Andrew Song: We face criticism 'cause of that. 'cause we have a more cowboy approach of saying, listen, we know this works. Mother Nature already did it to us. We just got to scale this up very slowly and, and when hearts and minds do it because, at the end of the day, no one actually does listen to scientists, in my opinion.

If we did, we would, we'd, we'd have sold climate change back in the 1919, uh, eighties when Carl Sagan, you know, stepped in front of Congress and said, hey, you know, this greenhouse gas thing is kind of, we should look into this.

Steve Hsu: Welcome to Manifold. My guest today is Andrew Song. He is one of the founders of Make Sunsets a Y Combinator Climate Startup. I'm gonna let him explain to you what the startup is about in a few minutes. Um, Andrew, welcome to the show.

Andrew Song: Thank you so much for having me.

Steve Hsu: It's great having you on. Uh, we met at this zany, uh, sort of festival conference at Light Haven in Berkeley called Manifest.

And, uh, we were actually at, I, I guess you were a speaker or some kind of VIP guest 'cause we met at the, the sort of VIP dinner. And, uh, as I was just saying to you before we started recording, I was really impressed at how well you could explain a really complicated topic. Like, uh, what your startup does.

But before we get into that, I just wanted to talk to you a little bit about your background, where you grew up. Um, I know you were a competitive swimmer as I was. Um, so let's, let's, let's start with that. So did you grow up on the West Coast?

Andrew Song: I did, uh, I grew up in San Jose, California

Steve Hsu: Okay. So you, you live not far from where you grew up.

Andrew Song: Exactly.

Steve Hsu: Yeah. And, uh, how did you end up at NYU and uh, how was the swim team?

Andrew Song: So, uh, I ended up at NYU because I could swim. Uh, I went to this, uh, high school called Bellarmine College Prep, which is like an all boys Catholic Jesuit school. And they were very formidable in the aquatics. Uh, I think there's no state level championships for, uh, for California, but there's like sectionals and stuff.

And so they were very dominant for like the past 30 years. They won probably like 29, uh, as like the equivalent of like a state championship. Um, and so I was recruited to swim for NYU. Um, and you know, on, on the side I was also an econ major. And so, uh, I swam all four years and the program was, um, was actually the first time we, I think they actively recruited, uh, from like these prep schools. And, um, I think by the, at least it's been a while, but at one of the seasons we were undefeated in, in one of our dual meets. Um, and so I wasn't the fastest guy on the team, but, um, I was definitely the shortest and, and, uh, but I would still beat some guys that were much taller than me.

I, we've met in person, I'm, I'm probably five seven on a good day. Uh, and so, um, and my specialty was breaststroke where, um, you know, it's more about the rhythm and, and the technique than it is so much about brute strength. I mean, strength obviously plays into it, but, um, that's what I specialized in, um, excel that.

Steve Hsu: Very cool. Now, correct? Am I, am I wrong about, this is NYU Division three.

Andrew Song: That's correct. And for, for funny reasons, uh, 'cause usually you're broken up by, uh, size of school. 'cause, uh, n NYU's roughly around 40 to 50,000, uh, total, total, uh, students. But, uh, because I think it was in the late seventies or early seventies, they were caught for price fixing, uh, their basketball games, uh, by the mafia.

And so they got knocked down to division three, um, like, and never came back up. Uh, so that's, that's the, that's the lore of NYUD three.

Steve Hsu: Wow, that's some great New York City color. The mob is involved in your sports program.

Andrew Song: Yeah, they were, I mean, like we were, we had basketball teams playing at the, uh, at the Madison Square Garden and, and then they were getting fixed. So shaving points,

Steve Hsu: Wow. Yeah. Most people don't realize how big NYU is. I mean, it's possible, is it the biggest private school in the country? Because that, you know, 40, 50,000 kids is like a, it's like a big 10 school, like where I teach, right. So,

Andrew Song: Exactly. Exactly. It's, it's, it's like, yeah, one of the largest private schools in the nation, I believe.

Steve Hsu: Yeah. So I swam division three also at a time at one of the tiniest schools in the country, uh, Caltech, of which my graduating class was I think 186 kids. And, uh, so we, we, I think my high school swim team was better than my college swim team. But, um,

Andrew Song: Same with me. Same. Same with me. So,

Steve Hsu: Now I also swam breaststroke. I did breaststroke and sprint freestyle. I had no endurance whatsoever, but I was a pretty good sprinter. And, um, I'm just, I wanna run some physics by you and see what you think of it. See, one of the reasons why you see, I think, uh, people who aren't that tall, I'm, I'm just over six feet.

So, um, the reason you see a lot of people who aren't that tall for swimmers, uh, being competitive in breaststroke is because instead of being a straight line when you swim it, the, the, the stroke actually breaks up the, uh, aerodynamic, the, the fluid dynamics of the stroke. And so the height advantage is not as great in, uh, for breaststroke as it is for the other strokes, because the other strokes, if you're a streamline and you have really good reach and leverage.

Um, that height really pays off, but for breast stroke it doesn't as much. And if you look historically, a lot of the top Asian swimmers back in the day before, uh, I dunno if you know this guy, Pan Zhanle, but, but before he became like the top sprint free seller in the world, there were most of the success for a lot of the Asians was in actually in breast stroke.

Andrew Song: That's correct. Yeah, the Japanese specifically were really good at it, so

Steve Hsu: Yeah.

Yeah. Cool. So, um, how'd you like New York? Was it a shock to you to go to New York from

Bay

Area?

Andrew Song: Oh, I mean, yeah, definitely because, uh, I was a sheltered, you know, kid from San Jose and I just wanted to get away from my parents. Uh, and I was the first, uh, I'm one of four kids, but my older brother stayed, stayed in state. Uh, so I was the first kid to like to venture out outside the state. And uh, I just wanted to visit, uh, experience the big city.

Um, and, uh, I really liked the team and that's why, that's why I joined. Uh, also just, I had some sites of actually working on Wall Street and so I thought, Hey, maybe, maybe this could be my foot in the door. Uh, but, um, uh, fate had a different, different,

Steve Hsu: So why are you not a hedge fund magnate right now?

Andrew Song: Um, it was, it was, it was, I, I, I, I graduated in 2008, uh, and so that's when the Peak Financial Alliance Yeah, perfect timing, perfect time to graduate in, in, in finance.

And so, um, you know, I, I, I still worked as an economic consultant, so, uh, that, that, that, it's like a, I guess the biggest firm you've probably heard of is like Analysis Group. But work for smaller, smaller teams that, um, essentially evaluated, uh, like catastrophic events. Uh, when bad things happen, uh, you have to calculate those damages and you can't get a lawyer or insurance person to actually calculate those damages.

You have to hire a person like me to actually calculate. Um, and so it was a very analytical, very, uh, collegial type of environment. Uh, generally I would be like generating reports for PhDs on, you know, you know, basic economic data. Um, and, uh, trying to share, you know, hey, if, like, we worked for some pharmaceutical companies where, um, unfortunately there's like 18 year olds that died because they were smoking and had bad birth control patches at the same time.

They had massive heart attacks. Um, these insurance policies, uh, insurance companies had to figure out how to calculate the damages and how to make these people whole. Uh, so they hired people like us, uh, to do that. So did that for about two years, but, um, got kind of depressed because. You know, I was essentially putting numbers on people, um, and it just didn't sit right with me.

Like I couldn't separate the two. Uh, and coming from Silicon Valley and, um, you know, 2010 was like, like the start of the app boom. Uh, like the mobile app boom. Uh, I thought, hey, like this could be an opportunity for me to go back, and try something.

