Thank you for joining us. I'm Dave Barden. I head up Telecommunications and Comm Infrastructure Research for Bank of America, based in New York, for the U.S. and Canada. I'm really pleased to have with us today, Andy Power, the CEO of Digital Realty, and Jordan Sadler, the head of IR, to join us for our kind of conversation today. So thank you guys for joining us.
Thanks for having us.
Pleased to have you. So, we're doing a little thing, real estate questions, for every company. We're trying to compile views from the real estate industry on these topics. And topic number one is, do you think the Fed is done hiking, and do you believe that rates may fall in 2024?
I'm gonna let Jordan lead off .
Wow.
For that one.
Well, we have a little bit of debate about it internally, but because I spent more years on Wall Street than him, I get to, I get to.
Did you?
I did. I'm older. One year. So, you know, I think the market is bigger than we are in terms of, you know, the expectations around rates. In terms of what the Fed's gonna do, I think the market's telling us that the Fed's generally done. You know, maybe there's one more increase. We'll see what happens on the CPI print. But in general, it looks like we're through most of the hiking cycle, if not all of it. So I think our expectation is that, you know, rates, particularly on the short end, are not headed significantly higher. And, you know, was there a second part to that?
2024, are rates coming down?
So, I think the market's also telling us on the short end, we'll see some cuts next year. So I think we'll probably start to see something like that as well. I don't know how aggressively, you know, without an event, but it would seem like we're, you know, pushing into a little bit of a moderation in the economy, and we'll start to see rates, you know, pare back a little bit, particularly on the short end.
As a former CFO, does that kind of perspective have a bearing on what Matt Mercier, your CFO, is gonna be doing, from a financing standpoint, fixed and variable, and all sorts of other stuff?
I'm sure it weighs into the calculus. But I think our playbook's pretty straightforward in terms of we're bringing in capital, largely private capital, to bolster our liquidity and delever our balance sheet and position us for greater growth that lies ahead. And I think that's regardless of rate cuts, hikes, sideways, whatever you want to call it.
All right, the second question, I think this is more for traditional real estate companies, but maybe it's relevant to you guys since you're active, is: Do you think that there's gonna be a meaningful pickup in real estate activity, transactions, M&A, in the next kind of in 2H 2023, 2024?
I'd just be guessing on traditional asset classes. It obviously seems like there's a lot of wait-and-see on valuations and waiting for debt maturities and other issues to come to force transaction activity. I think in our space, digital infrastructure, I do think you're gonna see a pickup in activity given the demand environment and the rotation of capital away from legacy asset classes towards digital infrastructure and trying to be a part of this growth opportunity.
Interesting. And I guess the final of the big three, this is not the way you typically get this question, but do you plan on—do you have some current run rate applications of AI in the Digital Realty business? And do you plan to be investing more, you know, more aggressively in the adoption and application of AI to your business?
I mean, we've been using AI in various shapes and forms for several years, from call it sales forecasting tools to RFP response tooling to automate faster, more efficient and accurate response times. Part of my mandate upon taking the seat eight or so months ago, the middle one was integrate and innovate, bring together our acquisitions, our data, remove the internal friction from our customers, and build that foundation for our external customers to have better outcomes and efficiencies. And AI is definitely a big part of that story. I don't think it's gonna be a massive dollar amount, but it's something that's been part of our history and definitely where we're going.
Okay, that's the big three. So I'll let you off the hook on those. So changing gears, you know, look, I think we'll start with, you know, the Digital Realty funding story. I think that's what everyone wants to talk about. At a high level, you know, I'm just gonna throw round numbers out. You know, we've got $1.5 billion of operating cash. We've got $2 billion of CapEx. We've got $1.5 billion dividend. And so on a run rate basis, we're kind of at a $2 billion deficit every year. And, you know, with rates having done what they did and where the leverage found itself after the Teraco deal, you had to get a little more creative, in terms of the funding outlook for 2023.
