DigitalBridge Group, Inc. (DBRG) Earnings Call Transcript & Summary

Okay. Good afternoon, everybody. Welcome to the Goldman Sachs Communacopia + Technology Conference. I'm Jim Schneider, I'm a telecom analyst here at Goldman Sachs. It's my pleasure to have DigitalBridge and CEO, Marc Ganzi, with us today. Welcome, Marc.

Thanks, Jim. Good to be here.

Maybe as a lead-up question, can you help us understand sort of your investment approach in the infrastructure space and how you're different as an asset manager versus a REIT in the space.

Well, it's an interesting journey for us. We started as a REIT in 2019. And what was becoming really apparent to us is the amount of CapEx that was required to support our customers. At that point in time, if you go back to 2019, we're kind of halfway through the journey of building public cloud. At that point, probably CapEx was somewhere in the order of magnitude of just about $0.5 trillion. Today, public cloud is about 80% built. About $1.3 trillion of CapEx has gone into that. And I think there's a broad acknowledgment even by our peers that are publicly traded REITs today, whether it's Andy or whether it's our friends at Equinix. To build these campuses and to build this type of infrastructure is not measured in tens of millions and today, not even hundreds of millions. We're talking about billions of dollars of CapEx. And so we felt like, at that point in time, that an asset-light approach would enable us to go quicker, would enable us to serve our customers better. And more importantly, it would enable our earnings to grow faster than in the format that we sat in, which was a diversified REIT. And so we made that pivot. It took us about 4 or 5 years to get there. We had to sell a lot of commercial real estate that we inherited. We had to clean up our balance sheet, get rid of about $17 billion of debt. Today, we have about $300 million of securitized debt. And so what it does for investors is it really allows them to participate in digital infrastructure at scale but not taking outsized risk, if that makes sense. And now we have a balance sheet that's levered at a very sensible level. We're not levered at 55%, 60% loan-to-value. We're levered somewhere between 3.5 to 4.5x FRE, depending on which numbers you believe. So we think that's the sensible approach. And also, the ability to have different sleeves of capital is so important today because a lot of our customers want us to show up for them and manifest ourselves in different ways that we hadn't historically thought of. So we think about our flagship fund and the different companies we own there. We think about InfraBridge, which owns infrastructure, digital and transport and energy. Credit is such a great hot product for us. It's really working well. We can provide capital to middle-market companies between $25 million and $100 million of EBITDA. Our core strategy is coming back into favor as interest rates moderate, and people want to own long-term assets. And we have a lot of stuff that we're working on in energy. And so it also gives us flexibility. We're not stuck. And I think this dovetails into what you and I were talking about earlier. A lot of our customers just don't buy data center space. They don't just buy fiber space. They don't buy space on our towers. They're buying holistic solutions to their network. And that's really something that we deliver at DigitalBridge that, unfortunately, an American Tower or Crown or an Equinix or Digital Realty cannot do. It enables us to have more conversations, more connectivity with our customers, and allows us to just grow faster. Our model is working, and we are growing faster on a CAGR growth than our public peers that are REITs.

Yes. In terms of differentiators, I heard you just raised to scale the diversification. What are some other advantages that you see for the asset management approach? And specifically, maybe what do you think is misunderstood about the DigitalBridge investment case?

Well, first, let's start with a couple of other things that perhaps I didn't say first. One, I like the notion where investors ride side-by-side with my personal capital. It creates really good alignment. I think as an asset manager where you've got the senior leaders in the company investing hundreds of millions of dollars into those products, you have spectacular alignment with our public shareholders but you also have that alignment of our private LPs that invest in our funds. The second thing that's really interesting about the asset manager approach is, if we are successful across the 46 companies we own and operate today around the globe, guess who really benefits? Public shareholders. Because you get to play in our carried interest. And so as we're successful and we achieve greater than a 2x MOIC or above an 8% hurdle, on average, we're generally getting 15% to 20% of the profits. And now, as you know, as part of our model, public shareholders grab a big piece of that chunk, of that carried interest. So as our funds mature, DigitalBridge is now 11 years old, we're now beginning to divest, as you saw in the first and second quarter. We're creating DPI. We are exiting some assets. And that's creating capital. That's creating liquidity. But it's also starting to create carry, and not carry on an episodic basis but carry on a consistent basis. And look, we're sitting on $86 billion of assets today across those 46 companies. We're managing $34 billion of equity at work. Generally speaking, digital infrastructure is a great asset class because generally, you deliver kind of between 2 and 3x MOIC. So if you took the midpoint of that, at 2.5x MOIC, and you look at the $34 billion of capital that we manage for the top investors in the world, you can begin to extrapolate certainly over the next 3 to 11 years, there's a lot of carried interest that's going to be paid out. And investors that have that patience and want to invest in the long term with us are going to be able to benefit and enjoy those profits side by side with our management team.

