All right. Welcome, everybody. Last year, Marc Ganzi, the Founder and CEO of DigitalBridge, he had a keynote for telecom infrastructure and outlined tremendous themes, and so we decided not to make the same mistake this year and invited Marc to kick off the conference, particularly from a telecom infrastructure standpoint, because he really just has such a unique perspective. So Marc, thank you very much for coming. Thanks for helping us to kick off this great conference.
Yeah, thanks, Brett. Good to be here again.
You have a very different business model in terms of how DigitalBridge has been designed to approach investing in telecom infrastructure, not just domestically, but globally. I was hoping you could maybe just start off and remind the group here about the structure you've put together, and then we can start talking a bit about some of the business opportunities that you've positioned the company to pursue.
Yeah. So look, we were three years ago, we were a REIT. We were a diversified REIT, investing in commercial real estate and in digital infrastructure, and we made the decision to focus exclusively on investing and owning digital infrastructure. We thought that was the right direction to travel.
I think what we couldn't anticipate three years ago was the velocity of the CapEx at which our customers were spending really to proliferate 5G networks, private cloud, public cloud. We'll talk about AI in a little bit, but just the CapEx has changed so radically, Brett, since you and I have known each other 25 years.
Think about, you know, a big year of CapEx was $100 million or $200 million, and this year, you know, we'll put $7.8 billion of greenfield construction to work around the globe, and that quantum of capital just seems to keep growing and growing, growing. So the business model had to change, and I think what we offered to investors about a year and a half ago was perhaps that you could invest and own in digital infrastructure in an asset-light model. And at that time, I think the story was a transition story. It was widely misunderstood, and I think 18 months into that journey, I think investors are now beginning to appreciate what we're doing. Because in this environment, where capital is so scarce and difficult to aggregate, our model is working. We're able to show up for customers.
We're able to meet those demands, and to be candid with you, we're playing offense. And I really like, in an environment like this, the ability to play offense. We've got a strong balance sheet. We're still investing out of some of our private funds. We have an incredible co-invest program, where we have the top limited partners in the world, the top institutional investors in the world, that partner with us and go deeper into specific opportunities, and we can talk about that a little bit, too. But the ability to go anywhere and the ability to form capital and to continue to show up for customers, that's our mantra right now. And look, it's, it's working.
You know, I think our ability to deploy not only the $7.8 billion in greenfield CapEx, but at the same time, we're raising $8 billion of equity this year. We'll place somewhere between $7.5 billion and $11 billion of debt this year in a very challenging market across securitization, private markets, and institutional bank markets. But it's again, this ability... What I've discovered, being a CEO for the last 30 years, is it's in these difficult environments, if you can raise capital, advantage. Advantage you, 'cause it enables you to continue to perform for customers, but it also allows us to go out and take advantage of opportunities that are gonna reprice themselves, and I think we're pretty excited about that.
So capital formation has been really important, and look, at the end of the day, our peers have followed us. I mean, if you look at what Equinix did in xScale, they went out and got GIC. You look at what American Tower had to do to fund CoreSite. They had to call on, you know, Stonepeak to go fund that transaction. Digital Realty is doing a series of private capital transactions, and so this is the direction of travel. I think if you're gonna be in digital infrastructure, and you're gonna make big commitments with customers, you've got to have the ability to form capital. And for us now, the story's clean. Our balance sheet is clean. We've made a commitment to deconsolidate our last two assets, which is DataBank and Vantage, which are both on schedule.
Having a clean balance sheet and having a clean story, investors are now, coming into the story in a, in a significant way. We've had a great year, in terms of performance. We had a really clean second quarter, and we anticipate the quarters will continue to be clean from here on in.
Just in terms of that last point on deconsolidation, targeted timeline, and then, to the extent proceeds come in, how would you think about recycling or reallocating the proceeds?
Yeah. No, it's, it's a good question. So, we've been really clear, DataBank will be first. That'll happen inside this quarter. We've been really direct about that. Vantage SDC, which is our stabilized data center REIT, private REIT that we own on a balance sheet with some institutional capital, that'll also fall probably somewhere straddled between the third and fourth quarter, but a clear plan to get that done this year. And in terms of recycling capital, I think our best ideas are continue to invest in ourselves. And so investing in ourselves really involves sort of four quadrants of thinking. One is buying our shares back, the second is paying down indebtedness, the third is continuing to reinvest, in our funds group, and then the fourth is accretive M&A. We've been really clear about that.
