What It Takes To Build A Self-Sustaining City On Mars | Saurav Shroff, Starpath

27:08So, you know, we sort of have to take like a safe bet on every single one and spend a ton of money. We said, okay, let's just first start, let's just make a rover in the shortest possible amount of time. So the rover behind you, the one that's white and gold we call rover 10 and that's because we've made nine rovers before it um the first one was literally made out of 80 20 which is like commodity aluminum beam that you can buy from home depot mcmaster car wherever um and we built it in three weeks so you guys just went to home depot bought some parts and then bingo we went to home depot we bought some parts and we're like okay let's try to make okay out of all these requirements like okay it needs to survive cold temperature this that whatever let's ignore them all and start with the simplest set of requirements like let's make something with wheels that moves.

27:54And believe it or not, Rover, Rover one, you know, to say that it moved would be true to say that it moved well, would be a stretch, a little bit of a stretch. Right, right. Exactly. And then, and, and Rover one took us literally three weeks to build. So we built it in a garage basically in San Francisco where the company started. We spent three weeks and probably$10 ,000 or less. I don't actually remember on Rover one, Rover one pretty much didn't work. We looked at it and we made a document called problems on Rover one to be fixed on Rover two. And it was like a long list. We then said, okay, how quickly do we think we can make a Rover two that then actually fixes all these problems?

28:30And it turned out the answer to that question was something like five weeks. Okay, great. So now we have Rover two, Rover two, I'm now remembering also made out of 8020. Now it has some sort of custom sheet metal steel parts. uh it like kind of drives it's got a unique suspension uh we like learned a thing or two about like the balance of the wheels how much did the cost of rover one versus rover two what was the difference like a few hundred bucks or a few thousand dollars rovers one through five were all built in like under eight weeks each and all cost under fifteen thousand dollars i mean they're just like we're just going as fast as we possibly can to understand you know like okay what are like Like, okay, what shape works?

29:08Like, what material works? Like, if I put the wheel outside of the axis of steering, like, does the wheel bend inwards? Does the wheel bend outwards? And we were basically just on an exercise, like, how quickly can we learn about what it takes to make a robot and where the actual limits are? Then with Rover 7 onwards, we started actually trying to build a machine that would work in space. So, Rovers 8, 9, and 10 were all progressive steps in terms of, like, okay, Rover 8 worked in a vacuum, kind of. It probably didn't actually, but like it sort of theoretically might have. Rover 9 also sort of like a little bit better, like worked in a vacuum, was like sealed against dust.

29:44And now Rover 10 is our first rover that actually came out of what we call, what in the industry is called a thermal vacuum test, which is sort of like the graduation exam for a moon rover. So it's like a thermal vacuum test where you put a robot in a big chamber. The chamber simulates the conditions of various planets. our test was for the moon um and you evaluate whether or not uh well a doesn't break okay thank god rover 10 doesn't break um but also like you know is our model of like how it how it behaves thermally how it behaves mechanically correct yeah um and so from rover one to ten it's like two and a half years or whatever um you know cost us no more than well i don't know that we can share publicly i'll say that it cost it i'll say that cost at least a hundred times less or at least 10 times less than you'd think um and uh now we have like less than a million dollars for all of them or something um not quite but you're not far off okay it's like a few million dollars for all of them and now rover 10 within a few million dollars um is the lowest cost of production and the lowest cost of development rover to ever pass tvac to our knowledge um and what is tvac uh tvac is thermal vacuum test so it's like that it's like what i was talking about a few minutes ago it's a graduation exam for rover it's a big chamber simulates the conditions of a planet tvax stands for thermal vacuum and it's like the final test that you put a robot through to figure out like it doesn't work and now we're building rover 11 which is even better than rover 10 in a bunch of different ways i could get into a technical list it would take forever but it's better than rover 10 it's easier to produce it's more robust it's faster etc etc um and rover 11 will be finished uh early next year and so for for rovers 1 through 10 uh i know that you were testing them in like a thermal vacuum chamber were you also like going out and if aren't they going to just be like collecting a bunch of basically dirt and so are you like going out to a desert to you know have them collect dirt how did you test them yeah it's a great question so actually right where you're sitting if you look behind you you can see a bunch of scuff marks in the concrete that's because right here there used to be a 25 ton pit of semi-cemented concrete so it's like it was like semi-soft concrete um and you know like you know we would drive the rover into the pit uh you know mine stuff like evaluate how well it mines stuff eventually we got rid of the pit because it was sort of like a safety hazard like you know every time the rover mines stuff it like releases huge clouds of dust um that are like you know i would say lifespan shortening uh to breathe in and so we removed it um and now we do sort of like more scientific testing now that we've got the mining portion down um for rover 10 we actually put a slab of that same concrete in a thermal vacuum chamber so it was like not just mining but it was mining while super cold while being like you know while in a vacuum uh basically under the simulated conditions of actually mining on the moon um and so i mean and oh my god this is a crazy question back during the rover one days we had uh literally a sandbox uh in our office Like people would come over and be like, oh, is this like the playpen for engineers?

32:42Like as a joke, it was not a playpen for engineers. No, no, it was not. So gross. I do not recommend a sandpit in your office. But we would use that to test like, okay, driving capability, mining capability, super, super crude. Like, okay, does this shape work? Does this shape not work? And, you know, like really, really basic stuff like that. But now we sort of elevated to the scientific level where we like calibrate the strength of the material, put it in a vacuum. It's arguably more boring nowadays, but you have to do the science to get it right. So how do you get like the best simulated soil composition of marsh and soil on Earth?

33:16Yeah, I mean, this is like a super long question to answer. The short answer is you make some guesses and you just take the worst possible case. There's like, you know, between 50 and 100 research papers on this. And they're all like kind of the same and that they're like, OK, here's our guess as to what it might be. um this is like the range of values we expect where the range is like you know one on the bottom and like 75 on the top well they've taken like martian soil samples so we kind of know what it is right right yeah so generally we know a little bit about mars from nasa's uh from nasa's missions to mars um and uh we know that uh we know that the martian soil is at least probably softer than soil on the moon and so really really what we've done is we've taken like okay we've read all the literature.

