Good morning and welcome to Citi's 2025 Global Property CEO Conference. I'm Mike Rollins with Citi Research, and we are pleased to have with us DigitalBridge and CEO Marc Ganzi. A couple of housekeeping items: this session is for Citi clients only, and disclosures have been made available at the corporate access desk. To ask a question, you can raise your hand or just hit the button on your mic, and we'll get your question included when we can. You can also go to live.q4.com and enter code GPC25 to submit questions. Marc, we'll turn it over to you to introduce your company and the team, provide any opening remarks that you would like to provide, and then please tell the audience the top reasons an investor should buy your stock today, and we'll get right into our Q&A.
First of all, thanks, Mike. It's good to be here with you again. I think I've been attending this conference for 22 years, I just figured out, so I hope we continue another 30 years. It's a delight to be here. As you just whispered in my ear, it's a home game for me, so I get to sleep in my own bed, so that's always an advantage. My name is Marc Ganzi. I'm the CEO of DigitalBridge. We are the largest owner and operator of digital infrastructure and digital real estate in the world. We manage $96 billion of assets across the globe and five continents, really with a specific subsector focus on data centers, fiber, small cells, towers, and adjacent infrastructure. The business today is in the form of an alternative asset manager.
We started as a REIT. We morphed into a C c orp, and now we sit in an alternative asset manager framework. And literally, as you know, Mike, that was out of necessity, not out of a plan. As a REIT, what we found was our alternative asset manager sleeve was growing faster than the REIT. And as we continued to take in third-party capital to meet the needs of our customers, our tax-free subsidiary got to a place where it unfortunately was growing faster than the core of the REIT. So we made the election to go to C c orp, C corp into alt, and that's, I think, we'll sit there for a while, Mike. I don't want to give you another zip code change for the next decade. So I think we found a place that we like.
Most importantly, it really supports, Mike, this notion that today in the digital real estate business, you really have to do three things really well. One, you have to form capital very quickly. That's something that our public brethren, whether it's in data centers and towers, have struggled a little bit with in the last four or five years. So the necessity for us to keep growing and to continue to support our customers was a function of being able to raise capital. The second thing you need to do really well today, I think, in the digital real estate space is you need to be able to be there early with land and power and the associated improvements, and you've got to be able to deliver for your customers on time. By the way, that's not a data center phenomenon.
That exists in our fiber businesses, that exists in our cell towers, our small cells, and of course, our data centers, and our 31 years of operational experience has shown us that if you can show up on time for customers and you can deliver space, power, cooling, and connectivity on time, you win. And this past year, we really won quite a bit. We were really successful in being able to deliver for our customers. The third thing we'd say that you have to deliver is the right outcome for shareholders, and we've replaced the word shareholders with stakeholders, and so this year, that's going to be on display more than ever, and so our stakeholders are not only you, our public shareholders, but also our private LPs.
And so delivering for both constituents is sometimes a bit of a tricky wicket, as they say, because on one hand, for our private LPs, we delivered a ton of DPI. We continue to bring capital back and recycle capital for our key LP partners. And for our shareholders, we have to continue to deliver value. And value for us in our current format is the ability to go out and raise capital, deploy that capital, grow our FRE, and at the same time, grow distributable earnings and grow assets under management. So as I look forward, why would shareholders want to own DigitalBridge? Well, first and foremost, we're the fastest growing digital landlord in the world. That is just a statement of fact.
If you look across our data center portfolio, our power bank of 16 GW, our 230,000 route miles of fiber, our over 500,000 wireless sites, we're in every continent, we work with every customer, and there isn't a place that we can't go with a customer and support them. It's a really unique architecture. It's a unique portfolio, and it's a unique roster of customers. And that ability to go anywhere and to show up for any customer and, most importantly, perform for them, Mike, is what sets us apart from our other peers in the digital REIT space. So that's DigitalBridge in a nutshell, and happy to go anywhere you wish today, Mike.
Great. Thanks. Maybe just starting off with kind of the portfolio opportunities and in terms of how they're performing, you gave an update on the earnings call around pipeline and bookings. And if you could share some of those details with us in terms of maybe an aggregate, but also how the major product classes in your portfolio are performing.