And there were three pieces to that. One was, divestitures, of which you've done about $300 million of $500 million. There was a kind of steady state JVs, of which you've done, I think, more like $1.25 billion over the $750 million target. Then there was development JVs, of which you've done zero out of 750. And then you kind of threw in a fourth extra, which was the equity raise, about $1 billion earlier in the year. So, I guess the first question is, with kind of the clock ticking on 2023, is development JV still a thing? And can you elaborate a little bit on, if it is, why it is taking so long?
Sure. So your numbers are pretty close to, I think, we're closer to $2 billion on the hyperscale stabilized, which was exceeding our expectations. And just because we have done zero of 750 sitting here, second week of September, I wouldn't count our efforts in that area as a total failure.
I didn't say that.
Listen, the way things played out, I think we, quite honestly, did some of the hardest activity first. And we didn't do this in a linear fashion. We went out with all these activities, both non-core, stabilized hyperscale, development hyperscale, to the right parties for each of those types of buckets of activity. We've outperformed on the hyperscale stabilized front. And the development, I believe, is proving to be the most attractive to the broadest pockets of private capital, given its return complexion. And the only thing I can say on timing is, we've won. It's been. As we've had this success, things have come our way in terms of valuations and structure of each and every development or each and every transaction we've done, as well as the ones that are underway.
They are more thoughtful transactions. It's very easy to just joint venture a stabilized asset with very few moving parts or say goodbye to a non-core asset to a new owner. We're looking for partners that are gonna be side by side with us, not only investing with us in what we put in the ground to date, but also going forward and are like-minded on those projects. So both being thoughtful about who the partners are, the projects, and the fact that, we've had other success that allowed us to be more deliberate on this, on development JVs, does not diminish my conviction and commitment to getting this done. I can't promise you two JV announcements on the next earnings call like the last one, but I see no reason why they wouldn't be coming, shortly thereafter.
So when you say that, do you think about these as maybe discrete projects, or do you think of this more as maybe a funding pool of partners, where it's kind of like, it's just kind of a revolving credit type of partnership?
For the development activities that we're contemplating, these are very much tied to a specific series of projects, not married to a new partner for regions or longer issue. Now, these partnerships thrive, we're happy to continue to grow them. I would say holistically, all the activity, whether it was the stabilized hyperscale joint ventures or this development contemplated joint venture, we had urgency, and we could not let perfection be the enemy of good in 2023. We've come across some great partners. I think the next partners we announce will be great partners as well. But I think the longer game here is to evolve this into more direct LP partnerships, and fund management models, for our hyperscale business.
That's not to take anything away from our partners to date, but that's where I think this business is going in terms of capital intensity, and I think Digital Realty has a major hand to play.
So that's great. So you've had success kind of getting the funding addressed this year. There's more that you'd like to do to kind of maybe create a safety net at the margin. But so one of the interesting things is that the reality, though, that you get pushback on is, as we think ahead to 2024, think ahead to 2025, and we do this math that, you know, the EBITDA is growing $200 million a year. We can lever that up maybe 5x. We can raise $1 billion of incremental funding using that, but that's kind of expensive capital still. These other methods of selling existing assets, that's dilutive. You know, selling steady-state JVs, that's dilutive.
The equity is dilutive and so as we look ahead, we see kind of. We'll talk about the fundamental, which are strong, but we've got these mid- to high-single-digit top line drivers, but they're kind of winding up being low- to mid-single-digit AFFO per share drivers because of the dilution that exists between creating that top line and how you get there. Is there a solution to that? Is there a time when this isn't a headwind to AFFO per share than it looks like it is today?
This company's gone through a tremendous amount of change in the last year or even predating that, in terms of reconfiguration of our portfolio and our capabilities. And we had to take steps backwards on a per share basis through selling outright or selling joint venture stakes, that we lose earnings and get redeployed. Just like almost everybody else in this conference, we're a REIT, so we really can't retain much of our cash flows, and we have to be, to be a REIT, that's in a sector with a platform that has tremendous new capital opportunities at our fingertips.
