Sterling Infrastructure, Inc. (STRL) Earnings Call Transcript & Summary

All right. Thank you, everyone, for attending. We are very pleased to have Joe Cutillo, the CEO of Sterling Infrastructure for a one-on-one discussion. We're going to talk about something we haven't discussed the entire conference, AI and data centers. Joe, you're obviously a leader in the space and you have a lot of forward visibility being part of what's become an ever-increasing bottleneck around advanced facilities. So it's great to have you on. Would love a brief introduction and then we can get into some Q&A.

Yes. So I appreciate everybody taking the time. Joe Cutillo, CEO of Sterling Infrastructure. We focus in -- when we talk about the AI space, in the site development and the electrical aspects of the buildout. And as we get into the questions, I can tell you right upfront, we've had a lot of questions around slowdowns, delays, pauses, all the political rhetoric. I can tell you first hand, we've seen no delays, no pauses, no pushouts, no cancellations. Every week, we hear from our customer base, which is the biggest hyperscalers in the U.S., along with the second tier below them. How do we do more faster? And so it's interesting in today's world in today's market. I read the newspaper, and I hear and see 1 thing, and I talk to our teams, and I talk to our customers and it's 180 degrees from what we're reading.

Yes. Well, you stole my first question, but there's a lot of nimbyism and it's manifested itself in a lot of different ways. It's moratory and it's power constrained, just general community uproar. What are your clients telling you about it?

Yes. I think the hyperscalers will admit 1 thing. They have not done a good job of getting ahead of, what I'll call, the public perception of what they're doing. I think there's a lot of miss that are very different from the realities. We're talking earlier with some groups on water consumption, what they're doing with the grid. There's very little water used in these, they're closed systems. So once you fill them, they're not using a lot of water. They're actually building out or funding a lot of the grid build-out to get power to them, et cetera, et cetera. So they haven't done a great job. But there's a very different reality versus the political rhetoric that we see in the newspaper every day. We're in the state of Texas, and to hear the Governor of Texas come out and talk about slowing down data centers. I don't know who had a heart attack first in the state. But the reality is what they've put in place, I think, is more guardrails and stuff around small projects. It's had no impact on the large hyperscalers or the projects that are coming out in the fourth quarter, first quarter, for sure, and we don't see anything happening in the future on that.

Yes, that -- and a lot of these projects would you say are like building 2 of a 10-building campus. You have a lot of the elements of feasibility already available power, et cetera. So a lot of this rhetoric is related to things that are in the common to your point, there's more form than substance?

Yes, it is. In the projects which people don't realize is 2 things. One, we're in a unique situation that we have 3 to 5 years of visibility of what's coming. So we're working with these customers for multiple years on projects before they even come out. We haven't seen any of those slowdown. And when we start a project, all the permitting is done, power is there, everything is completed. So it's not going to necessarily stop. It's done. The 1 thing we are seeing with the projects is the size and the scope of them continue to get significantly larger. A few years ago, when we got into this, we got interested in when they were around 100 acres. We were really excited when they got up to 300 or 400 acres and said if they could just stay there for the rest of our lives, we'd be the happiest people in the world. I'd tell you, we're working on jobs right now that are 10,000 acres plus. And when you start putting self power generation on 28, 29, they're talking jobs-talking campuses north of 30,000, 40,000, 50,000 acres.

And touch on that because you usually think of other folks like engineers having sort of the front-end part and that level of visibility. But in some ways, you're an N of 1 in terms of a service provider. And so you're working directly with the hyperscalers to kind of plan out some of the work they're doing. We've had this panacea of 2030 and then a cutoff. And now we're seeing data that supports not only growth over that period to 2030, but that continued sort of evolution beyond. What gives you that sort of demand outlook confidence over the next 3 to 5 years?

Yes. As I tell people, we're the canary in the coal mine. What most people don't understand is for the next generation of data campuses to start and break ground 5 years from now, they're working on them today. They're literally spending hundreds of millions of dollars upfront to get those ready. So we work with the hyperscalers on their 5-year plan. So we know what's coming out, approximately where it's coming out, and obviously, the timing. And people will say, well, why is that? Well, it takes 5 years today. First, you got to get land, you have to permit the land, you got to either run power or get utilities there in some way, shape or form. On the self power generation, you have to order a turbine, which now has a 5- or 6-year lead time. So they're ordering this stuff up front. So the time it takes from the decision point of spending money to actually breaking ground used to be about 3 years. It's now up to 5 years. Semiconductor similarly. We started doing the facility in New York. And I remember -- I tell the story, I remember when Micron announced are going to spend $110 billion or $150 billion on this facility. Our phone was ringing off the hook. Did you get the job? Are you going to do the job? And I said, there's not a job, right, first of all. A couple of months go by and everybody says, you got the job, you're going to do the job. We said there's not going to be a job for 5 years, and 1 year into it, people said, "Well, you must have lost the job. You're not talking about this job," and I said, "In 5 years, call me and we'll talk about the job." And they were right, I was wrong. It was 4 years and 6 months from the time the funding was announced until the time we broke ground. So I was off 6 months. But the reality is it takes that long to do it. And they're also worried about capacity because the size and amount that is coming at us they finally realized that there is a capacity constraint coming down the road to build these. So they're trying to give us a much longer look at it so we can build capacity in parallel with them.

