Okay, great. Good morning, everybody, and thank you for joining the Sidoti Year in 2025 Small Cap Conference. My name is Julio Romero, and I cover building products, industrials, and engineering construction here at Sidoti & Company. Really pleased to be able to host Comfort Systems USA. Their ticker is FIX. With us today, we have Julie Shaeff, Chief Accounting Officer and Senior Vice President, and Trent McKenna, Chief Operating Officer. So we'll do a quick overview of the company, then we'll hop right into Q&A. If you have any questions, feel free to type them into the section at the bottom of your screen. Happy to ask on your behalf. With that, Julie, Trent, always a pleasure. Thanks for being here, and the floor is yours.
Thanks for having us.
Thanks, Julio. Good afternoon, everybody. Just give you a—I think a lot of y'all know us, so I'll just kind of give a brief overview of the company. But Comfort Systems, what we really are is an assembled workforce. We have about 21,000 employees across the United States. About 85% of them have tools in their hands. They're pipe fitters, welders, technicians, project managers, and they go out to our customers' job sites and perform both electrical and mechanical installation and service on behalf of our customers. We're about 85% of our business is mechanical right now, where it's HVAC, process piping, plumbing, and about—I'm sorry—85% of our business is construction, where we're installing the mechanical and electrical for our customers. About 15% of our business is service, where we're going into our customers and servicing the equipment, performing preventive maintenance, and doing some small projects for our customers.
About three quarters of our business is mechanical, and that's where we do the mechanical, the HVAC, the process piping, and plumbing for our customers. And about a quarter of our business is electrical. The reason that electrical is smaller than mechanical, we've been a mechanical company since inception, since 1997. We started buying electrical companies in 2019, so that's why you kind of see we're a bit smaller on the electrical side. As far as who we do our work for, we're about 65% of our business is industrial. That includes data centers, chip manufacturing, fabrication facilities, tire facilities, other manufacturing, food processing. So we do that work for our customers. And then about 24% of our revenues is institutional. It's healthcare, education. We do some government work. And then the remaining is about 13% of our business is commercial. Don't do a lot of commercial construction.
Most of the commercial work we do is related to our service operations. This is a great cash flow business. Our first dollars that we generate from cash flow, we reinvest in our existing operations, whether it be investing in technology, investing in training for our employees. We'll spend a portion of our free cash flow on capital expenditures. We'll buy and expand our facilities, whether it be the shops, or we'll invest in some of the modular capacity, but it's a relatively capital-light business. Maybe 2% of our revenues is spent on CapEx. So we have a lot of available free cash flow, and over a five, 10, 15-year period, we'll spend about 75% of that free cash flow on acquisitions. The remaining we'll spend returning money to our shareholders through dividends and through share buybacks.
So what we really are is an assembled workforce, generate a lot of cash flow, and try to deploy it in the best manner possible on behalf of our shareholders. So Julio, that's kind of an overview of the company, and I think we're happy to take some questions.
Excellent. Thanks very much for the rundown. I'll hop right in. Maybe just starting off with what you guys are seeing on the demand front across your end markets, industrial, institutional, and commercial. And then also, if you could touch on from your seat kind of how you view the pace of data center and AI-related construction projects over the next several years.
So I'll jump right in, and then if Julie has anything to add, she can come in and add on to that. But we're seeing really, really robust pipeline looking forward into the next several quarters. There's a lot of projects that are being a part of that pipeline, right? So when we talk about our backlog, as we've mentioned in the past, we have a signed agreement with a scope and a price. When we talk pipeline, we're talking more about opportunities that exist. We're not intending to try to get all of those. We try to get the best, and we focus in on what we think is best, not just from a price perspective, but also from our people's perspective and from kind of a quality of project perspective. So really, really robust and strong pipelines.
And that's in, as you already mentioned, that's in technology with data centers and chip manufacturing. There is some of that on the horizon that we're seeing. They're very episodic and programmatic in how they build out those chip facilities. We see some future opportunity there. In pharmaceuticals and life sciences, one of our, I think, currently our largest contract is on a pharma project related to GLP-1. We have had some big announcements that we'll get our piece of with regard to GLP-1, but just traditional pharma also has a lot of activity as well. It's not just the GLP-1 drug development. And then food processing continues to be something that we see a lot of manufacturing as well. And healthcare is really strong. All of that just to say that we have really strong end markets. It's a very good demand environment.
