Digital Realty Trust, Inc. (DLR) Earnings Call Transcript & Summary

Mike Funk, the Head of Telecom and Infrastructure and Com Software Research here at the bank. Really happy to be part of the REIT conference again. So thank you to the REIT team for inviting us. Jeff back there in the back. I'm really happy to have Andy Power, CEO of Digital Realty with us. I think Digital is probably the first company I covered in this space about 13 years ago. So thank you again for coming out, Andy. We have Jordan Sadler, Head of Investor Relations as well. I'm going to kick off the Q&A. I'll leave about 5 minutes at the end for audience questions as well. So if you have any, you can hold them until then. So Andy, thank you again.

Thanks for having me. Appreciate it.

Absolutely. I wanted to kick off, Andy. So I relaunched in this sector earlier this year and what struck me was something you highlighted that you'd rather have a 7% growth for 10 years rather than 2% growth -- sorry, 10% growth for 2 years, right? And it certainly speaks to your philosophy. And I think the current market environment as well where there are so many big hyperscale deals available and a company to choose to chase them or not chase those deals. So how does that philosophy impact how you look at the market, the deals that you pursue and the deals that you're passing on?

Thanks, Mike. So just to clarify maybe I think we'd always want more and for longer as a common statement.

That's fair enough.

Just -- I think that where your question was going was something that I've said is I'd rather -- coming this year, we've turned the corner on our bottom line and have now, call it, put up a solid guidance at the beginning of the year that we've now raised throughout the year in addition to beating and on trajectory to repeat that movie into next year. And the question that's come to me is how high could it go? What's next? And my euphemism back was I'd rather be compounding our bottom line, which we view as very important to our long-term strategy and our cost of capital at 7% for 7 years versus just having 2 years at 10%. And that was in the context of, there's levers that we can pull in our business beyond execution in the broader market that go to the risk we have on our balance sheet and how we fund our business model. And that goes back to we've evolved our funding strategy, first from, call it, one-off joint ventures now to our first inaugural hyperscale fund. How much of development projects we keep on our balance sheet versus share with private capital has a near-term dilutive impact, but a longer-term extension of that runway, more projects that good returns further down the runway. And two, how much of the recycling of those hyperscale projects we move into private vehicles and lose NOI and FFO in large quantities when you do $1.5 billion or $2 billion joint ventures like we've recently done. So those are the levers I'd say, I think you were highlighting that I was referring to is becoming a consistent compounder at the bottom line per share growth and having that runway for growth supported by our backlog and our execution, but also the arrows in our quiver or levers to pull for our funding model.

That's a great answer there, Andy. And right or wrong, the market has gone very myopically focused on large AI hyperscale deals in the last couple of years. And you reported record numbers last third quarter, right, Jordan, for AI-related deals, even break that out for us. Thank you. But recently we see more of those deals have been go to markets that I would describe more as third tier, not primary, secondary data center market. So what are you seeing in the current market in terms of demand from those hyperscale customers? And do you think that demand comes back to the markets where you choose to compete where you can actually get a good return on your capital?

So the most important thing, I think, about our strategy is we believe -- we don't believe all data centers or all data center strategies are going to be created equal here. And we focus on markets that not only have robust demand, but diverse demand. They have workloads that are locationally or latency sensitive. And ultimately, many of these markets have some form of barrier to supply. Hence, we, for those hyperscale customers that you were referring to, which is in addition to our enterprise business which we love to talk about as well, gives us a value proposition to those customers and allows us to generate outside returns. That -- we've intercepted AI demand in those core markets while that demand had in the training phase did not have to go there. And the customers' feedback to us is why they choose us, why they chose these markets is the fungibility of the workload. If they get their AI demand wrong, they have other real business use cases, cloud computing that they can put in that capacity blocks. The cloud does not live anywhere or everywhere, I should say. It's locationally sensitive. There's architectures with availability zones in major metropolitan areas with radius restrictions within that metro. Hence, a vacant data center in a, call it, hinterland market cannot be backfilled necessarily with cloud computing versus a vacant data center in Ashburn, Santa Clara, Frankfurt, Singapore, Tokyo has ability to serve cloud needs.

And Andy, you mentioned the enterprise business, the 0-1 business. You've been increasingly highlighting as an important metric for investors to follow. So how does a 0-1 business strategy fit into the growth algorithm that you mentioned earlier and balance your business better than a pure hyperscale strategy?