34:03We know that there are like 700 possible configurations of dirt. And just like, okay, what's the worst for our equipment? The worst for our equipment is, you know, if the dirt is what's called subangular, meaning that the actual, you know, pieces of dirt are sharp at a microscopic level and that they're sort of like cemented and frozen solid. So what we do when we test is we just buy simulated subangular dirt, freeze it with liquid nitrogen with a calibrated amount of water in it to get it to be that super hard material. Now, all of our engineers think that in the real world, when we actually go to the moon and Mars, we're going to find that it's like pretty chill, way softer.

34:38And yeah, pretty chill, as you're saying. But the only right path forward from a like you can't engineer around a guess. You have to engineer around like, OK, the worst case is this. So we're going to build a machine that can handle that. And, you know, it's not so bad. Like like mining equipment is we might we mine harder and nastier stuff on earth all day every day uh you know for like lithium mining iron ore mining i mean everything um so it's not like there isn't a you know knowledge base on it you know among the human civilization of like how you dig up hard dirt let's run a hypothetical scenario in a hypothetical world where spacex does not exist and the capability of like reusable rockets doesn't exist would you be trying to build that is that something i personally be trying to build reusable rockets if SpaceX never existed.

35:29Correct. I think the real answer to that question, like I would love to say yes, like, oh yeah, I would do anything to make Mars happen. I think the real answer to that question is that like SpaceX is the reason why people believe that challenging things in aerospace are possible to do in the first place. Like if you take SpaceX out of the equation and you ask yourself, like, what does the aerospace industry look like? it looks kind of like a lot of other industries where you sort of assume the innovation has dried up like it looks i mean i don't know much about oil and gas but it looks like what i think of as oil and gas which is like the same thing over and over where the companies have probably done what can be done to make it more efficient already and it's kind of commoditized across every single competitor right i mean like that's i don't know much about oil and gas to be very clear but that's what i think of when i think of oil and gas and i think of i think sans spacex the aerospace industry looks almost the same.

36:23And so my real answer to your question is sort of a non-answer, which is I think sans SpaceX, you probably wouldn't even have, like you probably interview five aerospace companies a month. I don't think any of them would exist sans SpaceX, not just for the economic reason, not just for the economic reason that like, oh, they wouldn't be able to launch their stuff into space for a price that they can afford, but also for the reason that like SpaceX and the things that they've done have made other people in the world think like, okay, if you put your mind to it and you hire the best talent and you cut down on fluff and you cut down on bureaucracy, that you can actually produce an outsized return in terms of technology developed per dollar.

37:03I won't name names, but other companies are not the most confidence inspiring in. It's not like you look at some of the programs that you see in the rest of the industry and think to yourself, oh yeah, if I copy that i'm gonna be able to change the world right um but if you just copy you know starlink not for starlink but like you know you make a starlink for something else low cost high rate you know serves a big market you could see a world where a new technology like reflect you're talking about changes the world what's been the biggest like technical hurdle or technical challenge uh so far i mean there are a lot of sort of disjoint technical challenges I would say the biggest, you know, there's no particular biggest.

37:46My team will be mad at me if I cite one thing that's the biggest. The thing that I would say is probably the biggest is actually a broader problem with the US, which is the access to low cost, high rate manufacturing, particularly when you make real goods. The difficulty of making that good is sort of like a risk adjusted analysis of how good a thing is when you're designing it against how quickly and frequently you want to buy it. So like if you're designing, I'm just going to give you a simple example. If you're designing like a motor assembly for a robot, and this is fresh in my mind because last night I was watching an engineer design a motor assembly for a robot.

38:24You know, if the motor assembly cost$10 to manufacture, which it theoretically absolutely could if you fully automate the manufacturing process, well, not the whole assembly, but like it could cost like$100. Okay. Most of it could cost 10. Yeah. If it costs$10, you'll just buy one every day. You'll just like wherever you are at 8 p.m., you'll just buy it. In fact, if in the future, you know, manufacturing was so inexpensive, you know, that like you could make like an assembly like the one I'm talking about for$10, you would just have an optimist robot and a software program that automatically buys your design every day and assembles it the next day.

39:03So that whatever is wrong with the design, you immediately know the morning after. Right. But in the real world, you could have a mistake in your design that you might find out if you're good a month after you made it or a week after you made it. If you're really good, maybe a day, but in a lot of cases, like six months. And that's because if the design corresponds to a machine that costs$50 ,000,$200 ,000 or an old aerospace like$5 million to make, you're not going to make a lot of them. And instead, what you're going to do is analyze the shit out of it. and you're just going to be extremely careful and extremely slow at the design stage.

39:41And so what I would say is like, you know, this is kind of a cop-out answer because it's not one specific big technical hurdle. The biggest technical hurdle broadly is that the ability to fabricate the things that we design is not next to free. And if it was next to free, which I totally think it will be in, you know, five, six or seven years, the rate at which we could iterate would be, you know, at least double, maybe triple. The cost of our goods would be maybe 50 % better. And we'd be a heck of a lot cheaper at manufacturing or sorry, developing new technologies if we can manufacture them quickly at scale.

40:14You've been very clear and thoughtful about trying to make it so your iteration loops are super, super tight. And you're also like just trying to go through iterations, you know, make the robot in the first place, see what went wrong, see what we need to improve, do it again, and then just run through those iteration loops. How did you kind of come to that philosophy? And then what are you doing to, as the company scales, keep that sort of same philosophy? Two great questions. So I would say that, you know, we are not the inventor of the idea that iterating fast makes your end product better, cheaper, more quickly.

40:49You know, there are probably like 20 amazing companies out there that can be pointed to as the source of inspiration for this being a good idea. And now maybe like 200, right? If you count the new companies that do the same thing. So we by no means are the proprietor of this idea. We are just astute observers in looking at the fact that the companies with the highest market cap, the best products, the most talented engineers, the best ability to hire those engineers, the most compelling products, I don't know if I said that already, are the ones that iterate quickly. And you can name some names.

41:27um what we do and the second part of your question is is is also super important which is like okay when you grow when you grow how do you make sure that a hundred person org can also turn you know a design once every week or once every two weeks if needed um the answer lies mostly in making sure that the responsibility of your engineer is really clearly defined if it's the case for example and this is the case at a lot of companies that you have a product, the product has a product manager and, you know, the team has like six people and, or whatever number of people, it doesn't really matter.

