Fireside Chat with American Tower. I'm Jonathan Atkin with RBC, and pleased to welcome Steven Vondran, Executive Vice President and President, US, from American Tower. Welcome.
Thanks, Jonathan. Happy to be here.
Maybe just to kind of kick off for those that don't know you, give us a kind of a sense of your responsibilities with the company and what you did prior to the CoreSite acquisition, and then how your role has evolved and expanded.
Okay, sure. So Steven Vondran, I've been with the company for over 23 years. A lawyer by training, been in a variety of roles, operational roles, finance roles, legal roles in the company, and I assumed the job of President of US Tower about 5 years ago. And so at the beginning of that, it was really just the US and the tower business, and with the InSite acquisition, we expanded into Canada, so responsible for Canada as well. And then a little less than 2 years ago, we bought CoreSite. So now I also run our data center division for the company as well.
So, given your long tenure with the company, maybe you can kind of reflect on, you know, the 3G build-out, the 4G build-out, the 5G build-out. What are kind of the broad-stroke observations that you would have as we kind of look at prior cycles, up to the present day?
Yeah, what I would say is, it's a very predictable cadence that the customers follow when they're deploying a new G. And even before American Tower, I was in the industry before that, so I started back in 2G. And what you see typically in the cadence of a build is that first phase is overlaying their existing sites with whatever the new technology is. Then, after you get through that initial rush to cover the first number of POPs, you'll see them slow down a little bit while they start looking at their network and figuring out where are the holes in the network, you know, where's the quality issues that we need to address, things like that.
So you'll see them pull back a little bit, and then they'll start looking at that, and you'll see another wave as they try to cover the rest of the POPs. And as more people start using their phones and you see demand on the network, you'll see a capacity build. And then once that's done, you typically see an infill, where you'll see some cell splitting, they'll densify the network because they're starting to see some, you know, some capacity issues that can't be solved on the same towers.
And then, towards the end of the G, you see, because mobile data usage has risen, you know, you know, 20% to 30% to 40% a year for the last 20 years, you'll see them running out of capacity again toward the end of the G, and another whole wave of capacity adds. And what I would say is that 5G is really no different from our perspective than the prior Gs. And what you're seeing today is we had a big ramp-up in activity as people were pursuing their coverage builds, trying to get as many POPs as they could. And what you're seeing today, I've heard some people talk about a pause. We're not seeing a pause. We're still seeing activity on our towers, but you're just seeing that normal cycle play out, where activity is dropping a little bit.
I think it's a little bit different in 5G from the prior from 4G. In 4G, you had more of a layered start, so each of the carriers started at a different time, so those kind of waves and troughs were spread out a little bit more. In 5G, you know, one carrier had a little bit of a head start, but what you saw is everyone was starting at the same time, so the peak was higher, and so the trough's a little bit lower. So from my perspective, 5G is not following any different cadence than we saw in the others. Just a little bit of a timing difference in terms of when the carriers started building.
So giving your long tenure with the company, maybe take us back to maybe ancient history, prior to when MLAs existed. I think the first one
That's a very long time ago.
So October of 2010, from memory, is, is when the first ones got signed with AT&T, you, both you and Crown. But the operating model, you know, pre- and post, any, any, any kind of observations around how you actually run the business, and then you converted to REITs in 2012. Did that have any practical impact from an operating perspective?
The biggest change when you become a REIT is the requirement to pay out your taxable income as a dividend. From an operating perspective, there's a little bit of compliance around the edges in terms of how your contracts are written and things like that, but it didn't really change the operating model. When you go back to 2010 and kind of the comprehensive MLAs, which I think is what you're referring to there, that's really when you started having a couple of companies get enough scale that it made sense to do things on more of a portfolio basis versus an individual basis. Prior to that, you know, the entire industry was pretty much working on a site-by-site, à la carte pricing model, and there was some variability in terms of what those contracts said, but, but not as much.
You know, starting in 2010, there was an opportunity because both American Tower and Crown had the scale to look at a more portfolio-wide approach, and we found a different way of doing business with our customers that kind of streamlined things. That really started kind of a different era of the relationships with our customers, and we figured out that we were able to to monetize those agreements as effectively as if you were going pay by the drink, but in a way that was a lot more administratively effective for both parties.
Because of those administrative efficiencies and because of the effects of those MLAs, we're confident that we've driven more business than we would've in a pay by the drink, and that we're getting higher returns than we would have if we'd stayed with that kind of old operating model years ago.
