Redefining Real Estate in the Digital Age with Marc Ganzi, CEO of DigitalBridge
Willy Walker: Good afternoon, everybody. It is a real pleasure to have my friend Marc Ganzi joining me today on the Walker webcast to talk about all things digital. Digital infrastructure, the digital economy, and how DigitalBridge is meeting the needs of all of the household name companies that CNBC seems to myopically cover every single day. They don't talk much about the DigitalBridge and Walker & Dunlop, Marc.
Marc Ganzi: No, they don't, we are just the boring plumbers.
Willy Walker: They talk about all your clients. They don't talk about many of my clients. Let me do a quick bio, Marc, and then we'll dive into the conversation.
Marc Ganzi is chief executive officer at DigitalBridge and has been building networks since 1994. DigitalBridge today owns 26 portfolio companies that operate in the cell tower, fiber, data center, and small cell digital economy. Previously, Marc founded Global Tower Partners (GTP), which grew to be one of the largest private tower companies in the US before being sold to American Tower for almost $5 billion. Before GTP, Marc co-founded and ran Apex Site Management, one of the largest third-party managers of wireless and wireline communication sites in the United States. Marc graduated from the Wharton School at the University of Pennsylvania as a member of YPO and sits on the Broadband Deployment Advisory Committee of the FCC. Marc is an avid polo player, having won every major U.S. 20 and 26-goal tournament, including the 2009 U.S. Open, playing alongside brothers Facundo, Nico, and Gonzalito Pieres. You gotta love those Argentine names, Marc. Marc’s wife, Melissa, and two children, Grant and Riley also play polo.
So much to cover, my friend, and so little time to do it. I want to make sure we get through all the various things. I'd love to dive in with you, and I know we could spend an entire hour on any one of these. As I was going through and highlighting things I wanted to talk about today, Marc, these are either the words or the quotes from you that I'd like to touch upon in our hour. Towers, broadband, data centers, “Migration from 4G to 5G is extremely complicated,” data to the edge, broadband equity access, and deployment program, gigawatts. And we will spend a bunch of time on gigawatts. “The US will have a crisis in water and power,” 100% recycled water. Reno, six distinct businesses inside a data center: crypto blockchain, cloud migration, and AI migration from language-based models to inference. “We are not chasing China.” Will there be one winner in AI, like Google and Search? And finally, what on earth is a “Feel good version of a Sports Center.”
Let's start here. Go back to ‘94, Marc. And when you were looking at an office tower and saw that one-third of the income came from the actual tower on the top of the building and not the rents inside of the building, and how that got you headed down the path to spending an entire career focused on infrastructure and technology infrastructure.
Marc Ganzi: First of all, thank you, Willy, for the long and exciting introduction to my life. Look, the story starts like every, I think, good entrepreneur: things in life happen accidentally. And at that time, I'd come out of Wharton. I was working for a real estate developer, and we were focused on buying distressed real estate from the Resolution Trust Corp. I've dated myself officially, but as a part of buying some of those, as you remember, in those days, you had to buy pools of assets. You couldn't cherry-pick. You'd go in, and you'd select which block of properties you wanted to buy. And there was a Mid-Atlantic portfolio of assets we had looked at. And you are correct; one of them was an office tower. That was the PSFS Tower in Philadelphia. It's one of the tallest buildings in Philadelphia, but it was built in the ‘60s and part of a package of loans that we had to acquire. And there was this huge tower on top. I really didn't fully understand what it was, but I did know it was producing $700,000 a year of income in a 20% occupied office tower that looked like a good rent. We spent some time diligently on it. And I said, “Well, naturally, I've got to go find somebody that understands antenna leases,” and ended up meeting my partners, Alex and Jeff Ginsburg, who'd both attended Wharton. That's how I'd gotten to them. They were actually building something at that time called Cellular Networks, and so it was the only one I could find for my Wharton cohort list that was actually involved in something tangential to antenna leases. I found these two guys, I met them, I said, “Look, I've got a bunch of these antenna leases. What do I know? I'm a real estate guy. I know nothing about telecommunications or cellular antennas.” They explained it to me, and they explained what was happening at that point in 1994 when the FCC was getting ready to auction off these things called digital PCS licenses. And this was really the first migration Willy from one G to two G. And naturally, as the real estate guy in the room. I said, “Well, okay, this one rooftop is really valuable. How many other rooftops are going to be valuable?” And they said, “Well, funny you should say that. There's going to be 50,000 new towers built to do this new technology called Digital PCS.” And I was like, “Wow, that sounds like a real estate business. That doesn't sound like a technology business.” I stayed in touch with them. We ended up writing a business plan called Apex Site Management. Alex and Jeff are still with me today; 31 years later, we're still partners. One of them runs one of our portfolio companies. The other is my chief administrative Officer. And we've had an amazing three-decade run in trying to find ways, Willy, very simply put, how can you find that intersection between real estate and telecommunications? Everything that I've done in my career has been that intersection and trying to figure out how to translate between the real estate industry, which talks in one language, and the telecommunications industry, which talks in other languages. And by the way, telecommunications morphed into the Internet, into ISPs, and to cloud computing, private cloud AI. As you said, “All the logos that we deal with, all the tenants that we deal with around the world are much like what you do at Walker & Dunlop. It's the same thing. I'm in the tenant service business. I wake up every day, and I know I've got to serve my tenants better than anyone else on the planet does. We've just chosen to take this format, which is serving customers all around the world with their digital real estate needs. Whether it's a cell tower or I've got to go find the land, I got to build the tower. Whether it's a fiber route where I've got to go dig a ditch. What do you have to do when you build a fiber? You need easements. What's an easement? Easements real estate. It's not that hard. You're building a data center. You go to Loudoun County in Virginia to buy land. You have to deal with the zoning board. You got to get power. We deal with all the things that you deal with every day at your firm digitally.