I think this was a pretty unique year, Mike. I think this is one of the first years that I've seen really strong, high single-digit and double-digit organic growth from all of our four products at the same time. Usually in digital infrastructure or digital real estate, you'll have one product that performs quite well and maybe perhaps a down cycle in CapEx. I think to put this in a proper framing, what's happening with artificial intelligence infrastructure has really changed everything. I talked a lot about it last week at the Metro Connect Conference about ecosystem investing. Obviously, the lead sled here is data centers. I think at this conference, Mike, you've fielded more questions on data centers than I think you've fielded in your career. It's topical, right?
It's what's on people's mind, and it's what people want to talk about. If you start with that category, last year, our sector delivered just a little over 7 GW of leasing, Mike, as you know, across the sector. And at the same time, in terms of the activations, the sector's lighting up about 1.5 GW-1.6 GW in terms of actual data center halls delivered. So what's happening, Mike, in our sector today is there is a trade-in-supply imbalance of just over 1GW of power. And so while the sector is leasing to the tune of 7 GW-8 GW, it's only delivering 6 GW of capacity. So that's about a 1 GW supply and trade imbalance. If you pull that forward to this year, we've reforecasted the sector performance for this year, Mike.
We're forecasting about 8 GW of total demand on our sector across all of our peer group and ourselves, and again, we think the construction and supply chain issues that face us and power issues, we think the sector, again, will deliver about 1.5 GW per quarter. That seems to be the cadence that we're on, and so that creates, if you take that trade-in-supply imbalance from last year and you pull it forward in this year, we're going to keep falling behind, Mike, in terms of the ability to deliver on capacity, in terms of actual data hall deliveries. That's lit capacity against leased capacity, and keep your eye on that metric because that's the metric that we're chasing. That being said, data centers across our total portfolio last year grew over 20%.
And as I mentioned in our quarterly call, our backlog is up over 22% vis-à-vis last year. So that combination of very, very strong double-digit growth plus a pipeline that continues to grow. And then if you look at the announcements, just focus on four announcements, Mike, in the last 10 days. Alibaba, CapEx, up $58 billion. That's a big number. Apple came out with a $500 billion number in increased CapEx in terms of total infrastructure spend. Meta came out with another $200 billion of CapEx. And on top of that, then discussed financing over $35 billion of their CapEx with another financial sponsor. And so all of our customers are growing. And at the same time, Microsoft affirmed and reaffirmed its CapEx at $80 billion. So our forecast going into the fourth quarter last year for CapEx spend across the hyperscalers was $250 billion of CapEx.
We've now revised that forecast to $300 billion of CapEx of total spend. In terms of the velocity at which CapEx is being spent by the hyperscalers, Mike, in digital infrastructure, it's now over $300 billion for 2025. There's a lot happening, and it's pretty exciting. As you trickle down that ecosystem, the next logical place to go is fiber. We were pretty forward-thinking last week on our front foot about fiber. We're seeing a lot of really fantastic things happening in our fiber businesses. This is all related to AI connectivity. The new architecture to data centers is certainly more strand count to the data center. Our customers are leasing more strands from us than they've ever leased. They want more capacity. They want lower latency routes, and they want redundancy.
So if they wanted two-path redundancy into a cloud data center, they now want four-path. If it was 12 strands or 14 strands into the data center, we've gone from 48 to 124 to now 248 strands of dark fiber into a data center with multiple redundant paths. The amount of bandwidth that's going into data centers today is upwards of 10x what it was in sort of traditional public cloud. So these workloads are really bandwidth-hungry. And what we're finding with our customers is they initially procure a load into that data center. And much like you like towers, they come back with amendments. So they go from 12 pairs to 14 pairs to 18 pairs. And so your MRR goes up. And what I love about that is your CapEx is zero.
