Dwarkesh PodcastAustin Vernon - Energy Superabundance, Starship Missiles, & Finding Alpha
EVERY SPOKEN WORD
150 min read · 30,119 words- 0:00 – 2:25
Intro
- DPDwarkesh Patel
And you're, you're building this out of your garage, correct?
- AVAustin Vernon
(laughs) Yeah, yeah. (music) So the blockchain application, I'm actually, like... You know, what excites me personally the most is re-imagining enterprise software. (laughs) You're gonna be like Kardash whatever scale civilization, where you're using like immense amounts of energy. Like that's gonna have, you know, side effects, and you're gonna have to figure out how to manage that one way or the other. And I mean, one of those is eventually Earth may just be like a, a nature preserve, and we all live in space or something. But...
- DPDwarkesh Patel
Okay. Today, I have the pleasure of interviewing Austin Vernon, who writes about engineering, software, economics, investing on the internet. But not that much else is known about him. So Austin, do you wanna, do you wanna give us a bit about your, uh, background? I know that the only thing the internet knows about you is this one little JPEG that you had to upload with your recent paper. But (laughs) who, what, uh, w- what about an identity reveal, or I guess, a little bit of background reveal, to the extent that you're willing to, comfortable sharing.
- AVAustin Vernon
My degree is in chemical engineering, and I'm kind of like a life lo- long love of engineering, and also things like Toyota production system and stuff like that. And I've worked as a chemical engineer, like in a large processing facility. I've done a lot of petroleum engineering. Let's see, then now, you know, I taught myself how to write software, and now I'm working on kind of like more research, early commercialization of CO2 electrolysis.
- DPDwarkesh Patel
Okay yeah. So I'm, I'm, uh, I'm really interested in talking about all those things. Um, but I, I, so I guess the first question I have is Alex Berger, who's the, um, co-CEO of Open Philanthropy, um, he, he asked this question when I asked on Twitter what I should ask you. And you suggested I should ask you, "Why so shady?" So what do you... Uh, you, you have a, I mean famously, you have kind of like a anonymous personality, uh, pseudonymous thing you have on the internet. What's up with that? (laughs)
- AVAustin Vernon
Uh, yeah. Yeah, when, you know, he... I think he posted a tweet that said, you know, like, "I don't, I don't know who this guy is, or like if he's credible at all. But, you know, his stuff sure is interesting." And that, that really made me laugh, and that was hilarious. Um, yeah, it just doesn't, doesn't seem necessary. I think I'm fine with my, um, my ideas being well known, and, and communicating but I have, I have less desire to be like personally famous, so.
- DPDwarkesh Patel
Ah, gotcha, gotcha.
- 2:25 – 19:56
Starship as a Weapon
- DPDwarkesh Patel
Uh, I wanted to start off with, uh, a, a, a sexy topic. So, uh, c- c- let's talk about s- using Starship as a kinetic weapon. Uh, I thought that was like one of the, one of the more amusing posts you wrote. Do you wanna talk more about how this would be possible?
- AVAustin Vernon
Um, well, I think the main thing with Starship is like it's... You know, you're taking a, a technology, and you're making it about 100 times cheaper for cargo, and 1000 times cheaper for people. So when things like that happen that drastically, uh, you're just like look- looking at huge changes. And it's, it's really hard to anticipate what some of those can be, when the change is that drastic. So I think there's like a lot of moon-based, Mars-based stuff that, you know, doesn't really catch the regular public's eye. And I think they also have trouble imagining some of the like point-to-point travel that could be possible. But as far as like, you know, you start talking about it like as a weapon, and I think that's, um... You know, it lets pe- people know they should be paying attention to this technology. That, and we'd certainly do not wanna be second or third getting it. And then we should make sure that we're gonna be first.
- DPDwarkesh Patel
Yeah. I think you mentioned this in the post, but, um, so I, I, I would... As, as recently as the '90s, the cost of sending one kilogram to space was around 20,000. More recently, SpaceX has brought it to 2,000, and then there's like a lot of interesting questions y- you could ask when you ask, "What will be k- possible once we get it down to $200 per kilogram to, uh, to send into orbit?" Yeah, so, uh, uh, one of them might be to manufacture these, uh, weapons that are not conventional ballistics. But do you wanna talk about like why this might be an advancement over conventional ballistic weapons?
- AVAustin Vernon
Uh, well, regular conventional ballistic weapons are extremely expensive, (laughs) You know, this is more like a bomb truck, you know? But it's even like... Usually we think of like B-52 as the bomb truck, and this could be even, you know, cheaper than the B-52, um, delivering just like mass on target. Uh, when you think about like how expensive it is to fly a B-52 from like Barksdale, Louisiana all the way across the world, you can do it from South Texas or Florida with the Starship, and get, you know, more missions per day. And the fuel ends up being... Like w- when you go orbital, it takes a lot to get to orbit, but then once you're in orbit, your fuel consumption's pretty good. So over long distances, it has a lot of advantage, so I feel point-to-point works for like the longer distances. There's really like a sweet spot with these weapons, where you want it to be like pretty accurate, but you also want it to be cheap. Like, you're seeing that problem with like Russia right now, as they have some like, you know, fancy parade-style weapons that are really expensive, like multi-billion dollar cruise missiles. But they're missing like that, you know, $5,000 guided artillery shell, or like that, you know, like $20,000 JDAM, that you can just like pit massive... Or the, you know, the multiple launch rocket system, guided rockets. They're really like short on all those, 'cause I think they had, just had like a limited amount of chips they could get from the US into Russia to make these advanced weapons. But, uh, yeah, so the, it kind of, a Starship gives you just like a platform to deliver, like you could, you know, pit JDAMS in a shroud, or you could just like, you know, have, uh, the iron unguided kinetic projectiles. And it just becomes impossible for, you know, a ship to launch missiles to intercept yours if you cost, if your cost is so low. You can just overwhelm them.
- DPDwarkesh Patel
All right. There, there are a few terms there that, uh, neither I nor the audience might know. So what is, uh, what is, what is JDM, what is Shroud? And why are chips a bottleneck here? Uh, like, why, why can't it just be any micro-controller?
- AVAustin Vernon
So JDM is Joint Direct Attack Munition. So what we did is we took all, like, our Vietnam surplus bombs, and we put this, like, uh, little, like, fin kit on it and it costs like 20,000 bucks, which is cheap for a weapon 'cause it, um... You know, the actual bomb costs, like, I don't know, 3,000 bucks. And then you... It turns, you know, it into a guided weapon that before you're, you were probably lucky to get within 500 meters of the target. Now you can get it in with, like, two meters. So the number of missions you have to do with your planes and all that goes down by, like, orders of magnitude. So it's absolutely, like, huge advantage in logistics and, and just how much firepower you can put on a target. And the, uh... And, you know, like, we didn't even have to make new bombs. We just put these kits on all our old bombs. Let's see. Then the chips are a problem... There's, like, this organization called RUSI. I think they're in, in the UK. But they've been tearing down, like, all these Russian weapons they found in U- Ukraine, and they all have American chips in them. So, you know, technically they were supposed to, like... They're not supposed to be able to get these chips. And, you know, Russia can't make a lot of its own chips, and especially not the specialized kinds you might want for guided weapons. So they've been somehow smuggling in chips from, from Americans to make their advanced weapons.
- DPDwarkesh Patel
What, what is special about these? I would assume that, like, uh, they haven't, like... As far as I'm aware, the trade with China is still going on, right? And we get, uh, a lot of our chips manufactured from Taiwan or China. So why can't they do the same?
- AVAustin Vernon
It's the whole, like, integration. Like, you know, it's not just, like, a specific chip, but, like, the, the board. It's like... They're more like PLCs where, um, where you, like, wi- almost have, like, wired in, um, programming and stuff like that. And they come with, like, this... Like, the... To be able to do the guidance and all that stuff, it all kind of has to work together. I think that's the way I understand it. I don't know. I really don't have a really good answer for that one. But they're hard to replicate, is what matters.
