Ready?
Yes. All right. Good morning, everybody. My name is Stefan Diaz, machinery and construction analyst here at Morgan Stanley. We are really excited today to have EMCOR Group here, Tony Guzzi, Chairman, President, and CEO, and Jason Nalbandian, SVP and CFO. But before we begin, I just have a quick disclosure to read. For important disclosures, please see Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative. Great. Tony, maybe for the investors listening on the phone, the investors in the room who are not quite familiar with EMCOR and what you all do.
Yep.
if you could just please give them a quick rundown of the company.
Sure. When you think of EMCOR, we actually do the work. So we are a specialty trade contractor. We primarily put in mechanical and electrical systems, and then we fix them and help maintain them. We do that across a variety of sectors. We are pretty well diversified, from healthcare, the institutional sector, commercial, water and wastewater, transportation, and of course, why 90% of the room is here today.
We also are probably, you never know because there is private companies doing this, we are probably the largest specialty trade contractor, mechanically and electrically, serving the data center market. And we do that in just about every major data center market there is. And we can get into specifics in the Q&A. So when you think of us, we are a company of plumbers, pipe fitters, electricians, welders, HVAC technicians, sprinkler fitters, and the people to fix that.
When you go to our electrical business, we are primarily a medium voltage, or we would say inside the fence, inside the building electrician. That does not mean we do not connect to the substation and bring the wire in, but we are typically working inside the space. We also have a pretty significant low voltage business. So that would be structured cabling and all the low voltage that supports an operation like this.
On the mechanical side, we do heavy piping. Anything from a small pipe, which would be a plumber's job. Plumbing does not mean just bathrooms. Plumbing means medical gas, all those things, high purity piping, to very large diameter pipe, could be up to 64 inches. We also have a sheet metal business. Plumbing and pipe moves water and fluid, sheet metal moves air. The enclosures that move air.
On the aftermarket side, we have a fairly significant mechanical service business, which is two-thirds or more of our building services segment, and the balance being, for lack of a better word, operating engineers. People that are either going to be a Roth technician servicing a branch network or in the building. Then we have an oil and gas business as EMCOR Industrial Services, which it does more than just oil and gas, but it primarily was a downstream refining business with a little bit of upstream, with a very flexible workforce. We serve LNG and other things. Really good welders, really know how to work in mission-critical facilities.
Put all that together, you get to EMCOR, which when I think about it and how we've built a company, I guess the popular world and investment community, we would be your classic serial compounder. We invest for the long term. The other thing I think that is a little different than us, we have a very strong culture. We are a company led by a set of values. We train our leaders. Our core product, we do mechanical and electrical construction, and we do it very well, and service, but our core product is actually great field leadership. That is what distinguishes us versus other companies. Jason, you have anything to add to that?
I think that was a great opening summary, Tony.
Perfect. Thanks, Tony. Maybe just to start, if you could talk a little bit about data centers. I'm sure it hasn't been talked about enough here.
I figure I get, what do they do, about nine hours of meetings? What do you think about six hours and 20 seconds with data centers today?
Your network and communications revenue was up 45% in electrical last quarter and more than doubled in mechanical last quarter. Maybe if you could just give the audience some background. How has the work changed with the size progression of 20-MW builds? Now we're going to hundreds of megawatt builds and multi-building campuses. If you could talk a little bit about the content increase as these data centers get bigger.
Yeah. The multi-building campus is almost three to five years now. That also happens, and you can have cloud data centers in that multi-building campus, and they may put an AI data center. Let's be honest, as contractors, do we really know? We just know we're putting in something bigger.
Sure.
We know how to put the equipment in. We are not IT experts. We know how to install stuff and fix it. The predominance of our business still is cloud infrastructure. That in itself was a great business. We're growing 9% a year. Those things are somewhere between 20 and 50, 75 MW. It's probably where we've gotten really good at doing fixed price work. But there definitely is a shift to more AI, and put a number on that. Is it growing 30% a year, Jason? Is it growing 40% in the market? Does anybody really know? We are definitely participating in the expansion of the size of those data centers. Also, our ongoing cloud business is a great business for us.
We serve lots of markets. We used to come in and say, "Well, we're serving seven markets now mechanically and 16 electrically." I don't know how to do that anymore because there used to be one campus or two campuses in a state, and now that might be five or six campuses in a state. That's irrelevant now. I'd venture to say we're probably servicing in some way Fire protection, we're servicing everywhere in the country. Electrically, we're probably servicing 50% of the markets that matter, 60%. Mechanically, somewhere around 40%, 50%, a little less. The scope has definitely gone up.