42:02And each, you know, for each person on the team, you know, one person has a, has their hand in a thing, right? Then to go fabricate that item and test it out, you now sort of need like the either implicit or explicit approval of seven people. If instead what you do, and this is the model that a lot of, in my opinion, really great companies use. If instead what you do is you say, okay, I have a growing team, yes, but I'm not going to put more people on one thing. I will just make sure that the scope that each person is working on shrinks and the expected quality grows. Then it can be the case that like, okay, even though there's a 700 person team, you know, this motor assembly or this, you know, hinge assembly or this computer or this camera is actually the delivery of that product is only up to just one person.

42:50And if that one person is responsible for delivering it and is equipped with the money and the tools and the in-house machining and the in-house production to prototype fabricate that part quickly, then your iteration speed will be exactly the same. And then the second component is like culture, which is like, okay, well, you know, does your team like three months as an iteration cycle or does your team care about making the iteration cycle as small as possible? And both of those things are important for us. How big is your team, by the way, now? The team's about 16 people. We're growing as fast as we can find good people.

43:25Yeah. So if you're watching this, please apply. Yeah. And how old is the company? Is it like four years old? Yeah, the company's about three and a half years old. We started out super small, just the founders. Like I said, effectively a garage in San Francisco. I mean, it was like a really, really, really shitty place. like a rattly garage door every time someone drives by it's super loud and uh the the basic thing the basic thing for star path is uh we want to be we want to be ready to launch a huge amount of payload to mars that can produce on the order of low megawatts of power which roughly speaking is enough for a civilization of a few hundred people um right when the rocket's ready and so when we started the company we thought that date was in like two years yeah um we were wrong um was that actually positive that you were wrong and because you had an extra few years that was easier to basically work out kinks and stuff i mean i would say it's actually it's neither we've just we've just we've just calibrated the growth rate of the company according to when we think the rocket's gonna be gonna be ready it's sort of like a moving target and um from our standpoint it's way better to do that than it is to uh start developing the technology when we're certain that the rocket will be ready because at that point we'll be too late like the the real value of star path as measured to humanity is the date that a self-sustaining city on mars forms that can happen yeah and then like with or without star path so so without star path you might say like okay well it'll happen at some date our our goal is to make that as soon as possible you know to say that we think it can be done in 10 years would be i think a bit ambitious you can get a long ways in 10 years um and so so waiting an extra for context for those watching you can only fly to mars once every two years so so there's a window in 2026 there's a window in 28 and 2030 in 2032 well the 2032 one spills over in 2033 but you get the gist and so waiting basically missing a window is a huge deal yeah it's a huge deal it's missing two years where during those two years you could have let's say launched a mission gather data use the data to fix or improve the products and then scale production of those products for like a 10X bigger deployment the following year.

45:33So it's like, if you wait two years, you could conceivably be 10 times behind where you wanna be. If your plan, which ours is, is to grow 10 times window over window for the first few windows. So in 2030, we could be where we would be in 2028. Hypothetically, if 2024 it happened. Yeah, yeah, exactly. So it's like, you don't wanna miss a window. So from our standpoint, we're basically just like a cheetah hunting its prey. where the prey is the rocket launch and and we are the cheetah and the movement of the cheetah represents technology it's kind of a bad analogy but yeah we're tracking a moving target so when we started the company three and a half years ago we thought okay we're gonna be launching super soon there have been some benefits in going slower but you know my models liked it to i would have liked it to be faster which okay everybody would have um and uh and uh you know i i also will say that you know our opportunity like the opportunity for making a really really really big splash on the first mars window um is there now um and it wouldn't have been the case if the if the rocket launched sooner because we wouldn't have been able to build as much has has spacex said publicly how many starships they're trying to launch on the first mission yeah so you want to said publicly and like i said i'm not a spokesperson for spacex so like reference their website um elon has said publicly that they want to send a few test ships to mars in 2026 which i'm optimistic about uh and if you have like three or ten i don't know if there's a number i think like five or something hopefully it's enough yeah hopefully it's enough but no cargo so just just like in to try to get the ship to land it's extremely challenging problem to get a ship to land on mars it's like it's like hitting a bullet that's moving um we're like the bullet is like the size of like a you know like a it's extremely small in the context of you know how far you're traveling so it's it's very challenging we've done it before we've also missed before like mars missions in the past do not have a hundred percent success rate really even hitting the planet mars yeah um so and what what do you know what went into the misses like why that happened yeah yeah it's so straightforward so so basically when you when you're trying to get to mars what you're doing is you're slingshotting around the sun and you're slingshotting around the sun to try to catch mars on the other side of the sun from where you launch so if you launch let's say this is the sun and earth is moving around like this and mars is moving around in a bigger circle we'll just call them circles for simplicity when you when you go to mars what you're trying to do is slingshot around the sun to catch mars halfway around the other side of the sun um so like roughly speaking i mean this is not exactly true but like if you leave in the summer you'll and you'll get there in the winter it's it's really not true but for simplicity we'll say that it's true um and if you make your curve ever so slightly wrong you will like not only will you overshoot it by overshoot or undershoot insane yeah and that and that and that distance can be so big that you know maybe you get in mars you know like millions of miles yeah it could be like a few hundred thousand miles and if and if you miss you know you could like slingshot around mars which would be terrible if there was crew on board because they would be certified dead um it's a grim take and you know i'm optimistic about developing technology to make sure that doesn't happen but or if you undershoot uh you may sort of like you know wrap around mars and actually actually leave like go out of the solar system so there's like a lot of ways you can miss and the only way that you can't miss is if you hit the atmosphere because once you hit the atmosphere then you can use aerodynamic controls and land.

48:49No, not like an airplane, but the atmosphere can be gripped. But you can land at least. The atmosphere can be gripped, so to speak, and you can land. So 2026, SpaceX wants to send experimental vehicles and I'm really hopeful that they succeed. You know, there's been some recent successes that are really awesome. And then in 2028, SpaceX says that they want to send, you know, an even larger fleet of experimental vehicles. And then in 2030, they're saying they want to launch a large fleet of vehicles. And this is all public information, by the way. So yeah, you can look this up and find this information.