Everybody loves a real estate business where your TI is zero when your customer, when your tenant signs an amendment. That's what's happening in the dark fiber data center connectivity space. The last kind of key thing is mobility. Mobile real estate or mobile infrastructure real estate really had a very strong end of the year, and it had a very strong start to January. We had really incredible results across our tower portfolio in January. We're really seeing here in the U.S., and I think Crown and SBA and American all said the same thing. We've all seen increased uptake in our pipelines in terms of leasing activity. I know at Vertical Bridge, we've got built-to-suit going with all three customers for the first time in a decade. We have really, really long-term sustainable built-to-suit contracts with all three customers.
The good news is, as you heard on their quarterly calls, Mike, they're all deploying CapEx. They're back spending again. So we're seeing great activity from T-Mobile, from AT&T, from Verizon. And it's not just tower activity. We're seeing it in small cells. We're seeing it in indoor. And the entire ecosystem is really flourishing right now, Mike, on the premise that much like public cloud changed how we spent money in terms of infrastructure spend, AI is absolutely changing the narrative on how we build infrastructure and how we invest.
Really helpful. Maybe delving into a few of those topics. First, you mentioned the 16 GW of global power bank. Is that 16 GW fully available today where you have the letters, you have the utility, or does that 16 GW take time to realize just because it takes time to get transmission and time to get that energy going?
So our power bank is actually available to us and on demand. So these are campuses that exist that are built. We own the land. In many of those instances, almost all of those instances, Mike, we already have data halls built. So in places like Quincy, Washington, or Santa Clara, or Northern Virginia, or Reno, Nevada, or Tamboré, Brazil, we not only have the power, but we have all of the infrastructure in place. And it's literally going from data hall four to data hall five and six, or like in Bluffdale, Utah, going from data hall nine to data halls 10, 11, and 12. These are big campuses where we've been planning for the last seven or eight years for what we felt was a tsunami of activity that was coming. And so that's the power bank we have today.
We have another pipeline, as I mentioned on our call, a very deep pipeline of new activity that amalgamates to over 3-4 GW of new projects that could take us to 20 GW. And at the same time, as we mentioned on our call, Mike, we're spending a lot of time around digital power, which we think is the next frontier for our business in terms of being able to control our destiny with how we ultimately proliferate and bring power to our customers. We want to have control over that. We want to get off-grid, and we need to be energy independent to secure the velocity and trajectory of our customers.
And just to give our group here today just a semblance of growth for your data center portfolio, I think in the earnings, you mentioned you're almost at 4 GW of contracted power in data centers.
Occupied.
What's that?
Occupied.
Occupied.
Lit.
Lit. So it's online, delivering today.
Yeah. Customers in a data hall with a server attached generating power.
Of the leasing that you're anticipating for the category for the year of, I think you mentioned just your expectation for eight, what share of that could DigitalBridge potentially get in 2025?
I mean, I'm looking at the low side of that guy being 1.2 GW and the high side of that guy being 2 GW. We think when you have powerful platforms like Switch and Vantage and DataBank and Scala, we think we can take more than our wallet size. So again, our advantage is having that power bank today. And so when we sit with a hyperscaler in Tamboré, Brazil, they see data center seven, eight, and nine being built. They know they can go there tomorrow. You go to a place like Reno and you go look at the SUPERNAP in Reno, and you can see the data halls that are being built. A customer doesn't have to visualize what they're getting.
They know they're getting space, power, cooling and on top of that, the most advanced cooling system in the world, which is our new Evo system]. So we don't have theoretical conversations with customers. We're not tourists. We're not wildcatters. We're the guys that have been doing it for 31 years. And when a customer sits with us and we say, "We will put you in that data hall by this date," we show up and we deliver. And that's what really sets you apart in this game today because there is a lot of noise in the marketplace.
And then bringing it back to DigitalBridge shareholders, so as you lease gigawatts of capacity, can you boil down the potential benefit that that delivers to your shareholders?
Thank you for asking that. I think our shareholders benefit, we think, three ways. First and foremost, as we continue to build capacity, Mike, we got to raise the capital to fuel that growth. And so last year, we raised $9 billion of new equity that bears fee. And so every time we go out and we raise capital, we're putting on average about $0.60 of that dollar into data centers. And so it's a pretty easy translation really for our investors because for every dollar of capital we raise, we're taking about 90 basis points of fees on top of that. So our investors really get to see the flow through to FRE into earnings very quickly and very efficiently. The second area that they see benefit is ultimately in carried interest.
