Welcome to Citi's 2024 Global Property CEO Conference. I'm Nick Joseph, joined by Mike Rollins with Citi Research, and we are pleased to have with us Digital Realty CEO Andy Power. This session is for Citi clients only. If media or other individuals are on the line, please disconnect now. Disclosures are available on the webcast and at the AV desk. For those in the room or the webcast, you can go to LiveQA and enter code GPC24 to submit any questions if you do not wish to raise your hand. Andy, we'll turn it over to you to introduce the company and team. Provide any opening remarks, tell the audience the top reasons an investor should buy your stock today, and then we'll get into Q&A.
Thanks, Nick. Appreciate everyone coming here to see us today and dialing in. To my left is Matt Mercier, our CFO, and to my right is Jordan Sadler, Head of Public and Private Capital Relations. I'm Andy Power, President and CEO of Digital Realty. I think folks in this room probably know Digital Realty. We're the largest owner-operator of data center and interconnection infrastructure globally, supporting 5,000 customers across 50-plus metropolitan areas on 6 continents, supporting hybrid IT workloads for enterprise colocation customers to dedicated data halls and suites for hyperscale cloud and often AI customers as well. We've had quite a year since we saw you all last year, really advancing three priorities for the company. One, demonstrably strengthening our customer value proposition, which I think folks could see and will likely continue with record 500 new logos to our 5,000 customer base.
Pricing power in our less than a megawatt and greater than a megawatt categories. Accelerating fundamentals. We've been integrating and innovating, launching high power density solutions for our customers, adding to our Service Fabric. We've been diversifying, bolstering our balance sheet. We sit here today now positioned with acceleration in our fundamentals as well as our bottom line, which we believe will continue to keep getting better.
So think back for the last few years. Digital's been repositioning, optimizing, expanding. How do you look at the durability of the current asset portfolio, the sales organization, and the support functions to execute on your growth opportunities?
I would say we've worked very hard over many years to change the composition of this portfolio and this company into a real platform offering. You've heard supporting now 5,000 customers across the spectrum of capabilities with really incredible pricing power to our platform, multi-market customer expansions, and coveted locations that we view are irreplaceable. We've been adding in terms of geography and interconnection capabilities. Our Service Fabric had announcement with 70 new directory members and access to 200+ on-ramp destinations at the earlier this year. We've also been supporting the hyperscale and AI cloud providers with large capacity blocks. And I think the most recent change is we've changed the funding model. So these organic fundamentals, be it our pricing power, our cash mark-to-markets, our same-store growth, our EBITDA growth, falls to the bottom line over time.
Obviously, being able to garner fees off the capital we brought into the platform to support hyperscale deployments both on stabilized and development JVs. There's always more room to go. We're not resting on our laurels of what the work we did in 2023. There's certainly more to come. But I think we're certainly starting to see the fruits of that labor come to bear.
In reference to not resting, can you talk about today's joint venture announcement? Was this contemplated in the guidance? How does the return on Digital Realty's capital that you're investing shift from being a complete on-balance sheet 100% owner of the project to now being in this joint venture?
I'll turn to Matt in a second here to talk about the guidance implications. But if you go back a year's time, first off, what we did is we raised the bar of what good looks like in terms of capital deployment. And you could see that in our development returns of 400-ish MW now, 10% ROIs, and many projects even higher than that to the hundreds of basis points tuned. This joint venture we announced today with a long-term partner in Japan, Mitsubishi Corporation, is very similar to another transaction we did with a net lease REIT earlier last year. This was one of our lowest return development projects in our portfolio. Open book, build-to- suit to high credit quality customer, but call it high sixes % unlevered return on cost, simply a deal done in a different era.
It's initial cap project is $400 million, but the customer has takedown rates to grow it to close to $800 million. They're coming in for a 65%, obviously paying us various fees to the tune of, call it, 150-ish basis points increase to our returns. So again, moving off our balance sheet, the capital load for a project that is the lowest ROI I've ever heard and also allowing us to free up that capital to redeploy. Do you want to touch on how we included that in our guidance, Matt?