Steve Hsu: Now you, I think I read maybe in your LinkedIn, uh, uh, profile that you, you have some experience actually with hardware startups. Is that right?

Andrew Song: That's correct. That's correct. And so, uh, that's, uh, my, my first, the first company that I started was around, um, actually guitars that you can play in less than five minutes. And that was actually the Y Combinator company that I went through with, uh, magic or make, make sense as it's not a YC company, but, um, we have a lot of, we have a lot of shared investors like, um, Pioneer Fund, which is like a YC alumni group, uh, has invested us.

So, um, that's where the confusion might, might have come from, but. Um, yeah. Uh, both Luke, Luke, my co-founder and myself are, uh, YC alums.

Steve Hsu: I see. My bad. I, I think I had read that you were YC alums. I just assumed it was your present company, but it was your earlier company.

Andrew Song: That's correct? That's

correct.

Steve Hsu: So talk a little bit about, I mean, I think, uh, YC has become such a phenomenon now, it seems like for early stage stuff, it's like there's so many YC companies, especially in AI, like where I'm involved right now, there's just so many.

There's just like, uh, every year YC is just pumping out massive numbers of aggressive, uh, teams to do, in this case, AI. What, what, what is the, what is it like being a YC alum? What, what is the best thing about it? What's the worst thing about it?

Andrew Song: Uh, I think the best thing about it is the network. Um, you know that, uh, when you meet another alum there, they went through somewhat of a similar program. I mean, obviously it has evolved throughout the years. I was like, when I go to an alumni meetup, like I say, I'm, I'm winter, uh, 2016, people think I'm a dinosaur.

And so like, they go like, wow, you're old school. And, and, and, you know, uh, but at the end of the day, like it's something where even though the program has changed probably dramatically for, for scaling purposes. Um, you still have that kind of basic, um, foundation of, uh, being an alum and being part of a very selective group of people that, um, there was some sort of vetting process.

Like, I kind of joke, I say like, at least I know you're not an idiot. Like, or you're smart enough to get in. And so like, that's like, I think, 'cause I didn't go to a, somewhat, like you went to Caltech. I went to NYU, which is, in my opinion, like a very basic school. And so like, you know, like it was a very low bar in my opinion, to get to NYU

I just had to swim really fast. Um, but like, um, like you, you don't get that kind of clearing you do with like, you know, a, a Caltech or Stanford or Berkeley. Uh, and so, um, it's, it's, it's uh, it's a nice way to kind of get that pre-vetting, uh, to say that. Okay. Like, and, and the confidence of, um. These conversations are, can be candid, uh, they can be, uh, straightforward.

Uh, there's no bullshitting around A lot of, I think when I first got into startups, uh, people like to hype up their numbers or fake numbers or try to boaster like what their, like, their actual position, uh, rather than actually ask for help, uh, rather than actually understand like what are the, like the struggles that you're going through.

Um, and I think I can do that with any YC founder, but I have to be a little bit more guarded when I'm talking to other, other founders about that. So.

Steve Hsu: Interesting, interesting.

Um, you know, one of the mysteries I was hoping to resolve in the interview is how your present company, Make Sunsets, could be a YC company given what you guys do. And I think you've sort of cleared that up for me now because Make Sunsets didn't, didn't actually go through yc.

Um, so let's get into, well, so as we get into the, what your company does, which I think is really cool, it's actually a big concept thing, I think, which is, I think the thing my, I thought my listeners would get the most out of. Um, how did you get focused on that problem? Uh, because it isn't, it isn't, I mean, it's such a big ambition.

Um, usually the people that work in this area are people who are really focused on climate change. Like them, they view it as a real existential issue for humanity, and that's often their driver not necessarily wanting to become, uh, the next unicorn.

Andrew Song: Yeah, yeah, yeah. I mean, I, you know, throughout my career and, you know, growing up, I grew up in sunny California. Um, and I also, uh, was a boy scout growing up. And so, you know, I am fairly in tune with nature. Um, and, uh, I care about the environment. And so that was always in the back of my head. Um, also, you know, my, my parents are immigrants and they had a very scarcity type of mindset where it, and it kind of evolved into, into sustainability where like you know, back in the day, you know, you would use a piece of paper to do your homework. And so my dad, yeah. Shocking, right? But my dad would always say, use both sides of the paper before you use a new piece. And, um, you know, that was something where it was instilled in me. He's like, Hey, this is, this is a, um, this is a, um.

What's the word I'm looking for? This is a scarce resource that we need to, you know, keep, keep, keep track of, um, and not use so much. And it was, I think it was, coming from a more economic standpoint as my, my dad just doesn't want me to go through like, you know, pages and pages of paper, even though it is fairly cheap.

But, you know, we were, we were, they were, they were immigrants with, with four kids. So, uh, money was a little bit tight growing up, but, um, that then evolved to, actually, my first company that I wanted to start was called, um, Ovos, which was a, uh, a cooking app. Uh, because our family was so large, uh, and all of us actually played sports.

We would consume tons of, tons of food. So my mom would also like, just make a ton of food, but sometimes we wouldn't eat it all. And so I just see that waste being produced. And so I thought, hey, like is there a way for my mom to manage, manage this? Um, and so I thought, you know, back in, back in 2000, 2010, um, the app store was starting to blow up.

And, you know, you had this computer that had a camera that, you know. It was connected to the internet and, um, you know, people are interested in making new types of apps. And so I thought, is there a way for us to use optical character recognition to scan grocery receipts to then recommend recipes based on what she purchased?

Uh, so this was Prescale AI. This was pre, you know, this is using Tess Rack, which is like an open source, Uh, Tesira was the OCR, but, uh, O Open CV was the, um. The vision we used for, to scan the actual receipt, but then we then had to then feed it into, uh, the, the receipt data into, uh, uh, what do you call it, um, what was it called?

Uh, mechanical Turk, which is Amazon Service back in the day to then say, translate. What does OOGR mean? It can mean organic, it can mean, or, uh, it can mean orange. What does WHT mean? It can be wheat, it can be white. Um, and, you know, these are the types of problems that we were trying to solve because, you know, I wanted to figure out how we could prevent 30% of all food going straight to the landfill, uh, because I thought that was an important thing.

But then I, I quickly realized no one cared about, uh, saving their food. Like the, the demographic that I was going after wasn't the same, that owned really expensive smartphones at the time. Um, and so this was just a mismatch in the market. And so from that I learned, Hey, I gotta learn more about this whole tech thing.

Uh, and just started to become like a journeyman, uh, and. Went through various roles, uh, at different companies, uh, trying to be either like the first hire, the first non-technical hire and kind of scale myself up and be ready for, you know, when the timing I thought was right to, to start something like Makes Sunsets.

Steve Hsu: Interesting. Now this, that little theme I want to come back to where you said, uh, wait, nobody wants to save food. Nobody gives a shit about this. Um, um, I wanna come back to that in terms of what you're doing now. Um, but, uh, I guess the thing I was trying to get at is what did you really become aware of climate change?

The physics of climate change? Because really what your company's doing is, is really leveraging a certain, very specific kind of physics related to climate change. How did that awareness arise, with you and your co-founder? Like, and then, and then it must have then become like an actual direction that you're willing to like, you know, commit years of effort to.

Andrew Song: Sure, sure. Um, so it was actually, the inspiration was from my co-founder, uh, Luke Eman. He had read the book Termination Shock by Neil Stevenson, which talks about this, essentially a, a, uh, gas gas station magnet, uh, sends sulfur or sulfur dioxide, like artillery shells up into the stratosphere to Cool Earth.