I think we tried to make the argument this year that one of the great things about being in an asset light model, Michael, is not only do we have the direct flow through for the GP stakes in terms of how we perform for investors, but should we be successful and should we light incremental power and incremental data center space, we've actually created a theoretical algorithm for what it equates to in carried interest. In this quarterly call, we laid out a theoretical framework for one gigawatt of leasing translating to $1.55 of carried interest for our shareholders. As we begin to think through going from 3.8 GW today of lit capacity and pushing through that 16 GW, what we want investors to start thinking about is as we grow our asset base, investors win with us.
And the way you win with us is you participate in the profits interest. We've been very clear that our investors today, on average, are getting, as we go forward, about 1/3 of the carried interest going forward and 28% on the historical carried interest. So as we continue to scale AUM from $80 billion to $96 billion, we've already laid out a path for where we're deploying $16 billion this year. It gets you to a pretty easy cadence of we're talking about $112 billion of AUM by the end of next year. And you can sort of extrapolate what's equity and what's debt. But generally speaking, on a 50/50 debt to equity ratio, you can really see how we're compounding carry, Mike, at a really fast pace.
Now, the third thing is we got to actually manifest it. And manifestation is DPI, exiting investments clearly and cleanly, and then converting that into distributions for our shareholders. So those are the three ways that we can win for shareholders today. And most importantly, we want our shareholders to ride shotgun with me. And I think by aligning the carried interest between myself and our public shareholders, as we work hard to ultimately crystallize that, we want our shareholders to crystallize with us. That is really the optimal place we are now taking DigitalBridge to this year and the years coming up. As we telegraphed on the call, we've got to really turn the narrative around carried interest of one or two times a year into very consistent non-episodic distributable earnings attributable to carried interest, which is where we're going.
Then just one clarification on the carried interest opportunity. When you talk about improving that consistency and quantum of carried interest, is that a combination of unrealized and realized, or is it the focus for DigitalBridge to really ramp up the realized carried interest?
I'm more focused on realized. I think it's more tangible for investors to see a big exit where there's $200 million of carried interest and whether it's $60 million, $80 million, or $100 million flowing to them, they see it. They can feel it. It's tangible. I think this unrealized carried interest that's based on the principal of NAV is a little bit more difficult. And I think with the SEC sort of changing its rules on alternative investment managers last year, we had to adapt, as there's a new framework in Q3. And as we sort of worked through that new formulation, we had to take a markdown on the portfolio in Q3, and we had to take a markdown on the portfolio in Q4.
That sort of accounting change on how Tom was ultimately accounting for the NAV and for where we mark our assets is now over with. So I think what you're going to see is we come out into this year with a clean perspective on that NAV and a clean perspective on the marks. We're now building forward from here. And look, we had tremendous growth at the end of the year. So as you know, when you do a ton of leasing in the fourth quarter, guess what it turns into in the first quarter? It turns into cash flow. And so that cash flow imputes in organic growth, which means our portfolio companies really grew strongly in the fourth quarter, which translates into strong earnings in the first quarter.
I think this is one of these things with DigitalBridge where you can actually look at us and say we are a lot like American Tower and Equinix and DLR, which is the bookings in the fourth quarter translates into earnings in the first quarter. I think you're going to see that flow through happen here in the first and second quarter as we remark our portfolio and we get to grow. So the accounting changes we had to go through in the third and fourth quarter are now behind us. And now we can go back to growing NAV again like we did previously in the previous years.
When you talk about the opportunity to monetize data center assets and recognize some realized carried interest, one of the comments that you made on the earnings call was the estimate that there was $90 billion of other data center assets out there that might look to find a new home for their equity at some point, and you also announced your objective in terms of a product to go after some of that, and there's sort of a two-part question here. One is, does that quantum of supply create valuation risk for the data center category that there's just so much in assets that have to find a home, and secondly, if you could talk about the product that you see in terms of going after that opportunity.