Yeah, sure. So in terms of guidance, this was one of the development joint ventures kind of following Blackstone, which we announced at the end of December, which was $700 million. And where this sits in terms of guidance, it's not part of our disposition line item, which was at the midpoint about $1.25 billion. Where we're counting this is where we talked about development, CapEx spend, net of partner contributions, which was a little over $2 billion in terms of where we guided for 2024. So this amount where we closed roughly we closed on roughly $200 million. There's, call it, in the future, they're going to be funding another $100+ million or so as we continue to fit out. This is part of those net partner contributions that are within our development spend for 2024.
So just ahead of getting into some of the business fundamentals and demand drivers, is there a way to conceptualize going forward how you approach this funding model? You mentioned a focus on capital net of partner contributions. How should investors think about what this model looks like on a go-forward basis to appreciate the sustainability of investing, maintaining your leverage targets, and also managing any dilution risk for the shareholders?
I would frame it as this. When it comes to enterprise interconnection-rich colocation, we strive to have 100% ownership of those capabilities where we see the greatest pricing power and the greatest long-term organic or same-store growth, with rare exceptions being in markets that have incremental, call it, developing market or currency risk. When it comes to hyperscale, we're essentially building a private capital platform around hyperscale. This has been in the works for several years now from our, call it, billion-dollar joint venture with one of the Singapore REITs back in 2019 to our creation of Digital Core REIT to the transactions we did last year. We've supplemented the flavors of that from seating it with stabilized joint ventures at 5 or 6 cap rates. We're developing to 10. That's significant merchant gains we're harvesting from that.
We've also supplemented that with a development version in order to, call it, become more balance sheet efficient. And again, these are not non-core to Digital. These are great customers, parts of our campuses where we can put our capital at great returns. But they're the slowest of our herd once created because the customer credit quality is so strong, the leases are so long, the escalations are the lowest of what we can garner, the pricing power is the least of what we can garner. Hence, tapping into private capital platform when it comes to hyperscale is essentially what we're building there. We're wrapping that in a funding model where we're very cognizant of accelerating our bottom line FFO per share growth to mid-single-digit plus type growth.
Hence, we need to know that we can't do too many joint ventures that dilute too much NOI too quickly on that path. But as the pricing power on the cash mark-to-markets, the same-store growth, and we're kind of through this deleveraging change we're going through, I think we're setting up for a sustainable bottom line growth for years to come.
Great. Moving maybe to the demand side of the equation then, can you frame the size of demand and the funnel relative to past years? How much of this is a function of what you're doing specifically in the company to deliver better results versus just a rising tide of the landscape?
I'd bifurcate them into two rough buckets. On the less than a megawatt interconnection side, which is the preponderance of the customer count, enterprise customers, network service providers, we've been growing our capabilities in that category. You've seen that in those bookings in the less than a megawatt interconnection. You've seen that as the lion's share of the new logos, 500 we added to the platform. We're looking for double-digit growth in that category on a quarter-over-quarter basis. We've been absorbing just in that category, call it, 500 basis points quarter-over-quarter, so really growing the net kilowatts or cab equivalents in that category. That growth has been strong for some time and continues to be strong.
It's built on the back of moving from on-prem to off-prem, hybrid IT, cloud consumption, and now AI use cases as an incremental use case, which we're seeing many of across numerous industry verticals: finance and trading, retail, healthcare, a whole host of verticals. On the larger capacity blocks, cloud, compute, and certainly AI, large language models, and ultimately inference, it has always been lumpy. It continues to be lumpy. But it certainly has had a stairstep up in terms of demand as an industry and certainly at digital. And I think we're well positioned to capture that. Those customers are seeking large contiguous capacity blocks with urgency. And they're doing it at a time where numerous markets are seeing supply constraints due to power transmission, generation, NIMBYism, sustainability concerns, supply chain concerns, supply chain on the component side, supply chain on the transformer side.