So our results in the 0-1 megawatt enterprise colocation interconnection category are not an overnight phenomenon. It took massive investments over many years, spreading our business across now 50 metropolitan areas supporting 5,000-plus customers on 6 continents, the full product suite, investments in our go-to-market, investments in our business processes. And it also took a strategic pivot to make that the priority and the focus of the company. And we've now, in the last several quarters, started to see the fruits being born whereby we've been, in the $50-ish million per quarter or $200-ish million per year, call it inflecting 60s, 70s. And now we just had a record 2Q, which was $90 million, 18% higher than our prior record. And I could tell you why we value that is because we are offering tremendous value to those customers. Those customers are not building data centers on their own. Those customers are buying a platform with us with propensity to grow in more locations with us. Those customers are benefiting from our expertise in higher power densities and liquid cooling, and we can generate outside value for the customers and outside returns for our shareholders. And most satisfying piece of that is despite seeing the success come to fruition, it doesn't feel like we're anywhere near done where we could be generating that category. And I say that we're making significant investments in our systems, our business processes, our tooling for our sales teams, our penetration with partners and alliances. So we're not top ticking or anywhere near in that piece of our business today.

And I want to skip over to pricing, Andy. When I last covered these companies before moving to software for a few years, re-leasing spreads were viewed as a negative for Digital Realty. We're still working through some legacy contracts. They were largely flat to negative across your portfolio. And then more recent years, they've obviously ticked positive in a very strong way. Can you just walk us through your view on re-leasing spreads, how long they can remain positive in the range where they are and the contribution to growth?

So let's bucket again the business into the enterprise colo 0-1 megawatt for a second. We have inflected in that category, but to a lesser extent, but that's on the back of years and years of consistent inflation or inflation plus like growth in our mark-to-markets. We're now, call it, 4.5-ish percent mark-to-market in that category and signing new deals, uplifting pricing, normalizing pricing for our product set because we are a product of M&A and integration and adding more value to the customer and getting paid for that value from the customer. And I see that runway from a long, long way to go here because the addressable market is large. There is a long fragmented tail behind ourselves and one other competitor in the leading positions in the market. We are in many of those markets, neck and neck with our top competitor, but other markets, we're playing catch up. And we're maximizing that price volume equation to essentially make sure we're not sacrificing our velocity of volume at the detriment of price where we may not have the leading ecosystem in that market. On the bigger stuff, you've had legacy demands, digital transformation, cloud computing, now AI build upon each other, and it's come at a time when supply constraints have arisen to the greatest extent, quite honestly, in probably the history of data centers. And it's multifaceted, it's power generation, it's power transmission, it's supply chain for power equipment like substations, it's data center supply chain, it's use of water. And last but not least, it's governmental pressures, moratoriums and nimbyism in the space. Going back to, it's harder and harder to deliver that capacity, and hence, those things -- and also a mild inflation backdrop that's come through our space and our build costs have pushed rates to much healthier levels they are today from, call it, a certainly lower rate on the backs of a lower interest rate environment. I don't think this is going to last in a perfect state forever, and that's not how we operate our business, hence, my description of how we fund our business. But I believe, in many markets, this -- that are most important to our customers, this -- there will be a continuation of these themes in various shapes and forms. And how we operate in those markets, our history, our involvement deep and consistent with power companies, the community, all sorts of folks with a long-form approach -- long-dated approach is going to pay dividends to us because they're going to separate the people that came into data centers for a trade, broke some glass versus those who are here permanently. And I think we're -- again, our focus on these markets is insulating. One market may get some power relief, while another tightens up and vice versa.

And I want to come back to your comment a bit about the players that are under for the trade or a quick buck versus the longer-term durable operators. I'm going to table that for the moment because I want to continue to address what I think are, well, what are a lot of the questions I get from investors and concerns. Another one is the future proofing of data centers, right? So obviously, AI is more compute power intensive. We've been talking about power densities increasing significantly at least the next 5 or 10 years. So can you explain to us how your current facilities and then future facilities can address the rising demands for power, the greater need for cooling that those power densities bring because I hear from a lot of investors that, well, the legacy facilities just can't carry the next-generation workloads and require material capital investment to bring them up to spec.