Willy Walker: But in a very distinct vertical and in a vertical that's growing. One of the things that you pointed out is that your occupancy in your data centers has grown by 80% over the last year. And I think your rent is up close to 30% over the last year. I can go out and try to find any other asset class in commercial real estate, and they aren’t getting 80% occupancy growth and almost 30% rent growth over the last year. You're in a very interesting and very hot spot right now. Before we jump to that, though, Marc, let me just for a moment: would you run through… back when you started in the cellular space, you mentioned that there were going to have to be 50,000 cell towers built to keep up with the growth from one G to two G. What was when you started into the cell tower business, what was the projection of the number of cell towers we would need in the United States at that time to build out a nationwide cellular network? And then, what exists today?
Marc Ganzi: In 1994, there were about 8,000 towers in the US, mostly FM radio, cable heads, and towers. There really weren't a lot of cellular towers. People were plopping these other antennas on existing towers or rooftops. In the early days of cellular, like in a city like Philadelphia, someone like Comcast Metrophone or Bell Atlantic Mobile maybe had 50 towers and that covered the whole city. Now, you look in downtown Philadelphia, where each carrier probably has 500 different towers and small cells. The density as technology has gotten more advanced, and we've moved to better technology, from 2G to 3G to 4G to 5G. An initial cell tower in 1994 could cover 20 miles. You go to digital PCS and an average cell towers covering about three miles, and then you go to 3G, which is one and a half miles, and you go to 4G, which is three-quarters of a mile. And then you go to 5G, and a tower covers maybe half a mile. So, as these devices have gotten more sophisticated and as we run more through these, not just video and messaging and Microsoft teams, but I'm talking about active compute. I'm talking about applications. I'm talking about how you conduct your life, which is principally through your phone today. Much of what we do is through our phones, and this is ultimately where the cloud stops and ends, and it's where generative AI will stop and end. We do a lot of these things, and we put a lot of pressure on them. A lot of data is moving through those phones ubiquitously without your intervention and my intervention.
At any given time, I have 40 to 50 apps open on my phone. You have the same thing: whether it's an airline app, banking app, Amazon Prime, or whatever it is, your phone is running in the background. And it's running applications all day. And that data is going back and forth between your phone and your tower. Or if you have a Wi-Fi six hotspot like we have here in our office, you can move off the mobile network and move on to the Wi-Fi network seamlessly. Then you're using Wi-Fi to generate most of your applications, and you're staying off the major network, whether it's Verizon, AT&T, or T-Mobile. We started with an industry; there were 8,000 towers, growing to 50,000 to 60,000 towers. Then, at the tail end of digital, we'd get to 100,000 towers, then 3G, 185,000 towers, 4G moving at 265,000 towers in 5G, and we're moving to 400,000 towers. The industry just keeps growing. And in 5G alone, my customers are spending $1.3 trillion in CapEx. The way you would translate that is TI, which is tenant improvements. And its most fundamental, basic way is what we know is that over the next seven years, our customers will spend $1.3 trillion retrofitting or fitting out the existing infrastructure and building new infrastructure. And what that does is create more opportunities for us to go build new towers, build new small cells, lease land, get building permits, hire lawyers, and hire engineers. There's a whole ecosystem associated with mobile infrastructure. But make no mistake, it looks and feels a lot like real estate project management at its core.
Willy Walker: I think it's really interesting, Marc; as you talk about the proliferation of cell towers as the technology has advanced and we've gone from 1G to 5G, the distance between the towers has actually gotten smaller, not wider. At the same time, the height of the towers has gone from being very tall, as many of us remember back when they first started. And to your point about covering a 20-mile radius, they've gotten actually much shorter these days, down to the point where if you go around Manhattan, you have a company called ExteNet that has these small nodes that sit on every other street corner that allow you to use a small wireless network that allows you to move from one to the next. One of the things that I was fascinated with was the number of nodes inside Madison Square Garden. Talk about ExteNet and those small nodes and how important they are too dense urban centers to keep people connected.
Marc Ganzi: Small cell technology has been around for a while. In 1997, I ran into a guy named John Georges, who created the patent for turning RF, Radiofrequency into optical light and then optical light back and RF. What's interesting about small cells is that it is a very different business than towers because when you go into a venue like Madison Square Garden, we co-manage with the Dolans from a wireless perspective. We have over 180 small little towers, which are called a node. So nodes are like a low antenna, and it's a sector. And a node will literally cover one swath of seats. We have it so surgically pointed and then around the different levels of the concourses in Madison Square Garden, the locker rooms, and the infrastructure underneath it. We've got different levels of nodes and fiber, and it's in, and then somewhere in there is a huge media room. We actually have a server room where all the radios come back to one location, and then that fiber optic cable goes out to each carrier's network. With each customer, we have over 150 distinct radios inside Madison Square Garden. And what that does is when the game turns on, and there are 20,000 people packed in there, everyone has a seamless experience. We have the same thing at AT&T Stadium in Dallas, Texas, which is called Jerry World. We've got NFL stadiums, hockey stadiums, baseball stadiums, and basketball stadiums. And it's a complicated business. Not a lot of people do this. There are only two or three companies in the world that actually do this and do it well to the point where somebody like Jerry Jones will turn over their communications, or the Dolans will turn it over to us and say, “Hey, make it happen.” And we're responsible for the performance of the network. We actually run the RF, so the frequencies of the carriers run on our infrastructure. So, unlike a tower, we build the steel, and we put the antennas up. We get out of the way. In the small cell business, we're actually responsible for delivering the signal. It's a much bigger burden. It's a much higher responsibility. We had to move a little further up what I would call the stack. Customers rely on us. Verizon relies on me every time the Knicks or the Rangers play, we got to perform for 20,000 fans, of which a third are on Verizon, a third are on T-Mobile, and a third are on AT&T. Everyone's phones have to work because everybody wants to upload an image. They want to get on Instagram. They want to livestream something, whatever they're doing. We know we have to deliver a perfect experience for every fan who walks into Madison Square Garden, and that's hard to do. It's not an easy business, but if you're ever walking through Madison Square Garden or any sports stadium and you wonder why your phone works, it's because ExteNet is delivering that signal. And we're doing it through a shared network. That's the other challenge. AT&T, Verizon, and T-Mobile all share the same infrastructure at Madison Square Garden. It's interesting. We're not building three different networks at the same time. We're building one network that encompasses all the different spectrum bands and all the different radios. And by the way, the Knicks have their Wi-Fi on there. The NBA has a signal on that network. The NHL has a signal on that network. New York City's police department is on that network. And we also have public safety there as well. I won't tell you the agency, but we have agencies there, in case.
