Why construction productivity keeps falling — and what it would take to fix it

though uh even like he just really cares about fundamentally creating a a great product. So well we have another offer another author joining us in the TBPN Ultra Dome. Brian Potter is in the reream waiting room. He's the author of Origins of Efficiency. Thank you so much for joining us. We love Stripe Press.

We love Stripe. We had Privy on. We've had Dark Cash on. We've always enjoyed Strike Press's books and we're it's a pleasure to meet you. How are you doing? I'm good. Thank you guys for having me on. Uh thanks so much for joining.

Uh would you mind uh kicking us off with an introduction on yourself and the book and we can go into a bunch of questions about it, but I'd love to just kind of uh get a little bit of background for everyone on your journey to writing this book. Yeah. So my uh I uh am an senior infrastructure fellow.

I work for the Institute for Progress, which is like a progress think tank. Uh I'm best known to the extent that I'm known uh for writing this newsletter called Construction Physics, which is about buildings and infrastructure and and how to get stuff built uh in the US.

Um and my background, I my you know, before I did this, I worked in the construction industry. I worked as a structural engineer for about uh 15 years like designing buildings and parking garages and water treatment plants and stuff like that. Yeah.

Uh and the industry always seemed like extremely inefficient to me like you know everything is you know so labor intensive. It takes so long you know we're doing this similar work over and over and over again. Uh it should be much more efficient should all be done in factories blah blah blah.

Uh and then in 2018 I had the chance to join this like big exciting construction startup called Catera that had raised this was back when SoftBank was uh I remember give it up for it was kind of a precursor I think Hrien was drawing from it now like it was it was uh the like machine shop almost like uh yeah it was like you know it was this idea is like you know construction is inefficient because it's not done in factories right so we're going to into factorybased construction construction.

Uh it was run by all these former uh electronics manufacturing guys. So like not like software guys that think that you know that like oh I worked at Amazon so I know how to do anything right.

It was like people who knew about manufacturing um and they were going to sort of bring that knowledge to like the construction industry, right? So they raised a huge amount of money money got a huge check from Soft Bank uh raised like two3 billion in venture capital uh and it all went sideways, right?

It all uh it all went wrong and they they burned through it in in about three years and declared bankruptcy and there were various yeah reasons for that. At what point did you at what point did you leave? Uh I was there about until about a year uh before they went bankrupt.

So I was So you saw the Did you see the writing on the wall? Yeah. I when I was there I was saying it was like when my but my time there was like a year and a half of ups and then a year of downs.

And then after about the sixth round of layoffs I uh I uh with you know after our engineering team got cut by like about 90%. I was like need to time to bounce. Saw yourself out. Yeah. Yeah, but I wanted to understand, you know, why things had gone so wrong. Because part of it is just, you know, startups are hard.

Uh, building startups that are, you know, moving physical things around is is very hard. There's very oper various operational missteps or whatever.

Um, but also I kind of came to believe that sort of the thesis that they had built the company around was kind of either wrong or just like not complete enough, like missing very large chunks of it cuz people had tried to do similar things that Catera had done many many times, right?

If you go back over history, uh there's like a huge graveyard of companies of like, oh, you know, light bulb will just we'll build buildings and factories and it'll be so much cheaper and I'll be able to make a huge amount of money. I'll be the Henry Ford of Housing and just it just has never worked, right?

There's like this people have tried this over and over and over again and not been able to succeed. Will it work? Because I have an opportunity to invest. I'm a lucky I have a lucky opportunity. No, no. I I I I actually did meet a company that that's uh taking taking another crack at this.

Uh do you think it'll ever work? I I do think it will work but again I I you need to I wanted to understand why specifically it had been so hard in the past and why Catera and so many other companies had failed and what specifically you know would need to be true for it to succeed in the future.

So it was it bec basically the the where I ended up was like I need to understand what specifically makes it possible for an industry to get like more efficient over time and what specifically is happening why when that is is occurring and what is what sort of can prevent those things from happening and once I understand those mechanics I will be you know I will know what specifically would need to be true for some uh for some uh you know the construction industry or any industry to sort of improve improve over time.

And so that was sort of the genesis of the book is like what specifically does it take for some process to get more efficient over time. It feels like the entire book is kind of an abstraction on top of just this idea of like learning the learning curve. We've seen this in semiconductors.

Everyone who follows like the AI boom is uh acutely aware of the learning curve that happened at TSMC. Uh but you kind of draw a couple other historical analogies.

what stuck out to you as like particularly great examples of this efficiency in in in like going successfully and us actually driving down the cost and then what were the commonalities between that and and like what do they all have in common basically? Yeah.

So I kind of went through and I looked at like you know dozens and dozens and dozens of different industries and and seeing you know how they had improved their operations over time and what specifically was was changing uh in them when that was happening and sort of you know and I looked at like industrial improvement systems right so like lean manufacturing uh and like value engineering and all the in statistical process control and all these other things that like had you know specific ways you could try to make something more efficient and I kind of ultimately boiled all that down to like this c is, you know, a list of like a handful of things that you had to do to try to make some process more efficient.