So from my vantage point, power densification is a trend that's happened in our industry for a while and has been accelerating recently. At the same time, I'm not subscribing to this view that all compute GPUs are going to have to be at the most scientific experiments power density out there on the planet. And I think there will be -- we're in an environment when one particular provider is pushing the envelope of invention and power densities and cooling features, but you have a whole host of others that are building their own types of chips or trying to compete and are looking for efficiencies in the infrastructure from a power density standpoint. So I think that there's a world where numerous types of workloads from network to compute towards GPU AI today in training, ML, ultimately inference, incorporating private data sets will be in various private forms of power densities. We're a 20-year-old company. Every investment we made over those 20 years wasn't the perfect investment, and we acknowledge it. And we pruned our portfolio, sold outright 100% billions of dollars of data centers over the years. We exited numerous markets along the way that didn't have the attributes of where we're focusing today. What we found in the infrastructure arena, put aside the, call it, legacy telco hotels that may not support an AI cluster at massive power tens of liquid cooling but are running the Internet of New York City or Chicago or the Southeast, put those aside and have other uses, our campuses with larger format builds, contiguous capacity in terms of area, substations on site, runway for growth, that infrastructure, we found that we can -- of our own volition, go and densify the power. And we've done that for years in different shapes and forms. We've been doing liquid cooling for customers in the quantitative trading segment for at least 7 years, well before AI was all over the press. And every 2.5 years, we probably brought more power into that same exact suite for that customer, right? So I think we have a track record of that. We've done that on data centers that have been 15 to 20 years old. We've done that on brand-new data centers when they want to change the mix of air versus liquid on the fly. So I think we've got the track record and experience to do that. I don't think there's going to be a world where we're going to have the luxury to just say all the existing data center stock, no good. We can only move into new data centers. I just don't -- from the -- the data center is one piece of the broader technology, telecommunications, IT infrastructure. Remember, you got the servers and the compute, but then you have the distribution of the fiber optic networks and actually the consumption that goes all the way to your various devices today, in the past and we really consume this on in the future.

And you touched on power as well. I mean there are a few key parts to a data center and probably the most important is availability of power, right? We know it's very difficult today to procure more power. So you can talk about the advantage Digital has with the contracted power in some of your key markets like Northern Virginia.

We're very fortunate that we made some, call it, very forward big bets well before this inflection in demand happened. I was just in the D.C. area recently. And when we put together next to the Dulles Airport, sounded like a crazy bet in 2018, call it, spending $250 million on dirt next to the airport is now turning into some of the most precious capacity in that market. So being there early, being there consistently, being a good partner to all the partners that are delivering the capacity has been key. Northern Virginia is our largest market. We're very large in that market, and our runway for growth in that market is -- could more than double what we have today. But we're doing that same playbook in the Chicago market, Dallas, in Atlanta, in Frankfurt, Amsterdam, London, Paris, Asia Pacific, South America and South Africa.

And can we move for a minute to also managing growth? You mentioned before kind of selling assets, but you also have the off-balance sheet arrangements. You have the JVs, which you also used. So how do you utilize those to also manage the growth of Digital Realty? Or is it even a component to managing growth is a better financial structure for you?

So we made a decision that based on the fact that when it came to hyperscale, the opportunity was large and only getting larger, more capital intensive, more long term. Every piece of it was just getting bigger. And it was going to be bigger in multiple stages, the development stage, upon stabilization stage that we needed to evolve our capitalization and funding -- of funding of the company. Having relied mostly on the public company track record of raising public equity, we need more, call it, levers to pull for our business. We started with some great joint venture partners. This year had just recently announced our inaugural data center fund for hyperscale in the U.S., north of $3 billion upsized, oversubscribed. That was a vehicle that tactically, we seeded $1.5 billion of stabilized assets at an attractive valuation of call it, high 5 caps, but also seeded with some great development opportunities. So -- and that was -- we know they did this. We shared opportunities that we had on our balance sheet before the vehicle. I look at that as a massive milestone on what we're building in strategic private capital. And that strategic private capital that can be a funding mechanism for our own initiatives or also to scale other businesses or assets in hyperscale that we think are going to come to need natural homes. And we think there's an avenue like our LP saw of a business like Digital that is solely dedicated to data centers, skin in the game financially with these vehicles, through owner-operator experience and not a commingled vehicle. We're not investing in data centers plus office, multifamily, et cetera. We're not investing in data centers plus bridges and roads and airports. It's solely the swim lane that is our expertise. And we think that we have a good opportunity to scale that, that I think is going to help continue to allow us to support our customers' growth and grow our bottom line in an attractive fashion.

And you mentioned earlier, and I sort of want to come back to it. There are some companies that are basically this mantra of move fast and break things. It may not be long-term competitors. It's simply there for the trade in the space. But I wanted to separate those from some of the private companies that are more established, and we're seeing much higher levels of leverage at some of these companies. I'm seeing 10, 15x leverage. We're also seeing higher private market valuations in a lot of instances than public. So 2 questions here, Andy. Does your lower public market leverage level, does that disadvantage you at all when you're bidding on these deals? And then second, is there a case to be made for identifying and selling more stabilized assets into the private market and creating some financial alchemy of the valuation disconnect between public and private?

So one, and I think this is more germane to the hyperscale piece of the business.