First, from a sector perspective, Mike, I think there's been a resetting of cap rates that happened really in 2023. It started. I think in 2024, it manifested itself even. It articulated itself even better. I think when you have a stabilized large hyperscale campus, as a developer, you're thinking you can get a 4%-5% cap. I think as ultimately that data center moves past 82%-85% occupancy, it's , "stabilized". Then you're not selling it as a growth vehicle. You're selling it as a stabilized piece of real estate. I think it narrows on that bid-ask spread today, Mike, is somewhere between a 5.9% cap rate and a 6.8% cap rate, provided the lease is greater than 10 years and has an investment-grade counterparty. I think if you can put it in that box, that's the bid-ask spread.
Now, the question is, where can we buy assets? Where do we think the market is? And that dovetails into the product. And so our conviction call is that having built two stabilized data center platforms with Vantage, Vantage North America and Vantage EMEA, both of those have yield codes that serve as an off-ramp to stabilized products. And both those products worked, but I think the applied learnings of having launched those two products and having now run them for five years, we've learned a lot. And we understand what investors want in that product and what they don't want in that product. So I think having a clear pulse on insurance companies and pension systems and what their expectation is in this product, we now have four or five years of applied learnings. So it really gives us a head start in fundraising.
So I always say I like to go fundraise with a sniper rifle, not a bazooka or a shotgun. You want to know exactly where you're going and who you're talking to. And so that surgical nature of fundraising is a huge advantage we have in this new product. And so we know the right real estate investors we want to talk to. We've watched other real estate-like vehicles get launched in the data center space, and they raise money, but they can't find the right product. We actually have the other problem. We have all the product. We've just got to hurry up and raise all the capital. So on that new product, we do think that there is a marketplace where the bid-ask spread is narrowed. But most importantly, Mike, the sponsors that have built these big data center platforms and their infrastructure funds, they now have to return DPI.
We know that because our data center platforms were three or four years older than theirs. So they're now at the same place we were with Vantage three years ago. So we know the conundrum that they face, which is LPs are saying, "Look, I have no more co-investment for you. I'm out. I'm tapped." And ultimately, their credit ratings are also tapped. There's only so much debt you can take where your loan to value gets above 60%, and lenders go, "No, I'm not giving you our money." So the only avenue these guys have to grow is to return capital and to recycle capital. And so what we've said is, "Look, we're setting up a product. We think it's a big marketplace. It's a big opportunity." We're doing it in concert with Goldman Sachs like we did our AI sidecar vehicle, which was very successful last year. And we're creating a clearinghouse that's not going to own DigitalBridge assets. It's going to own other people's assets. So it's a neutral platform to ultimately create that opportunity to recycle capital. And at the same time, we're giving our clients, which are mostly insurance companies and pension funds, what they want, which is a stabilized yield between 6% and 8% and returns between 11% and 12% with no risk and with a sharp focus on long tenor contracts and only investment-grade data centers. That's the thesis in a nutshell.
One quick question on that. What defines an investment-grade data center? Is it the tenant credit quality? Is it the credit metrics of the actual data center in terms of leverage?
Yeah, it's a little bit of both, but I'd say the gating factor is the customer. So ultimately, we're going to securitize that portfolio like we've done 50x before. And ultimately, Kroll and Moody's and S&P defines what an investment-grade tenant is. And so we have a very good sense of that. And we have a really good sense of where the rating agencies are going with that product. So I think the recent Switch securitization was a great testimony to that. We were super tight in the single-A tranche. Inside 50 basis points of tightness, we have a really good sense of where that product is. I would say from a leverage level, we don't want to be over-levered. Again, the applied learnings of the two -Vantage vehicles have now given us a really good sense of where the LTV needs to be. I won't discuss it, but I would say our capital structure in this vehicle will be very conservative, Mike.
So we've got a number of questions that are coming in. I also want to prioritize any questions in the room here.
The in-the-room people come first because they came here to your conference. So I put priority on them.
Please raise your hand if you'd like to ask a question. We'll get to you in a moment. You mentioned earlier the importance of fundraising and the focus for the company on that. And just curious if you can give us an update since the analyst meeting in terms of how you're doing penetrating the largest tranche of LPs that you've been targeting. And I think you kind of, if I remember back to the meeting, kind of had a few different cohorts that you were targeting in terms of levels of size and opportunity. And just curious for an update on how you're doing on that front.