And if you could do any one of those things, you could kind of make it more efficient. And if you couldn't do those things, those paths were blocked. As it turns out, they are in the construction industry, you're you can't make your process more efficient, and it just gets more and more and more expensive over time.

And so, yeah, I looked at like a lot of different industries. I go really into uh you know, Henry Ford and how he sort of dropped the cost of of the Model T.

Uh I look at sort of the evolution of like nail manufacturing which is uh in the in the sort of the 19th century go back even farther and how they changed the technology to make over time to make nails which started out like hand forge nails like a blacksmith like hammer and steel and they found a way you know machines that could sort of emulate that process and they replaced those machines with even better machines and uh so on.

So there's like dozens and dozens and dozens of examples in the book of sort of specific things uh that have gotten cheaper over time and the lessons that we can kind of learn from those things. How naive is it to just say uh what's remain stubbornly high costwise? Housing, medicine, education.

What do those have in common? Regulation. How how naive is it to just throw regulation is the problem at those particular industries? Uh that's a big part of it for sure. I mean the problem is that like everything has gotten more regulated, right? Like manufacturing included.

Um so it's it's like that's like part of the puzzle, but it doesn't really tell you the whole thing because even in PL, you know, the problem, you know, to take it back to construction, uh the problem of like construction productivity and not getting cheaper to build stuff is really something you kind of see around the world.

you like I I have a graph in there that's like construction costs in like a variety of different countries and they all kind of this you know scary line of going up and uh to to the right over time even the countries without building codes and ownorous HOA you know less labor yeah or like different regulatory regimes and stuff and there's certainly places that like do better than the US in in various things like in various ways of building the US is like very far from the efficient frontier um but we have a very hard time of like pushing ing that efficient frontier for so like regulation is like a big part of it but that's kind of one of the sort of things I think a takeaways from the book is that it's not just regulation like you could have all the you know remove all the regulation you wanted and you'd still run into these sort of various physical constraints and market constraints that prevent these sort of efficiency improvements uh in some cases how are you thinking about energy in America we've we've we've gone through this AI boom now where uh we've scaled up the existing capacity of data centers.

We're building new data centers and it feels like the last link in the chain is can we build a 100 nuclear reactors in America in 2030 to stay on track with like the most aggressive projections.

Um is there anything unique about uh obviously energy production is a construction problem but is there anything unique that you found in the energy industry that uh folks might be able to learn from? Yeah, I'm well, you know, I write a lot about energy on on the on the newsletter.

Uh I don't have a background in energy, so it is a lot of me like groping my way towards like some understanding of of how this industry works. Uh I'm a really big solar guy. Solar has like a really lot of nice properties that like makes it easy to sort of uh make efficiently at like very very large scale.

There's this really interesting paper um but basically it's this like big graph of like the sorts of energy technologies that have become cheap and the sort of energy technologies that have not become cheap and the ones that have become cheap are these sort of things that like you can make repetitively in very large volumes and you don't need like a lot of customization of and so like solar panels which are like you can make in like really really really really enormous numbers and you can kind of plop down wherever it doesn't need a lot of like sightsp specific customization uh are kind of in this like very cheap quadrant.

And then something like a nuclear reactor which you make in like much much smaller numbers and like needs a lot of like specific design for the specific reactor that you're building is sort of in the much more expensive uh quadrant.

Um and so solar and like the batteries which like really complement them really nicely is like a really good way to sort of make this stuff really cheap. these cost curves have like gone like down like a lot and there's like no sign that those are stopping anytime soon.

And so um you know that just you know those this aligns with like so much of what we know about what what it takes to sort of make something inexpensive that I kind of see that like biting off a very large chunk of the of the energy uh that we produce uh in the US assuming you know take it back to regulation assuming that sort of regulation interferences don't kind of get in the way.

How how often did you find uh capital being a constraint lead to more efficiency?

I think every startup founder has a has like an example of a time when like maybe if they they threw uh you know a hundred people at a problem they would have gotten a different solution but they only had a handful and so they were able to create some novel uh uh a more efficient way of doing something uh or we saw this with like deepsek and and having having potentially fewer chips and creating a more uh efficient architecture.

Was that a common theme at all in in uh in the Yeah, it's it's interesting. I think there's kind of like two sides of it.

One is that in some cases like what a repeated theme of the book is that like scale is really really very important and if you can the more you can make of something the more opportunities you have to make that less expensively and often scale is like very very expensive right so like one of the sto the story of like container shipping over time is a story of like needing really really enormous investments to like build these big giant ships which are like cheaper per container that they're transporting but very expensive overall.

all and also like really really big expensive terminals to sort of handle those ships. And so only like a certain number of like countries could like invest in these like giant terminals that were needed to sort of service these huge ships.

And so you know cost of transporting these goods fell a lot but like there was winners and losers in who sort of gained gain from this technology development. It was really the people that could afford uh to put the money into it to do it.

Um, but then on the other hand, you also see cases where kind of like you talked about with Deep Seek, people working under these constraints were able to come up with like really improved ways of of doing something that were much cheaper and much better than what was uh what came what came before.