Willy Walker: I'm unsure. There's a little Homeland Security there.
Marc Ganzi: It’s an interesting business. And again, it's real estate. What did I have to do? I had to negotiate a long-term lease agreement with the Dolans. Part of that lease agreement was taking a demise premise, which is the radio room where I actually have a lease; I have a square footage and then associated with that square footage. I have the right to the chases and conduits and I have the right to place the antennas, and then it's a revenue share. The Dolans are my partners, and it has worked out really well. It's been a great relationship with them. We've had a long relationship for over a decade.
Willy Walker: You talk about Jerry World; if the Cowboys continue to lose, you may not want it to be called Jerry World. He might want to be calling at AT&T Stadium. I digress. As a faithful Washington Commanders fan, I had to put that poke in on Jerry World.
Mark, real quick, I don't want to spend a ton of time on this, but you know a ton about this, particularly given the FCC commission. You said so, the Broadband Equity, Access, and Deployment Program or BEAD, which is part of the Inflation Reduction Act. I read the testimony of Brendan Carr, who was the FCC commissioner for the House Subcommittee on Communications and Technology back in July. And in it, he basically, if you will, poked at the Biden administration, basically saying that $42 billion had been put towards this rural broadband connection program and that almost a thousand days later, not a single home, not a single business had been connected to it. I've heard you talk about this a little bit and how slow it has been in the rollout of it. Should expectations have been higher or is the whole process of laying down cable and connecting rural consumers, the rates that they're going to pay, the subsidies from the government, and all those complexities. After reading Carr's testimony, I went back and looked at both sides of the aisle as they relate to why we are where we are on this. And there's no doubt there's a lot of red tape. There are a lot of rate issues to deal with, and there are a lot of supply chains that have been very slow in rolling this out. But it is quite something that almost a thousand days after $42 billion went to it, we don't have a single home or business connected to it. Anything I'm missing in that read?
Marc Ganzi: Look, it's not exactly as Brendan characterizes it. I think Braedan has his agenda; he's a friend, and I know him really well. Commissioner Carr's a good guy. I think everyone who serves on the FCC is well-intentioned. I think I'm well-intentioned. I think we're all trying to do the right thing. The challenge is the machinery that is Washington. It's not a Biden problem. It's not a Republican-controlled House and Senate problem. It's the nature of how these appropriations go out, which is just slow. And the culprit here is really NTIA. NTIA is responsible for doling out the capital and then getting it out to the states. Then, the states have to go through their appropriations and doling out again through the PSCs. The machinery of how we've organized this was just wrong from the get-go. To use a fiber analogy, the wiring was bad. Because you can't expect NTIA to deploy $37 billion and things like that (snapped fingers). It just doesn't happen. Some of these grants are now being distributed to the states. We've actually won a few of them at our portfolio companies. In Louisiana, Vertical Bridge, one part of the wireless part of this is also creating wireless solutions. It's not just building fiber to the farm. There's actually a process by which other technologies are used as well. We combined a fiber tower built in Louisiana to help rural communities get connected. And that's worked out really well. That was a $100 million grant that was won by Zillow and Vertical Bridge. They're doing it in connection with the state of Louisiana. It's working well. One $120 million grant in Nevada to build high-speed connectivity between Reno and Las Vegas which was really about creating that commerce road. And that's worked out amazingly well. We've got five customers already on that network who have signed up for it. And so it's starting to happen. It's slow. And I think the frustration from Brendan's perspective is, “Could we have gone faster?” Sure. “Are we set up to go faster?” No. And I think, ultimately, there were some things that could have been done to accelerate it. I would have liked to have seen more money for E-Rate. So that's the program that brings fiber to schools, particularly bringing technology to grade schools and high schools where we're very underfunded. We missed that. There should have been a swim lane dedicated specifically to the Department of Education where we could go a little faster supporting our kids and supporting, candidly, schools that are in inner cities, which is the opposite of putting fiber to the farm. I really wanted to put fiber into areas that were underserved from an economic perspective. There are a lot of communities where fiber penetration in the U.S., in the urban core, is less than 20% Willy, whereas the rest of the US is about 87% connected to fiber and broadband. By the way, 87% of the economy is not even in the top ten in terms of fiber connectivity to the home. We have a long way to go. And I'm not pessimistic. I was part of the machinery that helped create the programs. And I think we've got a lot more work to do. Hopefully, this will continue, and we can deploy the 437 billion. It's $65 billion in our total program. But the appropriations are about $37 billion. We'll get there. And the states are moving slowly, but it's coming. And for an industry guy, it is a little frustrating because I'd like to see this money in the hands of school districts. I'd like to see it more in the inner city. We can light up underserved rural communities, but those aren't the people who are screaming for equal access to broadband. The people who are screaming for it are our kids and people who don't have the income to get broadband equality. That's where I would have liked to have seen the balancing act. I'm kind of vocal about that. I don't really like to hide my feelings, and I'm delighted to serve the FCC. It's been a real honor and a privilege to serve with those commissioners. And I got put there by Obama. Trump kept me on. I've stayed on through the Biden administration. And the thing that I love about what I do, Willy, is what we do isn't red, it's not blue. Digital infrastructure tends to be purple. We do have an industry where I think the aisles come together, and we do recognize that there's a purpose to what we do, which is providing connectivity and infrastructure for all Americans, irrespective of political affiliation. That does make me feel good at the end of the day.