Yes. Now AI. That's been a consistent, maybe unrelenting, theme at this conference, I would say. A lot of companies are just in the early stages of trying to figure out what AI means for them, very early in their journeys. At the same time, we've seen hyperscale CapEx go higher year after year after year. We'll get into the different segments. But maybe from your vantage point, owning towers, data centers, fibers, et cetera, how do you see all this kind of playing out in terms of a holistic portfolio of different assets?

Well, I think we managed a lot of third-party capital. We have close to 300 clients around the world. And the first question I get at every fundraising meeting today is, "All right, tell us what to do in AI." I think the narrative on AI is pretty similar to the narrative of the last 30 years of investing. There's always something that comes along every 5 to 7 years, and everybody goes, "Wow, I got to go jump and do that." My experience is you never get too high in the highs. You never get too low in the lows. And generally speaking, the AI trade is kind of an interesting place. Maybe it's a little ahead of itself. Maybe we overcorrected. But the reality is history has a way of repeating itself. And having built digital PCS, having built the early stages of the Internet and having built a lot of companies that are part of these phenomena, it takes time, right? It took a lot of time for the Internet to make money. It's taken a lot of time for public cloud to make money. It took a lot of time for mobility to make money. Remember the days in wireless when all the carriers lost money. AI is no different. And so I tell people don't get euphoric around AI and the next great idea because, for every 10 ideas in AI, 8 are going to fail. That's just kind of the algorithm of investing in early-stage venture investing. But what is interesting is that, if it does follow the same arc that the public cloud followed, it took the hyperscalers about 6 years to start to monetize cloud earnings, right? So they started building that infrastructure in 2011. Those businesses began to cash flow in '16 and '17. That was really the catalyst moment. So when we look at AI, we have to say to ourselves, "Look, we're 2 years into this. We're in maybe the bottom of the first, top of the second, in a nine-inning baseball game." So don't get overly euphoric about it. Now the challenge to that is to make all of this work requires infrastructure. And all of the verticals that I just talked about, fiber, mobility, cloud, we know those were multi-decade journeys in terms of building that infrastructure because we started in 1994. That's our heritage is building networks. So I guess we have perhaps a little bit more sober perspective on it. And on the CapEx side that you explained, you're right. I mean CapEx, back in 2023, hyperscale CapEx was around $186 billion. It got to about $196 billion, $198 billion last year. It's going to $230 billion this year, and it's going to over $250 billion in 2026. So CapEx is rising and these guys aren't monetizing AI yet. But my suspicion is, if you believe that AI infrastructure got started 2 years ago, the logic or the algorithm would point to 2028, 2029 being the crossover point where they begin to monetize. We have a late-stage venture growth business. We invest in companies that use our infrastructure in stage Series C, Series D. We announced a joint venture with Intel called Articul8, which is a great business that focuses on generative AI and sort of private cloud environments, which has actually gone much better than we thought. We're getting bookings far faster than we thought. So maybe AI goes a little quicker and maybe I'm being perhaps a little bit too punitive. But what is interesting is, if you look at that CapEx that was spent last year, that almost $200 billion of CapEx, it wasn't just in data centers. Everyone thinks that AI CapEx is just related to big hyperscale data centers. It's not. It's ecosystem investing. So one of the great benefactors of that has been fiber. There's been a tremendous resurgence in enterprise fiber for us. And that's been primarily in dark fiber routes where we're seeking either new routes to data centers. AI requires multiple authentication path, which is redundancy, instead of a 2-path redundancy, a 4-path redundancy. We're going from hyperscalers purchasing 4 pairs of dark fiber to 12, 24, 48. So they not only want diversity and route selection, but they want more dark fiber. And so who wins? Well, for us, it's Beanfield, it's Everstream, it's Zayo, it's all of our businesses that are connecting data centers to those cloud environments or back to those PoPs. So we think fiber is one of the interesting sort of tangential winners. We do think, in time, mobility is a big winner. We do believe that most of generative AI applications will ultimately reside in mobile products. And so that ability for generative AI to work and gather information and gather data and ultimately make decisions, it doesn't happen at a desktop. It doesn't happen in an office building. Decision-making happens in the consumer's hand or it happens on the device. So that's a connected car. That's an IoT network. That's a machine-to-machine connection. We believe that generative AI will consume 3x more data than public cloud did on mobile devices. So if you once again go back in public cloud and you look at what happened to mobile data traffic between 2015 and kind of circa 2020, mobile data traffic went up 10x, Jim, in that 5-year period as cloud adaptation moved to the device. The same thing will happen in generative AI. So we believe there'll be a 3x increase over the amount of usage that you see on devices today. So we're very positive on towers, extremely positive on small cells, very positive on fiber. And the key thing to this is ecosystem benefits. And so when you get back to the DigitalBridge case and why is it interesting, it's interesting because we're building AI infrastructure holistically. I'm not just saying, "Okay, I got to turn on capacity in Johor," or "I've got to build a fiber route from Manassas, Virginia to Dublin, Ohio, because I got to connect 2 data centers or 2 clouds," it's really about where we've evolved as a firm is a holistic relationship with the customer. So when I sit with Google or I sit with Amazon, I sit with Microsoft, I sit with Deutsche Telekom, it's like, "How can we help evolve your network?" I'm not looking for a single order. I'm looking on how I can weaponize the 46 companies and the $84 billion of assets we manage to the benefit of our customers. That is very unique. That is the most unique digital infrastructure platform on the planet. And so we've got to get out there and evangelize the story a little more so the stock will perform. But we're very excited about what we're doing. And our customer relationships have never been better because the things that we're doing and the ways that we can show up for customers is very unique. And that's, I think, really the story today.

