Welcome to our next session. My name is Jonathan Atkin. I cover the communications infrastructure sector at RBC. With me for the next several minutes of Q&A is the Chief Financial Officer of Digital Realty Trust, Matt Mercier. Matt, welcome.
Thank you. Thank you, Jon. Appreciate being here.
We are going to cover a lot of topics and see how many we get through. Maybe starting with agentic AI and what you are seeing there. Sounds like there are some early demand signals. When do you start to see that show up in bookings, things like cross-connect, ServiceFabric, and when do we get from preseason to early innings to maybe the bulk of the demand?
Yeah. In terms of agentic, I would say we are still in the early innings, but I think we have seen, especially even over the last couple of quarters, definitely an uptick in what we are seeing around diverse deployments within our 0-1 MW that are taking up interconnection as well. Maybe to frame it, I think you are seeing demand is broad and diverse today across our product set. Which I am sure we will get into in terms of hyperscale, but also in our 0-1 MW , which I would say more the agentic, although some of those cases I think are starting to see greater than, or larger than 1 MW deployments. But we have seen an increase in demand across our enterprise and service provider segment, deploying diverse network-oriented deployments across multiple markets globally.
I think we're actually starting to see that pick up in terms of discussions we're having across our global portfolio. I think some of the ways to characterize that are, the last three quarters, we've set records in our 0- 1 MW business. That includes records within our interconnection as well, to your question. Just last quarter, I think was our record quarter in terms of interconnections. We're seeing that show up not only in cross-connects, but also in ServiceFabric demand. We still see a lot of opportunities. We add partners and capabilities across that product, but also in our, what we call our bulk fiber type product, which allows customers to be able to bring fiber across not only our campus, but also to other campuses, even outside, in some cases, the Digital Realty portfolio.
That's part of stitching together what has been more training related deployments that are outside that, bringing them into more of our interconnected assets and campuses, one of them being here in Chicago, but across our broader global portfolio as well.
The infrastructure requirement to support that growth, what does that look like versus, say, traditional colocation?
The one thing I think we've seen, our portfolio, I would call out two things, at least, that I've seen. One, I think you're seeing larger space and power requirements in some of these core markets needed to ultimately support inference and more agentic AI type workflows. So what would typically have been a standard average 300 kW or below deployment within our 0- 1 MW, which would be landing in some of our most interconnected assets across our portfolio. We're now seeing greater than 500 kW a megawatt. In some cases, 1- 5 MW is now starting to become more of a standard in terms of what we're seeing some of our enterprises as well as our service provider type customers take in order to support what is becoming a more diversified inference-oriented type architecture across our platform.
Two is back to what I was mentioning before. I think in addition to seeing an increase in cross-connects, we're also seeing an increase in bulk fiber as the bandwidth needs increase to be able to connect to training facilities into these more dense, interconnected carrier, telco-heavy facilities that we have in multiple markets across the globe.
More to the land pipeline risk. Is it just a broad topic area? There are many U.S. metros, and then globally, I think you probably see this as well, although it gets a little bit less attention from investors, but they've all tightened, or many have tightened data center permitting. Your current pipeline is not at risk, but you did flag concern about replenishing capacity in roughly the 2029- 2032 period. How many years of pipeline visibility do you have before these constraints materially affect delivery?
Yeah. Look, I think we've gotten ahead of the game in terms of what we've done, especially over the last several years, both from a capital as well as a land and power perspective, in order to put ourselves in a great position to extend what we see as a great opportunity set for several years to come. You mentioned we have maybe a set context too, as well. We have 3 GW of operating capacity today. Behind that, we've got 1.4 GW under development. So right there, that's a 50% increase in what the portfolio could look like within the next, call it two years. After that, we've got 7 GW of land capacity that we'll be able to bring online over the next, call it two to 5+ years in order to continue our pathway of growth across our global portfolio.
That 7 GW of land is in various stages of power procurement. There's probably almost 3.5- 4 GW that we have in hand today in terms of ESAs ready in hand. We're working through the rest of that. Again, that's already another doubling of our overall capacity that we could bring online over that, call it, two to five-year time horizon. Yes. I would say there's never been a better time to be in this industry, but it's never been probably harder in terms of bringing on power and related capacity.
