Sama Senra — Sam Altman on Building OpenAI to a Billion Users

Reading copy · always available

The conversation, with a map.

The outline on the left jumps to each theme and cut. David's provisional speaker labels are off by default; turn on the blue highlight when you want to inspect the attribution.

Watch on YouTube ↗Transcript generated with Option AFK ↗

Want to make a transcript like this? Try Option AFK.

Likely David (opt-in)Unmarked = not confidently attributed

Tobi Lütke as an AI-native CEO

Hands-on experimentation, product feedback, and the question of whether AI will put software businesses up for grabs.

01.01

The hands-on CEO

Why Tobi is already working inside the future he is selling.

David?I just brought up Tobi Lütke and the fact that uh I I recorded with him previously. David?Why'd you say that you think he's one of the most interesting CEOs right now? One of the things that struck me the most about Tobi is in the very early days of AI, and then at every moment along the curve of its developed, he has been the most forward-leaning CEO. He's in there like writing the software himself. He is experimenting with it. He like sends us extremely detailed feedback on the product offering, on the capabilities of the models. He was before anybody else was saying this, he was like, We are not an NPC company, and thus we are going to adopt agents. Otherwise, you know, we're totally screwed. We're gonna build it ourselves. Every time I talk to him, he is at the edge of what anyone, CEO or not, is doing. He builds himself, he understands it, he has like a great deep feel, and he is always six eight months ahead of like any other CEO. Do you remember when he wrote I think this is probably like a year and a half ago, maybe 2024, he wrote that letter saying that like the first thing you have to do is see if AI can solve your problem. And then even back then, it was my, I don't know 18 months, 24 months ago, people went crazy. They thought it was ridiculous. That was ridiculous. This is my point. Like he's just consistently been ahead. He has been correct. He's leaned in. He's very no bullshit. So there's like no hype. There's nothing other than like, here's what it can really do right now. Here's what I think it'll be able to do soon. Here's how I'm gonna push the company here. And just like extremely deep understanding of where it's at. I never even thought of that. How much of a benefit it has to be for somebody in your position where you have somebody like that giving you intense and very direct and clear product feedback. A lot of people send product feedback. He is the only person at the intersection of like CEO of a large company and extremely accurate, detailed on the cutting edge feedback. Yeah, he told me, I don't know if it was on the episode uh that we're if it was in the episode or if it was after, but he said he was very um adamant. He's like, we're gonna look back in two thousand twenty six and I wanna actually your opinion on this. I didn't think to talk to you. About this. We're gonna look back on 2026 as the year that every business was up for grabs. He said somebody was going to build the AI native version of Shopify, and he said, and it's going to be me.

01.02

An AI-native Shopify

Rebuilding the company around agents, feedback, and current capabilities.

David?And so at night, he is literally trying to rebuild if if you started from scratch, what would you do with the current technology? That was the other thing I was gonna say about him is he does it himself. Like he is using these tools himself. He's writing software himself, he's trying the models himself, he is trying to like reimagine his workflow flows himself. Most CEOs, when you get to that level, have like teams of people that are managing teams of people that are trying to implement the thing and they're trying to like make you happy and they're trying to like, you know, smooth the rough edges. I think it's very hard to get the feel if you're not actually doing the thing. And he he does it so hands-on all night long, as far as I can tell. I'm not sure if 2026 will be the year that every business feels up for grabs. I might disagree with him a little bit there. Uh but but I I I get the spirit of that and I do understand that it feels like that's happening.

01.03

How much is up for grabs?

The first hard question: whether software businesses can be rebuilt from scratch.

David?Do you think that's even possible? David?Like whether it's 2026 or 2046? I mean, obviously not literally every business. I think there are some things that are very anti-AI. Like the better AI gets, the more some businesses that have nothing to do with AI, I think will be harder to compete with because we'll really want these like authentic non-technological experiences, or we'll care more about sports teams or whatever. So no, not not everything. But I think there will be many software businesses that are very up for grabs. David?Would you disagree that on the timeline then? David?What do you mean? I disagree on the timeline. I think it's going to take a little bit longer. Okay. David?Can you say more about that?

Why adoption lags behind capability

Economic inertia, old computer habits, and the missing iPhone moment for AI.

02.01

The economy has inertia

Technology can move faster than institutions, habits, and businesses.

I love startups. Like I think startups are the coolest thing in the economy. And uh I've spent my career trying to like really understand startups. And I thought when we got to GPT-4, which was back in twenty twenty three, I think. Uh the That very quickly after that, there was going to be much more disruption in software business being up for grabs right away than it turned out to be. And the thing that I think I was wrong about a few things.

02.02

Habits are the bottleneck

The technology is here; changing how people work is the slower transition.

Uh but one of them, in terms of the speed, one of them is the economy just has so much inertia. People keep doing the same things they're doing. They keep buying from the same uh you know company. They keep sort of wanting to use their tools in the same way. I think that's actually a positive in many ways and it's gonna make this big transition in front of us go smoother and slower and I'm grateful for it. But I think it means we've all been too ambitious on timelines, even with this incredible technology. I think AI is one of the most incredible technologies humanity has ever invented. Society and the economy will adapt more slowly. Yeah, it's funny, we were talking before we started recording that there's all these parallels to history. Obviously, I read history for a living. When you were just talking, I wasn't even thinking about OpenAI and AI and Sam Altman. I was thinking of like reading this biography of Larry Ellison and in like the 80s, he was just like, guys, this isn't a software problem problem, it's a people problem We have to convince them like we can install software, they they're not using it. We have to change their behavior. The technology is there. It's like we have to now adapt humans so they actually start using the technology. My own example of this was uh after Netflix came out and started shipping DVDs, even before they started streaming, it was amazing to me that people still went to Blockbuster. It was incredible to me. Like I would just watch this because I kind of I drove by a Blockbuster on my way to and from school. And it was amazing to me that people still did it. And you know, like that is an example that has stuck in my head of like force of habit and the way people do things and changing behavior is just much harder than like the tech nerds realize. David?So if we go back to this uproar of Tobi writing that, you know, open letter or that letter to the people inside of his company, you're adopting this faster than anybody else because you're partially inventing them, right? David?So like do are is there something where you're actually shocked at your own? behavior or like I know there's a better way to do this. I'm even creating the product that could be better and yet I still can't get over this like habit this force of habit. 100%. Okay. I I love that I only've ever asked me this before. I have been waiting for this question. The thing to me that feels most psychologically inconsistent about myself is that I have for 20 years been using computers the same way. I now have a magic thing called codex, so do you, so does everybody, that means I should completely be using my computer in a different way. I should not be clicking around, uh, you know, pasting from one messaging app to another. I should not be scrolling mindlessly through my emails and trying to figure out which one is like least painful for me to open and respond when I don't want to be dealing with it. I should not be like keeping a to-do list and doing sort of this like these road computer tasks in the same way that I have for so long. And yet there's like something in my mind that is encoded that like doing this kind of stuff is what it means to work and what it means to be productive. And if you asked me, I would never say I like doing it that way. I would in fact I would say the opposite and I and I think I would mean it. But like by revealed preference, I have a better way to do it now. I can do it faster. I can be using Codex for more of just like my day to day, like gotta get through this stack of emails, gotta do the stuff on my to-do list, gotta, you know, deal with all these things. And I still do it that way. And it makes no sense, other than I must like secretly like it or feel good about it. David?What do you think is gonna has to change for you to actually adopt your own product in a more deep way? I don't really know. I I I mean it's happening gradually, and this might be the right answer, which is these things have to happen gradually and totally changing someone's like ingrained habits and workflows is difficult. Uh I think there are better products we can build with this technology that will They will make it more seamless to do that. But right now it feels like Like we're all kind of straddling these two worlds of, you know, we still have a computer, we can use the old way, and we have Codex that can use our computer in this amazing new way. And we're like not sure which to use when for what. And I think this is mostly a product failure. The phase that we're in now reminds me of like smartphones before the iPhone. I was like an early adopter. I had like a palm trio in you know, two thousand three or 4 or whatever. I never had a sidekick with them. I thought I wanted one. I never had one. And a lot of the technology was there. Like it was missing multi touch, but mostly it was missing like the product ideas that made the iPhone the iPhone. And I feel like we are now in a world where we have all of the technological pieces, but we have not had the iPhone moment of like completely changing how someone interfaces with technology. David?We were talking about Tobi's like Tobi's out here building these himself, right? Uh we have a mutual friend in Josh Kushner. He says he's like there's a there's a big comparison to be made between the way that Steve Jobs thought and the way he ran his company to the way that he thinks that you do. He wasn't the one obviously writing the code, he wasn't the hardware, but he look he's like, I am patient zero. I am making products that I myself want to use. And essentially like everything that we saw with Apple is just basically what he wanted. There's this great story in one of the books um where the he He they he w they were supposed to have a meeting on uh I think one of the MacBook like the new MacBook laptops. And the team prepares like all this huge presentation for Steve and they're like really nervous because of his commanding presence. And he walks in, they think it's gonna be like an hour meeting. He walks in and he shows them the laptop and he's like on, off. Press the button, it comes on, presses off like immediately and then he then tries to open up the the MacBook. There's like a delay. He goes, make this meaning the MacBook like that. And then walks out the room. And that's the whole meaning. Like that there's a lot of examples in uh the history of Apple like that. David?How do you approach it? David?Like how do you improve the product? David?Like are you just doing it through your own needs? David?Like how do you think about this?