So from our perspective, it's all about making the right go, no-go decisions. We spend a lot of time with our teams making sure that they're focused on how to make those decisions correctly for their markets and for their people and for their expertise. And so that's what we're focused on going forward, is making sure that we pick the right projects so that we can continue to have success.
Excellent. And you talked about your backlog and your project pipelines. Your backlog's at a record high, $9.4 billion. Your project pipeline, which goes out beyond the reported backlog, is historically high as well. And from an end market perspective, it's obviously skewing towards the technology front. But can you also talk about the backlog in terms of a quality perspective, the types of designs, the scopes, and the complexity within the backlog at this point?
Yeah. So it's skewing, like you said, towards the technology side with the data center customers primarily. But the interesting thing about the way we, at least the way I think about our business, is we're really just about delivering craft professionals to a work site so that we can then construct the facility. And from a craft professional standpoint, the difference between welding pipe in a data center and welding pipe in a healthcare facility or an industrial facility, there is no difference, right? And so it doesn't matter to our craft professionals. So we're very capable of moving in and out of end markets. Right now, the data center end market favors us in some ways just because of our size, our sophistication, and our scale, right? So that's helpful to us from that standpoint.
But overall, in long-term thinking, as you think about, okay, the value of Comfort Systems, it really is embedded in its ability to be the place where the best place where a craft professional wants to build a career. And we have some advantages from that because of our scale and also our ability to kind of manage the locality and the localeness of a business so that a craft professional wants to be part of that business while at the same time benefiting from the scale of Comfort Systems. So that's all just really long-term helps us from a strategy perspective be able to meet demand. And then something we've mentioned before on things and something we're very deliberative about as far as our strategy. We could see a big demand environment coming. No one would have anticipated this sort of demand environment.
But because of that, we acquired Kodiak, which is they specialize in contract travelers and traveling labor. So in the craft professional world, you have a laborer that wants to stay within about 100 miles of their home. That's how they like to do it. And then you have a laborer that wants to travel, wants to go somewhere and spend four to six months working on a project and then go back to where they call home. So those two people, we needed to supplement Comfort Systems on the traveler side. So that was why we brought in Kodiak. And they've been really helpful in helping us flex into all this demand and be able to meet the requirements that our customers have.
Super.
I'll add two things also. Right now, we're able to pick the work that we want. We're turning down work every single day. So from a quality standpoint of backlog, and this has been going on for a couple of years now, but it really is some great projects. We're picking to work with general contractors that we believe run really good projects, that are safe, that we have a good opportunity to have good outcomes on these projects. We're also able to negotiate some good terms on these contracts right now. And that includes having a good schedule of values so we can kind of get it overbilled and being able to collect the money in good order. You can see on our balance sheet, we're significantly overbilled. So yeah, we're being able to take advantage of this environment and pick good work with good terms.
Great. Thank you for highlighting, Julie. And you've talked about project selection for a while, but I keep hearing more about your ability to your preference for customers or repeat customers you've worked with in the past. What makes a good repeat customer aside from getting paid early? What makes a good long-standing relationship with that customer?
It's really, price is an important factor when we look at projects, but it's not everything. What we really are looking for is project work that's really good for our employees, that our employees are happy to work on, that it's going to be safe, it's in a good geography, it's easy to get to. The projects, we think we'll have a good chance of being executed well. The general contractor runs a project that's efficient where we have opportunities to have really good productivity. All those things really contribute to a good outcome on these projects, and I will say right now, we're just seeing our skilled craft professionals are performing really, really well. I mean, the execution is fantastic, and it's a tribute to them.
But I also think it's been the fact that we've been able to pick the work we're on has contributed to those good outcomes.
Excellent. I wanted to dive a little bit into some AI-related trends and your viewpoints on some aspects that we often get inquiries on from investors about how it relates to Comfort Systems. The first one is on generative AI. As that becomes more mainstream used from a business aspect, used by creators, Disney today had an announcement about an agreement with OpenAI. As that becomes more mainstream, are you seeing any inflection points or derivatives in demand related to the broader use of AI tools within your business?
