A Large Homebuilder Hedges With Robots, Micron Warps a Region, and the 11% Nobody Can Pass On - October 2026 Reviewed as Noted의 전체 녹취록

these regional contractors, 00:04:43.295 --> 00:04:49.846 are, is there a recognition of, like, are they, is any bit of their strategy 00:04:49.846 --> 00:04:55.056 being informed by these kind of regional growth statistics in terms of where 00:04:55.056 --> 00:04:57.877 to invest and where to resource themselves? 00:04:58.365 --> 00:05:04.189 I'm not hearing it like that. I'm hearing it more like we could maybe do some of that too. 00:05:04.485 --> 00:05:09.466 Like healthcare contractors saying, you know, I could probably become part of 00:05:09.466 --> 00:05:12.106 a JV for a data center, right? 00:05:12.106 --> 00:05:18.336 Like the kind of bigger story is that lots of times people don't have enough 00:05:18.336 --> 00:05:22.406 people to do the data center job, so they're JVing all over the place. 00:05:22.406 --> 00:05:25.852 And that starts to open up partnerships that they might not have otherwise done. 00:05:26.462 --> 00:05:29.766 So I'm hearing about people from, you know, the East Coast that are setting 00:05:29.766 --> 00:05:34.386 up shop in Texas. Now, we'll see what happens, whether that's a pull is quite 00:05:34.386 --> 00:05:36.583 to Texas as much as it was, but you get my point. 00:05:37.106 --> 00:05:40.879 So I think it's a little bit more like people thinking outside of their region, 00:05:41.396 --> 00:05:46.086 because there's just so much demand in these pockets where, you know, 00:05:46.086 --> 00:05:51.295 the combination of permitting and other factors are driving primarily data centers. 00:05:51.867 --> 00:05:56.256 It's just skewing the whole market. Like this amount of money going into something 00:05:56.256 --> 00:05:57.985 is just warping literally everything. 00:05:58.473 --> 00:06:01.766 You know, how long is that going to last? Certainly another year or two. 00:06:01.766 --> 00:06:06.436 I mean, even if the new Congress throttles it a little bit, they're not going 00:06:06.436 --> 00:06:07.416 to throttle it that much. 00:06:07.416 --> 00:06:12.126 And it's just going to make that what was happening already with transparency 00:06:12.126 --> 00:06:15.376 and sort of water and electricity use was kind of happening already. 00:06:15.376 --> 00:06:18.404 It's just going to now be regulated more than it was. 00:06:19.687 --> 00:06:24.458 And, and, and Hugh, do you, do you, don't you think that it's like drawing the, 00:06:25.018 --> 00:06:28.471 best and the most valuable workers away? 00:06:28.842 --> 00:06:36.298 I mean, I think about who is, who's capable of putting their wife in an RV and driving to Texas. 00:06:36.298 --> 00:06:40.648 Like the guy, the guy that's going to do that has, he, he's got a family to 00:06:40.648 --> 00:06:43.658 support. He's got some kids that are in college. He's like, like, 00:06:43.658 --> 00:06:47.468 this is a major commitment, right? It's a big life change to drive across the 00:06:47.468 --> 00:06:49.810 country and take on a job for two years. 00:06:50.048 --> 00:06:56.188

It's not the 20-year-old apprentice in San Diego that is going to community 00:06:56.188 --> 00:06:58.395 college to try to become an electrician. 00:06:59.156 --> 00:07:04.003 I wonder about that. I wonder about that because, again, the volumes are so big. 00:07:04.415 --> 00:07:07.798 I'll give you an example. I do some work with SmackDown, and not them, 00:07:07.798 --> 00:07:12.948 but a related group that they're part of just got a bunch of money and, 00:07:12.948 --> 00:07:15.318 in fact, have been offered, and we're talking millions. 00:07:15.318 --> 00:07:19.578 From, I believe it was Meta, but anyway, one of the hyperscalers to really drive 00:07:19.578 --> 00:07:23.518 a lot of training, right? And they've gotten the same type of offer from apparently 00:07:23.518 --> 00:07:26.148 from other people. I don't know the details. I just literally met people at 00:07:26.148 --> 00:07:27.226 their conference last week. 00:07:27.717 --> 00:07:31.188 So no, I think it's pulling everybody. If anything, I would say some of the 00:07:31.188 --> 00:07:32.564 senior folks don't have to. 00:07:33.092 --> 00:07:36.868 People with a lot of experience where the move is, I don't want to pull my kids 00:07:36.868 --> 00:07:37.817 out of school or whatever. 00:07:38.334 --> 00:07:42.008 They may be less likely to do that. Whereas maybe it is the 25-year-old who's 00:07:42.008 --> 00:07:43.681 more likely to go because they can. 00:07:44.538 --> 00:07:47.288 I don't know. I think it's, that'd be the kind of thing you really want to see 00:07:47.288 --> 00:07:49.944 the data on because there's probably more than one thing going on. 00:07:50.536 --> 00:07:55.470 Man, they're desperate for folks at, you know, at scale. So I think, 00:07:56.478 --> 00:07:58.483 I don't know. I think that'd be a good thing to see some data on. 00:07:59.784 --> 00:08:02.836 I was curious, this was something I think we talked about last week, 00:08:02.836 --> 00:08:08.936 Stephen, or last month, Stephen, is like the statistics were completely skewed 00:08:08.936 --> 00:08:12.910 towards these megaprojects, but everything else seemed like it was down. 00:08:13.305 --> 00:08:18.466 So like anything outside of megaproject territory or like data centers and we'll 00:08:18.466 --> 00:08:23.206 say like industrial type projects, that was down. Has that remained kind of 00:08:23.206 --> 00:08:26.160 true? Did we see that anywhere this month? 00:08:27.125 --> 00:08:30.816 Yeah, I don't think I pulled the same data source, but I think it would be interesting. 00:08:30.816 --> 00:08:34.196 And I think anecdotally, I think that is true. And I think the question there 00:08:34.196 --> 00:08:39.546 is, is it because owner demand is down in those verticals, or if just all the 00:08:39.546 --> 00:08:43.332 resources are being drawn away from those verticals, and they can't support them? 00:08:43.763 --> 00:08:48.116 I think cost of capital is absolutely going to be part of that because we already

00:08:48.116 --> 00:08:51.703 have had interest rates go up, which, you know, that's just kind of across the economy. 00:08:52.028 --> 00:08:56.756 But also there's just so much capital that's going to be invested in anything 00:08:56.756 --> 00:09:00.816 real estate related. And if you've got this much demand, it's just crowding 00:09:00.816 --> 00:09:03.902 things out. I'm sure that has something to do with it. Yeah. 00:09:04.803 --> 00:09:08.341 Construction finance is so weird anyway, you know, like the bank does a little, 00:09:08.341 --> 00:09:10.111 and then this guy does a little and that does a little. 00:09:10.111 --> 00:09:14.311 So, you know, it already is a, is a kind of a complicated picture. 00:09:14.311 --> 00:09:18.641 And I have a feeling that between all of this other, other demands on capital, 00:09:18.641 --> 00:09:22.131 it's just crowding some of it out. So things that were on the, 00:09:22.131 --> 00:09:24.173 on the borderline, maybe get pushed back. 00:09:25.241 --> 00:09:30.231 Yeah. So, well, the, the, the thing for sure, I I've been watching the census 00:09:30.231 --> 00:09:35.391 numbers on this stuff for a couple of years and we saw in march this year by 00:09:35.391 --> 00:09:37.819 march this last this year there was, 00:09:38.531 --> 00:09:43.351 major indicators that the that the flip was underway right that that, 00:09:44.154 --> 00:09:47.991 these large industrial and we're not just talking about data centers we're talking 00:09:47.991 --> 00:09:51.382 about chip fab a little bit other re-industrialization right, 00:09:52.491 --> 00:09:57.971 compared to non-residential or compared to residential starts being down compared 00:09:57.971 --> 00:10:02.124 to all of these home builders having to give major discounts on, 00:10:03.031 --> 00:10:06.871 on home purchases and all that stuff you know they're trying everything they 00:10:06.871 --> 00:10:10.561 can to keep some of these other markets afloat from a capital point of view 00:10:10.561 --> 00:10:13.085 and it's just still not it's not holding. 