You know when you are doing a 200-MW data center, and I always try to ground people on what 200 MW really means. The city of Pittsburgh runs on 300 MW. So just the scope of one data center may use. We know we are doing more mechanical scope, probably 1.5 to 2 times if that was just a cloud data center. It is the heat, right? The heat, it needs to be cooled. And electrically, that might be 1.25 to 1.5. Jason, maybe.
I think the only thing I would add is the pace and timing of these projects, right? If you look at how fast these projects are moving today versus what they would just three years ago, I think that is another factor in terms of what has changed for us.
It definitely informs our investment. So it is always important to remember who we are. We are contractors, and so by nature, we are opportunists, right? But we are strategic opportunists. So in a market, we are going to make sure we take care of our core customers, but we are also going to go to where the best margin opportunity is for us to deploy our labor over that one-to-three-year period.
That is sort of, for a good contractor, your long-term investments are in building your workforce, your training, building your leadership team, investing in the right software, investing in the right prefab, or at least knowing how to expand that, because you got to be ready for the moment. So we got to be where we are in these fast-growing markets. We just didn't fall in it. We had to be ready for the moment.
We were ready for the moment because we invested in all those things I talked about over a long period of time. The way you attack a market like data centers is a couple different ways. One is you have people that were always great at it. They built some of the first data centers in the country, the EMCOR team did. We built the AOL data center in Virginia in 2000, 1999 to 2000. We did the first Equinix jobs down there. And then we did the first big financial data centers. And that team is really what allowed us to expand into the number of markets we are today, and that happens two ways.
It happens through you have great electrical and mechanical companies that you already own that do really healthcare and high-end institutional lab work, manufacturing work, and they can be trained to do data center work, and their teammates do that with them. You do that through acquisition, where someone might be just doing the day 2 work, and because of our scale, training, and capability, we can make them a major data center player.
Some we acquired at a very reasonable cost many years ago, eight, 10 years ago, but we knew that was starting in that market. Then it informs your capital allocation even today. We announced five electrical deals and $600 million in revenue trailing, $100 million, give or take, of EBITDA. The two biggest being Schmidt and Connelly. Schmidt in Central Texas, Connelly outside Chicago, and there we already have a good presence in Chicago. Just for the record, both of those are now closed. We closed them in August.
Early August.
Early August. What will happen there? Well, they already have a great customer base. They did a little bit of data center work, but the market's there outside Chicago to do more, and they can team with, well, we have a great company there, Gibson & Lyons-Pinner, and those three companies will team together to do some other work. In the case of Central Texas, one of the companies we've expanded in the data center market was our Morley-Moss Inc. company.
We bought it back in 2018. It's now 10 times the size it was. Great data center builders, that team. They will now bring Schmidt, who is a very sophisticated contractor and executes well in the field. Their growth will mainly come from more exposure to the data center market. Jason, then, so you think about markets, people thought about Texas, right? But we think Texas is magic, too.
That's why we do outside of the oil and gas business, over $1 billion of revenue in Texas without Schmidt. So we've spent a lot of time through capital allocation preparing ourself for the moment. That's not just data centers. That's high-tech manufacturing. It's healthcare. It's significant institutional work. Then we have the aftermarket business that we'll do single line of service trade in any of those sophisticated studies. Long-winded answer. Jason, you got anything to add?
I think I'm good.
No, I think, Tony, that's really helpful, and I think that's a pretty good segue. You mentioned your contractors.
Yeah.
You go for the jobs that offer the highest returns. Let's say, in a hypothetical situation, if data centers were to slow for a few quarters, do you have the ability to move the workforce around into areas like semis, healthcare, more institutional work? Are you turning down jobs in other sectors to be able to service this data center and market?
The answer to the last question is, of course. Nobody can do everything in a market. You can't. The critical bottleneck for us to grow, I'll come at it that way, is actually supervision. We'll find the tradespeople, and we'll find the right ones because they'll want to work for us. But we have to train that foreman, that superintendent, that project manager to be able to expand.
By nature, EMCOR is diverse. Maybe Jason can talk a little bit about our diversity demand, and that will get to the question. But you also have to be able to reposition your workforce. That also means you have to cut it. A big part of EMCOR is, we try to keep our cost structure as variable as it can be. That's the other thing good contractors do. We lease our prefabrication facilities.