You know, throughout that period, you perhaps thought about doing outdoor small cells. You do a lot of indoor DAS from SpectraSite and kind of growing that business, but what's been the calculus around abstaining from that opportunity, whereas one of your peers obviously is
Sure.
leaning in?
So we do have a few outdoor systems... and we were experimenting with kind of the outdoor DAS model, which has kind of morphed into small cells in a lot of areas, and we were experimenting with that a decade ago. And what we realized in the outdoor space is it's the fiber cost is the largest component of the cost of that business. And when we looked at it and looked at the incremental returns that we could drive, we just found there are better places for our capital. So it's not that it's a terrible business, it's just not a business that meets our return criteria compared to other things we can do with our CapEx. And that's we test that thesis occasionally.
We'll go back and look at the market and try to see if there's a place that we could drive outsized returns, and we really haven't found opportunities to scale it in any way that would make sense for us.
So a couple more tower questions, then we'll pivot to data centers and CoreSite, which, you know, also falls within your bailiwick. Maybe kind of fast-forwarding now to the current decade, and you know, you've talked about a long-term 5% organic growth rate. What are some of the swing factors that would cause you to deviate from that? You know, the reported results are arguably more decoupled from activity levels, given the holistic MLAs that you have with many of your customers. So how do we think about the growth rate and what could push you above or below that?
Sure. So let me just kind of reiterate what we've said publicly. What we said is that we anticipate an average of at least 5% OTBG from 2023 through 2027 on average across that period. And we've also said publicly that we have about 75% visibility into that number. And so I'll kind of break that down a little bit further with you. So, that 5% is 6%, if you exclude the spread churn. So when you kind of back into that, we have a 3% fixed escalator. That's one component of that, and then we have our comprehensive MLAs that are the other component of that.
Our historical churn has typically been in the 1%-2% range, so I think that we can imply in that math is that we have a 3%-4% new business growth on top of that. Now, if you break that down into, you know, what are the opportunities to outperform or underperform, what we've baked into that, that 5% over that time period is what we expect the carriers to do. So we've looked at 5G. We have a lot of history with prior Gs, and we've kind of factored in what we think the data growth rates are going to be over that time period, what the technology is capable of, the equipment that's available in that timeframe to meet those needs, and we made some assumptions about the activity levels on our towers.
A lot of that's underpinned by our comprehensive MLAs, so that does give us some smoothing through the, the waves and the troughs. And, and so we are expecting a, some business outside of that that's baked into that guidance. Opportunities to outperform would be that if the customers do more, if there's more demand on the networks than we're expecting, if mobile data grows faster than we're expecting, and they have to add more equipment, there could be potentially some upside, especially from new sites. There's also, we have not baked in meaningful new business from a new customer. So if you saw a new greenfield build by a cable co or some other new player that came into the market, we haven't built in meaningful upside from there.
And with DISH, we've baked in our contractual minimums that are in our comprehensive agreement with them. So if you saw DISH, you know, suddenly become a more meaningful player in the market, and they outperformed what their network's capable of, there's some upside there. On the downside, there is some new business baked in from what we expect to have happen. So if people stop using their phones, if kids decide they don't want to do social media on them, and you see a slowdown in those networks, you could see a slower pace to that build. But again, that's where our comprehensive MLAs give us a lot of protection over that term period.
So when we look at the market today, you know, I've mentioned the peaks and the troughs, and we know that people have talked about a pause, but, you know, that's not what we're seeing. What we're seeing is still activity on our sites, and we think it's just part of the normal cadence. But we look at that, and we feel very confident that that the numbers that we put out there are what we're going to see over that time period. Again, on average, there might be a little variability year to year, but we feel very good about that.
On domestic, on domestic M&A, we had a panel yesterday, with private developers, and financial sponsors, commenting on how there's still quite a gap between, private market TCF multiples and, what the public guys trade. What are you kind of seeing as, somebody who regularly does, you know, kind of small tuck-in acquisitions, and any kind of observations on, the pace of that kind of activity in, the broader industry?
Yeah, we're seeing that as well, where there's still a dislocation between private multiples and public multiples. Having said that, we are seeing those multiples come down a bit. So I think you are starting to see some of the private capital players raising their return criteria, maybe challenging some of the growth assumptions in some of those models. So we're seeing them come down a little bit, but not enough for us to do anything in a meaningful capacity in terms of M&A in the US.