Yes. Now there seems to be a lot of capital still in the sidelines that wants to invest in infrastructure, whether that's private equity, infrastructure funds and otherwise. Do you believe, at this point, there are enough good assets and projects out there to invest in that can drive good returns? Are people kind of stretching to find levels of returns that they find acceptable? And maybe talk about what your threshold is.

Well, look, for 30 years of investing, we wake up every day, and we have a decision to make as allocators. We can go out and we can buy assets and we can execute a brownfield strategy and create great platforms or we can go up and do greenfield. And look, in the last 24 months, greenfield has been fantastic for us. I think we mentioned on the last quarterly call, we've got over 100 data centers in development, $29 billion of CapEx committed. All of that tethered to a long-term contract, 13, 15, 17 years in duration. I've never seen single-tenant yields on data centers where they are today. And some of that is a supply and trade imbalance and some of that's about power. We can talk about power in a second. But I'm actually finding, and we saw this on display last week, we had a decision to make on a big data center asset in Asia where we could have matched another GP and we could have paid a 2.3% cap rate to go chase that asset or we can go into that market and we can go back up behind our best customers and go build single-tenant locations at 11%, 12% yield. What's a better use of capital? Is it chasing something to an IRR that's for us doesn't work, where we get uncomfortable and we over-lever the asset, which we weren't willing to do? Or do we go with one of our management teams and really focus on our customers and focus on 6 critical markets in Asia where we can go toe to toe and compete? We chose the latter. It's better for us. It's in our DNA to work with our customers and compete with other platforms. And so I think there are situations that we look at, if you look at the last 4 deals we've done on an M&A side, we actually feel like we've been chasing value. And I do think for investors today, as we look ahead to close to $80 billion of LBO debt maturing in digital infrastructure in the next 36 months, we do think there's going to be value. And so we don't want to chase assets to a 2% cap or a 2.3% cap, I'd rather work with our customers and get a 15-year lease and build an asset that I know and control and that ultimately, I can lever and securitize and take that 11%, 12% cash yield and turn it into a mid-20s IRR, which we can do quite well. But it doesn't say that we're turned off for brownfield. I mean we did a recap of Vantage with Silver Lake that's gone really well. We bought a small really interesting business that lights up fiber to multi-dwelling units in the Southeast called OpticalTel, really happy with that acquisition. We just announced the first big tower deal in Japan, JTower, very excited about that, to have a platform that will focus on tower development. And so we are doing M&A, but at the same time, if I had to sort of frame where we're deploying our CapEx today, I would say, 50% brownfield and 50% greenfield. And I think you just have to be really disciplined right now. This is a moment in time where discipline will really matter.