When we build them, we count on a three-year or less payback. We're doing it with an eye towards that market that we're in today, clear-eyed about what it is. Maybe go through some of the diversity demand and how things would actually shift.
Yeah. I think to some extent, the work is the work, and we have a track record of being able to pivot from one sector to another, whether it was when we moved from warehousing and distribution coming out of COVID to the semiconductor space and then into the data center space. We could do that in reverse as well. The other thing is all the investments we've made in prefabrication and some of the construction tools we have, none of them are tied to a single market sector.
You could take those same fabrication shops today that are maybe serving data center customers and use them to serve a different sector. But to Tony's point on diversity, if you just look at EMCOR, let's just take the first half of 2026. We've grown organically about 18.3%. But if you strip out high-tech manufacturing, which is semiconductors, biotech, and you strip out network and communications, which is data centers, we're still growing organically around 9%. There's strength in other markets, and those are things like institutional, which is heavily weighted towards some of the colleges and universities we're working for.
Manufacturing and industrial, which includes both the reshoring or nearshoring work we're doing, as well as some of the food processing work we're doing. Then we're starting to see a resumption in demand for warehousing and distribution. For us, that's in commercial. But it also goes beyond that. If you look at the first half of this year, we saw some sizable growth in our building services business, which isn't tied to the data center market, and that's up about 5% organically. Our oil and gas, our EMCOR Industrial Services business, is up 15% organically. I think that just speaks to our diversity and how we are able to serve all of these markets.
Yeah. If you think about data centers, there's a lot of press out there, and I'm not a political prognosticator. I have my own opinions, which I don't necessarily need to share with you today. It's irrelevant as the CEO of EMCOR. You have to understand where we are in a job cycle. Before we get engaged, someone usually almost always has the power, the site. Usually, in the EMCOR world, they may talk to us about something five years down the road, but we're not listening, really. For us, there has to be a job. There has to be how they're going to prosecute or execute that site, what our role is going to be in it.
My view, and I think our constructive view is, for the next three years, they pretty much know what they're going to build. That won't be affected unless we really go draconian, which I don't see happening, by moratoriums or anything else. I don't think the political thing is what's going to drive this, by the way. It might be an antecedent of it. All the power we said we need to put in in these places, that were sold off for gas turbines through 2031.
I know everybody in the room probably knows that because you track things like that, and the three or four major producers. We're talking large combined cycle things. That's what's going to power this long term, or eventually nuclear. That's what powers a base load system. That will be the obstacle if we don't do that by 2030. It won't be all this other stuff. Because when we talk to our customers, and we're talking both to the people that will actually occupy the co-los or the people that are building themselves, they're doing both. They can't keep up with their product developers.
For the most part, our customers aren't people like Anthropic and OpenAI. Our customers are who you would think they would be. For the people like us in this room, they can't keep up with demand, what you all want and what we need to run our business, we think, long term. The cloud business is a pretty good business in its own right. Of course, my personal view is I always ask people, and I pull out my phone, "What are you willing to give up?" Once we answer that question, we can have these other talks. The power question is the real question.
When you get to there, behind the meter, in front of the meter, I don't think it matters because most of the behind-the-meter solutions will be dispatchable power eventually. The behind the meter means we're funding that solution. You don't want a data center in your neighborhood. You certainly don't want a combined cycle power plant or a modular nuke. We don't want the data center, but we're going to put a modular nuke at the end of the cul-de-sac. Really? I think the power will happen. It almost has to happen. But there will be people that win and lose in this. I think the good news about who we are is we are very well positioned for where it's going to matter. That's intentional. That's not an accident.
That's really helpful, Tony. I do want to let the audience know, if you do have a question, feel free to raise your hand. We'll get a mic over to you, and you could ask. Maybe just moving along here, outside the data centers, we've seen some recent announcements on the semi side, your high-tech manufacturing business. Where do you think we are in the semiconductor and the pharma build cycle, and where do you see demand progressing in high-tech manufacturing as we go into 2025?