00:10:14.506 --> 00:10:18.141 Sponsor break if you spend any time running a project you know the answer is 00:10:18.141 --> 00:10:20.791 almost always sitting in the contract document somewhere, 00:10:21.354 --> 00:10:26.941 catch is that it's almost never in one document It's split across the contract, 00:10:26.941 --> 00:10:30.551 the specs, and some flow-down clause three layers deep that nobody connects 00:10:30.551 --> 00:10:32.383 until it's already cost money. 00:10:32.534 --> 00:10:36.411 That's the whole premise behind Document Crunch. That risk doesn't live inside 00:10:36.411 --> 00:10:39.100 any single document. It lives in the gaps between them. 00:10:39.349 --> 00:10:43.854 And most teams never look at all of them together until they're standing in a dispute. 00:10:44.545 --> 00:10:49.105 The average construction dispute in the U.S. now runs about $60 million, 00:10:49.781 --> 00:10:52.172 according to Arcadis. It's a lot of money. 00:10:52.573 --> 00:10:56.535 And the root cause they point to isn't fraud or bad intent. It's people not 00:10:56.535 --> 00:10:58.157 fully understanding what they agreed to. 00:10:58.384 --> 00:11:01.263 So Document Crunch builds a platform around closing that gap. 00:11:01.455 --> 00:11:07.068 Crunch AI reads the documents, surfaces the risks, and it cites the exact page and clause that's big. 00:11:07.416 --> 00:11:11.245 So you're not just taking its word for it. Project Assist lets your team act 00:11:11.245 --> 00:11:14.495 on it, ask questions across the entire project document set, 00:11:14.495 --> 00:11:18.874 hand draft your submitals, notices, and RFIs right from the chat window. 00:11:19.101 --> 00:11:23.475 And the whole thing runs on your own company standards. The office and the field 00:11:23.475 --> 00:11:26.868 are working off the same playbook for bid through closeout. 00:11:27.158 --> 00:11:32.165 No disconnects. It's live now across the U.S. and Canada. Go to documentcrunch.com 00:11:32.165 --> 00:11:35.517 forward slash construction conversations and book a demo. 00:11:37.425 --> 00:11:42.885 You mentioned how this might impact kind of the labor pool long term as we kind 00:11:42.885 --> 00:11:46.816 of optimize for a very specific maybe variety of build, 00:11:47.545 --> 00:11:52.005 go out on a limb like do you have any kind of hypothesis on what that looks like of, 00:11:52.665 --> 00:11:56.769 for the workforce as we kind of optimize for speed and optimize for these very 00:11:57.053 --> 00:12:01.445 intricate kind of mission critical oriented workflows yeah i'm not sure that 00:12:01.445 --> 00:12:04.624 it that i meant that that it optimizes for that that, 00:12:05.209 --> 00:12:08.925 project type so much as it's just people are making a lot of money doing this 00:12:08.925 --> 00:12:15.645 yeah so i think you know in five or ten years somebody who came in as an electrician or as an hvac, 00:12:16.605 --> 00:12:20.255 you know worker they're going to have all that capability and they'll have dealt 00:12:20.255 --> 00:12:25.515 with really complicated projects that were fast and like like really tense so 00:12:25.515 --> 00:12:28.665 these are going to be they're thoroughbreds compared to you know someone that 00:12:28.665 --> 00:12:30.085 may have come up in something else, 00:12:30.578 --> 00:12:33.225 and i think there's a there's I wouldn't call it a perfect storm, 00:12:33.225 --> 00:12:34.595 but there's more than one thing going on. 00:12:34.595 --> 00:12:38.857 Like how many, how many, I got a nephew who's, who's 15. 00:12:40.012 --> 00:12:43.208 Does it make sense for him to think about college? I, you know, 00:12:43.528 --> 00:12:47.152 like maybe, but it's much more of a question than it was. 00:12:47.552 --> 00:12:49.618 You know, you look at what it, what it's going to cost. I mean, 00:12:49.618 --> 00:12:53.468 it's not crazy to think it's 400, $500,000 to get. 00:12:53.468

--> 00:12:58.588 I mean, obviously you're talking a pretty expensive school, but not a crazy one either. 00:12:58.588 --> 00:13:03.208 You know, think about a half a million dollars to get a degree in, 00:13:03.648 --> 00:13:09.078 you know, in studies of something or whatever. Like, I just think that the automatic 00:13:09.078 --> 00:13:11.450 let's go to college thing is just not what it was. 00:13:11.897 --> 00:13:17.308 And what that's going to do from, you know, to the trade workforce, 00:13:17.308 --> 00:13:19.914 I think it's just like you look at someone who doesn't have that. 00:13:20.245 --> 00:13:22.978 They're making $150,000 when they're 22 or whatever. 00:13:22.978 --> 00:13:25.916 I mean, like some of it, especially when you add in these crazy bonuses, 00:13:26.329 --> 00:13:28.358 you know, thinking about what you were talking about, Aaron, 00:13:28.358 --> 00:13:34.478 where they force people to use local, what do you call it, local resources who may not want to come. 00:13:34.478 --> 00:13:37.598 So they're giving them like a morning bonus and then an afternoon bonus. 00:13:37.598 --> 00:13:40.298 I mean, you hear these nutty stories about keeping people engaged. 00:13:40.298 --> 00:13:44.708 Like, that's a lot of money by the time you're done. So I just feel like we're 00:13:44.708 --> 00:13:47.028 going to have more of a workforce than we otherwise might have. 00:13:47.408 --> 00:13:54.309 But we've had a secular problem with our workforce shortage for like 50 years. It's not new. 00:13:54.709 --> 00:13:57.428 So I don't know that this is going to completely solve it. I think it's just 00:13:57.428 --> 00:13:58.761 going to make it better than it would have been. 00:13:59.701 --> 00:14:03.058 I think something else that gets lost in the data is because there's been so 00:14:03.058 --> 00:14:07.178 much growth in data centers and some element of manufacturing, 00:14:07.825 --> 00:14:12.243 it kind of shadows a lot of the infrastructure work that is actually going on. 00:14:12.359 --> 00:14:16.218 You know, I've talked to a number of infrastructure GCs in the last couple weeks, 00:14:16.218 --> 00:14:19.418 and there's projects across the country, and they're large projects, 00:14:19.418 --> 00:14:22.778 but they're in the billions, but they're not in like the 25 billion, right? 00:14:22.778 --> 00:14:26.978 And so I think rewind 10 years ago, and that would have been the project on 00:14:26.978 --> 00:14:28.257 the face of Construction Dive. 00:14:28.570 --> 00:14:31.757 And now just because of the scale of how Mission Critical has grown, 00:14:32.007 --> 00:14:35.612 that's now maybe a non-mega-mega project, which is interesting. 00:14:36.922 --> 00:14:42.021 Let's move to robotics. So Promise Robotics announces a multi-year agreement 00:14:42.021 --> 00:14:43.801 with D.R. Horton to support more 00:14:43.801 --> 00:14:48.648 than 1,250 homes. Horton is the biggest builder in the U.S. by volume. 00:14:49.182 --> 00:14:52.625 The model is factory-as-a-service. Horton doesn't own the plant. 00:14:53.159 --> 00:14:58.661 I think that's interesting. So I guess Horton does about 85,000 homes a year. 00:14:58.661 --> 00:15:01.834 So this is certainly not a majority of their build per year. 00:15:02.258 --> 00:15:04.050 It's about 1.5%. 00:15:04.514 --> 00:15:06.657 So very much in the pilot kind of phase. 00:15:07.230 --> 00:15:12.151 But I think it's an interesting kind of model that we're seeing. 