Yes. Makes a lot of sense. Maybe just kind of switching to data center for a second. The bulk of what we've seen in data centers has been training. It's been about training, training, training. And now people have talked about inference a little bit more. How do you think about kind of the trade-off between investing in wholesale data centers versus retail data centers? How do you think the transition between training and inference goes? How long does it take to kind of get there? And how do you think about balancing your investments?

Well, it's interesting because I think a lot of investors don't understand that the data center marketplace is actually 6 different business models. And so we wake up every day and we say there are 6 different ways we can deploy capital into data centers. Certainly, managed services, hybrid cloud. You mentioned the retail colo. You have hyper edge or Tier 2, Tier 3 near-edge compute. You've got public cloud and you've got private cloud. Now where are those big AI workloads being built? They're being built principally in training models. Most of that is in private cloud. A lot of people think that AI just sits out in the public cloud, but that's not true. All the hyperscalers are building those initial language models in a private environment. Why? We're not turning it loose yet. That's inference, right? And that's when you get to generative AI. And those big, big sort of language-based models, which are kind of 400 to 1 gigawatt workloads, those data centers are not exactly readily available. And so you have to be very careful about how you do those and how you build them. And we built a few of them, and they're very hard and they require a lot of CapEx. But if I think about the 6 sort of different business models and where we can put capital, we're sort of risk off on managed services. We do have a little bit of hybrid cloud that we do. We're very negative on retail colo. We think that's an aging and dead asset. We love hyper edge, which is what DataBank does, which Equinix does. Edge computing for us, that's our fastest-growing vertical. It's actually faster than private cloud and public cloud. We're obviously very long on public cloud. We've got Vantage. We've got Scala. We're very excited about what we're doing there. And then probably our most intriguing business is our private cloud business at Switch, which has just absolutely defied gravity in the last 2 years. So I think as investors, you get to choose, right? You can wake up and decide where you want to allocate capital. And ultimately, the returns are very different across those 6. Not all data centers are created equally. And so if you're looking at sort of managed services, those businesses trade at 4x. If you're looking at retail colo, it will trade at 11 to 12x. If you're looking at hyperscale data centers, it's going to trade in the mid-20s or, as we saw last week, at 2% cap rate. So it's a little bit of pick your opportunity and choose wisely and really understand your risk tolerances and understand how that correlates to returns. But I think for us right now, we're super focused on edge. We're very focused on private cloud and public cloud. And those are the 3 areas where we're investing, the other 3 areas we're sort of risk off.

Yes. Fair enough. I guess maybe just trying to think about how you think about, kind of on the other side of things, potentially monetizing assets over time in your portfolio. There's obviously been some rumors out there last several days. But how do you think about kind of like those things? Anything you want to comment on that front?