There are other digital-like GPs out there that also subscribe to our asset-light model that are subscale. And so we've been looking at a series of different acquisitions out there, and as multiples have come in a little bit in our peer set, those acquisitions get more attractive and more accretive. So, you know, my philosophy in managing the balance sheet and deploying capital is to be incredibly pragmatic, have a high hurdle rate. Our hurdle rate has gone up because of the cost of capital. So I think you're gonna see us do all of those things, and we've demonstrated our ability to pay down debt, to buy our shares back, to do M&A, and then last but not least, continue to invest in our funds, which are performing really well.
We're trying to get a slide up that gives maybe a global picture of what the business looks like, and you alluded to this: you really are the only way that public investors can invest in a business that has global exposure to telecom infrastructure. And your portfolio companies hit, I think, every type of asset class within telecom infrastructure, and you're in most of the key regions. What is it about being global and diversified in your portfolio that you think is the right model, when historically, the public companies have been very singularly focused on one thing with one type of corporate structure, generally a REIT structure?
Well, look, I don't think the traditional way of owning digital infrastructure was the wrong way to own it. I think there was a myopic approach to towers and fiber and data centers that has made sense for the last 20 years, but the world's changed. In a software-defined environment, where our customers are buying different parts of the ecosystem in different parts of the world, the ability to show up and manifest ourselves on a global scale, $74 billion of assets managed globally, 40 different investments. We operate in Asia, we operate in Europe, we operate in North America, now have built two campuses in South Africa. So, you know, it's just for us, it's just the direction of travel at the end of the day. Our customers just don't wanna buy one rack in a data center.
They don't wanna buy one peering point. They don't wanna buy two strands of dark fiber. They really want you to show up and be able to deliver holistically, a network. And that's, that's happening. I think we've, we've seen situations where we can show up, you know, with Zayo, with DataBank, and Vertical Bridge, and go build a network for somebody. And that ability to deliver a network, I think, is, is what is the changing narrative. That's gonna be incredibly important in AI, and we'll get to that in a second, but I think, you know, everyone thinks, okay, AI is just about big hyperscale, you know, high power density, you know, big campuses. No, it's more than that. There's a lot of different aspects to delivering AI, and so we wanna be the landlord of choice to the world's leading logos.
We have rapidly become that in the last three to four years. Just at the velocity of the capital that we've put to work and the success-based CapEx that we're deploying, we're just getting more at bats than the strategics. Because we have the geographic footprint, we have the ability to show up with a product and go anywhere, and we have the capital. So when you can deliver geography, and you can deliver network, and you have the capital to invest in customers, you win. And I can guarantee you, our competitors can't say that today, but we can.
You referenced earlier the nearly $8 billion of capital that your companies will be investing this year. Where is that going? What’s being overweighted right now?
So that'll be about 65%-70% in data centers, and that'll be distributed, you know, probably about 70% into public cloud and 30% into private cloud. And when I say private and public cloud, keep in mind, DataBank does both, from an edge perspective. So, we're deploying a lot of that capital, as you can imagine, behind the world's most important hyperscale, logos, who are deploying networks, you know, all around the world. I think if you look at that CapEx that we're gonna deploy, we currently have, roughly about 2.8 GW of in-construction activity right now, so that's, you know, a lot of power.
And again, we're delivering that on four different continents at the same time, all very synchronized and going, you know, in the direction of travel that the customers want. And there's more behind that. We've actually seen our leasing backlogs at DataBank and Vantage and Scala and Atlas Edge and AIMS. All of our data center platforms are up anywhere from 2x-5x. So to me, backlog is really the health of any business, whether it's towers or fiber or data centers, your leasing backlog really tells you the story. Just take, for example, DataBank, which is our domestic U.S. edge computing business.