00:15:12.151 --> 00:15:15.525 So I guess, Hugh, the question is, we've talked about robotics quite a bit. 00:15:16.094 --> 00:15:23.001 I think as model kind of capabilities have improved and maybe the brain side 00:15:23.001 --> 00:15:27.741 of robotics has improved, is this a breakthrough moment for kind of the business 00:15:27.741 --> 00:15:29.241 model associated with robotics? 00:15:29.241 --> 00:15:32.904 Or is this just kind of one more iteration towards something else? 00:15:34.011 --> 00:15:37.774 I would probably say it's the latter, but it's probably an important one. 00:15:38.192 --> 00:15:43.511 You know, like, I think breakthrough is a, is a strong word only because construction 00:15:43.511 --> 00:15:48.101 is so complicated in terms of who's going to buy what and what project is this 00:15:48.101 --> 00:15:51.422 right for and so on. But I think the intuition you're getting at, which is. 00:15:52.096 --> 00:15:57.098 The ability of a facility, a factory that has robotics in it to be a lot more 00:15:57.098 --> 00:15:59.067 flexible than it was and much more. 00:16:00.158 --> 00:16:04.808 Able to reconfigure for, you know, something different is clearly better than 00:16:04.808 --> 00:16:07.828 it was and going to get even better, you know, pretty soon. 00:16:07.828 --> 00:16:11.448 I mean, whether it's, and we see a lot about humanoid robots, 00:16:11.448 --> 00:16:15.478 but I think the point is what goes into those doesn't need to be humanoid. 00:16:15.478 --> 00:16:19.422 It's still that those advancements can live in something that's a little more specialized. 00:16:19.916 --> 00:16:25.258 So thinking about factories that are are you know much much more like you know, 00:16:25.758 --> 00:16:31.324 anyway customizable i think is is absolutely right the business behind it like one of i remember when, 00:16:31.980 --> 00:16:34.989 there was a couple of companies right before covid, 00:16:35.582 --> 00:16:39.558 spent some money out of chicago one in particular in chicago and they had trouble 00:16:39.558 --> 00:16:44.478 with it because i'll just say who was it was skender but that but they had they 00:16:44.608 --> 00:16:47.378 built a factory and it was pretty cool. 00:16:47.378 --> 00:16:50.146 And it was, there was a lot of smart people that were involved in it. 00:16:50.715 --> 00:16:54.488 It wound up running into trouble because of financing, not because of, 00:16:54.488 --> 00:16:57.258 I mean, the demand wasn't what they hoped it would. And obviously

COVID had 00:16:57.258 --> 00:17:01.218 an impact on that, but the point was, it wasn't that it failed in the market 00:17:01.218 --> 00:17:03.466 is that it failed in the financing. 00:17:03.850 --> 00:17:07.721 So the problem a little bit with stuff like this, that is a really different business model, 00:17:08.203 --> 00:17:12.978 is sometimes they, they, they struggle with some of the things that make it possible to do, to do, 00:17:13.458 --> 00:17:17.028 the business you know and maybe it's financing maybe it's something else maybe 00:17:17.028 --> 00:17:20.198 you know what i mean maybe it's you know the architect's ability to deal with 00:17:20.198 --> 00:17:24.201 it who knows but the point is there's so so much complexity that goes into, 00:17:24.741 --> 00:17:26.715 something like this that that. 00:17:27.438 --> 00:17:30.749 They may have had on the horton side had to do an awful lot to make it work. 00:17:31.614 --> 00:17:35.298 That that that limits how much this means for the rest of the industry in the 00:17:35.298 --> 00:17:39.238 short run so i think it's a really important thing and and it Maybe there's 00:17:39.238 --> 00:17:43.724 five more steps like this before it becomes the way that housing gets built. 00:17:45.388 --> 00:17:48.262 Well, I think, you know, how they, in one of the press releases, 00:17:48.262 --> 00:17:49.922 I think they talked about D.R. 00:17:49.922 --> 00:17:54.012 Horton kind of passing on drawings for the different templates of houses and 00:17:54.012 --> 00:17:58.992 Promise Robotics kind of consuming that and building out a plan and a factory 00:17:58.992 --> 00:18:03.892 to respond and create these kind of components that get assembled on site. 00:18:03.892 --> 00:18:06.962 And so, you know, variability kills that plan. 00:18:07.113 --> 00:18:08.982 And so I think that's why they're starting regionally, right? 00:18:08.982 --> 00:18:10.109 They're rolling out in Phoenix. 00:18:10.875 --> 00:18:14.466 You know, Tyler, you've kind of leaned into AI and agents. Yeah. 00:18:15.092 --> 00:18:19.702 On a, in your personal life. Have you like any humanoids? What's your, 00:18:19.702 --> 00:18:22.907 you have any robotics? I'd call them ever. No. 00:18:23.499 --> 00:18:26.976 I mean, yeah, we're seeing massive investment over into the robotics space. 00:18:27.423 --> 00:18:33.772 The, the AI side of this conversation within DR Horton, I kind of look at it 00:18:33.772 --> 00:18:37.922 and I'm like, okay, I mean, I can see how AI might be useful in some of this for sure. 00:18:37.922 --> 00:18:42.562 Like from like having the robotics, being able to tap in to something, but I don't know. 00:18:42.562 --> 00:18:49.212 I go back to when i was detailing steel and when i was working with fab shops that would have to, 00:18:49.712 --> 00:18:53.419 run robotics in their fabrication shops and the amount of time, 00:18:53.883 --> 00:18:59.682 it took to just set up the way everything flowed through the shop and i

think 00:18:59.682 --> 00:19:03.762 it's really interesting too that i think hugh you brought it up it's like almost 00:19:03.762 --> 00:19:05.121 like this this factory could. 00:19:05.947 --> 00:19:07.521 Shift around depending on what 00:19:07.521 --> 00:19:10.718 it was building right and so like the lines could change and everything, 00:19:11.310 --> 00:19:18.025 and it'd be really hard to optimize a factory like that but i guess you could potentially use ai, 00:19:18.501 --> 00:19:22.461 to identify where the bottlenecks are occurring and then re kind of configure 00:19:22.461 --> 00:19:26.501 how your shop is set up for that too so i don't know it's just kind of an interesting 00:19:26.833 --> 00:19:30.581 interesting angle this like factory is a service thing that could be reconfigured 00:19:30.581 --> 00:19:31.627 depending on what you're building, 00:19:32.162 --> 00:19:36.881 but also like the modular side of things has been talked about for years and 00:19:36.881 --> 00:19:40.911 years and years is this thing that's going to change the game and like i've 00:19:40.911 --> 00:19:45.139 had umpteen thousand conversations about it and i've yet to see it actually, 00:19:45.749 --> 00:19:50.971 change the game because it requires people to slow down and think and people 00:19:50.971 --> 00:19:55.451 don't like to slow down and think they just like to go go go a lot of times 00:19:55.451 --> 00:20:00.250 especially in our industry so and yeah to the point like how many people in the, 00:20:00.819 --> 00:20:05.151 in the accountable chain from you know owner through design through engineer 00:20:05.151 --> 00:20:10.757 and all that have to change to make this thing work right you know that's what i so who knows, 00:20:11.418 --> 00:20:17.804 well well here's the thing yeah okay so at dr horton nothing has to change. 00:20:18.931 --> 00:20:24.531 As a publicly owned company dr horton is primarily operating as a transactionalist 00:20:24.531 --> 00:20:27.971 they don't even they don't even own all the lots they sell. 00:20:28.791 --> 00:20:35.268 They built Four Star in 2017 as like a way to shift the capital heavy land development off. 00:20:35.680 --> 00:20:41.051 And this is just an opportunity and a test to figure out how to shift manufacturing 00:20:41.051 --> 00:20:45.381 or how to shift the production of the construction site, right? 00:20:45.381 --> 00:20:48.971 So really what they're doing is buying an option. 