I'm going to hit it here, and Jason has. We've done some interesting analysis that, first of all, semiconductor work, there's about six geographic markets that matter. We're in four of them, and we're in all of them fire protection and sprinkler-wise. It's lumpy work, and the initial fab they build on a site comes in at a big number as far as you get a contract. Then quite frankly, once you've been there, sometimes it comes in in smaller numbers. You necessarily don't see a backlog, which you may see in revenue eventually. Where do I think we are in a and look, we can participate. In three of those places, maybe three and a half, we're very good mechanically. Electrically, a couple of them, and fire work and sprinkler everywhere. It's also very difficult work.
Doesn't mean you can't make good margins at, doesn't mean It is very difficult work for the most part, pretty difficult customers. They're all difficult. The data center people are difficult, too. But they're good customers, because they value what we do. But sometimes we're making a trade-off. In one state, we made a trade-off because there's a bigger data center build happening. We're really good at it. We're going to sit out the next round of the fab, probably come back for the third one as we continue to build up our labor force. Pharma's a little different. I think a chunk of what we're doing today is tied to GLP-1s, and there you have to be positioned in the right places. We are. We're in New Jersey, we're in North Carolina, we're in Indiana.
We're here in Southern California, where they're doing sort of more of the bio research. So we're where we need to be. There, we're both mechanical and electrical. I'd say probably general. Like New Jersey, we have more of an electrical presence to do that work. They're mechanical. North Carolina, we have sort of both. Southern California here, probably more electrical. But all of them are really good markets for us. But again, go back to what my statement was earlier. We're contractors. We're going to look at that market. We're not going to exclude ourselves from something that matters. But over that two-year period, we're going to try to get the mix right to maximize margin and our return to you. Jason?
Yeah, I think to Tony's point, some of the growth we've seen recently in data centers did come at the expense of semiconductors. But if you just step back and you say the high tech business, which is roughly 40% semi, 40% biotech, and 20% some EV battery work that we're doing, if you just look at that sector overall and you say today versus three years ago, we're still operating at a base that's almost 1.5 times what it was in 2023. If you kind of did a three-year CAGR from 2023 to today, you'd see that we're growing there at almost 11%. So there's still a lot of opportunity. I still think to Tony's point, we're well-positioned.
If you snap the line in 2024, it had shown we were growing 30%, right? Because we had a big job, right? No, we like it. We've been working in Arizona in semiconductors. I mean, I've been here 22 years, so 1998, 1997.
Wow.
Something like that. So that's the thing about our capabilities. We have deep capabilities in some of these markets. Where we didn't, we acquired the right companies and were able to take our folks, and they were good teammates, and they brought people along. We share knowledge probably better than any company in the industry. We share resources. I mean, what we can do on the VDC side to share resources across our company, we have 1,500, 1,600 BIM VDC people. Scale matters. The type of work you're talking about, whether it be data centers or semiconductors or pharmaceutical or healthcare, especially, the virtual design construct, VDC, that then takes it through to prefab, that's a real skill. That then informs our CapEx spending and what we do there.
Tony, I think that's a really good transition. You just mentioned VDC, prefab. I think in the last earnings call, you mentioned that a good portion of your CapEx-
Yeah
is going towards prefab. Maybe what's the right way to think about your prefab capacity expansion, and then how should we think about CapEx as we go into 2027, 2028? Does this investment step down or?
I don't think it steps down. Because we're growing. The absolute dollars don't step down. About 20%-25% of that CapEx is just maintenance CapEx. It's leases, it's building upgrades on office space, it's specialized vehicles that we can't lease. It's things like that. The rest, the growth in that CapEx as we build out our modular and prefab capability, which we had, and then we just continue to grow it. Maybe go through the math on that, Jason?
Yeah. I think if you look at, I don't know, let's take a five-year look. If you look, let's say our revenue CAGR is about 14%, 15% over a five-year period. Our CapEx CAGR is nearly twice that. It's about 28% to 30%. That delta is those investments that we're making in our shops, and I think it takes two forms. Sometimes it's the addition of new fabrication space, the expansion of existing shops.
But the other element of it is us reinvesting in existing shops to make them more productive. How do we add more automation? How do we change out equipment to make sure that the layout is best so that we're getting the most productivity and the most efficiency? So I think, we're not capital intensive. Our CapEx is 0.6% to 0.65% of revenues, and I think that holds into 2027 and 2028. The way I would think about it is any CapEx growth in excess of revenue growth are those investments into the future.
The rest are just maintenance capital.
Right.
