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

Okay. We will go ahead and get started. My name is Cameron McVvey. I cover communications infrastructure here at Morgan Stanley. And it's my honor to welcome Matt Mercier, CFO of Digital Realty. Welcome, Matt. Thank you. Great to be here. And before we get started, I'll read this. For important disclosures, please read our Morgan Stanley research disclosure website. If you have any questions, please reach out to your Morgan Stanley sales representative.

Okay. And we will get started. So Matt, Digital Realty reported strong fourth quarter and full year results in February. There was a record 0 to 1 megawatt bookings and a near-record backlog. I guess as you think about the past year and the year ahead, what would you identify as the 1 or 2 main drivers behind recent results? And what are your key priorities for 2026?

Yes. I mean -- so first, yes, so thanks for the setup. I mean we came into '25 with roughly guiding to a little bit over 5% bottom line core FFO growth. We ended up the year basically almost 300 basis points above that. And I would say the setup for '25 is very similar to kind of where we're setting up for '26 and partly why you're seeing a congruence in terms of where we started our initial guidance close to 8% for 2026. And the key drivers, I would say, are: one, continuing our momentum on our 0 to 1 megawatt interconnection business, which I'm sure we'll dive into more. We've seen great growth and execution across that segment for a number of reasons in terms of our -- the number of markets that we're in, our connectivity profile and capabilities and product set. And as a result, we grew the signings as an example, there over 30%, interconnection bookings over 20%. So we see that as a very stable, sticky connectivity-rich business that we're seeing overall demand increase from an enterprise perspective. And we see long-term tailwinds in terms of not only continued digital transformation from that same enterprise base, but also, I think, near- and long-term scaling for AI, in particular, inference, which from our standpoint, we're still in quite the early innings. So I think that's a foundational layer of our growth. In addition, we also did well on signings that we did in late '24 and through '25, we had over $1 billion in overall bookings at our 100% share. That led to a very healthy backlog of over $600 million in revenue that we've got expected to commence in '26. So again, kind of derisking the majority of our revenue profile for '26 and therefore, supporting that overall bottom line growth per share. And then I think all those things, largely the favorable supply-demand dynamics are proving to be, call it, a pricing tailwind, favorable pricing dynamics throughout both of those segments, leading to improved mark-to-market story, which is benefiting our same-store portfolio. So I think all those things together are what's giving us conviction for continued bottom line growth that we put up for '26, but also continuing that in '27 and beyond. As we've said, that's a clear and key focus for us in sustaining that momentum on bottom line growth.

Great. That's helpful. And just a follow-up on the '26 guide with -- it's about 8% growth at the midpoint on a core FFO per share basis. When you think through the key building blocks here, what do you think could drive you to the high end of this guide?

Sure. I mean it's a couple of those components that we just talked about. So I'd say the 2 that probably have the most near-term, call it, revenue through to bottom line impact are: one, outperforming on our 0 to 1 megawatt and interconnection business. So those signings generally have a shorter sign to commence lag generally in like the 1 to 3 months, right? So the more that we can outperform in that, that should accrue to further in-year revenue growth as well as longer-term growth, but has a near-term impact. And then I would say two would be outperformance on, call it, mark-to-market renewals, and that could be both within our 0 to 1 as well as our greater than a megawatt category. As again, we're -- I would say we're in a favorable supply-demand dynamic where pricing continues to be in our favor. There's broad-based constraints to bringing on capacity. So that should continue for an extended period of time. And I would say those are probably some of the 2 near-term items that would be levers for us to outperform within at least the current year.

Got it. That's great. Now I wanted to ask about the power-based occupancy. It was guided to improve 50 to 100 basis points at the end of the year. So I think around 89% or so. I guess what's driving this transition to reporting power-based metrics? And how should investors interpret occupancy progression given the shift?