I believe when it comes to data centers and cloud and hyperscale and certainly the AI to come, that a combination of both public and private capital sources is the most prudent way to fund this business and maximize value for our shareholders and our partners. What we're doing this year that's different is what we talked about, is sharing the development piece of the equation with private capital sources, which will make the balancing more efficient and allow the organic fundamentals, which, by the way, just recently inflected. I mean, that's the inflection is not a long ago story. I'm speaking many quarters and conferences of negative cash mark-to-markets and poor capex to revenue line growth.
In 2022 into 2023, we've seen this inflection in a backdrop with demand outpacing supply and going towards pricing power and higher ROIs. So this development JV is gonna allow us to call, hopefully turn the needle into 2024 and 2025 and beyond, and allow us to accelerate our bottom line growth at the same time.
So let me ask this question: With respect to these development JVs, do you look at this as a force multiplier, where you plan to spend the same amount of CapEx but get a lot more bang for your buck? Or do you plan to use the JVs as a substitute for the capital that you might otherwise have spent, which will kind of maybe accelerate, deleveraging and other things?
It's gonna be both, quite honestly, right? The supply-demand dynamic, assuming we do what we said we're gonna do, which I think our track record so far has been pretty good, the supply dynamic, demand dynamic drives where the capital allocation goes, so once we get the balance sheet back in order. I look at that dynamic, and I see the opportunities increasing at better ROIs. So we haven't finished 2023 to put a budget for 2024, but the CapEx intensity overall could be the same, even potentially increasing, but I think our pro rata share of that will go down.
I.e., the partners we bring in today, we're gonna raise an investment for their share of what we've already extended into the ground, in the shells or certain completed halls, and then they're gonna fund alongside us and carry some of that non-increasing value through the development cycle. So force multiplier will obviously bar fees to the platform, lighten our balance sheet, and that's very much a tool in our toolkit in delevering, i.e., carrying non-producing assets through the year.
One of the funding vehicles that hasn't been open, that was a big opportunity when it was launched, was DC REIT. Could you talk about how does DC REIT kind of fit into the current and future profile of funding the business?
Obviously, very disappointed that the DC REIT has not been able to be a bigger part of our story, because I think it elegantly meets both theirs and our objectives in terms of shareholders. It is obviously had a major customer bankruptcy. That process, the specter of that process has been playing out for a while now. The technical bankruptcy process didn't start until the beginning of summer.
They most recently had an extension to that process, but will bring it closer to the end of the year. So it's kind of we're seeing the light at the end of the tunnel. I don't know what's gonna happen on the heels of that definitively, but I think any form of clarity of that, of any realistic outcome, I think is a good thing for DC REIT, and I think that with already you've seen a rebound in its cost of capital, even prior to resolution of its tenant bankruptcy. I think it is a, breathes a second life, where we can continue to monetize or vend assets into the vehicle at third-party fair valuations. And it takes our hyperscale exposure that we have a significant amount of on our balance sheet and allows it to generate external growth.
With respect to Digital Realty, specifically, and specifically with respect to that Cyxtera bankruptcy, you guys took about $0.05-$0.07 out of the output for AFFO per share. At the time, you said it was kind of a worst-case scenario kind of budget. I think, based on the process unfolding, is there reason to believe it's not a worst-case scenario? Or if you don't know it yet, is there a moment between now and when we finally get the end of the bankruptcy, that you could have clarity on that?
I wouldn't call it a worst-case scenario, but I don't, I don't think when we, when we put those numbers out there in July or what I do with the facts and circumstances now, I don't think it's gotten any worse. Potentially, maybe it's gotten better. We're kind of coming to the end, as I mentioned. We have not had any of these rejections to date. I think 2 of the Cyxtera tenants are pre-petition, which I would not count on recovery of. The majority of that, there could be recovery through lease being accepted, or if leases are rejected, we can inherit the underlying customers in this capacity. I think what we know more of now is, I think the latest 90-day extension just got triggered.
