All right, let's get started. Good morning. Welcome to the Wednesday morning session of Nareit REITw eek 2025. My name is Greg Williams. I cover cable, wireless, and telco, as well as fiber at TD Cowen. I'm joined in this session by the President and CEO of Uniti, Kenny Gunderman. Kenny, thanks for joining us.
Greg, good morning. It's good to be here.
Maybe we could just start with finishing off the quarter and your bookings. With all the GenAI hype, there might have been some big expectations on big hyperscaler deals. You did a million in bookings a few quarters in the past, though you have informed us in the past that there's lumpiness in these deals. Help us with what the bookings could look like in the foreseeable future. Will we see a million quarter bookings again? Help us with just the hyperscaler environment, whether it's going to training data centers, like what you're basically selling.
Sure. I'm mindful that we're at Nareit, and usually there's a few folks in the room who don't know the Uniti story as well as others. We're one of the largest independent fiber providers in the country, so almost 200,000 route miles of fiber around the country, very uniquely positioned to benefit from both the fiber-to-the-home theme, which I'm sure Greg will get to in a minute, and the hyperscaler theme. I'm always interested to see where the first question comes from, whether it's fiber-to-the-home or hyperscalers. Glad to dig into that one first, Greg. Yeah, we had a really strong quarter, strong demand, strong bookings, but I totally understand the question. The hyperscaler demand has really come from a very small, de minimis number just a couple of years ago to now a TAM in our fiber space of almost $20 billion, $15-$20 billion.
Almost overnight, in 12-24 months, it's evolved into a very large TAM. A few years ago, we were selling 6- 12 strands of fiber to the hyperscalers, and it was a de minimis part of our business. Today, we're selling 30, 40, 50 times that amount of fiber to hyperscalers in single deals, and they're coming back and buying more on top of that. The strand count type of deals that we're talking about here are deals that we haven't seen, the types of deals we haven't seen in the fiber industry. That's a huge increase. To that point, Greg, it's right to question, when do you start to see those numbers in the bookings numbers, for example, or in the other vanity metrics that I like to call them, revenue and EBITDA?
I've cautioned people not to expect to see those necessarily because, A, they are lumpy, and B, they're the types of deals that don't necessarily run through bookings in a traditional sense. Because in addition to dramatic increases in strand counts, we're also building greenfield more as opposed to selling lit fiber or existing capacity. We're also selling more long-haul as opposed to metro. These deals, in some cases, are structured in a way that are different than traditional deals. For example, for us at Uniti, the rest of this year, we have two really sizable hyperscaler deals that are going to hit probably in the fourth quarter, and they never hit bookings. These are two of the biggest deals we've done with hyperscalers.
They never hit bookings over the past 12- 24 months, but they'll hit revenue and EBITDA in the fourth quarter of this year. When we think about guidance for the year, we're right in line with guidance, as we said in our last quarterly call. Analyst estimates are actually low for the second quarter because our year is back and loaded because of these really large mega hyperscaler deals that never made their way through bookings because they're structured in a unique fashion. Make no mistake, with that said, the demand from the hyperscalers has been terrific. We think that $15 billion-$20 billion that I just mentioned is growing to $40 billion-$50 billion in the next several years.
When you consider that the wireless carriers historically have been one of our largest customer bases, they spend as a group roughly $30 billion, and that's across the board, not just in fiber. That just gives you a point of reference about how big of a customer opportunity that is for fiber companies like Uniti on a go-forward basis. We are very excited about it. They are terrific customers. They are buying substantial amounts of fiber to build data centers that enable AI learning, as Greg mentioned. Ultimately, that will enable AI inference, which Greg also mentioned. We have said, and Greg, I think you agree with this, but the inference phase of AI is going to be more exciting to us at Uniti than the learning phase.
That is when you'll start to see the demand and the opportunity flow through bookings and revenue and EBITDA in a more traditional sense with recurring revenue.
Got it. Thanks for the intro to the company too. I always say Uniti has twin engines, right? It's the commercial fiber and it's the fiber to the home, both going through nice upcycles.
I like that. I'll use that.
Back on the GenAI demand, you were just at our conference a few days ago. At the risk of repeating some of it, you mentioned that the funnel has 20% of the funnel is AI-related. Do you expect that to grow and take advantage of that demand? You have lots of other business, wireless enterprise, cloud on-ramping. Do you keep it intentionally at 20%, stay diversified, or are you just going after anything?