- DPDwarkesh Patel
All right. That's interesting. Um, yeah, and I guess that has a lot of interesting downstream effects because, uh, for example, India buys a lot of its weaponry from Russia, right? So i- i- (laughs) if Russia doesn't acc- have access to these, then it- other countries that buy from Russia won't have access to these either. You had an interesting speculation in the post where you suggested that you could just keep these kinetic weapons in orbit, like a Sword of Damocles really, uh, almost literally. Yeah, that sounds like a really, uh, scary and risky scenario where, I don't know, you could have orbital decay and you can have these kinetic weapons falling from the sky and destroying cities. Um, do you think this is what it will look like in... or could look like in 10 to 20 years?
- AVAustin Vernon
Well, yeah. So the, the advantage of having on orbit is you can hit targets faster. So if you're launching the rocket from Florida, you're looking at, like, maybe 30 minutes to get, get there. So, you know, the target moves in that time. Whereas if you're on orbit, you can have them spaced out to where you're hitting, you know, within, like, a few minutes. Um, so that's the advantage there. When you actually look at like the... You, you really have to have, like, a two-stage system I think for most because, um, if you have, like, a really aerodynamic rod that's gonna give you good performance in the low atmosphere or get going too fast and just, like, burn up before you get there. Um, you know, tungsten's maybe the only thing that you could have that could go all the way through. Uh, that's why I like the original concept, use these big tungsten, like, rods, the size of, like, a telephone pole. But, you know, tungsten's pretty expensive and, like, just the rod concept, it kind of limits to what you can do, um, if you just do the rods. So a lot of these weapons will have, like... That's why I was talking about, like, with the shroud, like, something that actually slows you down in the upper atmosphere. And then once you're to the velocity where you're not just gonna melt, then you open it up and let it go. So if you actually had it, you know, fall from the sky, some may make it to the ground, but, um, a lot would burn up. So the, uh... A lot of stuff that makes it to the ground is actually pretty light, you know. It's, like, stuff that can kind of, like, float and has a large surface area. Um, yeah. So that's like the whole thing with Starship, like, the... Or not Starship, but, um, Starlink. All those satellites were meant to completely, you know, fall apart on de-orbit.
- DPDwarkesh Patel
I see. Like, one of the implications of that is that these may be less powerful than we might fear because if, um... Like, if kinetic energy is mass times, uh, you know, velocity squared, then you have to... If there's an upper bound on the velocity and then the velocity is the component that grows, um, the kinetic energy faster, then it s- it suggests that you can upper bound the power these things will have. You know what I mean?
- AVAustin Vernon
Yeah. So, so even the tungsten rod. Sometimes people, like, you know, they're not good at physics or something, so they don't, like, do the math, um. (laughs) They think it's going to be like a nuclear weapon, but it's really... I think even the tungsten rod, I might have put it in there, I think, if I'm remembering correctly, like 10 tons of TNT or something. It's like a big, a big bomb, but it's not, um... You know, it's not like a super weapon. So it... That's what I think I said in the post. It really has like... It's like advanced missiles where they're almost more defensive weapons. So I can keep you from hitting your ship somewhere, you know, and like, yeah, I could, like, try to bombard your cities, but I can't, I can't take ground with it, you know. I can't even, like, police sea lanes with it really. I'd still have to have regular ships, you know, if I had this air cover essentially to, you know, go, like, enforce the rules of the sea and board freighters and stuff like that.
- DPDwarkesh Patel
Yeah. So I... You, you speculated in the post, I think, that you could have, like, potentially the- these... Uh, you could, like, load this up with shrapnel and then it could, like, explode next to, uh, an incoming missile or an incoming aircraft. Um, yeah. C- could these get that accurate? Uh, 'cause that was surprising speculation to me.
- AVAustin Vernon
Uh, yeah, I think, like, for ships, I think it's pretty... you know, like, I was, like, watching videos of, uh, you know, how fast a ship can turn and stuff because you'd want to, like, release your shrapnel, even if you're gonna, like, do an initial target on a ship to, like, try to kill their radars and stuff. You'd want to do it above the ceiling of their missiles. So it's like how much are they going to move between your release, where you stop steering, and, and that. And it's really, you know, it's like maybe like a thousand feet. So that's pretty simple because you just like shrapnel the area. The aircraft, you would be steering all the way in. So it, it's maybe actually... I'd say it's doable, but it'd be pretty hard. Yeah. And you'd actually maybe want to even go slower than you would with the ship attack. You know, you need like a specialized package to do the aircraft. But you can see these aircraft on, um... like if you have enough synthetic aperture radar and stuff like that, you can see them with satellites and then guide it in the whole way. You could even, like, say load, like, heat-seeking missiles into a package that, you know, stop, you know, unfurls right next to them and launch conventional missiles too, probably. And that's, uh, that'd be pretty hard to do some of this stuff, but it's just like kind of, you know, the things you might be able to do if you put some effort into it.
- DPDwarkesh Patel
Yeah. The reason I find this kind of speculation, uh, really interesting is because, uh, when, when you look at the, uh, uh, modern weaponry that's used in conflicts, it often seems like th- it just seems like directly descended from something you would have seen in World War II or something. It, it doesn't seem, um... Like, like if you think about, like, how much warfare changed between like, uh, 1900 and 1940, it's like, yeah, they're, they're not even the same class of weapons anymore. Um, so it's interesting to think about possibilities like these where the entire, um, the entire category of, uh, weapons has changed.
- AVAustin Vernon
Oh, right. And that's because, you know, the same thing, like, you know, our physical technology hasn't changed that much. So it's, it really has just made more sense to like put better electronics in the same tanks we have than to build... Like, you're just not going to get... We haven't learned enough about tanks to build like a new physical tank that's way better. So we just keep upgrading our existing tanks with better electronics. So they're, they're much more powerful, they're more accurate. You know, a lot of times they have longer range weapons, they have better sensors. So the tank looks the same, but, you know, it may be as like several times more, like, killing power, whatever, what have you. But, you know, the Ukraine war right now kind of... You know, they're using a lot of like 40, 50-year-old weapons. So that especially looks like that. (laughs)
- DPDwarkesh Patel
Yeah, yeah. Um, which, which kind of worries you if you think about the stockpiles our own military has. I'm not well-educated on the topic, but I imagine that we don't have the newest of the new thing, right? Like, we'll probably have maintained versions of decades-old technology.
- AVAustin Vernon
Uh, you know, I mean, we spend so much, we've got relatively... This, this kind of gets into like there's a lot of debate about like how ready our military is and for certain situations, it's more ready than others. Um, I'd say in general, most people talking about it have the incentive to, um, downplay our capabilities because they want more defense spending or just, there's lots of reasons. So I think we're probably more, more capable than, than what you might see from like, you know, some editorial in The Hill or whatever. And I think just like us just sending a few weapons over to Ukraine and how successful they've been using them, I think shows a little bit of that. But there's, there's so much uncertainty when it comes to, uh, to fighting, you know. Especially when you're talking about like a naval engagement where there, we just don't have that many ships in general. You can have some bad luck. So I think, I think the... You know, you always want to be a little bit wary. You know, don't want to get overconfident.
- DPDwarkesh Patel
Yeah. And if, um... like if, if the offensive tech we sent to Ukraine is, uh, potentially better than the defensive tech, um, like, it, it's very possible that even a ballistic missile that, uh, China or Russia could launch could sink like a battleship and then kill 2,000, uh... you know, a thousand or whatever soldiers that are on board. Or I, I guess, I don't know, you think this opens up avenues for defensive tech as well or...?
- AVAustin Vernon
Yeah, I mean, generally the, the consensus is that, um, defensive technology has improved much more recently than offensive technology. And they're still trying to... This whole, like, strategy China has, they call, um... it's like area denial, anti-access area denial, A2AD. Um, and the... so that's basically just like missiles have gotten better because the sensors on missiles have gotten better so they can keep our ships from getting close to them. But you know, they can't really challenge us like in Hawaii or something. And it really goes both ways. I think people forget that. So yeah, it's like hard for us to get close to China but, you know, Taiwan has a lot of missiles with these new sensors as well. So I think it's probably tougher for China to get close to Taiwan than, than most people would, uh, would say.
- DPDwarkesh Patel
Oh, interesting. Yeah. Can, can you talk more about that? Because every time I read about this, people are saying that if China wanted to, they could, uh, uh, knock out Taiwan's defenses in a short amount of time and take it over. Um, yeah. So can you talk about why that's not possible?