It appears that based on the bankruptcy filings, that there is a potential buyer that's not Digital Realty. They're also very much committed towards an alternative path of the lenders becoming the equity. The pro forma balances they've shared, actually, I would say, surprised me the upside. I think it's like roughly 2x debt to EBITDA. They obviously, if they take the business back, they appear to be looking to do it, and they have a much more conservatively levered balance sheet, which is a good thing.
I guess the theory is that even if there are leases that are rejected, that have, you know, some short-term dilutive effect, that we're in an environment where they're pretty healthy again, in general, right? Because those represent re-leasing opportunities that would be, you know, future growth.
So one thing that's different about us when it comes to data center land is, we don't believe we built or acquired the 300 best plus data centers on the planet Earth that'll stand the test of time. We've actually went into our portfolio and continued to optimize and monetize things that we don't believe are core to our strategy or have longevity or pricing power for the long run. We've sold $700-ish million of non-core dispositions, a large piece of had concentration to that customer in standalone one-off sites, outright going 3 or 4 years back. The majority of our remaining exposure are in either a combination of multi-customer locations, thinking highest demand pricing power locations in Chicago or Singapore or Frankfurt.
Or if they are standalone customer locations in tight, tight markets like Santa Clara, where SVP won't have power until 2030. So, and we've been to this movie many, many times in the evolution of history of, of data center advances. Reseller is a really bad word, word coming out. WeWork models essentially go upside down. But in this case, instead of just, office tenants, we have people with critical infrastructure in there. They want to stay, and we can help support those end customers. And we've done this, we've been through this. We know how to operate through this, and, navigate it.
So, you mentioned something I think that people debate a little bit, which is the strategy. I think that people look at the data centers in kind of a bifurcated way. You're either a retail guy who caters to smaller customers with focus on interconnection, and you have a large sales force that beats the bushes to find those customers. And or you're a wholesaler. You build large data centers with relatively fewer tenants, and it's kind of an old school real estate model, and, and they have longer leases, and, and, you know, I think that people think that in the old Bill Stein days, you know, the old Digital Realty was kind of an old school, wholesale real estate business.
You know, Andy Power comes in and, you know, we're buying interconnection, and we're buying, you know, retail assets. And I think that the question now is, is Digital Realty a wholesale data center company, a retail data center company, or a hybrid? And if it's a hybrid, why is that the right approach?
I'm not gonna stand alone as a characterization or question, suggest, take credit for all this activity. But I think what you've seen from our team here is we have continued to build the capabilities to have a global enterprise colocation interconnection that is adding 130, 140 new customers each and every quarter. We've supported 5,000 customers across 50+ markets on six continents, and doing better and better in that capability in an arena where there really is a small, small set of competition on that stage. At the same time, we also believe in the addressable market when it comes to hyperscale, and we are serving the hyperscale cloud customers in 20, 30, 40, 50 different locations.
While we don't pray to the god of trying to have 100% of the market share, we pick our spots as to where we can provide value to them and create excess return from that value add. Places where they've landed their Availability Zone, where it's harder to do business, where we can future-proof their runway for growth. And what's different or newer, or is that we're trying to accelerate that growth in the less-than-megawatt interconnection, the customer count, the connectivity, the penetration of the enterprise customer base. And we're trying to take our hyperscale business and re-tweak our financing strategy to it, and bring in private forms of capital with non-stabilized assets and development, to maintain more pricing power and same-store organic growth on the public company DLR.
Just on that point, on the hyperscale opportunity, there's been a lot of capital being poured into kind of the what I would call, like, the bespoke data center market. Companies like Aligned and other guys are out there with private capital, building custom data centers for the hyperscale guys. Is that a threat, or is that a reflection of the opportunity, where a rising tide can float all the boats, and the tide might be a lot larger than we see even in the public markets?
