Great. Good afternoon. Welcome to our fourth annual Fiber to the Home Symposium. My name is Greg Williams. I cover cable, wireless, and telco. I'm joined in this afternoon session by Dan Peyovich, the President and CEO of Dycom. This session will be a 45-minute fireside chat. I will try to leave some questions for the audience. Just submit anonymous questions into the portal. I have a screen over here that shows you the questions. I can try to get to them. We have a lot of content here. Dan, without further ado, thank you for joining us. I believe you have a Safe Harbor Statement as well.
Yeah, great. First, thanks for having me, and great to see everybody. If I could, I'd just like to point everybody to the disclosures on our IR website related to any Safe Harbor Statements, any forward-looking statements I might make today. With that, let's jump right in, Greg.
Great. Let's start with your brief overview of your company. What is Dycom's area of focus, specifically its role now in fiber to the home deployment as this event is fiber to the home?
Yeah. Dycom is a leading provider of specialty contracting services. We work across all 50 states in the U.S., limited to the U.S. We do work for the telcos, the MSOs, typically designing, engineering, program management, construction, and then the maintenance of fiber networks, HFC networks, you name it. It's a business that, again, from the perspective of the maintenance side and the new installations, and of course, fiber to the home is certainly getting quite active and busy these days.
Can you give us just maybe a brief overview of a mix of customers, whether it's by type or even specific customers, and what's keeping you most busy?
Yeah. I think for folks that follow Dycom, you would see that we typically report out our top five or our top ten customers. They're the names that you would imagine, whether it's AT&T or Comcast or Charter or Verizon, Frontier. I could go on and on. It is important to point out that we have hundreds of customers that we're active. We do rural work. We do work for very small co-ops, very small ISPs, all around and across the country. As I said before, that work can come in all shapes and sizes. In some cases, we're doing the engineering work and some of the pre-planning work. Of course, we like it when we can be involved through all stages of a project.
If we can get in early, we can help with the planning ahead of it, do the engineering work, do the permitting work, program management across the entirety of the project itself, and then, of course, the construction. I think one of our focuses is to get that maintenance work that follows on with the customers as well. We want to be there in perpetuity, making sure that we can keep their networks up and running and keep people connected.
Right. How much of the work is actually fiber construction versus the planning, the pre-planning, the engineering? What is the business exposure to this, and how does it compare to last year? What does the 2025 setup look like?
Absolutely. I know everybody follows a lot of the stats that are out there. Industry sources would say that there was about 10 million homes passed last year. That has been increasing year- over- year. Of course, as you look at the curves, over the next five years, a lot of growth and opportunity still exists there. For us, as I said, we have this underlying maintenance and services business that we do. We do not give exact sizing on that, but it is a large part because we are focused, again, on the long-term aspect. In addition to that, as I said, we do do the engineering. The fiber to the home builds have ramped up considerably over the last several years. Our customers, what we really learned coming out of COVID was that every home needs to get a high-speed broadband connection.
The customers have certainly rallied around that. Separately, there has been a lot of federal and other state funding that has rallied around that. We talk about it really having become a utility that at the end of the day, everybody wants to make sure that they are up and running and they can meet the needs and the speeds that we need to operate in today's world. With that, we have seen and followed many of our customers as they ramped up their program over the past few years. There is a lot of complexity to that, making sure that we are hitting the passings and delivering on the expectations that the customers have. Our customers are talking out about publicly, and it takes a lot of pre-planning. Greg, to your question, where we can get in the engineering, that is fantastic.
Where we don't, we still want to be in a place where we're going to plan with the customers and plan out how we go attack and look at different parts of their geographies to ultimately bring in the number of passings. We see that only continuing to grow. We've commented that over the past year, about 14 months now, our customers have added an incremental 37 million passings to their build program plans. 37 million is obviously a massive number when we talk about 140-150 million homes across the United States. 37 million, in addition to their current plans, and for folks on the call, there's approximately 76 million homes that have been passed today with a single fiber connection. There's about 12 million that have more than one fiber connection. 37 million incremental to the point is a significant number.
As we look at that ramp over these next 5+ years, we see continued increase. In fact, for us, we did give an outlook for the year. A large part of that outlook and that 10%-13% increase that we expect in our revenue is related to these fiber to the home programs continuing to ramp. Our customers continue to find success both in the penetration that they're getting and certainly in the ARPU as that's grown over the past few years.