I think there's a whole host of elements where that inflection is obviously flowing through to our pricing power.
When you look at how much inventory you have, development, can you frame how much in megawatts Digital currently has available to market and sell to customers?
So on the less than 1 MW category, we have a sizable amount relative to the demand because you're always pre-building ahead of that. So I'm going to zoom away from that because I don't want to mix apples and oranges here for your comparison. Our total land bank, even net of our joint venture with Blackstone, which was only 20% of that land bank or capacity and when I say land bank, I'm calling shells, some built-out suites, and land capacity that can be brought in certain years north of 3 GW of capacity. The prime, call it, I need it now, I need it large type capacity blocks, which I think your question is kind of honing on, starts in the probably most sought-after market being north of Virginia, call it 100+ MW between two campuses in Loudoun, another 200 MW in Manassas.
We've got equivalent, call it, 50-ish to 100 blocks in Dallas, I think 50 in Santa Clara, Frankfurt and Paris, and Amsterdam and Marseille, call it in the 50 MW blocks. Seoul, excuse me, yeah. Seoul, we've got close to 50 or 60 MW blocks. Tokyo, Osaka, similar sizes. So there's several chunks out there. Those aren't all built out. They're not sitting vacant, obviously. But they're called prime suspects or targets for customers with urgency around large contiguous capacity blocks, certainly well-suited to AI.
That's really helpful. On the earnings call, you had mentioned that Digital Realty was in negotiation to lease up roughly 100 MW of your inventory in Northern Virginia. Can you provide any context of what you're seeing drive that demand? Are there any updates on how you're progressing with that?
Northern Virginia was a market that got severely disrupted through power transmission issues. Despite that, it took some time for the market to fully get absorbed. Time, I guess, is not exaggerated because it was months, not years. You've seen the rates in that market recover from the $70s to the $130s to the $140s to the $150s. I venture it will be going even higher, close to $160 or maybe slightly higher than that. Those capacity blocks we have in that market, we were fortunate. They were delivering late last year into this year. We had the luxury at the time to be waiting for the market to move our fashion. We were fortunate that that is what transpired. Now it's a time where there's very few options in that market.
There's urgency from customers with just general cloud compute, enterprise demand, certainly AI demand, all trying to tackle that same need. I don't have a new lease to report on your conference here. I would say there's a high conviction around those capacity blocks.
So you mentioned AI a few times. Just maybe zooming out for a moment, can you unpack the ways in which Digital Realty can benefit from this broadening of Gen AI demand directly and if there may be also some indirect opportunities to benefit in terms of the way customers will now need to deploy their workloads in the future?
I believe we are able to tackle the AI opportunity better than anyone else in this space, public or private, given the breadth of our capabilities and infrastructure here. The popular versions are the big capacity blocks that you've heard about. We just kind of ran through the major markets where the large language model training is going to happen. Ultimately, inference may be built on the back of that, again, in a backdrop where there's supply constraints. So these bigger is in prior years, bigger meant better terms for the customers. It's kind of been flipped on its head a little bit in today's era. So with 3 GW of capacity blocks, many of which we can activate pretty expeditiously, I think we're well positioned in, call it, strong ROIs, 10 ROIs or better, able to harvest large gains from those type of builds through JV partners.
You extend that to the enterprise use cases. We're live with enterprise use cases of AI this year. In 2023, we retrofitted close to three megawatts in one of our U.S. locations and went live with a liquid cooling solution. At the end of the year, I was touring with a customer in Paris with a live liquid cooling, which meant we were helping that customer years before to get ready. Quite frankly, our CTO was talking about this AI trend seven, eight years ago. So we've been doing this for a long time. And the breadth of our infrastructure, which didn't come from just cage and cabinet only but came from hyperscale and diversity of power densities and infrastructure capabilities, suits us very well in an environment where, quite honestly, the form factors are just getting bigger, bigger in smaller deals. Small T-shirts are becoming mediums.