What I would characterize the private capital as really solely flowed towards hyperscale business. They've shown very little of interest in mustering a run to create a third or fourth horseman to ourselves and our major competitor in the enterprise colocation. And it's been going after a small subset of big customers with big deals, has been the movie. And the likes and names you create mentioned are solely focused on that. And they're a product of the demand just being so robust and diverse, right? And they filled in gaps where we elected not to play. We, there's a report that came out just recently on the Phoenix data center market. That's a market where we still operate, and we have a highly connected downtown destination.
We got a campus in Chandler, which is this very supply-constrained market. Years ago, we said: You know what? We're not going to chase hyperscale into the deserts of Phoenix. We sold a land site we owned, we sold a land site we inherited from DuPont Fabros. You know, we ceded opportunity. A lot of hyperscale business got built there by 10 or 12 different names, all doing the exact same thing and all fighting and commodity like dialogue and conversations with the hyperscalers, which is fine. We picked our spots. It's at Santa Clara, Ashburn, Paris, Frankfurt, Singapore, Osaka, Tokyo, São Paulo, South Africa, were places where we could support these customers in a different way and generate better returns. So that's the little difference in what we're doing versus some of our private.
So let's talk about the inflection. You know, you're right, you're right. We had spot market pricing was below embedded pricing in the base. You know, retention pricing is always north of spot pricing because no one really wants to leave the data center, but you got to give them a break. But all of a sudden, what we started to see in the last basically year, is that spot market pricing started to kind of inch north of where the embedded lease pricing was. We had positive re-leasing spreads. I think you've been guiding to about positive 4%, for 2023. I want to dig into that a little bit, but maybe you could just for the benefit, like, what made that inflection happen, and how sustainable is it?
Listen, on our less than a megawatt category, we've had a long trend of positive mark-to-markets that have just inflected more positively. I would say that goes to our history and our success in the colocation business, being able to obviously curate our customer base and push the needle on pricing in that category even further. The bigger new news is that I've been saying for a good chunk of 2022, and rapidly accelerating in 2023, is the pendulum on pricing power has moved back to the providers, and that's by and large pre-AI demand hitting the scene.
Where demand remained robust and intact and diverse across regions, across major markets for bigger deals, while supply just ran out of gas on multiple angles, transmission lines into Northern Virginia, 2030 in Santa Clara, generation in Dublin, moratoriums in Singapore, nimbyism, environmental concerns. A whole host of factors that allowed us to have greater intestinal fortitude to push back on rate, because it wasn't just commodity-like competition. So, that has been evolving and I would say, accelerating as we went through 2023. And I'm of the mind, I don't think that dynamic is gonna rapidly resolve itself. So I, and that, again, was a pre-new AI demands kind of coming onto the scene.
One of the contributing factors to the rising, you know, rental revenues that you're getting is not power. You pass power through. Some of your counterparts in the enterprise colocation market have different strategies which have yielded some pretty strong returns for them. Is there any part of you or your business or corners of your business that you might rethink the power pass-through approach to that element of the business?
By and large, in our call it less than 1 MW interconnection or less than 1 MW colocation piece, we do have a very similar contract, maybe not identical, in terms of the frequency that we're able to reset the power cost of the contract. Certain contracts are priced with, like, a margin to the power. But we're similar boat when it comes to called shared infrastructure colocation. On the greater than 1 MW, the conventional standard has been pure power pass-through + C, and I don't think we're gonna change that dynamic anytime soon.
So your, your guidance is for the re-leasing for the year to be kind of 4%-ish. And that's a composite function you've been talking about, kind of positive in enterprise colo, and not negative in the above 1 MW. Could you remind me what the re-leasing was in 1Q and 2Q for greater than 1 MW?
You keep me honest here. I think we were up at 6.6 and 1/2 -ish or something on the greater than a megawatt. I think the less than megawatt was 4.8, versus the trailing twelve, about 4.2.
Right.
So both acceleration. I think we called, I believe we called out on the call, we did have a shortage from renewal in the mix there that maybe inflated, like, 100 basis points, but it's still pretty darn healthy.
Right.
So.
So, the guidance.
That's 2 view.