Steve Hsu: Wow. I gotta, I gotta tell Neil next time I talk to him that he's, does he know you, he inspired your startup?

Uh,

Andrew Song: Yeah, yeah,

Absolutely. in his substack he wrote about us, um, saying, these are, this is the only company that's actually trying to attempt something like this. And so we got a little hat tip from him, from that. So,

Steve Hsu: Okay. So your co-founder read the story where you just like, well, Neil's a serious guy and he did study physics and electrical engineering, so maybe he's not, maybe this idea will actually work. Like how much due diligence did you guys do on the actual science before you.

Andrew Song: Yeah. I first thought Luke was insane to even propose such an idea. 'cause I, I, I know, I, I knew this, I knew Luke since 2016. 16, 2015. And he came, came to me with this idea back in 2022. So we had known each other for quite a while, and he would always share ideas with me like, Hey, this startup sounds, this idea sounds interesting, like, do you wanna pursue it?

I was like, no, that's, that doesn't interest me. but I knew he always had sustainability in mind, and when he proposed this, I thought like, there's no way this works. Uh, so then I started reading, you know, a bunch of academic papers for like, like six weeks, like due diligence. And then, the thing I just kept on running into is like, why hasn't anyone tried?

clearly it happens in nature. Clearly we're doing it badly in the troposphere, uh, with sulfate emissions at ground level. Uh, why isn't anyone trying to do this in the stratosphere? At least small amounts, um, to then validate, okay, like, this is safe, this is scalable, uh, which it is. That's kind of where we're at right now, trying to figure out how to scale up, um, to a meaningful amount.

Steve Hsu: Awesome. So at this point, the audience is really mad at me because they're like, wait, I still don't quite know what Andrew and Steve are talking about, and I've been sitting here for 10 minutes. So why don't you give your elevator introduction to what the company does?

Andrew Song: Sure. So, Make Sunsets, uh, launches reflective clouds near the ozone layer to cool earth. Uh, what we do is mimic what volcanoes, specifically strata volcanoes have been doing for millions of beers. The ones that we particularly are interested in, uh, 'cause there are volcanoes that, uh, can cool the earth too much.

But the ones that we're specifically interested in are pinatubo back in 1991, which injected roughly 20 million tons of sulfur dioxide into the stratosphere and cooled the plant by 0.5 degrees Celsius. There was even a more recent instance in 2022 called Hunger Tonka that injected roughly 400,000 to 700,000 tons of sulfur dioxide into the southern hemisphere and cooled.

Cool the southern hemisphere by 0.1 degrees Celsius for about a year. And so, um, we're essentially just trying to mimic that natural process where we're putting a, uh, um, a reflective material, uh, that has been shown in nature, uh, to do it and apply this essentially sunscreen until we can decarbonize our energy stack. Like, uh, you had previous guests like Casey Hanmi. He and I are good friends, uh, companies like that to scale up. Casey is getting there, but, um, we, there's a, there's a time where it's gonna be a little bit of an overrun and we need to be able to mitigate the worst effects of climate change until companies like Casey's can scale up.

Not only um, the heating is happening now, as scientists have pointed out. We're currently at 0.1 to 0.4 degrees Celsius, of warming since the pre-industrial levels. And, um, like there was a New York Times article saying that the rate of warming has been accelerating, uh, for every decade. The world currently warms up by 0.25 degrees Celsius.

And so, um, that's bad. Um, I mean, there's no magic number where that's like, you know, rate of, or, you know, 0.5 versus 0.6, like those are like. Going higher is not good, but there's no magic number where we have to say, okay, like this is the line that we're crossing. or, or we're trying to try to draw a line on.

Um, what we're trying to say is that if we know how to warm up the planet using carbon dioxide, AKA greenhouse gas gasses, we should also figure out a way to cool it down.

Steve Hsu: Great, great explanation. So let me, let me dig into, uh, some of the details here 'cause I'm sure my audience wants to hear it. So. Um, just to, just to recap, we know from these examples, these volcanoes that you mentioned, that the sulfur dioxide, if it's ejected into the atmosphere, it increases the reflectivity of the earth and so that the earth reflects more of the sunlight, uh, into space and therefore it cools the earth.

Um, we have some guesses at how much what, what the effect size is per say, a million tons of sulfur dioxide that reaches the atmosphere. Um, presumably we know how persistent, how long that effect lasts. I guess it was a temporary effect. After each of these volcanic, maybe

just

say

Andrew Song: called residence time. Exactly. And so this is why VCs love us. It's recurring revenue. You gotta reapply that sunscreen year after year.

Steve Hsu: Right. So, so a typical molecule, let's say you get it all the way to the stratosphere, uh, how long is that gonna persist or at least, uh, continue to increase the reflectivity, uh, of the stratosphere? How, how long is that last?

Andrew Song: Uh, so it depends on the latitude as well as the altitude, uh, and also the particle size. But, uh, based on the modeling, based on real world examples, anywhere from one to three years, uh, versus, yeah. So this is again, uh, if you inject it near the ozone layer, so 20 kilometers and above, uh, but if you do, if you actually inject sulfur dioxide in the troposphere, uh, so this is the air that we breathe.

This is where commercial airliners fly. Uh, this is where most weather activity happens. Uh, you still get some reflectivity, you still get some cooling, but that residents time then, uh, gets, gets cut down to 10 days, uh, to like, or a couple weeks, uh,

because of the rain clouds, because of, uh, there's no, uh, stratospheric wind circulating, uh, and dispersing these aerosols all around the world.

Steve Hsu: Right. And can you steelman the critics, uh, particularly the environmentalist critics of what you're trying to do? Are there people saying like, this amount of, uh, SO two in the stratosphere is gonna have these negative consequences for the environment and therefore we shouldn't do it?

Andrew Song: Yeah. Yeah. And that's, that's like the, the big argument that we get all the time. Luckily, climate scientists and atmosphere scientists have modeled all these types of scenarios. like we will personally never try and like, at least as far as I know, I'll never try and cool the planet by more than 0.5 degrees Celsius.

Um, if we even are so lucky to get that far. But that's, in my opinion, the do not pass go line, because that's what Pina Dubbo did. and in 1991 I was six years old. Like, I don't remember anything bad happening in 1991 weather-wise. But you know, everyone's mileage is like, you know, mileage may vary, but, uh, in my opinion, it was a fairly nice, nice time in my life, uh, in terms of weather.

Steve Hsu: A Beautiful, that was a beautiful summer in 1991. That was the year I finished my PhD at Berkeley. Drove across the country to Cambridge, Massachusetts, and the weather was fantastic.

Andrew Song: Yeah, there you go. And so, and it was 0.5 degrees cooler during that time. Um, and so that's, that's like, you know, we're not saying we want to create an ice age. and like that's a lot of what the ACA academic papers like, you know, try to model out because they want to know what is that upper bound of, of like, doom.

But we know that this is something that can be injected in, in various amounts, in small, small quantities that can scale up easily. And then again, it falls out in about a year or two. And so, you know, we can, we can actually modulate how much cooling we really want and that's the most exciting thing.

It is reversible, uh, just like volcanoes, just like, uh, what's currently happening where, uh, back in 1979. We were injecting roughly 140 million tons of sulfur dioxide into the troposphere. And then, uh, the Clean Air Act kicked in, uh, the EPA clamped down on sulfur emissions, and now we're down to roughly around 72 million tons of sulfur dioxide worldwide.