Yeah. So the cohort group that you mentioned, the top cohort is the top 20 LPs in the world today. And I'm proud to say we represent 18 out of the 20 largest LPs in the world. And most of those LPs have multiple accounts with us, multiple fund products. And I think that's really the focus that we discussed in analyst day is one, we had to widen the aperture. And so we took our sales team from 20 people to 38 globally, really set up a great team in Asia, which is now starting to bear fruit for us. And I think getting the big accounts is important because they can continue to allocate to you across credit, real estate, power, all the things that we're doing. The top and most sophisticated allocators in the world, they want exposure to digital.
They want exposure to what I would call digital power, which is a hybrid of renewable and some not renewable, but a pragmatic approach to how you bring power to the AI economy. I think investors also want private credit. I think one of the big surprises for DigitalBridge this year was the acceleration of our private credit platform. And again, there you've got upwards of close to now I think it's a little bit less, but it was over $90 billion. Today, I think it's about $70 billion of digital infrastructure LBO paper that was written in COVID that has to be refinanced.
And so one of the things that we've seen with our credit platform that's working really well, Mike, is working with banks like Citi, like TD, like Deutsche Bank, like Barclays, like JPM, where ultimately the new regulations don't allow these banks to lever up these companies at 9x, 10x, right? And so where their comfort zone exists is in that first tranche, in that investment-grade tranche between four and seven turns, depending on the asset class, towers, fiber, whatever it is. Where there's a real opportunity is in that second piece of capital going into the capital structure, triple B and double B, and that's where our credit team is thriving.
And so the ability to deploy billions of dollars of capital in great tower businesses, data center businesses, where we can get that desired coupon between 9%-11% and get a total return between 12%-15% is really an outstanding product. And so what you see, what you will see from us is obviously we're really successful in the fourth quarter in private credit. That's continued into the first quarter. And I really see that business, as I mentioned to you, it's a business that I think we can get to 20 billion of AUM pretty quickly. And so we're scaling our private credit business because, again, it's about solutions. It's about showing up for customers. It's about showing up for really good CEOs who are kind of between $50 million and $100 million of EBITDA, Mike. That's the sweet spot in the middle market and DigitalBridge has become the preferred lender of choice.
We're closing a bunch of loans right now. We're scaling our private credit platform. I like this combination of private credit, power, AI infrastructure, and having these arsenal solutions that we can offer to our top 20 investors is really what they want. At the same time, going wider, right? Digging deeper into insurance pockets, going after real estate investors, going after renewable energy pockets of capital, which we've never done before. We're finding these swim lanes, Mike, where we can go and raise new capital. In the fourth quarter, we brought a bunch of new logos on our platform. We're doing that in the first quarter. We have to bring anywhere from 25-100 new investors on our platform this year. That's what we're doing. We widen the aperture. We go to focus more myopically and surgically in these swim lanes, and we keep our attention on our top 18 clients, the top 20 allocators, as you said. That's the strategy for fundraising.
Great. So we've got time for a few more questions. I think on my list to give a little preview is we've got a tower question, a data center question, and a fiber question that's coming in from our group.
Good.
If there's anything else that we want to prioritize, please raise your hand. And then we'll get into the fun rapid fire. So first on towers. So tower growth, here's the question. Tower growth has been relatively benign as 5G buildouts were completed a few years ago. With recent developments in AI and potential acceleration of AI applications, when will there be a need for carriers to accelerate investment and why?
So, I think, as I said earlier, we're seeing that already. We had our best January, our best domestic January. And my 30 years of doing towers, my partner, Alex Gellman, and I, we were talking about it, but we had our best January. And by a lot, it wasn't by like a little. It was a significant amount. And I think the way we sort of parse that is it was continued amendment traffic. We're not done building 5G. Actually, that statement is false. We're still densifying 5G networks with overlays. And what we're now seeing is infill. So just like we saw in LTE and LTE+ , Mike, you begin to see as you turn up 5G and you start seeing the pockets, you get the infill macros. So what we saw in the surge in demand and applications was infill macros.