Mediums are becoming larges. Larges are becoming extra larges. Our infrastructure is well suited to cater to that. Last but not least, early innings is probably an exaggeration of where AI is for the data center capacity. All this leasing is in buildings that are getting built, right? It's not like they turned on gigawatts of capacity which is sitting on the shelves and these models are being going. As this evolves, you're going to move from training to inference. You're going to move from a ChatGPT or equivalent that's using public data or the internet to private data. You're going to evolve from public consumers using the most of this to enterprises using this.
I believe the proximity of the data that sits inside our four walls, not just the on-ramps we have but also the availability zones we have, as well as the hybrid IT we have for our enterprise customers, positions us not only from growing our space and power needs but also connectivity needs. That's still to come. We don't know where this is going to go just yet. It seems like what we built here is going to be proven even more valuable over time.
When you look at the portfolio, the inventory you have, you talked about the under one megawatt highly connected. You've talked about the hyperscale. These enterprise and what's sometimes referred to as some of these wholesale deployments that kind of sit in the middle, is this an opportunity where you also still see growth from your customers and Digital wants to play in that market? Or over time, do you see the company evolving to maybe more of a barbell strategy where you're really focused on highly connected and you're really focused on hyperscale but the stuff in the middle just starts to decline in terms of exposure?
I believe, first off, I think we're the only company in the data center space that actually prunes its portfolio and outright sells, exits markets. We sold $ billions in that category of infrastructure that we didn't think were long-term holds or places where our customers wanted to grow. We've whittled that list down tremendously to a very, very, very small portion of what we have today. At the same time, I don't think this is a vanilla and chocolate ice cream game. And if you fall in between, that there's no such buyers. And the fact that, as I mentioned, these form factors are growing in size, we're actively able to reposition a hyperscale, a scale, a wholesale towards the colo or towards the enterprise use case. So I don't believe that there's this tweener category that no one wants to lease, quite honestly.
If it's that category, we've sold out of it, quite honestly. That's what we've been doing for some time, not just last year. So if you look at our portfolio, we're actively looking at capacity blocks. I'm not sure I'd call them hyperscale. But maybe a hyperscale customer was in there. They churned. They went to a bigger building. They went to X or Y. And we can say, "You know what? That's part of our now of our colo roadmap or maybe for our high-performance compute liquid cooled roadmap." So repurposing space to higher and better use, which is a very modest capital investment, if anything, and driving much better returns in addition to a stickier and growing customer base.
One of the questions that we've gotten a bit of since the earnings is trying to think about how to compare the same-store NOI growth opportunity this year, the guided range of 2-3 from last year's performance. There's a new slide, I think, in your deck today that unpacks kind of the impacts, pluses and minuses in 2023 and how those play into 2024. Can you just take us through the high points of what's important for investors to know and then how that relates to the long-term stabilized growth opportunity for your portfolio?
Sure, Mike. I'll take that one. So yeah, we put a new slide in our presentation posted to our website today this morning, trying to give some more color around this topic. I think the key highlights are when we've had, as I think most of this crowd knows, we had negative Same-Store Growth for the last couple of years, meaning from 2022 and before that. 2023, we saw we started to see the fruits of the positive pricing dynamics that we're seeing in our overall market. We started to guide in 2023 at roughly 3%-4%. We outperformed that.
We discussed that and we kind of laid out in the slide, a big part of that outperformance was ultimately where power pricing was driven by power pricing largely in our EMEA region where we got close to 200 basis points plus of improvement related to that, plus some benefit from also CPI, which was also much higher in 2023 on the backs of inflation that started to accelerate in 2022. Those things have now, when looking at now transitioning into 2024, those benefits are now have become, call it, a headwind as we've seen both CPI and, more importantly, power pricing within the region come back to more normalized levels.