Yeah. I mean, so we're using tools already deeply in the business. What I see from AI is I think there's already, even if you just said, "Let's freeze the models today and we're not going to improve them at all," it's going to take the construction business several years, maybe even a decade, to fully integrate the efficiencies that have been already created in the models. So this is a dumb analogy, but I like to use it because it helps me understand it. When I was first starting off in my career, there were still lots and lots of people in the workforce that didn't use the keyboard because they hadn't learned to type, right? And keyboard is one of the most, it's a huge efficiency tool, right, and massive productivity, but they hadn't learned to type.
So a lot of our workforce will need to learn how to use these tools because there's already the efficiency and productivity baked into them, and they'll just continue to get better, right? And so it's all about making sure we get them in their hands, that we get the use cases. So we have them throughout the entire organization. We're using them and really trying to dig deep on where are we getting the best benefit. And our theory is if we can get it in the hands of people that are doing the work, that we'll then start to see the use cases that are best, and then we'll share them across. And I think we have a real benefit from our scale there because we can be doing it at a much larger scale than our competitors.
And so that really helps us to be able to identify those use cases and then share them across the entire platform. So I'm really bullish on what AI will mean to us from a productivity and efficiency perspective, but it's going to take a little bit of time because you got to learn to use it, right? Just like how people had to learn to type, right? So that's part of the equation.
Super helpful there. And then maybe even a little bit from a different angle there, maybe from a broader, aside from internal perspective, but from a demand perspective. I know you have plenty of work out there, but does generative AI becoming more mainstream and more used by a broader array of applications, has that created any kind of inflection or change in terms of more people knocking on your door for your services?
Well, I mean, we have so much demand in the data center area right now that I'm not sure we'd notice if there was an increase. I mean, I've said this before, and I'll say.
That's interesting. I mean, the fact that it's so much you can't even.
Right. To me, from my standpoint, the announcements they make, they're trying to build, they're at least announcing more build than they could feasibly do, right? And what I mean by that is just the U.S. to support it from a labor, from a regulatory, from an electrical grid, utility. It's all more than can be done. And so from our perspective, that's a demand environment where we're not, let's say you add 10% to that, well, it's still not going to get done. So I mean, so from our perspective, it's not really increasing or flexing demand. We're just going to continue to see it. Yeah, I think everywhere I go, it seems like people want to try to say, "Well, is this a bubble?" And there might be certain companies that are in a spending bubble.
But as an AI bubble, I don't think we're anywhere near it because I think to my point earlier, which is that even just in the construction industry, it's going to take years and years and years for the workforce to understand the productivity tool that has already been developed, and that tool is just going to continue to get better. And the one thing that's been true about AI from the very beginning is that it requires a lot more compute, and it requires a lot more data than anybody ever anticipated it would. And it just continues to, as they iterate on it, continues to require more and more of that. So I'm no expert on that, so I don't want to pretend like I am. But what I can see from my demand lens is that that's not going away anytime soon.
Okay. That's extremely helpful. And you talked about that even if you freeze earlier, when we were talking about internal uses, you were talking about that even if you freeze models, the LLMs where they are now, right? There's still a ton of runway for it to be applied internally. I assume that also extends to your customers who are asking for services too. Maybe aside from the models perspective, the chip technology continues to advance, right? You hear about these Tensor processing chips that are being designed more for machine learning. If the data centers that are being built, if the chip technology changes internally, do you guys have to do anything differently with regards to electrical requirements or cooling requirements? And are you seeing any of that show up in designs or scopes or anything of that nature?
Yeah. I mean, the data centers we're doing now are much denser than the data centers we did five years ago. And so there's just a lot more work for us inside of them, which means just more piping, more electrical, more scope, right? And again, no expert on chips, right? But what we have seen is that data centers that were built for a prior chipset, were built for a prior design, they're not really able to just kind of retrofit those into the AI model, right? They're either having to kind of turn those data centers into legacy data centers that house my Instagram pictures or whatever, or they're having to really just go ahead and say, kind of tear it down all the way down and then rebuild it for an AI deployment, right?
Now, who knows what, again, that's more, they're way bigger experts on this than we are. But what we're just seeing is that what we used to do for if we had X amount of square footage in a data center, what our scope would look like is now three or four times larger than it would have been for that amount of square footage because of the density of these data centers.