Their backlog has gone up 5x in 12 months, and we literally can't build data centers fast enough to keep up with the edge demand that's happening in these secondary and tertiary markets, as public cloud is now proliferated out to the edge. And that won't slow down. I think maybe CapEx moderates at some point for the hyperscalers on the edge side, but we do believe over the next kinda three to five years, edge is where you wanna be. And the way we define edge, of course, is these, you know, tier two, tier three markets, and that's working. DataBank's had a great year. They're at about 298% of plan, and it's, you know, it's September.
Yeah.
I don't think I have a portfolio company that hit 300% of plan by September. That's a, that's a new one.
What's driving it? I'm sure that AI is part of the answer, but clearly, you were beginning to see the backlog fill before everyone incessantly talked about AI. So holistically, how are you thinking about what's driving the funnel?
Well, I think for DataBank, it's really about edge workloads. It's not about AI. Edge AI comes probably another two to three years down the road. So literally, what they're doing right now is supporting public cloud, supporting enterprise, supporting mobile infrastructure. You know, when you move into a decentralized RAN, you can start putting those radio access network cores in different locations. And keep in mind, where most mobile users are is the edge anyway. So if you think about where Nokia and Ericsson are deploying those software-defined networks, they're deploying in our data centers and on the edge.
Low latency environments that can quickly cross-connect out to towers and small cells, and that ability to take, you know, workload from 10 milliseconds down to two milliseconds is, as, you know, any of the CEOs will tell you tomorrow or on Thursday, they'll tell you that's a big difference in terms of the user experience. So, DataBank just happens to be the right company at the right time, and the right—and the, with the right product. I think on the AI side, that's where really Vantage and Scala, which build big hyperscale campuses, are benefiting. And for Vantage, you know, we're growing the fastest, probably in Europe, has been our best theater.
North America, close second, and Asia, third, and the GCC and Africa, fourth, and then Scala is our Latin America based hyperscale data center operator, and that business has also exploded. So, but those leases are now not measured in 5 MW or 20 MW. Those leases are now measured in 50 MW-300 MW. And then on the other side, we have Switch, which we took private about a year ago, and that business has also exploded, as you know, enterprises and even the hyperscalers want private cloud environments. So a lot of the initial big workloads for AI are sitting in private cloud. They're not in public cloud, because really, those language models are in a phase where they're learning.
We haven't reached sort of the intuitive phase of these models, and so those environments need to be highly secure, and I think this is an area where Switch has really performed well. I think I said it a year ago at this conference, that Switch will be better served being a private company than a public company. It was hard for public investors to get their minds wrapped around what Rob was doing in the, in the private cloud, but, you know, since we've owned the company, we basically have hit, y ou know, our five-year leasing forecast, we hit in one year. Just to give you a sense of how fast, you know, the leasing is coming for Switch. And look, I think Switch also is one of those unique stories where they have the right product, and they have a highly secure environment.
That Tier 5 environment, where you've got, you know, 100% uptime, is really important to the, to the AI folks in the initial phases. They want to know they have a secure place to, to build out those language models. And the other thing that's really working for Switch right now is it's 100% renewable. And so you and I were talking before this about power and how difficult it is to procure power today. You look at arguably the two most important data center markets in the United States, say, here in the Valley, and also out in Virginia. Both of these markets are shut down. The next upgrade to the grid here in Santa Clara with SVP is effectively, it was 2026, and now that's being pushed out to 2028. And you can't even talk to Pacific Gas and Electric.
They won't, they won't be able to deliver you 20 MW of contiguous power. So you have a real power problem here in Northern California, which is really arguably the most important area for innovation in the United States. Then you go to Data Center Alley in Virginia, and the governor has done a great job there. Senator Warner's done a great job there. That looked like it was gonna be a five-year problem. They've now got it reined into about three years, and they've created some other counties where you can develop data centers, but where the customers want to be, most importantly, that part of the grid is shut down until 2025, 2026. So, it's really brought to the forefront the scarcity of power, and most importantly, the relevance of renewable power, and Switch being 100% renewable.
Like, for example, in Reno, we've got 11 million sq ft of data center capacity. We've leased into about 1.7 million sq ft of that. So I've got about 9.5 million sq ft of expansion space that's 100% solar. We share a solar farm with, with Tesla there. And that market is effectively, depending on your, your, the route of fiber you take, you're anywhere from two milliseconds to 10 milliseconds right here to the valley. So Reno's becoming really the proxy for Silicon Valley for, you know, data center activity. And so a lot of the big workloads that you're gonna see from, from the AI folks are gonna end up in, in Reno, I think. Well, I, I know. It's going where the power is, basically.