Both. We're a wholesale fiber provider largely. We like that model because our customer segments range from the hyperscalers like we talked about to the wireless carriers who provide mobile broadband and also fixed wireless and the satellite companies, the fiber-to-the-home providers who are building fiber to the home but need backhaul for that traffic to get back to the core, international ISPs, large enterprises, large healthcare campuses, government agencies, et cetera. We have a very wide array of customer segments. As a result, no single customer segment represents actually more than 10% of our revenue or EBITDA. To Greg's point, not one customer segment represents more than 20% of our sales funnel. That's not necessarily intentional.
We do not try to get down to the precise percentage, but it just happens to work out that way because we focus on a diverse set of customers. The demand across all customer segments is generally growing. It may ebb and flow. We had, Greg, as you know, last year, the wireless carrier spending was down relative to other segments. We never noticed that at Uniti because other customer segments made up the difference. That is the beauty of a diversified business. Ultimately, when you look at the hyperscalers, the important thing to note is not necessarily that it is 20% of the funnel. It is the fact that it came from virtually zero to 20% of a very large multi-billion dollar funnel within a very short period of time.
When you consider that some of the transactions and opportunities that we have with the hyperscalers are not unique and do not get reflected in bookings in the traditional sense, like I said, if you actually reflected those types of deals in the funnel, that 20% number would be a lot higher. We do not look at it that way in the funnel, but if you did look at it that way, it would be substantially higher. I think I said this last week, Greg, and if not, I will say it here, the funnel of opportunities for the hyperscalers for us is as good as it has ever been. When I say that, I do not just mean the Uniti funnel. I mean looking at the opportunity that Windstream has with hyperscalers in their wholesale business because we are about to merge with Windstream. That deal is closing soon.
Windstream has a large wholesale fiber business themselves, and there's a lot of synergies in bringing those two companies together. They also have a large focus on the hyperscalers, and there's some really large transactions that we're working on together with Windstream already with hyperscalers.
Got it. The deal construct and the yields on these hyperscale deals, have they changed? We spoke, I think this time last year, we spoke about really high upfront or NRC costs or fees going to you. Yields could be north of the typical 5-10% range. Has that evolved, or is it sort of the same construct?
It's definitely the same construct, and it's evolved a little bit. I would say it's evolved in a favorable way. As a wholesale fiber provider, we're a shared infrastructure provider. We build fiber, and we usually have an anchor customer when we build that fiber. We don't build fiber for demand on the come. We always build it with an anchor. We target a 5-10% initial yield on any capital that goes into the ground. On day one, we've got an initial yield that's usually a 10-20 year contract locked in. We have a plan to sell the second, third, fifth, tenth customer, et cetera, on top of that fiber. We call that lease-up. Our plan is to have a lease-up strategy where we get to 10% plus yields within a very short period of time.
For those of you who follow us, you know we track this each quarter and show investors that our initial blended anchor yield is around 7% on our greenfield builds. Our blended all-in yield is close to 30%. We have been really executing well on that anchor lease-up strategy. Greg is rightfully asking the question, how do the hyperscalers fit into that model? We approach hyperscalers as anchor customers generally. They sometimes lease up existing fiber. They sometimes buy waves. They sometimes buy fiber in the traditional sense. The big deals that we are working on with them are anchor deals where we are largely building a lot of new fiber. Those yields blended with lease-up that we have already started to see on those opportunities are already approaching 20%. We are well ahead of what we would say is our traditional model.
We're not only executing on the anchor builds for the hyperscalers, but we're also really executing on the lease-up of those networks and, again, approaching nearly 20% yields.
Yeah, it's good to hear because there's a concern out there that the fiber that's connecting to training data centers is kind of built in the middle of nowhere because the training data centers need a ton of power. Where can you get power in remote areas? If you're building out the middle of nowhere, are you passing the schools, the universities, and the business campuses, hospitals to get that? I guess you're choosing deals that have the good lease-up opportunities then?
Yeah, absolutely. You're touching on another theme that's important to Uniti. So Uniti, we've got a nationwide network, but we focus on tier two and tier three markets. We don't focus on the NFL cities. We like to go to the smaller markets. One of the reasons I think we're benefiting, I would say in a disproportionate fashion with the hyperscalers, is for the reason that Greg just mentioned, the power grids in places like New York City and Virginia and Chicago and other places are stressed because they're just bigger markets. In order to service these mega data center campuses, they require substantial amounts of power, heating, and cooling. Using already stressed power grids is a challenge.