Well, I always tell everyone in our Monday morning meeting that everything is for sale always. And the junior partners in the firm look at me and they go like this, and they're sort of like, "Why do you say that?" Well, it's really simple. We're investors, right? I think we do compete against asset managers. I want to be clear, there's a difference between being an asset manager and an active investor. We're the latter. We're not the former. My job is not to collect fees for 12 or 13 years and not monetize assets. There are other people that are GPs and that are in the infrastructure space and that trade as an asset management, and they want to collect management fees for a long term. We think our job is a little bit different. We think we have a very high standard that's placed on us every day. We have over 300 LPs. We have a fiduciary responsibility to create great returns for them, which we've been doing for many decades. The duality of being a public GP and being an investor is part of the game or part of our job is we do have to return capital. So we've had 8 exits in the last 24 months. We've returned over $8 billion of DPI. And in the first quarter this year, we had some DPI. The second quarter, we had DPI. Look, if we're going to keep fundraising to the cadence that we're raising, part of that mission, part of that job, is I have to return capital back to my LPs to get new capital. So there will be episodically quarters where we have a great outcome and we exit something at a great price. We lose the management fees but we gain carried interest. And I think that's a big part of our story is that we're sitting on a significant amount of carried interest that I don't think people appreciate or the Street appreciates. I think what you were talking about earlier is there was a story about Switch. All my one-on-one meetings today, I think the first question I get is, "Tell us about the Switch IPO." Everyone's ready for another publicly traded data center logo. I won't comment on Switch's IPO or non-IPO. What I would say is great assets and great management teams can find their way back into the public market. It's been a lot of fun working with Rob and Thomas and Madonna Park. They've done an amazing job. And it's such a unique asset because no one does private cloud like Switch does. And nobody can do it at scale. And I think the thing that we saw in Switch that maybe public investors missed was a massive land bank where we could build out 12 million new square feet of data center space. And we had a runway clearly into 4 gigawatts of power; and not just regular power, renewable power. So that combination of renewable power and a massive land bank is what we recognized when we diligence-d the company. And now we've executed. We've gone out and more than doubled the EBITDA of the business in 2 years, way greater than that, and it has a huge pipeline and it's executing. And also we gave a great talented CEO, Rob, who wasn't exactly in love with dealing with public investors because Rob is a very mercurial talent. But what we've done is we liberated Rob and we freed him up to go do what he does best, which is sell to customers and build great data centers. And when we can work with CEOs and we can help them catalyze their dream and help them go in the right direction -- we also gave Rob billions of dollars of capital, which as a public company. Switch was stuck. It didn't have billions of dollars of capital to make 400-, 600-megawatt commitments to big customers. So we then had a big wallet. He didn't have to go to public investor meetings. We freed him up and he just absolutely crushed it, as did Thomas and as did Madonna. So it was one of those unique stories where you can take a company private, make it better. And perhaps we do bring it back into the public format because Rob, I think, is a little older. He understands a few more things. And also the business is really cash flowing now, which is great. So it could be Switch 2.0. But I want to come back to where we started this conversation, which is what is our job, when do we sell assets. We sell assets when we can get a 30% to 40% premium to NAV, which is held on our books. It's almost our responsibility to sell an asset. So if anyone comes to me and gives a 30% to 40% premium to NAV, we'll sell. We did that with Wildstone, our digital billboard business in Europe. We sold it to Antin. And we love that business, and we love the CEO. And the business is growing at 25% CAGR. It was the only consolidated owner of digital media infrastructure in Europe. Antin saw that. They bought the business. They paid us a massive premium for the business. And recently, I was in Paris meeting with Alain and his partners at Antin, and they said, "Wow, we're so happy. That business has absolutely crushed it for us." That makes me happy because we actually sold the business to them. We achieved our outcome for LPs. Yes, we lost fees, but we gained EPI. We printed a great return. But most importantly, we printed carried interest for our public investors. And it's not like that that's super unique. We're going to go back out and do it again with a different management team. But I like outcomes where everybody wins. And to have a buyer come back to me 2 years later and say they're really happy with an asset and they're happy with the management team means we've also done something nice because we've actually sold somebody a business that's high quality. And we're not afraid to do that. We don't need to hold the asset for 10 to 15 years.

Great. I've got about 25 more questions on data centers, but maybe I'll forestall that for a second and maybe ask about fiber, another important part of your business, which you referred to before. Lumen has announced about a $5 billion deal with Microsoft and other partners. Kind of interconnect data centers, you referenced that as an application area. Maybe when it comes to these kind of deals with this just kind of general characteristics, what type of returns are you underwriting? Can you see more deals like this in the future or maybe many more deals in the future? And this is a dark fiber deal, how interested are you in participating in that kind of deal, in particular?