But again, that's where we've been, I think, well ahead in bringing land capacity online, broadening our capital sources, and putting us in a great position to be able to continue our momentum of delivering what we've recently talked about, which is called 10% bottom-line growth, which we've done last year, we expect to do this year, and expect to continue that for several years to come.
So will the moratorium topic work its way into any of your releasing discussions and even indirectly affect things like renewal spreads? Do you expect that or not?
Probably the short answer is yes, in some regards. But look, I think there's a few things that are happening right now. One of which is just, there's just a broad supply-demand imbalance, right? And so we've seen pricing continue to pick up across the majority of our global markets. It started a few years ago with Virginia, which was kind of the initial foray. Although I would say last week I was in Europe, I was in Zurich and London. We did a couple of property tours, including some customers. And it actually reminded me, Europe has seen a level of constraints well ahead of what we're seeing in the U.S. Maybe different degrees, but Digital, we've been operating in 50+ markets. We've seen where constraints have come up over the last several years. Europe's one place, Singapore's another.
I think one thing that we've seen, which probably goes to where your point is, as demand continues to improve and supply becomes constrained, pricing is what tends to move. So we've seen that in terms of our new signings and our development yields, which have picked up over the last, call it, year to two years. We're seeing that in terms of renewal spreads. Just this last quarter, I think we set a record in terms of where our renewal spreads were, which were over 60%, are greater than a megawatt categories, and very healthy within our 0-1 MW .
On top of that, we're seeing an environment, at least in our portfolio, where over the next couple of years, our expiring rents are on a downward trajectory versus market that I expect will continue to increase as a result of where inflation is heading, where interest rates are going, and again, back to what has been a very favorable supply-demand backdrop that I think all brings itself to a better pricing environment.
Maybe talk about Neoclouds. I think all $1.4 billion of your signings in first half 2026 were IG-rated traditional hyperscalers. Now you've got IG project-level ratings, hyperscale credit backstops for Neoclouds, and how does that kind of factor into your underwriting framework?
Yeah. I would say we've probably taken a cautious approach in terms of our exposure to Neocloud. I'll say that from the perspective of excluding any sort of backstop arrangements. Look, I think that's partly been because of, we've also had an ability, given where our supply is and where demand is, we've been able to be choosy in terms of the customers that we bring on. We've been able to pick some of the more investment-grade, high credit quality customers to bring in our portfolio. We do have some Neocloud exposure.
We've actually seen more of that recently in smaller deployments where we've, somewhat back to your first question, where we've helped them connect some of their training workloads through some of our more interconnected facilities as part of a broadening and diversity of workloads that we're seeing across the Neocloud, but is also the overall hyperscale set. So that's where we've, more recently, I think, picked and choose in terms of where we think we can best satisfy their needs in terms of our portfolio, which is more smaller, more connected workloads in some of our more gateway-type facilities.
As you look at your capital plan going forward, what is the rough mix across region, and what are you looking for in terms of demand signals or maybe permitting constraints factor this, but how could that shift versus the current mix of APAC versus EMEA versus Americas?
Look, if you went back a couple of years, you would've seen the majority of our actual development pipeline was in Europe. That is now shifted to the U.S., just given the size and scale of AI and cloud deployments that we've seen across the U.S., where we continue to develop across our global portfolio, EMEA and APAC. I think we are seeing a dynamic where APAC as a region feels more welcoming to AI. Maybe not in all countries, but especially in contrast to maybe the U.S. at the moment. So I think we are looking to expand our presence there, which has been 10% of our portfolio. We've had a view. We continue to want to expand that. We've done that more recently, but more on our more highly connected assets.
We've bought some assets in Johor, in Malaysia, with some expansion capacity behind that, but on a relatively smaller scale to, again, what you are seeing in the U.S. We've continued to do deals across Japan as well. But a lot of these markets are becoming, again, more constrained. But I think we see great demand in APAC from across our product set, again, both scale, hyperscale, and more of our enterprise connectivity-oriented play. We are seeing more diverse workloads continue to pop up within EMEA. Again, on a smaller scale, last quarter, we set a record within our 0-1 MW within our broader EMEA region, and we are starting to see those workloads start to tick up within a number of the core markets across Europe. The U.S. as well, in terms of larger workloads continue to dominate the headlines.