- AVAustin Vernon
Uh, well, it might be, but I think it's because of the uncertainty thing. But Taiwan, you know, it has actually one of the largest defense budgets in the world and they've recently been upping it. I think they spend like, I don't know, 25 billion a year and they added like an extra five billion. They've been buying a lot of anti-ship missiles, a lot of air defense missiles, stuff that like, you know, Ukraine could only dream of. I think Ukraine's military budget was like two billion. They have, you know, professional army. They're gearing... And then the other thing is they're an island. So, you know, like, Russia could just like roll over the land border into Ukraine. But, you know, I mean, almost... There are just very few successful amphibious landings in history.... like some of the recent, most recent ones were all, you know, the Americans in World War II and Korea. So like, the challenge there is, is just, you know, it's kind of on China to, like, execute perfectly and do that. And so if they, like, had perfect execution, then possibly. But you know, if, like, maybe their air defenses on their ships aren't quite as good as we think they could possibly be, then you know, they could also end up with half their fleet underwater within, you know, 10 hours.
- DPDwarkesh Patel
Mm. Interesting. And how has your view of Taiwan's defensive capabilities, how, like, how has the Ukraine conflict updated your opinion of what might happen?
- AVAustin Vernon
I didn't really know much about it and then, you know, I like started looking at, like, Wikipedia and stuff and, like, all this stuff they're doing. And so you know, the... Taiwan just has, like, a lot of modern platforms, like F-16s with AR anti-ship missiles. They actually have a lot of their own. They have indigenous fighter bombers, indigenous anti-ship missiles because, you know, they're worried we might not always sell them, you know, them... They've even recently gotten these, uh, long range cruise missiles that could possibly target leadership in Beijing. So I think that makes it uncomfortable for, you know, the Chinese leadership. Like, if you attack them, you're gonna have to go live in a bunker. Um, so there, there's lots of like... You know, but again, there's... I'm not, like, a full time military analyst or something, so there's a lot of uncertainty around, around what I'm saying. There's not... It's not a given that China's just gonna roll over them.
- DPDwarkesh Patel
Okay. That's comforting to hear. Let-
- 19:56 – 42:12
Software Productivity
- DPDwarkesh Patel
let's talk about an area where, uh, I have a little bit of point of contact. Um, I thought your blog post about software and the inability of it to increase productivity numbers, I thought that was super fascinating. So, before I ask you questions about it, do you wanna... Do, do you want to lay out the thesis there?
- AVAustin Vernon
Yeah. So the, uh... Yeah, so if there's one post I kind of, like, felt like I caught lightning in a bottle on, it was that one. It really... Like, everything I wanted to put in it just, like, fit together perfectly, um, which is usually not the case. But yeah, I think, I think the idea is the world's so complex and we really underestimate that complexity. And if you're going to, like, digitize processes and automate them and stuff, you have to capture all that complexity basically at the bit level. And that's extremely difficult. And then you also have, like, diminishing returns where, like, the easily, you know, automatable stuff goes first and then it's, like, increasing corner cases to get to the end. Um, so you just have to write, like, more and more code basically. And so that's why we don't see, like, runaway productivity growth from software is because we're fighting, you know, all this increasing complexity.
- DPDwarkesh Patel
Yeah. Have you, have you heard of the waterbed theory of complexity, by the way?
- AVAustin Vernon
Uh, I don't think so.
- DPDwarkesh Patel
Okay. It's, it's, uh, it's something that comes up in compiler design, but the idea is that there's, like, a fixed amount of complexity, um, in a system and if you try to reduce it, uh, what you'll end up doing is just you'll end up migrating the complexity elsewhere, right? So I think an example that's used of this is, is when they try to program languages that are, uh, not type safe, something like Python. You can say, "Oh, like, it's a less complex language," but really, you've added complexity when, when, I don't know, two different types of numbers are interacting, uh, like a, like a float and an int, right? Um, you've added complexity there, uh, that... I mean, as your program grows, that complexity exponentially grows of all the things that could go wrong when you're making two things interact that are... In a way that you, you were expecting not to. So yeah, the, the idea is, um, you, you can just choose where to have your complexity, but you can't get rid of that complexity.
- AVAustin Vernon
Yeah. Yeah, yeah. That's... I think there's like this... There's kind of, like, an interesting thing when you start pairing it with management theory and, like, how it kind of starts tying into some of my other posts is that for a long time, when you add up, like, all the factors, the most complex thing we are doing is, is, you know, high volume car manufacturing. And so we got a lot of innovations and organization from car manufacturers, like the assembly line. Then you had Sloan at GM basically, you know, creating the way the modern corporation is run. And then you have Toyota production system. But arguably now, creating software is actually the most complex thing we do. So there's, like, all these kind of, like, squishy concepts that underlie, like, Toyota production system that software has had to learn and, like, reimagine and adopt. And you know, you see that with, like, Agile where we can't have long release times. We need to be, like, releasing every day, which is like, you know, we're limiting inventory there. Or, um, yeah, there's just... There's, like, a whole thing especially that's showing up in software that existed in car manufacturing where you're talking about, you know, reducing communication. So, like, Jeff Bezos kind of, like, now famously said, you know, "I want to reduce communication," which is counterintuitive to a lot of people. This is, like, age old in car manufacturing where you have, like... Toyota has these cards that go between workstations and they tell you what to do. So people normally think of them as limiting inventory, but it also tells the worker exactly what they're supposed to be doing, at what pace, at what time. And the assembly line is like that too. You just, like, know what to do because you're standing there and there's a part here and it needs to go on there. Um, and it comes by at like the pace you're supposed to work at. And there's... It's, like, so extreme that there's this... I think it's a famous paper, but it's by, like, List, Cyverson, and Levitt. And they went to a car factory and like, you know, studied how, like, the defects propagated in cars and stuff. And once a, like, a car factory gets up and running, like, it doesn't matter if you... what workers you put in there, like, if workers are sick or you get new workers, like, the defect rate is the same. So like, everything is just like... All the knowledge is built into the manufacturing line. And there's, like...... these concepts around, like, idiot proofing and everything like that, that, um, are very similar to, like, what you'll see, you had Uncle Bob on there. So Uncle Bob, you know, says, like, only put one input into a function and stuff like that because you'll mix them up otherwise. So it's kind of like this, the Japanese call it, like, poka yoke, and it's like you, you make it where you c- you can't mess it up. And that's another way to, like, reduce communication. And then software, of course, you have API. So I'm really interested in this overall concept of, like, reducing communication and reducing, um, co- how much cooperation and, like, and everything we need to run the economy.
- DPDwarkesh Patel
Right. Right. The, speaking of the Toyota Production System, like, one thing they do to reduce that defect rate is, if there's a problem, um, all i- all the workers in that chain are forced to go to the place where the defect or problem is and fix it, uh, before doing anything else. And I guess the idea there is this will give them context to understand what the problem was, how to make sure it doesn't happen again, and also prevent a buildup of, um, inventory in a way that, like, keeps making these defects happen or just keeps, uh, keeps accumulating, um, inventory before the place that can fix the defects is able to take care of them.
- AVAustin Vernon
Right. Yeah, yeah, exactly.
- DPDwarkesh Patel
Yeah. Um, but, you know, I, I think, uh, one interesting thing about, uh, software and complexity is, uh, you, I think you said a little bit earlier that software is a place where complexity is, is the highest, uh, in the world right now. And one of the interesting things is, yes, that's true, but you, y- uh, I guess software gives you the choice to interface with the complexity you want to interface with. And I guess that's just part of specialization in general. But you could say, like, for example, um, uh, a machine learning model is, like, really complex. Uh, but ideally, you get to a place where that's the only kind of complexity you have to deal with. You're not having to deal with the complexity of, like, how is, how is this program compiled? Um, like, well, how, well, you know, like, how are the libraries that I'm using, how are they built? You can al- you can, like, fine-tune and work on the, uh, complexity you need to work on. It's similar with, like, app development, right? I, uh, Byrne Hobart has this blog post about Stripe as solid state. I, I forget the exact title of the blog post, but the basic idea is that Stripe hides all the complexity of the financial system. It charges a higher fee, but you can just kind of treat it as an abstraction of a tithe you have to pay, and it'll just take care of that entire process, and you can focus on your comparative advantage.