Well, we do that. We do that at Everstream. We do it at Beanfield. We do it at Zayo. So we're very skilled at building dark fiber networks. I think, look, for Lumen, it was a necessary transaction. The structure of it, I didn't love. I wouldn't sell my network, which is kind of what they did. I think the original Level 3 transport network is one of the best networks out there. So selling a piece of that maybe was the right idea for Kate and what she needs to do with that business. But it's something we wouldn't do. We wouldn't sell the infrastructure. Instead, we would sit with the customer and we'd say, "Tell us the diverse routes you need. We'll tell you where we have network. If we don't have network, we'll go build it for you." But I think when you're in the infrastructure business, the last thing you want to do is sell the assets. So we have a great relationship with all the hyperscalers. We're building some fantastic routes. And I think what's interesting today versus where the fiber business was 3, 4 years ago is we started at all of our fiber businesses with the mantra about 2.5 years ago that the only way the fiber business is going to work is if our customers start putting up NRC, nonrecurring charges. And so getting our customers to invest in new routes and giving us that CapEx upfront, it's really important to us because it defrays our going-in CapEx. Because our CapEx at Everstream and Zayo was just too high, it wasn't sustainable. So my first goal, my first priority, was reduce that CapEx going in, create shared infrastructure and create shared risk. We have to share in that risk with the hyperscalers when we're building dark fiber routes. And then the key to that for them is they have money. They can put up CapEx. And what we tell them is, "Look, we want to reduce your MRC, so your monthly recurring charges are lower." And by reducing their charges, they have less hit to EBITDA. And everything for them is about earnings and revenue growth. So this actually was a good way to reset the relationship with a lot of our hyperscale customers. And I think what we see is we see a more efficient use of capital. Our return on invested capital, which used to be a null set, is now inside of 48 months at Zayo and Everstream and Beanfield. We've gotten all of that CapEx recaptured now inside of 4 years. So sometimes you got to have that discipline again to step back and reset the relationship. The good news is there's a lot to everyone. I'm happy for Lumen. I'm happy for Kate. I hope they succeed. I want everyone to succeed. But ultimately, when you're doing these big greenfield projects, they're complicated in fiber, particularly if you're doing a long route, a long-haul route, that's greater than 500 miles. So we're finding that those long-haul routes are really lucrative. The mid-mile routes are doing really well for us. So a lot of that metro loop where you're cutting laterals off those metros and out to data centers or out to towers or out to RAN hubs, that's proving again. That's coming back again. That business is coming back as 5G densifies. And more importantly, as the mobile carriers create C-RAN and O-RAN off-ramp where you have those cloud on-ramps, those are proving to be really good areas to invest in fiber. So fiber has really come back. I'm talking about the enterprise side. I won't talk about residential. And dealmaking is coming back. I think you saw in the last 2 to 3 weeks, we've had a lot of deals. Frontier and Verizon, I think, was a great trade. I think both organizations won. The MetroNet trade was fantastic. I mean, for our friends at Oak Hill, that's a great outcome. I mean I'm really happy for them. And hopefully, that works out for T-Mobile as well. So there's a lot happening in fiber. I think you're going to see more fiber assets trade. And it is one of the tangential winners coming out of AI, early innings.

Yes. How do you think the fiber market evolves? Well, I don't know if it's clear, if I'm even right, but it seems to me like there's not every customer in the world that's a hyperscaler who can handle dark fiber and management of that kind of asset. So as this kind of percolate down to enterprise and down the stack beyond this kind of very small set of customers, how do you think the market changes?

Well, people still need lit services. You hit the nail on the head, not everybody wants dark fiber. So from that perspective, we still see growth in lit. Pricing pressure is really tight. You have to have good relationships. You have to have unique routes. And you've got to be able to leverage scale in that business, in lit services. Lit services is a big boys game. So we still do it. It's profitable for us. It's net of churn. It's still profitable and we're growing. But in terms of the dark solution or data center connectivity solutions, that dwarfs that growth in terms of lit services. But I think as a fiber player, whether you're Beanfield in Canada or you're Everstream in the Midwest or you're Zayo with a nationwide footprint, you've got to have that full suite of services. And there's other products you have to have. I mean I think about SD-WAN. SD-WAN is a really important product. You have to have the ability for some organizations to create a software layer to their connectivity. Most organizations are doing that now. So our SD-WAN product is growing. We bought a business called QOS in Zayo. That's performed well. And also, you've got to be cognizant of what's going on with NTIA and some of the broadband grants. We've won some of those important opportunities, won 1 in Louisiana, we won 1 in Nevada. We're bidding on a lot more stuff. The NTIA is finally releasing that $35 billion out there into the wild. It took some time to get there. But I think what you'll see in the next 24 months is a lot more of that Build Back Better America grants, part of the broadband bill initiative. So there should be more E-Rate opportunities. That's been a very lucrative business for us where we light up schools, and we get 10-year contracts that are sort of government backed. There's a lot going on in fiber, but it's definitely starting to turn, and we'll see. It's been a battle coming back out of COVID, but I think we're fighting our way back.

Yes. And then on the tower business, I want to ask you how that one plays out because clearly, I think we're about 50%, a little bit more, upgraded in terms of 5G on the domestic tower portfolios for most of the carriers. Given the sort of uncertainty out there about when more spectrum becomes available, how do you think this kind of plays out in their investment plans? Do you think that we see a build-out of that and then a lot more densification? Or how do you think that kind of goes?