Super helpful. I guess as you're thinking, again, shifting back to implementing AI and automation internally within Comfort Systems and the workforce, what can help that integration? Is it just more focus on AI internally? Is it a combination of maybe adding on some companies like Kodiak, which help from craft labor location standpoint? Just talk about that broadly, about how you can kind of become more efficient and other things of that nature internally.
Yeah. Yeah. Kodiak's a great example of us looking and saying, "Okay, super strong demand environment. We're going to bring Kodiak in. Kodiak specializes in these travelers." That's going to let our operating companies feel better about flexing up and down depending on what their demand environment looks like. So that's been really successful. And then on top of that, to answer your question about the AI piece, we had this large influx of ability around deploying our craft professional, but that was logistically very challenging with the recruiting and with the bringing in, etc. So we acquired a very small company called Pivot.
One of the big reasons why we brought Pivot in was because Pivot had this tech stack that really helped us with the Kodiak travelers and allowed us a quick entry into some AI tools that really helped us with recruiting, with retention, with making sure we know whether our travelers database, that's 30-35,000 names big, whether that database is actually still relevant, so it's reaching out to these employees and to these potential employees and keeping them in warm, for lack of a better term, so all of that plays into kind of making sure that we're using this technology to really help drive that productivity and efficiency, and this is just one example, right, of where we're using it.
Super helpful. Maybe shifting to modular a little bit. I think last call you mentioned you'll be at 3 million sq ft by early 2026. And I think, Trent, you also mentioned on the last call that you prudently consider adding more based on strong demand. Can you maybe dive into that a little bit more to the extent that you can? Just talk about the bar that would be needed to be cleared to kind of think about adding more modular capacity beyond here.
Yeah. I mean, I think based on what we've done over the last couple of years, right, we'd almost have to make a public announcement if we weren't intending to maybe consider at least some expansion, right? Because we've expanded every year for the last several years that capacity. So we're always looking at that potential expansion, and there'll be announced here or there smaller expansions and things like that. But long term, what would we be looking for bigger expansion? Well, there's a couple of things, right? We'd want to be sure that we have the commitments from our customers for long-term capacity usage, right? Because we don't want to build it and then find out they're not going to come, right? So we want those commitments.
But then secondly, we want to be sure that we can staff it, that we have the right talent profile in the area where we're building the facility, and that we have the confidence that we can bring in the right craft professionals to be successful. Because what we don't want to do is expand and then have our quality degrade or expand and have difficulty with the ability to staff those facilities. So that's why we're real prudent in how we think about it. And we make sure to have a very detailed and robust plan moving forward. And it's easy to build the buildings. It's easy to install the cranes. It's easy to put the robots in. It's hard, though, to find the people that want to do that type of work.
We want to make sure that we're absolutely solving for that before we do any sort of expansion in the future.
Makes sense. You mentioned earlier, Trent, your largest single booking in the past couple of quarters was in pharma. I thought that was notable. Can you maybe expand a little bit into the current pharma landscape and what you're seeing there at the moment?
Yeah. So that's a GLP-1 facility in Indiana, and it's in the middle of a cornfield. I'll tell you that. It's shocking to see it being built there from my perspective because I think previously that might not have been placed in the United States. I think it's an interesting trend that we're seeing. There was a GLP-1 announcement around Houston that I think we'll probably get our fair share of when it becomes time. So that's just the GLP-1 side of the equation. The traditional pharma side, there are a lot of projects that we're pursuing. There's a lot of projects that we're looking at. There's a lot of projects that are being announced that are all in very good geographies for where we have historically been able to have success in pharma. So that is all. I think there's a lot of reasons it's happening.
I think in part, it comes out of this isn't like they're taking production from overseas and mothballing it and then moving it here. It's more just they're making the decision to build new facilities in the U.S. And I think that's probably a function of business leaders having experienced COVID and the supply chain disruption that that represented. And so they're erring to the side of, "Hey, if we can figure out a way to make this pencil out in the U.S., let's do it here." And I don't think the tariffs are hurting that in any way. So all of that is a long way of me saying pharma continues to really be very, very strong opportunities for us, especially forward opportunities in that market.
Right. Because you mentioned in the past, the longer lead times and longer planning stages it takes to get to that.