- AVAustin Vernon
Yeah. And it's, it's really actually very similar in, like, car manufacturing and Toyota Production System if you really get into it. It's, it's very much the same conceptual framework. Also, there's, like, this whole idea, like, in Toyota Production System, everyone wo- everyone works at the same pace, which we kind of talked about. But also, like, your content, your work content is the same, like, there's, there's, there's no room for not standardizing, like, a c- a way you're going to do things. So everyone, like, gets together and they're like, "All right. We're going to... This certain part, we're going to put it together this certain way at this little micro-station. And it's going to be the same way every time." And that's part of, like, how they're, you know, reducing the defect rates. And then if you, you know, like if your assembly pro- is too long, like it's longer than what your, like, time allotment is to stay in touch with the rest of the process, then you just keep breaking it down into smaller pieces. And so, you know, each, each person only has to know, like, a very small part of it. And even the, you know, even like the, the overall engineering team, you know, has all sorts of strategies like this too. There's all, they have all sorts of, like, tools to, like, help them break up all these processes into very, like, small parts and to make it all, like, hold together. Still very, very hard. But, um, it's kind of the, like, a lot of the same ideas, 'cause you're taking away, like, the complexity of making, like, a $30,000 car, or 30,000 part car, where everyone's just focusing on their one, their one little part and they don't care how, what someone else is doing.
- DPDwarkesh Patel
Yeah. But the, uh, uh, but the interesting thing al- also there is y- it seemed like there, you need one person who knows how everything fits together because one of the pro- uh, uh, from what I remember, one of the tenets of the Toyota Production System was you need to have a global view. So, I mean, in that book, um, was it The Machine or the other one? The Toyota Production System book? But anyways, they were talking about examples where people would try to optimize for local efficiencies. I think, uh, they, they especially pointed to, like, Ford and GM for trying to do this, where they would try to make machines run all the time. And locally, you could say that, "Oh, this machine," or, "Oh, this process is super efficient." You know, it's always outputting stuff. But it ignores how that, uh, that added inventory or that process had a bad consequence for the whole system. And it, so it, it's interesting if you look at a company like Tesla that is able to do this really well. The interesting thing is that Tesla is run like a monarchy, um, and this one guy has this, like, total global view of how the entire process is supposed to run, where do you have these inefficiencies. You have some great examples of this in the blog post. But yeah, I, I think one of the examples that... Oh, I think, was it the Toyota Production System book? But anyways, this guy goes to this factory and, the author, and he asks, um, "Is this, like, an efficient factory?" And the guy's like, "Yeah, this is totally efficient. There's nothing we can do adopting the Toyota way to make this more efficient." And so then he's like, "Okay, let me look." And he finds that in one of the, uh, so, uh, they're like treating steel in some way, but it's only, it, it should only take a couple of seconds, um, and the main process does only take a couple of seconds. But some local manager decided that it would be more efficient to ship, uh, their parts out to get the, ge- get the next stage of the produ- uh, process done somewhere else. A- and, uh, so thi- this is, like, locally cheaper, but the result is that it takes weeks to get these parts shipped out and get them back. And so that means that the actual time that the parts s- spend getting processed is like 0.1% of the time, which makes the whole process as a whole super inefficient, right? So I don't know. It, it seems like the implication is you need, uh, you need, like, a, a very, like, monarchical structure with, like, one person who has a total view i- in order to run such a system? Or am I getting that wrong?
- AVAustin Vernon
Um, not necessarily. I mean, you do have to, like, make sure you're not, um, optimizing locally. But I think it's the same, you know, you have that same constraint in software, but I think a lot of times people are just like running over it because processing has been getting so much cheaper, (laughs) you know? And I, like, people are expensive, so like if you could save development time, you know, it, it just ends up, you know, the, the trade-offs are different. When you're talking about, like, the tyranny of, like, physical items and stuff like that, it, um, you know, it get, the constraints get a little more severe. But I think, uh, you have, like, the same, the same overall, you know? You still have to fight local optimization, but the level you have to is probably different with physical goods. I was thinking about, like, the, the smart grid situation from, like, a software perspective. And, um, like, there's this problem where, like, okay, I'm putting my solar farm here and it's impacting somewhere far away and that's then, like, creating these, like, really high upgrade costs, you know, that cost two or three times more than my solar farm. Well, you know, the, the obvious thing would be, if you're doing software, is like you're, like, gonna break all these up into smaller sections, and then you wouldn't be impacting each other and all that and you could, you could work and focus on your own little thing. But the problem with that is, uh, wi- if you're gonna, like, disconnect these areas of the grid, is that the equipment to do that is extremely expensive, you know? Like, it's not like I'm just, like, gonna hit a new tab and open a new file and, and start writing a new function. Um, and not only that, but you still have to, like, actually coordinate how these, this equipment is gonna operate. So if you just, like, let the grid flow as it does, everyone knows what's gonna happen because they could just calculate the physics. So if you start adding in all these checkpoints where humans are doing stuff, then you have to, like, actually interface with the humans, and, and the, the amount of things that can happen really starts going up. And so it's actually, um, a really bad idea to, to try to cart all this stuff off just because the, f- like, the reality of the, the physical laws and the equipment you need and everything like that.
- DPDwarkesh Patel
All right, interesting. And then I think you have a similar, uh, sort of like cosine argument in your software post about w- why, uh, ver- vertically integrating software is, is beneficial. Do, do you wanna explain that thesis?
- AVAustin Vernon
Yeah, and I think it's just like, you know... It, it actually gets to what we're talking about here, where it allows you to, to avoid, like, the local optimization. Because, you know, a lot of times, we're at- you're trying to build, like, a software MVP or you're, like, tying together, like, a few services. They don't do quite what you need, so if you, like, try to scale that, like, it would just break. Um, but if you're, like, gonna take a really complex process, like car manufacturing or distribution, retail distribution or, you know, like, the home buying process or something, you really have to vertically integrate it to be able to create, like, a decent end-to-end experience, um, and avoid that, that, you know, local optimization. Um, and it, you know, it's just very hard otherwise, 'cause there's no... You, you just can't coordinate effectively if you have, like, 10 different vendors trying to do all the same thing. You end up in, like, just constant, like, vendor meetings where you're, like, trying to decide what the specs are or something, instead of giving someone the authority or giving a team the authority to just go start building stuff. And then, you know, if you look at these companies, like, they have to implement these decentralized, somewhat decentralized processes when they get too complex. But at least they have, like, control over how they're interfacing with each other, you know? Like Walmart has vendors control their own stock, you know? They don't, like, tell the vendor, "We need X parts." It's just like, "It's on you to make sure your shelf is stocked."
- DPDwarkesh Patel
Yeah, yeah. So I, I, what was really interesting to me about this part of the post was... I don't know, I guess I had this vision of, um, or I, I had heard of this vision of where software is heading, where everybody will have a software as a service company and they'll all be interfacing with each other in s- uh, some sort of cycle where they're all just calling each other's APIs. And, um, and that, that like, yeah, basically everybody and their mother would have a, a SaaS company. Um, and it's, it's, uh, the implication here was, from your argument was t- given the necessity of integrating all this complexity vertically in a coherent way, then the w- winners in software should end up being a few big companies, right? That compete with each other, but still, uh...
- AVAustin Vernon
I think that's especially true when you're, like, talking about, like, you're combining bits and atoms, um, you know, li- maybe less true for, like, pure software. The physi- like, the physical world is just so much more complex and so the constraints it creates are pretty extreme, you know, compared to like, you can maybe get away with more of, like, everyone and their mom having an API in like a pure, pure software world.
- DPDwarkesh Patel
Right. Yeah. Yeah, I, I guess wh- uh, you might think that i- in the other kind of wor- e- even in the physical world, given that people really need to, um, focus on their, uh, comparative advantage, they would just try to outsource the software parts to these APIs. But is there any scenario where the learning curve for people who are not in the firm can be fast enough that they can keep up with the complexity? Because, you know, there's huge gains from specialization, um, and competition that go away if this is the world we're forced to live in. And then I, I guess we have a lot of counterexamples, or, or I, I guess we have a lot of examples of what you're talking about. Like, Apple is th- you know, the high, high- most, uh, the biggest market cap in the world, right? And famously they're super, uh, vertically integrated, um, and yeah, obviously they're, their thing is combining hardware and software. But, um, yeah, i- is there any world in which we can keep, uh, th- that kind of benefit but, uh, have it be within multiple firms?