OpenAI's highest-leverage priorities

Models, compute, chips, energy, financing, supply chains, and why research resembles startup investing.

03.01

Models and compute

Why research progress depends on chips, energy, capital, and supply chains.

Most of my effort right now is on research and compute. I would love to be able to spend more time on product. We have great people thinking about the product here, but the most important thing that we can do is to create smart models and to be able to run them efficiently and abundantly for a lot of people. If we can get that right, I believe that everything else will follow. Philosophically, I'm very inclined to say, you know, try to find the like the high-leverage difficult problem that will continue the exponential. And for us, this is like models and compute. I also just think those are like problems that naturally suit me.

03.02

The infrastructure problem

The physical constraints beneath an apparently weightless software revolution.

David?Why do they naturally suit you? To scale compute in the way that we're doing this requires like it's a complex supply chain. There's like a lot of interesting partnerships uh to figure out which I like doing. There's like interesting financial challenges of how you're going to uh finance what is probably already or at least rapidly becoming the most expensive infrastructure project in history. Um the technology questions that go into building our compute at this scale, from you know, designing your own chip to the supply chain of fabs and people that make racks to sort of the power systems for these things. I'm always been interested in energy, uh, all come together. So there are a lot of problems that are interesting across technology, business, policy, supply chain, logistics, all together around building compute at this kind of scale.

03.03

Research as investing

Power laws, outliers, and judging ideas before the evidence is obvious.

So I used to be a startup investor. And the thing in my career that I have found closest to startup investing is managing a research program. There are all these ways in which they're really different too. Like, you know, the average researcher and the average founder have, I think, on the surface, look different for obvious reasons. But there's like a lot of similarities about how you find the non-consensus bets, how you decide where to have conviction, how you understand what exponential growth looks like, how you manage like outlier talent and how you how you identify it even more. This is the research building that we're in and it's where I sit. Great. Say more about why the parallels between where you learned and startup and With doing research. One big one is the power law. So people talk about this all the time in investing, which is you have to kind of like reprogram your brain, because we don't seem naturally well suited to think this way. Where you know your best investment will outperform all of your other investments put together. Your second best investment will outperform everything else put together after that. And AI research at least is like that as well. When we started, people thought it was totally unlikely or almost impossible that AGI was possible. David?What year is this 2015? 2015. I mean, we just got you know hammered in the by like all of the intellectual giants of the field for saying that we were going after AGI. And then when we started really focusing on large language models, we got hammered again saying this is completely ridiculous. And we understood, or I understood, at least from my kind of like startup background, and I think other people understood in other ways, this point of high risk bets are okay as long as you take the ones where if they work, it's super valuable. And research research looks this way. The kind of people that make great researchers are sort of non-consensus, fresh approach, high energy sort of non-standard is the word that keeps coming to mind. Uh people you gotta say more about non-standard. David?Can you be more specific? David?Are they spiky?

The value of real non-consensus thinking

OpenAI's 2015 bet, power-law thinking, and the difference between conviction and performative contrarianism.

04.01

What non-consensus actually means

Independent conviction is different from making a slightly different version of a crowded idea.

You don't want to fund a founder who has A very slightly different take on the same idea as the last thousand people you talked to has tried to convince you and maybe convince themselves that somehow they're completely different and doing something totally new. But it's mostly like trying to, you know, fit in with the herd and be on the same track as everybody else and do what they're supposed to do, which is start a startup. And, you know, they've heard Peter Thiel say enough times that, you know, there's like something that you're supposed to be doing different that they kind of of try to emulate that, but they don't really mean it. Like it's it's very clear to me when you have someone who just thinks differently than most other people and has is is willing to stand by convictions that are very unpopular, may well be wrong, but if right, like at least they're gonna be really right, versus someone who is like a you know, thin veneer on the same idea that everybody else has.

04.02

OpenAI's original bet

The 2015 decision to pursue AGI when very few groups were doing so.

In 20 late 2015, when we were starting OpenAI, there were very few AGI. efforts in the world There was DeepMind, one or two others that I can think of. It was like a very non-consensus thing to do. In that same year, there were probably odd to pick on it just because it it came to mind, but there are other categories too. There were probably many, many thousands of founders starting photo sharing apps. You know, that was probably like not as good of a thing to do. Today, a lot of people want to start AI labs. There are some handful of people, you know, two, three, whatever, doing something completely new. Um, that actually wasn't possible until the AI got this good, but uh doesn't seem like a good idea yet. And like that is the thing that as a startup funder, I always wanted to fund, and the thing that mostly worked for me. There's a similar thing for researchers. There were a lot of researchers that would chase whatever the last thing was that worked. And there were a small number of researchers that had high conviction towards a new idea. And uh I think we were and are the best research lab for those people. I remember talking to Demis about this a few months ago and he thought, you know, it's in terms of like chasing after the way that you guys are. It's like, you know, there's the big three players, the the money, the capital required is like you're not, there's not gonna be like a fourth bigger player. But he's like there's like this 10, I forgot what the number was. I'll just make it up. Say 10% chance that there's just some monk researcher that's gonna approach it in a way that it's just this angle we've never even considered. Totally. I don't know how to put a number on it, but there is some chance. I don't think he I'm making the number up, but it was like a small percentage. There is some chance. I want to tell you about the presenting sponsor of this podcast, Ramp. I have been reading a lot about SpaceX lately. SpaceX is one of the most valuable businesses in the world and one of the main themes in the history of SpaceX is constantly attacking and questioning your cost. Ramp helps many of the most innovative businesses in the world do exactly that. The median company running on rampenses by five percent. And one thing SpaceX has demonstrated is that a religious dedication to controlling costs can help actually increase revenue because you can pursue opportunities you couldn't otherwise. And we see that in the ramp data too. The median company running on ramp also grows their revenue by 16%. So when you're running your business on ramp and your competitors are not, you have a massive competitive advantage that compounds over time. Ramp is the only platform designed to make your finance team faster and happier. Many of the top founders and CEOs I know run their business on ramp. I run my business on ramp and you should too. Go to ramp.com to learn how they can help your business save time, save money, and grow revenue. That is ramp.com. I found one of my all-time favorite quotes when I was reading the book Zero to One. The quote says: The single most powerful pattern I have noticed is that successful people find value in unexpected places, and they do this by thinking about business from first principles instead of formulas. That is exactly what AppLovin has done with their advertising platform. AppLovin connects you with over a billion potential new customers inside mobile games. AppLovin allows you to capture undivided attention. AppLoven ads are full-screen video ads that are watched for an average of 35 seconds that is retention That blows other ad platforms out of the water. And you can launch on AppLovin in minutes. You set the goal, and AppLovin achieves it. There's no complex setup, no expertise needed, and AppLovin scales quickly. They can put your ads in front of over a billion potential. Customers. Other businesses have seen immediate results, have scaled to hundreds of thousands of dollars of spend per day and increased their revenue by millions. So you want to get started quickly before all of your competitors are on AppLovin. And you can do that by going to applovin.com. That's applovin.com. David?Okay, so what is confusing to me? You went from founder to investor to back to founder. David?But why in 2015, like what got you interested in artificial intelligence to begin with that you're saying, hey, this is such a non-consensus thing? People think I'm fucking crazy. I'm gonna do it anyways.

Altman's lifelong interest in AI

Childhood computers and robots, early AI disappointment, empowerment, science, and abundance.

05.01

A lifelong attraction to AI

The childhood fascination that preceded the modern wave by decades.