Look, Microsoft, Amazon, Google, Salesforce, Oracle, these guys have it right. Apple. They've made it very clear that if you don't have renewable power, they're not gonna do business with you in the future. But that's, for them, it's no longer a nice-to-have, it's table stakes. And so we're continuing to build renewable data centers with Rob and the team at Switch. The good news is, Vantage is now learning what Rob's done, DataBank's learning, and having these amazing ecosystem of companies where they can pass knowledge to each other is really beneficial. And Rob's really smart. He's got an incredible nose for innovation, and the company has over 300 patents, as you know. So I think that those applied learnings are gonna help in the edge, they're gonna help in hyperscale.
It doesn't solve the problem. I think if we look down the road, depending on the reports that you believe, whether it's 35 GW or 38 GW that's gonna be built over the next seven to eight years, we know that the existing transmission infrastructure in this country is inadequate to satiate what's coming from AI. So there is this kind of cliff moment, where it's maybe three to four years down the road, where the existing grid infrastructure in the U.S. just runs out of the capacity to deliver the power. We can generate the power, but we just can't deliver it, and so that becomes a real problem. Look at what Texas is doing. Texas is redoing their entire infrastructure grid right now, which is ERCOT, which was a big project in Texas, but that's gonna take 10 years.
So the thinking has to happen now. We have to be out in front of this, and I don't think we can sit there and wait for government to come up with a solution. This is gonna be a private sector-based solution, and that's, for example, what we're doing at Switch. We're not waiting, you know, on the state of Nevada, we're not waiting on the state of Michigan, we're not waiting on Georgia, we're not waiting on Texas. We're creating our own power narrative, and ultimately, the goal for us is we need to get behind the meter.
You know, right now we're in front of the meter, but the goal is, in the next two to four years, I've said it very clearly, the vision for DigitalBridge is, we not only have to be the most important landlord in the digital space, but we also have to be the largest landlord that delivers, you know, carbon neutral solutions for our customers. Because I think if we can help solve that problem for our customers, the key is they're gonna come back and do more repeat business with us, which what I've learned in 29 years of this is, if you take care of the customer, they'll take care of you back, and we've got really good customer relationships.
So how's that shaping where you want to invest capital geographically? Obviously, in the Nevada desert, you have a lot of sun and a lot of space, but that's a unique situation.
That's a unique situation. I think we've used hydro, we've used solar, we've used wind. In Europe, we're looking at a lot of wind solutions. So for example, in Dublin and in Cardiff, where you've got a lot of wind, there, we're looking to deploy significant amounts of wind farms adjacent to the data centers, and right now, sell into the grid or eventually go around the grid and create our own substations. We've done that in Brazil. Scala has done that, where we've sourced all of our power through hydro, and then what we've done in Campinas is we've created our own substation, and we've become our own power company. And now what we've done is we've just bypassed the national infrastructure.
We have our own infrastructure, and so we deploy that into our, into our campus, and at the end of the day, you know, that's what the hyperscalers want. They want you to deliver 100%, you know, green source solution for power. And this is not just for ESG, this is business continuity, like t his is survival. I don't think it's ESG. And some people think a year ago when I said that, that I was being a bit dramatic. Y ou know, I'm just doing the math, right?
There's just an infinite amount of power, and we see demand, you know, just sort of stepping up each year in terms of total quantum. Maybe that plateaus and the industry is delivering, you know, 5 GW-6 GW a year. I mean, we're on pace as an industry to deliver about 5 GW-6 GW this year. It'll be the most power that the data center sector has delivered in the history of this business, will be this year, 2023.
And then at the same token, you know, we're all investors in the room, so you got to pay for it. You know, how do we pay for it at the same time? How do you form the capital to go out and do it?
Yeah. Well, and speaking of paying for it, I mean, we've been dealing with inflationary cost pressures now for over a year, particularly energy. Cost of capital has gone up based on where interest rates are. How much success have your portfolio companies, we'll stick with the data center space right now, had at passing through the higher cost of doing business and of investing, to make sure you can still get the returns that you need to get relative to your cost of capital?