49:18But yeah, Elon's been very clear that he really wants to basically scale the number of starships heading to Mars, like maybe even by more than an order of magnitude every single time, because that's the only way. Like I think he said something like it needs to be over a thousand starships going in order to make life multi-planetary and like a self-sustaining city. Yeah. So we can just work out the math roughly live. the approximate number of tonnage of tons that you need to get to the surface of mars from earth to make the to make a city that's self-sustaining is about a million there's like some debate that that number might be a hundred thousand or ten million or whatever but nobody thinks it's like you don't want to undershoot you don't want to shoot first of all and nobody thinks it's like drastically different so a million tons to the surface of mars is what it takes to make life multi-planetary roughly speaking if you want it to be permanent like okay the city will not die even if the ships stop coming from Earth for any reason.

50:09Each starship can carry thereabouts 100 tons to the surface of Mars per flight. So what this means is that you need to have, if you had a fleet of 1 ,000 that could carry 100 tons each, you can now carry 100 ,000 tons per fleet trip, meaning each time the fleet goes and comes from Mars. And then if the fleet makes 10 trips, then 10 trips of a fleet of 1 ,000 ships will be a million tons, which is a self-sustaining city. You know, I actually have some optimism that that number will actually be a lot bigger. I mean, StarPath's business, of course, is not modeled around, you know, there being a 10 ,000 or 100 ,000 size fleet of starships.

50:49But I do have some optimism that it might either be a bigger fleet in number or a bigger fleet in terms of the size per ship. Because, you know, it's my opinion, like once you have a business that works on Mars, like once you have it, like once you have the cost of transport, below the threshold price that people want to pay. That million ton city roughly correlates to a million people. I think you could sell way more than a million people on going to Mars. Especially if it's only like, you know, you have a minimum, you have to stay there for, let's say like four to six years, and then you can come home if you want, or like some number of people can come home at that point.

51:24Right. I mean, first of all, I'd say like, you probably want to make sure that your supply of transit from Mars back to Earth is bigger than your demand like the last thing you want is for people to feel trapped because nobody will go somewhere that they that they think they'll feel trapped so you always want to make sure that the ships coming from mars back to earth when there's crew on mars uh are greater than the number of people that like you should have empty seats you know obviously from business standpoint you don't have empty you just want to basically like fill it with cargo or something yeah yeah you don't you don't want to have it be the case that that that somebody who wants to do something can't because now you go from the greatest adventure of all time to like the greatest prison yeah the greatest prison of all time, which like, it's our responsibility as humans, as friends of each other to not imprison our friends.

52:06Okay. So relatively straightforward, but, you know, going back to my comment on scale, like, you know, I totally understand Elon's take that, okay, you want to have a million people to have a self-sustaining city and roughly speaking, agree with it. That being said, when airplanes were made, people said, okay, airplanes are extremely costly. And, you know, they're going to be used in these niche circumstances where like you really need to get from one place to another or maybe have a lot of money. And now airplanes are less expensive. Energy is a little bit cheaper. I think I forgot exactly how many, but it's like over 100 million annual passengers or more than that.

52:41Yeah. So there's hundreds of millions of passengers per year. Now, obviously, flying on an airplane is very different than flying to Mars. But my basic point here is just that it's very hard to predict like a million is a number that's based on what's required to make life multi-planetary. But if you think about it from a business and demand standpoint, if you shoot for a million, you're actually way more likely to overshoot than undershoot because a million means that you've built a compelling product that is broadly liked by people who buy it. And so if a product is broadly liked by people who buy it, you make it better over time.

53:11There's no reason why you couldn't sell 10, I think 100 is probably your limit where it's like, okay, at 100 million people, you're now looking at a a meaningful portion of this of the species right like over one percent um but uh yeah so like you know i i have i have like you know hopeful aspiration um i'd say i'm cautiously optimistic that the future actually has way more than a thousand starship size vehicles or or maybe a thousand but they're way bigger yeah and like by future you mean like near term i would i would assume that eventually you'd want like tens of thousands every launch window yeah i mean there's really like no reason to not have an insane amount it's super challenging yeah by near term i mean like 20 years um it's super challenging to predict like what happens in 20 years if i could predict what happens in 20 years i would just go to the you know i would go to polymarket and just you know become a billionaire um as in i would just you know make some bets on uh you know 0.0001 probability and then like pay off 10.1 sort of thing um so it's hard to predict but but from first principle standpoint, it's like what the challenge we're looking at is, okay, how do we make a product, which is transit to Mars, including obviously return journey, what you do when you're there, and everything in between food, water, everything.

54:26That's so compelling that people actually want to buy it. And obviously, we're starting by building the thing that's the most important, which is the propellant production plant. But when you look at that problem, it's not obvious that you couldn't solve it you you couldn't close that for five or ten or twenty million people um you know it there's if you could close a hundred thousand there's no reason you couldn't close 10 million um so i'm definitely optimistic that'll be more um you know time will tell ironically like on making starship uh rapidly reusable and like going from mars to earth it seems like the propellant is a huge deal but then actually if it takes like 95 percent of the energy that's going to be used on mars just in the propellant making process the hugest big thing is just making enough energy um and so for you on making solar panels how are you thinking about like really ramping the scale of of manufacturing these solar panels so that you can kind of make all that energy yeah okay so a totally totally correct input assumption that you need to make a shit ton of energy on mars uh to make a city there um the approximate number is about 100 kilowatts per person it's like 10 to 100 kilowatts per person can you put that in like context like the average house is like two to five maybe sometimes 10 kilowatts like a big a big house with air conditioning is 10 kilowatts um you know small house with no air conditioning is like two kilowatts um when i was growing up my house pulled an average of 1.3 um that's because my family's like super interested in so it's going to be like 50 to 70 times average consumption on earth on mars yeah but i gave you the per household number on mars and i gave you the sorry on earth and the per person number on mars so it's actually like it's actually more like 200 times 200 times like 200 times the power per person because most of that energy is spent on powering the plant that makes the fuel for that person that they will eventually use when they go home and i mean the rockets are like the size of a swimming pool so you're talking about like swimming in rocket propellant, just to give you a sense of scale.

