, or there's threats from the city of Sydney that they're not gonna issue any more will-serve letters, but our intelligence indicates that those grids can actually sustain the workloads that are coming. And remember, Asia's a lagging market-
Yeah
... on cloud and AI. It's behind Europe, and it's behind the U.S. So as a laggard economy in AI infrastructure, we want to be in Asia. So we built a big team there. We've up to... We're over 45 investment professionals in Singapore. Very active team, really good team, led by a great guy who used to run private equity for TPG in Asia. So he's doing a great job building and running the team, and we're really happy about that. I would say other emerging markets, we're sort of risk... You know, we're kind of neutral on LatAm.
We're sort of watching a couple of- there's countries we like, there's countries we don't like. We definitely like towers in LatAm right now because of the 5G CapEx cycle. We're definitely keen on wholesale fiber. We've been building a lot of hyperscale activity in Brazil.
So there's good pockets in LatAm. We still think that's a good place to go. We did, in the last fund, build some data centers in South Africa. We've continued to be intrigued by the GCC and Africa, sort of, you know, Middle East, Northern Africa, plus Sub-Saharan. That's some of the fastest-growing wireless data adoption markets in the world. So in terms of penetration and subscribers and cloud adoption, Africa is the fastest-growing continent because they're so far behind, so it's like a bit of a, you know, the-
... the statistics are a bit skewed, but I think the institutional acceptance of Africa and the Gulf is coming. You look at the statistics, particularly out of Saudi Arabia, the UAE, Qatar, and Kuwait, those are the fastest-growing YouTube markets, and they're the second fastest-growing social media markets in the world. So we see a lot of maturation coming in the Gulf, and there's gonna be good investment opportunities there, and we're already involved in some of that already.
Gotcha. You know, one of the things we've been talking about is potential for rates to come down. You know, you earlier were talking about the need for... We'll, we'll wrap it up, we'll wrap it up.
That was me waiting for a rate cut.
Yeah, there we go. So if some interest rates come down, do you think we get an unthawing in the M&A environment that then drives further DPI back to LPs, which then drives actually a acceleration in fundraising activity?
I don't think there's a direct correlation between a 100 basis points reduction in interest rates and DPI. I don't. What I do think is this is some of the most volatile markets we've seen in a long time-
... because you have this perfect storm of political uncertainty here in the U.S. You have two major conflicts in the Middle East and Europe. You have, you know, inflation persisting, and you have high rates. And so you put that all into the supercomputer, and you get a lot of vol, which is what you're seeing in the last 30 days. And I think that vol continues. I don't think that subsides. Now, if interest rates come in 50 basis points, the market's gonna yawn, and volatility will persist. If interest rates come in 75 to 100 basis points, I do think that calms markets, I do think that stimulates lending, and I think in turn, that stimulates M&A activity. But at the same time, we're also watching multiples.
I think in the last panel you heard this disconnect between private M&A multiples and public M&A multiples, and there's a big disconnect there, particularly like in fiber, where you saw a private trade, you know, last week that was done at KKR paying like 22x-24x for a cable business, essentially. And a good cable business, mind you, but and then in the same token, you see some of these public equities in fiber trading for 4x. So that's a pretty big value disconnect. 4x to 22 or 24, that's about as big of a disconnect as I've seen. I think the disconnect in data centers and towers is not that big. Alex mentioned, you know, they're seeing private M&A multiples between, you know, anywhere between 22x and 32x, and you see the public comps between 16x and 22x.
That arb is kind of starting to rein in a bit.
Yeah.
I think the interest rate trade has already happened, and then in data centers, you know, we've seen the great recovery in DLR. Equinix is performing well, and private market trades are happening in the mid-20s to low 30s. So there's a 6 to 8 turn, you know, disconnect there. But there's still a pretty big... You know, we're not seeing a, a parity in public and private multiples like we saw 10 years ago. So that's not happening, and so I think that's a function of the fact that there's a lot of private capital sitting on the sidelines looking to activate that capital.
All right, last question for you is: you've been operating in this theater for over 20 years, communications infrastructure, and you've seen multiple cycles as part of this. As you think of the magnitude of capital that's being put to work in support of AI and other use cases, how is this cycle different and perhaps similar from prior expansion cycles? And as part of that, what words of advice would you give to the audience?
Well, I think we finished this conversation where we started, which is, "Don't get too high and don't get too low," right? The hype of the AI trade is definitely a bit overhyped, but what I would tell you is the CapEx numbers are about as good as we've ever seen. And so the optimist in me, to end the conversation, is that there's a lot of work to do in this room, and there's a lot of capital that's gonna be deployed. There's a lot of capital that's gonna be raised, debt and equity. So we have a big responsibility in this room as the industry, 'cause I think you've done a great job assembling the industry.
We've got a big responsibility to invest that capital intelligently and stay disciplined, because there were plenty of times in the last 30 years where people built infrastructure that was empty, and speculation kills the industry. Over-leverage kills the industry. We're not completely out of the leverage cycle from what happened in COVID, Michael. There's a ton of LBO debt rolling. There's a bunch of companies that have senior unsecured notes above that. We have businesses that were levered at 12x, 14x, 15x that historically were levered at 4x - 5x investment grade. So you've got this extra compounding effect of 8 to 12 turns of leverage, where interest rates have blown away from us. And make no mistake, a lot of these sponsor-backed deals are blowing holes in their covenants, and we're already seeing restructurings.
They're happening quietly, but there's a ton of LBO debt that has to be addressed, and a lot of that comes next year. It's 2025, 2026, and 2027 that we have all the LBO debt from COVID rolling, and at the same time, interest rates are getting cut, so there will be a chance to refi, but you're gonna see some blowouts. So I tell folks, "Look, we're in a great space. We're really fortunate." I've always told you this, that I'm, like, the luckiest guy in the world. But we have a big responsibility, and we have a... You know, on the horizon, we have a series of issues that are coming up that are gonna face our industry that we all need to start thinking about.
The waste of water, CO2 recapture, transmission grids, power, will-serve letters, and just the enormity of what we need to build. It's a lot, but it's exciting, like, we should all be pretty pumped up about it.
It's very exciting. Marc, I could talk to you all day. Thank you very much for being here. I really appreciate it.
Thank you.