And so because of that, uh, there's been a lot of papers and articles recently saying that, uh, we've unmasked this cooling because we're, quote unquote cleaning up the air now. And so, uh, we're still injecting CO2, which captures the heat, which traps the heat, but we're, we're, we're cleaning up the SO two in the troposphere.

And so all we're proposing is saying, go ahead.

Steve Hsu: Sorry to interrupt you. Give us the number again of what you need in the stratosphere to get that 0.5 degrees Celsius cooling.

Andrew Song: So, so, um, this is based on the IPCC uh, modeling where essentially, um, for every trillion tons of CO2 you put into the atmosphere, it warms up the plant by roughly 0.45 degrees Celsius. That's the average. and then, uh, based on David Keith's work, uh, he's one of the leading academics in this field.

One gram of sulfur dioxide in the stratosphere offsets the warming effect of one ton, one metric ton of CO2 for a year. And so that's a one to million leverage. So you need roughly around 1.1 million tons of sulfur dioxide to be injected continually into the stratosphere, to cool earth by 0.5 degrees Celsius, which again is nothing.

This is nothing.

Steve Hsu: Right. So just, I just wanted to make your point that a million tons in the stratosphere, and after a year or two it ends up maybe back at sea level, um, that's negligible compared to the pollution that we already inject into the troposphere. Right. It's a, it's a,

it's a. it's a. small

Andrew Song: And we're saying just let's just shift some of it up, up a little higher. That's all. So it's not like we're adding anything to the system. We're just shifting it higher. where there's less of the downside effects of, uh, experiencing sulfur dioxide in, in the, the air that we breathe.

because generally what happens is that when you inject sulfur dioxide into the stratosphere, uh, there's this thing called a brew does in circulation, where essentially if you, if, I don't know if this is a, it's a video podcast, but, um, if pretend just visual, just pretend my head is the globe. Uh, if you inject near the equator, what just happens is that the aerosols essentially just travel to the polls and it takes about that distance alone, that is why it takes about one to three years and then it falls out at the polls itself.

But since the amount is so little, um, that amount is negligible to anything down below. So you can continue to just reapply this sunscreen over and over again. Uh, we've been applying this sunscreen, we've just been doing it the worst way possible instead of applying our entire body with sunscreen, which is applying just on our faces, like with our eyes and our mouths open, which is, which is not good.

Steve Hsu: Now I think that this is the part when we were having dinner together, I was kind of stuck on for a while. I was like, what is actually happening with physics? So, you know, you take the contents of our atmosphere and you just increase the concentration of SO 2, just slightly. But there's a significant increase in reflectivity, I guess in waves.

It must be visible light wavelengths. Um, I don't know, maybe you can say a little bit about that.

Andrew Song: Sure, sure. So what we're actually influencing is watts per square meter, so it's called radiated. Forcing is the unit of measure or is the, is the metric that we're actually trying to influence. And so, um, currently the earth is absorbing way too much sun versus reflecting it away.

And so by adding increasing albedo, meaning how reflective something is, that's how we can actually prevent the sunlight from even hitting the greenhouse gas layer, which is what is trapping the heat. So we're kind of up the chain in terms of like the, uh, greenhouse gases where we can technically continue to increase CO2 levels, uh, to higher amounts without the warming effect.

If we can apply the right amount of, so

Steve Hsu: Right. I think the part that was, I was kind of hung up on, you know, as a physics guy, as you were explaining this to me, is how could such a small increase in concentration of sulfur dioxide actually have this. You know, pretty significant, you know, the effect of changing the albedo of the, this, the earth's, uh, atmospheric layer.

And, you know, it still probably needs a little bit more, uh, at least for me to understand it at the micro physical level. But, um, I, I guess it must be that SO two is just extremely absorbent and a missive

Andrew Song: Reflective. Reflective. Yeah. I mean, just think of very, very tiny mirrors. I mean, it doesn't have to be SO 2. Other candidates have been calcium carbonate.

Um, and also another one is diamond dust. Um, but the, for various reasons, scientists keep going back to SO 2, um, as the kind of ideal candidate for now.

but we're not married to SO 2 as a particle in particular. It's just the most well studied, naturally occurring is why we selected the aerosol to send up recently.

Steve Hsu: Got it. Got it. Okay. So I think we've explained what you guys are trying to do. You're trying to get the stuff up in the atmosphere. Now maybe talk about the mechanisms that you're currently using, because at the moment people can actually pay you

Andrew Song: Mm-hmm,

Steve Hsu: to inject some of the stuff into the stratosphere. But talk about the technology you've developed to do that.

Andrew Song: Sure, sure. It's very, very basic and you'll be shocked, um, uh, how simple, dead simple it is. Um, we've had, you know, people come out and they go like, really? This is it? Um, like, and we're just saying, yeah, this is how simple it

Steve Hsu: You're not firing.

Andrew Song: what we're. No, no, not yet. Not yet. Um, it's just this, that's just too James Bond to terminate shock.

The reason why Neil Stevenson used, uh, uh, artillery shells is 'cause he's a gun guy. So, um, you know, it fits, it fits his mom but, uh, no one, no one gets mad at balloons. For the most part. For the most part, everyone likes balloons. Uh, we specifically use, uh, natural biodegradable latex balloons that the weather surface surfaces use to get measurements of the weather.

And so, uh, essentially all we do is we put sulfur dioxide in these balloons, um, and then weigh it, to understand how much mass we put in. And then put lift gas in. So it could either be hydrogen or helium, uh, and then attach some telemetry to the balloon to verify natural altitude, 'cause that's the most important thing.

And then based on the physics of the balloon itself, since it's natural latex. As you go higher and higher into the atmosphere, the balloon expands because the pressure outside the balloon gets less and less. And so the balloons are actually built to actually explode at a certain altitude, which we can calculate based on the, the, the payload, uh, the weight of the payload, the, the, the lift gas, uh, the size of the balloon.

Um, and then also, uh, what's the last one? Um, those are the main three. but essentially then you can actually time where to actually make the balloon explode. And so we always aim for 20, 20 kilometers and above.

Steve Hsu: This is literally a problem that one could assign in freshman physics, actually.

Andrew Song: Oh yeah. Oh yeah, yeah, yeah. Exactly. Exactly.

Steve Hsu: and the mass of the balloon, et cetera, and the tensile strength of the latex, et cetera.

Andrew Song: Exactly. Exactly. Exactly. And so they're, they're rated to go up to 30 kilometers. I think the highest we've ever gotten a balloon to go is about, uh, 35 to 40. Uh, so they get, they get pretty high up there. Um, the higher, the better. But, um, you know, 2020 is like everyone says, this is the prop, proper stratosphere.

And then you won't, you'll get the one to three year residence time. Um, and so these balloons are fairly small. Our payloads are currently around 1.5 kilos. Uh, so, you know, that offsets the warming effect of 1500 tons for a year. Um, so it, it's, it's sizable.

So that's roughly 68,000 trees in one of these 1.5 kilo payloads. So 68,000 like trees, if you assume that each tree can offset Yeah, the tree lives for one year, a tree can offset roughly around 22 kilograms.

So based on that math, um, because I, 'cause most people don't understand what 1500 tons of CO2 looks like. Um, so, you know, if you can, everyone can just look outside and see a tree. Uh, so there's just man, 68,000 of those in, in this one single balloon that's about six feet in diameter, uh, has that kind of leverage.