56:27You need to make a lot of rocket propellant. Yes, exactly. So going back to your question, how do you plan on scaling solar power production? You're definitely right on the input that you need to make a lot of energy. And that's, by the way, the key, sort of just for context for those listening, StarPath just announced a commercially available solar product for satellite companies. So we make our own solar panels now. And if you're a satellite company, you can buy them for about a tenth of the cost that they could be otherwise purchased from existing vendors. We did not intend to enter this marketplace, but the reason we entered this marketplace is because of what you're saying, which is that we needed an extremely scalable - The biggest constraint.

57:05Yeah, it's one of the biggest constraints. I mean, I would say technology is the biggest constraint, but from a manufacturing standpoint, the good that we need to produce the sheer biggest quantity of measured in count is solar cells. We need like 100 million solar cells on the surface of Mars to make even a moderately sized city. Let's actually put that into context as well. How many solar cells do we have on Earth? Oh boy, I actually don't know the answer to that question. But a lot. Okay, so the approximate, the unit of measurement for solar on Earth is watts. And I believe that the total production of solar modules in the US is on the order of about 10 gigawatts per year.

57:42I could be wrong. Okay, don't quote me on that. The total production of solar modules for this is a better question all space applications combined is about 40 megawatts 50 megawatts about about 50 million watts is all of the power that's used on every spacecraft launched from earth today out of which the majority of that is starlink and that's like a rough number you might as well just call it 100 i mean i think starlink is already like over 90 of all satellites in orbit right yeah so starlink's starlink's majority obviously they account for the majority of the power. But just to give you a sense of scale, all spacecraft that orbit Earth, every year, the spacecraft that are manufactured, it takes about 40 million watts of solar panels to power those.

58:23For Mars, you want to have, you're not going to believe me, about 40 gigawatts of total power. 40 gigawatts. So 1 ,000 times more power. Now, out of that 40 gigawatts, we'll probably produce about four using solar energy. So, or yeah, so about a 10th. So about a hundred times more power than all of satellites combined. And then past four gigawatts, there's roughly speaking a threshold at around, it's actually less than four gigawatts, but there's a threshold around a gigawatt where it becomes more cost-effective to develop a nuclear reactor. And so, and so early on, we'll power Mars using solar cells.

59:01And that's the product that we produce in the biggest single quantity today. It's not necessarily the most important product we build, I would say that that's probably not true, but it's the product that we produce in the biggest scale. And we have a mostly automated production line that makes those cells at a pretty magnificent rate. So we can make millions of watts per year today. And in the future, we're going to scale that all the way up to maybe 400 or 500 megawatts. And then that sort of 400 or 500 megawatts per year production rate will level it out. And the remainder of the growth for the city on mars will actually be powered by nuclear um and that's because nuclear gives you a couple beneficial traits the first is that your total power produced per mass is a little bit better so you're able to cram more power onto a rocket which as you're trying to scale the power up a lot you're going to want are you talking about like just shipping uranium basically exactly yeah so if you just well you would ship early on a complete nuclear reactor um but that's fucking nuts you can ship a nuclear reactor on starships well to be clear you you can't yet like there's a huge regulatory over uh you know there's a huge regulatory you know burden it's also it's also the technology doesn't exist yet um so you there doesn't exist for example 10 megawatt nuclear reactor i can buy today which i would if i could and then put it on a start like a small modular doesn't exist yet yeah i mean people are making it there are a lot of great companies actually in this area that are working on nuclear reactors some for space um and so that technology will come around um you know it's unclear yet whether starpath will be in the manufacturing business of that product or the buying business of that product.

1:00:34But the advantage of solar in the early days, the reason why we use it is because the cost and schedule are better. The cost to get a solar panel that works is extremely low compared to a nuclear reactor. And we can do it today. In fact, we've done it today. And there are no rules that say you cannot launch a solar panel to space. In fact, a solar panel is launched to space four times a week on a Falcon 9. Every Starlink satellite has a solar panel. Every satellite that goes to space has a solar panel. So it's really important. Are they launching four times a week now? uh something like that it might be like three times a week so maybe i'm exaggerating but yeah they're about they're about like a you know between 100 and 200 falcon 9 launches very like proven like this is a known thing people do it all the time extremely known extremely known yeah um and so um yeah so so how do we scale well the production is automated automated production is easy to scale you just buy more lines so if you have a line that produces some amount and you want to make 10 times you just buy 10 lines you just factory factorial it and just basically copy paced exactly you factorial the production of solar panels um which is something that we're actively working on right now and uh in the long term you'll factorial the production of nuclear reactors and you know there's some great companies out there that i recommend you look at that that are working on that uh for earth applications um and uh and so yeah scaling the production of solar panels actually not the most challenging thing in the world the more challenging part which is the problem that we're solving today is dialing in one production line or in our case two um that can make uh you know solar panels well um and then once you have a production line that works copy and pasting it like you said factorial style is not the challenge there's some making the first one is that yeah making the first one's the hard part we're mostly there ironically i think i think elon basically says like uh prototyping is easy production is hard so you're kind of thinking that actually it's just going to be um prototyping is hard production is going to be relatively easy relative to that?

1:02:23Well, I mean, I would call the creation of one production line production. And that is hard. And that is hard. And that is hard. So I would agree with Elon that production is hard. Prototyping is easy. You want to make one solar panel with your hands. That's a pretty doable thing. It's doable. It's a pain in the ass. I wouldn't recommend it, but it's doable. And then, you know, if you want to make a million, you need a machine. Yep. Okay. So what do you think the biggest risks are for your business? Like making it actually work? Yeah. I mean, there are like sort of two categories of risk. And I basically price one of those as zero and the other one as being a combined challenge of both.

1:03:02So the first challenge is that, can you make technology that works? And I price this at zero because, as in zero risk, because even if you can't make technology that works, it's just a question of how many tries you need. We think it'll take us two or three tries. We could be wrong. And even if we're wrong, the business will still work. um and uh okay so that that's one category of risk the second category of risk which i think is far more important to understand and evaluate is okay the business explodes it's worth 50 100 billion dollars or more if people want to go to mars and you've built a product which is a transportation service a return service and housing etc when you're there that's compelling enough that people want to pay for it.