They are looking for areas that are in rural America or suburban America, which happens to be, in some cases, in or near a lot of our markets. We are building network in or near our markets that are additive to our network. Having said that, there are also some locations that they are building that are way the heck out in the middle of nowhere. We are not building there because we do not view that as a shared infrastructure model where there is lease-up potential. There is plenty of lease-up potential in the areas that we are building for the hyperscalers.
Is it still predominantly training rather than the inference that we spoke earlier? Like the other two sizable deals that'll hit in the fourth quarter, those sort of training? Do you see some of your RFPs? I think last week we spoke some of it's Nebula, some training, some inference. I guess the next question then is, when do we move to that 80/20 towards inference, do you think? Two years, three years?
Yeah. Greg, we do not know for sure is the short answer. As usual, I will give a longer one, which is we think it is largely training today. Part of the reason we know that is because the training data centers tend to get built in the more remote areas because latency is less important for training than it is for inference. That is where the majority of the data centers are being built that we are connecting to today. I think it is fair for us to call that training. With that said, we believe inference is probably going to be here sooner than we thought because I listened to all the hyperscalers and what they say publicly, including NVIDIA, which is a really important part of the supply chain here.
Last week, they had earnings, and they were talking about how the demand and the procurement of inference chips is substantially higher than expected. They are already seeing, I think, the inflection point in their business, which means it will not be long before we will see it in ours. As a result, we will start seeing more edge data centers being built by the hyperscalers because that is where the data centers need to be closer to this end user so that latency is substantially better. In addition to that, and what is probably most exciting to us is when the inference phase hits, we will not just be pointing to data centers from the hyperscalers.
We'll really be looking at just increased demand across the board from our customers who will be using AI, whether it's all of us as individuals using ChatGPT or Grok on our mobile devices or healthcare campuses using it or government agencies using it or large enterprises using it. I think, Greg, we'll probably stop even talking about AI. I think it'll be such a foundational, one of those foundational elements of broadband growth that we see that drives overall demand. It will be infused, really, I think, in virtually every workstream. I think that's a really exciting time for both the fiber industry and the data center industry for that matter.
You're short of answering my next question, but can you help the audience in the room and explain why inference would be more exciting for you? Is it just the demand size? Is it the margins, the CapEx intensity?
Yeah, it's really all of the above because I think when inference, so to be clear, inference means you're using AI. You in this room are using it, and various institutions are using it, which means when you do a ChatGPT search, that you expect a really quick response on that. When an enterprise uses AI, they need a really quick response in order for that to happen. All of that demand is going across either starting with a wireless connection, but definitely backhauling to fiber, or it just goes immediately to a fiber connection if it's fiber into the home. Excuse me. Excuse me. It's driving a substantial increase, we think, in demand for fiber.
You are broadly increasing the number of customers who need to buy large bandwidth pipes into their homes, into their buildings, into their upgrading wireless towers, upgrading small cells, all those things needed to drive incremental demand. Ultimately, to your point about capital intensity and margins, we do think a lot of that will be lease-up, right? The networks that we already have in place at Uniti, and again, we have close to 200,000 route mile network around the country, but we also have close to 5 million connected endpoints in the network, including homes, buildings, wireless towers, and small cells. That is when those connected endpoints are going to be essentially used as on-ramps to drive traffic onto that network.
I think a lot of the inference phase should come at more capital, and it should come in the form of lease-up, which should come at better capital intensity and higher margins.
Can we talk about the fiber demand that's not AI, whether it's cloud on-ramping, enterprise connectivity, wireless? What are some of the other verticals that are more exciting and just speak to the demand backdrop for those?
Sure. I'd love to talk about hyperscalers because that's an area that's really attractive and sexy to investors now. The reality is, last year, and actually including this year, our biggest customer segment has not been the hyperscalers. It's been the fiber to the home providers. This is another theme that I'm sure Greg's going to get to, but there's a lot of capital that's being spent to build fiber to the home by Verizon and AT&T and T-Mobile and others. Ultimately, building fiber to the home is only part of the story. You've got to build backhaul. You've got to have metro fiber, and you have to have long-haul fiber to connect that fiber back to the internet or to the core. That's where Uniti comes in.
A lot of the people who are building that fiber to the home are not actually building the backhaul. They're using providers like Uniti for that backhaul. As a result, the past couple of years, they've been a terrific customer segment for us with the demand that's growing at the edge of their networks. They're procuring demand for our fiber from us. Terrific customer segment, not nearly as sexy, but very attractive. The wireless carriers were down last year as a customer segment, but we foreshadowed that we think this year that there would be growth in wireless. We're really seeing that. Early indications are that our original estimates of the growth in the wireless segment are probably conservative. We're probably going to exceed growth, our expected growth in the wireless segment.