Well, I had been interested in AI whole my life. I was like a very nerdy kid. I like you know, I was like the kind of kid that spent Friday nights like playing on my computer and watching sci-fi, reading sci-fi. And I always thought that AI would be um like the most amazing, kind of craziest thing. I never thought I'd actually get to work on it, but I always I always like loved it. Uh I even came to college sort of to study it. I worked in the AI lab. The summer between my freshman and sophomore year. And nothing was working. In fact, like very memorably a professor told me, could try all of these things. There's all these directions. The one thing we know doesn't work is deep learning. We tried that for a long time. You know, it's the it's the kind of most guaranteed way to have a bad career. And you know, I was like an impressionable freshman in college, whatever. I I like I assumed that was true. So, you know, pursued these other things. It was clear to me at the time. This is now like kind of two thousand five. That AI was not working. And I happened to like accidentally get into startups, but then very much fell in love with it. And so I I wouldn't even call it a career detour because it was super helpful uh looking back and becoming like a startup investor was was great. The sort of normal career path in Silicon Valley is you're like a or not the a a pa a common path is that you're a founder and then you like sort of similar time. And you become an investor. We don't want that. I heard you say that you're going to work on this for the rest of your career. Yeah. I'm hopefully we're starting, we're going to come on the show multiple times. I'm going to hold you to this. We don't need more founders that retire and invest. We have a lot of different. We have too many investors. But what I was gonna say is I the the fact that I got to go in the other direction. I was like an investor first, and then I ran a company is pretty unusual. Very unusual. And I'm super grateful for it. Uh, because the like you get this unbelievable set of learnings and pattern matching if you really study and watch companies as an investor that have Has been super helpful to me running OpenAI, but it's it's like it's the opposite normal direction. So it's just like a very rare thing, and I strongly recommend it.

05.02

The investor detour

Why trying to be a startup investor clarified what he wanted to build.

David?Why is it helpful? If you're running a company, you have faced some number of similar decisions in your past, like some number of like crux decisions and you've seen what works and what doesn't. And if you have to like, you know, make a high-stakes strategy shift or like fire an executive in a really messy way, you have like whatever your own limited previous experience was over the last five or ten years that you've been doing it. But as an investor, you kind of watch all the crux moments. So you don't get the kind of like operating practice that you do just day in and day out running a company, but you've seen a lot of the like big crux moments. A lot. Kind of all day long, you see those. So just the the the wealth of the data set that I had there was awesome. David?So that plays in your head when you have a decision make? Yeah. I'm like, oh, this is what happened when this company had a similar thing, or I saw this founder make this mistake, or this founder got it really right. We were talking about that you studied the Industrial Revolution. We were talking about some great biographies that we both read earlier. Uh my friend Daniel X says this about me because like I think the benefit of me doing this project uh my other podcast called Founders for 10 years is like so he's like, you're like an L L M trained on history of greatest entrepreneurs, yeah. But with the temperature turned up 'cause you're fucking crazy. Because I'm like super passionate about it in a weird way to be like obsessed with dead entrepreneurs. Uh but it is helpful. It's like, oh, like I'll be talking to a founder and they'll talk about something they dealt with. I'm like, oh well, like Carnegie did this and Rockefeller did this and like you might want to try this. David?Like and do you find that you have like one big inside all of that or it's just like for any given scenario you have like what all these people did and how it comes together? I think it's depending on the personality of the founder. Yeah. Right. And so like when I was reading, I've read your blog for years. I think like you're a great writer. You're very succinct. I like like how the brevity is just really appealing to me. And I love numbered lists. It's weird that we both write in the same way. And I feel like I'm reading your blog and I'm like this is the exact conclusion that I would come to based on all the reading. And like there's this handful of principles that could be applied, but it's like it really depends on like who the founder is and what they want to do. This is what I'm trying to say. It's like let's go back to what we were talking about. I still like like you're you're making a big jump because you know you're one of the from what I hear, one of the best investors of all time in Silicon Valley. You could just be rich and not really have to work because investors are kind of lazy. Kind of not. No, it's pretty it's I I I having done both, I think I can say it's like much, much, much harder to run a company than be an MP. Exactly. And that's what should people should be doing, in my opinion. I agree. So uh then you're like, fuck that. I'm gonna not take the easy route. I'm gonna do the hardest thing ever, the thing that people think is impossible, thinking it made fun of. Yeah. I still need to understand. David?Okay, so you're into an kid, like why would it appeal to a kid? You were living in St. Louis at the time. Yeah, I was living in St. Louis. David?Why would AI appeal to you back then? Well, I think like it appealed to every kind of computer nerd. Like I don't think it's that unusual about me. It just felt impossible. I think most people would say like of course that'd be the coolest thing ever, but it's totally impossible. I think the like the weird thing about me was like, okay, let's try. Uh but I think everybody thought it everybody would think it's awesome and something to something to go for. David?So wait, that was a personality trait of yours as a kid? That I think even if you told you that you couldn't do something, like your initial response was resistance. David?Not resistance, but like, uh, are you sure? David?Like why not? Let's try, let's see what happens. Maybe I can't. Maybe we can.

05.03

Impossible things and empowerment

Optimism about science, abundance, and giving people more agency.

I was like a very optimistic kid. And also, like, the more something seemed impossible, like the more intrigued I was. The idea that we could invent a technology that would let us do everything else, that would just empower people in this way that no other single technology could. That always seemed like innately incredibly appealing to me. It's like, I want that thing, I want to be able to do everything else. Like I think another thing about that I kind of was like a personality trade as long as I remember is like, you know, it's like it is interesting to like really give people a lot more power, a lot more ability. In some sense, this is the whole arc of technology. And I was for sure always a technology nerd. But like AI is the strongest version of that I can imagine. David?What did you think that it would enable back then? Like when you were a kid, like this seems like a cool technology. I want to do X, I can't do X unless AI is invented. It's always hard to remember like how much of this is the stuff that I actually thought at the time versus like how much you're trying to build right now. Yeah, like how much my current work has like colored my memories of it. For sure. As a kid, I was like very into robots. You know, we had like a robots club in my school uh and the robots at the time were like laughably I even remember at summer camp, we had this like little turtle that you could control with a computer on the floor or on the table. And I thought that was just the coolest thing. There's something about like physical stuff moving controlled by a computer that I always thought was amazing. Now I am extremely interested in what AI can do to advance scientific discovery. In my memory as an adult, I think I thought that was cool as a kid too, but it feels just implausible and I assume that's an example of where like the memories have gotten more colored. But now the fact that we can have AI go discover new physics and cure diseases and what it's already doing from math. Like I think this is I think this will be one of the most important areas, even more important than automation of other tasks that AI can do just to help us understand. More things we were talking earlier about this this book, The Beginning of Infinity, and rereading that book from today's vantage point, I'm like, man, AI is really gonna help us do this, do this important thing of understanding everything for as much as we can. I was definitely interested in the sort of like Star Trek version of like huge prosperity and abundance and and what AI could do to drive that. Maybe the memory of being interested in science is more real. I also loved science and just the this idea that we could like, because we were smart, we could figure out how to understand the world and make predictions and do things that we couldn't without this deep understanding. I don't know. That seems like innately awesome. It's interesting how consistent over time what humans want from AI. Cause like the something you're describing is very similar, I just reread the biography of Claude Shannon for the second time. And I had forgotten, because I hadn't read the book in maybe five years. And I forgot that him and Alan Turing used to meet every day for coffee when they were both at Bell Labs. This is like nineteen forties. Yeah. And they would just talk about AI. And they were both obsessed with their it's they thought it was inevitable back then and they thought it was gonna happen like fifteen years from there.

What AI will—and will not—replace

AI solving problems once thought impossible, while humans continue to value human connection and analog life.

06.01

The Shannon/Turing vision

The old dream of computers that could reason, discover, and extend human capability.

David?So like 1955, that we're gonna have computers, which didn't exist, right? They had the analog versions, uh that are that are going to be smarter than humans, and that anybody that thought that wasn't gonna occur, they thought was absolutely ridiculous. David?And they're like, well, what would you want the computer to do? He's like, solve math problems, write poetry, like solve uh cure diseases. Like you hear this over and over again. I have read a bunch of things that those guys wrote at the time, and I am so sad they are not here to see it. Because they were so right about everything. We're finally at the moment where yeah, it is solving novel math problems. It is discovering other stuff. It is, you know, you can argue about how good or not, I would say not very good, but it is writing poetry. It's here like to what these guys I think they would have said. All right, you've done it. Like this is it. We've got it. Um and that would have been so cool. Yeah, this is the weird thing where everybody's just like, oh, it'll never do X. David?Like I talked to uh people in the music industry, it's like it's never gonna make great music and and then they're like, Well Do you think it's gonna like make a podcast? I was like, of course it's going to. It's going to do everything that we can do, at least to I would say better than like even right now, like better than what we can do. It's a very bizarre thing where it's like it will never surpass what is happening at this current point that I happen to be alive. There's a deep human psychological flaw there. But here is a I think a more interesting question. Let's say it does make a great podcast. You know, two AIs are having a more interesting conversation than you and I are.

06.02

Human connection

Why more capable machines may make real human relationships matter even more.