Sure. Yes. So for those who maybe -- I'll add a little bit more color for those who may not have heard us talk about this on the last call. So for almost through our history, we've reported occupancy more on what you would call more of a real estate-oriented basis, which is on square footage. But if you look at a lot of the other metrics and key metrics that we report on, whether that be leasing, development capacity, like almost all of our other key metrics are power oriented based on per kW generally metrics. So we just thought it was time given how much power is a part of the overall dynamic and the overall marketing messaging around our space to align all those things together and really make some consistency across those key metrics, which are pricing, signings, development deliveries and therefore, utilization/-occupancy. So that was the main impetus behind that change to create more consistency and uniformity in how we look at our business and what really drives our business. In terms of what that -- the relative change in occupancy, which is kind of why we've guided towards more of this relative versus absolute within our occupancy guidance is really not any different. I mean part of the -- the other part of the reason for that change is that we've been seeing increasing densities within our space, right? So you can have the same amount of power or more power within the same amount of space. So again, -- and creating that consistency within those metrics is, I think, is what we talked to not only our investors about that they thought largely would be helpful. So -- but that doesn't really -- the relative change, like we guided last year toward basically a similar 50 to 100 basis points occupancy. We were basically right in the middle of that. We're guiding to another 50 to 100 on top of that this year. That relative change is consistent whether you're measuring it based on power or square footage.

Great. I wanted to shift a bit to AI inference. And in the past, you described the 0 to 1 megawatt plus interconnection business as the most strategic priority. The last quarter set a new record in terms of bookings. And so what's driving the momentum? And how do you see the segment evolving with both AI inference and edge case workloads?

Yes. So I mean, we're -- again, to set the stage a little bit. I mean, we're seeing AI come into play. I mean we -- part of our strategy is that we're able to satisfy a broader spectrum of workloads all the way from multi-megawatt hyperscale up to giga scale type deployments down to single cent cabinets through our global portfolio. Over the last 2 years, we've seen -- we started to talk about more about kind of AI, AI-related workloads in terms of signings as a percentage of our bookings. For the last 2 years, more of that was oriented towards our greater than a megawatt, where we saw training come into play more early on, more prevalent. So we've talked about, on average, we've seen roughly 50% of our bookings, again, going back the last, call it, 1 to 2 years within that greater than a megawatt category be within training. And so that's where we started to see more of that workload start early. Inference, we've seen pickup, but I would still say we're in the early stages of that. So as a comparative, last year, we talked about within our 0 to 1 megawatt in interconnection, which is more of that kind of retail colocation enterprise type workload, which more people associate kind of that where that inference might land. Last year, meaning -- I mean, 2024, we had maybe mid-single digits in terms of like the percentage of those bookings that were AI-oriented. And we categorize that based on discussions with customers, types of chips they're deploying, the level of density they need. And we saw that -- we did see that increase. So last year, 2025, we saw roughly 20% of those -- of the bookings within that category were driven by AI had some level of inference workload. So a minority, but a growing percentage of that share. And I think we'll see that continue to pick up into '26. It still feels early from our side. We're still seeing where they're looking for -- enterprise customers are still utilizing the large language models from the larger hyperscale and companies that are really focused on that technology. And where we're seeing more of the inferences in financial services, health care, some level of content. So companies that are, generally speaking, look to get ahead of the curve and have the resources and the wherewithal to be able to deploy AI specifically within their stack, whether that be for their own customer use cases or internally for trading algorithms or other needs. So I think it's an early -- we're in the early innings, I think, of what we see as AI inference, but we see the potential as that continues to roll out as we expect to see more enterprises adopt some level of what we saw and how we saw cloud rollout over its many years, where it started really with the same customer base. The hyperscalers are driving the majority of the cloud in the early days. You then had enterprises that started to adopt their own private cloud, and now you're kind of in a position where hybrid is like the preferred architecture. We see that same thing rolling out in some level for AI, where you've got same large hyperscalers and some new companies driving the large language models, the training, which has some level of inference embedded within it. You're seeing a very small cohort of enterprises that are driving their own enterprise-level AI and inferencing. And we think over time, you're going to see private AI and then some level of convergence into a hybrid AI rollout, which is why we're setting ourselves up in a portfolio that can satisfy both those large training deployments all the way down to more inference -- smaller inference workloads, whether that be 100 kW, 800 kW, 4 megs or 50 megs.

Got it. That's great. And I guess with this broadening enterprise AI adoption, how important really is latency? How often does it come up within customer conversations? And how do you think you're positioned in a latency-sensitive world?