What we looked at it is, there were three vectors that we were focused on, and we had a data center offsite last December, where we talked exactly about this, which was, one, how do we get to our lowest cost of capital? We put our six CEOs in a room and came up with a game plan, and it's worked well, but it's hard. The second is, how do we provision as a collective group? So looking at the 350+ data centers we own worldwide, and the amount of power that DigitalBridge portfolio companies are consuming, I think we're probably number one now, or certainly they're close to Equinix, in terms of just square footage and power, not in terms of probably EBITDA and earnings. They're a little bit older than us.
But the key to that is showing up as a global data center provider so that we can level the playing field in negotiating in terms of chillers, generators, and the respective, you know, hardware or componentry that we put into those data centers from electrical work. And I think we've been able to do a good job of getting our costs down. It took some time to get there, but I think this year's been a good year. Last year was a really tough year. And at the same time, as you know, Brett, rents are, rent's just going like this.
So we've seen a complete repricing of our new inventory, and we haven't seen any pushback on that pricing. And so what we've done is we've priced effectively ahead of interest rates, and we've priced ahead of inflation. And so it's, it's no surprise that when you read the reports from CBRE or Jones Lang, you've seen rental increases in the order of magnitude of 20%-25% in some of the big flat markets in Europe and the, and the core markets here in the U.S. There's been some markets where we've priced, Brett, above that, where rents have moved significantly up.
But at the same time, we're also assuming that the cost of capital is going up, and that we don't see cost of capital declining until probably second quarter next year, will be the first time we think we see a real rate cut, where we can start seeing, you know, the cost of debt go down. So, to that end, we securitized most of our debt in 2020 and 2021 and 2022.
And so we have no material debt maturities in any of our portfolio companies in 2023 or 2024. And in fact, you know, 78% of our debt is fixed, and it's fixed on 2020 and 2021 interest rates. So we, we kinda had a hunch this was gonna happen, and so we, we cleaned up the balance sheets of all of our portfolio companies. And the most important thing is we spent the last two years deleveraging, so we had a basically a 47% loan-to-value ratio. Today, we're at about 43%, and in this market, it's really hard to delever. Well, how do you delever? You go out and you lease, you cut costs, and that's what we did. So there's a way to get there, but I think the key from our perspective is having the youngest fleet in the industry.
Our average data center is less than three years old, across the planet, so we have the youngest inventory out there. We're a lot younger than DLR, we're a lot younger than Equinix, and so having a younger set of assets with the right power density, the right fiber, the right cooling configuration, we've been able to price a little higher than our peers. We don't give out rental rates, but we, we can tell you with, with total certainty that rents today are very strong. And three, four years ago, I think the industry, we were, we were price takers, and I think today we're, we're really the price maker, and that's been the, the shift in what's happening in data center land.
Another side effect of the higher cost of capital is just really chilled the M&A environment. And it's normal. We see this every time rates go up. The sellers of assets pause. They don't wanna take a lower price. They wait. Where are we in the cycle of M&A coming back into telecom infrastructure, and do you think this is gonna be a big opportunity for the DigitalBridge portfolio companies?
I think it's a, it's a huge opportunity. We are doing M&A. We're doing tuck-ins, small tuck-in M&As. We're very, very active right now in investment committee. There's three to five new deals a week. You are right, people are sitting on the sidelines. They're kind of doing this. They're putting their hands underneath their legs and waiting. I think it's, you know, this isn't like 2008, 2009, right? This is more like 2001, 2002, more like dot-com crash. And I think it took us to come out of the dot-com crash in probably 24 months to get out of that. And I don't know what we're gonna call this correction, whatever we wanna call it, the Powell correction, maybe. I don't know. We'll have to figure it out. The COVID correction? But we're probably maybe halfway through that M&A stalemate.
Remember, in 2003, the M&A market really picked up, and the way the M&A market picked up was mostly through restructurings.
Yeah.
So we were super busy at that point in time in my career. I think I was, I was at Deutsche Bank, and we were doing mostly, restructurings. I was on the private equity side of, of DB. That market took two years to sort itself out, and then it was 2003 to 2005 until we got back to a normal M&A cadence, which was 2005, 2006. So really, if you look at that dot-com crash from 2001 to 2005, it was actually a four-year window where it took multiples four years to recover coming out of the dot-com crash.