So our, call it, roughly 2.5% same-store guide that we gave for 2024 is impacted by, call it, 200 basis points from EMEA power pricing impact as well as some portion of CPI that has come back down to more normalized levels, as well as, call it, roughly 50 basis points of incremental property taxes, largely in Chicago and Northern Virginia markets. Chicago, we started to see that last year. We're expecting to see some more of that this year. So that roughly, call it, 2.5 midpoint guide is more like 4.5-5 on a normalized basis for 2024. And I think that's more representative of what we should expect going forward from our same-store cash NOI growth.
Just while we're on the topic of this, so if you take that 4.5-5.5 normalized range, if that's the right range going forward, what does that mean for the core FFO per share opportunity before any interest rate headwinds come into play? Then recognizing that we're in a higher rate environment, what that ultimate realizable growth could be for shareholders?
We said on the earnings call, we said since then, we're teeing up to accelerate the growth to be mid-single digits+, right, in 2025 and doing it in a framework that's going to continue to generate that for several years to come. That's what we're essentially doing. So take those 4, 4.5, 4-5% same-store growth, modest G&A. We got development capacity coming on, putting on a leverage benefit or headwinds depending where we are in refinancing debt to get the bottom line growing mid- to high-single digits.
Maybe just staying on pricing. Is it possible to quantify the mark-to-market for the portfolio today? And then as you think through the multi-year renewal cycle, if you can quantify kind of how to capture that opportunity?
Again, bucketing the two categories, less than a megawatt interconnection, we're churning those contracts every year, every 2 years, every 3 years. So there's a pretty good refresh. There are episodic markets like Singapore or certain U.S. markets or even Europe where there's certain tightness where you have outsized ability to push rate. But I wouldn't say there's a massive embedded uplift just yet. And we do obviously look at where our competitors are pricing and make sure we're maximizing the value for it in that category. So maybe some uplift versus but you kind of capture it pretty quickly. And we've been capturing it. On the greater than a megawatt, our installed base at expiration is 117.
That's in place in 2026 at expiration.
We're doing deals now in Northern Virginia, $135, $140, $150, $160. That's been what the trajectory is. That's a huge portion of this. We have some markets that are less than that. We certainly have markets that are higher than that. I think we're several $ per kilowatt in the money on the greater than a megawatt category.
Then as you look to renew, are there changes to the duration of the leases? And are you trying to move more towards inflationary protections within the leases?
We've done a few things. First off, when that same market is the most hot market in the world now was not quite so hot. We've basically inverted our curve. And we were incentivizing shorter-term contracts. So we had better opportunity to capture an uplift in that. That's no longer the case. So we're now looking for a push for longer-term in those markets, for bigger capacity blocks, that is. Over the last year, we've been on a big push to expand CPI Indexation into our bigger longer-term contracts. Even if we have to get it with a cap and a floor, we're trying to push that insulation into our risk. We're also pushing other features, reimbursement for taxes, insurance, base stops, sometimes Triple Net. So trying to, again, take advantage of this opportunity to put better terms into the longer-term contracts.
Just back to the one other quick question on pricing. For the under 1 MW, while there might not be a sizable mark-to-market, what would be your expectation for average cash renewal spreads in that, recognizing, as you mentioned earlier, the performance has been pretty durable there?
I mean, there are a huge portion of that guidance, which is 5% Cash Mark-to-Market. So I'm just saying there's not like a 20% uplift in that category because we've been pushing them 4%, 5% for years and years.
That's helpful. Maybe back to just the topic of AI. One question that does come up regularly is how to think about the opportunity to accommodate more dense workloads across the current portfolio and if there's a risk of obsolescence of some of these facilities as you look out over a multi-year period of time. You talked about some of the opportunities of repurposing, reconditioning. Can you kind of share how Digital Realty thinks about these questions?