So the guidance out there is kind of a zero-ish kind of re-leasing benefit from over the greater than 1 MW, but the first half of 2023 is kind of in that 500 basis points. What has to happen in the second half of 2023 to make that zero happen?
Technically, we only guide in aggregate. We're certainly coming out of many innings of pricing pressure that zero felt like an aspirational target for a long time, and being on the north side of zero, is a new move.
It feels weird?
But I can tell you, the pricing is, it has moved dramatically. I gave you examples on the most dramatic moves, Ashburn being in the $70s-$140. And it feels like it could continue to run. I wouldn't overindex on, and I've heard this before, well, if you're this much in the first half of the year and your guide is still +4%, that means you're deceleration in the back half of the year. We're handicapping the greater than a megawatt leasing, because the sample set is only a fraction of the size of contracts, and these are the customers renewing before the contract comes due. We can't dictate if they sign it in the third quarter, the fourth quarter, or January first. Right, so,
But when you look at that pricing landscape today, and you look at that embedded base, I mean, is there a real risk inside your contracted portfolio that there could be down pricing? Because it feels like prices everywhere in the world are up relative to what the embedded price is, even that was signed five years ago is. So is it even possible?
I think it's a tough, it'd be a tough set of hands to have a negative Mark-to-Market, likely in a greater than megawatt category.
That's a pretty strong thing for the last decade.
So, but we'll see. I mean, again, if I got a bunch of really good ones, I have one contract's negative, and if all the good ones slip to the next quarter, and that one got signed, like.
People will kill you.
Yeah. But I can't control that. But, but I, if I'm guesstimating, I think it's probably tough to.
So I have to do it, and I apologize, to end this on the AI opportunity. We just had a two-day AI conference here, the first-ever 2023 global AI conference. One of your peers was a participant in that. I know you guys have talked publicly about your role. I think that the, I just want to kind of say that, what's your response to the critique that the embedded base of data centers are just old dinosaurs, they don't have the power density capabilities, nor does your development pipeline, and that the whole AI thing and the assignment of value to companies like Digital Realty, that there is any opportunity, AI is misplaced. How do you respond to that critique?
First off, I've said this before, so I put AI in the category of, unlike a lot of the buzzword bingo that has come through our asset class or industry, be it crypto or edge or 5G or autonomous vehicles, to date, AI does feel very real, and I think Digital Realty is gonna have a demonstrable role in that. The first inning would be a massive overstatement of where we are and where this is going. We are seeing demand. We've been catering to AI historically. We're seeing new AI come on the scene. The hyperscale cloud providers are a major piece of this. The training of large language models is what's happening now or just about to happen now. It's probably a better characterization. They need large, continuous capacity blocks.
That's priority number one, two, and three. We got over 3 GW of that, so there's a role. And while it's shown to not be low-latency, location sensitive today, we are seeing customer preference near Availability Zones where they have compute because they don't know where this is going. Inference in the future is gonna be a multiple of training, private data sets, and the thesis to be tested is the proximity down the road, which is a future state. That is not gonna be a 24 inferences will be up and running, and we're gonna be living in this massive AI world. But having proximity to the Availability Zones of cloud providers and Digital Realty, hybrid IT , or even the network nodes, and that may be a stretch of its need and proximity.
I don't see how this is a t the very least, a rising tide is lifting the industry, and I think Digital Realty will benefit in this arena. And I don't see this, having spent time on the GPU versus CPU, and I don't see us emptying out 300 data centers and putting in new GPUs. Even if you take the most bullish projections of someone's highly incentivized to sell GPUs, there's still a massive GPU for numerous workloads that's gonna exist. So, and we've done the work as to what our infrastructure would need to be retrofit and already tailored our new designs to cater to it. And we can do that with the four walls we have today, and the capacity we're growing for the future.
I think that's a great place to leave it. Thank you, guys, everyone, for, for joining us today. If you're interested, we did put out a piece on AI and its impact on the data center industry last week, so you can find that, in the Bank of America Research Library. Next up in this room will be Ventas. So enjoy the rest of your conference, everybody. Thank you so much.