Willy Walker: Let's dive in Marc to data centers. I want to look back to Reno because you just mentioned Reno and why Reno has become such a hotspot as it relates to the growth in data centers. But you're currently the largest private owner of data centers in the country. You've got 210 data centers that consume three gigawatts of power, and you have another hundred under construction that will consume another three gigawatts. When I listen to you talking about the size and scale of all this, to be honest with you, when I hear three gigawatts, I don't really know how to size that to get a sense of how much power is that and what do we use? I went out and pulled a couple of things together. And I want you to dive in to help everyone understand how important power is to this entire industry.
A lot of people have heard about Microsoft's investment in Constellation Energy to get Three Mile Island back online. In looking up what Three Mile Island generates, we see that it generates about 800 MW of power per reactor. And so you'd need two, three-mile islands or four reactors from Three Mile Island to power what you have already in data centers in the United States, or 2x that to do what you're currently building right now. And even when you think about that, I think about a big nuclear reactor, and I say, “Okay, it's got to be producing a lot of energy.” But try to scale for people just the intensity of power consumption at data centers today. And why the quote that I said previously that we are going to have an energy and water crisis in America is so important for the growth of this industry?
Marc Ganzi: Well, look, you framed it right. And you're using Three Mile Island as the analogy. I'll tell you the way we describe it: New York City consumes five gigawatts of power. New York City, the biggest city in our country, every day, our data centers, we wake up, we consume about 4.1, 4.2 gigawatts of power. We're building another. We're bringing another 3.7 gigawatts online right now. So we'll be pretty darn close to eight gigawatts of power, which would be a factor of about 1.6 times the power that New York City consumes. Just to give you a sense of a frame of reference. And our pipeline today is about another five gigawatts behind that. In theory, we're going at 12 to 13 gigawatts, which again continues to add to our assertion that we're the largest.
Willy Walker: That's just you?
Marc Ganzi: That's just DigitalBridge.
Willy Walker: There was something that I heard you talk about, Marc, that Elon Musk's prediction is that the data center industry will acquire 80 gigawatts of power once it gets fully scaled. Is that where we're heading? Is that the right sizing of it?
Marc Ganzi: That's right. Sam Altman will tell you there are 50 gigawatts needed to power AI, and Elon says, “80 gigawatts.” And I think Elon's number is Sam's number plus public and private cloud. AI will consume 50 gigawatts. Private and public cloud will consume another 30 gigawatts. The industry sizing is about 80 gigawatts. And where we are today, probably at about an industry of 15 to 16 gigawatts. We were a year ago at about 13 gigawatts. I think we've added another three gigawatts of capacity this year.
Willy Walker: And look, for two seconds, as a reference point, you mentioned the public cloud. Today, the public cloud uses about 13 gigawatts.
Marc Ganzi: Yeah, that's about right.
Willy Walker: Public cloud and data centers are about the same right now, where data centers have now jumped up to about 15 gigawatts, growing towards that 50 or 80 gigawatt number, whereas private cloud continues to grow at an accelerated pace.
Marc Ganzi: Private cloud is growing super-fast. Okay. You've got three swim lanes of growth happening at the same time, and then all the data is a fourth one that really confuses you and messes up the whole conversation.
Let's start with the public cloud because it's something we all know. There isn't a day that goes by that none of us… we interact with, whether it's Amazon, whether we're logging on to Microsoft Azure, we're logging here on Zoom, and we're doing this call on Zoom. We all touch the public cloud every day and all these applications operate in the US cloud or they operate in Azure Cloud or Google Cloud, or Oracle's Cloud. And folks like Zoom use one of those clouds to create this opportunity for us to talk. And that's been going steady and hard since 2011. The public cloud started over 12 and 13 years ago. And so we're probably 70 to 80% built in terms of the public cloud infrastructure. And it's consuming about 13 gigawatts. It's the biggest user of data center capacity today is the public cloud.
Now, what's a private cloud? The private cloud is much like the public cloud, but it's a private cloud where only your data sits, and it's very secure. And we have a company called Switch, based out of Las Vegas, that does it. Equinix has a little bit of this private cloud, and what's been happening and the reason Switch has exploded and grown so fast is big corporations and big government agencies they've lost a little trust in the public cloud. It's not to say they don't put those workloads out there, but there are certain workloads that Federal Express, American Airlines, JP Morgan, Dell, or the U.S. government want to know that those workloads are private, they're sitting in a very secure location, it’s not commingled with the public cloud. It's just staying resident through their own VPN, running through an unsecured VPN or a virtual private network connection, and you're accessing that data only for your internal use.