When I think about that backlog, I guess, is 37 million homes. The cadence going forward, I mean, I look at 10.5 million when I add up all of my providers this year. It is a peak year. I see that repeating sort of next year. AT&T, Verizon, etc., still keeping the pedal to the metal. After that, do you see it sort of tapering off? Do you see BEAD sort of filling in that gap? I am just kind of thinking about the build curve, if you will, in the next five years.
Yeah. I think some of that still has to be played out. Our customers have talked about their aspirations. Some have been specific about what that cadence is year- over- year. We certainly see this being greater than five years to get all of that work done ultimately. At some point, when you take that, you mix it with some of the hyperscaler things, when you mix it with some of the rural, there is going to be pressure on the market. At the end of the day, we've seen in the fiber to the home builds over the past few years, there's challenges around permitting. There's challenges around inspections and locating and giving municipalities. I do think that will continue to be somewhat of a bottleneck as we move through it. Our customers have done a great job overcoming that in the builds.
I think they've got a really good idea, certainly an idea of how these builds can continue and what that future looks like. We just think ultimately they're all going to be on a little bit of different cycles, a little bit of a different timing. As those numbers, the actual passings do start to come down over time. I think it's important to remember that at some point in time, the lot lines are going to get wider, the cost of the passing. I know it's talked about a lot about cost per passing. Of course, what we do is cost per foot, right? It's cost per mile, cost per each. As those lot lines get wider, the homes get further apart.
It is important to remember that in those outer years, generally speaking, and again, this isn't for every single customer, but generally speaking, you're going to get to homes that cost more ultimately to get passed. All that to say, we see continued revenue growth opportunities, certainly over the next five years.
Even though it does start subsiding a couple of years from now, you mentioned it, the maintenance opportunity. Can you maybe peel the onion there too? What is the maintenance opportunity for Dycom as you just finish up the builds years from now and you pivot to more of a maintenance mode?
Absolutely, Greg. Even before maintenance, there's the drop activity. That's the work that's getting it from the street to the side of the house. What we generally see is that takes about four years to hit some kind of terminal penetration rate. After we get through a given neighborhood, you've got about four years of coming back and doing that drop work to get people connected over time. You have that that also combines with maintenance. Listen, for sure, does fiber require less maintenance than copper? I'll just get that out of the way. Ultimately, it does. It's a passive network. Copper has more pieces and parts that we maintain on a given day. The important part here is, one, there's a lot more plant going in the ground.
Our goal, and certainly part of our strategy, is to continue to capitalize on getting the maintenance and service contracts that follow. What I always like to say is there's always going to be roads that get moved. There's always going to be new neighborhoods that get built. There's always going to be, unfortunately, somebody that hits a pole or knocks a pole over or puts a backhoe in the ground somewhere they're not supposed to. That activity will always continue regardless of whether the networks themselves are passive or active. It's been a priority for us. We really want to be long-term partners for our customers. We want to be long-term partners in the communities. We do prioritize getting that maintenance and service work and having that in perpetuity over time. We try and earn that work with our customers every day.
Like I said, we continue to see a ton of opportunities. More plant gets installed nationwide.
Right. With that more plant installed in the near term, I guess what we're also hearing is private equity-backed ISPs. They may have slowed down in 2024, but it seems to be picking up again in 2025. We just had a panel moments ago about private-backed capital still seeing that 8 million-10 million builds a year. Is that something you'd agree with? That's a loaded question, but.
Yeah, it is a little loaded question there, Greg. Yeah, I think everyone is different. We certainly saw a couple of years ago from more of an overbuilding standpoint. We saw a couple of years ago a lot of people talking about it. It's certainly some activity there. That has tapered off. There are some folks that are doing really well, have robust plans around how they're attacking that work. Certainly from the private equity side, there continue to be conversations. We have seen folks react to the overbuilding becoming more challenging as more people have tried to enter that space. Of course, as the ILECs are building out their footprints themselves. I think there is some competitive spirit there for sure.
Got it. I wanted to pivot to the cost structure. I was a little surprised both the private panel and private equity panel and the Verizon panel. Cost per home passed. Verizon noted it could be down 10%. Some private equity folks were saying that inflation's sort of behind us. I'm just curious. At the same hand, I pulled an FBA report recently that said we're up to $18 a foot versus $16 for buried at least. So I'm hearing a lot of different things. And curious to hear, do you resonate with the Verizon's cost per home pass going down or cost going up?
Yeah, I'll give you our perspective, Greg. First is the industry has become much more efficient. When these fiber to the home builds started and everybody really started to hit the market at once, it takes time for the industry itself. That's both the customers. It's certainly us. It's the municipalities. It's the entire supply chain to figure out how best to implement that work. I think upfront, there's a lot of capacity that was behind it. Trying to get that in place at a speed, you're not as efficient until you get that learning curve in place and you really start going. Many of these customers now have been building for several years. They have that figured out over time.