So a kind of example of that would be like Toyota's manufacturing methods which were like Toyota production system which evolved into lean manufacturing.

uh those kind of were created in this environment where like they couldn't develop these like mass production methods that Ford had used because their car market was so much smaller and it was so much more varied.

They couldn't just make a million of a given model or whatever that they had to find ways of like producing this stuff efficiently that didn't require this like massive capital investment basically. And so that was sort of the genesis of that those those ideas.

And so yeah, I think there's definitely cases where yeah, you need like a lot of investment to sort of find ways to make this cheaper, but then there's also cases where it's like also working under constraints of not very much investment has has uh been important as well.

Are you at all optimistic that uh this data center boom will teach a generation of people uh uh uh that you can build big things quickly and efficiently if you just basically put your mind to it because there's like a lot of from from an energy standpoint just like you know if you look at what what Elon has done with um Colossus 2 he's basically doing the impossible.

a lot of people would have like looked at that project and said it's not possible and so that sort of it feels like that sort of mindset of like we're just going to make it happen.

Uh these is being applied to data center development but then presumably those people can say I'm going to build a bridge and uh they can imply that same kind of approach elsewhere. I I certainly hope so. We're certainly building like an enormous amount of this infrastructure like it's really really unprecedented.

there's all these crazy stats like you know data center spending is now uh exceeded like office building spending or or something like that which is which is totally wild. I guess one thing that worries me is that historically people have been like you know not really cared about data centers.

They've been happy to just like let them get built and the jurisdictions sort of collect the tax revenue for it and and not really worry about it beyond that.

uh as like the buildout of them is like going forward and there's like more and more of these data centers and they're bigger and larger uh you're really starting to see like a grassroots movement of people like you know the nimbies sort of now being opposed to data centers in a way that they weren't before.

So like Virginia which historically has like been you know a major place where data centers get built and has basically been fine with them getting built there.

Now you're starting to see like residents oppose them more and more and you're starting to see, you know, grassroots movements around in different states uh springing up to oppose these things.

So that worries me a little bit and I hope the sort of forces of getting these things built and enthusiasm about building infrastructure are are stronger than that or um but uh you know it always seems like uh the NIMBI forces are are quite strong. So hopefully they uh they uh they don't build momentum. They're OP.

I have one last question. Um there's this post by Rune who's talking about Dan Wang's new book. Uh and he says the general elite consensus now is that industrial process is a technology that lives in the heads of people.

And he goes on to say that it was a mistake to let so much lowv valueue industry be offshored due to the loss of tacid process capital.

And I was just wondering what your thoughts were on this idea of industrial process knowledge that there might be a few key people that actually know how to build something at scale and uh just what the ratio how how steep is the power law of human capital when it comes to largecale industrial manufacturing efforts.

Yeah, I think it's dead on.

And I think that's absolutely very important and I talk about that at various parts uh in the book how it's often really hard to transfer like manufacturing or production technology from one place to another place in part because it's hard to like pick up and lift these uh process knowledge which is just in the heads or like embedded in this web of relationships and so it doesn't necessarily even exist in explicit form, right?

It's just like this is this system that turns out to work very well and you can't just like recreate it because we don't essentially know how it how it came to be in the first place.

And then you know we talked you talked about a little bit about the learning curve earlier and that's kind of this really similar idea where a lot of your improvements to some technology over time come from just like the factory floor and learning how to sort of do this um better and better over time, but it's very coupled with actually physically doing the the work.

And so that's one thing that I yeah I I think is is really important is that often times just technological progress is coupled to sort of this like process factory knowledge of actually having the experience uh doing things.

One one really fun sort of example of this is um during the the early days of of the space race where the US was having like a really hard time uh building their rockets and there's a part where like you know because of various political things uh uh the uh Navy was going to send up their rocket uh first they were going to be like the first ones to sort of launch a US satellite into space and Wernern von Braonn who was the German rocket scientist who then had been brought over to the US and was working for the army He goes to some like, you know, military leader and he says, "Look, you can tell these Navy guys they can do whatever they want.

They can take my rocket and they can paint Navy on the side of it and do whatever they want, but they need to use my rocket and not theirs because my rocket will work and their rocket won't. " And then what ended up happening was they didn't listen to him and the Navy launched their rocket anyway and it didn't work.

It blew up on pad and then so finally they listened to Wernner von Braonn and just launched his rocket and that's when we finally got uh a satellite into space using Ver Wernner von Braonn's uh rocket and then of course Wernon Braun was like a major force in the Apollo program as well.

So it was like you know the German the German rocket knowledge that had accumulated during World War II was like very very important and both the U both the Soviet and the US uh their early rocket development efforts were basically built on this German knowledge that had been accumulated.

So this process knowledge and like this you know expertise that gets embedded in the heads of these of these people working at the sort of forefront of technology uh is not easy to sort of recreate. Um, I think it's very very important. Well, thank you so much for stopping by the show.

The book is Origins of Efficiency from Stripe Press. It's available now for purchase. Highly recommend picking it up. One click on Amazon. Thank you. Stripe checkout hopefully. Hopefully. Uh, we will talk to you soon. Have a great rest of your day. Thanks. Thank you so much. Um, really quickly, let me tell you