Yes, and part of that is even just alliance agreements with some of these hyperscalers because it's almost like mutually self-assured destruction. You need them as much as they need you. And so in some ways, they're trying to help you plan in phase so that you can execute for that.

Yes. No, it's absolutely. They have -- we -- it took us a couple of years of trying to explain to them that this was coming, but they've got it. And as a result, it's interesting because in a lot of cases, we're actually working on projects directly for the hyperscaler before a general contractor or electrical contractors even picked just to get speed or reduce the total cycle time of the project. At the end of the day, time is money for these guys, and they are really worried about how to speed up these projects and get more done faster. That's all we hear every day from them.

So that sort of business model structure visibility alliance agreements, if you think contractually in terms of your backlog or what I'd call shadow backlog with next phase opportunities related to existing backlog, what kind of visibility do you have there?

Yes. So first, the great thing with how we do this is our jobs are phased. So we're not stuck into 1 large long-term contract. For us, that's the greatest thing. It reduces risk significantly, allows us to adjust pricing along the way as necessary and doesn't enable us to get into a lot of risk on the project. To the outside world, it's challenging for us because none of it shows up as backlog, only a small piece does. So we're going through with the groups today, an example of a job that we're currently on. And when we talk about our backlog and future phase work of being $7 billion, I will tell you that, that number is extremely low relative to what it really is. And I'll give you 1 real-life example of a project that we're on. The project starts out, the hyperscaler -- or it wasn't hyperscalers, another group, bought 600 acres of land. And they said, we're going to build the data campus on 300 acres. So we start the 300 acres. That 300 acres, approximately $300 million to complete. We knew that, it was going to be the minimum. However, we put $10 million or $15 million in backlog. We only talk about future phase where we have 30% of the design complete. So there's probably another $50 million of backlog or a future phase, but we know it's going to be $300 million. So we're off there by a fair amount to what's going to happen. However, what we don't talk about is other 300 acres. As soon as they're finished with the first 300 acres, they're going to develop. So that's roughly another $300 million that's not in future phase, not in backlog. In addition, they have purchased 800 incremental acres attached to that. So that will be, call it, another $700 million or $800 million of work there. So instead of a 3-year project, that's going to be an 8- to 10-year project. Now to put it in perspective, the projects we're bidding today are 5 to 10x bigger than that project in totality. So as we talk about is 2030 or 2032, we're sitting there today looking at projects that we're starting or getting ready to work on over the next 12 months, there will be 8-, 10-, 12-year projects. And that doesn't include the electrical spend that comes on the back end of it, which will extend it out another 2 to 4 years. So if you take a look at it, everybody is worried about the slowing down or stopping, worst-case scenario, which wouldn't happen. But if it stopped tomorrow, we'd still be working for at least 5, probably 7-plus years on some of these projects. So it's really, really fascinating how big they are in that. Our challenge is how do we articulate to the outside world that we had that much work in front of us that we still have to do that's not technically backlog or under contract.

Right. But in some ways, you have proof points with historical work that you've done that sort of [ meat ] off into.

We have 100%. When we plant our flag, as we say, when we plant our flag and start a project, we have hit rate to complete everything that's done on it. Just in this year alone, we've picked up 4 major campuses that we didn't start that we're in [ valuing ] the customer out because it's either in a geographic footprint we weren't in or was in the time frame that we didn't capacity to start.

You touched on a good point. So what is the bottleneck? Is it customer demand? Or is it just the ability in terms of labor, resources, capital to prosecute the work that they had?

Yes, a little different for site versus electrical site is more around project management skills. Though equipment is going to start getting tight towards the end of this year and next year, that will be a challenge. We've done some stuff with CAT to get our fair share or maybe more of the equipment coming up. On the electrical side, it's labor, it's pure electrician. But the piece that gets a little confusing is a country when we finally have started more technical schools, getting folks into trades, young kids and even some middle age adults actually going back for career changes. But the problem is, it takes 4 years to get an electrician certify. So we have now hundreds if not thousands or thousands if not tens of thousands of young people and apprentices going through the electrical program, which is great. However, there's just 1 fundamental flaw with that is you have to have a journeymen for apprentices. So in the state of Texas is an example. Every journeymen can have a maximum of 4 apprentices. So if I don't have enough journeymen, it doesn't matter how many apprentices I have. And it's going to take 4 to 5 years to get these folks to become journeymen. So we have this 4-year window that's going to be very challenging until those classes start to graduate and become journeymen to really build the population of electricians that we can use in the industry.

So what's to solve? Is it modular work? Is it guest worker programs like -- you can look at the Bureau of Labor Statistics. And by 2030, we're going to have something like a 30% shortage in electrician. So what's going to solve for it? Or is that just going to constrain the cadence of activity?