For us, what really is an index there and can turn to cost savings, if we have good outlooks to what these builds are and there's a lot of time behind them, that helps a ton with getting labor attraction. Certainly, it helps from a productivity standpoint. All of those parts and pieces that ultimately drop to the bottom line to make it less costly for our customers, you need that. You need to get through the learning curve and you need a good outlook in front of you. We are in a good spot from fiber to home builds, right? We are in a good spot for where we are. We are in a good spot when I talk about the 37 million incremental passings. We are in a good spot that those can be added into builds. Certainly, there are ramps associated with it. Those ramps take some time.
We're not in a place that we were a few years ago where you're starting from zero. I think all of that goes in that plus column, right? I think when the customers talk about cost savings, the labor pressure is certainly a lot less than it was a few years ago. The CPI is certainly less than it was a few years ago. All of those different inflation factors are better today than they were. On the other side, I think this is where it becomes unique to the different builds and where customers are in their individual programs. Certainly, I wouldn't be commenting against customers and their programs. As I said before, overall, the passings, the less aerial mix that you have and the more underground that you're going to have is going to drive costs up.
The further that those homes get apart, that will drive costs up. I think you have, at the end of the day, I think you have a little bit of both. I think each customer and each part of the build that they're looking at are going to be different. For us, like I said, I think in the out years of the overall fiber to the home build cycle, you are going to end up with homes that are just going to be further apart. For us, again, when we price by the foot or by the mile, that's going to mean they're going to cost more per home. On the other side, you have efficiencies that you're gaining over time.
A lot of puts and takes. There's the learning curve. There's the efficiencies, the better pre-planning. But then you're saying more buried, less dense. Maybe in the next two years, is it sort of, does all of those puts and takes sort of neutralize out, or are we going to see, you think, any meaningful rising costs for home pass?
I think the challenge here with many things is it just entirely depends on the build itself. I really think this is an opportunity where we can partner with our customers and often do partner with our customers to help work through that process. That is not to say that they do not do an excellent job themselves. They do. We can bring a lot of insights from the municipality, from a labor standpoint, from a productivity efficiency standpoint. We can try and optimize that curve ultimately. I think, Greg, that the direct answer to your question is it depends on the customer. It depends on where they are in their build and which part of the build that we are going to help them with.
In the longer- term, certainly you're going to get to places that just the homes are going to be further apart or there's going to be more underground as part of it.
How would tariffs impact your business? I mean, I imagine most of it's a pass-through because it's going to be on the equipment side. Maybe I'm wrong.
Yeah. For us, for the most part, the materials that we put, the plant that we install is procured by our customers. We handle some component of logistics. We're bringing it out to the sites and obviously doing the installation. From a tariff perspective, there could certainly be cost impacts on their side that could affect how much work they want to go do and the velocity with which they want to do it. There could also be timing if everything's got to be coming from the U.S. They're trying to move away from offshore for some of the items. Now, much of it is already coming from U.S. companies. All those things could impact timing. For our own equipment, at the end of the day, we buy a lot of equipment to do the installation. We've got great partnerships with the different vendors and manufacturers there.
Obviously, we're in constant communication with them about tariffs and potential impacts. As it stands today, we feel good. One, coming out of COVID, a lot of those folks brought a lot of their supply chain into the U.S. or certainly looked for redundancy where they couldn't. I think that helps. As we look at a shorter-term impact from tariffs, we're not hearing concerns today. If it's longer than six months or something that's maybe unprojected today in some of the discussions and analysis that's coming from the new administration, that could change. As it stands today, we feel good certainly about what this year looks like in front of us.
Can you talk about the pricing environment? Are you charging more customers 2024 or 2023? Is there upside in your pricing?
I mean, there's always inflation mechanics for all of us, right? Wage escalates every year. Even though it's less than it was a couple of years ago, those pieces do come into play. Equipment typically costs a little bit more. All of those things certainly come through at the end of the day. Overall, though, like I said, we really look towards the efficiencies that we can bring to customers. Our goal for them is always to try and drive the total cost down, right? Of course, at the same time, we're looking for efficiencies internally, whether that's from operating leverage, whether it's from the work that we've been doing. We've had, obviously, a huge focus on safety, huge focus on quality, and then the efficiency that we work towards in the field. Our goal is that we can win for both, right?