00:20:49.531 --> 00:20:53.169 From D.R. Horton's point of view, they're buying an option on production. 00:20:53.477 --> 00:20:57.231 I mean, as soon as they announced this deal, within two weeks, 00:20:57.231 --> 00:20:59.891 they announced a 45% reduction in starts. 00:21:00.824 --> 00:21:05.691 Hmm. Right. So what they're really and then what that does is that says, 00:21:05.813 --> 00:21:09.018 wait, Promise just made this deal. Did they know that Horton was going to be 00:21:09.018 --> 00:21:12.018 doing a 45 percent reduction in starts? 00:21:12.372 --> 00:21:17.338 If they did, they probably designed that factory in Phoenix specifically to 00:21:17.338 --> 00:21:20.678 be at an 85% utilization rate, right? 00:21:20.678 --> 00:21:23.708 Like they're, they're designing, that's a micro factory in Phoenix too. 00:21:23.708 --> 00:21:28.524 That's nothing like the ones in Montreal or Alberta or potentially Toronto that, 00:21:28.978 --> 00:21:34.388 that promise adjacent folks have been built, have been building. 00:21:34.388 --> 00:21:38.338 So realistically, factory as a service is almost a distraction. 00:21:38.338 --> 00:21:47.049 What this is, it's really an option that D.R. Horton is exercising or purchasing. 00:21:47.304 --> 00:21:52.528 And the question is, what does the contract actually look like? Who is short on the put? 00:21:54.251 --> 00:21:57.716 Fascinating so i think the point is, 00:21:58.471 --> 00:22:04.166 they see a slowdown and they're essentially like hedging their risk exposure 00:22:04.271 --> 00:22:05.844 is is kind of another way of saying that, 00:22:06.465 --> 00:22:10.368 yeah that's exactly what's happening they they either see a slowdown or generally 00:22:10.368 --> 00:22:15.558 they want to be asset light as a as a business and don't want to be correlated 00:22:15.558 --> 00:22:20.488 to the housing market in the the production side of the housing market, right? 00:22:20.488 --> 00:22:24.078 They don't want to be correlated to the asset ownership, which again, 00:22:24.078 --> 00:22:27.558 looking at the bigger picture here is really been the challenge, right? 00:22:27.558 --> 00:22:32.108 Like Katera's problem was that they invested in all these assets and couldn't 00:22:32.108 --> 00:22:37.160 respond to the lumpy demand that is like, that is so common in construction. 00:22:37.862 --> 00:22:41.246 What's Katera? I've never heard that name. I'm just kidding. 00:22:42.251 --> 00:22:44.127 There's a base crater that we can take you to. 00:22:46.466 --> 00:22:47.447 I mean, I do think. 00:22:49.967 --> 00:22:53.369 There is, you know, to the point you made previously, Aaron, 00:22:54.367 --> 00:22:58.777 I think the maturity moment for robotics is when we stop talking about the robot 00:22:58.777 --> 00:23:02.860 and we start talking about the output, right? And I think this is on that path. 00:23:03.440 --> 00:23:08.177 And I, you know, obviously we'll see how it goes. 00:23:08.177 --> 00:23:15.047 But I think if you talk about Dusty, you talk about Rugged, you talk about all 00:23:15.047 --> 00:23:18.807 of the different robot applications, I think the success moment is when we stop 00:23:18.807 --> 00:23:21.367 talking about the robot and we just talk about their productivity, 00:23:22.387 --> 00:23:23.886 the cost profile, et cetera. 00:23:24.345 --> 00:23:28.777 And so I think that's what I'm interested to watch here is not so much the robot 00:23:28.777 --> 00:23:31.822 since more of the outcome. Well, and it's so much the

timing. 00:23:32.322 --> 00:23:35.793 Yeah, go ahead. I was going to say, if I may real quick, the examples you gave. 00:23:37.067 --> 00:23:40.081 Dusty and Rugged and, well, let's start there. 00:23:40.540 --> 00:23:42.984 They prove the limits of robots. 00:23:43.593 --> 00:23:47.977 Because if you think about it, they had to choose something where there's nothing there. 00:23:47.977 --> 00:23:53.227 Like the one time when people sweep the floor and clean it up and remove all 00:23:53.227 --> 00:23:57.697 the complications and so on. Like it's the one, it's the cleanest, 00:23:57.697 --> 00:24:01.449 most simple environment in, on a, on a construction site. Do you see what I mean? 00:24:01.838 --> 00:24:06.187 Like the fact that that's the one that's working illustrates the limits of, 00:24:06.187 --> 00:24:09.390 of robotics, at least as they are now, you know? 00:24:09.687 --> 00:24:12.747 And again, factories, the whole point is you can control the environment or 00:24:12.747 --> 00:24:15.567 even a fab shop in a way that you can't on a, on a job site. 00:24:15.567 --> 00:24:19.886 But anyway, just, it's really interesting. The ones that work illustrate why 00:24:20.043 --> 00:24:21.593 it's a, not a general solution. Yeah. 00:24:25.349 --> 00:24:29.519 Even in the fab shops, proving that a robot is going to support the work that 00:24:29.519 --> 00:24:33.619 you're doing almost always depends on measuring how much of the output that 00:24:33.619 --> 00:24:36.595 you're going to get, how much standardized output you're going to get on that one thing. 00:24:36.862 --> 00:24:40.147 Because if you still have to shift your line and do a lot of these, 00:24:40.420 --> 00:24:42.489 you have to kind of redesign what a robot looks like. 00:24:42.489 --> 00:24:46.289 It's still often much better to just keep people moving around. 00:24:46.289 --> 00:24:49.309 If you finally get to a place where you have enough demand, enough throughput 00:24:49.309 --> 00:24:55.723 on a piece, now it makes sense to put machine tending in or to roboticize and fully automate a line. 00:24:57.076 --> 00:25:02.829 It's interesting i'll throw in there the the humanoid angle of this if we play 00:25:02.829 --> 00:25:06.969 that out like let's say okay humanoids come into the industry and they're starting 00:25:06.969 --> 00:25:10.107 to you know walk around our job sites and do work for us, 00:25:10.719 --> 00:25:15.969 i think the the issues then flow back upstream to the planning side right which 00:25:15.969 --> 00:25:20.152 is okay we have to be able to tell the robots, 00:25:21.109 --> 00:25:26.689 with some level of clarity what they're supposed to do and so this idea of like 00:25:26.689 --> 00:25:31.149 oh we'll just beat it to fit paint it to match out in the field like robots 00:25:31.149 --> 00:25:36.018 are they they may do it but they're not going to have the intuition of a human to say, 00:25:36.690 --> 00:25:39.807 okay here's how we're going to beat it to

fit and paint it to match, 00:25:40.499 --> 00:25:44.589 as readily available i don't know maybe it'll change eventually but i see this 00:25:44.589 --> 00:25:45.559 kind of tagline for ugc tyler what's that. 00:25:47.789 --> 00:25:52.433 Beat it to fit, paint it to match. Oh, that's literally every subcontractor's. 00:25:53.869 --> 00:25:57.769 I mean, that's been the whole line for years in steel is like, 00:25:58.129 --> 00:26:01.385 we don't know, just send it out and we'll beat it to fit, paint it to match. 00:26:01.890 --> 00:26:04.773 Like that's kind of where we're at in some ways. 00:26:06.774 --> 00:26:09.924 And it's that, and it's that ability, usually the reason you're having to beat 00:26:09.924 --> 00:26:13.504 it to fit is because you had to build something before you actually really understood 00:26:13.504 --> 00:26:15.634 the site conditions or, right. 00:26:16.063 --> 00:26:21.834 And, and, and unless you start to, when, when we start to get to real-time data 00:26:21.834 --> 00:26:26.144 information and we're, and we're accurately comparing expected versus ad built 00:26:26.144 --> 00:26:28.404 as built on a real-time basis. 