Sorry, yes, it is. I should be clear on that.

The -- going back to our strategy of not judging ourselves on market share for hyperscale, we -- like we do in enterprise colo, we pick our spots. We're in 50 metropolitan areas; 30 of them, we have something we can add a lot of value, we can generate outsized return. That value could be -- they've installed a tremendous amount of capacity with us. They've got great operational relationship. We've got a runway for growth that no one else has. It's a tougher place to do business. They need to connect on our campus to somebody else that's a customer, whatever it is, that's our angle and how we play. In that arena, we don't just look at the minimum -- it's not just about the cost of capital of the commodity competitor. We're looking for alpha in those returns. And you can see that in our development schedule. I believe we're probably generating returns better than the average private capital that's taken the cost of debt and the cost of equity and what we're doing when it comes to hyperscale. It's intentional. And that sacrificing volume. We're saying in this market, we can't do that. We're not going to be in that market.

But focus to be a good steward of capital, to your point.

Yes. And now even with that playbook, we federate capabilities. There are certain markets where we bring alongside private capital. When we went to Latin America, we brought a financial partner with us. When we went to Japan, we came together with Mitsubishi Corporation to create MC Digital Realty. In the last couple of years, we did financial joint ventures, including development joint ventures. So we try to, especially with the fund taking it to the next level, have the best of both worlds of have the benefits of the public company, the access to capital, all the benefits of being a public company gives to our team members, our shareholders, our partners and customers, but also have capital for hyperscale that is the same level playing field as anyone else in private capital when we use that. So -- and that example of leverage, our fund is not high levered, but it's a higher leverage attachment than digital.

Makes sense. Sorry, Sarah, please go ahead.

So [indiscernible] but [indiscernible] funds which is stabilized assets capital come in?

I think we would have been successful based on -- albeit I'm biased -- based on the portfolio we put together, tremendous diversity. This is a hard asset class to get diversity in because the assets are so big. But I would say when we approached moving from joint ventures and one-off vehicles to a fund, which I think is better for the investors, better for our platform and ultimately our shareholders, I think we said, you know what, we could spend a longer time going down the road that you have described to get successful or we could bring something that solves a problem because we needed to fund some of our development, have some stabilized assets and we can really start with a great foundation and build our brand when it comes to private capital, and then build upon that with other vehicles more akin to what you described.

[indiscernible]

I think that's -- to me, that's not just a data center phenomenon. The overall robustness of desire for capital moving towards core, core plus types of vehicles relative to, quite honestly, investments in certain segments of real estate has left that less desirable than going towards the development. That's a product of risk, demand drop, demand supply, interest rates. I don't think that is a permanent fixture. And I think we, at Digital, have a role to nurture that not necessarily 100% with this vehicle, but incremental vehicles as we scale private capital.

Certainly to Mike's point about disadvantage versus private capital. The disadvantage really more not of returns, but not growth or really returns to equity holders because ultimately, their leverage just get some higher returns to equity holders, [ not ] -- they don't necessarily underprice you because their cost of capital is necessarily better or worse than [indiscernible] leverage, right?

I think in this current environment, we got some very -- a landscape of very sophisticated financial private equity firms backing platform in the space, many of which we partner with. They are not tone deaf to the supply-demand dynamics, right? They are not necessarily cutting to their cost of capital minimums. I think the nuance of you could say, difference or potential disadvantage of solely being in the public markets going after hyperscale is something we lived in. The cycle of development, you spend money in the public [ wrapper, ] it doesn't earn any earnings for a while, especially as projects take longer digestion periods. Land is becoming a bigger portion of that. And then getting that algorithm in a development-fueled growth model, I think, is challenging. Hence, we said we need to tweak our model in many shapes. We need to make that colo interconnect the engine that led the business. We need to fund, when it comes to hyperscale, with private capital partnerships in order to get that growth algorithm that I don't think a private -- private capital aren't looking at their per share growth algorithms.

But their growth is higher. The growth can be higher, the returns aren't. Overall return to the total invested capital is [ high, ] but the growth -- the [ 2 points ] of growth for you [indiscernible] higher leverage and higher growth.

So said otherwise, we have a development partnership with a little famous group over on Park Avenue. And they haven't ever gone -- we haven't gone to them and said, hey, we want to quote the customer, and say, no, cut the rate, go lower return, Mike. They understand -- they're economic animals.

And on the enterprise business, I mean how long of a tail of growth do we have for the 0-1 megawatt? Is that a 5-year tail, a 10-year tail? I mean the question used to be the incremental migration of enterprise into cloud. That's kind of 100 basis points a year accelerated. But how much more do we have in that growth tail do you think?