David?Do you think people will care? Or will they want the one with the real people because we're all obsessed with people and the fact that it's not real people Yeah, for this is like more interesting. It's like, oh, these two people that that I may be predisposed to like or dislike or having conversations that's interesting to me. I think for like strict like reference, like maybe my other podcast where I'm just saying, hey, this is some interesting ideas I read in this book, that could maybe get disruptive or other cases, but I especially for people that were born before this happened, maybe it's different for your son. David?You know? Maybe for me, it's just like I think humans are gonna always be drawn to humans. I really deeply believe that. I think there's like a lot of other jobs that could face significant transition, but stuff that's about people, stuff that's about people's connection, connection to people and people liking other people, that stuff feels like it gets uh more valuable in the post-day world, not less. I may be actually the wrong person to talk about this though, because I kind of deeply desire, even though my entire work is digital, it's you know, broadcast all over the world, it's just like I deeply desire like more like more of an analog life. I like reading physical books. But it's like when I was talking to Kelly, I was like, I didn't get on Zoom, call me. me or we'll talk in person. Like I like physical shit. I don't like I'm like that too. I don't ebooks. Yeah. I don't like Zoom meetings. I like to be like with people in the real world. I definitely think there's a subset of weirdos and there's probably a lot of them that live in this city that you know don't like humans and only want to com uh communicate with computers, but it's like I think that's a tiny percentage of humanity. I think it's a tiny percentage of humanity. I th this this is why I think like the world is on the whole not gonna be that different even with superintelligence. Like people are still gonna be very fundamentally wired to care about other people to want to be around. other people um to interact with other people. And you know, there will be some people who just like get obsessed with the models and just think humans are in the way or a you know danger to be contended with or whatever. And for most people, it'll be the whole point. I think it's very important that when we do find people like that, to them be called out and make sure they don't acquire power. I I certainly agree with that.

AI safety, control, and human power

Loss of control, centralization, cautious deployment, real-world feedback, and the case against fatalism.

07.01

Two risks

The tension between losing control of a powerful system and concentrating power too tightly.

Maybe the two big risks that I'm most worried about with AI, which are a little bit in tension. One is like a uh a loss of control, where you know, AI somehow just becomes too powerful in a way that we can't guarantee the control we want. And the other is power gets too centralized, where you have, you know, one company or model or person with too much power. And in in both of these, the fundamental thing is like I think it's a very anti-human position for either of these things to happen. The right approach is to say like we want people deeply in control of the future. We want people deeply empowered. People are the whole point of this all. Like that this is we are not gonna sit here and, you know gradually hand over control to an AI model because we don't trust or like people. And you know, uh it's like a very misanthropic thing to say we're gonna just like put all of our trust in this model and let it have all the power on decision making over the world. But I think there are some people in the world who think that's the right outcome. There's like another version of this, which is because we don't trust people, we have to limit who gets access to this technology and how they can use it and all of these terrible things could happen. And and you know, out of fear of those, we are going to concentrate power in the hands of a few companies and they're gonna, you know, we're not gonna let other people use this, but we'll give them some benefits. Like my my caricature of this is I think there's some people in the AI field who who effectively say, we're gonna give the world a cure to all disease and we're gonna make stuff really cheap in exchange for people giving up their their autonomy and impact over the future and power, uh, and also in the name of safety, and also like just absolutely rampant inequality. Like there will be people that have access to huge amounts of wealth and power, and other people just get a pretty good everything. And this is like a terrible sales pitch. This is a very anti-human sales pitch that somehow people feel willing to make. David?Aaron Powell Why do you think they feel willing to make that? I think it's like fear and power. I I think you can when people talk about the risks of AI, I think there are a lot of people who are so nervous about the magnitude of those risks and get so taken by that and feel a need to protect the world from that, um, that they're like, you know, we should trade off a lot of liberty for safety here because this is unlike other risks we've seen. But then I think that also ends up kind of like a way to justify a lot of power seeking behavior. Everything when I read, like when I was just saying the Claude Shannon, what Claude Shannon was saying or Alan Turing, at least in the books that I've read, it's more of like an optimistic like we're gonna invent things that make our lives better and can do things for us. You are totally right that if you go back to the Claude Shannon, Alan Turing era, uh they talked about how wonderful AGI would be and all the things that it would do. Uh and when we started, we really had a lot of pressure from the Doomers.

07.02

Doomers and solvability

Taking the danger seriously without treating it as a reason to give up.

Now, the part of the Doomers that I agree with is that this is a powerful technology, and we should err on the side of side of safety, and we should act with caution at each level of technology. The part of the doomers that I don't agree with is that it's an unsolvable problem. If you go back to the beginning of OpenAI, um, I think there would have been two widely held opinions. Number one, not at all, and certainly not in 10 years were we're going to build something that was very AGI-like. And then conditioned on if we didn't have a lot of We're not going to be able to make it safe. You know, if you had an AI that was smarter in many ways than a lot of the smartest people, most of the smartest people, then you know, the Doomers would say surely at that point the world would have been destroyed, the alignment thing would have failed, and there were just these very confidently held positions about what would have happened. A decade on uh we have built something that I think most people would say at the time would have said us very AGI-like. And you know, a lot of good things have happened and the kind of crazy bad predictions of the world ending have not happened. So I think that should update people's predictions about the future. There are still higher stakes challenges in front of us to solve. But our approach, this is another thing I learned from startups, of the way you do things is to put things out into the world, get feedback from real customers, see where they break, see where they don't break. That is the way you make a good product. That is also the way you make a safe product. And we have made way more progress on AI safety than I think most people thought we would when we started. David?Why? Because so many people, there's billion people using your products on a weekly basis. And each time we get a new level of model, we put it out in the world and we see what, you know, what works, what doesn't work, where people need us to relax the guardrails because they have good things they want to use it for, where we have alignment failures, where we have safety systems failures. ChatGPT has only been out like less than four years. Billion people use it. And sensitive important stuff. And the fact that we can deliver something that is broadly considered safe, of course, there's issues with it. Like in that short of a time frame with such a powerful technology, I think there is no way we could have done that in Ivory Tower. And you know, this is how I believe you build good, safe, robust, useful technology and products. And I think is a great a great learning of Y Combinator. And it would have seemed to most of the AI safety people, you know, totally impossible to get to this stage and still have to do that The level of safety guarantees we have. No. I do think it gets harder from here, but I don't think you're gonna solve it by disconnecting yourself from reality. David?Why does it get harder from here? Because we're as about the smartest people in the world are as about as smart as the smartest models in the world and that's gonna flip right now. Yeah, direction I think that's right. I think that the the models are just so incredibly capable and improving on such a steep trajectory that the unknown unknowns. Maybe they don't get harder relatively, but from an absolute perspective, they they seem harder. And I think we'll have to make a bunch of difficult decisions about, you know, when we delay development, when we sort of say, okay, you know what, let's have contact with reality now or let's let's wait longer to really study this more. Sorry to someone recently uh and something that stuck in my mind is that the you know the FAA FAA has helped make flying incredibly safe. Flying on the surface seems like this extremely dangerous thing. And you probably get on an airplane without giving it much thought. And this was certainly, you know, airplanes are not that old in the long trajectory of human history, and this was certainly not the case at the beginning of airplanes. They have extremely robust accident reporting, extremely clear-eyed. You're never, they never try to like, you know, hand wave over something, they want to extract as much information as possible. And in some sense, I think with any new technology, an approach like that works works very well and is often underappreciated.

07.03

Safety through deployment

Deploy carefully, learn from real-world failures, and improve the system.

So when we started deploying our models, when we said we're gonna put ChatGPT out in the world, we know the model's imperfect. Um we know it hallucinates, we know it can do these other things, but we also know that The world has got to experience this technology. We've got to learn how to make it safe. And we gotta put the power in people's hands. We cannot just use this to impose our worldview. We cannot use this to go sit in a lab and try to think through all the impacts, which won't work anyway, because society and the models are gonna co-evolve. Like we have to all do this together as this as this joint product. And then we'll do very good accident recording. We will study when something goes wrong. We will put out a very clear post-mortem. We will learn as much as we can. We will not only improve our own technology and products, but we'll try to share those learnings with other people building AI. I think it's worked surprisingly well so far. And that was like Good examples from history of technology, good examples from startups. Deal is how the best founders turn the world into their talent pool. I've been studying how history's greatest founders operate for a decade, and one thing they all have in common is they understand that recruiting and hiring the very best talent is your most important priority. A players recognize other A players, which is why top companies like Ramp, Shopify, Eleven Labs, Uber, and Doordash all use Deal. Many of the top founders I know have personally invested in Deal after using their product. And what they discovered is that Deal is the best company in the world at building infrastructure for global hiring. Deal will help your business hire, pay, and manage any worker anywhere in the world so you can retain the best talent anywhere and spend the rest of your time focusing on what you do best, delivering value to your customers. The founder of Eleven Labs has a great description of the value deal can give your company. He said we built Eleven Labs to break down language and communication barriers with deal enabling us to hire and support exceptional talent anywhere we can accelerate our innovation and bring more voices stories and ideas to every corner of the world deal is trusted by over 40,000 businesses. Learn how they can help your business today by going to deal.com forward slash CENRA. That is deal.com forward slash Senra.