1:03:48And this is a much more important question to ask because it really will lead you down the path of developing the correct technologies that people want to actually buy. And so the biggest risk of them all is that even if you make a fully and rapidly reusable rocket, a low cost rocket propellant production plant, a habitat that's beautiful and delightful to live in, and a food production system that makes delicious food and all the other things that would make the best case outcome for everything. Make it sort of a Garden of Eden yeah even the risk the biggest risk for star path is that even if you do that the value proposition to customers to travel to mars for 100 200 300 400 000 in the long term okay the prices early on will not be that is is not a good enough value proposition to convince people to go um and if that's the case um then uh you know start start path won't work but that is not a reason not to try right like to say i'm not going to try to build a city on mars because i don't think people will like it is like the equivalent of saying, you know, I will not build, you know, abundant electricity because I don't think that the needs, the power needs should go up.

1:04:53Yeah. I think, I think the, the first pitch for people going to Mars is roughly similar to like, do you want to be a Navy SEAL? And like, do you want to go through that program to try to prove that you are like the best and push yourself to the limits? Well, I mean, I, I, I think the Navy SEALs are incredible. So I wouldn't say that like, I would, you know, if I was going to Mars, I would not compare myself to a Navy SEAL. I would put a Navy SEAL above me. But I think the pitch is similar to that, which is like, do you want to do, do you want to embark on what is almost certainly to date the greatest adventure that any human has ever done, despite for millennia humans dreaming about an adventure as epic as this?

1:05:31Do you want to do that? And I think we can get, you know, for the first few ships of hundreds or thousands, not hundreds and not hundreds of thousands, but hundreds or thousands of people, like that pitch will work. And so I'm completely not worried about the first 10 ships or the first 100 ships. What I'm worried about is how do we make a product that is so compelling in seven years time, where once you're past or eight years time or 10 years time, or once you're past that initial few ships, you know, I say few, it's still like a lot. But once the product goes from completely novel to less novel, how do you still make it the case that uh people want to go to mars for the prices that you can sell a round trip ticket to them for yeah i think the novelty will stick around for a while i don't know if it would ever go away honestly like if you're right i hope you're right i think well i i think for even if even if you said um only novelty for the first million people to go to mars that's still how many you know in perspective of number of humans alive that's like it's like under 0.001 it's under Yeah, exactly.

1:06:37And I think it will be a novelty. It's actually, sorry, it's just over a tenth. No, it's under a tenth of a percent, my mistake. Yeah, yeah. Yeah, it's not clear. It's not clear to me. But from our perspective, it's very simple. It's like, okay, start with the things, start with the basics that matter the most, energy production, propellant production, which by the way, we haven't talked about this yet, but propellant production also implies the production of water and oxygen for drinking and breathing. So it's like, because - That's just part of the process. Yeah, it's just part of the process.

1:07:03So like oxygen as a liquid is fuel for a rocket. If you turn a tiny bit of that into gas you now have oxygen for breathing and then of course we talked about how water is a pre-product of oxygen so if you just take a little bit of water out of the system before you turn it into oxygen um that's for humans that's for humans to drink it's a negligible quantity like you know you will not you will not out drink a rocket engine so in in the process of making the rocket fuel you basically make way more water than you could ever need for humans yeah so what we say but the sort of like the sort of like line that we repeat is by produce by creating a rocket propellant production plant you create a virtually unlimited supply of water for drinking and oxygen for breathing basically like use as much damn oxygen as you want you can have a leak i mean you'd have other problems if you had a leak in your habitat but um you know from a from a production standpoint you could have a leak in your habitat and it wouldn't be a big it wouldn't be a big deal um but yeah from our standpoint simple it's like okay start with the basics the thing that matter things that matter the most prop production life support power From there, four years, six years down the road, start answering the question of how do I make sure that this business continues to have a compelling product offering in the worst case that the novelty does wear off?

1:08:12Like, okay, you and I hope the novelty never wears off. People are eager to go once they realize it's accessible and it's safe. And they're not gonna die. And they're not gonna die, right. There'll be some people who are willing to go - Once you can get over the hurdle of you're probably not gonna die on takeoff or landing. Yeah, like I would be willing to go even if I was going to possibly die. what what percentage about 20 you'd go one in five or one in uh i die one in five times with one in five well that would be 20 yeah wait really 20 chance yeah so 20 is my threshold for going like if i was 50 50 likely to die i think i would say you know let's send some robotic i don't think i would do 20 i think i would probably do like 0.01 yeah like an airplane basically well no no i think an airplane is like once every 12 million passengers or something like extremely extremely good it's insanely i think i think we don't necessarily want to be selling the public on this like one in five chance of death i'm just saying like my threshold is one in five for me personally i don't expect that to scale and um you know i think we should aim for a safety of like 100 100 safety or 99 you know like we should aim for basically zero there there will be some okay like practically speaking you can't you can't say like i'm going to colonize a planet and then nobody dies ever um there'll be some mistakes but i think we should aim for safety that's like you know i think airplane's good a good airplane's good safety it's a good benchmark like yeah everyone goes nobody cares about it i think the really important thing is basically you you and i think elon's very conscious of this you can't have you can't set a precedent for the first starship that goes to mars with humans on it blows up because if that happened that would be horrible for the mission right yeah i mean if you look back at um demo two which is dragon's first flight um there's like i mean i think there's documentary about it at this point but basically like the opinion that spacex team had at the time according to you know what you could find online it's like okay what like whatever we need to do to make sure that the rocket doesn't blow up we're gonna do that um so that's a great safety posture for the first few missions and then after the first few missions hopefully maybe you do like maybe you do maybe you like only do your risky stuff on missions with no humans on it yeah optimist is on board yeah optimist is on board or like you know our robots on board and uh you know we say this all the time it's like our our technology like because we've we've taken the idiot index production so low um is low cost of manufacturing enough that it's completely fine for us to take some risk that the rocket blows up like in in the normal aerospace world, you know, like if you're a satellite customer, like your satellite costs so much money that, um, you, you know, if you're the rocket company and you blow up the satellite, it's like a nightmare.