00:26:28.404 --> 00:26:32.954 And we're getting we have an information system that's capable of knowing exactly 00:26:32.954 --> 00:26:36.594 where the people are and where the robots are and what work needs to be done 00:26:36.594 --> 00:26:40.774 and and adjusting decisions and being able to make changes at 00:26:41.154 --> 00:26:47.244 a rate of 10, 10 changes an hour, because that's the kind of things that that 00:26:47.244 --> 00:26:49.104 that humans can do on a job. 00:26:49.995 --> 00:26:55.054 The robots and the humans are going to always be out of sync and reality is 00:26:55.054 --> 00:26:56.556 going to be out of sync with the robots. 00:26:57.411 --> 00:26:59.894 I mean, what we just described, though, is back to your point, 00:26:59.894 --> 00:27:04.348 Hugh, robots do really well when there's a lack of variability and when there's a clear plan. 00:27:04.753 --> 00:27:11.234 So it seems like residential is kind of the perfect environment for an increase 00:27:11.234 --> 00:27:17.204 in robotic kind of engagement, but maybe not the kind of R&D incentive. 00:27:17.383 --> 00:27:22.994 Well, I think it's also worth asking, if we assume, which we kind of just did, 00:27:22.994 --> 00:27:26.324 that they're not replacements for what a human can do, and I think Tyler put 00:27:26.324 --> 00:27:29.762 it perfectly about intuition and just ability to problem solve, 00:27:30.379 --> 00:27:36.074 does that not mean that the role that they play is not what a human would, 00:27:36.074 --> 00:27:37.374 but a slice of what a human would? 00:27:37.374 --> 00:27:40.234 And maybe things that you wouldn't normally pay a human for, 00:27:40.234 --> 00:27:43.754 like, I need you every night, go through the site, clean things up, 00:27:44.520 --> 00:27:46.064 check this, check that, whatever. 00:27:47.241 --> 00:27:50.587 You know, things, if you think about what we use AI for, even sometimes it's 00:27:50.587 --> 00:27:51.557 stuff a human would do with. 00:27:51.557 --> 00:27:54.486 Sometimes you're doing things that if I had, if I had, you know, 00:27:54.776 --> 00:27:57.577 budget, I'd have people do it, but I can have AI do it for me. 00:27:57.577 --> 00:28:01.377 So I feel like there's going to wind up being an opportunity to say, 00:28:01.719 --> 00:28:04.837 we don't have to look at what a human would do. We can say, what is something 00:28:04.837 --> 00:28:07.457 that I would have a human do if it were cheap? 00:28:07.937 --> 00:28:10.728 And now that it is, I can, I can have the robot do it. 00:28:11.059 --> 00:28:13.973 That may not be everything, but I think that's something even back to, 00:28:14.467 --> 00:28:20.617 if you think about what rugged and Dusty do, what they lay down on the slab 00:28:20.617 --> 00:28:23.150 is way more complicated than what it replaces. 00:28:23.638 --> 00:28:28.827 So it allows for things that a human wouldn't have done that helps the job in other ways. 00:28:29.164 --> 00:28:31.910 So I have a feeling that over time, you'll see some of that too. 00:28:32.217 --> 00:28:36.707 You know, you'll see that let's keep the robot away from live work and live 00:28:36.707 --> 00:28:40.385 building, but there's still an awful lot that goes into a job site that it can support. 00:28:40.973 --> 00:28:45.857 So what you're saying is that we might reduce counseling bills for people going 00:28:45.857 --> 00:28:50.637 forward for like you know my dad could have used a robot one day to get it to 00:28:50.637 --> 00:28:52.455 hold the flashlight in the right place, 00:28:52.943 --> 00:28:56.497 like we could have just used a robot for that and so then i'm not going to counseling 00:28:56.497 --> 00:29:00.917 as a result so i love this i think this is great that's so funny we're saving 00:29:00.917 --> 00:29:03.317 society over here yeah no not. 00:29:04.037 --> 00:29:08.677 There still one of these episodes it's gonna be like an episode of mari i'm 00:29:08.677 --> 00:29:12.717 gonna bring in papa campbell just like midway through the episode and just be 00:29:12.717 --> 00:29:15.117 like, Hey, Papa Campbell. 00:29:15.577 --> 00:29:20.695 I don't even know his actual name, but not Papa Campbell. That's for sure. 00:29:20.927 --> 00:29:25.497 Oh gosh. All I hear are great stories that need to be brought to the fore of your youth. 00:29:26.457 --> 00:29:28.627 How come no one's seen Tyler for a couple of weeks? I don't know. 00:29:28.977 --> 00:29:31.137 I have no idea. I was ever against that podcast. 00:29:33.522 --> 00:29:40.912 All right. So from robots to maybe AI, let's talk about we've got adoption of 00:29:40.912 --> 00:29:45.002 AI in different levels throughout the industry, but at an individual level, 00:29:45.002 --> 00:29:47.022 it's happening. We've seen that, right? I think so. 00:29:47.782 --> 00:29:52.222 E&R

back in October kind of went, I think through a survey and a couple other 00:29:52.222 --> 00:29:56.932 means, went out to insurers and brokers and kind of asked how the adoption of 00:29:56.932 --> 00:30:00.634 AI was potentially influencing different exclusions that were showing up in 00:30:00.745 --> 00:30:02.178 construction liability policies. 00:30:02.746 --> 00:30:09.582 And the answer was not yet. I guess there were some professional liability policies that were being, 00:30:10.182 --> 00:30:12.472 and the professional liability policies were not showing up, 00:30:12.472 --> 00:30:18.302 but the general liability, at least according to Gallagher, some exclusions started to appear. 00:30:18.302 --> 00:30:25.616 So I think the question is, Hugh, if your team uses AI to save a bunch of time, 00:30:26.245 --> 00:30:30.563 and that leads to maybe a multi-million dollar mistake... 00:30:31.279 --> 00:30:35.472 How do we adjudicate that mistake in today's reality? 00:30:35.756 --> 00:30:40.505 Yeah, I don't think that's even a question. And I'm glad you bring it up because 00:30:40.505 --> 00:30:44.436 it is a question people ask. But we have to keep in mind, these are just tools. 00:30:44.881 --> 00:30:48.435 As smart as they sound, these are just tools. So it's still the human's job. 00:30:48.435 --> 00:30:52.869 It's still the PM's job. It's still the company's job. I mean, a funny analogy. 00:30:53.171 --> 00:30:59.635 There was two court cases recently about whether if I'm talking to ChatGPT or 00:30:59.635 --> 00:31:04.295 whatever, in prep for a meeting with my lawyer, is it covered, 00:31:04.295 --> 00:31:06.490 does that covered by attorney-client privilege? 00:31:07.053 --> 00:31:10.315 Which if you think about it is a really weird question to ask. 00:31:10.315 --> 00:31:14.385 It's like, well, if I was writing notes in, I don't know, in Microsoft Word, 00:31:14.705 --> 00:31:15.836 yeah, it wouldn't be covered. 00:31:16.405 --> 00:31:21.395 You're just using a digital tool, right? But the fact that these AI can kind 00:31:21.395 --> 00:31:24.195 of sound like a human sometimes, it throws people off. 00:31:24.195 --> 00:31:27.556 And the crazy thing is one of the, these two cases, one of them, 00:31:27.910 --> 00:31:31.817 the judge found that, yes, It is covered because it's like you were talking to a human. 00:31:32.159 --> 00:31:35.675 That's insane. And I can't imagine that's going to survive. The other one, 00:31:35.675 --> 00:31:38.635 the person said what I just said, which is, in fact, I'm taking what they said. 00:31:38.635 --> 00:31:41.476 This is like using Microsoft Word. So back to this. 00:31:41.859 --> 00:31:45.425 At the end of the day, just because AI was involved doesn't absolve anybody 00:31:45.425 --> 00:31:49.275 of any responsibility. And if anything, you know, hopefully there won't be big 00:31:49.275 --> 00:31:53.562 court cases. At some point, something will happen because it's construction and stuff happens. 00:31:54.195 --> 00:31:57.495 Hopefully it's not too big a deal and no one gets hurt. But you're going to 00:31:57.495 --> 00:32:02.275 wind up having examples of people thinking that, you know, they let it, 00:32:02.275 --> 00:32:03.565 they didn't check the AI. 00:32:03.565 --> 00:32:08.260 And I think it's so easy to forget that it is still imperfect. 