So you have gone through the acquisition of the Verizon portfolio. If there were at some point an opportunity to acquire carrier assets, any sort of lessons learned from prior cycles? And how would that inform your view on... You know, the carriers do still build some of their own towers, so there are some opportunities to do sale-leaseback deals. Is that fundamentally of interest? What’s your thought on that?
It really depends on the terms and conditions of the pricing of what's out there. You know, we evaluate everything that comes to market, like every other company does. Yeah, I think what we've been very clear about is our capital allocation priorities right now are funding our dividend 'cause that's required as a REIT, and also delevering. We've committed to a delevering cycle. We also have an internal CapEx program that delivers some of the best returns that we can get. And then beyond that, if there's M&A that's more attractive than the stock buyback would be, then we would look at that. But, you know, I think we've also said publicly that right now there's nothing that we see in the M&A market that's compelling enough to change our capital opportunities at the moment.
If there's any tower questions, I'm happy to take a few now, and we'll have some time at the end. Barring that, I'll maybe pivot over to data centers and CoreSite. So, maybe discuss a little bit about what you're seeing in terms of the demand environment for kind of retail enterprise interconnect.
Sure. Look, we're in a very high demand environment where supply is constrained, and that's really true across the 10 markets where CoreSite operates. I think what we said publicly is we had a record year leasing in 2022. We put out some supplemental information with our earnings, showing the pre-leasing going into the construction that we have going on. And so I think what you're seeing in the environment, it's a couple of things. Certainly, generative AI is putting demand on the entire ecosystem, and that means there's less total space available to everyone. For CoreSite, you know, we are an interconnection hub. That's really our business. We're not in the hyperscale business.
We have a few hyperscale customers, but we try to balance an ecosystem that's balanced between, you know, it's retail scale and hyperscale, but it's also enterprises, clouds, and networks. And we try to balance that because that helps us drive the highest returns on invested capital in the industry. CoreSite, traditionally, as a public company, had the highest returns in the industry. We're operating with the same philosophy there, and so we're still, you know, maintaining that discipline in terms of who's in there and the kinds of returns that we can drive on it. We're still seeing a lot of demand from hybrid cloud deployments, and that's where enterprises either are finally going off-prem.
They're finally getting rid of their existing data centers in their own buildings, or they went cloud, and now they're realizing that it's cheaper to be in a hybrid cloud environment. And so we're still seeing a lot of demand for that, and that's true across the portfolio. There are also power constraints in some markets, and that's driving prices up in those markets, and it's also creating you know, demand for different types of applications that we evaluate case by case. And sometimes it makes sense to do those, and sometimes it makes sense to pass.
What has been the most notable change in the business since it's come under American Tower ownership versus when it was independently operated?
I'd say a couple of things. One, you know, we have the same management philosophy and the same approach to pricing. We've been able to fund their CapEx programs a little bit more robustly. And, if you look at that, you know, what we spent last year and this year, it's quite a bit more than what they were spending historically. And, that's gonna give us great returns because we're in high-demand environments. So we've got a number of projects under construction. We've talked about SV9, it's our new ground-up facility in Silicon Valley. And also, we're building computer rooms and all in most of our existing facilities where demand is outstripping supply.
Probably the biggest change, I would say, in the way the business is running, and it's more of a function of the environment than American Tower, I think, is we have a lot more of us pre-leasing on the stuff that we're building. So if you look at all the data centers we have that are under construction today, they're about 36% pre-leased, and that's a high-water mark for the industry there. But I'd say the biggest difference with American Tower running it is that we're funding the CapEx a little more robustly and taking advantage of the great market conditions that we're seeing today.
They, they used to talk about targeted development yields when, when they were independent. You know, given, given the higher pre-commit that you're seeing, that, that might suggest the deal sizes are getting larger. So any, any comment on deal sizes, and then, and then do you still have kind of the same calculus around underwriting new business opportunities?
We'll start with the calculus first. Yes, we're still the same yield criteria, and in fact, we, you know, we try to drive yields up whenever we can. In a favorable pricing environment, you can do that. In terms of deal size, what we're seeing overall is that everyone's using a little bit more space and power than they did before, particularly enterprises in a hybrid cloud environment. We've seen the sizes of those installations increase quite a bit over the last, say, five years. This year in particular, where people are concerned about future capacity, we're seeing people actually lease into kind of future demand because they're worried about the ability to grow. We're seeing the deal size grow a little bit.