Well, we invest in towers globally. We're in 5 different continents around the world. So we're building infrastructure in Asia. We're building infrastructure in Europe, North America, Latin America. And each of these regions, Jim, is growing at different speeds and velocities. I think if I had to stack rank our favorite markets today, we think the U.S. is in a great place right now. So we're investing heavily with 3 of our big customers. We've build-to-suit relationships in place with all 3 of them. And we actually see that more macros are coming on because there's actually a lack of spectrum. So that lack of spectrum creates frequency reuse, which is a lot of what we saw in congested 3G networks, congested 4G networks. And now that 5G has turned on, we're seeing the same thing. So history again has a way of repeating itself. We think there's probably 65,000 to 80,000 new macro towers that are needed in the next 3 to 5 years.

New towers.

New towers. And then on top of that, we think there's another 300,000 to 400,000 small cells that are needed from an outdoor perspective. The small cell outdoor market today is somewhere around 800,000 nodes, give or take. So that's going from 800,000 to 1.1 million, to 1.2 million. This lack of spectrum, particularly in the mid-band and the higher bands, is a real complication for 5G, maybe less so for T-Mo given their rich spectrum position. But if you look at what Verizon and AT&T have, what we're seeing from them is a commitment to investing in the network, and we'll be there for them because Boingo can do it, ExteNet can do it, Vertical Bridge can do it. We've got great companies that face those customers, and we're having really fantastic conversations. We don't sleep on Latin America. We have 2 businesses down there -- we have 3 businesses, Mexico Tower Partners, ATP and Highline, all of them growing at double-digit CAGR growth. So that's mostly amendment traffic, some macro, some BTS. But we have ATP growing at 20-plus percent. We've got MTP growing at 12% and Highline in Brazil growing at 16%. So those are all 3 businesses growing at double-digit organic growth rates. We really like Latin America from a towers perspective. And then Asia has been quite good. Southeast Asia, where we have the largest private tower company there called Edgepoint, now over 56,000 towers, going to 80,000, big build-to-suit pipeline. Finally, colocation is happening there. It took a while for colocation to get going. And soon we'll be in Japan, of course. So 9 tower companies around the world, all performing. We've got a great relationship with Deutsche Telekom in Germany. That business has produced better than we thought, a lot of colocation now as the DT towers open up a little bit. And we're getting Vodafone. We're getting Telefonica. We're getting one-to-one. Belgian Tower Partners, we have the largest tower company in Belgium, that's performing quite well. And then we have a business in the U.K. called FreshWave and another business called Digita in the Nordics that owns mobile infrastructure up there. So we're very steeped in towers. Towers is our sort of lineage. We've been doing towers for 30 years. And I still, to this day, tell investors, I haven't found a better business model than towers. If you can find a better one, share with me. I know data centers are superhot right now. But think about an industry that's been as resilient as towers has been for 30 years. And I think it will continue to be resilient for the next 20 years.

Yes. And just maybe to close on that point, do you think there's parts of your portfolio or parts of the infrastructure space that are kind of being overlooked or maybe like not prioritized by investors that people should be paying attention to and over what time frame?

Well, it's a value question, right? It gets back to what's your entry cost and how you get in. I mean, we've been looking at the satellite infrastructure space. We talked about digital media infrastructure, which is digital billboards. Suboceanic cabling is coming up for a significant day of reckoning. All of that infrastructure that was put in 20 years ago is aging. So glass, particularly glass underwater, suffers from dissipation over time. And so a lot of these really, really important routes that were laid 20, 25 years ago, they're aging out. So there's a huge opportunity to recycle capital in suboceanic cables. Nobody talks about that. That's a big opportunity. It's a threat at the same time. When you think about people cutting cables up in the Nordic Sea, we won't talk about who those people are, but the business is a good business. It's misunderstood. Satellite infrastructure, misunderstood. Digital media assets, misunderstood. So there's parts of the ecosystem that a lot of people don't pay attention because right now, it's super easy to be a frontrunner and go do data centers. But there's a whole ecosystem of trillions of dollars of CapEx that we focus on and that we're investing in every day. Again, it's about being completely focused on ecosystem, having capital to go anywhere and most importantly, showing up for customers.

Yes. Sadly, we're out of time. That was an awesome overview. Thank you, Marc, for being here.

Thanks, Jim. Appreciate it.