That we can drive the ultimate cost down to them while at the same time, and as we've shown these past few years, that we can show margin improvement as well.
Maybe talking about equipment and labor, what's the latest you're seeing in terms of equipment? Maybe we can bifurcate it: equipment first, any bottlenecks that you're seeing that we've seen in the past. What do you expect in terms of equipment and supply to get you to do your job over the next year?
Yeah. Equipment, it took a while to get through a little bit of a COVID overhang there and to get folks caught up. With all of our equipment and manufacturers, we're in a good place now. We certainly got our orders are being met. We're not seeing any of the delays that we saw before. Overall, they feel good about their ability to continue to keep us in front of the equipment that we need as we continue to grow. Very different landscape than it was a couple of years ago. Like I said earlier, we've got some fantastic partnerships there. I certainly appreciate the relationships we have and our ability to keep the equipment out in front of the builds. I think I'm going to get to labor next.
Yeah, let's talk labor.
Yeah, I'll jump right into that because I think the two do go hand in hand. We spend a lot of time planning. There's a lot of strategy behind it. We want to make sure that in all cases, we're keeping labor and we're keeping the equipment side in front of our customers' builds. We're very proud of some of those constraints that we had coming out of COVID in calendar 2021, calendar 2022. We were able to ensure that our customers' builds weren't impacted by our inability to either get equipment or to keep labor in front of the curve. We've invested huge amounts into our labor force. That's from certainly a recruitment standpoint and how we go through recruitment, but also on the training front. We've completely rebuilt all of our training programs throughout the business.
What we're trying to do is certainly provide technical expertise, but also provide opportunity for our folks so that they can see when they join Dycom what that opportunity looks like for a full career and not something that's a short-term opportunity just to come to a build. We have made investments there. We have made investments in facilities and in training facilities around the country so we can continue to train the talent. All of those hopes are, of course, one drops to the safety side, the quality side, and the efficiency. Two, we want to make sure that we're keeping people in a long-term perspective. It's a huge effort, right? I mean, there is turnover in the work that we do. I talked about we have a lot of folks that we're taking from an environment where they weren't working out in the elements.
They weren't working in the heat, packing tools up and down ladders, working with complex machinery. We really feel like we've put a lot of time and energy into bringing them in the business the right way so we can train them into the skill sets, but also get them acclimated to the environments we work. I think the big part is we got to show them opportunity. These builds, the fiber to the home builds that are out in front of us, that's a great landscape and backdrop so that they can see that there's many years of work to go do. Within Dycom, they can see the opportunities to rise through the ranks. Many of our top leaders with tools and worked their way up. That's something we're very proud of. I started with my tools in my past.
From an opportunity set, we really feel like that differentiates us in the labor space.
What is your mix of in-house versus outsourced labor these days? Maybe tell us about the benefits or the pros and cons of each side.
Yeah. We do self-perform a large part of our work, the majority of our work. We do use subcontractors. We do not publish the percentages. This is important, Greg, we use subcontractors in different parts of the business in different ways. We use them to augment. When a new build starts, we might bring in more subcontractors upfront. If we are ramping up in-house talent for particular builds in particular areas, we might use more subcontractors. There is just a general subcontractor mix. We do look at those as long-term partnerships. We want to make them successful. We have helped grow many of our subcontract partners into larger businesses, brought different levels of expertise to them so they can continue to be successful.
All of the work we do, there's not any work that we do here that we only do with subcontractors. Every type of work we do in-house, our people are trained to do it. Again, the majority of our work is done with in-house forces.
Got it. Maybe talk a little bit about BEAD. Dycom shares were down a few weeks ago with nearly everything else for that matter when we heard about Microsoft and we heard about DeepSeek and then the BEAD concerns. On the latter, again, earlier this month, we saw potential modifications to BEAD guidelines. Some changes can be positive, right? Because the ISPs can be seeing streamlining rules and no rate regulation and union concessions. Some are negative. It could be more delays. Of course, more would go to satellite. That is the biggest part, I think, why the stock was impacted. Maybe you can share your thoughts on changes in BEAD. I understand BEAD is not in your guidance at 10%-13%, but it is still something we all got to think about and how you are positioning for these possible changes.
Yeah. Thanks for making that point. It's not in our outlook that we gave for the year. We still do think that there's a possibility of some BEAD revenue hitting the industry this year. We still think that there's a possibility, but most likely it's going to be in calendar 2026. It's going to be when that program ramps up. Satellites are going to play a part. They were always going to be a part. They're certainly going to be more of a part now. The reality is, and you've heard it from some states that have talked about like Louisiana when they say, "Hey, if $3,500 per passing is a number, we can make all of our," and I think it's 80% or 90% of their homes are fiber, "we can make that work." That's not a high bar from what was getting talked about.