That is just the wireless carriers coming back and building more towers or starting to build a little more small cells. Again, anytime you build a wireless tower or a small cell today, you need a fiber connection. I always say you cannot have wireless unless you have fiber. We are seeing some of that. Ultimately, Greg, those are two customer segments that we have been excited about this year. I do think over the next several years, we are going to start to see large enterprises become a bigger customer segment for Uniti. Today, large enterprises are not a huge focus of ours, but it will be increasingly because we think they are going to be important users of inference going forward and just general broadband growth.
Got it. That's helpful. I wanted to shift gears to Kinetic real quickly. Last week, you spoke about you had real conviction in the three and a half million homes, if not more now. Can you talk about how we should think about the cost to build and upgrade those homes at Kinetic? Windstream is known for lower density markets, more rural. One would have thought the cost per home pass would be well north of your $600-$700 range. Just speak to why it could be so much lower than others and just the cost per home pass in general.
Yeah. Kinetic today services close to 4.5 million total homes. Today, about 45% of those have fiber built to them. By the end of this year, we'll have about 2 million homes built with fiber of the 4.5. We have said publicly that we're really going to accelerate that build. We are big believers in fiber to the home, big believers in that as the superior product relative to cable, relative to fixed wireless, relative to satellite, anything else that's out there. Fiber is a superior product. We believe that by getting to 3.5 million homes by 2029 and covering roughly 80% of the footprint, that's a terrific business model for us. I've said we're not going to stop there, but that's the goal that we've set out publicly.
I feel great about the returns on that number. A big input to that model is the build cost. How much does it cost to build to get to that home? Today, Kinetic is building at roughly $650 per home. As Greg knows and a lot of you know, most people are spending well north of $1,000 per home to build. We get the question a lot. Why is Kinetic able to build at a lower cost? There are really multiple reasons. One is these are smaller markets. Building along a street in Greenbrier, Arkansas, is just cheaper than it is in New York City. We see that in our Uniti Fiber model. Also, over the past 10 years, Kinetic has built a substantial amount of fiber already, building the backhaul, building the metro fiber.
As I was talking about earlier, the fiber to the home carriers are, in a lot of cases, not building that for themselves. Over the past 10 years, Kinetic has built a lot of that. That is the network that Uniti owns today. We are already saving 10-20% on the build cost because that investment's already been made. Thirdly, Kinetic builds a lot of their homes internal with an internal sales.
An in-house team.
With an in-house team. That saves another 10-20%. On a go-forward basis, Greg, that $650 is probably going to increase because as we accelerate the build, we are going to be outsourcing more of that build just to the time value of money to capture that demand. We are going to need some outsourced help. That $650 is going to increase, but we still think it will stay south of $1,000 over the next number of years. Ultimately, that is a huge input. ARPU is an important input. Right now, Kinetic has higher ARPU than most in the space. We think there is some pricing power in these smaller markets. ARPU has been growing nicely, 1%-2% a year. We expect that to continue as well. When we put all that together, it is a terrific model. The returns have been great.
We expect that to continue over the coming years.
Got it. Cost per home pass might go up, but still south of $1,000 because you're going to be outsourcing more of it. The reason you're outsourcing, you don't have the engine to go out and accelerate. Again, you ran the NPV on these things, and it's better to do so.
Absolutely.
Capture that. You mentioned ARPU. It is higher. You mentioned you have pricing power in your markets. Is there a fear of vulnerability if lower-priced fixed wireless comes into those neighborhoods and ARPU goes the other way?
Yeah, yes and no. Fixed wireless, the demand for fixed wireless over the past several years has been better than expected in the industry. I think even if you ask the wireless carriers, they've probably outperformed what they expected. It's a good product, right? It's very easy to turn it up. It's convenient. It's affordable. It's somewhat reliable. It has decent speeds, 100 meg speeds, which when you compare that to cable and you compare that to DSL, that's a favorable comparison across the board today. As a result, Kinetic and other fiber to the home providers, cable providers have definitely lost share to fixed wireless. By the way, as a wholesale fiber provider, we've also benefited from that fixed wireless demand because that means our wireless customers are buying more backhaul to service that. We're sort of hedged a little bit.