1:10:53Like you have, like you have to get insurance. The insurance costs a lot because of course there's big risk and the payouts huge. Um, but for us, it's like, we want to build, we want to build equipment that's effectively commodity as at least for us to produce to the point where, you know, if you blow up like one in 10 robotic missions, like we don't care. I want to be very clear. I'm not suggesting you should blow up one in 10 human missions. I think you should blow up zero in 10 human missions, but for a robotic mission, one in 10 blowing up, no problem. Even, even one in three blowing up is no problem.

1:11:21That's just a, a 33%, you know, or 50 % overhead, um, in terms of, uh, building 50 % extra stuff. Yep. Yep. It's like no big deal. So I know when when Airbnb started, I think Brian, Joe and Nate basically didn't hire anyone for I think it was like the first two years. And they basically just built the company, extremely conscious about not hiring the wrong type of people. And they were very focused. I think at the beginning, Brian had this line where he would ask someone, you know, if you knew that you were going to die in 10 years, would you still like what would you want to do? Would you still be wanting to work on this company or would you go do something else?

1:11:59If something else, then probably go do that. How, you know, I know you said you basically have 16 people now and you're very conscious about how you're hiring. How have you kind of found people that are very aligned with the mission and are excited about this kind of future that you're trying to build? Yeah, I mean, I would love to give you some like magic bullet answer. The short answer is just not hiring a lot of people we interview. It's like, it's tough. What's the biggest reason that you don't hire someone? The biggest reason we don't hire someone is we don't think that um you know basically we don't think they have the mission alignment we want we want people who work at starpath to look at the problem we're solving and go holy shit this is the next big thing and um we want people who work at starpath to like read about how the technology we built is quote-unquote impossible which is like commonplace for us to find on the internet that like you know something that we've done in the real world is is quote-unquote impossible in the eyes of someone else and be like yeah yeah whatever it doesn't matter to me um you know we're gonna just keep on doing it yeah we're gonna keep on doing it um and so you know we find tons of smart people um tons of people who are extremely good at um you know the things that we want them to do and we still make tough choices which is to not hire some people even if they're really good um because we want extreme mission alignment um and uh okay but you you asked it like sort of an implicit question under there which is like if i found out that i was dying in 10 years what would i do um and it got when you're talking about it it got me thinking and i think the answer is if i knew that i was going to personally die in 2035 okay so that leaves the mars 28 window 2030 2032 and 2033 2034 i think right um no oh yeah no no it'd be like 2035 so like if i was going to die okay four or five mars windows i think we would we would stick to a similar plan for the next year and a half two years which is to get uh equipment onto the surface of the moon for testing um to be confident that the machines were going to produce that scale actually work and then as soon as we have confidence for any particular machine just like do whatever is needed to finance the living shit out of it and make like tens of thousands of units and store them in a fact and store them in a warehouse that's what i do it's just like okay well we know the machine like it's not like making a city on mars is like this like super complicated thing where we don't know what the implementation is we know the implementation it's like make robot that produces power water oxygen oops and then like like a long trailing list of other stuff um we know how to make the robot that does the robots that do the bulk of that um so if i was going to die in 10 years i would just i would just i would just do everything i can to make as many of those robots um before i die so like like maybe you only need 100 000 to make a city on mars i just make all 100 000 right now which by the way in automotive terms is small numbers like it's very normal for a car to make 100 000 units a year it's like in fact low like yeah no that would be a very bad car yeah a shitty car would be like uh i think ferrari does really low production on a few models and stuff but yeah like a model three i don't know the numbers but i think it's like millions it's like a million yeah or something well yeah and they want to scale it to you know or at least you know whatever the autonomous vehicle is it's going to be like 20 plus million a year or something like that well those are crazy numbers and i hope they achieve them but my point is just like 100 000 robots is like small fish in the game of um robots at scale which which automotive is the best example of robots at scale i love this uh jeff bezos like the regret minimization framework where he basically says like production project himself out to age 80 and look back and say like what would i have regretted if i didn't do like what actions would I regret not taking?

1:15:43What are those actions for you? It's hard to answer without just saying like what I'm doing now. I would say at least on that list, okay, so obviously on that list is doing whatever we can to build the city. Definitely also on that list is doing whatever I can personally after we finish developing the technology to go live in the city. Like one of the things I talk to my parents about sometimes is like once we're done making the technology here, I kind of want to just go do like I don't know if this like pc but like a bit of like slave labor like I want to I want to just go to mars and just like do the shittiest few jobs grunt work yeah just grunt work like because at the end of the day like you know numbers on a screen going up are important for businesses but they don't feel the same as like you know the type of stuff you do as a kid which is like build a fort with your friends and so like I want to do like the manual the shitty manual labor on mars so that i can feel like the whole spectrum of satisfaction the equivalent of working at mcdonald's on mars oh for sure yeah 100 like if mcdonald's has a dude if mcdonald's has a mars think about how fun it is like okay at mcdonald's on earth you're like frying fries which is like not that interesting but the equivalent of mars okay well for some it's interesting i used to work at a burger store when i was younger it's called five guys it's like i love frying fries i actually love cutting the fries but we don't need to get too far into that um it's really fun um like the equivalent of cutting potatoes on mars will be like repairing the cutting teeth for a mining robot where like each tooth that you repair corresponds to like you know seven tons of mine material so like just imagine how how like rewarding that is to like sharpen a tooth put it in a bucket and be like dude this bucket represents like the oxygen production for like 75 people you know that would die without these teeth or or or like you know like you hypothesize like what do you think what do you think the like yeah just like what do you think like the the manual labor jobs on mars will be obviously most of it will be automated but a couple things here and there will still be manual we need like labor jobs for 100 to 1 million people 100 000 to 1 million people sort of situation yeah yeah so i would just want to go to mars so my bucket list for minimizing grad is to go to mars and do one of those things um and then maybe like you know if if possible like if easy you know get on a spaceship and like visit jupiter or something not not the planet just orbit obviously you land on jupiter you're dead as a doorknob but um you know i'd like to maybe go see some of the gas giants up close um i mean there's some pictures we have that are like sort of like dynamic super resolution where like we have shitty pictures and we've like made them better using after effects and whatnot um that are so so cool like various cloud patterns and storm patterns on jupiter so maybe i'd want to go see those but it's like from a feasibility standpoint it might be like well you know a lot of money but definitely going to mars and doing some labor is on the list um and uh and past that i don't have much on my bucket list isaiah from valor talked about the idiot index in the nuclear reactor industry being like 140 to one or more um what is the india index for space and how are you kind of like lowering that yeah um it's hard to say but it's at least 100 to 1 and 100 to 1 is sort of like past a threshold that you and i were talking about today that you know i won't name you know it's like it's past a threshold where you definitely need there's no excuse for 100 to 1 idiot we love we uh we we talked about before this started we talked about the retard index which is which is not a door if you're yeah yeah that's just me but if you're above 60 to 1 idiot and do you want to explain what the idiot index is oh yeah okay for those who are listening the idiot index uh is a is a is a index that can be defined for any finished good which is the ratio of the cost of the finished good so maybe your iphone's a thousand dollars to the cost of the raw inputs raw inputs now practically yeah practically you want to define the idiot index where the denominator so the numerator the finished good product is the price that you pay.