Maybe if we could just take a moment to talk about your RPOs. They've been growing meaningfully faster than your revenue guidance. Maybe how should investors think about conversion versus history, and do you think revenue growth should begin to catch up to RPO growth in the next few years?
I think a big piece of the delta is just how much work we've booked in the last two quarters.
Yeah.
We had significant bookings in Q1 and Q2, and I think we're carrying today more revenue in RPO than we did historically. We used to start a year, and we used to say, "We need to go book and earn 60% of that year's revenue in that year. That number has steadily come down as project sizes have went up and our pipelines have expanded. In recent years, it's come down to about 40%-45%, and I think that's what's driving.
Yeah
The disconnect between RPO growth and revenue growth.
The other thing, I think it's always important when you think about us. Those numbers over time are all apples to apples to apple. Everything is consistent because our RPO definition is the accounting definition. We're not making guesses in those RPOs. We have a contract.
It's executed.
It's executed.
There's a firm price.
Or it's a change order that's been approved. We're not making guesses in those RPOs about what might happen. We might be on a data center site, and we know we did building one, and they're going to build five buildings, and we're probably going to do one, three, and five. Maybe other people in this space would say we're putting all that in there and guessing at it. We don't do that because, one, you don't know.
You still have to perform, right? It's sort of like working with you all. We build one building, we had to do it great. You earn the right to go do the next one. And what I can say is we have pretty good track record of doing that. A lot of people lead that way, because we finished work for other people, but no one's ever finished our work, especially in the data center space.
Sure.
We're on the conservative side with RPOs, maybe versus other folks in our space.
RPO?
Remaining performance obligation.
We just call it the backlog.
When people say backlog, there's people that will put in, they're guessing what their T&M work's going to be in a year. They're guessing what the next couple jobs are on a site. We don't do that. It is an accounting definition. We have a three or five-year service agreement. It's the non-cancelable portion of a service agreement, which is usually 90 to 120 days. This all came out, what, Jason, about eight years ago?
Yeah, 2017, 2018.
Somewhere in there. We were so close to it anyway, we just went with the accounting definition. That's true with our numbers too. We report GAAP numbers. We figure you're all smart enough to add back the amortization and see what that number is.
Maybe we could spend a minute or two here on margins. Last quarter, you posted healthy operating margins.
I think it was the best ever, right?
Close to 11%, but I think on the call you noted that these levels aren't necessarily run rate going forward. Maybe if you could just explain to the audience why margins move around, maybe quarter-to-quarter, and what's the right band for investors to think about margin trend?
For a long time, I've had one statement always, and I'm going to say it now. It's like with that statement at the beginning.
My edge flavor.
Safe harbor. This is our safe harbor statement. This is not a quarterly business. We also don't get to say we're not a manufacturer. We don't get to say we made 10.6% margins, and now we have all these standard costs in, and we know what they are, and that widget's always going to be, or that air conditioning is always going to be that for the next three quarters because we know what our costs are. Here's what we know, is our operations for a very long, our margins for a very long time, and our RPOs a very long time, can bounce around quarter to quarter. Then they do settle into bands because that's what the composition of the market right now maybe is.
I think Tony's point's very valid. It's a project-based business, so just mix or project timing, even execution can have an impact in any given quarter. But for us, we think if you want to look at a sustainable margin guidance for us and say in the near term, where do we think margins can settle? We really do believe a rolling 12 to 24-month average is pretty indicative of what our business can do.
If you start at the consolidated level, it gets you right around where our margin guidance is today, 9.5%-9.8%. That's operating margin. If you looked at each of the segments, our electrical business would be somewhere between 12% and 13%. Our mechanical business today would be somewhere between 12% and 12.5%. Our building services business would be 6% to 6.5%, and industrial would be right around the 3% margin.
It has a pretty heavy amortization load yet. Part of what is going on there.
If we could just talk a little bit about contracts. Recently, we have seen maybe a little bit of a shift to guaranteed maximum price, more cost-plus contracts as opposed to fixed-price. Why are we seeing this shift to these type of agreements, and do you think this is maybe more structural, given these jobs might be more complex given the AI data center build-out?
Yeah. We are still predominantly a fixed-price contractor and always will be. You get into a guaranteed maximum price area sometimes in healthcare, and then sometimes complex manufacturing jobs will take you there, and the more complex data centers will take you. We have one client that that is how they build. Now, they are very rational and reasonable what they allow us to put into cost, so their margins tend to be a little better, and it tend to look like a fixed-price job from a margin standpoint sometimes. But what leaves us and them there? You take risk and return. We try not, quite frankly, to operate in the high-risk, high-return box. It is a dangerous box to be in when you are deploying highly skilled labor. What puts you in that box?