One, I think the misnomer is that AI workloads and GPUs are going to totally erase CPUs from the planet Earth, which is not the case. So we have many applications and use cases in probably the most network-dense locations, which I don't see GPUs replacing CPUs, quite frankly. So those locations are legacy telco hotels that were not built as data centers and will likely be called supporting the internet for years to come. By and large, where this is applicable is our campus capacity blocks. Much of that capacity, it already has a chilled water loop, which is the major, most important ingredient for bringing liquid proximate to the chip. So once you have that set up, it's pretty easy to retrofit. And that's what we've been doing. Those are those examples I mentioned, be it Paris or markets in the U.S. e.g., Chicago.
We've been doing this for many years. For infrastructure on those campuses that do not have a chilled water loop, there are also designs to retrofit those and bring liquid closer. These do not look like astronomical capital needs. And they also would be prerequisite that the existing customer needs want it, i.e., they want to leave because they can't get what they need. So I don't see a major obsolescence factor here coming to our portfolio. And again, we've been selling outright billions of dollars of data centers for stuff that, not because it's necessarily obsolescent, but we certainly use that as a key criteria of what we exit.
I'll just remind everyone, if you want to ask a question, just hit the button on your microphone. The light will pop up. We'll try to get to your question. Another question just on the AI trends and just demand trends. What are you seeing in terms of geographical interests? Are you seeing a bigger focus out of the primary data center markets into secondary or tertiary markets to go after a lot of capacity, cheap power? Or are you seeing still the demand in core markets? And then what does that mean for where you might expand your presence over time?
So I would say the AI trends when it comes to large language model training is certainly, first off, been very U.S.-focused. But we are seeing it globalize. And I think that's going to continue. Two, there's obviously a push and pull between core markets and newer markets. Even when the customer has the ability from an application standpoint to position it outside of a core market, we still see many customers coming back to those core markets because they're using dual uses, including cloud compute that has radius restrictions for Availability Zones. And the infrastructure is so much more prevalent in the core markets, right? It's not like these non-core markets had a bunch of data centers just sitting idly there or all the ingredients that go towards data centers. Our strategy has been to stick to our knitting. We're not in this business for an AI trade.
We're into this business for a long-term growth for our customers and obviously our shareholders. In the core markets, be it AI, be it cloud compute, be it enterprise hybrid IT, we see robust and diverse demand. We see the able to capture more than our fair share of AI as well as to cater to numerous other use cases and be able to deliver the greatest value for our customers.
So one other question I want to try to squeeze in when we get to our rapid fires. So in the past, you've talked about some of the markets. You alluded to it earlier in Northern Virginia of dealing with power constraints from the utility companies. Where are you seeing those constraints? And from the timelines that are currently given where things might loosen up, is that starting to come in, stay the same, or even possibly push out as you look at this as one of the variables influencing supply and pricing?
Whether it's power, generation, transmission, supply chain on components for substations, NIMBYism, sustainability concerns, those markets are Northern Virginia, Santa Clara, Chicago, Singapore, Seoul, Amsterdam, France, Frankfurt, Dublin, numerous markets. They're very pervasive. I don't think I can think of one where things got better. So that's been in the midst of this for some of these for 12, 18 months now. We don't know if there's definitive push out. But my odds are this is not going to get course-corrected quickly or overnight and has good potential for dragging out even further.
Sure. Any questions in the room before we do a rapid fire?
All those what I just discussed about the 3 gigawatts, that's land we owned, permitted. And anything that has any near-term delivery potential likely has a Will Serve Letter from a utility. So we have some form of contractual right to that power allocation. That's what I'm referring to, those 50s, 150 utility blocks I just mentioned, that would be the case.
All right. Rapid fire. Same story in NOI growth for data centers overall next year in 2025?
Hit that one.
Yeah. I mean, there's two players, right, in the public sphere. So I would go with where we talked about from a normalized in the 4%-5% area.
Two players. But will the property sector have more, fewer, the same number of public companies a year from now?
Jordan?
Same. And then what's the best real estate decision today? Buy, sell, build, develop, redevelop, or buy back shares?
I think build.
Great. Thank you very much.
Thank you very much.