00:32:08.260 --> 00:32:12.445 The whole reason it works is because it can handle kind of probabilities. 00:32:12.445 --> 00:32:14.285 Well, you pay for that on the way out. 00:32:14.616 --> 00:32:18.807 And I think as people are learning to be vigilant, I see a lot of people that are not. 00:32:19.022 --> 00:32:23.635 I see a lot of people that are enamored with how great what comes out of it 00:32:23.635 --> 00:32:25.065 can be and they don't check it enough. 00:32:25.506 --> 00:32:30.158 So yeah, to me, there's not even a question there. And I think hopefully, 00:32:30.709 --> 00:32:35.028 certainly owners of contractors are not under that impression. 00:32:35.764 --> 00:32:35.868 But, 00:32:37.908 --> 00:32:40.718 this is really a question about the insurance companies, though, 00:32:41.873 --> 00:32:49.618 and either how they can create ambiguity in this conversation or reveal preferences 00:32:49.618 --> 00:32:51.258 and actually understand what's really going on. 00:32:51.258 --> 00:32:58.104 Because an insurance company's biggest problem isn't a couple of one-off cases, even severity. 00:32:58.562 --> 00:33:02.928 Their problem is correlation. Are we going to start seeing a bunch of common 00:33:03.248 --> 00:33:09.789 mode failures happening that we could go in and potentially deny. 00:33:11.594 --> 00:33:15.198 Because these teams are relying too much on AI? 00:33:15.198 --> 00:33:20.188 So, yes, there is a conversation here to be had about how we treat it and how 00:33:20.188 --> 00:33:21.207 we look at it and how we... 00:33:22.223 --> 00:33:28.700 How we babysit it basically the but the bigger question is is how are insurance companies going to, 00:33:29.680 --> 00:33:32.809 find ways to get out of some of these dno claims, 00:33:33.634 --> 00:33:37.240 yeah i mean are you saying insurance companies do that aaron i well they're 00:33:37.240 --> 00:33:41.670 more pressed to think of yeah i mean there's a great point there's kind of three 00:33:41.670 --> 00:33:45.460 things right which is like if you are an ai first company and you're automating 00:33:45.460 --> 00:33:46.753 the crap out of everything, 00:33:47.217 --> 00:33:52.680 insurance companies probably need to completely change how they approach the work that you're doing, 00:33:53.515 --> 00:33:58.291 and they don't have an actuarial approach to that yet the other thing is this like, 00:33:59.220 --> 00:34:02.860 ai ai can't really be defined today without going, 00:34:03.540 --> 00:34:07.050 with a fine tooth comb through everything that's happening right because like 00:34:07.050 --> 00:34:11.320 you said like when you when

we were introducing ourselves you we've been building 00:34:11.720 --> 00:34:16.640 machine learning tools and ai tools and you know Essentially, it's stochastic, 00:34:17.540 --> 00:34:19.618 prediction about something, right? 00:34:20.920 --> 00:34:25.855 That exists. Honestly, the software that contains the most of that in construction today is CAD. 00:34:26.070 --> 00:34:30.580 And just in how it finishes lines and how it connects, you know, 00:34:30.580 --> 00:34:32.816 there's machine learning all over that tool. 00:34:33.188 --> 00:34:38.568 So what it's really doing is it's creating where it's shifting the cost of litigation 00:34:38.835 --> 00:34:40.880 back to the contractor and, 00:34:41.413 --> 00:34:48.222 the professional to prove and to actually litigate what AI means. Hmm. 00:34:49.952 --> 00:34:55.826 I mean, Tyler, you are knee-deep in helping contractors, one, 00:34:55.826 --> 00:35:01.410 make sense of their data, building AI kind of centric workflows. 00:35:01.874 --> 00:35:06.256 How much of this, two questions, how much does this go into how you orient your 00:35:06.256 --> 00:35:09.926 contract or contract with those contractors? And also, how are you setting them 00:35:09.926 --> 00:35:14.539 up to maybe protect themselves? Or has this kind of come up in any of those conversations yet? 00:35:15.062 --> 00:35:17.476 I don't know, man. And I've been working through this a lot because like, 00:35:18.196 --> 00:35:20.579 I think, yeah, the accountability component. 00:35:21.556 --> 00:35:26.736 I've been wondering if we're going to find ourselves contracting people out 00:35:26.736 --> 00:35:31.323 more and more instead of just having like these massive teams of W-2 employees. 00:35:31.857 --> 00:35:34.127 Because like, think about it. If I come work for your company, 00:35:34.464 --> 00:35:39.905 I'm transparent and I'm saying, hey, listen, I use AI tools to get the job done. Then I screw up. 00:35:40.389 --> 00:35:44.406 Right. I mean, it's pretty easy to cast blame on me at that point. 00:35:44.592 --> 00:35:48.836 And so like, hey, it was Tyler, it was his team, and he was the problem here. 00:35:48.836 --> 00:35:51.208 He didn't check his stuff before he implemented it. 00:35:51.748 --> 00:35:55.546 So I almost wonder if there's that lens of like people are going to start doing 00:35:55.546 --> 00:35:57.025 more contract type labor. 00:35:57.906 --> 00:36:01.814 We all have like professional liability in our sub. Yeah, I don't know. 00:36:02.534 --> 00:36:08.844 But then also with a W-2 worker, I think there are two angles of this. 00:36:08.844 --> 00:36:13.384 There's like the AI that you're using day to day that you're chatting with, 00:36:13.384 --> 00:36:16.064 right? Like you're talking to and collaborating with. 00:36:16.517 --> 00:36:22.055 But then there's also the agentic AI, which is going out and doing things proactively. 00:36:22.328 --> 00:36:29.903 And that's, I feel, the biggest fear within this story here

is the agentic side, 00:36:30.275 --> 00:36:37.074 the proactive AI that nobody is really saying yes to or no to. 00:36:37.074 --> 00:36:41.274 So I don't know, man. It's just kind of an odd spot. 00:36:42.514 --> 00:36:46.764 I was just interested to hear what Hugh and Aaron had to say about this topic, 00:36:46.764 --> 00:36:50.494 too, because I think everybody's kind of going, it's kind of up in the air right now. 00:36:51.874 --> 00:36:54.849 I think that if I can tee off of some of what you just said, 00:36:55.121 --> 00:37:00.234 the idea that you have an agent that completes something, I don't think anybody's 00:37:00.234 --> 00:37:02.684 ready for it. And I don't, I know the technology is not ready. 00:37:02.684 --> 00:37:04.094 I don't care what vendors say. 00:37:04.454 --> 00:37:08.554 It's just not what, what you, the better way of thinking about it is our, 00:37:08.954 --> 00:37:11.194 an agent will automate up to 80%. 00:37:11.194 --> 00:37:13.934 Like it gathers things, it drafts things, it pulls things. I mean, 00:37:13.934 --> 00:37:16.981 there's stuff like, you know, let's have an agent that pulls all the, 00:37:17.498 --> 00:37:21.576 you know, details from, from invoices. I mean, you know, stuff like that. Sure. 00:37:22.029 --> 00:37:26.264 But, but, you know, you come down to the question of, can I create a checkable standard? 00:37:26.264 --> 00:37:30.411 Can I verify it? And so little is actually verifiable. So as a result, 00:37:30.754 --> 00:37:33.743 before you even get into the conversation of who's liable for what, 00:37:34.074 --> 00:37:36.134 it's just not something you can reliably automate. 00:37:36.134 --> 00:37:42.094 So I think this, this change and talking about, you know, agents as if they 00:37:42.094 --> 00:37:45.484 are going to complete things, A, I think overstates what they can do, 00:37:45.484 --> 00:37:49.886 but also is a great way to freak people out, as opposed to saying the stuff you hate doing. 