You have not built one before, you have not built one in that location before, you have not built one for that customer before, you have not built one with that engineering team before. It does not have to check all those box to put you into high-risk. That changes. One could be weighted more than the other. But four or five of those things together, if you are hitting two or three of them, you sit there and go, "Huh, maybe we want to be in a GMP environment," and we got to convince the customer.
What is happening today, because again, these are really smart customers. There are no surprises. We are putting each other into the decent return, good outcome box together. Because what do not they want to do? One, they do not want us not to work for them, right? They want us to be on the team because we're good, and we want to be on the team because they're good.
On these more complex designs, these larger AI designs, it wouldn't be in either one of our interests probably to work in fixed-price, especially in the mechanical world, because they're figuring out how they're going to do this the most efficient way possible, and what the cooling system needs to be, and the piping system needs to be to do that. That may change mid-job. Or they want to share information openly with each other so they can get an idea as they're building the next design what things really cost, versus we just gave them a fixed price for the whole job.
In those guaranteed max price jobs also typically have a reasonable change order mechanism, and that allows the job to keep flowing. Because you don't want to get into, we don't want to price an acceleration change order, and that change order could come either a little change in design, or they may want it done faster. They don't really want to catch us, right? They want to have an open book about what that's actually costing.
For the largest job, they've been very fair on what cost looks like. Sometimes they'll help pay part of our fabrication cost to expand our facility if it's on site. The general conditions tend to be fairly generous because they want to attract the workers. Part of what you're doing in some of these sites is you're trying to build a workforce. To build that workforce, you need to know what you're going to have to pay them. In some of these sites, they haven't had a project of this scale.
So what are you going to need to do to attract labor there to build this over the next couple of years, knowing that you're not going to keep, you'll hopefully keep your supervision, but most of that labor will cycle in and out of there over time. That's what takes you to that box, and that's going to be part of our mix, I think, for a while, Jason, because of that, both from our choice and theirs. We don't just get to dictate that, but if we think it's in that higher risk, higher return, or even higher risk, moderate return, we will sit down with the customer and try to push it that way. You almost have to, right?
Great. Well, thank you, gentlemen. We're actually. If we could sneak in one more question.
Yeah.
You could just say it. We're almost out of time.
Yeah, far away.
Mic is coming.
The question is, given all the debate in the U.S. on question on especially the debate of increasing inflations and wages.
Yeah
People in your business are so busy that some of your people and also independent contractors can have jobs and increase salaries by two, three, four times and work all over the place, especially more the specialized workers.
Yeah, I don't think.
Given your.
Yeah.
Nice $100 million kind of hours worked, and to figure out how, let's say, the salary cost and how that
You really can't do that by any mathematical equation.
No, but the question is more like how you sort of manage the cost
Sure
Because, and then also whether because you're a price-fixed operator, how you sort of manage that, because that's probably one of the most important thing to.
On a fixed-price job, we would never take that unless we knew what our labor cost was going to be on that job on a dollar per hour basis or what that mix of labor between apprentices, helpers, and everything to get us to a composite labor. We would never do that. One of the reasons you end up in a GMP world is because you sometimes need to structure the package of wages, base wages, overtime, and per diem to get the people there.
To retain people, is that difficult? To retain people, because there is a lot of demand of your current people.
The trade worker themselves, about 40% of our workforce, 50% is actually they work for us all the time, and the supervision is usually ours. That is who we focus on. The tradespeople, a chunk of them will come in and out of a job, right? If it is a two-year build, they may leave. They work for six months, all that overtime, leave for four months, come back for six months, and then leave again.
Actually, the base wages have been very pragmatic of what folks have asked for in the union contract. That is not where they are focused. Their focus is what the package is going to look like on that job, and they want to stay competitive, too. As far as us or wages in the salary side, it really has not been a whole lot different. We have terrific retention at our CEO level and our direct reports. Why? Great company to work, and we are very fair in how we pay people. Where people have the opportunity at the salary level at EMCOR or the superintendent level is in the bonuses.
And would you.
We're actually out of time. I'm sorry.
You can join one of the meetings maybe.
Tony, Jason.
Thank you.