What we're going to see is there's going to be a couple of things that happen. First, no question, we just tripled the size of our modular and prefabrication facility. Anything we can build in the factory and get out, we can say 15%, 20% of the cost and labor. So that's a piece of it, right? That's 1 piece. How do we continue to grow it. Two, what we're going to see -- and what we're doing, we're hiring -- and just in Texas alone, we hire 80 electricians a week. And it's not that we're losing 80 electricians a week. We grew that business 140% year-over-year. We're using 80 more electricians a week. How are we able to do that? What's going to happen is if you think of the electrical business, there's a lot of small electrical companies out there today. And they may do work on maybe a hotel like this or a small office building or a small industrial building, highly competitive and not high dollars in revenue. Those crews are mostly traveling crews. And what we need are traveling crews to go out to these facilities are not in metropolitan areas. They're in pretty rural areas. What we're able to do is take those electricians, we can pay more $5, $10, $20, depending on where it is, an hour or more. They get 60 hours of work. And instead of moving from project to project every 6 months, they're out on a project for 2 or 3 years, right? So they got more stability in their life, they're making a lot more money, and we can kind of [ route ] them. Now what's going to happen is the small electricians or the small electrical businesses are really going to start struggling to get leaner. So the big guys will be successful, more so than the small guys until you deplete that pool, right? And then you have a whole other problem. So there's a few things. The challenges that get through this next 3 to 4 years until the graduating classes start, but we believe that there really needs to be some immigration reform on top of it. That's the fastest way for us to bring in skilled certified electricians from around the world that can help us out.

Yes. And that's after data centers has created a huge sucking sound for any other form of infrastructure development. You talked about the Micron project. How does that change your view around semiconductor facility and other sort of derivative forms of advanced facilities in the U.S.? And then we talk about reindustrialization of the country. But if these data centers are sucking up all the craft labor because they're paying better prevailing wage rates, how is that going to marry with all the other work that's yet to be done?

Yes. I think in some ways, look data centers are the hottest thing. But I think people are actually too fixated on just data centers. When we look at the macro of what's happening, data center is the biggest piece today, but it may not be the biggest piece forever. And there's a lot of other big stuff coming. So '28, we'll start seeing pharma being built. Those are going to be big, big factories and big projects. 2030, we got semiconductors coming, right? In parallel to this, you've got the grid being built out. You got 3 or so, at least 3, there's I think 5 or 6 that have been permitted, but 3 LNG plants are probably going to go in the Gulf Coast. We keep getting asked about the 8 nuclear plants that they're looking at starting around 2030 to 2032 coming on. The whole point of this is an electrician is an electrician is an electrician. A pipe fitter is a pipe fitter is a pipe fitter, right? And it's not just data centers that's fighting for these, it's completely different end markets that are all growing very rapidly that are going to be fighting for the same skills and these same traits. So it's going to get more challenging. The bottom line is who pays the most is going to get them. And it's really, I think, that simple at the end of the day.

And that speaks to project selection that this audience will appreciate being selecting amongst a variety of opportunity sets in front of them. Are you sort of essentially -- have you become effectively like a fund manager and you're saying who are the guys that are going to me the best terms, the best [ mining ] multipliers which are the best clients, and I'm going to just work for them and everyone else.

We have. I mean we're very selective. We turned down a lot more projects and people realize either end customer contractual terms, we're not [ one to ] do that. But we're not beholden to data centers. If chip plants come out and they're much better margins and much better customer base, we'll move to them. We've moved multiple times through our journey. We focus on margin growth and cash flow, right? For us, that's the most important thing. And as we look at the market, not only today but over the next 7 to 10 years, the opportunities are only going to get better. They're not going to get worse. So why would we go as we say, practice with someone when we can make real money somewhere else? And we're not going to be alone on that. I think that's going to be the interesting movement of resources and assets and capital on how that all plays out.

And are you sort of locking in sort of like what some of the memory providers are doing and saying, hyperscaler, I have this kind of pricing mechanism. I'm going to select the projects that I'm going to do for you but I'll give you capacity. Are you -- is that the next iteration?

Yes. We're staying away from any long-term commitments. We've been asked multiple times to make some long-term commitments. We don't want to do that. This thing is going to continue to play out. You could have expansion by customer A in a geography you're not in versus customer B in a geography, and we want to stay in certain geographies and stay away from some as well. So we're a believer that if we can deliver projects faster than anybody else on time, every time, we can pick who we go to versus getting locked into that.

Especially [ you know ] how fungible that labor pool is.

Yes. It doesn't matter for us. It doesn't matter if it's a semiconductor plant or a data center plant or a pharma plant or a battery plant. It's irrelevant.

You touched on margins. I think there was some investor focus on the infrastructure margins. Operating margins were around 24%. What -- walk me through a color what that margin profile looks going forward. You're obviously integrating CEC and what structurally supports this sort of margin profile and how is this sort of competitive environment evolving? I know that's a loaded question.