Yeah.
To the pharma side.
Yeah. Yeah. Thank you for mentioning that, Julio. Yeah.
Absolutely. And maybe some other end markets, healthcare, education, can you maybe just touch a little bit on those end markets and what you're seeing there?
Yeah. Super strong in both those markets as well. Not the demand environment that the data centers are seeing, but a robust demand environment. And one thing I like to mention because I think everybody focuses so much on the data side, and that makes sense. The data center market's very strong. But if you look at our 50 operating companies, less than 20 are doing data centers because just the regions and the areas that those other companies don't have data centers being built. So they're doing really good business in healthcare. They're doing really good business in education and institutional. So all that is a way of kind of saying, "Hey, those markets are really strong as well." So it's robust kind of across the board.
The only place we see weakness, and we've mentioned it before, is in commercial where you don't see the downtown office buildings. You don't see the hotels. You don't see those types of projects. Much of that in our demand, in our pipeline right now.
Makes sense. But you do see some of that on the service front, I believe, right? And then.
Yeah. Most of our service, so when you look at our service revenue, most of that is in those types of end markets. Yeah.
Gotcha. Julie touched on it earlier on capital allocation historically been pretty consistent. Can you dive into that a little bit more and talk about your priorities for cash?
Yeah. So over the long term, our growth really has been through acquisition, adding that talent, adding that assembled workforce. So we continue to look at really good opportunities. Again, we'll probably spend about 75% of our free cash flow in a five, 10, 15-year period on acquisitions. We just closed. We announced with our last release that we closed on two additional acquisitions, two electrical companies on October 1st. There's a lot of companies that are kind of in that $100 million-$150 million range that are in our pipeline and things that we're looking at. Companies in that $3 million-$500 million, those are going to be more episodic. There's not a lot of those. So those will happen when they happen. We're always looking for good companies. We're not really a fixer-upper. So it's very relationship-based. So we don't have quotas for any year.
We just continue to get to know these companies over a period of time and hope as they decide they want to be sold, that we'll be there and we've built. We're looking for companies that have a culture very similar to ours, right, that really value their employees, and we're a contractor. We're just like them, and we're a good place for these companies to come, so they are getting a little bit more expensive. These companies have a higher value than they did five years ago. We're happy to spend a little bit more money on these companies, but it's really just looking for companies just like us that will continue to grow, and then on an annual basis, we'll continue to do dividends. We've been increasing our dividends pretty steadily over the last couple of years.
The yield hasn't gone up quite as much because of the stock price, but we do continue to increase dividends and then share buybacks. We'll do opportunistically.
Yeah. Have you guys ever think you've ever bought anything from private equity, right? I think.
I can't think. Not any of our contractors. We've never bought from private equity.
Gotcha. It's just been relationship-based, like you said, and probably the right time to.
It's yeah.
Yeah.
In a lot of cases, we're buying it from people that started these businesses. They can be multi-generational in some cases. They're Baby Boomers in a lot of cases. They're looking for an exit strategy. Now, we won't buy something when they're ready to retire. We want those folks to work for us for at least three-to-five years and in a lot of cases for a longer period than that. Because again, we're buying assembled workforce. We got to make sure that workforce comes to be part of the future of that company. Yeah. We'll just continue to develop those relationships and hopefully continue to build the business that way.
Yeah. That's the secret sauce, if you will. Just quick 30 seconds on what aspects of the story you feel investors maybe should hone in on as we look to 2026.
What do you think?
I'll say real quickly, and Trent, you can pipe in, is we tend to talk a lot about the different end markets, data centers and pharma and stuff. I think it's important to remember that our workforce can do any of these things, right? It's not like we have a workforce of electricians and pipe fitters and welders that go out and do data centers, or we have individual operating companies that just do data centers. These companies can do any of this work. An electrician, honestly, they don't care if they're doing electrical work in a data center or a hospital. So I think that's an important part to understand about Comfort Systems is the fungibility of doing these different types of projects.
I would just say short term, the demand environment is really, really strong. Then long term, I think to Julie’s point, we want to be the best place for a craft professional to work, and that’s what adds long-term value to Comfort Systems.
Building legacies, as you guys have mentioned in the past.
Right.
Thank you guys so much for taking the time.
Thank you.
Thanks.