Why people both use and distrust AI

Fear of rapid change, poor industry communication, autonomy, and AI's potential to create more small businesses.

08.01

Why use and distrust coexist

People can rely on AI while still feeling that the change is unsettling or poorly explained.

David?One thing that has to be disorienting for you, because I it doesn't make any sense to me. Like you know, essentially what I focus on is just entrepreneurs. David?Entrepreneurship, right? So all the podcasts I make are for the benefit of entrepreneurs. David?I'm glad other people listen, but it's like heavily focused on just trying to serve, trying to find useful information, whether it's in a biography of a dead entrepreneur or talking to somebody like you, that other entrepreneurs can benefit from this conversation, right? Or any of the podcasts that I make. So in that, there's not like a lot of entrepreneurs love AI. But like what I I'm trying to figure out, like if you can help me reconcile, it's like everybody uses AI. Everybody hates AI. David?What the hell is going on there? Well, people are always afraid of like rapid socioeconomic change. We talked about the industrial revolution earlier. I love I love reading about previous technological revolutions too. And like people did not have universally warm and fuzzy feelings to the change that was happening during the throughout the Industrial Revolution. I think it's probably a good feature of human society that we have some built-in inertia. We have some skepticism of rapid change. I think that probably helps society in times of turmoil or in times of, you know, localized craziness or whatever. Um so some of it is probably good. And I think a feature of human biology, which I believe in never trying to fight too hard. I also think a lot of the people building AI uh uh you know, have been off saying there's a 25% chance we're gonna destroy the world and yeah, we're gonna race ahead to do it because otherwise those bad guys will do it first. Or, you know, it's like, man, this thing is going to be really terrible and there's gonna be 50% of the jobs are going to go away in the next year and we hope you all are okay, but seems really scary. Like we have not, as a field, done a very good job of explaining to people what the benefits are and how the downsides can be mitigated. And we certainly have not done a good job even Even if people have had answers like, you know, saying, well, there's going to be universal basic income or work will be optional or whatever, there's been very little discussion from people about how and why it's important that people have more power and f personal freedom in the world, not less. And that matters a lot to most people.

08.02

The autonomy problem

The value of keeping people able to influence and design their collective future.

The ability of people to influence their own future and collectively to design kind of where society is going to go and the autonomy that comes with that is very important. And I don't think a lot of people in the AI field they feel it for themselves, but they don't spend much time thinking about it, reflecting on it, acknowledging how important that is to other people. And so to go back to that like character Jason of the sales pitch, characterization of this caricaturization of the sales pitch. I don't know if that's a word from earlier. It is. Um the like dear peasants, we will bequeath upon you these gifts of a cure for cancer and material wealth and wealth and some things and you know great entertainment. And you stop complaining and we'll make all the decisions about the future and just trust us. We'll be better and like dictators. Not good. Not good. As a lover, entrepreneurs and sort of like a student of what has made Made this incredible economic miracle of recent centuries work. Really empowering people to go do new stuff and to push on the things they believe in, and to have the freedom to create companies and invent technology and sort of pursue ideas and like the system around that that makes that happen, there is like nothing I believe in more strongly. And I think even if people don't see themselves as an entrepreneur ever, if they they may never want to start a big company, they do understand how important that is. And then when you hear people kind of implicitly or explicitly saying there's going to be less of that with AI, because a small number of people are going to have the power, but they're going to make great decisions and keep everybody safe, I think that's very scary to them. I also think that even if maybe most people don't want to start really big companies, a lot of people want to start smaller companies. And that has been hard. That has been something that has required a fair amount of privilege and luck and resources to be able to do.

08.03

The small-business upside

AI lowering the threshold for more people to start something of their own.

And we are about to see the greatest boom in people starting. In smaller businesses that we have ever seen. I think AI is empowering that. Now, for some reason, the field, including us, has not talked about that enough, even though we see all these signs of it and it's it's great, and we have not built enough products to accelerate that. But I think we're gonna see a lot more of that. It's funny, uh we talked about Tobi Lütke earlier at the end of the conversation, and I think that's in the episode. He he mentioned something I didn't think about. He's like, Oh yeah, you and me are in the same business. He's like, we're both trying to create more entrepreneurs, he's building infrastructure. for entrepreneurs I'm building educational and inspirational podcasts for them. And the crazy thing with what you just said, it's like not only I think out of any new industry, it's not that as an industry, the A industry, which it's doing the worst job I've probably ever seen. And I think part of it is just the ability you guys have to get out there and talk about the stuff that you're seeing and educate. There's this great book called the Intel Trinity, and it talks the story of Intel, and it's called the Trinity because the three main players are Bob Noyce, Andy Grove, and Gordon Moore. And when they, there's a great story in the book that I never forgot. They go from inventing, I think, the integrated circuits to the microprocessor. And they realized that that technology was so important and it would scare the their potential customers. So they went out, they stopped, those three people stopped doing what they're doing and went out and started educating potential customers, investors, the entire country. And they said at one time they were putting on more classes than like the local community college had in their entire course catalog. That's how much they made it a top priority. It's like we're gonna get out and educate about this new technology. David?And it's just like, well why isn't it AI doing that? I mean no excuses we should be doing more. I think we've like tried versions of this. We haven't gotten it quite right. Well you're doing it right now. We're like this is a point of one of the points of reason I wanted to talk to you about because like I use AI all the time. I think it's fascinating, but you have such like this you have a view that makes mine look like it like the the view of an ant. Like you have there's so much stuff in your head that I want to like get out and be like, hey, you have all this context. It's like and you're inventing this incredible technology. I would love to know how other people are using it. Actually won before we even get to other people, I heard you say something that was interesting that I think ties to what you just said about kinda like we're inventing technology to tech. Technologies increasing rapidly, but the adoption should be slow and deliberate. David?And I heard you on another podcast saying, hey, I'm even considering like how much should I let AIC every single thing that's on my computer? David?You want to talk about that?

The next AI product opportunity: context

Models are capable; the next frontier is giving them enough personal and organizational context to be useful.

09.01

Context is the bottleneck

Smart models still need the right personal and organizational context to be useful.

With the latest generation of models, I don't want to say they feel smart enough, because I think we should always aspire for them to get smarter, but they are pretty smart. And I feel more limited at this point by the amount of useful context the AI has on me. Like, I want the AI to know as much as it can to help me. I want it to be doing things I can't or don't want to do on my own. Like, I'm not gonna read every post on our internal Slack. I'm not going to go read every story a customer has to tell about where ChatGPT works for them or failed them. I can't. And then there's like other stuff of like I just, you know, I probably could read more research papers than I do, but like, oh, it's like takes a lot of mental energy and you know. But I would love to have an AI agent that is constantly trying to be helpful to me and that can look at and understand more context than I can or that I have time for or energy for to do on my own, and can help bring that context to bear and give me good advice when I have to make a decision. So I think we focused correctly so much on model intelligence that on the product side, we have not yet thought enough about what it means to give a model more context than any person could have and help advise that person on their big decisions.

09.02

A different way of working

The coming shift is not just a better chatbot but a new working relationship with software.

My sense is we are just on the precipice of being able to able to see a very different way of working with AI on a on a on an axis where people just can't get this good. There are plenty of like very smart people, but there is no one that can read like, you know, tens of thousands. Thousands of pages of context in some small number of seconds, and really like you use that accurately. And this is something that AI can do that just is gonna be very new and an incredible supplement. You've read all these biographies. There are probably times where you vaguely remember something that if you could remember a specific anecdote from one of them, it would like really help an entrepreneur in one moment for that particular entrepreneur right when you were talking to them. But maybe you forgot it or maybe you don't remember it exactly right. I built my own AI tool. So I use it internally. David?So you know what it is? It's only trained on since 2018 I've kept every single note and highlight from every single book that I've ever used into this database. And I would search it for that. And then when your the work that you guys do came out. Then I added so I have it trained on that, then every note, every highlight, and then all the transcripts for my episodes of founders. I use this thing every single day. That's awesome. To the to make every single episode. David?So I'm working on Claude Shannon, right? The Claude Shannon episode came out on two or three weeks ago, whenever it was. David?And I'm asking questions about all I was like, hey, what did Bob Noyce say about this? David?Or what did Rockefeller do about this? And I've I I made the episode. I read the book. I took the note. I don't remember because it was like seven years ago. It's incredible. This is what I mean. I was like, eh, it's fucking awesome. That is so cool. David?Let's get into like how you think about running the company, right? David?So you're spending your time, you said your your main focus is getting more compute and then research, right? Okay. So you want the models to be the best in the world. David?But how do you think about like, do you have to build your own products? David?Like you built Codex, right? David?What are I don't even know the the fucking product lines. David?Like where's all the revenue coming from?