So you, you mentioned hybrid cloud as a, as a driver, rapidly moving, you know, demand profile around, around, various types of AI. Anything qualitatively that you can share, around the, the sales pipeline and, and the composition of it? Has it changed in even the last couple of months?
. Well, I'm not gonna give you too many specifics on the sales pipeline, but what I would say is we've seen some increased demand for generative AI-type applications. Now, again, we're not the appropriate forum for the big server farms. And so if you think about, you know, the large learning models and the large compute, that's gonna go to a hyperscale facility. So it's a much cheaper cost of operating than, of course, we're not a discount provider. Where we're seeing demand is we provide the interconnection environment that feeds those large learning models. So you'll see somebody who builds a server farm somewhere else, and they wanna be in a highly interconnected facility like CoreSite, because that's how you feed all the information to it. You're pulling it from various sources. So we've seen some demand there.
You know, AI has been a driver for a while. Generative AI is kind of the biggest thing in the headlines now. But you've had traditional machine learning, you've had other kind of AI applications that have been in our facilities for a while, but we are seeing an uptick in, you know, people reporting AI as the driver for their interest right now.
So if there's questions, please raise your hand. So on that last answer then, you are seeing some generative AI demand. Would that manifest itself in things like cabinet adds, or more like cross-connects, or a combination?
It's both. What we see is, we see some specific installations to support the AI, and when you have those come in, you see a lot of cross connections. So we're seeing an uptick in use of our OCX platform, which is our fabric, and also an uptick in the internal cross connects for those applications.
There's questions, feel free. Your sites are the destination for, you know... We had a fiber panel yesterday, where we talked about network as a service. So you, you've always been, you know, a destination for some of those interconnect hubs. Is that something that you see as a driver of your cross-connect business? What's your view on other companies that leverage your connectivity platform to kind of extend the reach of their networks?
Sure. So, you know, you don't come to CoreSite and say: How much money are you gonna save me? We're not a discount provider. You say: How much money are you gonna make me with your environment? So people come to CoreSite to connect to other people. That's the main driver for being there. So the networks are a component of that, but the clouds are as well. And we have the second most cloud on-ramps next to Equinix, and everyone else is kind of a distant third. And so the cloud on-ramps combined with the networks, and then also with the existing enterprise ecosystem, is really what the secret sauce for CoreSite. And so I would say the fiber guys are definitely a part of it, but the cloud players are just as important as the fiber.
Is the nature of your cross-connect business more around, you mentioned the OCX platform, or is it still kind of bilateral physical cross-connects? Any way to give us some color on that?
You know, right now, the physical cross-connects make up the majority of the cross-connect revenue, but OCX is a growing platform, and we're seeing higher growth in OCX than we are seeing on a percentage basis in the fiber cross-connects. So look, they're both very, you know, healthy ecosystems that are growing. The beauty of OCX is that we can extend it out into smaller facilities. So as we think about, you know, the future of edge computing, OCX can provide that connectivity back to that connection-rich environment, in a, you know, kind of a one-off basis.
Looking at the portfolio, you're in a number of major internet gateway markets. Maybe give us a bit of an update as to where you have recently commissioned capacity. You mentioned a 36% pre-commit, but you do have inventory in a number of markets. What are kind of the high-demand ones that you're seeing?
So the kinda New York market has been kind of a robust growth for us, particularly in financial services. We're seeing, you know, a lot of growth in Silicon Valley and L.A. Those have traditionally been high-growth markets for us, and that's not slowing down at all. And then our Northern Virginia campus in Reston is seeing a lot of demand. Again, we're not in Loudoun County, so we're not subject to the same power constraints that are occurring there. And so we've got some capacity there that people are excited to take advantage of in Northern Virginia. And actually, here in Chicago, we're seeing higher growth in Chicago than we've seen in a while. So it's really across the board, we're seeing some elevated activity, but those kind of key markets, New York, California, and Northern Virginia, are the primary drivers.
So, number of your peers that used to be independently listed are now privately held. Has that? How has that kinda shaped the market dynamic in your various business segments within CoreSite?
You know, most of the... Really, the only competitor that has the kinda same type of business is Equinix. Most of those other companies that went private are more focused on hyperscale, or they're focused more on retail, and they're not really as interconnection dependent. So we haven't seen a huge difference in that. Now, you're seeing more capital flow in the ecosystem, and that's good because demand is going up. You know, it's not even linearly, it's more exponentially, is how demand's going up right now. And again, all of those installations need to be connected back to other people at some point. So I think it's actually healthy for the ecosystem, for those companies to go private, for some of that private capital to flow in and meet that, you know, kind of large hyperscale need.