On a go-forward basis, I actually think, Greg, that the fixed wireless losses that we've seen over the years are going to be an opportunity for us to take back subscribers because as we all start using more AI on our handsets, we all start using more bandwidth-hungry applications, you're going to need more than 100 Mbps speeds in your home. At that point, those subscribers are going to be looking for, I think in a lot of cases, a new home. They're going to be looking for a fixed broadband solution. We're going to be there with fiber in order to bring those consumers back.
The $50-$60 per month that fixed wireless is that you're paying today for fixed wireless, I think in a few years, that's going to be still affordable relative to fiber, but the speeds and reliability of fiber are going to outweigh that cost savings. By the way, I think that's one of the reasons the wireless carriers are some of the biggest investors in fiber to the home today. I think they see that same theme coming.
Yep, the convergence.
Yeah.
I do want to talk about convergence. We have time. Before that, I just want to talk about financing these builds. Then we'll turn it to the audience. There's concerns of how are you going to finance these three and a half plus million builds. ABS, Asset-Backed Securities, has been a newer subject in the commercial fiber and fiber to the home world. You successfully raised ABS in the past. That sounds like something you can be doing in the future. Help us with the ABS capacity you have now, where it can go, and how would you use it?
Yeah, the ABS, great question. The ABS market has been sort of the gold standard in shared infrastructure communications over the past 10-20 years. The tower companies have been users of that, the data center businesses. Now fiber companies are, fiber to the home providers and commercial fiber companies, wholesale fiber companies. We did our first ABS at the beginning of last year. Zayo, another large fiber provider, has done an ABS. A number of fiber to the home providers have. We at Uniti have said that in the near term, we think there's another $1 billion of ABS capacity that we have. Also, Greg, we've said it a little bit, but we actually think the real capacity there is three to four times that $1 billion number.
$3 billion-$4 billion of capacity.
$3 billion-$4 billion of capacity in the.
You're charging the commercial side or fiber to the home side?
Both.
Both.
Both the fiber to the home and the commercial side, that well exceeds how much we need to spend to build to get to those three and a half million homes. We have more than enough capacity in the ABS market to fund that build. I am not suggesting that we are going to use ABS to fund the entire build. I think that is a great arrow in our quiver. We anticipate having a healthy mix of both high yield and ABS in our capital structure on a go-forward basis. ABS is going to be a terrific tool. When you consider that we have some expensive debt in our capital structure, we had to raise debt at times in volatile times in the past. For example, we have 10.5% secured debt in our capital structure.
The ABS transaction that we did a number of months ago was less than 6.5%. So there's a.
Huge refi opportunity.
A huge refi opportunity in ABS will be a big part of that.
In short order, right? In the fall, that's something that's callable.
Absolutely. Yeah.
Are there any questions in the audience? Yep.
I'm sure you're unhappy with your stock price. What is the market not getting right? Or what are the levers you could pull to maybe?
I'll just repeat the question for the webcast. In the audience, they're asking about your stock price. Investors clearly aren't happy. What levers can you pull? What's the market getting wrong here?
Yeah, great question. I think I don't think the market is necessarily getting our story wrong. I think the market's not really paying attention yet. I think that's a big part of it. Because for those of you who know our story, we're on the verge of merging with Windstream. Windstream has been a private company for the past, really, six years, if you include some, really, virtually for the past six years. The equity market is not as knowledgeable of that business. That's going to represent three quarters of the business on a go-forward basis. There's asymmetrical information out there on Windstream. Secondly, we're de-REITing. At the corporate level, we are going from a REIT to a C Corp There's a transition in the shareholder base as a result of that.
There are also some technical elements related to the merger that are going to impact the stock at closing. We hear from investors directly that there is a lot of enthusiasm for the story. There is a lot of excitement for the story. There is also some concern about the technical overhangs that exist. The good thing about that is a lot of that, if not all of it, gets washed away once we get to closing and beyond. As a large shareholder myself, I do not like to see where the stock is, but I am not deterred or I have not lost any enthusiasm or excitement for the story. I think we are just in a period of time where there is some technical overhang, and we are going to work through it. The fundamental intrinsic value of the business is there.
It is tough to get the bid support with the dilution that is coming up when you close the deal, right? Who wants to necessarily buy in front of that?
Right.
I guess the answer is you're going to be a much larger company too. So that's sort of an offset.
Absolutely.
We are just about out of time. So Kenny, thank you very much.
Great. Thank you, Greg. Thank you all.