And so to get to say the million ton, uh, you know, payloads that we're looking for to actually offset, you know, 0.5 degrees Celsius Celsius cooling, uh, there are balloons that NASA currently flies that have payload capacities of, of one metric ton. So one metric ton for the people who are visually. Who wants to see what that is, is about the size of a Toyota Yaris or a Fiat Fi vintage Fiat 500.

That's about a ton of ways. So there are balloons that actually can go up that have that kind of pale capacity that we can leverage to use to send up the sulfate aerosols. but you can also use artillery shells.

Uh, you could also use supersonic jets, 'cause character. It doesn't have to be supersonic, but generally you want something that's supersonic.

'cause then you get the, when you, when you have a supersonic plane, you have to go higher because of the, the, the wind resistance. If you fly too low with supersonic, you're running too much air, uh, into, into too many air molecules. So you get less efficiency. Uh, but it could be a U2 spy plan as well.

But then the trade off is the payload is very small.

Steve Hsu: Now for you guys, you know, if you were someday to do, or humanity was someday doing this at scale. So a million tons a year. Are balloons probably the most feasible way? And at that point, what is all this? I, I know you keep saying biodegradable natural latex, but what a million such balloons going up every year.

What, uh, you know, are there environmental issues associated with that?

Andrew Song: Absolutely. And we should, uh, you know, run, we should solve those problems as we run into them. Uh, and so, um, there are plastics that are biodegradable. Um, another thing that is actually one of your guests, again, I'm gonna reference Casey Handier 'cause he's the one who kind of led us to this. But, um, we could just burn the balloons up there once they reached the altitude.

Uh, since we use hydrogen as a lift gas, uh, we can actually have a time charge or even just have some telemetry saying like, okay, we know the rate of. The rate of, uh, climb for this, um, for this balloon. Uh, and we know it takes roughly around 120 to 150 minutes based on, uh, you know, a balloon going about three to five meters per second, um, to get to the stratosphere.

Then, you know, that's when we ignite this payload, um, and then release the hydrogen or burn the hydrogen, which then burns the balloon, but then also releases an O2. Um, and so to do this at scale, we're not gonna be doing this launching individually. Um, as you referenced, our backgrounds are in hardware.

Like we wanna nerd out. We want to build robots that can build, like, launch balloons like, you know, with automation, but we're just not there yet. And so, um, but to visualize how that robot balloon launching robot would look like, uh, imagine a balloon, uh, your, your Amazon packages that you get, uh, you'll get these little air cushions inside of them.

Um, imagine that machine that does it is just essentially a balloon sausage. Um, and instead of injecting air into it, you inject hydrogen plus sulfur dioxide. And so now you have this balloon sausage that's just going straight up into the air. And these balloons are about this, the length, the diameter, about 10 meters, and about the length of a a a a kilometer.

So these are really, really big balloons, but, um, you know, we're not breaking any physics here because, um, you know, NASA regularly flies, like payloads that are around three to four tons of, of, of payloads. So like three x three and a half x larger than, uh, what we're currently proposing.

Steve Hsu: Yeah, I love ambition . Ideally, is it near the equator or is it, does it matter? Uh,

Andrew Song: Correct. Um, and so, so you want to do it closer to the tropics. Uh, so then you get that longer residence time. Um, there are proposals to actually launch closer to the arctic. To then get, uh, target just the ice caps. However, then that residence time gets cut down to roughly about six, six months.

Steve Hsu: it. Okay. But somewhere in some fairly isolated area, there's this giant factory that's just like streaming up these tubes of plastic into the sky. And every now and then, every now and then one of these sausage tubes ignites. And then, and, uh, excellent.

Um, in terms of the dynamic near term dynamics of your company, I assume you're, you're doing R and D to figure out what that scalable, sort of, massively scalable solution would look like.

But are you also trying to become profitable based on carbon credit type, uh, financing? Like, just explain like.

Andrew Song: correct. That's correct. That. Yeah. So, you know, when we looked into this space, uh, we quickly saw that the CapEx to get into any kind of carbon removal or climate tech intervention type of company, the CapEx was enormous. And,

Steve Hsu: knows.

Andrew Song: as Casey knows, yeah. You know, us being the kind of scrappy startup guys we are, the first thing we thought is like, how can we make this a profitable business from day one?

Um, and how can we sustain ourselves financially until that day, that day comes where yes, we have the plans of how to scale a one-ton balloon, but how can we work our way up to there? And it's, it's just bigger and bigger balloons. and so currently what we're doing is we're selling what we call cooling credits, where one cooling credit costs anywhere from one to $5 to offset

the warming effect of one ton of CO2 for a year, So this one, cooling credit. we'll, we'll group together every month, group together, put it, put the equivalent into the SO 2, uh, S 2 into the balloon, and then send a report saying, Hey, we sent your balloon on this day.

Uh, it reached this altitude. Uh, these are your serial numbers of the, say, 1500 grams that we deployed in this one certain balloon. Um, you know, thank you so much for helping us Cool Earth. So it's essentially the kind of, uh, business model that we've had. Hopefully what happens is that then we have a critical mass of people where corporations will start to adopt this as part of their sustainability goals.

I don't think that's gonna happen anytime soon. Uh, but you know, again, if you want to currently remove a ton of CO2 using direct air capture, it costs anywhere from $400 to $500 per ton. Um, this is not an apples to apples comparison. Again, ours only lasts for a year. You know, direct air capture can.

Be, you know, hopefully captured, sequestered for over a hundred plus years. but you know, the, the, the cost curve is so dramatic in terms of like, we're currently charging a dollar, uh, we can get it down to 1 cent sub 1 cent, director captures hoping that they get it, get it down to a hundred dollars per ton removed in, you know, 10, 20 years.

Um, we can do that with just the big enough check if, if, if someone is willing to, you know, launch a really, really big balloon with us.

Steve Hsu: So, yeah, it's an important distinction that, you know, if you capture the carbon or you stop someone from emitting the carbon, you know that carbon was gonna be circulating the atmosphere for a really long time, whereas your effect is, you know, transient maybe one year effect, right?

Andrew Song: We're essentially decoupling the warming from the CO2.

Steve Hsu: I mean, given this disparity, uh, in the, in the timescales, is there any hope that, like the EU might make what you're doing eligible for the equivalent of carbon credits and things like that? Um, or is it just, does it just have to be right now the goodwill of someone who just feels good 'cause they spend a dollar on, uh, cooling things a little bit.

Andrew Song: I mean, if we can just look into my crystal ball right now, uh, that's the hope.

Um, people realize that we're not gonna build power fast enough. Um, and, uh, you know, we're not gonna, we're gonna even, and no one wants to install AC units in, in the U right now, uh, for, for some reason or another.

Um, but like, then people will start to realize, okay, is there something we can do on the micro macro scale to, to, keep the energy sort like our, our energy grid, uh, stabilized until, until that day comes. And so it's, it's something where. It's gonna take time. I think people are just wrapping their heads around instead of putting, or instead of taking something away from the atmosphere.

And putting something in instead in itself is a huge hurdle unless you understand physics. Like I, uh, the reason why I wanted to jump on your podcast is 'cause I think most of you, your listeners are physics, uh, physics guys or girls. So, um, so like them, they'll quickly understand, like, you know, radio forcing and lots per square meter and all that stuff.

but it takes a while for the normal person to understand, wait a minute, you're not taking anything away from that unless you're actually adding something. and you're not actually solving the underlying problem. You're just kind of kicking the can down the road, which again is what we need right now.