For example, a lot of banks are now creating for your financial data, a private cloud just for Willy's data. Willy Walker's financial data, let's say your bank at JPMorgan, is going to sit in a private cloud. If it’s in the public cloud, that information sometimes gets corrupted, and people could jump into the public cloud, take their financial data, and pull it out. There's been a lot of cleanup related to what we call reverse cloud. The theme for the last decade has been every CIO and every CTO, including yours who runs Walker & Dunlop, is like, “Hey, Willy, we got to get everything out in the public cloud.” And so you guys quickly migrated. I don't know what platform you picked. You picked one of the platforms. But Walker & Dunlop operates probably on one of the public clouds. I don't know which one you guys use, but whether is it AWS or Microsoft you use one of them. The problem was your CIO, probably a couple of years ago, said, “We have all this client data, all this landlord data, we have all this tenant data, we have this property management data, but we have to keep it secure.” And so some corporations have said, okay, I'm going to put my active compute, which are my applications, in the public cloud. So the stuff that helps us run Walker & Dunlop every day, we're going to move that to the public cloud because you have to it has to be nimble, it has to be flexible. And all your employees and all your clients have to use it. Now, the private cloud pieces, Walker & Dunlop financials, Walker & Dunlop client data, your tenant data, and your landlord data, all of those have to be siloed off and secured. Why? Because at the end of the day, you get hired because clients trust you. They trust you to manage their real estate. They trust you to do their leasing. They trust you to do their brokerage. Whatever it is that you do, it's a relationship based on a fiduciary level of trust. That trust is only as secure as the data secure. And you've got to keep your client data very secure. Maybe you're not doing this yet. But guess what? Dell is. Federal Express is, and Oracle has a private cloud business. And so you're seeing a lot of this very secure data that needs to be pushed into a private cloud environment. And you've got a public cloud, you've got a private cloud, and now you have AI. Along comes AI, this monster, massive amount of compute where we're building these language-based models, where we're getting smarter and smarter data sets that are beginning to think for themselves and move to what we call inference. Inferences when data starts thinking for itself.
And so along comes AI, and Sam Altman's really smart. I agree with him. There are 50GW of data centers needed to build AI. And so you got the 50GW of AI, you've got the other 30GW of private and public cloud. And you just have the sector just exploding. And all of that Willy goes against an environment where land is tougher to get, permits are tougher to get, pour is tougher to get the right fiber solutions, and data centers are tougher. And so our sector is coming of age, but it's coming of age in a moment where resources are getting more and more constrained. And then water. You're using a lot of water to chill these data centers. Where is that data center located? Oh boy, if it's in Nevada, guess where that water is coming from? It's being leased from Colorado. It's really expensive. You're going to tell me you're wasting the water. Why can't you recycle and clean up the water through osmosis? And so our business, it's beautiful. It's great. It's exciting. It's moving really fast. But it has complications. And that's where the specialization comes into play. It is real estate, but it's incredibly specialized real estate.
Willy Walker: Two things I want to dive into before we move to get under the hood of data centers and the six distinct businesses inside of them. And you actually talked about three of them right now. But I want to dive into that a little bit deeper. But as we talk about energy, a couple of things. First, you have a partnership with Microsoft, which I believe is in Oregon. We're using 100% renewable water at a data center. And I think I'm right on that. Is that correct?
Marc Ganzi: We do that? Yes. It's actually in Santa Clara. It's in Silicon Valley.
Willy Walker: Switch, which you talked about, is 100% renewables. Switch is this private cloud provider of all those applications that you just talked about, Oracle and Dell, and all that. I think it's 2016, and it has been fully renewable. Is that a competitive advantage to Switch versus the other competition that is out there? We want to turn off the grid.
Marc Ganzi: 100% It is. We've been able to do that, and it creates a set of solutions for the customer who wants to get carbon neutral as fast as they can. Getting to net zero is a high priority for Amazon and Microsoft. By the way, the same thing for American Airlines, JPMorgan, and the US Federal government. When we bring a customer in a Switch, we're providing them with a lot of connectivity and a ton of security, and we're giving them 100% renewable power. That is why Switch is doing quite well: it's in the solutions business. Everybody wants to be in the solutions business, and that combination of security and clean energy is really important. Those are two things that the customers at Switch really value. And look all of our data center businesses are following that model now. Whether we're building our own microgrids adjacent to solar, wind, and hydro or whether we're providing a PPA directly into the customer, where we bring that renewable power, bring it onto the grid, then we bring it through the grid back into their data center. We're getting creative. We're learning. We're getting better at this. I've committed my portfolio to be net zero by 2030. It's going to be hard because even a data center that's 100% renewable still uses a little bit of grid power to equalize. Because you have this thing called intermittence. Solar, wind, and hydro are unpredictable. They're not perfect sources of power. The sun's not on 24 seven. The sun goes away for half the day. Wind can die down. And sometimes, with hydro, we don't get the results we want. So we've been honest with our investors. We've been honest with our customers that we do have 100% renewable data centers. But ultimately, all of our data centers do have a backup, which is batteries, and another backup, which is generators; and then a third backup, which is connected to the grid. We can never go down, Willy. It's really important to us that the promise or the commitment we make to our customer is the power is always going to stay on. That's critical.
Willy Walker: As you talked about, where are data centers going these days? You mentioned Herndon previously. Virginia has done an incredible job of attracting a lot of data centers, particularly in Northern Virginia. Duke Energy has also been a big partner there. I looked up Duke Stock, thinking about derivative investments based on the data center surge, and Duke's up 27% over the last six months. I think the number Marc is that 24% of the electricity generated in the state of Virginia today is being used by data centers. Almost a quarter of the total energy generated in the state of Virginia is going to data centers. And as you said, a number of data centers are moving to other parts of the country where electricity and water are more abundant or potentially cheaper. You mentioned Montana in one of the interviews that I'd heard you talk about. I went and looked up Northwestern Energy in Montana, and their stock prices have been up 18% over the last six months. But talk for a moment about the investment play of buying DigitalBridge and investing in Nvidia as it relates to the chips that are powering the AI revolution, etc. But as it relates to the infrastructure behind what you're doing, there's the construction, there is the power, there is the water. I looked up Corning and Corning is up 63% over the last year. There is a cable that's coming into all these places. Talk for a moment about those derivative investments.