And then the third leg of that stool is BTS. Our JV with Verizon is very important. Our national agreement with T-Mobile is really important. And we just signed a nationwide deployment agreement with AT&T. So Vertical Bridge is building for all three carriers at the same time. So we had actually a very strong start to the first quarter. My suspicion is in talking with some of the CEOs, with Hans and with Mike and some of these other guys and John, they're going to continue to spend capital this year. And I think ultimately, as AI ultimately proliferates to these devices, we see mobile data traffic picking up three to 5x over the next five years.
And I was at an investor conference in Miami about a week ago, and Masa Son was giving a big speech. And he said, "10x mobile data traffic." That's what SoftBank is anticipating with generative AI to the device. And remember, Mike, AI flows through five different ways: consumer, enterprise, you have applications, government, and do not forget about device-to-device IoT networks. That is the fastest area of growth in generative AI, where devices are talking to each other and a human and an enterprise is not involved. We're going from 28 billion connected devices to over 59 billion connected devices in the next five years. The amount of mobile devices that are tethered to mobile networks is exponentially increasing at a very alarming pace.
Another question that we've been getting. I'm going to summarize a couple of questions for data centers and AI. If you look at some of the recent headlines around low-cost AI LLMs or the idea that one hyperscaler may have canceled some contracts, are we at a point where all of this capacity that's being deployed needs to get digested? Is there a risk that there's a little bit of a lull in demand for a period of time? Or do you view kind of, you mentioned this earlier, but how do you view the situation there?
I think people need to remember that the data center sector is characterized by two types of data centers: data centers that are self-performed by our customers and data centers that are performed by the industry. The category where our customers are self-performing, Mike, they're building in remote locations, right? So those LLMs that you're talking about are happening in places like North Dakota and Idaho and Alabama. These are not places where you build highly connected, highly scalable data center campuses. These are singular-focused data centers that are specific for a customer need, and we don't get involved in that. That is 6% yield territory, and that's not where DigitalBridge hangs out. We're looking at development yields between 9% and 12%. We're looking at multi-tenanted campuses, connected in good markets where we can scale and grow.
And so we don't really want to get involved in those LLMs where they're going to be remote and where the residual value from a real estate perspective isn't that sexy. So we try to stay in places where we know we can develop, where the land cost is going to appreciate, and ultimately where we have strong experience in reletting, Mike. That's super important. So for example, in Santa Clara, we've already gone through our first wave of renewals because those leases were signed eight, nine years ago. And by and large, we're getting incredibly high renewal rates in the high 90s%, much like towers. And so that's what we want to see. You want to be in locations where you've got great proximity to permitting, scarcity of power, highly connected areas where the land value is high. Those are the places that DigitalBridge focuses on.
And so in that capacity, when you switch lanes and you come over here to the third-party pure-play portfolio, which is our industry, again, look back at what I've told you around the trade and supply and balance. Until that levels out, until the sector starts delivering lit capacity to where it leases, rental rates are going to be unstable. And in large favor, they will favor the landlord. That's what the industry has shown us for the basically last decade in data centers.
In fiber, it seems like the public markets have had a hard time understanding and valuing fiber infrastructure, maybe relative to some of the private markets over the last decade. Are there changes or parts of the fiber business that you just think are underappreciated in terms of their ability to create value over time?
I think, again, it's a lot like data centers, Mike. Everyone early took one paintbrush and said, "That's data centers." It's not true. There are six different business models in data centers. The same thing applies in fiber today. There are actually five different business models in fiber in terms of how you can make money. And ultimately, y ou've got wholesale fiber to the premises, which are these long-term contracts with MDUs, whether it's an HOA or an apartment building. You get 10-year contracts. That's very infrastructure-like. So for example, that's like our business Fibernow or Hotwire here in South Florida. And that's an infrastructure business if you really think about it. And then you've got the three different enterprise levels of fiber. You've got long-haul transport, you've got metro, and you've got enterprise. Different characteristics, different contract terms, different returns. The music's rounding me out.
Yeah, I wish we had more time, but Marc, thank you so much for joining us today.
Thank you.
Thank you.