We need more time. Um, there's catastrophic climate events happening all the time because we're introducing more energy, uh, into the system than taking it away. Um, so that's, that's what we are.

Steve Hsu: I think one of the numbers we should have mentioned by now is my fault that we didn't get there already is, you know, the total cost to, you know, cool. Say, say to cool the plant for one year or half a degree Celsius. Um, it's, it's rather modest. Is it like $50 billion or something?

Uh.

Andrew Song: It's, it's, less than that. It's like one, one to $2 billion.

Steve Hsu: So it's one to $2 billion.

So, yes. So I think that was another thing that struck me when I first spoke to you about this, is that this is eminently affordable, like government programs easily spend that much on CO2 related stuff, right? So, um, so if you could just, if you had the tech ready and you just said to the EU or, you know, some cry babies who actually care about this stuff.

Just kidding. Um, just give us one or 2 billion and we'll, we'll, we'll, uh, counteract the, you know, huge amount of the, the CO2 related heating. Um, the amount of money is just shockingly small that's involved here.

Andrew Song: Right, right. And to give you, uh, give you some context. I mean, one, 2 billion sounds like a lot of money, but, um, there was a, a, a Bloomberg article saying that since May of 2025, if you walk back 12 months, the US alone has experienced about a trillion dollars worth of climate damages. And so it's like,

Steve Hsu: Yeah,

Andrew Song: I, again, like to just do the math.

Like I, I think that's a good bad, like a trillion dollars. Trillion dollars versus a billion dollars.

Steve Hsu: you

Andrew Song: Um,

Steve Hsu: If you took all the developed countries together and you convinced them that, okay, a hundred billion a year in climate related damage is happening to you guys, and for 1% of that you could kind of alleviate it for coming years, um, you know, it seems like a no-brainer.

Andrew Song: Yeah, it is. And I think, I think the, the things that we're currently facing up against, and the, the critique that we get is that, the scientists who are researching this, they wanna do more modeling. They want to make sure that every stone is unturned. They wanna make sure they're at 0.9 9, 9, 9, 9 9, you know, like nine nines of degrees of confidence before even a single molecule gets deployed.

Uh, and so we face criticism 'cause of that. 'cause we have a more cowboy approach of saying, listen, we know this works. Mother Nature already did it to us. We just got to scale this up very slowly and, and when hearts and minds do it because, at the end of the day, no one actually does listen to scientists, in my opinion.

If we did, we would, we'd, we'd have sold climate change back in the 1919, uh, eighties when Carl Sagan, you know, stepped in front of Congress and said, hey, you know, this greenhouse gas thing is kind of, we should look into this. But, um, it's, it's something where I think. There's modeling that can only go so far.

You have to actually show it and prove it. Um, and this is, that's where the, that's where the hangup is, is that people don't want to experiment on themselves or, you know, there's all this, this kind of, you know, kind of consent. But once people realize the risk involved and the, the, the little amount that we have to put up in the cost, um, I think that's when people start to kind of warm up to stratospheric aerosol injection.

Steve Hsu: Yeah. Now of course, I, I, I, I have not looked carefully into all this research, so I'm just accepting what you say and it, it doesn't sound completely crazy to me. But, um, you know, if, if you say to these governments, okay, there's maybe a billion dollars or a hundred million dollars of r and d required, and then we could start putting up not a million tons, but a fraction of that, and then look to see the effect, see if it's really distributing itself the way we think it should see that it actually, there is a slightly measurable decline in the rate of heating or something.

Um. You know, there, there's, there's steps that one could take along this research program that I think don't jeopardize the world environment, right? They're small steps, definitely way smaller than like a volcano going off. And, and surely we should explore that direction, um, which involves both the RD of how, how you would make these sausages, uh, if you chose to do it, but then also like testing the impact

of,

you

know,

Andrew Song: The monitoring of it.

Steve Hsu: Yeah, exactly.

Andrew Song: Exactly.

Steve Hsu: um, I mean, I would think that satellites and stuff could measure the change in the albedo. Like if you had a slight change in the albedo, I would guess they have enough sensitivity that they could, they could at least notice it well before you have to put up enough to cause a 0.5 degree Celsius, uh,

cooling,

Andrew Song: Correct, correct. Correct. Yeah. So that, that's why we, we, we say that, uh, right now our, our kind of intermediate goal right now is to scale up to that one ton payload. Uh, because based on the conversations we've had with scientists, roughly around 10 to a hundred tons of sulfur dioxide can be measurable by satellite.

And so the specifics. Yeah. And so, you know, you have to have good timing. You have to follow the plume, like all this kind of stuff. But the, the, the satellite that we would leverage is called Tropomi, which is, uh, a European Space Agency satellite that was specifically to monitor, you know, weather activity plus volcanic eruptions.

and so that there, there is that level of sensitivity. There's even, uh, there's a current project called, in the UK, sponsored by Project Aria, where they're trying to actually figure out what is that, uh, that mini amount of sulfur dioxide that you have to deploy to be detectable by current satellite infrastructure.

Um, and so they're gonna go the modeling route. We're gonna go the actual physical route and show that it actually works. and that, that, that, yeah, go

Steve Hsu: there's an alternate universe not too far from ours where I became director of Aria

Andrew Song: Really interesting. They're looking for a CEO right now, so.

Steve Hsu: And then I approved and then I, because my Aria was a creation of my buddy Dominic Cummings when he was, uh, in power in the

UK, uh, with Boris Johnson. And I was one of the people mooted for the directorship.

and. Yeah, it's funny. And uh,

Andrew Song: Small world.

Steve Hsu: In that alternate universe, I approved the R & D project involving your company, made sunsets, and then this thing is chugging along.

But,

uh,

Andrew Song: Yeah, yeah. I mean, they've raised, I think, was it 45 million pounds or something like that? So, I mean, it's a good first step. Again, it's everything, there's no deployments involved. It's all modeling. Um, and, but at least there's some real money going behind it. And there's actual government support, which is exciting.

but. In my opinion, like I kind of look at and, you know, we can step into kind of more in your field, but you know, I I, I was an early employee of scale AI, AI back in 2018. Like before, before that time it was shit data and shit data out until you actually had well labeled data. That's when machine learning models actually got useful.

And so I feel like that's the same thing that's happening currently with stratospheric aerosol injection and their modeling. Trouble is that they're using very stale data or data that's just not clean. It's obviously done via volcano, but you also then have to, you know, see out all the CO2 and all the other counter counteracting, uh, aerosols as well as the greenhouse gases.

So, you know, when Pinatubo injected 20 million tons, yes, it cooled the earth by 0.5 degrees Celsius, but there was also a ton of CO2 tons of CO2 millions of tons of CO2 that actually went up as well. So it wasn't a clean, uh, you know, a clean reading, uh, is, is how, what I'm trying to say.

Steve Hsu: Yeah, I mean, in general I'm very skeptical of climate modeling. Um, I, we, I think you mentioned a couple previous guests I've had on the show. So I've had Casey, Handmer, who's, uh, on the carbon capture side, Klaus Lochner, who's kind of crazy. Well, actually they're both physicists, but Klaus is kind of a crazy genius, I guess.

He's the guy who first proposed carbon capture as a, as a possible thing. And he started out as a particle physicist at Los Alamos. Um, but then I had a third guy who also, I was trained in general relativity, but then became like one of the leading climate and weather modelers. Uh, now his name escapes me, but I'll, I'll put the link in the show notes.