Marc Ganzi: Look, we call it an ecosystem. And I think whether it's investing in power generation like Siemens or Caterpillar, those are good tangential stocks. Those guys are backed up for two to three years on their orders. And a lot of people don't think of power sets. But ultimately, remember that at a data center, we are investing in the power and the gensets in those backup resources. So whether it's Kohler or Caterpillar or Siemens, we have these sophisticated power systems that provide the primary power but also the backup power. That's one of those interesting adjacencies. Certainly, Corning's an interesting play because they're a leader in manufacturing fiber. Dicom is a company that does a lot of fiber installation work. A lot of people don't really look at Dicom. That's kind of an interesting business, and whether it's core, we've NVIDIA, Arm Allied, all these guys that are building GPUs, Intel, that are building chips and GPUs, that's also super important as well because that part of the ecosystem is growing really fast.
As you know, NVIDIA has been backed up for three years. Arms due to deliver its first chip in 18 months. MOS is putting a ton of money into that. Core is an outsourced provider of GPUs. We're a big lender to Core. We do a lot of work with them. We do work with all these guys. These are all customers of ours, so we know them really well. And I would just say, “From an ecosystem perspective, all these businesses are performing well, and they need power, and power is in short supply.” The reason why a Constellation or a NextEra, a Duquesne Power Light, Puget Power, and all these utility companies have record orders in terms of new commitments to providing power. And on those PPAs, the prices are going up because, at the end of the day, Microsoft and Amazon want to be locked in for 15 years on power. And it's really a great leverage moment for the utility industry. And they know that they're playing their cards correctly. And at the same time, the DOE on the other side of the fence, we've spoken to the DOE, and we talk to them all the time. They want to make sure this doesn't hurt consumers. So, ultimately, we want to see utility prices stable. We don't want to see it go up because of data centers. And there's a lot of conversation about how the data center sector can contribute to the grid. How does it become a good partner to the grid? And I think this is something the DOE is super focused on, which is why you see all these forms of alternative energy popping up. Which is how do we not disintermediate or disturb the grid and maintain our artificial intelligence superiority in terms of where we're going? We are in an arms race to build the best AI models.
Willy Walker: Final thing, before I dive under the hood, if you will, of data centers, have you quickly describe the six different businesses that run inside of them because I think it is fascinating. Most people, I think, think that data centers today are all about AI, and they're actually distinct businesses in there. Then, you'll see the pricing model that you've changed to. But before we jump to that, one final thing on location.
I mentioned Reno, Nevada. Reno, Nevada is, I believe, two milliseconds from Santa Clara in terms of latency. One of the interesting things here is that latency as it relates to AI, really isn't a huge issue. In other words, when we read Michael Lewis's book on trading and trying to front-run trades, the location of the data centers is exceedingly important because two milliseconds is two milliseconds too late. Whereas in AI, the location of that two-millisecond latency between Reno and Santa Clara isn't that big of a deal. And yet, at the same time, Reno, as a city and as a state, is allowing for the building of data centers from a regulatory standpoint, from a permitting standpoint to grow very dramatically versus in California, maybe not so much. Is that the reason why it's all going to Reno? Because power can't be that much cheaper in Nevada than it is in California.
Marc Ganzi: It is cheaper.
Willy Walker: It is?
Marc Ganzi: It's materially cheaper, first of all. And it is way cheaper. Power is cheaper. Zoning is way easier. You don't have to build everything earthquake-proof like you have to in Santa Clara. Our construction costs go way up in Santa Clara against Reno. And look, the reality is you can't get 50 megawatts of continuous power to a piece of real estate in the Bay Area anymore. It just doesn't exist. There are two power providers: Silicon Valley Power and Pacific Gas & Electric. And if you can't get that power, you don't even option of the land. You don't even try. It's nothing to try. So along comes Reno, and Tesla gigafactories there. The Switch Super nap is there. Google's going there. Other companies are going there to build data centers, a lot of renewable power. We have a ton of solar they're coming into our data center. And you have a state that's like, happy to see you. Nevada is like, “Hey, come to Nevada.” California's like, “How can we tax you and make your life more difficult?” California is literally a very difficult place to visit. Love California. It's a great place to visit. But they've made it literally impossible for you to do business there, which is why we moved our headquarters from L.A. to Florida. We wanted to be in a state that actually wants us to be here and wants us to do more business here, not less business. I think you have to create the right environment. You've got to have power, land, zoning, a proactive government that wants you there. And that's why people have flocked to Reno because it's a better submarket than Santa Clara.
Willy Walker: And there's obviously a huge number of jobs that are involved in building data centers. But then, at the end of the day, once the data center gets up and running, it actually is running servers that, for end-use, might disintermediate human beings or make jobs go away. Where do you play out on that as it relates to the impact of technology broadly and AI specifically as it relates to job creation as being a threat to the American workforce?
Marc Ganzi: Look, I think the labor force, in general, Willy is shifting. I think the jobs that you and I coveted coming out of college 30 years ago are very different from the jobs that kids are coming out and getting today. There's a whole different set of skills. And again, it is related to digital. IT jobs, web programming jobs, and database management jobs. Even in big corporations, big IT organizations are built to run the data and keep it secure. If it's a car manufacturing company, make sure the manufacturing line moves more efficiently and faster. So I don't think digital is disintermediating job. I think there are a lot of job creations in digital. There are many jobs creation in the management or the asset management of digital assets that didn't exist ten years ago and didn't exist five years ago. Look at cryptocurrency by example. That's a whole industry now.
Willy Walker: But that you've got out of. You don't like crypto. You had a crypto; you had a mining company.