For -- so where we're seeing more clear workloads in the enterprise -- and I should say that, that also is -- we're not only seeing inference happen from enterprise-oriented customers, but the same companies, hyperscale companies that are taking hundreds of megawatts on -- that are training, they're also taking similar to how they do with cloud, smaller deployments within our 0 to 1 segment more in our highly connected facilities for inference-related workloads. So I think the theme is that latency continues to be an important factor for not only enterprise deployment of cloud and AI, but also for hyperscale deployment. So they still feel that it's important to be, for the most part, as close to the eyeballs as possible, so major core markets, as close to as many networks as possible so they can transact and transmit across a number of telecom providers and also being as close to as many different customers as possible. So curating these sites and campuses that have multiple customers that they can then connect with. So that all comes back to why we've been focused on creating a portfolio that is highly interconnected across 50-plus major metropolitan markets. adding capabilities within our interconnection ecosystem through Service Fabric, bringing partners onto that fabric to enable further differentiation as well as value to our end customers because ultimately, we believe whether it's for cloud workloads or AI-oriented workloads that are still relatively new, especially within the inference, but expected to grow that we can satisfy that and we can bring value to our customers over time.

Great. Okay. And on that point, I wanted to touch on liquid cooling. And just curious if you could provide an update on the rollout of liquid cooling solutions, both in new and existing capacities. And I guess, how important is liquid cooling from an AI inference basis?

Yes. So right now, I would say the majority of liquid cooling needs that we're seeing is more targeted to our newer deployments. So -- and again, you would say most of that is probably more oriented towards training, but there's likely some level of inference happening in those deployments as well. So we're seeing more liquid cooling, 50 to 150 kW type rack levels that we can satisfy, particularly in our new sites and new developments. We're seeing some level of liquid being requested in some of our existing facilities, which we have the ability to handle and be able to retrofit for. So we've done high-performance compute type deployments within our existing facility. We have the ability to do that broadly across roughly 30 markets that we have today. We've done it today in, I think, 14 of our sites. I think it helps that we've had sites that were built when water cooled chillers were still in vogue. So we're able to tap into that. But I think also there's a level of workload today that's still able to be satisfied with air-cooled air-cooled technologies, particularly at least what we're seeing on the inference side, where the densities haven't quite reached that, call it, 50 kW average that we're currently seeing from a -- or 50 kW and above training. I think even hear Equamax talk about it's in the 10 kW, that's all relatively manageable within an air-cooled environment.

Got it. Okay. On sovereign AI, it seems like that has been recently highlighted as a growing theme. How significant is it an opportunity to capture the government demand globally?

Yes. I think we see that, again, as a growing area and need, especially internationally, and we've actually captured some of that today. Where we see most of that is it typically doesn't come directly from government. They're typically working through, in some cases, a hyperscaler or an intermediary. So we -- that's where we see most of it through a partnership that governments have with some of the -- even the larger hyperscale or other partnerships. So I think the benefit that we have is take our EMEA portfolio. We're in -- not only are we in all the flat markets, but we have a presence through the majority of the countries through -- within the take that through to APAC, where we have a strong foothold in Singapore. We have capacity in Tokyo, Osaka, Seoul, Australia. So we have an ability to satisfy workloads at a country level, which is where you're starting to see that more sovereign level being requested. So I think our ability to be able to have capacity available in multiple global markets and the ability to partner with a number of different hyperscale as well as sovereign, whether that be cloud or sovereign AI needs, puts us in a position to be able to satisfy those demands across our global portfolio.

That's great. I wanted to ask on pricing. And when you think across your customer demographic, how do you think about the price elasticity across the customer base? And then secondly, how do market rates currently look for when leases come up for renewal?

Yes. I mean it's -- look, the pricing is largely a supply-demand typically equation through most of our core markets. So right now, we're in a phase in a cycle where that is -- the demand is high, supply is constrained, and we have a favorable pricing environment at the moment. We've seen -- we've largely seen -- when you look at it from a couple of different angles, like our mark-to-markets continue to improve. So that's -- we talked about this a little bit, I think, earlier in terms of last year, we had -- a little over 6% mark-to-market on our global portfolio. That's a mix and a weighting between our 0 to 1, which is typically more inflationary, 3% to 4% because the contracts are more 3 to -- I'm sorry, 2 to 3 years in term. So they're always usually close to market. So they're escalating more like an inflationary type rate even when they come up for renewal. And then where we're seeing sort of the higher mark-to-market is in our greater than a megawatt portfolio, which was north of 10% last year and essentially expecting a very similar type of setup, like I mentioned before, coming into 2026 in terms of the guidance that we set forth. Now I think as you look at sort of broader pricing dynamics across our global portfolio, I think you're seeing, call it, spillover effects. 1 to 2 years ago, you saw significant increases in pricing in Northern Virginia as an example, as that was one of the first markets that saw a very heightened level of demand versus basically a supply constraint that happened almost overnight due to power transmission issues. And now we're starting to see that spill over, I think, into a number of our other core U.S. markets where we're seeing mid- to high single-digit type annual increases in the U.S. markets. Go over to Europe, and you're seeing roughly kind of mid-single-digit type increases is not quite the same heightened level of demand versus supply challenge, but still healthy demand, constrained supply. And then you move to APAC, like a market like Singapore that's highly constrained, very competitive in terms of mark-to-market pricing, still highest set of demand. So we see really good mark-to-markets within that market, in particular, in APAC as well as others. But Singapore, I think Singapore stands out. I would say the one other thing I'd add to that is as we look out, in particular, from a mark-to-market perspective and the opportunity at digital, like we see an improving mark-to-market profile as our expiring rates we're comparing against markets start to step down in '27 and '28, '29. So I think that gives us kind of continued confidence in being able to have pricing power given the supply-demand dynamics and having that as a lever to support our continued bottom line growth going forward, definitely.