But telecom infrastructure is a much more established category.
It is.
Money is far more sophisticated and deeper. So do you think we could see a faster recovery into a normalized M&A environment?
Well, I think we will. I think it doesn't happen till next year.
Yeah.
I think interest rates have to sort of reveal themselves a little bit more. I also think this notion of private lending against bank lending, against securitized lending, you know, the banks are relatively closed for business. Regulators have put a very high bar on the carrying cost of bank debt on balance sheets. So what we're seeing is, the banks are still sort of trying to really go out and syndicate a lot of the bank debt they put on in 2001 and 2002. And so until these big banks sort of flush through their balance sheets, you're gonna continue to see the rise of private debt capital formation and the rise of unitranche loans. And so we're seeing that in the marketplace.
I think, you know, we're constantly pricing out, you know, providers of capital, like Apollo or Ares or even some of our peer GPs like Blackstone or Brookfield. What you do see is there's a lot more collaborative approach to lending, and I think you're seeing a hybrid of senior banks and private lenders coming together. And we haven't seen that in a very long time. But that's a function of really just an evaporation of the liquidity in the system, and I think that, again, that will continue for at least another year. I know, for example, w e have a series of credit funds. Our credit fund is doing incredibly well, way ahead of plan.
Returns are up, yields are up, and we've got a pipeline that's, you know, 30 deals deep, and we're saying no more than we're saying yes. So, we're being very selective, as are other folks like Apollo and Ares and people that do this, you know, on a day-to-day basis. I think in terms of the M&A market, just to come back on that, I think everyone, for years and years, used to think M&A was just a game that was played by strategics, and I think the chessboard has shifted. And I think you have infrastructure GPs, but now you have LPs on a direct basis doing deals. And so, we see very sophisticated, large sovereign wealth funds. We see sophisticated insurance companies and pension systems being more active on a direct basis, and so that does add another buyer set to the ecosystem.
I mean, the deal we did for Beanfield with OMERS is a great example of that. There were strategics that showed up to that auction, there were GPs that showed up to that auction, and then there were direct LPs that showed up to that auction. The two best bidders were two Canadian pension systems.
The highest bidder for that asset was OMERS. We were happy to partner with OMERS. Some of these transactions also require you to be creative, and I think, again, one of the areas where an asset-light model works is we have over 200+ global LPs. We have the top 10 global investors in the world. When there's a really interesting M&A trade, we go partner with them, and that's again a solution set that we have that others don't have.
All right. We're already running out of time. I wanna talk about fiber.
Yeah.
You know, I've covered the telecom space for a very long time. In their public iterations, the public fiber companies have never really lived up to public market investors' expectations, one of which was Zayo, which you now own and operate.
Can you talk about how you think about where fiber is in the evolution of real telecom infrastructure, value creation? Maybe talk a little more about what you're doing at Zayo.
Yeah. Well, first, look, fiber is infrastructure. Whether in its most basic format, it is the critical sort of arteries that connect everything, connecting cell towers to RAN hubs to data centers. So it's vital, it's critical, and I think where the street has gotten amiss with fiber is that a lot like data centers, there's five or six different business models embedded in fiber. So not all fiber businesses are the same, just like not all data center businesses are the same. So again, we come back to the same thing you and I have talked about for 20 years. If it's digital infrastructure, it's got to fit sort of four quadrants, right? Which is, first and foremost, do you have a long-term lease? Second, do you have high exposure to investment-grade counterparties? Third, do you own the fee? Do you own the land?
And then four, do you have a unique permit? Do you have a barrier to entry? I think why you've loved towers for so long is, if you get a great building permit in Burlingame by the airport, no one's gonna build a tower next to you. In the data center space, if you have a great Will Serve Letter from the power company for, you know, 90 MW in the middle of Santa Clara, nobody's gonna be able to build next to you. The problem with fiber is, there's no privacy to permitting, as you know. As long as you're filed as a CLEC in a state, you can go on the same pole, run the same fiber as the ILEC, as the cable co, and as the overbuilder. So there's no uniqueness.