- AVAustin Vernon
So this is like a post I've got on my list I want to write. The, the, the blockchain application I'm actually like... You know, which excites me personally the most is, um...... reimagining enterprise software (laughs) because, like, the things you're talking about, like hard typing and, like, APIs, it's just, like, basically built into some of these protocols. Um, but I think it, it just really has a lot of exciting implications for how much you can decentralize software development. And you can... But, you know, the thing is, you can still do that within the firm. So I think, uh, I think I mentioned this as like, you know, if the government's going to place, like, like all these, like, rules on the edge of the firm, like, it- it makes transactions with other firms expensive. Um, so if you... Internal transactions can be cheaper because they're avoiding, like, the, the government, you know, reporting and taxes and all that kind of stuff. So I think you'd have to think about how these technologies can reduce transaction costs overall and decentralize that, but also, what are the, the costs in between firms?
- DPDwarkesh Patel
Yeah, it's really interesting if there are... If the cost are logistic or if they're, um, if they're based on the knowledge that is housed, as you were talking about, you know, within, within the factory or something. Um, because if it is just, you know, uh, logistical and stuff, it's just like you had to report any outside transactions, then yeah, that does imply that a technology surely like blockchain could help. But if it's just that, yeah, you need to be in the same office, and if you're not, then you're going to have a hard time keeping up with what the new requirements for the API are. Then maybe it's that, yeah, maybe the inevitability is that you will have these big firms that are able to vertically integrate.
- AVAustin Vernon
Yeah, like, for these big firms to survive, they have to be, like, somewhat decentralized within them. So I think you have... You're, you're going to the same place as just, like, what, what does it... Like, you know, what's our fr-... Like, how are we viewing it? What's our perception, you know? So even if it's like a giant corporation, it's going to have, like, very independent business units, um, as opposed to, uh, you know, something like, you know, a 1950s corporation.
- DPDwarkesh Patel
Yeah. Byrne Hobart, by the way, has this really interesting post that you might enjoy reading when you're... While you're writing that post. Um, uh, it's like type safe communications, and it's about that Bezos thing, about how, uh, yeah, h- how his, his strict style for how to communicate and how little to communicate. Um, there, there's many examples in, uh, Amazon protocols where you have to... The only way you can, like, uh, put in this report is... In this place, you had to give a number. You can't just say this is very... Like, you got to say, like, "Well, we, we project a x% increase or whatever," so it has to be percent. Or, and y- you know, there's many other cases where the- they're strict about like what type definition you can have, something have in the... In written reports or something. And it has kind of the same consequence that, uh, type-strict languages have, which is that you can keep track of what the value is through the entire chain of the, the flow of control, so.
- AVAustin Vernon
You've got to keep, uh, work content standardized.
- DPDwarkesh Patel
(laughs) Um, so we've been hinting at the Coasean, uh, the Coasean analysis to this. I think we just talked about it indirectly, but yeah, for the people who might not know, um, the... Yeah, so the... Coase has this paper called The Theory of Firms, and he's trying to explain why is the case that we have firms at all. Like, why not just have everybody compete in the open market for employment, for anything? Like, uh, why do we have jobs? Why not just have... You, you can just like hire a secretary by the day or something. And the conclusion he comes to is that if you... By having a firm, you're reducing the transaction cost. So, you know, people will have the same knowledge about like what needs to get done. Um, you obviously are reducing the, uh, transaction cost of like the, the contracting, finding labor, um, blah, blah, blah. And so the conclusion he comes to is the more the transaction costs, uh, are reduced within people in a firm as compared to the transaction cost, uh, between different firms, the bigger firms will get. Um, and yeah, so I, I guess your, uh... That- that's why the implication of your argument was that there should be bigger tech firms, right?
- AVAustin Vernon
Yes, yes, definitely, because they can basically decrease the transaction costs faster within, and then even at the limit, you know, if you have large... You know, large transaction costs outside the firm between other firms that are artificially imposed, then it will make firms bigger.
- DPDwarkesh Patel
And then, so what does the world look like in that scenario? So will we just be like these Japanese companies, uh, these huge conglomerates who are just, uh... You, you rise through the ranks as... From the age of 20 until you die? Or how... Is that what software will turn into?
- AVAustin Vernon
Uh, you know, I- I- It could be. I mean, I think, I think it will be lots of very large companies unless, you know, there's some kind of change in, in the inter-firm transaction costs. And again, that could possibly come from blockchain-like technology. But you probably also need, you know, better regulation to make that cheaper. Um, and then you would have smaller firms. But again, I'm not... Uh, you know, in the end, it doesn't really matter. Like, you'd be like working in like your little unit of the big, big mega corp or whatever. So it, it may not... I don't know what that would look like, you know, like as like a personal level, but...
- 42:12 – 58:11
Car Manufacturing
- AVAustin Vernon
- DPDwarkesh Patel
Yeah, yeah. Um, okay, so speaking of these Japanese companies, let's talk about, uh, car manufacturing and, uh, um, everything involved there. Yeah, so yeah, I- I... We, we kind of, uh, hinted at a few elements of the Toyota way in lean production earlier, but do you, do you kind of want to give a brief overview of what that is, uh, so we can compare it to potentially other systems?
- AVAustin Vernon
You know, I think like all these kind of like lean Toyota process-like systems, they- they do have a lot of similarities and, you know, mostly you want to even out your production so you're producing very consistently and you want to...... you know, break it into small steps. You wanna limit the amount of inventory you have in your system so that there's... and when you do this, it makes it easy to see, like, how their process is running and limit defects and, and, you know, the, the ultimate is, you know, you're really trying to reduce defects because they are very expensive. That's maybe... it's, it's, it's a little bit hard to summarize. I think that's my best shot at it there, quickly, off the top of my head.
- DPDwarkesh Patel
Yeah. I, I... the, the, uh, the interesting thing about the, the Toyota system... so at least, um, when that... the machine was released, they talk about... that book was released I think in the '90s, um, and they went to, like, the history of Toyota. And one of the interesting things they talked about was, ther- there was a brief time where the company ran... I think it was... was this after World War II? Uh, but anyways, uh, th- they, you know, the company ran into some, uh, troubles. They needed to reduce, um... they needed to lay off people to not go bankrupt. They had much more debt on books than they had assets. And so yeah, th- they, they wanted to lay off people, but, um, uh, the... obviously, the (laughs) people were not happy about this, so there were, like, violent protests about this. And, um, in, in fact, I think the US... uh, the US written constitution, like, gave strong protections to labor that, um, they hadn't had before, so... which, which made it, um... which gave labor even a stronger hand here, uh, and so... Anyway, so the Toyota comes to this, uh, agreement with the unions that they'd be allowed to do this, like, one-time layoff to get the m- company on the right track, but afterwards, they could never lay somebody off. Um, and then so they... which would mean that, like, a person works at Toyota, works there from the time they graduate college or high school till they die, right? Um, and I, I, I don't know. Like, that's, that's, that's super intense in a culture, I mean, in software where you have average tenure at a company is, like, one year (laughs) th- the difference is so much. And, uh, there's, like, so many potential benefits here, I guess a lot of drawbacks too, but one is obvi- obviously if, if you're ta- talking in a timescale of 50 years rather than one year, the, um, the incentives are way more aligned between the company and the person 'cause like any, any, anything you could do in, like, one year is not gonna have a huge impact on your stock options, uh, in that, in that amount of time. But if you're planning on ho- if this company is your retirement plan, then you have a much stronger incentive to make sure that things at this company run well, uh, whi- which means, yeah, you're probably optimizing for the company's long-term, uh, cash flow yourself. Um, and also, yeah, there, there's obviously benefits to having that, um, uh, that knowledge build up in the firm from people who have been there for a long time. But yeah, that, that was an interesting difference, uh, one of the interesting differences, at least.