But I think we've had a great mix of business across our hyperscale as well as our 0-1 MW in interconnectivity portfolio.
You compete globally against a number of private developers that operate much higher leverage than you do. Given what is going on around cost of capital, what are you seeing on, say, the competitive front, and how that might be reflected in discussions with your hyperscale customers?
Look, I think this goes back to almost probably two or three years ago. We set out on a plan. When I took over as Chief Financial Officer, we were seven times levered, and people thought that was high in public market context. We have set out on a plan over the last several years, I think, to do two things related to this. One, we brought our leverage down and we brought our growth profile up. Two, we broadened our capital sources. We started expanding our private capital business, started out with joint ventures and expanding that from a development perspective. More recently, we did our first closed-end fund last year, where we brought in over $3 billion of equity capital.
I think we saw what was a broadening and increasing scale in terms of capital need in order to support what has been an incredible opportunity set across, again, our global business. We now, I think, put ourselves in a position to be able to access a broadening set of capital that enables us to tap more recently across, I think, what has been the most attractive sort of cost of capital in a global set. Just this Monday, I was in the market. We did a Swiss bond offering, 2%, CHF 600 million of capital. Pretty attractive in terms of what you can do and in terms of what I think other private players can do. I think the bottom line is diversity, I think, is key as you start to enter environments here where interest rates are rising rapidly.
They are up over, depending on which one you look at, they are up over 100 basis points year to date, and a lot of that has happened, call it, in the last three months. I think we have a great ability to tap across the best pockets of capital to continue growth across our business. Ultimately, that, I think, is also going to accrue to, I think, hopefully a continued better pricing environment from a rents perspective as well.
Audience questions.
Do you see a trend towards project financing, like bank CBO, off-balance sheet financing for data center build that is anchored against customer contracts?
Yeah, I think the question was: Do we see project finance for data center builds? The short answer is yes. We haven't been as prolific on that front. Just given our public company, we've been able to issue bonds. As we start to look at and think about within, but we have done it within our JV, they call it more private capital vehicles. Yes, we have done that both through bank and bond project finance markets. Maybe this is adding more than you were ultimately looking for, but I would say that those mark-- Access to capital is still abundant and available, but I would say it's becoming more discerning, and maybe even more so with where interest rates are going. But particularly from, I think, an underlying customer set. I think the highest invent-- I almost view it as there's three.
It feels like there's almost becoming three tiers now. There's the high investment grade customers, AA or A and above. You can get deals done as long as your contracts are structured appropriately. You then have lower investment or BBB level. Some of those are getting harder and more discerning in order to get capital against them. Then you have, call it more Neocloud, no ratings. I think that's becoming much harder across, at least from what we see from a broader landscape perspective across the globe.
Other questions? Please.
Does the increased uncertainty in the regulatory kind of NIMBYism environment change your views on the right level of investment to make into the pipeline going forward?
Okay, so I was told to repeat the question, so that's why I'm doing it. Does the increased level of NIMBYism change our view on how much we look to build a pipeline? Look, I would say no in terms of what we're seeing today, right? The majority of the land in our developable capacity that we have is, broadly speaking, in major core markets across our global portfolio, right? Where we've had a long history of experience, we've seen the demand trends. We're having constant discussions with our customers, feel very good about the long-term potential for that capacity. I can't say that I would have maybe that view for other people's portfolios. I think they'll have to make that decision.
But I think in terms of our ability to access, I think demand across a multiple product sets and multiple customers, right, we're not beholden to 10 or 20 customers. We have now 6,000 across our portfolio, right? We're not in five markets. We're in over 50 markets across the globe. So I think diversity of demand profile, customer base, geographies, especially in this type of environment, I think should accrue to our benefit.
Thanks very much for your time.
Oh, thank you. Appreciate it.