00:37:50.537 --> 00:37:54.512 That allows you to do your thinking. Like that's where AI helps is it brings 00:37:54.512 --> 00:37:57.392 it together. Maybe it drafts the thing, maybe it gives you a version or, 00:37:57.392 --> 00:38:01.322 or, and you know, a take or a critique or whatever, but ultimately it has to 00:38:01.322 --> 00:38:03.572 leave a human's hands for, for most. 00:38:03.892 --> 00:38:07.179 And I think we also underestimate how much judgment people bring to things. 00:38:07.806 --> 00:38:10.942 I think you're right here. 00:38:10.942 --> 00:38:13.762 I think the judgment thing is there, but you know, I've, I've, 00:38:13.896 --> 00:38:19.682 there's these things that databases can do in terms of mutual exclusivity and 00:38:19.682 --> 00:38:23.492 collective exhaustiveness that humans are incapable of doing. 00:38:23.492 --> 00:38:28.722 We're great at judgment, but when it comes to massive amounts of information 00:38:28.722 --> 00:38:33.667 and sorting through huge complex scenarios, we aren't that good at it. 00:38:33.772 --> 00:38:38.162 And it's possible that if we actually get more discreet with some of these things, 00:38:38.162 --> 00:38:42.857 and we're not just talking about, you know, judgment scenarios, 00:38:43.147 --> 00:38:46.456 that the machines are going to be way better than us at it. 00:38:46.891 --> 00:38:49.822 It's just most of the inputs aren't that good. That's, that's been to your, 00:38:49.822 --> 00:38:51.172 maybe that's what you meant by disagree. 00:38:51.172 --> 00:38:54.602 That's exactly, that's exactly my point is that most of the inputs aren't that 00:38:54.602 --> 00:38:59.432 good, but if the inputs were actually digital and immutable and, and, 00:38:59.792 --> 00:39:04.372 you know, we had software that wasn't destructive and had traces and logs and 00:39:04.372 --> 00:39:07.104 all that kind of stuff, the inputs would be a lot better. Yeah. 00:39:08.219 --> 00:39:13.293 So I think one of the challenges with the kind of human in the loop model, 00:39:13.293 --> 00:39:17.393 other than that it's maybe dismissive of our humanity, is, 00:39:18.193 --> 00:39:22.573 when you have outputs that are plausible, which these AIs produce, 00:39:22.573 --> 00:39:25.918 agents produce, I think, to maybe your point, Aaron, 00:39:26.551 --> 00:39:30.633 often the human is kind of not really doing a proper review. 00:39:30.633 --> 00:39:34.039 Or there hasn't been definition of what proper review means. 00:39:34.527 --> 00:39:40.833 And so I think maybe, Aaron, put yourself in the seat of you're managing a large GC. 00:39:41.917 --> 00:39:46.243 As you see kind of AI adoption happening at individual level, 00:39:46.243 --> 00:39:52.653 at project level, how are you maybe carrying this conversation with your insurance provider? 00:39:52.653 --> 00:39:57.532 Are you being more proactive internally or trying to put the hedges in place externally? 00:39:59.523 --> 00:40:05.543 I mean, I definitely think as a large GC, you have the benefit of working with 00:40:05.543 --> 00:40:07.573 your insurance provider. 00:40:08.013 --> 00:40:12.603 And that's a big part of this, right? Because the whole thing that the insurance 00:40:12.603 --> 00:40:15.203 provider wants to know is what's actually going on. 00:40:15.203 --> 00:40:18.553 And most of the time they're putting these governance docs out there and they're 00:40:18.553 --> 00:40:23.663 doing this stuff in the renewal and the disclosure forms because they haven't 00:40:23.663 --> 00:40:26.993 been able to build a statistical model. They haven't been able to see what's 00:40:26.993 --> 00:40:29.605 happening to understand where the risks are, right? 00:40:29.964 --> 00:40:36.033 And so if you can cooperate with the insurance company now, everybody's afraid of what happens. 00:40:36.033 --> 00:40:41.123 You know, we start putting the little trackers in cars and pretty soon our rate 00:40:41.123 --> 00:40:44.303 doesn't come down, but everybody else's

goes up and, you know, 00:40:44.303 --> 00:40:49.274 just the whole it's there's definitely a lot of competition things to consider there. 00:40:50.506 --> 00:40:56.476 But yeah, if you can, if you can reveal the right information at the right time 00:40:56.476 --> 00:41:01.562 and work with them, both, both parties are going to make it work. 00:41:02.019 --> 00:41:05.096 Yeah. Cause they want, they want to see your business grow. You know, 00:41:05.096 --> 00:41:09.576 they want to see the business grow. They want to see efficiency and productivity 00:41:09.576 --> 00:41:14.673 come to the business because it ultimately means more work for them too. Yeah. 00:41:15.729 --> 00:41:18.466 We're running out of time, but I do want to get to this next one. 00:41:18.466 --> 00:41:24.756 I think, so input prices rose 11.2% year over year. 00:41:25.354 --> 00:41:29.076 And I think that's no surprise. I think, you know, you go to a grocery store, 00:41:29.076 --> 00:41:31.896 you talk to your wife, like everything's getting more expensive. 00:41:31.896 --> 00:41:35.727 And there's, you know, there's the Iran war, there's some tariff tailwinds. 00:41:36.193 --> 00:41:40.286 I think that's not the surprising part. I think the interesting part of this 00:41:40.286 --> 00:41:45.620 kind of ABC analysis reference that one in four contractors, 00:41:45.968 --> 00:41:47.986 as a result, expect lower margins. 00:41:47.986 --> 00:41:51.036 And I think there is a narrative out there that when prices go up, 00:41:51.292 --> 00:41:57.131 those prices are, that increasing cost is passed on to the customer, in this case, the owner. 00:41:57.709 --> 00:42:01.076 But at least 25% of the contractor community doesn't think that. 00:42:01.453 --> 00:42:04.286 I guess maybe to start a variety of this conversation... 00:42:05.432 --> 00:42:08.329 Why do we think that isn't always just passed on to the owner? 00:42:08.789 --> 00:42:14.599 Is there a cap on some of these verticals in terms of what we can charge from 00:42:14.599 --> 00:42:15.869 like a square footage standpoint? 00:42:15.869 --> 00:42:23.419 Or are these contractors who are seeing a squeeze in their margin not incorporating 00:42:23.419 --> 00:42:28.179 the appropriate business model response to increased costs? Do we have any hypothesis 00:42:28.179 --> 00:42:29.946 there? Aaron, you want to start us off? 00:42:30.949 --> 00:42:35.189 Well yeah i mean we spent 20 years before you know up up until what basically 00:42:35.189 --> 00:42:40.506 up until covid we spent 20 years in basically a low low inflation environment 00:42:40.936 --> 00:42:45.369 yeah maybe 25 years right where business models were like were, 00:42:46.069 --> 00:42:50.309 not thinking about escalation unless they were just in the annual union, 00:42:50.909 --> 00:42:55.609 labor escalation scenario right and so i think there was some tools that were 00:42:55.609 --> 00:43:02.016 lacking and And then obviously, in COVID, it was like everything was this force majeure conversation. 00:43:02.623 --> 00:43:07.129 And then finally, now we're having to actually put these kinds of cost escalation 00:43:07.569 --> 00:43:13.620 clauses or some kind of mechanism into the business to reclaim that. 00:43:14.788 --> 00:43:18.693 I think the real number here is that those inputs are up 11.9%, 00:43:19.099 --> 00:43:24.018 but over the same period, bids are only up about 5.4%. 00:43:24.018 --> 00:43:32.108 And so there's really about a 6.5% wedge or leakage in this number. 00:43:33.688 --> 00:43:36.028 And I don't know if it has to do with where it's coming from. 00:43:36.028 --> 00:43:38.848 It may be that a lot of it's coming from diesel fuel. 00:43:40.328 --> 00:43:45.828 Some long lead times, unexpected escalations that are beyond what was already 00:43:45.828 --> 00:43:49.578 in a clause, and it's gotten into a point of sharing. 