I think as more of that information comes to bear, we continue to believe that there's going to be a significant fiber opportunity related to BEAD, related to getting rural America connected and bridging the digital divide. It's going to take some time to play out. We can speculate what percentage of the $42 billion+ the match, let's call it $50 billion in total. We can speculate how much of that ultimately goes to fiber. I think some of the other programs that like Louisiana awarded $500 million, that was not to fiber. I think some of those certainly might not make it through the new administration and some of the work that's being done there. The actual fiber builds, getting America connected, everybody's behind that. That's been a bipartisan supported issue for a long period of time.
As you pointed out, that left side, all of the requirements that were included in the BEAD program that would add cost, that certainly could add time, if those measures get relieved, whether that's the pricing cap, whether it's some of the requirements around the labor force, any of those things could, one, increase the appetites from our customer base. Many of them have talked about how some of those program requirements didn't make it very attractive. We could see more entrance certainly from some of the larger ISPs. The second part, that all helps the economics. If those get removed, remember that those costs per passing would come down in line with it, right? There are more homes that I think could possibly qualify.
I wouldn't wager a guess on exactly what the number against the $50 billion is, but what I would say is even if it's pick some number of billions that come through the program, that's still very large for our space, and that's a lot of work to get done over four years. We still see it as a real opportunity. Obviously, pieces that need to get worked out. We don't want to get ahead of it. That's why we didn't include it in our outlook. Still optimistic that the consumer today really would prefer to have fiber. We hear that time and time again from our customers. We certainly hear that across the space. They prefer to have fiber where it makes economical sense.
Again, we think that whether it's BEAD or some of the other federal programs, I mean, there's still a lot of RDOF money out there. Their states are using their own funds to connect some of these rural homes. So we still think that rural fiber opportunity is significant.
Yeah. When we spoke with the Louisiana folks at Metro Connect, they said, to your point, a vast majority of the projects won were fiber. A vast majority was underground. Obviously, they've got weather, so they need to stay underground with hurricanes, etc. It does seem to bode well that the preference by far would be to get it fiberized rather than satellite, which might not be the long-term solution. I mean, if I had to contextualize it, maybe you're saying instead of maybe being really great, it's just maybe a great opportunity still, even with the satellites coming in.
Yeah. I kind of do, if you have an opportunity, I'm not sure my hands are on the screen, but if you have an opportunity, any opportunity that's like this, and then you start moving around the edges, it doesn't change that there's a large magnitude that's got to come through our space at a time where, again, you still have the 37.5 million incremental passings that our customers are going to do with private capital. That's going to have, again, they won't be on the exact same curve, but they will certainly overlap. You have the AI hyperscaler, and that will overlap as well. You have a lot of pieces coming together. Again, yeah, we think the rural is going to continue to be a large part of the opportunity.
You mentioned the other subsidies like RDOF. Maybe you can expound on some of those other opportunities because they're pretty large in size too that we all so focused on BEAD, but there's plenty to do on all the state-level work.
Yeah. Inclusive of BEAD, there's about $100 billion of funded programs out there. BEAD is roughly half of what we see. That does not include, again, some of the states now using their budgets to backfill and get some of these rural homes connected. Completely absent of BEAD, those continue to move along. We talked about RDOF. I think that the number for phase 2 of RDOF, I think it's $10 billion-$15 billion that's still behind it. That has not been awarded yet. That opportunity is still there. All of this just comes together to meet the need, right? I think it's really important, Greg, just to come back to that. We got to get homes connected with high-speed broadband, right?
Everybody needs the connection so they can watch events like today so they can have their video conferences and all the technology needs they have at home. We still think fiber is going to play a large part of that. We are certainly glad to see our customers, from a private capital standpoint, have continued to increase their velocity and the ultimate volume and the passings. On the rural side, we still think that the dollars, whether it's BEAD, whether it's RDOF, whether it's EACAM or some of the other funding mechanisms from the states, are going to help to bridge that with subsidized funding.
What does your competition look like these days? How is that changing? Who do you see out there bidding on this stuff?
I mean, I would say overall, it hasn't changed much. We're in all 50 states. I believe we're the only company that's in all 50 states in our space. As I talked about before, we work for a number of customers. We really get a good picture of what's going on state to state, municipality to municipality, customer to customer, program to program. We've got a really good insight whether we're actually building the work or just see the opportunities come through. Generally, we have another large competitor. Generally, ours is much more of a regional competitive model. A lot of regional companies, a lot of local companies that are still out there today. Of course, over the years, we've acquired a number of those and brought them into the Dycom family. We still see a competitive set that is largely regional or local.