- AVAustin Vernon
I mean, I, I don't... I think there's, like, a diminishing returns to how long your tenure is gonna be. Like, maybe one year's too short, but there's a certain extent to where, you know, if, like, you grow faster than your role at the company, then it's time to switch and, you know, maybe that's like... it's gonna depend on the person, but maybe like five years is a good, a good number. And so if you're, if you're not getting promoted within the firm, then your human capital is being wasted 'cause you could go somewhere else and, and, and have more responsibility and perform better for them. Um, another interesting thing about, like, you... that story is almost all lean turnarounds, you know, or like where we're gonna implement something like Toyota production system, they come with no layoff promises because, um, you know, if you're gonna increase productivity, that's when everyone's like, "Oh gosh, I'm gonna get laid off." So instead, you just... you have to, uh, increase output and take more market share is what you do.
- DPDwarkesh Patel
(laughs) It's, uh... it's, it's like, um... it's kind of like burning your bridges, right? So you... this is the only way.
- AVAustin Vernon
You, you really... like, the process really requires, like, complete buy-in 'cause a lot of your ideas for how you're going to standardize work content come from your line workers, um, 'cause that's what they're doing every day. So you can't... if you don't have their buy-in, then it's going to fail. So that's why it's really necessary to have those kind of clauses.
- DPDwarkesh Patel
Yeah, yeah. That, that makes sense. Was it in your post or in the book where they talked about, um... No, I think it was in your post where you said if, if somebody makes their process more efficient and therefore they are getting, like, more work allotted to them, then obviously they're gonna stop doing that, right? So, um, (laughs) uh, w- which means that, I, I don't know. Do you had to give more dime time... downtime to your best workers or something, uh, or the people who are most creative in your company?
- AVAustin Vernon
I was just gonna say, like, you know, if you're, like, a, you know, worker at a plant, then... usually they have, like, small... a lot of times, like, for that level of employee, like, actually small rewards work pretty well, like, um... a lot of people used to, like, on drilling rigs, used to, like, if you met certain targets, like, give the guys, like, $100 Walmart gift cards. Um, so sometimes, like, small... to reward, you know, new ideas, stuff like that, it works. But because the whole system has to grow together, like, if you just improve, like, one part of the process, it doesn't... it may not help you. You know, you have to be, like, improving all the right process and stuff. So normally, it's much more collaborative. Like, there's some engineer that's looking at it and like, "All right, this is our... this is where we're struggling," or, "We have our defects here," and then you go get together with, like, you know, that supervisor and the workers in that area, then, you know, you all figure out, like, what improvements could be together. 'Cause usually the people already know. Like, this is like, you know, you see a problem at the top and you're just now realizing it, and then you go talk to the, the people doing the work and they're like, "Oh yeah, I tried to tell you about that, like, two weeks ago, man." (laughs) And then you figure out, you know, a better process for them.
- DPDwarkesh Patel
Based on your recommendation and Steven Molina's recommendation, I recently read, uh, The Goal. And after reading the book, I'm much more understanding of the value that consultants bring to companies potentially because before you could think, what does a 21-year-old, uh, who just graduated college, what do they know about manufacturing? Like, what are they going to tell this plant that they didn't already know? How could they possibly be adding value? And afterwards, it occurred to me that there, there are so many abstract concepts that are necessary to understand in order to p-... you know, like, uh, to be able to increase your throughput. And, uh, so now I, now I, I guess, I can see how, like, somebody who's generically smart but doesn't have that much industry knowledge might be able to contribute to a plan. Like, how, how, w- why, what value consultants could be bringing?
- AVAustin Vernon
Yeah. I think, I think there's, like, we get, you know, this applies to consultants or, like, young engineers, like, a lot of times you, you put young engineers, like, just right in the thick of it. Like, you know, on, like, working in production or process, like, right on the line where you're talking to the, you know, workers the most. And there, there's really two advantages. There's several advantages to that. One, the engineer learns faster 'cause they're, like, actually seeing the real process. And, um, the other is there's, there's, like, easy opportunities for them to still have, um, a positive impact on the business because there's just, like, $100 bills laying on the ground just from going up and talking to your workers and learning about stuff and figuring out problems they might be having and things like that, that could, that could help you lower costs. I think, I think there's a lot of consultants that, you know, I, I don't know how the industry goes, but I would guess there's, like... You know, I know Accenture has like 600,000 employees or something like... Or maybe I don't know if it's that many, but it's just a large number, and a lot are doing more basic tasks. And then, you know, the, there are some people that are doing, like, the high, more high-level stuff. There's probably a lot less.
- DPDwarkesh Patel
Yeah, yeah. There, yeah, there was a, there was a quote from one of those books that said, "At Toyota, we don't, like, consider you an engineer unless you need to wash your hands before you can have lunch." Um, (laughs) um, yeah. Okay, so, uh, in, in your, in your book about... Oh, sorry, (laughs) not your book. In your, in your blog post about the car manufacturing, you, you talked about Tesla and then, you know what was really interesting is that I, I... In a footnote, I think you mentioned that you bought Tesla stocks in 2014, which also might be interesting to talk about again when we go to the, uh, market, uh, and alpha part. But anyways, yeah, so it would... Uh, okay, so what, what... And then you talk about Tesla using something called metal manufacturing. So if, if you wanna... (laughs) First of all, like, how did you know in 2014 that, that Tesla was headed here? And then, yeah, what is metal manufacturing and how does this di- uh, differ from, uh, the Toyota production system?
- AVAustin Vernon
Yeah. So yeah (laughs) just, like, was goofing around and made that up. Someone actually emailed me and they were like, "Hey, like, what is this metal manufacturing? I want to learn more about this." And it's like, "Well, sorry, I just kind of like made that up." Um, so I thought it sounded funny. But yeah, I think, I think it's really the idea that there's, there's this guy, Deming. Yeah, W. Edwards Deming. And he, like, had a lot of, found a lot of the same ideas that Toyota ended up implementing and, like, they, you know... Toyota, you know, respected his ideas a lot. And, and America never really, except for the software industry recently, never really, like, got fully on board with this, uh, in manufacturing. And so this new... And of course, it's, like, software people that are, you know, coming and implementing this in manufacturing, and it's, like, the real American way of doing things. 'Cause when you look at, like, these manufacturing processes, like, the best place to save money and optimize is, like, before you ever build the process or the plant. It's, it's very early on. And so I think if there's, like, a criticism of Toyota, it's that they're, they're optimizing too late and they're not, like, creative enough in their production technology and stuff. They're very conservative and, like, you know, that's why they have, you know, hydrogen cars and not battery cars, even though they invent- you know, came out with the Prius, which was, like, the first, you know, large sales hybrid. Um, so yeah, I think this, this whole, like, what Tesla's doing with really just making Deming's ideas our own and really just, like, Americanizing it with, like... 'Cause, you know, like, "Oh, well, we want to cast this because that'd be easier. Well, we can't because we don't have an alloy. Well, we'll invent the alloy." You know? I love it. It's great. Mostly, I just like Tesla because they do such... Like, I, I agree with their, like, engineering principles and stuff like that. And so I didn't know that their, the company, would come to be so valuable. Um, it's just like I was just always reading their stock reports and stuff and, like, well, I at least need to buy some stocks so that I'm... so that I, you know, have a justification for spending all this time (laughs) reading their 10-Ks and stuff.
- DPDwarkesh Patel
I wanna get a little bit, uh, more in detail in the, the exact difference here. So, so a- so lean production, I guess, is yeah, they, they're able to produce their cars without defects and without, um, you know, matching demand or whatever. Um, and then so but wh- what is it about their system that prevents them from making the kinds of innovations that Tesla's able to make?
- AVAustin Vernon
It's just, it's just too, too incremental. It's, it's like, it's, it's so hard to get these processes working, so the faster you change things, like, it's, uh, it becomes very, very difficult to, like, change the whole system. So one of the, one of the advantages Tesla has is, well, if you're making electric cars, like, you have just a lot less parts, so that makes it easier. And then also they're like, you know, once you start, like, doing the really hard work of basically, like, digitizing, you know, like, stuff... Like, you know, they don't have speed limit dials, you start just removing parts from this, from the thing and, and you can actually then start increasing your rate of change even faster. And, uh, it makes it hard to get behind, you know, if you have these, like, old dinosaur processes. But some... I think there's someone, there's, like, a YouTube channel called The Limiting Factor and he actually went into, like, the, the detail of, like, numbers on what it costs for Tesla to do their gigacasting, which saves, like, tons of parts and deletes, like, you know, zillions of ro- thousands of robots from their process. Um, and if you already have, like, an existing stamping line and all that, where you're just changing the dyes based on your model, then, like, it doesn't make sense to switch to-... the casting. But if you're building new factories, like Tesla is, well, then it makes sense to do the casting, and you can build new factories very cheaply and comparatively and much easier. So there's a little bit of, like, you know, they have lots of... They just have lots of, like, technical data, I guess you could say, in a software sense.