1:19:48And the denominator is the price of all of the commoditized inputs. So you don't want to call like a chip one cent because like technically it's one cent of silicon. You want to say like, okay, the chip is like two bucks because that's the lowest you can, that's a commoditized good. And we already know that you really can't get a chip below two bucks, at least for now. And in aerospace, the idiot index can be like hundreds or sometimes thousands to one where like you buy a product and we buy products like this all the time for you know for for early missions where you know we haven't yet vertically integrated it where like the product might be like ten thousand dollars where if you had the design for the product you could fabricate it for 50 or 20 where like the idiot index is like hundreds and hundreds to one and the idiot index is a good sort of gut check as to whether or not um the finished good has like if you're if you're like if you're is a good deal and if you're if you're basing some sort of economic model or constraint against the price of an item you also want you also want to know the idiot index like you don't want to say something like oh it's impossible to build a city on mars because you know like nasa's moon mars rover cost a billion dollars well it's like the idiot index which okay it's hard to call the nasa mars rover idiotic because it's not it's like an extremely capable machine but at the end of the day it's very expensive to build um and so if you wanted to make a thousand of them you would not take the price of making that thing and multiply it by a thousand you take the price of that thing see how much you can improve it which might be itself a thousand times and then multiply by a thousand like 500 million to like 500 000 yeah so you could nasa could have a robot that costs 500 million you could make it for 500 000 actually ours costs less than 500 000 um and then you would multiply that number by a thousand to see like okay what does it cost to make a thousand um and uh and nuclear the idiot index is super high mostly because of this this is like the big irony mostly because of regulatory yeah it's it's not just it's such like that's only half the irony it's mostly because of regulatory and further furthermore the the majority of the regulatory burden in making nuclear reactors comes from people who care about the environment which is like which is like fate loves irony like the nuclear reactors are like one of the best technologies for protecting the environment because you have effectively emission-free energy production like basically zero emission relative to all other forms of energy production even better like it's better than solar it's better than wind and i'm a big fan of like every energy form of energy production in the world less than probably like coal mining um and it's even better than those so like like nuclear is like the ultimate technology and you know regular regulation makes the idiot index super high and aerospace the idiot index is super high because almost all the goods are sold at extremely low volume so like if you've made a thing you know it costs 100 bucks to make but you spent 200 grand making it and you're only going to sell five well it's at least like 100 it's at least 40 grand and if you want to make a profit it's going to be 200 grand right per unit um and so you're asking about like okay what's our experience like buying in the case of solar panels solar panels from you know existing vendors i don't know what their profit margins are but all i can say is that because they're only selling to a few satellites every year they're only selling to a few satellites every year they have no incentive to make the price small and when the when the price is like a thousand dollars a watt where on earth we make solar panels for like 44 cents per watt okay you can go to home depot and buy a panel for like under a dollar a watt um uh you know the incentive to make the production cheap isn't really there um and so yeah so so the indian index is super high and more or less like the first order of magnitude improvement you can make on the idiot index for any given good in aerospace is by just trying to make it cheaper like if all you do is you ask like okay what are the expensive parts like what are the parts of this that make expensive is it labor is it like a unique material that there's only one supplier of is it like a finished good from another company that has a high idiot index and you just remove that you'll you'll you usually get a 10x improvement.

1:23:48So if all you want is 10x better, all you got to do is like, for most applications is just look at the thing, ask why it's expensive, find the thing that costs the most, which usually accounts for 90 % of the cost, and remove it in one way or another. Maybe you remove it by like testing a cheaper alternative and it works. Maybe you remove it by integrating the production, like vertically integrating the production of that costly good that's making expensive, or some other creative solution. And this is like, you know, I won't speak for other companies but this is like roughly speaking how like all of the products that have seen scale in the space industry of which there are a few have reached that just like okay what's expensive okay delete that uh add something else that's you know maybe maybe like 20 worse and a thousand times cheaper in some cases better and cheaper too um and and basically we do the same thing now in some cases you want to make you want to make like a a more than 10x improvement uh maybe like maybe in the case of rovers it's like okay 100x improvement whatever and and so in that circumstance you you sometimes have to take like a completely different approach.

1:24:46Like, okay, maybe the reason why this is expensive is because you have a requirement that isn't real. Maybe it's the case that you think you need, you know, fault tolerance or redundancy, but actually, you know, if you work out the economic model and you say like, okay, 90 % or like 10 % of my machines will fail completely, that that's actually much cheaper than, you know, making it so darn reliable that only 1 % of your machines fail. And so sometimes those are like, you know, know, tricks you can use to make your cost basis way lower. But it's pretty case by case. And actually, the first thing gets you like most of the way there.

1:25:22Like if you just ask like, okay, why is this motor$40 ,000, which is like a, you can shop for motors online for spacecraft that are$40 ,000, like easily you can find that. And you just like, you just like ask yourself like, okay, which step of making this motor like costs 40 grand or 20 grand and either just delete the motor entirely and make a new one or or something in between you can get like usually from 40 ,000 to four pretty easily