A lot of these projects, Greg, are getting more and more complex, right? The fiber to the home programs are complex in trying to make sure that you're going to promise and ultimately achieve month over month the kinds of passings that your customers want to see. If we think about the rural work and some of the pieces and parts that go along with that, if we think about, again, some of the AI builds, all of that creates complexity that is not the same as you see every day in the maintenance and service work or what the industry might be used to. We think ultimately that that favors Dycom. We can differentiate in that space, bringing in program management, like I said, bringing in that end-to-end from engineering, design, planning, all the way through construction and the maintenance side.
What I would say ultimately is we feel like we're well positioned and we believe we do a good job of differentiating ourselves across a competitive set. We believe that we can, one, we've shown it in the growth we've had these past few years and continue to do so in the years to come.
With the competition being the customer themselves, and we talk about them doing in-house versus outsourcing their own builds. We just spoke to an ISP who was saying they could do 30% lower cost per home pass when they do it in-house. You kind of answered some of it. It's complex business, but help us understand the risk of competition doing it and your competitor, your customer doing it in-house versus outsourcing to you guys.
Yeah. It's pretty rare that we compete with our customers directly on their own work. Most of those relationships are symbiotic. By that, I mean they might have part of the work. They're typically going to do that with union forces. There is another part of the work that we get. That can vary based on our agreements. It can certainly vary based on customers. Sometimes we do all their work. To your point, there are customers that do a lot of the work in-house. We believe that our offering, one, with the volume that we do nationwide and with our footprint, we believe from a cost standpoint that we can be highly cost competitive against our customers' insourcing. Certainly, most of them outsource the work and the model and the industry that we exist in is there for a reason.
We think that that will continue over time. Every customer is different. Certainly the way that they attack the challenges that they have and how they set up their business can be different. At the end of the day, we do believe that the service that we offer, the reliability, our workforce that's across so many municipalities and so many states, whether we're rural or metropolitan, is something that differentiates us and provides a lot of value to our customers.
I wanted to pivot to the GenAI fiber opportunity or the commercial fiber. I know this is a fiber to the home symposium, but I'd be remiss not to talk about GenAI. I know lots of folks on this call or meeting are also interested. Huge opportunities out there. We cover Lumen and Uniti, and you guys are building out a lot of fiber on behalf of these providers to connect to. I guess right now it's the training data centers to connect to. They're sizing this at $40 billion-$50 billion. Huge pie in the sky numbers, but.
Big numbers. Big numbers.
Yeah. Same idea. We're going to work with a fringe here. I guess help contextualize the opportunity for Dycom. Maybe we'll talk about the training first, and then maybe we can talk about how that pivots to inference.
Yeah. I think really what the conversations that we're having, we're working with some of our customers today. We were very excited to talk about the Lumen award that we received a couple of quarters ago and the overpull work. That is work that we're actively doing today. I like to call that kind of first inning and what we think is going to be a long game ultimately to get all the data centers and to get these long-haul networks reinvigorated for things that have not been done in 25 or 30 years. Greg, the way that we look at it is that the hyperscalers and our customers, they're looking downfield. These are long-term projects. None of these happen overnight. Again, you're talking about things that have not been done in decades. They are trying to get ahead.
We really believe that what we're building today, the conversations we're having about these future networks to connect the data centers nationwide is really about inference, right? It's about getting this high-capacity fiber networks in place that aren't there in the same capacity today. I'm sure you've all heard some of the fiber counts that are going into these conduits that are being talked about, they're massive compared to what the existing plant is. To have them be private networks so that they can not be jumping from different pieces of equipment so that they can bring that latency down and try and get to this ultra-low latency. Again, we believe that's all for moving to inference and moving from the training phase. It's just a longer-term build. These networks to connect all of these cities separate from the overpull where the conduit's there.
When you're doing new construction on these long-haul routes, getting through the permitting, getting through the engineering, getting through all the planning side of that, it's going to take years. It's going to take time to get these data center clusters connected around the country. I want to just make a quick point. There's been conversations around permitting and easing permit restrictions. I think that is a great thing. Ultimately, whether it's fiber to the home, whether it's these long-haul routes, ultimately a given community or municipality, they only want so much work and activity happening. So many streets torn up in their cities at a time. Even if you remove some of those constraints, the municipalities are only going to let the work go so fast. Back to the AI, huge amount of planning to go into these. These are linear routes.