Marc Ganzi: It's not for us. Crypto wasn't for us because the cash flows were too short. They're kind of month-to-month cash flows, and we don't really own a lot of real estate where the cash flows are month-to-month. If you look at a data center at Switch, we're signing a 10 to 15-year lease. You look at the crypto mines; we own a nice one that we sold. Our average customer was there for less than 30 days. The rents are really high in a crypto mine, but the customer's volatility is really strong. So you're getting a lot of churn and a lot of turnover. And to be honest, I kind of like a data center with five customers with 15-year leases versus 50 customers with one-month leases. That seems like a better piece of real estate for us. And so, look, we built a great business in Iceland. It was very valuable to somebody else. They paid us a high teens multiple. And we said, “I'm very happy at a blended high six, low seven cap rate to get out of a piece of real estate. That's all month-on-month leases. You would probably make the same decision as well. Faced with that data, we made 3X our money.
Willy Walker: I'm surprised you're also not big on the blockchain. And it's so interesting in the sense that a lot of people have talked about the blockchain being the future of secure data and therefore, being able to get around.
Marc Ganzi: I'm a big proponent of blockchain.
Willy Walker: You are.
Marc Ganzi: It's not something we invest in directly. Other people are smarter and better at it, and venture and private equity do that. We just are not set up to do that. And so we have friends that do it at Vista, that do it at Kleiner Perkins, that do it at Sequoia. There are guys understand blockchain way better than we do, so I'd rather give my money to them and let them invest in blockchain. And the same thing. A lot of those partners that tech firms invest with us because we're really good at building a data center. But make no mistake, blockchain technology sits in our data center and runs on our fiber networks globally. So we are part of the blockchain, but we're the infrastructure or the plumbing behind it.
Willy Walker: Just real quick on data centers. And when someone walks into one, many of us have never actually been into a data center, but the pictures we have typically seen are racks of computers.
Marc Ganzi: Think blinking lights, blinking lights.
Willy Walker: Lots of lights and lots of cables. And I've heard you talk through sort of the six distinct businesses inside of it. If you could just run through quickly from racks and servers to hybrid cloud to enterprise data center, run through that. And then more importantly, how you've changed your pricing model, because I found this to be super interesting as it relates to charging for space and now charging for energy.
Marc Ganzi: Look, the data center sector is really six different business models wrapped all into one. The ecosystem played out all the way down the IT stack, and you have the first business model, which is called managed services. That's basically your small business. You don't need your own IT infrastructure. You hire someone to manage your whole IT stack. And it's not a great business that trades at about six to eight times EBITDA, but there's a place for it. People need to have their IT managed, and so we call that managed services.
You move up the stack to the second swim lane, which is called Hybrid Cloud. So Hybrid Cloud is a company that will host your cloud for you, whether it's private or public. They'll also host all your IT Infrastructure. It's an integrated solution that gives you the cloud and your own internal operational flexibility. We call that hybrid cloud.
The next move up is called enterprise. So, enterprise data centers have been data centers for around 25 years. You're Florida Power Light, your Georgia Power. You want to go put your servers in somebody else's building, and you want to stay out. You don't want to own your own IT infrastructure. So big enterprises and big corporations will go put their servers in somebody's data center. And that's called enterprise or co-location, as it were. Good business trades at 8 to 12 times, maybe 13 times. Contracts are sort of three to five years in duration, where managed services are not clouded, and the contracts are year-to-year. So, the shorter the cash flows, the less valuable the business model is. The last three verticals are quite valuable.
So, the fourth business model is called Edge Compute. So edge computing is where you have these data centers. They're located in secondary and tertiary markets, sometimes in the suburbs, where the cloud customers and private cloud customers are starting to deploy technology on the perimeter of the network. As the workloads start to shift and move, it's not just in New York, San Francisco, Atlanta, and Dallas. You start to see these secondary markets form, and that's called the edge. And you're starting to put resources on the edge of your network. That's a very good business. That's a publicly traded REIT called Equinix, and that's our company called Databank. They focus on the edge and provide that connectivity that goes from a core market out to an edge market. Very good business. And those businesses are trading at, you know, 26 to 30 times EBITDA or AFFO. Equinix is a really valuable REIT, as you know. It does a great job.
Then, you move into the public cloud. So, this is called hyperscale data centers. These are data centers between 50 MW and 200 MW in a campus setting and they're principally built for Google, Oracle, Microsoft, Amazon, Meta, and Apple. Everyone that's operating in the public cloud is the hyperscale. And the hyperscalers like to be in areas where they get a lot of power and privacy, and that's kind of the fifth vertical, and leases generally, there Willy are like 10 to 17 years in duration. That's good business. A lot of capital like KKR, JIP, Blackstone, Macquarie Equity, and Brookfield, are all putting money into that vertical because of its infrastructure. It's real estate, it's infrastructure. A long-term lease is very safe. You can't get hurt.
Then, the final frontier is the private cloud. That's where Switch and Equinix has a small private cloud business as well. That's one of the fastest growing verticals, really sticky, highly secure, Tier five, difficult to build, high CapEx, high barriers to entry. And those businesses, again, trading at 28 to 40 times EBITDA. Very valuable and very difficult to replicate. 70% to 80% of the tenants are investment grade. Long-term leases, high investment grade concentration, and really difficult technology to replicate. Those are the six different verticals that exist in the data center space today. If you're thinking like a landlord, how do I play in the data center space? Those are the different business models that you can play in.
Willy Walker: Super interesting. And when I hear you lay it out that way, it gives me a really good sense of when we hear about data centers and we sort of think of just AI, and we don't really understand what's going on and where the actual scale is. And then also when you laid out what those different businesses sort of trade at. So much of that is based on those long-term contracts. It's just fascinating. Thank you for laying that out.
One of the things you know better than most, Marc, is the AI Arms Race. And a lot of people talk about who we are trying to catch up with. And China seems to be out there as the likely adversary, if you will, in the AI Arms Race. And as I said at the top of the call, you're not a big believer that we're in an arms race with China. Why do you think that's the case?