It just helps the rest of the ecosystem develop and grow more.
Maybe the last question from me is around the, I think you talked about 1,000 sites to tower sites now, where you have enough land, power, fiber, either in place or readily deployable, that could serve as the host for kind of mini compute hubs?
Sure.
So where do things stand in terms of trials underway? Anything in the last year that's shaped your thinking about that opportunity?
Yeah. So I want to be clear, we're not developing 1,000 sites. We've got 1,000 parcels that we think are appropriate for, kind of, these medium-sized hubs. What's happened over the last year, since we bought CoreSite, it's opened up a whole new set of customers to talk with and to understand what their needs are, and particularly looking at the cloud enterprise and large enterprise side, what they think the future holds for them in terms of edge compute. And there are a lot of great conversations about what that's gonna evolve into and how it's gonna evolve. We're still early days on this, so, you know, there's no kind of breakthrough to report. You know, we're not forecasting any revenue jumps on that, and we're not forecasting a lot of capital spend on that.
Right now, we're in the experimentation phase. We do have one facility that's kind of a sandbox to play in, that we're developing, that we're gonna break ground later this year. It's a one-megawatt facility in a tower site, and there's a lot of interest from those partners and kind of coming into that facility and saying, "Let's see what's possible here." You know, we believe the edge will develop. Everyone that we're talking with believes the edge will develop. It's a question of when and exactly what that looks like. And, you know, we've got a seat at the table with some of the most influential players out there. We're working on proof of concept with a number of kind of systems integrators and large, you know, carriers and clouds.
Nothing to report publicly yet, but the private conversations and some of those projects are going really well. So we're excited about the future on that. But again, it's not, it's not tomorrow.
You mentioned the power constraints before in Virginia. Just curious, when you think that gets better, and what are you doing to prepare?
So we're just in a different, on a different transmission line, so we're just not subject to the same constraints that are there. And, I don't know exactly what's gonna happen. I think that you've got some different timelines that different people have put out there, but probably asking Dominion would be the best, best chance on that. But I think it's gonna be a while, and so I think in the meantime, we've got the available capacity and a little bit more runway on power than some of the other facilities do.
Can you just update us on the update on the capacity for the process?
So, well, what I will say is, fortunately, I run the US, so I really can't answer that. I'd just refer you back to what Rod said publicly, and that is that we are looking at strategic options in India, and until there's more to report on that, we really can't get into any specifics about the dynamics of the process.
Question. So the question was, competitive dynamics vis-à-vis, companies like Equinix-
Yes.
that focus on enterprise and retail.
Look, with every one of those deals, there's gonna be some sort of nuance to it, and a lot of times, it's contiguous available capacity is one of the drivers to it. Pricing is certainly a factor in it, but also who they need to interconnect to. And so it may be that we have the particular enterprises that that person wants to connect to in our facility, so you win it based on that. And then there are some design differences. Most of our facilities are bespoke facilities that we build, and we have raised floor platforms, and that offers a little bit of flexibility in the design that a slab build doesn't. That's not usually dispositive of the deal, but that can... Depending on what they're trying to deploy, that can be a factor.
So there's really a couple of different drivers of it. Oh, gosh, I don't, I don't know the exact number on that. What I would say is our the legacy facilities, so like things like One Wilshire, that's in a, a, an older building, those are on slab. But anything that we've built in our campus environment since then, so most of our Silicon Valley campus, most of our campus in Northern Virginia, those are all on raised floor platforms.
Anything, and this kind of cuts across both business segments, but anything from a macro perspective around labor availability, supply chain, that either affects your ability to deliver or your customer's ability and velocity to move in? Anything kind of different now than earlier in the year?
It's getting better. It's still not fixed. The supply chain in the data center business is still constrained. Labor hasn't been a big issue for us, although I'm glad I say that, but, you know, some of the components have a long lead time. I think what's different is that we've adjusted, so we're, you know, typically, CoreSite and other companies were ordering just in time for delivery, and now we're actually stocking some inventory, and we're placing orders for long lead time items like generators, in anticipation of future builds in future years, because those timelines have been elongated. But it is getting a little bit better on some of the components that were holding things back before.
We are out of time. Appreciate your spending a half hour with us.
Thanks, Jon. Appreciate it.