No one's been focusing on this, quite honestly.

Should we be? Or is that a bad question?

I focus on it. Our company is focused on it. And when you focus on it and you see the results inflect on it, I think it's going to be paying less dividends from making that decision. But I'd say competitively, the landscape has not been a focus. Part of that is -- we call it -- there's been a lot of consolidation in that space. It's a harder road to build upon. I mean it's been a good 10 years at Digital, making this pivot to just getting the results we have today. You look at the addressable market in that it is very, very large. It's still growing at a healthy clip. You look at the share ourselves and Equinix have in that market, it's still a relatively small share of that market. There's a long tail of competitors you probably never even heard of from net telcos who are still quasi in the space to one-off names. You still have an on-prem piece. You still have cloud adoption. And with cloud adoption has multi-cloud and hybrid cloud adoption. And then you still have AI to filter into that space, right? And I think AI is going to be different in terms of enterprises adoption of it. I mean you could elect -- I don't think Bank of America does this, but Bank of America could elect to put their data center in this building if they wanted or their old office buildings.

We elect not to.

Yes. Not a great decision, but it's doable. If you were going to put in 150 watts per cabinet or even higher in liquid cooling, I would say that, this is not a great building to do this in, right? So I think that's a nuance that I think we will eventually see in terms of influencing enterprise adoption for its AI.

I have one more quick one, then I'll open it up. I do have 3 questions from the REIT team that we're asking consistently across meetings I want to get to before we close. And Jordan knows this. He knows this thing. So you kind of touched on it, Andy. How does the marriage of the enterprise business and the wholesale maybe create synergies or benefit Digital Realty? A lot of talk about inference developing. I think we're very early stage, probably even before first inning right now on that. But are there synergistic benefits as we make that transition? Or is that too much of a leap to say there'd be any kind of relationship there?

I think that at many times in the history of our company, we had decisions well predating me and even after my joining of which road to go down in terms of strategy, one road, the other, both. And I look at it as a few simple things. Whether it's hyperscale for cloud computing or AI, it's a large, attractive addressable market. Enterprise, you probably heard me sing the praises of, I think, of that business and what we're going to do in that business. Putting them together lets you fish in 2 large pools of opportunity. And we're doing it in different fashions, but it allows us to scale our infrastructure, scale our platforms, scale our capital. I think there are synergies there. When we're in front of a customer and we can have discussions about network nodes, on-ramps, what they're doing in AI, megawatts of compute, et cetera, in a holistic fashion, we're spreading our resources and solving more for the customer in a differentiated fashion. So -- and this is a virtuous cycle. All these -- I think we're not in the business of investing in island data centers, right? We're about numerous customers that connect on our campuses, connect to enterprises and service providers and build upon each other's technology and innovation to support their end customers at the end of the day. And that's why we focus where we focus and not go where we don't focus.

I [ do not ] think investors miss that as we all focus too much on the headlines coming about single deals, individual transactions. So it's a good point. I do want to get to the REIT team questions before we run out of time here. So let me go ahead and read these off to you. These are going to be multiple choice, Andy, so it should be simple.

My favorite questions.

You can't be wrong.

Just go with C.

Yes, just go with C all the way down. When the Fed starts to cut, do you expect borrowing rates for long-term debt to, a, decline; b, stay flat; or c, potentially rise?

I'm going to pass this over to Jordan yelling over here.

First go down and eventually go higher.

Okay. So what part of the curve are you playing? I don't know if that was an answer. I'll let Jeff interpret that one and try to put that appropriate.

What was that? A was go down.

It was decline, stay flat or...

Decline. Okay.

It was not a good standardized testing.

No, I was better at the essays.

Where did you go to college [indiscernible]?

I did -- I went to Digital Realty University.

Perfect. Last year, the majority of companies stated they're ramping up spending on AI initiatives. How would you characterize your plans over this year? And that would, of course, be internally for expense reduction or efficiency? Sorry, higher, flat or lower is the multiple choice.

It's going to be higher. I think we're very fortunate. We've been integrating our systems because we're a product in numerous mergers. And we had to do this. We had to remove the friction from our internal customers, our sellers, our partners and our internal customers. But it also, I think, will make us AI-ready faster and more effective when AI tooling comes.

Okay? Industry prediction. Do you believe same-store NOI for your sector will be higher, lower or the same next year?

Higher.

Higher. Perfect. You answered exactly the same as Equinix earlier, by the way. So that was very consistent. We have about 1 or 2 minutes left. Are there any quick questions from the audience? If not, we can let Andy and Jordan get the rest of their day back. So thank you all again for coming out. Really appreciate it.

Thanks, guys.

Take care.

Thank you.