Marc Ganzi: I think they're hyperscalers, and they've got their AI national champion, who's building their infrastructure. Remember, AI is about managing data. It's about putting data into models that allow it to learn, self-learn, get to inference, and then get to generative AI. The journey we're on is probably longer than ten years. And ultimately, it's not about do I have the data centers. Do I have the power? Do I have fiber? But ultimately, do you have the data? You got to have data. Data is a whole new commodity. And so given that most of the great businesses in the world are based here in the U.S., we have the data. We have the advantage. We're also the leader in entertainment, in music, in content. And so we're voracious consumers of that content, and we create content every day. So, creativity and having data are commodities. And it's not to suggest that China doesn't have its own content. It's not to suggest that China doesn't have its own great corporations in its own great data sets. But at the end of the day, all of those things are owned by the state. The state controls that data. Then, the states are going to build their own language-based models. And based on that, the data is going to get to a certain set of outcomes.
I'm thinking more down the road, perhaps than others are thinking. My belief is probably more in the Elon belief that ultimately, the country that moves the fastest to protect its data and manage its data and creates those models in a safe environment that's secure and allows those models to grow in due course. And you feed it, you're going to get to the right set of outcomes. But remember, it’s the old adage that I learned at Wharton when I went to my first computer sciences class. I think in 1993, was GIGO, garbage in, garbage out. If the data is not good and the data is not clean, you're going to get a lot of false outcomes. So in an environment where the state is protecting the data and shaping the data, and you're manipulating the data, you're really not allowing AI to grow the way it needs to grow. And remember, it has to grow. It has to self-learn, so you can't manipulate it.
There's a great book called AI Superpowers. Everyone should read it. It takes you back 30 years in time to the invention of AI, how it came about, and, ultimately, the different schools of thought that have evolved around, do you control it, do you let it go. Then, the evolution of chat and Open AI and what the Chinese are doing with their models. And look, I think we got to run our own race. I don't think we need to be consumed with trying to compete with China to determine whether we get AI superiority. I think we have a lot of advantages. I think we have great corporations and great data. And it's really our ability to continue to feed that, foster it, and grow it in safe environments with the right data sets that are going to create the right outcomes. I'm very skeptical of China's ability to grab and weaponize data from Democratic-controlled corporations that aren't going to give their data to China and they will have to go steal that data. They have to manipulate that data. Ultimately, he or she who controls the data is the one that's going to be able to put that into the right language model, that allows it to learn, that allows it to get to inference, and then allows you to move to generative AI. I don't want to belittle the audience. It's just complicated. Because there's so much data and the ability to manage that data and the ability to weaponize that data.
I think in a military context, it gets more scary. I think I'm thinking about it as a businessman, an entrepreneur, and somebody who runs, I think we have 46 companies today. I'm always thinking about how I keep my data secure and how I make my businesses better and more efficient. Every CEO struggles with that today, which is the efficiency of AI against opening up my data and allowing it to potentially be manipulated and corrupted. And so I encourage people when thinking through these decisions to do their homework, go slow, and don't have to go fast. Everyone thinks like we're in this rush. You're not. You can take your time. Ultimately, you and I will sit down three years from now, and 80% of your business will be run by generative AI. Whether it's brokerage sales, property market, whatever it is. A lot of your decision-making, a lot of your grunt work will be handled by models that will make those decisions, that will enable your executives to be more effective and to move faster.
Willy Walker: Yeah. The final one, because we're almost out of time, is you don't believe that there is one winner in AI. We've talked about telepathy and you talked about those three major cellular carriers. And he sat there and said, “Verizon, AT&T, and T-Mobile, there are only really three of them. Yes, there are some others, but they're really only three of them.” In search, we really have Google with eight, with Microsoft playing a kind of bridesmaid role, but not really very much, though. But yet at the same time, you think that we will have competition and that there will not be one dominant player in AI. Why do you think that is?
Marc Ganzi: I think there are different platforms. I think there are different models being built. I certainly think what Sam is doing is interesting, and he's got a lot of money, but don't sleep on the other players. Everyone has the capability to build their own models. Everyone has the capability to proliferate that technology. We've got a partnership with Intel called Articulate, where we're building our own generative AI business models, and we're killing it. We're making money. We're signing more accounts. Our monthly recurring revenue has quadrupled in less than two quarters. There are other people out there doing it. Nobody has a patent on AI. You've got to build the models. You've got to put the time in and you got to have the data. If you don't have the data, you don't go anywhere.
Data is the new currency, Willy. If you have your data, keep it safe, and control your data it is your corporation because eventually, you're going to be able to weaponize that and really enable you to create comparative advantage.
Willy Walker: Super cool, Marc. I appreciate you taking the time. It's nice watching everything you've done. You came out to Sun Valley and spoke at our client conference two years ago and had everyone's eyes spinning.
Marc Ganzi: That was super fun. I brought my daughter that day, and I was, like, so excited to, like, she sometimes watches me public speaking, but I think that's the first time she'd been where I'd gone to a conference. I was like, “I'm so excited she's there.” And, of course, you put Jamie Lee Curtis in front of me. My daughter is like, “Jamie Lee Curtis was awesome, Dad, you're okay.” But Jamie Lee Curtis is way better. The whole plane ride home, I had to figure out why I wasn't as effective as Jamie Lee Curtis. But that was super fun, and I'd love to come back. It was great conference you run an amazing conference.
Willy Walker: I'm super appreciative of you taking the time today. You're coming to Sun Valley, and I look forward to seeing you sometime soon. Thank you so much. And not that you need it, but great luck to you and the team at DigitalBridge for all your undertaking today.
Marc Ganzi: Thanks, Willy, I appreciate it. Be well. Take care.
Willy Walker: Bye.