But, um, that whole episode is about the limits of simulations

Andrew Song: Modeling. Yeah. Lack of

Steve Hsu: Yeah, and I think the best data would be if you could really track a plume using a satellite and measure how it affects the reflectivity. I think that would be amazing if you could actually do that.

Andrew Song: Yeah. And that, that would take probably about $20 million. So if there's any billionaires on your listen to your podcast

right now

Steve Hsu: Well there are, there are some no guarantees. But, um, yeah, so I think that would be an amazing experiment. So in, in the event that we really do need to take aggressive measures to directly, uh, counteract global warming, you know, it's, it's good for humanity to do that, basic science to figure out what's, what's the most feasible method, and yours might be the most feasible method.

Yeah,

Andrew Song: And if it's not, we're willing to shut down the company. This is why we call it Make Sunsets. We wanna sunset this product as soon as possible. Um, because we want companies like Casey Handmer's terra formation to, to scale up to meaningful amounts, because that's what we need. We can't keep applying this sunscreen.

Uh, but we have, we, we, we've actually been doing it. We've just been doing it in the wrong altitude and too much of it,

Steve Hsu: My, my view on climate change, well, one, I, as I've, as I've stated in some of the other podcasts, I have a lot of, uh, I think there's a lot more uncertainty in the modeling than the actual I-P-E-C-C type people would admit to. But the, the, the other thing that people I think are not aware of, you need like, uh, o you need like, sort of, uh, overlapping areas of expertise to understand this.

The rate at which China is shifting to renewable energy production, uh, they're gonna pass 50% of electricity generation from mostly wind and solar. this year. I think, um, it tells me that again, if you do these sort of back of the envelope, large scale calculations, it tells me in a few decades 'cause, 'cause that that energy is the cheapest energy you can make.

It's cheaper than burning stuff. Um, within a few decades, I think we will not be pumping substantial amounts of additional carbon into the atmosphere because eventually these new technologies, which are clean, will actually take over just based on, just purely on economics without any additional subsid subsidization just because the technology's getting better.

So really the next few decades that we have to really. Worry about. And I can imagine a scenario where, okay, first of all, it could be, there's not really a big problem over these next few decades. 'cause I think the models are very uncertain. But even if there is a huge problem or we just decide we wanna spend money to not have these climate, uh, catastrophes and stuff, um, I can imagine a world where they go to your solution and they just use it for 10 or 20 or 30 years until the rate of CO2 emission goes way down because of these renewable energies and Casey's stuff never gets off the ground. It's totally, we don't, we don't really need to remove it. It's there. And we had some warming, but we just lived with that. And then, uh, the, the amount of CO2 production by humanity as a whole just goes way down because of these new technologies.

I mean, I could easily imagine it happening that way. I think the thing is that, when I talk to most people, even people who care about climate, know a little bit about climate. They don't necessarily know the economics of what's happening with electrification

Andrew Song: Oh yeah, it's

tremendous.

Steve Hsu: Look at what's actually happening in China where they, the car fleet, transportation fleet, and renewable energy production are all, it's clearly past the tipping point.

You, you know, where it's gonna go. It may take them another or two, but it's definitely gonna get there. And then the rest of the world will figure out, oh, this is actually a better way to do it. So for me, it's a finite time problem. The CO2 thing is eventually actually gonna be solved just by pure, uh, scaling of tech energy technologies.

Andrew Song: I, I, I agree. I agree. And to, to be fair, Casey, uh, will remove CO2, but then use it to convert it into, uh, hydrocarbons to then burn it and

Steve Hsu: Yeah, no, that's

true.

Andrew Song: So.

Steve Hsu: So for him, yeah, if his thing gets fully working, he's actually making, making useful stuff out of the, this removes so

Andrew Song: Instead of putting, yeah, putting it to the ground, which I think is, is not the way to go right now. Uh, maybe it's, maybe it's an incremental step. Like, 'cause you know, that's, that's the first part of Casey's equation,

Steve Hsu: Right, right. But you know, it is amazing. Like, uh, I, well if I were, you know, if, if I were a philanthropist and I were advising you, I might say like, hire a really good lobbying firm in Europe because they're, they're wasting far greater amounts of money on stuff that's potentially less impactful than a mature version of your technology.

Right. So if, if you really could like, you know, counteract that much, uh, heating, spending only a billion dollars a year or $2 billion a year, it's like, wow, that's the best solution anybody's ever thought of.

Andrew Song: Yeah. Yeah. I mean, it wasn't me. It has been studied since the 1970s. Um, there's been over 2000 act papers written about this. Um, there've been scientists who dedicate their lives to try and discount this as a climate intervention. Um, and then got converted and became like proponents of it, which is, which is kind of crazy.

Um,

Steve Hsu: Is there any entity, like maybe an academic lab or something that is spending as much energy or making as much progress in this direction as your company?

Andrew Song: no, uh, sadly, I mean, well, I, I maybe being unfair here. Um, you know, there, there, there was Scope X, uh, which was led by David Keith, and, um, it was, it was a Harvard backed, you know, effort. To send instrumentation to analyze aerosol plumes. Uh, they raised $20 million from Bill Gates, but unfortunately they got shut down.

They couldn't even launch a single balloon. Um, and it was because of, uh, the pushback that they got. There was a native, there was a native group in, in, uh, I believe, uh, Norway or Sweden, who said, you can't launch in our backyard. And so they got cold feet and said, okay, I guess we can't do this.

And then kind of, you know, closed the shop and went from there. So we saw that and we're like, what, what? Like, sorry if I swear, but what the fuck? Like, just do it somewhere else. Like we're doing this in the north, we're doing this in Northern California. Like right now

Steve Hsu: That's where the, that's where the Silicon Valley attitude comes in.

It's like

Andrew Song: Yeah, exactly. We don't have the baggage of trying to get published here. We're just trying to solve a problem.

Steve Hsu: Incredible. Incredible. So, um, what, what should we look for in terms of big milestones for you in the next year or two? Is it, is it getting that one ton balloon up there?

Andrew Song: Yeah, exactly. And it's really like, either we'll get it through, like we have roughly now 900, 900 plus customers. These customers include Palmer Lucky, Emma Sheer, who was the one day CEO of OpenAI at one point, and co-founder of, uh, Twitch. Uh, we've got, um, who else, uh, Chen Wong who, um, you know, is the ex CEO of Reddit, but now is a tree guy.

Um, like those types of guys are kind of like helping us out and um, you know, we just need more of them. We just need to get more smart people to understand the physics of this and the economics and how scalable it is and to be talking about it. Um, and from there I think it's just a matter of, you know, shots on goal like this where I get opportunities to talk to people like you and to a broader audience to understand it's like, this is something we should do and try.

And if it doesn't work, it goes away in a year, um, and it doesn't cost much. So that's, that's where we're at right now and. We've been essentially like in, you know, cockroach mode trying to do that. Like I pay myself $60,000 a year right now. You see my kids there, you know, they're cute, but you know, they cost a lot of money and so in the Bay Area, and so, you know, we're, we're sorry.

Steve Hsu: They can eat.

Andrew Song: They can eat and they can eat. Yeah. Yeah. I think they, they gobbled like, I think like 10 s'mores. Like we were talking, we were talking and so, um,

Steve Hsu: I, you know, you, you need to get on the, um, feel good, do-gooder, ted effective altruism, et cetera, et cetera, circuit, which I'm sure you're aware of. But, uh,

Andrew Song: yeah, yeah.

Steve Hsu: know, um, one step at a time, I guess

Andrew Song: Exactly. Exactly. I think it's just like, we're still obviously trying to figure out our messaging. Um, you know, there's a lot of block blocking and tackling from like, the conspiracy theorists. Like, I don't know if you were following the news over the weekend about, you know, the whole Texas flood stuff, um, but like M-G-T-M-M-G-T.

No, no, no, no. I mean, they're actually, they're throwing shade on my, uh, my friend, uh, Augustus Doricko. 'cause he actually has operations of cloud seeing down there. But, you know, based on the evidence that he showed, um, clearly they were not responsible. I mean, this is just climate change. Um, this is what's happening.

It's what's gonna happen more frequently. So, yeah, it's, it's just something where we get, we get bunched in with, um, natural weather events that are, that are affecting people. Um, they think it's us. And like, to be clear, we've only deployed roughly around 160 kilograms of sulfur dioxide. So this is, this is a drop in the bucket,

like

Steve Hsu: Just proof of principle.

Andrew Song: Exactly, exactly.

Steve Hsu: Yeah. Hey, let me, since we've gone through whatever listeners are still left after we're at an hour now, um, whatever listeners are still left, I, this is a good chance for me to explain. I think that with the background that we've covered now, that the easiest way to explain why I believe there is so much uncertainty in climate modeling, which is un largely unacknowledged, um, by most modelers.

Although the, the, the guest whose name escapes me, who's one of the most famous guys in the whole field, um, acknowledged this in the interview that I gave, uh, he gave me the following. That when you model you, you don't, the simulations are not good enough to figure out what it does to cloud cover.

So, the dynamics of clouds is very like micro local stuff. It's very, it's not something we can easily simulate. There's tons of very nonlinear fluid dynamics and, and, and temperature flows. It's non-equilibrium, uh, thermodynamics. So we don't know whether as the planet heats due to CO2 trapping, the level of cloud cover gets to be greater or less.

And, and even the distribution of the clouds as a function of altitude, we don't know. So.

Andrew Song: Yep. Like serious cloud thinning. Yep, yep,

Steve Hsu: Very complicated physics, uh, which, you know, if you really corner a modeler, like if a room full of physicists corners one of these climate models and starts asking questions, they have to admit this is true. Um, so they, there's lots of stuff not known about how the heating is going to affect cloud cover.

Cloud cover affects reflectivity just the same way your SO2, uh, uh, stuff affects it. And so we don't know the sign, we literally don't know the sign of the feedback loop. So, as the planet heats, does it become more reflective because of more high clouds in the atmosphere, or does it become less reflective, in which case there's positive feedback, it becomes hotter and less reflective so it absorbs more sunlight?

Or is there a natural shutoff mechanism that happens because of the evolution of the cloud cover and it's just unknown? So even the sign of that effect, I would claim is not, maybe people claim to kind of know it, but I don't think it's known for my confidence.

Andrew Song: Yeah. Yeah. So, yeah, clouds, like I, when I, I've talked to a lot of climate scientists and, uh, atmospheric scientists specifically, like clouds are such a, yeah. They, they'll readily admit that clouds are the most difficult thing to

Steve Hsu: if, if I, if, if I gave you the biggest supercomputer in the world and I said, you know, here's a cup of coffee and I'm gonna pour cream in it now.

Okay. And you just need to, I'll give you all the initial conditions of exact, uh, profile flow of the cream flowing into the coffee. You then deduce the pattern of, you know, white stuff in the dark coffee, you know, 10 seconds later.

It's basically beyond what people can do. Okay. And the idea that I'm gonna do this for the entire earth with clouds, uh, with CO2, with, uh, all kinds of like, uh, heating of the deep oceans and currents that evolved due to, you know, there's so much stuff going on that for them to say they know what the cloud like.

After the earth is heated, say one degree Celsius from CO2 in the atmosphere, what has happened to the cloud cover? What has happened to the average reflectivity of the earth? Nobody knows the answer to that. And so that, that tells that that's related to whether there could be a nonlinear takeoff where, oh, we, we get in this terrible situation where the earth just becomes Venus or it actually just stamps itself out and it's like, who cares?

Like, oh, we have a higher CO2 level, but we only didn't get that much warmer. Um, our initial projections were just wrong. And there, there are people who look at historic eras where the CO2 levels in the atmosphere were much higher than they are now, but the earth wasn't unlivable and just say, ha, these guys, you know, don't know what they're talking about.

So, this, I think at a purely technical level, uh, a true criticism of this field, not, not through any fault of their own, because it's just hard, it's hard science, which requires simulation of a very non-linear system. Um. But I just thought I would throw that out there because I think that's essential, if you're a scientist and you're interested in understanding climate, um, and its effect on human civilization, that's a key point.

And we've sort of set it up through the discussion that you and I have been having for the last hour.

Andrew Song: Totally, totally. But to be clear, directionality of SO2 in the stratosphere always cools the planet down.

Steve Hsu: Yeah, no, no, you I I, I totally get it. I should have done the homework before I got on this call, but, um, the, the, the notion that you start to increase the, uh, you know, the, the, the, the level of Yeah. SO2 dense. Well, suppose I just take normal atmosphere and I just start sprinkling in a small concentration of SO2, and then I just ask, what is the reflectivity of that gas that is presumably knowable both through laboratory experiments and probably through theory, just direct calculation.

We could. Probably figure it out. Um, and I, I would guess there's low uncertainty in that effect. I, I would guess people are very confident that the sign of, of what you're doing is known, which is different from the sign of

Andrew Song: Everything else.

Steve Hsu: aggregate effects of clouds, uh, contingent on some further heating of the atmosphere of the earth.

Right. I think we don't quite know the answer there,

Andrew Song: yeah, yeah,

Steve Hsu: Great. Well, Andrew, it's, it's been great having you on the show. What is the best way for people to contact you? We'll put your company's website in the show notes, but if some billionaire urgently wants to talk to you about some testing in the atmosphere or something, what, what's, what's the best way for them to reach you?

Andrew Song: Uh, yeah. andrew@makesunsets.com or you can find me on Twitter. Um, I'm just making, uh, our, our company profile. I, I'm, I'm the one who monitors it and, uh, tweets out of it. So,

Steve Hsu: Okay.

Andrew Song: just made some, made sunsets. Exactly. I mean, it's happened before, which is kind of crazy. So like, yeah, like Twitter's a great channel for that.

Um, like you don't find that on LinkedIn. Everyone's getting hounded on LinkedIn for trying to ask the billionaire for a job rather than,

Steve Hsu: Yeah, I,

Andrew Song: geoengineering experiments.

Steve Hsu: I've known Neil Stevenson for a long time. He actually, uh, he, he attended the same, he grew up in Ames part the earlier part of his life in Ames, Iowa, where I grew up. So he actually, I think, attended the high school that I attended. Um, he would be a great spokesperson for you if you, because he's, he's pretty technical, so if you walked him through all the details and he became convinced directionally you guys know what you're doing, then you would be a great spokesperson to spread the word that this is actually a viable thing that needs more work.

Andrew Song: I, that'd be amazing. Uh, if, if you can set up an intro, I, I'd love that. Um, yeah. So,

Steve Hsu: You guys? Well, I don't even think you need me because if he blogged about it, you should just, I'll, I'll give you his email address. You can just, you can just ping him and just say, hey, we're the guys, can you help us spread the word that this makes sense, or, we'll, we'll educate you about it and then if you believe us, then maybe you can help us.

Right. So,

Yeah.

Cool. Um, Andrew, thanks a lot for being on the show. I wish you guys all the best.

Andrew Song: Thank you so much, Steve. Have a great one.