Okay. On the international market theme, you've entered several new markets recently, Malaysia, Portugal, Israel, Indonesia all in the past year. Can you walk us through your decision framework for a new market entry and how these markets fit the global interconnection strategy?

Sure. Yes. So I would say the common themes that you've seen, and we even announced another acquisition this morning, site, a highly connected site in Bulgaria. But I think the common themes you see then all the ones you mentioned, Indonesia and Malaysia, I think there's a couple of highlights. One, in particular, we -- we've been looking to grow our presence in APAC. That's just as a region, finding sites capacity where we can grow there. APAC is roughly, call it, 10% of our revenue today. We'd like to continue to grow that region. We see a very healthy demand dynamic and want to expand that as an overall portion of our overall portfolio and customer customer in a number of markets. So that kind of hinges some of those new market entries. Two, continuing to build our overall connectivity platform globally. So all these sites have been largely some of the more connected sites within those markets, giving us a strong, I think, solid base for not only enterprise customers, but that path to continue as AI continues to roll out throughout the globe, having those connectivity points of presence and networks, we think, is going to be very important. And then third is we have expansion capacity that typically we've been able to procure as part of those acquisitions. So both being able to continue to drive our enterprise 0 to 1 signings growth, but also having potential capacity for larger-scale deployments within those regions to is creating a broader connected campus ecosystem within each of those markets. So that -- those are the kind of the key things that we look for in terms of expanding within new markets globally.

That's great. Okay. I wanted to hit on power. It's a growing theme that a lot of investors are interested in. How does Digital think about using grid connections in utility providers versus behind-the-meter power solutions, about fuel cells or turbines? Where do those bridging solutions fit into the general strategy?

Yes. I mean, so I think first and foremost, like our view, and I think -- and I believe this is a broader industry view, and maybe not everybody, but from who we speak to, I think it's a general consensus, if you will, that at the end of the day, long-term utility grid power is the best primary source. So that's, again, from our standpoint, from customers we talk to, other industry participants, developers, operators. But as we've seen, we're in the grid is -- the grid hasn't necessarily kept up with the needs. So you mean ourselves as well as our customers and other developers, we're looking at ways to bring and bridge ultimately the power capacity divide that exists today. But we see that, again, more as a bridge versus a permanent solution. And we are looking at a number of our larger markets where it makes sense because of the investment that we need to have, the regulatory approvals, the partnerships we need in order to bring a site online that and bring on a developer and a partner that understands and knows how to do this since it's not our core business. We do see a need to try to bridge that capacity in the near term and put us in a position to be able to continue to bring on capacity in some of these more highly constrained markets like a Northern Virginia or Dallas in the U.S. So we are -- we as well as others are looking at bridging capacity, whether that be gas or fuel cells that we can bring to bear and be able to bridge and kind of rotate as power comes online to new -- to the next set of incremental sites. So that -- those are things that we are looking at. None of those solutions are easy. You still got to get -- it's another complexity within the overall development and procurement process, but it's a necessary thing given the power constraints we have in a number of markets in finding these bridging solutions to be able to bring capacity online sooner and be able to rotate that through to the next set of development sites that we have or able to bring online in those markets.

Got it. And on that point, there have been headlines around grid operators implementing some of these stricter requirements to access the grid for various data centers. this -- I guess, how do you view this? Is this a headwind or because you're scaled with access, more of a tailwind?

I mean I would say -- and I believe you're speaking more to like there's either financial or other commitments that grids are looking to just to maybe clarify for whether that's take-or-pay, deposits, guarantees, the -- because there's a lot of operators out there who have -- and developers who have come and said, "I need this much power and either not taking it, and that's created some of the concern, congestion. So I think from a digital standpoint, while I don't necessarily love it because it's incremental potential financial commitment that we have to put up, I think because this is one of many areas where an investment-grade balance sheet, I think, is a strategic advantage because, one, typically, that means we have to -- if we do have to post anything, we're able to negotiate a lower amount. If we do have to put up anything, typically, we can do it at a lower cost. And I think it just gives us some incremental leverage and bargaining power with major power suppliers, probably first and foremost, because we're -- we've already been doing business with a lot of the utility providers in most of the markets throughout -- in particular, in the U.S., but also throughout Europe and Asia, just given that we've been in business for over 20 years. So we've shown that we've not only -- we're procuring the power that we've asked for, we're utilizing it like we've said. So I think overall, it's a benefit for us, both from an operational perspective and already having established relationships with most of these utility providers. And two, I think, ultimately, from a financial perspective, given that we're an investment-grade counterparty versus some others who aren't able to put that up.

Great. I'm going to open it up to audience Q&A, but -- so think of a question. But before I do, I wanted to ask, on the earnings call, you called out a minor interest expense headwind related to newly raised debt. How should we think about market appetite for raising incremental capital, just both from a credit and equity market perspective?

Yes. I mean, so there's probably a couple of different angles on that one. But I think, one, we've, in essence, put to bed the headwind that we talked about. I mean it's still a headwind, but we had done that refinancing in late November, early December in terms of euro bonds that we had coming -- maturing in January. So we got ahead of that. Granted, still a headwind from an overall bottom line because we're taking out 2% debt and essentially refinancing it at a little over -- basically right around 4%. But that's in our guidance, already taken care of. So kind of check that one off the list, at least for 2026. I would say, in addition, this is another, I would say, advantage of being a public company. So we have access to multiple pockets of capital, both public bonds, hybrid securities, asset level financing. And we have great relationships with a number of banks in our bank group to help support us. And so I think from -- that's an advantage, I think, of being in the public sphere versus some of our maybe private competitors who are, in essence, probably overly reliant on asset level type financing and therefore, even more -- have to be even more diligent, if you will, on the type of customer and credit quality that they're able to finance in this environment where we have potentially a little bit more flexibility. Now don't get me wrong, that's not to mean we're very conscious about our -- the quality of customers we have. When you look at our top 20 customers, the majority of those are high credit quality customers. But I think it gives us really goes back to that we have a broader diversity of customer base and ability to finance across different types of products.

Great. Okay. We have a couple of minutes left, if there's any audience Q&A. If not, I did want to ask about hyperscalers if you think about this year, they've announced these massive CapEx plans, which creates some worry maybe around bottlenecks on power, construction, equipment. How do you manage supply chain risk? And what is elevated hyperscaler CapEx mean for pricing power?

So I think first, like when you -- probably something to note, like when you look at the what is almost half or more than half, like what is it, $600 billion-ish area, well over $0.5 trillion, which is crazy when you say it out loud that hyperscalers are spending. Not all of that is going into building data centers. Actually, a good majority of that is actually going into the CPUs, DPUs, servers and equipment that are actually going to move into a data center. So I think that's -- it's important to kind of distinguish what level and figuring out how much is in each category, that's a little bit of more art than science, but I think it's good to keep that in mind. And from a Digital Realty standpoint in terms of what that means, that just means -- I don't think it's any different really from how we've approached it historically, meaning we've lived through the pandemic. We've seen long lead equipment lead times elongate. We've gotten ahead of that. For a number of years, that's really not gotten any better. So we've been on top of it for a long time. We look ahead in terms of what we need to procure given the megawatts that we expect to deliver, working with our partners to strengthen that -- those supply chains. So that's something we've been doing for a number of years, something we'll continue to do as we deliver even more and more megawatts into our operating portfolio. And has that -- and I'm not sure exactly where you're going with the pricing part, but I would say we have seen increases in the cost of that long lead equipment over the last couple of years, but we've also seen a corresponding increase in the pricing that we're able to achieve on that deployed or related to the deployment of that capital and keeping and/or improving the overall yields that we're getting on our development pipeline.

Great. And we are at time. Matt, thank you so much. Yes. I appreciate it.

Thank you. Hopefully, that one Great. Good to see you.

Thanks again.