It's not like fiber to the home. I refer to that as more of a shotgun approach where you can bring in a lot of homes from different communities to get to your total. Here, you're going end to end. Getting that set up the right way, working through that so you have good crew flow, so you have the right equipment there. Again, you're working with municipalities on traffic control and the permitting side. We think this is certainly over five years, probably 10 years' worth of work. Meanwhile.
Is that the training side or is that the inference side then?
For inference.
Both. Okay. It pivots, yeah.
Yeah. Yeah. Because I think, Greg, I think while we talk about being in the training phase today, the networks that we're building are really for the inference phase, right? The data centers are working through the training today, but pushing out to that edge. Like I said, reducing that latency. These networks are really getting there to meet the need because it's going to take five plus years to get all of this up and running and get it to a level that the hyperscalers need.
Right. That's the big conversation too is the training feels like it's direct routes, but it could feel more one-and-done, whereas the inference would be more in availability zones and major metro markets. Is there a lot of new routes being built then, you think, in inference? Because one would surmise that, of course, training, you need a new route to the middle of nowhere where they're doing the training where the power is. But on inference, you've got parent and child data centers already built with availability zones. Do they need new incremental routes? They need shovels in the ground for that. Is it you're saying absolutely yes?
I think we look at it just a little bit different. We talk about training and we talk about inference for sure, but we also talk about existing infrastructure. There are the data centers that are there today. Again, the conversations with the hyperscalers are they want to get their existing cloud connected, right? They want to get their data centers connected. It is not to say that they are not today, but with a higher capacity, with redundancy of these routes and moving towards a place where the fiber will be on an ultra-low latency network. That work is happening just to connect what they have today. The other side that you are talking about, which absolutely is related to inference, is are they going to move data centers to the edge?
This is where I come back to get everything connected and get all these clusters around the country connected today, connecting NFL cities for all the hyperscalers, getting these networks some level of redundancy. I mean, they're talking about triversity, quadversity, multiple connections on different routes to data centers. That is a very large-scale opportunity. If you add in bringing data centers closer to the edge, absolutely it is incremental, but you're already starting from a huge, huge driver coming into our space. We would certainly love to see that happen. The point I'd like to make is whether that part happens or not, there is still a ton of work to be done.
Right. So you're saying it's not just the edge deployment for AI and inference. It's also what you mentioned, triversity and quadversity, meaning you need three or four separate physical demarcation points for redundancy and disaster recovery, etc., right? That can be a huge opportunity.
At least two. Exactly right, Greg.
I mean, it's not like people define it as training versus inference, but from your sense, are we still, when do you think we'd move more to the inference mode? I mean, I've heard that we're at like 80/20 training inference, and that'll flip to 20/80 training inference. When do you think that sort of flows and flips?
Yeah. I'm going to leave that up to the experts that, again, these hyperscalers, the conversations we have with them certainly are more around future planning, Greg. When the actual, when the circuit flips, so to speak, compared to whether the builds need to happen and where we are in the build phase, again, we just think those two are independent. I'll leave exactly when that happens. We could also talk about where quantum is going to be and how many years out the potential for quantum networks is and quantum computing. I'll leave that to the experts that are spending all their time on it every day. We're just excited to be part of it and have great partnerships in that space. Again, we just think it's going to be a significant driver in the telecommunication sector in the coming years.
With the training phase that we're in, the hyperscalers are funding this with a ton of upfront capital or NRCs in our fiber world, more so than the past. Past, it could have been 40% or 50% upfront, and some of it's 70%-80% upfront. I've heard cases where certain hyperscalers are throwing 100% upfront. Does that, in turn, flow to Dycom then to build faster as well? I'm just trying to understand if it flows down to Lumen or Verizon, then in turn flows to you. Does that expedite your build? I'd assume so. I'm just curious how it flows.
Sure. Some of the customers have talked about it. From a velocity, how quick they want to get these networks in place, the answer is as quickly as possible, right? As quickly as possible. They want to stay ahead. This is not the majority of the spend. When we talk about hyperscalers spending $300 billion-plus related to AI infrastructure, this is not the majority of that, even though it is significant to our space. I think that is another important note, right? What is here is a lot of time to get these networks to go nationwide. Is the impetus there to go fast? Absolutely. Are there bottlenecks and barriers that will keep it from going overly fast and have it compressed into a couple of years? Absolutely. There is no way you can get it done that quickly.