00:43:49.875 --> 00:43:54.318 So a lot of this really has to do with, like, are you structuring your escalation 00:43:54.318 --> 00:43:59.288 clauses in a way that is actually consistent with what's going to happen in the market? 00:43:59.288 --> 00:44:04.978 And I think owners are probably just more financially savvy about this stuff 00:44:04.978 --> 00:44:06.686 and about the range of risks. 00:44:07.310 --> 00:44:11.140 The other thing that happens here is the question about whether or not trades 00:44:11.394 --> 00:44:16.642 and GCs are both aligned in the way that they manage the escalation causes. 00:44:17.810 --> 00:44:22.768 That was actually going to be my question is, I think, drilling into what's 00:44:22.768 --> 00:44:23.876 under a number like that. 00:44:24.027 --> 00:44:28.628 Like the problem always with numbers in construction is it covers an awful lot of different things. 00:44:28.628 --> 00:44:32.498 And I wonder if the leakage is happening on the trade side more than it is on 00:44:32.498 --> 00:44:36.386 the GC side, and if there's specific trades more than others where, 00:44:36.980 --> 00:44:40.628 you know, there's either something unexpected or they tend to be small companies 00:44:40.628 --> 00:44:42.028 because of the type of trade it is. 00:44:42.468 --> 00:44:47.118 I just, I wonder a little bit about what's, I mean, if it's a number that big, 00:44:47.118 --> 00:44:49.518 it's reflecting a lot of companies. So it can't just be like, 00:44:49.518 --> 00:44:51.052 you know, only the drywall guys. 00:44:51.441 --> 00:44:56.015 But nevertheless, it'd be interesting to understand what dynamic is going on under it. 00:44:56.335 --> 00:44:59.748 And where that squeeze is coming from, right? Like the squeeze could be coming 00:44:59.748 --> 00:45:03.948 from the fact that a bunch of residential firms are now competing much more, 00:45:04.448 --> 00:45:09.705 much more aggressively for, for work or, you know, all the trouble in the, 00:45:10.348 --> 00:45:13.415 in the other commercial and multifamily markets, right? 00:45:14.804 --> 00:45:18.389

That's a good point. Yeah, that was going to be my flag here is, 00:45:19.075 --> 00:45:24.604 you know, if we're looking at our margins getting pressed, it has to be, 00:45:24.964 --> 00:45:27.929 yeah, I mean, the competition has to be driving some of this. 00:45:28.376 --> 00:45:30.310 It's just a very competitive marketplace. 00:45:30.855 --> 00:45:34.344 And then they're just trying to get the next job, right? 00:45:34.344 --> 00:45:38.814 So I'm going to lower my prices in order to get the next thing, 00:45:38.814 --> 00:45:42.721 because if I don't have something, then I'm going to go out of business, 00:45:43.145 --> 00:45:48.404 right? So it's better to have something that I can work with than nothing at all. So, 00:45:50.776 --> 00:45:55.864 especially if the pressure is labor, right? I mean, keeping your bench alive 00:45:55.864 --> 00:46:00.041 is one of the hardest things that a trades contractor does. 00:46:00.395 --> 00:46:04.374 Well, okay. On that too. So if we look at the hyperscalers and kind of back 00:46:04.374 --> 00:46:07.174 to what we were talking about earlier, if these younger kids are able to go 00:46:07.174 --> 00:46:11.673 out and get a job there and they're getting primo money for that, 00:46:12.298 --> 00:46:15.844 like the expectation goes up so yeah as a result if you're not changing your 00:46:15.844 --> 00:46:20.887 prices then you're getting like just that compression that's gonna naturally occur there, 00:46:21.524 --> 00:46:25.374 a corollary might be it'd be interesting after the fact to see if change orders 00:46:25.374 --> 00:46:29.824 went up like we lost it in the bid and we made it up later because that never 00:46:29.824 --> 00:46:31.515 happens in construction so who knows. 00:46:32.504 --> 00:46:36.564 I mean the study i'd like to see then maybe i'll submit this to you aaron because 00:46:36.564 --> 00:46:41.295 you have like a you have a monster computer that's just churning through economic data, I believe. 00:46:41.446 --> 00:46:49.772 But the data I'd like to see is what's happening to, let's just focus on builders of data centers. 00:46:50.137 --> 00:46:53.234 Is there margins going up over the last 10 years? Because it should. 00:46:53.234 --> 00:46:55.895 If it's not, that's probably a problem, right? And, 00:46:56.867 --> 00:47:00.478 I mean, simple economics as demand rises. Yeah. 00:47:00.838 --> 00:47:05.398 Well, there's some confounding factors that are going on there because a lot 00:47:05.398 --> 00:47:08.629 of those builders also have brought on a lot more self-performed labor. 00:47:09.203 --> 00:47:14.323 So absolutely, most of those top tier GCs, their margins are increasing. 00:47:14.665 --> 00:47:19.378 But what's interesting is their margins on self-perform are actually lower than 00:47:19.378 --> 00:47:23.304 the trades contractors would have had in that context. 00:47:23.722 --> 00:47:26.848 And that makes sense because they're right at the learning curve. 00:47:26.848 --> 00:47:28.428 They're coming into a new business, 00:47:28.428 --> 00:47:30.938 that kind of thing. They don't know how to price for it the same. 00:47:31.758 --> 00:47:35.098 I mean, I can imagine some of that, right? And they don't know how to plan for it either. 00:47:37.236 --> 00:47:43.599 When you don't have that sort of natural tension between the trade and the GC, the GC wins. 00:47:43.750 --> 00:47:47.948 And so the GCs for a lot of these self-performing, now not in every case, 00:47:47.948 --> 00:47:52.058 but in a lot of cases, the GC actually lowers the margin of the trade because 00:47:52.058 --> 00:47:56.187 they're forcing them into bad schedules and they're forcing them to cover for 00:47:56.496 --> 00:47:58.718 like, right. I mean, they just basically become a relief. 00:48:01.118 --> 00:48:04.101 Yeah. I feel like that's a classic one. 00:48:04.269 --> 00:48:10.388 It's like always the low bid, but why, why is it? Yeah. 00:48:11.318 --> 00:48:15.408 Well, we're out of time guys. I've enjoyed this. We missed two stories, but I'm impressed. 00:48:15.408 --> 00:48:20.878 We got through four hugh where can people find you linkedin tiktok yeah link, 00:48:21.418 --> 00:48:26.551 not tiktok linkedin is great my name you see okay aaron, 00:48:27.450 --> 00:48:31.419 Yep. You can find me, you can find me actually find me on LinkedIn. 00:48:31.419 --> 00:48:36.111 And then also you can see what we're doing at Lou L U H dot tech, 00:48:36.587 --> 00:48:41.523 or you can also find me at the Asher hub.com. 00:48:42.331 --> 00:48:46.119 The Asher hub. That's a cool name too, man. You guys got a corner market for 00:48:46.119 --> 00:48:48.209 cool names. Tyler. What was you? 00:48:48.759 --> 00:48:53.039 I have one called a AI edge dot construction. That's that's so since we're giving 00:48:53.039 --> 00:48:54.619 out websites, that's a good name. 00:48:54.822 --> 00:48:57.279 No, you can catch me on Broadway. 00:49:00.419 --> 00:49:05.619 Short run yeah very short run this isn't i don't i don't think it's not meant 00:49:05.619 --> 00:49:09.119 to be mean but i could actually see you doing broadway i could see you doing 00:49:09.479 --> 00:49:13.879 some quantity of drama what are you saying i'm dramatic what are you saying that, 00:49:15.299 --> 00:49:16.278 way, 00:49:17.892 --> 00:49:21.779 you can find me on linkedin broadway you can find me on linkedin you can find 00:49:21.779 --> 00:49:27.979 me at fieldproof.com all one word field proof is a great website i highly recommend 00:49:27.979 --> 00:49:32.699 field proof it's a new it's a new website too so go check it out it's got animations all right guys, 00:49:33.318 --> 00:49:36.745 no more tyler that's enough self-promotion and another thing check, 00:49:38.419 --> 00:49:40.738 it i just end the recording all right thanks guys.