OpenAI's platform strategy

A single interface plus an API, models across the cost curve, and a deliberate choice not to compete with every customer.

10.01

Platform over product

A direct interface and an API can support an ecosystem without competing with every customer.

Actually, I think we should be more of a platform company than a product company. David?Like we will build products, of course, but how many products you have? Let's let's back up. David?How many products do you have now?

10.02

ChatGPT, Codex, and the API

Where the consumer product, developer tools, and model platform meet.

We just merged ChatGPT and Codex together. So we used to have like ChatGPT, Codex, and the API. Okay. You know, and Codex sort sort of inf unfortunately named, but and that was not just coding. It could kind of do any kind of work, which confuse people. I'm confused by that. Yes. Okay. As were many other people. What I think most people want is the sort of like single interface to To their own personal or their company's AGI, they can kind of help them with whatever they need. And then the ability with an API to build anything they want on top of it. And that is the platform that we should offer to the world. We're gonna sell great AI. At every, at every point on the cost curve, cost performance curve, we will be the best. You know, you want really high-end AI to discover science, that's great. You want really like inexpensive AI to do like, you know a massive amount of volume of work that maybe doesn't require genius level intelligence. We got you covered there too. And you know, thinking about this as a sort of people talk about different ways, a new utility, a new commodity, whatever you want to call it. Like people want to use a lot of AI and they want it at a low cost and they want it to be fast and to work well and have their context and be smooth. We we got you. And then there's like a single product, which is I need to ask the AI something. Eventually maybe it's the AI should proactively offer me things. But you will have this interface, which started as a chatbot and now also has code me agents. And I think at some point we'll feel like a more persistent agent. To this AI that is running on whatever you need it to run on. But that's it. I don't think we should go build every product category. I don't think we should like go try to compete with all of our customers. I don't think we should try to like subsume the entire economy. I think we should offer this platform and try to have 100 million new businesses and 8 billion people use it in all kinds of new ways. So one kind of direct interface to the product, one API for people to use it however they want. Those eventually come more and more together too. And then it's all about what people do with it, build on top of it, whatever else. David?What mistakes did you make to have to learn that?

Focus means killing good products

Why Sora and Atlas were sacrificed to prioritize Codex, general intelligence, and upstream infrastructure.

11.01

The cost of killing good ideas

Focus is not abstract: it means choosing what will not get the team and compute.

I feel like um you've had to kill some good ideas to and sacrifice going after the great with your full like intensity and focus. Yeah, I I I think killing the good ideas, like sacrificing the good ideas to go after the great ideas is is kind of the hardest lesson for any entrepreneur or business to learn. It sucks to kill good ideas. And no matter how much you think you're gonna do it, you you people everyone. Like kind of maybe just by the nature of who chooses to be an entrepreneur, seems to do terrible at this. I'm terrible at this, I know I'm bad at this.

11.02

Sora, Atlas, and focus

Why even good products can lose to a more important opportunity.

But last year, for example, we killed Sora, which was you know good product and fun and cool, but used a lot of compute and not as important as codex where we put the compute. Um we killed our web browser called Atlas, uh, which again, I think was a great product, it was the best web browser, but not as important for us to focus on as somewhere else we could put that talent. Um in a world of limited compute, limited people, limited resources, we thought really hard and we said, you know what, the the general intelligence for knowledge work and eventually for science, most important thing we can do, anything that goes into making that upstream of generating that intelligence, building our own chip, building our own data centers, um, you know, writing good infrastructure software, certainly training models, obviously. That's all really important. But then let's just offer this AI as as a service and get people to Use it for intellectual pursuit, for work, for scientific discovery, to be more productive in their personal life. And let's have the kind of flexible general platform and not do a lot of other things.

The people who help founders think

The orangutan theory, Paul Graham, Peter Thiel, simple genius, and the value of shipping early.

12.01

The orangutan theory

A thinking partner becomes more valuable as the work becomes more complicated.

Anybody com engage in complicated work, and you're gotta be the you know top of the sorts of top of the list of anybody alive right now, need somebody to help organize their thoughts. It's extremely beneficial. You see this in every single biography. You see this history. It's like you need somebody to talk to. There's the there's actually a funny story of how extreme this can be. Uh Charlie Munger has this thing called orangutan theory. David?You've ever heard of this? Where he said a relatively smart human could go in, uh sit down with a rag and tank, tell him all his problems, tell him everything else of mind, and then, you know, the reggae tank obviously says nothing about nothing else. The human leaves and the human's better off. Just the the the idea of just being forced to put your thoughts into, you know, into into some kind of structure. Now obviously with a very intelligent partner. Munger. This role for Buffett. Buffett's one of the most intelligent people who ever lived, greatest investor of all time, still needed to organize his thoughts to somebody else. You are going through I can't think you're you're you have almost like a singular lived experience, especially for somebody as young as you are. David?So I'm curious like who plays this role in your life? David?Like who do you go to that can even remotely empathize with what the hell you're dealing with on a day-to-day basis? Kind of three categories here. One, a lot of the researchers that have been here forever, we've kind of like all been through it together and we've developed this set of like shared language, intuition, standards, whatever you want to call it, and that I have not been able to replicate with anybody outside of the company um when it comes to like the shape of what's happening and what might happen next and where the technology is likely to go. In in terms of questions of just like, you know, business and the world, for a long time in my career, Paul Graham and Peter Teal have been two of the people that I have learned the most from about lots of different phases of my career and are still the two people that I go to if I really have like a very non-obvious problem that I'm stuck on. Uh and there I have not found anyone else after a lot of looking, that has the same kind of ability to to just to like think in a super nonlinear way. Like, you know, if what LLMs do are predicting what word comes next, those are two of the people that I can predict the least what word is going to come next. And that is a super valuable skill. You go with like a, oh man, I feel really stuck and I've kind of thought through all these options. And someone that can tell you like, eh, I think none of those options are good. Here's this thing that now seems totally obvious and correct that you didn't think of. Just a completely different view that you haven't heard anywhere else. Is this like more of like a prompt for your own thinking as as opposed to like explicit advice do It's often like here is a specific thing. David?Really? Yeah. David?So what would be an example that you could share from like Peter?

12.02

Paul Graham, Peter Thiel, and simple genius

The outside thinkers who helped founders see what was in front of them.

Peter's very fascinating to me. He is very fascinating. And I it's it's actually who I thought of. The reason I thought of this question just now is because you're like we had to kill these good ideas for the great, we're cut an atlas, we're compute constrained, we have to foc fouscus fo,cus, is like that's something that it's very obvious when you listen to him talk about the importance of focus and like if you have something that's working, making it work better and going down this line, like uh taking an hour away from that to like explore something else is too expensive. You should just go deeper on what's already working. Like there's a lot of value at the extremes. After we launched Chat GBT, sort of this weird thing, because people didn't really know what to use it for. And it was growing super fast, but it felt like very unstable or kind of almost like low-value growth. Like people were using it to Just because they were like interested in talking to it and what they could do. And so there were a lot of people in the company who were like, uh, this is, you know, we gotta figure out something else. This is not sustainable value. And remember talking to them about this like list of five or six other things that we could focus on instead. This is like maybe two months after ChatGPT launched something like that. And he was like it's an obvious mistake to do anything but this besides the fact that it's growing, which is rare and great. It was not growing as fast then as it did start after. He's like the power of this is the power of the Google text box. It's like a text box you can type anything into and it does the right thing. And the fact that it doesn't match the current Silicon Valley wisdom of you know, you gotta have like feeds and you have to have like a network effect and you have to have because we had none of these things, and that's why everyone was worried. You know, you have to have like a you know, way that people are gonna build up more con this before we had memory. People are gonna build up more context, or people are gonna get locked in, or people you're gonna have all the these like all that stuff, like people have just been chasing the Google business model for twenty years and this is the first thing that's come up and you know. Clearly the empty text box worked for Google. David?So why don't you just double down on that? It's growing like it's very flexible and you know it has all of the signs other than it doesn't fit the current Silicon Valley wisdom. And I was like, okay. And so we went super hard on chat to PT, uh, and it was great. It's like a simple genius to what you just said. Sometimes there's like more complex genius, but that was an example of very important simple genius. David?What about some advice that Paul Graham or some guidance or a direction he kind of pushed you in. David?When you just said that, the thing this is like a meme for many YC founders, uh where you would go see him for office hours and he would say, you know what you should do, and he would like shake his hand, his finger like this, You know what you should do, you know what you should do. And sometimes the thing that came after that was great. Sometimes the thing that came after that was terrible. But the important thing was there was a kind of there is a creativity and open landscape and just uh, you know, let's try a lot of things. We talked about the spirit of Of iterative deployment. And we talked about, you know, what what how like in the same way startups he really I think pushed the startup ecosystem into this world of you gotta like get ship a a a v1 year and bar stuff early. And it doesn't matter if it's you know it could be much better. You'll you'll you'll get it much, much better because of the feedback to customers. David?I don't even think I asked him before we launched ChatGPT like, hey, do you think we should launch this thing? But I knew what he would say. I knew it was still early. I knew it was still embarrassing. And I knew the right thing was to like get it out and get it in front of people. David?So wait, wait, your mental model program is so complete you don't even have to do that. Just in this one case. That's that's the one where you would say like there's certainty. Let me tell you something funny. David?Right before he died, a few months before he died, I went to Charlie Munger's house and had uh dinner with him and I was like, How often do you talk to Buffett? He goes, Never. David?I go, what? He goes, we talked every day for hours and hours. Buffett can just pretend to pick up the phone call to call me and he already knows what I'm gonna do. Say. That's obviously comes after 65 years of working closely together, but I thought it's a good. That is really a funny story. No, there there's many times that I couldn't predict what he's going to say, which is that's why I think it's valuable. But in terms of the like Launch when you're embarrassed of a product. I know what he's gonna say there. Like that one. Uh that that that has been like I won't say the most valuable piece of Y s of tactical YC advice, but it's been up there. I'm astonished looking back at all of my data points of YC founders over the years, how much the ability the the like moving fast and being iterative correlates with success. Okay.