From a pure kind of dollars perspective for us, whether it's long-haul routes or any of the work pretty much that we do, the vast, vast majority, 90+% of the work we do is unit-based. So by the foot, by the each. Whether it's a hyperscaler route, we're going to install X number of feet or X number of miles or X number of pieces, and then we're going to build that no different than we do on the fiber to the home, no different than we do on the maintenance work.
What are your OpEx and CapEx needs to help meet these aggressive goals?
Typically, we've been in the 4%-5% range. We do purchase the bulk of our fleet. We do that because, one, we've got great partnerships with the equipment suppliers. We help them from the R&D side for what we need out there in the field. Two, we want to make sure that our crews can be efficient and effective. On the resale side, we do do well reselling with a five- to seven-year range. We do not see that changing significantly, whether that is the fiber to the home ramp-up, whether it is AI, or whether it is rural. Some of the equipment types will change. Certainly, when you are doing larger fiber installs, the pieces of equipment themselves might be bigger. If you look at it in the entirety of our spend, we continue to think that is going to be in line with our revenue growth.
Right. There was a little bit of risks or snafus in the last couple of months, whether it's DeepSeek, scaled models. Google just came up with Google Gemma 3, which is massively scaling AI. Microsoft's chatter about that. Do you see any temporary pauses by hyperscalers kind of maybe reconsidering their models? I guess from your perspective, the pipes get filled regardless. Just your thoughts would be really helpful.
Yeah. This is, again, where I go back to this sort of do it a third or a fourth time. You're talking about tiny movement on the very outskirts of this. Importantly, again, in talking with our customers and talking with the hyperscalers through the entirety of DeepSeek and some of these other pieces, the conversations haven't changed. Certainly, they've come out publicly after each kind of bump in the road and talked about how their focus is still there. There's going to be iteration in the space. It's going to move in different ways. We all know it's not going to be perfectly linear. All the conversations we're having, the need has not changed.
The hyperscalers themselves, sometimes they build their own data centers, and sometimes they build their own fiber. If they build their own fiber, are you building it on behalf of them, or are they also direct customers?
Yeah. No, it's a good question, Greg. We believe that the bulk of the work is going to continue to go through the ISPs, through the core customers that we have today, because they're used to building and maintaining these networks. I think that that's generally going to be where the hyperscalers would like to be. It doesn't mean that there couldn't be one-offs or you could be inside the fence or within a given cluster, that there could be some work opportunities directly for the hyperscalers. For the most part, we think that's going to be through our existing customer base.
We're talking about AI and your opportunity, but what about AI for Dycom itself? We're another year into the AI frenzy, but admittedly, from where we sit as investors, we're trying to figure out, is AI really starting to materialize? Obviously, it's really early, but what are the opportunities at Dycom, and when would you see tangible benefits to these?
Yeah. No, it's a really important point because not only do we want to build the AI networks, it's something that we've been working on internally for some time, right? Innovation is core to Dycom, has been for a long time, continues to be. We want to stay out in front of the industry. I get it, right? The work that we do at the end of the day, it's boots on the ground, right? We're digging holes, we're putting conduit in it, and we're pulling fiber through it. There is a lot of complexity there, and there is a lot of opportunity to innovate. We've been training. We've been training AI for several years on the quality front. We train it to read pictures, and we're taking thousands of pictures every day.
We've trained those models so that they can find quality issues ahead of work getting buried, ahead of work getting covered up, ahead of work getting inspected. That's been a huge benefit for our teams out in the field, and certainly, that flows through to customer value and obviously, ultimately, shareholder value. That's just an example. I don't want to give away all of our inside scoop and the things we're working on, but what I could tell you is that, like everybody, we're working hard to try and stay out in front of AI. We're finding a lot of good applications. If there's one thing that we're really good at, we create a huge amount of data every day with all the work that we're doing across all the customers.
We've got massive data sets that we're looking to see how we can really bring those in a material way into the business. That ultimately, again, going back to one of your earlier questions, we want to try and drive the cost down for the customer. We want to try and make the work more efficient. Certainly, we want to make it safer and make sure we continue to deliver the quality that Dycom's come to be known for across the industry.
That's interesting because you would have the advantage of knowing the soil versus rockiness. If you have that data from AI, it could be invaluable whether you monetize the data yourself or you have sort of an AI-managed service plan. That'd be interesting.
We've been very active in all that, Greg.
Great. With that, we're just about out of time, but thank you very much, Dan, and thank you everybody for participating.
Yeah, Greg, thanks for having me, and really appreciate it. Good to see you all.
Yep. Take care.