- DPDwarkesh Patel
Yeah, that's super interesting. Uh, the, the analogy is actually quite... it, it's like, uh, well, Microsoft has probably tens of thousands of, uh, uh, software engineers who are just basically service its, servicing its technical debt and making sure that the old systems run properly. Whereas a new company like Tesla doesn't have to deal with that. The thing that's super mis-, uh, um, interesting about Tesla is like, it's, uh, wh- what is... Tesla's market cap is like way over a trillion, right? And then Toyota's is like 300 billion, and Tesla is such a new company. Like the fact that you have this, uh, Toyota which is like legendary for its production capacity and it's, uh, production system rather. And you, like this, this like company that's like less than two decades old is like worth many times more is... (laughs) it's, it's kinda funny.
- AVAustin Vernon
Yeah. I would, I would say that in that measure, I don't like market cap. You need to use enterprise value and the... when you start these comp- these old car companies have so much debt that if you look at enterprise value, the, uh, it's not so jarring. Like literally, you know, like I don't know. I can't remember what like GM's worth, like 40 billion or something, and then they have like $120 billion in debt. (laughs) It's like... so their enterprise value is, is like five times more than their, than their, uh, market cap.
- DPDwarkesh Patel
Uh, what, what is enterprise value?
- AVAustin Vernon
Enterprise value is basically like, what is the value of the actual company before, like, you have any claims on it? It's the market cap plus your debt. Simple, uh, the most simple. But basically, you know, if you're the equity holder, like, and the company gets sold, like, you have to pay the debt first. So you only get the value of what's leftover after the debt. So that's why market cap is... when Tesla has very little debt and a lot of market cap and then these other guys have a lot of debt with less market cap, excuse the, the comparison.
- DPDwarkesh Patel
Yeah. And then, uh, what, what... I mean, one of the interesting things, it's similar to your post on software, um, is that... yeah. It seems like one of the interesting themes across your work is automating processes often leads to decreased, um, decreased eventual throughput because you're probably adding capacity in a place that you're just adding excess capacity and you're also making the money-making part of your operation less efficient by have it, making, having to interface with this automated part. And it's, it's... it sounds like there's a similar story there with car manufacturing, right?
- AVAustin Vernon
Yeah. I think, I think if we tie it back into like what we were talking about earlier, automation promotes local optimization and premature optimization. So a lot of times, it's better to figure out, like, you know, instead of, like, automating a process to make a really hard-to-make part, you know, you should just figure out how to make that part easy to make. And then after you do that, then it may not even make sense to automate it anymore or, you know, get rid of it all together, then you just delete all those robots, so.
- DPDwarkesh Patel
Yeah, yeah. That's interesting.
- 58:11 – 1:17:25
C02 Electrolysis
- DPDwarkesh Patel
Okay, so let, let, let's talk about your, uh, let's talk about your other project, uh, that you're working on right now, the CO2 electrolysis. Um, do you wanna, do you wanna explain what this is and what, like what your current approach is, how... what, what, what is going on here?
- AVAustin Vernon
Yeah. So I think just overall, like electro fuels right now are like super underrated because you're about to get hopefully some very cheap electricity from like solar or, you know, it could be maybe some wind, possibly even, if we get really lucky, some nuclear geothermal. And it will make sense to make like liquid fuels or natural gas or something just from electricity and air, essentially. Um, so there's, there's many... there's like a whole spectrum of, of ways to do this. So, um, CO2 electrolysis is one of those in way- you know, it's basically you take water, electricity and CO2, um, and, and a catalyst, and then you make more complex molecules like carbon dio- or carbon monoxide or formic acid or ethylene or ethanol or methane or methane. Those are all options. Um, but it's, it's important to point out that (laughs) right now, I think if you added up all the CO2 electrolyzers in the world that they, you know, you'd be measuring their output in kilograms per day. And of course, like the products I just mentioned, we make millions of tons per day of, so there's like a, a massive scale up if it's gonna have, um, a wider impact. And, and so there's some debate. I think the debate for the whole electro fuels sector is how much are you gonna do in the electrolyzer? So one, one company that I really like their approach that is different than mine is Terraform Industries, and they wanna make methane, which is the main constituent of natural gas. But they're just making hydrogen in their electrolyzer and then they, you know, capture the CO2 and then put it into a methanation reaction. So everything they're doing is like already world scale basically, like, you know, we've had hydrogen electrolyzers power, um, you know, fertilizer plants without, you know... provide them with the hydrogen that they need. We've had... you know, methanation happens in like all ammonia plants and several other examples. It's well known, very old. Um, and methanation is like hydrogen and CO2 combined to make water, um, and methane. Yeah, so their approach is like the more conservative, but if you add... if you do more in the electrolyzer, like I'm gonna make the methane actually in the electrolyzer instead of adding this other process, you could potentially have a very... a much simpler process that has less CapEx and scales downward better. You don't need... traditional chemical engineering, like, heavily favor scaling. So-... with the more, like, Terraform processes, you know, their plan is, like, absolutely ginormous factories, you know? (laughs) These, these could take a long time to build, so like, one of the things they're doing is, um, you know, they're having to fight, like, the, the complexity that creeps into chemical engineering every f- every step of the way. Because if they don't, they'll end up with a plant that takes 10 years to build, and that's not their goal. Um, you know, like it takes 10 years to build a new refinery because they're so complex. Um, so yeah, so that's, like, kinda where I am. I'm like more on the speculative edge. Um, and it's not clear yet which products will be favorable for which approaches.
- DPDwarkesh Patel
Okay, yeah. And then you're, you're building this out of your garage, correct?
- AVAustin Vernon
(laughs) Yeah, yeah. So that's where, like, the electrolyzers... Everything with electric industry is, like, a flat plate instead of a vessel, so it scales down. So like I could have a pretty good idea of what my, you know, like, 100 square centimeter electrolyzer is gonna do if I make it quite a bit bigger. You know, I'd h- have to worry about, like, you know, how my flow might interact in the larger one and, you know, make sure the mixing's good. But it's, it's pretty straightforward, because you're just like making your flat plate a larger area. Whereas the, uh, you know, the scale, uh, it's different than scaling a traditional chemical process.
- DPDwarkesh Patel
I, I'm curious what, how cheap energy has to get before this is, um, this is, uh, e- efficient. And I, uh, uh, i- if you're turning it into methane or something like that, presumably for fuel, is the entire process, uh, energy positive? Or, uh, th- like, how cheap would energy, e- electricity need to get before that's the case?
- AVAustin Vernon
Uh, so yeah, so the different products and different methods have different crossovers. So like Terraform Industries, they're shooting for like $10 a kilowatt h- or megawatt hour, uh, for electricity. Um, but again, their process is, it's simpler, a little less efficient than... A lot of the other, like, products are a little like... also have, like, better premiums, like just worth more per ton, uh, than methane. So your, your crossover happens somewhere in between $10 and $20 a megawatt hour, which is... I mean, that's pretty... Right now, solar is maybe like 25. Maybe it's a little higher, because panel prices have gone up in the last year, but, you know, I think the expectation is they'll come back down. And so getting down to like 15, we start having crossovers for some of these products like ethanol or ethylene or methanol. Um, yeah, it's not, it's not science fiction.
- DPDwarkesh Patel
Yeah, I think, uh, in, in Texas where I live, the, um... That's where it's at, right? The cost of energy is like 20 or something, uh, dollars per megawatt hour?
- AVAustin Vernon
Well, (laughs) not this summer, but, uh, (laughs) yeah, r- recently. A lot of times in, in Texas, the, uh, the wholesale prices are around like 25 to 30.
- DPDwarkesh Patel
Gotcha. Um, okay. So a, a lot of the actual details you said about how this works went over my head, so what, what is a, what is a flat plate? Or I guess, e- before you answer that question, can, can you just generally describe the approach? Like, what, what is, uh, what is it, what you're doing to, for, to convert CO2 into these other compounds?