Y Combinator's lasting influence

YC as an operating system and philosophy: iterative deployment, technical founders, speed, and founder leverage.

13.01

YC as a philosophy

The cultural operating system matters more than a fixed startup checklist.

You've mentioned YC way many too many times in this conversation. I I have to explore this because we talked before. It's like listen, I'm not a journalist, I'm enthusiast. I don't have a list of questions. I'm like, I have a foul world class founder across from me. I want to know what the hell is in this person's mind and I want to like extract information out selfishly for me. So like why like I just I'm shocked at how much you reference it in conversations. Like how impactful going through YC and then running Y through YC, so being you know affiliate with them has is clearly on your life. David?Can you like you expound on this? There is some band that wasn't that successful. They didn't sell that many albums, I mean, but they like influenced all of the musicians that came out of the band. I think it's called the band. Literally the Rick Rubin told this story. I think it's is it might be the Velvet Underground. But but you know the idea I'm getting whether it's from the band or whatever. It's not fair to talk about YC in this way, because YC measured by like traditional metrics and market up created or whatever is one of the hand would be one of the handful of the most valuable tech companies. But the degree to which YC totally influenced everything that has happened in the last 20 years of The tech industry and startups, entrepreneurship, whatever. I think is is only like sort of understood. OpenAI is an example of that. Not just from like how we have shipped our products in the world, but like the philosophy of how we run our research lab. I think if you go talk to like many of the other people running like this generation of large tech companies, they would tell you similar stories, even if they didn't YC. David?But what's happening there? David?Is it an operating system that YC is giving you because you hear, you know, do these five things or whatever, or is it more like a philosophy of building companies? This is the confusing part for me as an outsider. I mean there there is all there is some of the operating system of what to do.

13.02

Iterative deployment

Technical people in charge, shipping early, and learning in public.

But I think it was the philosophy of how to run companies, the idea of iterative deployment and technical people in charge and sort of being willing to bet on young people with a lot of energy, ambition, but maybe less experience throughout all levels of the company. And then it was also the change, the related change to the whole ecosystem that happened. In the pre-YC tech ecosystem. So if we you know ran the clock back to 2004 and then projected technology forward to 2016, um, but not anything else about the shape of Startup ecosystem, what it meant to be an entrepreneur, how capital flowed, all of the, you know, who got to run companies, all of those things. I do not think OpenAI would have been possible. I think the changes that YC induced in the whole ecosystem, you know, more leverage going to founders, young technical founders having the ability to raise lots of capital, um, the ability to sort of like work on ambitious things without a very proven resume. I don't think opening out would have been possible. So this is kind of like a big change. David?Is this all tied to the fact that you think there was a benefit in you going from founder to investor for a long period of time back to founder? There are all those benefits too. And I wouldn't say I really went from founder to investor to founder 'cause like the first time I was founder didn't. work out that well. So like still starting a company. You learn some lessons from failure, but I think you learn way more from success. No, you got like, hold on, you we're not moving on from that. You gotta say more about that. There's some I can't believe I think it's Anna Kranina, some great Russian novel. I'm very embarrassed not to notice. Starts with like all unhappy families or unhappy in their own way. All happy families are the same. Yeah. That explains why it jumps into memory. Um but I think this is really true. Like when I look at the lessons of where I have failed at something I learned something generic about grit and determination and something not to do. But there's most things don't work. So there's like a lot of reasons why things don't work. And there's I think it's like harder to put put together the correct causation. And when I've had something really work, when I understand like what parts of why combinator really worked or what parts of OpenAI really worked, trying to apply those lessons going forward has been much more helpful to me than trying to apply the anti-lessons of what didn't work. And so, you know, you you should of course learn as much as you can from every data point. So learn from the failures, learn from the successes. But in my own experience, uh I have when I have tried to apply those lessons, the lessons I learned from success were very good and I should have applied those more. And the lessons I learned from failure were either fairly generic and I kind of already knew them or or got in the way of something else. And and I think this is like generally true for a lot of people. Yeah, but isn't it like we already kind of know what we should do or what should we avoid, but it's like the reminder, the constant reminder. So like the the best description of my other podcast founders I ever heard is like it's church for entrepreneurs. If you really think about it, I used to drop it on Sundays and I should go back to doing that. But it's like really I'm just telling the same, it's the same personality type it's appeared throughout history. Yeah. It's just like now this person happens to be building ships. And this person built technology, but like and they live the same personality, that's for sure. I'm kind of obsessed with this idea of things that last for a long period of time. Yep. And like uh, you know, companies, the best companies can last a long time, but not as long as cities, and uh cities don't last uh cities and countries don't last as long as like religions. David?And I'm like, so out of all the man-made things, what has lasted longer? I would say I can't think of anything other than religion might be another answer. So then I started studying, I grew up my mom was fundamentalist Christian, so I was forced to go to church my entire life. And I just started analyzing like what all the main religions in the world have in common. It's like, oh, we have a shared base of knowledge. Usually some kind of book. Right. We meet with like minded fellow believers at regular intervals. And it's not like I go to church on Sunday. It's like, okay, we talked about Jesus last week, but let's talk about this other guy. It's like no, go back to these same books, these same stories over and over again. So I read your blog and you even said this uh something about like when YC ended. It's like you're repeating the same thing, you're telling it to them all the time, and then they leave the church, you know, to to for this analogy, and then they stop doing the same stuff it's like it's not even the lessons, it's like the constant reminder that this is important. I extremely strongly agree with that. But I think it is better to be reminded of the thing like talk to your users more, you know, ship products earlier, get more feedback, um, hold a higher bar for who you recruit and who you hire and more quickly this is like but it's the positives that I think are good. You have one of the greatest tweets. I have a I say to my phone you're like, you know, skip the conferences, the fucking dinners, everything else, just like essentially make the product or sell and sell the product. If you're not making it and you're not selling it, like that's all you actually have to do. And I think it's like there again, it goes back to like that simple genius. So then this is the other part that I find most fascinating because somebody asked me uh yesterday, they're like what's your usually there's some kind of historical equivalent for every founder I meet that I can like, oh that guy's kinda like Vanderbilt, that guy's kinda like Rockefeller or Ludwig or any of these people. David?And it's like what's your historical equivalent for Sam? I was like there isn't I can't think of one 'cause like I don't know him enough well. understand how he thinks yet. David?What do you think now? Well we're this is the first of hopefully eight conversations so we'll I'll tell you on conversation seven but this is very rare. I just talked to Doug Leone and he talked about one dude that he hired it's the founder of New Bank and he was a fucking like an associate in VC and then leaves and founds one of the most successful companies. I'm like, I've never heard of that. David?Doug, have you? David?And did he dedicate his life to this? He goes, no, that's the only one. So again, it's very rare. Mostly people go from founder, sell their business, unfortunately, and then investor, as opposed to run the business till you die, which is my preferred method of things. David?What I'm curious about this is like when you just said I learned more from successes, right? Well, it's the successes because you were exposed to what ten thousand different companies in that decade or decade and a half that you were doing this. And you saw obviously maybe the half a dozen or the dozen were the best in the world. David?So like are you taking their successes as well as like instructive? No, no, totally totally. Uh the And I think people do I mean, you're a incredible student, but there's a lot of pretty good students of entrepreneurialism and entrepreneurialism. And people I think often try to go look for those lessons of the things that that really worked. And as you said, it's kind of the same thing over and over again, like done in different industries, but you have to be reminded of it a lot. And it's unglamorous. I had no understanding going into this conversation. I think I have a slightly better understanding going into this. I told you before we started this is just for my own edification. But like even the people that influence you, where it's just like Peter Thiel saying, No, dummy, he's obviously going to say he's like, No, dummy, this is working. David?Why are you doing anything else but the thing that is working? So there's gotta be examples where you're like, hey, I've given this advice to other founders a million times. And then you catch yourself, oh shit, I'm not even applying my own advice at this point in time. Totally. Yeah, I'll give many examples of that. David?I I I I think it's also instructive to like what what was the new thing?

Building OpenAI from uncertainty

The small, directionless beginning; research bets; leaderboards; demos; and the path toward GPT and scaling laws.

14.01

A research lab without a playbook

The early uncertainty and the attempt to learn from older research institutions.

What didn't you have the advice for and the thing that was really different about OpenAI than anything that I had pattern matching. For is it was four and a half years from when we started the company to when we launched our first product. David?The opposite of YC advice, right? Yes. Okay. Yes. And although there were all these ways which managing a research team was similar to selecting and advising founders. Learning what it to whatever degree we learned it, because I don't think we did it perfectly. Like how you manage through this part of the world where you don't have the external signal from customers, and you're just trying to like You know, do what would normally be the catastrophic startup device if not uh not shipping a product for four and a half years. That was very difficult. David?And we tried all of these things about how we like had a how we replaced the signal of do customers actually like the product for is our research actually working? One of the things that worked actually is during the um Dota 2 days when we were trying to use RL to beat this video beat people at a video game, we put up like a leaderboard and people could just see how different ideas were performing and what was, you know, that was like objective and real and people wanted to like go up that. But we had to try all of these things to To uh basically like simulate end users. And that was a totally interesting new problem. I had no pattern matching for. David?How did you work your way through that? David?What was your thinking? David?Like, how'd you do this?

14.02

How the lab learned to learn

Asking people from great labs, testing approaches, and discarding what failed.

We asked a bunch of people who had been at great research labs of the past. And it had been, you know, OpenAI started at sort of a time when everybody in Silicon Valley, as their vanity project, including me, uh, wanted to start a research lab. And there were all these books about the heyday of Bell Labs or Xerox PARC that were very popular. Everybody was talking about this. There was a huge amount of discussion. In fact, I even see one of the books over there about uh about Bell Labs. But there was not a ton of people that had like in living memory how to actually do it. So We talked a lot to Alan Kay. We talked to a handful of other people and we got some advice from them about, you know, what made a really good research research lab. And some of it was really good. Some of it didn't translate as well to the current moment. Well, we also didn't have this giant monopolistic profit profit printing machine like Bell Labs was spun out independently. Uh Polaroid did a lot more research when they had uh essentially like a monopoly on its photography. David?I just read the biography of the founder of Honda, right? Ha the guy created the most successful m uh uh motor vehicle of all time, the Honda Cub has sold uninterrupted for like sixty years millions of yeah.ck Fingu vehicles. And his whole thing, and he he ride the same conclusion Bill Lobbs did, that he thought that research and development had to actually be separate. It was a s it was spun out of the company and had separate ownership. Yeah. Just like Bell Labs did. We did not have that. No, you did not have the when I think back to those early days, I mostly feel like I was trying and failing to raise money. That's like my dominant memory of the early days of OpenAI. So much effort, so frustrating. I wish we had some sort of cash machine like that. I remember like one of my clearest memories of all of OpenAI. So announced coming out of twenty fifteen, but the first day was right after New Year's in twenty sixteen. And twelve of us or eleven of us showed up at Greg Brockman's apartment, you know, like nine, nine: thirty on a Monday or Tuesday morning, something like that. Let's say it's January 4th. And everybody's there, and it's been this like big effort, and everybody walks in with a lot of excitement. It feels like the first day of school, whatever. David?And then very quickly, people like look around the room and they're sort of like, well, what do we do now? Someone says, okay, we should get a whiteboard. Greg, you know, gets someone to go off and find a whiteboard. Whiteboard comes. Um, look around again. David?You know, what are we supposed to do? now And you just feel the energy in the room collapse. And none of us know what to do. Like there's no it was not like building a product startup. It was not like let's build this product and let's talk to customers. It's like okay, we've said we want to make AGI. We should write some papers. Okay, let's write some papers. Maybe we should think about some ideas. Okay, let's think about some ideas. You know, everybody's got their like moments of I have no idea what I'm doing. That was one of mine. Like I have you know, we have just launched this thing. None of us have any idea what we're gonna do. So we did we know how to do and eventually we figured out a lot of things didn't work. Eventually we figured out a kind of like rhythm for making and then evaluating research bets. And far from perfect, obviously, uh, but we did find a gradient that we could kind of progress along. And we figured out how to get, you know, the resources that very smart people needed and how to make sure that we were like not completely getting lost in the wilderness. And over Some number of years, mostly chaotic stumbling, we eventually made most of the big discoveries. You know, what started as the unsupervised sentiment neuron paper uh turned into GPT-1 and then eventually GPT whatever, the scaling laws work that gave us the confidence, not only by the compute, but the understanding about how to scale up our models sort of came together. And through and many other things too. Through this process, we learned things like that idea of leaderboards that worked. We also learned the incredible power of external demos uh for like an eminent person that the researchers really wanted to impress. And then we learned a bunch of things that like didn't work, like fake deadlines. That has to be so disorienting to live through that experience. You're at twelve people in an apartment, don't even have a whiteboard, don't know what to do. Fast forward a decade, you have a billion people using Very strange experience. David?Do you keep a journal?

Documentation, memory, and legacy

Letters to a child, recording the journey, the podcast as a recurring reminder, and the closing advice to keep building.

15.01

Letters to a child

The private record of a life becomes a way to remember what the work felt like.

When my kid was born, my first kid, I would like, you know, get home at the end of the day and be rocking him to sleep and just like talk to a kid or whatever. So I would just like need to come up with things to talk about. So I would like just tell him about my day and what we were struggling with and kind of like what I was worried about and what was happening. It was kind of like fun for me to do and I was like, this is sort of interesting and someday it'll be like interesting for him to have this. So I started writing him like every Sunday I would like write him a letter. I would like talk just to talk, and then I would like write it down. If I only ever did like eight of them or something. David?How many kids do you have? Two. Okay.

15.02

Why founders should document the journey

Write down the uncertainty now, before the polished story replaces it.

Bezos has this great line about building Amazon. David?He's like, we're trying to do stuff that we can tell our grandkids about that we're proud of, right? And those things are hard. The fact that you were writing to your son. Oh man. Keep writing the letters. And if you don't do that, this is real quick. Just because I've read enough books about this. David?Most of the time, guess what? Founders don't write autobiographies when they're 40. They write them when they're 70 and they're looking back and they wish they could do it again and so much has been lost to the stance of time. They all repeat this. They're like, I wish I journaled. So even if you don't do it, you have enough resources. David?What I would do? Write have, a book written, even if it's for internal purposes only. David?You ever read The Little Kingdom by Michael Mortz? I never read it. Oh, you have to. It's like the first six-year history of Apple written by Michael Mortz. He's a phenomenal writer. He's a great writer. He's a crazy writer, winds up being one of the best venture capitalists of all time, I guess. But like the you just the point is to have that book ends. Steve hasn't even made that Apple yet. So you get like what actually happened. You're gonna want this. You might not want it now, but you're damn sure gonna want it when you're 60 or 70. David?The thing that was so interesting was like the mindset of writing to your kid, like you really can't hide behind anything. Like you're like, I really care what my kid's gonna think about me. So like this thing happened, like you know, didn't feel great about it, better do it differently next week. Like it was a very interesting extremely interesting mental framework. David?Maybe I'll find some way to do it again. David?No, maybe you're gonna do it. Okay. Sam, thanks for taking the time. This was awesome, man. Thank you, appreciate it. I hope you enjoyed this episode. Please remember to subscribe wherever you're listening and leave a review. And make sure you listen to my other podcast founders. For almost a decade, I've obsessively read over 400 biographies of history's greatest entrepreneurs searching for ideas that you can use in your work. Most of the guests you hear on this show first found me through founders.

How much software?

When?

Human agency

Concentration

Where to learn

Failure reporting

The bottleneck

The agent

The product shape

The customer relationship

Human media

Analog life

Entrepreneurship

Power

The idea

The timing

Resource allocation

The hard cut

Who needs one

The role

The operating system

Shipping

The data point

The lesson

Replacement signals

The timeline

The pace

The benefit

When to record

The purpose