So it sort of fails in that fourth and final quadrant, which is, you don't have anything that's unique or exclusive in terms of being able to box out your competitors. I know some people would tell you in the cable business, the cable guys feel like they have—some of the cable guys feel like they have a unique franchise agreement. In some cities, they do have a franchise agreement, but most cities now have a second franchisee. So you don't close the zoning door behind you in fiber. And even if you build a really unique sub-oceanic cable route, somebody can come right behind you and build another sub-oceanic cable route in the same place. So I think that's one of the areas where there's a lot of competition. And when you have a lot of competition, you can't set terms.
A lot of the fiber agreements tend to be between two years and five years in duration, except for Dark Fiber, which as you know, is 10 to 25, 30 years. The term of the cash flows is shorter. You're less exposed to investment-grade counterparties, and you don't own the fee, you don't own the land. You're typically running through an easement. While it's infrastructure, it just kinda has some small blemishes compared to towers and data centers. And so I think that's where investors get a little sideways with the fiber risks. And look at the industry hasn't been a particularly great, you know, industry to own because most of the stuff trades at eight to 12 to 13 times.
And then there was this euphoric moment where infrastructure capital came into fiber in the last three or four years, where people started paying mid-20s for fiber businesses. And we challenged that. We sort of said two to four years ago, so we're out, we're not doing that. And so we stayed out of resi fiber, and a lot of people made big bets, huge bets, particularly in Europe, where you're seeing a lot of bankruptcies now, and you just can't pay mid-20s for fiber assets. It just doesn't work. The dog done hunt. But in the situation with Zayo, where we paid 10.8x, that did work for us, and we sold the data center business at 16x. And so we, we, we delevered the business. And, and we also knew that Zayo had a really good and unique plant. It had unique long-haul routes.
It had unique metro rings, and we knew that the business wasn't being managed well. We knew we had to change the product set. We had to change the sales team, change the commission structure. We changed the capital structure, or we're going to probably change the capital structure again. And, ultimately now Zayo is posting, you know, strong, high single-digit, low double-digit growth for the first time, organically, for the first time in three years. So it's taken us three years to fix it. I'm not here to tell you we're completely out of the woods, but the products that are working are SD-WAN, data center connectivity, you know, fiber to node, fiber to towers, fiber to data centers. E-rate's working. We just won a $96 million grant from the NTIA to go build out some rural networks.
And so we've had to work a little harder. We've had to be a little more creative, but now we've got the engine fixed. We finally have the right management team in place. We had to fix the back office. We had to reinvest in the network. We spent over $190 million in voluntary CapEx the last two years to fix certain pinch points in the network. But fiber is hard. It's not an easy business. What I love about data centers is you sign a 15-year lease with an investment-grade tenant, and they control the data hall. In the fiber businesses, you know, that's not the case.
Is the end market supportive of owning a fiber asset? Do you see, for example, AI or other themes driving up demand for the infrastructure?
Well, what we do see is, particularly on certain low-latency data center routes, we're seeing customers go from, you know, four pairs to 12 pairs, to 24 pairs, now to 200-400 pairs. So the good news is, like a tower lease, as a customer takes more capacity on that network, yes, our rates do go up. We have the ability in our, on our master service agreements with those carriers, as they take more pairs, we get more rent. So that's working. That's great. That actually feels like towers a little bit. The gift that keeps giving, the amendments. So that's going well. I think, you know, SD-WAN is working. But it's again, it's hard. I think if you're, if you're focused on the hyperscalers and you're building the right routes. And the other thing is capital efficiency.
I think the previous management team was, you know, sort of believed it was an open checkbook, and I challenged that about a year and a half ago. I said, "Look, we got to stop putting up all the CapEx," because we know in other successful digital infrastructure business models, you have a combination of NRC and MRC. So now what we've done is we've -- we're massively lowering the CapEx at Zayo. We'll cut CapEx by... Gross CapEx, we'll cut by 50% this year.
Net CapEx, we'll cut by 65%. Our paybacks, instead of being, you know, the previous management team thought a five-year payback was good. Today, we think sub-24 months is a good payback, and that's what we're getting. Really good capital efficiency, which is why we're turning the story.
All right. Well, that's a great place to end. Marc, thank you so much for being here.
Thanks, Brett. Good to see you.
Bye.