- AVAustin Vernon
Well, yeah, like, so it just... I mean, it literally just like looks like a, you know, an electrolyzer. You're like, you have two sides, an anode and a cathode, and they're just smushed together like this because, um, the electrical resistance. And if you put them far apart, it be- makes it... uses up a lot of energy. So they, you smush them together as close as you can, and then you're basically just like trading electrons back and forth. On one side, you're turning CO2 into a more complex molecule, and on the other ti- side, you're taking apart water. And so when you take apart the water, you, you know, it kinda like balances out the equation, balances out your electrons and everything like that. I need to, I probably need to work on that, uh, on that elevator pitch there, huh?
- DPDwarkesh Patel
(laughs) Um, I guess what the basic idea is you need to put electro- uh, you need to put like power in to convert, uh, CO2 into these other compounds.
- AVAustin Vernon
The inputs are electricity, water, and CO2, and the output is usually oxygen and like whatever chemical you're trying to create. It's along with some side reactions.
- DPDwarkesh Patel
And then these oth- uh, these chemicals you mentioned, I think, uh, ethanol, methane, formic acid, are, are, are these all just, uh, fuels or are they, uh... or wha- what are the other uses for them?
- AVAustin Vernon
Uh, so the idea, a lot of people are taking like a hybrid approach with carbon monoxide. So this would be like 12CO would be... They've raised a lot of money to do this, have like a hundred employees or something. Um, you can take that carbon monoxide and make hydrogen, and then you have syngas to make liquid fuels. So like they want to make all sorts of chemicals, but one of the main volume ones would be like jet fuel. Let's see. Formic acid is like a, it's like the small, it's the little, the small fry of all these. It is, um, like a additive in a lot of things, like, uh, preserving hay for animals, stuff like that. Um, then ethanol, you know, there's people that wanna like... There's like this company that makes, um, ethylene, which goes into pra- plat- uh, plastics. It makes like polyethylene, which is the most produced plastic. Or you can burn it like in your car, although I think ethanol is a terrible vehicle fuel. Um, but then you can also just make ethylene straight in the electrolyzer also. So there's kind of like a, there's many paths so, you know, which path wins is, is kind of like an interesting race to see.
- DPDwarkesh Patel
Yeah. The, um, the, the, the ability to produce jet fuel is really interesting because in your energy super abundance paper, you talk about, um, uh... You know, like you would think that even if, even if we can electrify everything in solar and wind and it becomes super cheap, that's not gonna have an impact on the prices to go to space, for example. But I, I don't know. If a process like this is possible, then it's like some way to, um... I, I, I guess in financial terms you could add things like add liquidity and then, like, turn basically this cheap, um, solar and wind into jet fuel through this indirect process so that like, uh, the, the price to send stuff to space or to, uh, ru- uh, I guess just, you know, have like, uh, cheap, uh, plane flights and whatever, all of that goes down as well.
- AVAustin Vernon
... it basically sets, like, a, a price ceiling on the price of oil, you know, and whatev- whatever you can produce this for is, like, the ceiling now. Um, which is, like, uh, uh, maybe the, the way I think about it.
- DPDwarkesh Patel
Yeah. Um, there's a lo- uh, do you want to talk a little bit of, like, how your background (laughs) led into this project? That y- this is your full-time thing, right? So we're... M- I don't know if I read about that, but, uh, w- well, what, what, w- where did you get this idea, and, like, how long have you been pursuing it, and, you know, what's the progress and so on?
- AVAustin Vernon
You know, I've always loved chemical engineering, and I love working at the big processing plant, 'cause yeah, it's like kid in a candy store. Like, I would just, like, m- you know, if I had extra time, I'd just, like, walk around and look at the plant. (laughs) Um, so thought it was so cool. But the, uh, like, the plant where I worked at, like, their u- uptime was, like, 99.7%. Like, it just... So like, if you wanted to change anything or do anything new, like, it terrified everyone, 'cause they're like... And you know, that's how they, like, earned their bonuses, was, like, run the plant, you know, 100% uptime all the time. Um, so that d- that w- just wasn't a good fit for me, and also, like, you know, so I thought, "Well, I always wanted, like, my own chemical plant." But you know, it's like billions of dollars to build plants, so that, it was, like, a pretty big step. So I think this new technology of like, you know, there's like a window where you might be able to build, like, smaller plants, you know, until it, it optimizes to be, you know, hard to enter again.
- DPDwarkesh Patel
Oh, and, and then while, why will it become hard to enter again? Uh, uh, what will happen?
- AVAustin Vernon
(laughs) Well, you know, if someone figures out how to build a really cheap electrolyzer, they, you know, just keep it as intellectual property, then, you know, it, it would be hard to, to rediscover that, you know, and compete with them.
- DPDwarkesh Patel
Uh, and then so how, how long has, have you been working on this?
- AVAustin Vernon
Uh, about, uh, not quite a year. But yeah, I actually thought... got this idea to work on it from writing my blog. So when I wrote the, uh, heating fuel post, I didn't really know much about... There's another company in this space, Prometheus Fuels. I'm like, "Oh, this is an interesting idea." And then I got talking to, um, a guy named Brian Hellegaman, and, uh, he's like, "You should, you should do this, but not like what Prometheus is doing." And so then I started looking to it, and I liked it. So, I've been working on it since.
- DPDwarkesh Patel
Yeah. It's, it's interesting because if energy does become a, as cheap as you suspect it might, and if, um, I- if this process works, then yeah, this is like a trillion dollar company probably, right? I- i- if you're gonna get the patents and everything.
- AVAustin Vernon
Uh, yeah. I mean, maybe. There's like... Th- uh, with chemical plants, there's like a certain limitation where, like, your physical limitations (laughs) . Like, you know, like you hit... There's only so many places that can have a good, like... or good places for chemical plants. Um, you start getting hit by like transportation and all that. So, like, you know, you can't, you can't just like produce all the chemical for the entire world in Texas and like transport it all around. It wou- it wouldn't work, so there you're talking about like a full globe-spanning thing, and then at that point, you know, if you're like building factories all over the world, someone's going to, you know, like figure out what your intellectual property is and all that. So, you'd have to, like, keep, keep innovating, you know, to stay ahead of the competitors, and I think that would limit your... You know, ultimately it's a commodity, so you're making commodities, so you don't have the same kind of defensibility that, um, you know, other sectors do.
- DPDwarkesh Patel
I see. Yeah, yeah, yeah. Okay. There, there's not, like, net- network effects, I guess.
- AVAustin Vernon
Yeah. So, so yeah. So not only, like, if you try to d- you know... So like what... You know, when I was talking about this is not quite consistent maybe as what I just said about, like, harder to enter, so you... But I think, like, what happens is, like, the scale starts increasing as you go on. So there are certain... Even though, like, this is easier to scale down, there's certain elements that are, are very much hard to scale, and then there are organization as well. So, but you only need a few competitors to... Basically, you'll end up with, like, early on, a few competitors that continue to grow against each other and limit the, the margins, and it hard... be hard to be like the fifth, you know, 30 years down the line.
- DPDwarkesh Patel
What is the state of this project right now? So are you guys planning on starting a company and, um, yeah. Like, what hu- what are the, what are the milestones you guys are shooting for?
- AVAustin Vernon
Right now it is just me, but um, you know, I have like a, a family of engineers. We're all engineers, so it's kind of like, you know, loosely supported, um, by... right now by, by other people. And my family to... as well, they're participating some. Um, but yeah. Basically, I just have to, like, get... You know, I've already done a lot of the theoretical design work at a, just like a very cursory level to make sure it makes sense and like, you know, the costs will be reasonable and stuff like that. So then right now, it's like working on the electrolyzer to basically meet the targets you need for like reliability and, and product concentration and, um, energy cost. And also then just like is it manufacturable? Because right now a lot of the electrolyzers like they use in the, in the labs, like they're literally smaller than a postage stamp and they're very difficult to make, so...
- DPDwarkesh Patel
Okay. I see. And had you started working on this, uh, before or after you had quit your job?
- AVAustin Vernon
Um, oh yeah, after. I, I quit my job like five years ago or something. I was doing like software stuff in between.
Episode duration: 2:24:12
Install uListen for AI-powered chat & search across the full episode — Get Full Transcript
Transcript of episode YSgfu7XprIM
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome