[0:00:00] All right, everyone. So how many of you are Stanford students? Raise your hand. Okay, pretty good. We got a lot. How many of you in your years at Stanford lived in Donner? We got a few. Well, Sam was actually a Donner resident. And yeah, so why don't we start there? Do you remember moving in on your first day at Stanford over here?
[0:00:25] I do. I actually, I haven't been back to campus in so long. I'm feeling inspired to like go walk by after this because I have like such fond memories of that first day. It was like, you know, I'm sure like for all of you, it was a very strange experience, but it was like awesome as soon as I got here. Do you have any favorite memories, any building climbing, tunnel diving? Yeah, I was mentioning this backstage. I like, it's sort of, you know, when you go to a place you haven't been in a while, all these memories that you forgot about come back.
[0:00:55] And very near here, like early in freshman year, a bunch of us went steam tunneling, which I don't know if people still do, but it was like a cool rite of passage at the time. One of the people that we went with was like not told you had to put the cover back on. So they kind of left the cover off. We went down and looked around and whatever and came back and there was a car, professor's car that had like fallen into the open manhole. I thought I was going to get like, you know, kicked out. It was like first few weeks of school.
[0:01:25] But I had totally forgotten about that, and it's fun to see. My memories are, Stanford is awesome. And I just have very happy memories of that time. And kind of even the parts that I didn't think I was gonna like at the time, like staying up all night working in the basement of Gates are very happy memories now. I think that's like a core CS student type of thing. I have also pulled all-nighters in Gates. It's- It's a good memory in retrospect. Wait, yeah.
[0:01:55] Every time I walk by Gates, I just think back to that all-nighter, and Gates comes with certain memories now. Did you have any specific clubs that you were a part of here at Stanford? I did a few, but honestly, the best thing I did was just hacking on projects with friends. If I could go back and do it, I would have done less clubs and more just building stuff. Yeah. All right. Well, we went through the archives of Sam's time here at Stanford.
[0:02:27] And I don't know if you guys know the club BASES. It's the Student Entrepreneurship Club here on campus. And they found this document, the Business Symposium. Whoa. And it was for the next generation of business leaders. So this is from spring 2005. This is like right before you dropped out. And in this document, there's this one entry. And it says, viendo.
[0:02:55] And Viendo is a service for cell phones that provides personalized information about hot venues, a map of locations of friends nearby, and compatible new friends and dates in the area. And keep in mind, this is 2005. So this is like before the iPhone came out. So this is kind of like a crazy idea at the time. And it's co-founded by four students, as you can see here. Peter, Alec, Nick, and this guy named Sam up there as well. Is this looped? Yeah.
[0:03:24] I forgot about this also, but the important thing here is not about building stuff, but it's about how much Y Combinator did to change what it took to get a project funded. I was like, you know, I came here, I was not that interested in startups, and then it kind of like, I was like, it was in the water and I got excited about startups, but it was a super different time then, and you would like talk to people and they'd be like, oh, you need like a business person.
[0:03:53] You gotta get these connections to VCs. It was like, you gotta get serious. You need a bunch of MBAs. You gotta write a business plan. And it's like, you need to raise this big round and you're gonna have these VCs. They're gonna tell you what to do and whatever. And I was like, okay, I guess I don't know any better. And these smart people are saying, that's what you have to do. So I'll believe it. This was right around this time is when Y Combinator launched.
[0:04:21] I'm sure you all already think this from just hearing from Gary, but if you want to start a startup, you should definitely do Y Combinator. And it's an amazing thing. But before YC, like in the dark ages, this is what you had to do. You had to go do business plan competitions, which are... Business plans are a waste of time. Business plan competition is an even worse waste of time. But this is kind of what you were stuck with. This was how you could get things out in the world. And in the...
[0:04:51] 20 years since, things have changed unbelievably for the better. Yeah, so how long were you working on this idea before you discovered Y Combinator? I was hacking on a project. That was another thing. This, like many other things, did not start intending to be a company in any way. This was like a sort of late-night dorm project. Oh, wow. But over a while, and then it got serious when YC... Do you remember if you won this competition? We did, yeah. That's pretty good.
[0:05:19] So then how did you end up getting in contact with like YC? You were in the first batch, right? You just like applied online. I actually, I think it was, maybe it was like you filled out this weird form that PG had that you had to like paste into an email and not change any formatting because he had some weird parser for it. I think it was like that. So all three of us are sophomores right now. Could you walk us through a day in your life back then?
[0:05:54] I mean, it was really boring. I don't know. I went to class or didn't, and I hung out with my friends, and I built stuff. Yeah, again, very fond memories, but it was very simple, lovely, boring time.
[0:06:13] So when you were considering dropping out, did you feel like it was risky? I mean, were you excited? I mean, it sounds like school maybe wasn't the whole environment that you were looking for. What was kind of like your process behind that? Look, I feel a little bad saying this because obviously it worked out for me and I don't, you know, survivorship bias is a real thing. I didn't think at the time, and I don't think now, that dropping out is like a particularly risky thing to do.
[0:06:44] I don't know if Stanford still has this, but at the time it was like you could stop out for eight years and come back kind of no questions asked, which I think is a great policy, and it certainly made it easier to get my parents slightly on board with the idea. I think you should just work on whatever the sort of most interesting, highest rate of learning, highest impact thing you can is, and if that's being in
[0:07:12] School and learning a ton of stuff and being around interesting people and working on interesting stuff, great. If it's doing a startup, great. If it's dropping out to do AI research or whatever, also great. I think one of the trends of the last few decades, couple decades at least, has been that the default path has gotten relatively less good and the option space of what you can do and still
[0:07:39] Be really successful and happy and have a great impact on the world has increased and increased and increased. And as the society gets richer and freer and more capability with a better technology tree, I think that will just keep going. So it didn't feel like a crazy or risky thing to me at the time. I think if I were a student today, it would feel like even more of a, that's totally something I could consider. My parents didn't like it.
[0:08:09] Yeah, that makes sense. That makes sense. Wow. So you raised over $30 million for this idea, which is kind of considered seed round today. And I was going to say, that's like a pre-pre-seed round. It was a lot of money at the time. Yeah. How did you do this? I mean, what did the pitch look like? It's like 17 GPUs now or something. Wait, no, that's exactly right. Well, one thing that is cool about Stanford, and particularly so at the time, is it had a good on-ramp. It was very well connected to the whole
[0:08:40] VC machinery. And Y Combinator also was helpful there. There were less VC firms and less money at the time, but they were extremely focused on Stanford as a source of ideas and people. Do you remember any stories from that YC batch or from founding? Did it go really well at first, or what was that process like? Actually, it felt like still being in college.
[0:09:12] Everybody was pretty young at the time. We kind of would stay up late talking or working on computers. And it felt shockingly not that different. And I've heard a lot of people say this that have done YC or not that have dropped out to do startups. If you're a college student, if you're a CS-ish college student, and then you do a startup, the last year that you did school and the first year of doing a startup eventually gets very different. They're shockingly the same.
[0:09:43] So I mean, we have a lot of students here from Stanford, but we also have a lot of non-Stanford students. And I agree with you. I do think Stanford has a pipeline into this VC culture and all this money, right? I mean, what would you recommend to students coming from other schools if they were interested in going into this same process? That was the case at the time. I think that's no longer the case. I think the sort of Stanford special access, I mean, maybe there's still some of it, but I think it's mostly over, maybe totally over, maybe even a slight negative selection for Stanford students at this point.
[0:10:12] I think YC did a lot for this, but also VC just matured as an industry, and there's so much money in the world that wants to fund startups now that I think it's no longer the case at all that it's a very Stanford-y thing. So fast forward a few years, and you're now the president of YC, and then all of a sudden, boom, OpenAI is born. What happened? How did you guys come up with the idea? So I was studying AI when I was here.
[0:10:42] And it was not working at the time at all. I thought AI would be the coolest, most important thing, but I worked in the AI lab the summer between my freshman and sophomore year, and I was like, okay, this is not it. And I got distracted and did a startup with something else, and then I became an investor for a while. But I always thought that AI was at least one of the two or three most important things that could happen.
[0:11:13] I was paying attention to it. This is a thing I recommend and something I've heard many people describe over the course of your career, which is you get conviction on the things that you care about early, a superset of them. Some you'll lose interest in. Some will develop more than you thought. But the landscape of the world just is going to keep changing as you get more experience.
[0:11:38] Continuing to focus and stick your head up and look and not get into tunnel vision on one thing seems like it's really important and really correlates with success over a long period of time. And so I was paying attention to all the areas I was interested in. In 2012, the AlexNet paper happened, and I, like everybody else, was like, oh, that's kind of cool, or very cool, but it didn't update me as much as it should have. Strong priors are a real factor.
[0:12:08] I then watched over the next couple of years as deep learning models kept getting better with scale. And I still didn't update enough, but now I was at least like, this is like an asteroid coming and a crazy thing that people are not paying more attention to. And so in 2014, we started saying, we should really do something about this.
[0:12:34] It was incredibly unpopular at the time. Like we had said, we want to make an AGI lab and all the like elders of the field were like, you're insane, you're a scammer. This is like, you know, AGI is a hundred years away. None of these approaches are going to work. And we really said, let's just push on this one idea that scaling deep learning seems to matter. We didn't know at the time how beautifully predictable the scaling laws were, but it was at least clear back in those days that if you threw more compute at something, you got better results most of the time.
[0:13:04] And we kind of just decided we were gonna push on it as far as we could and see what happened. We were also very worried about what it would mean for like how we're gonna solve the alignment problem, what it would mean for society. We didn't know that we were gonna make a product. We just thought we were gonna be a research lab. So we kind of had a weird start, but it was really this, it was this mindset of if AI works, it will be unbelievably important. And we finally have an attack vector that maybe can go very far. It felt
[0:13:34] And still feels kind of crazy to me that it can go as far as it's turned out to go. Deep learning, the fact that these models can sort of do unsupervised learning in particular still seems sort of magical to me. The fact that next word prediction can be used basically to discover new scientific ideas, it makes like intellectual sense, but to see it work is still like a very crazy thing.
[0:14:05] But that was it. It was we have this one idea, let's push on it. So you see all these ideas now, but as you mentioned, it was incredibly unpopular at the time. I'm really curious to know what the first pitch for OpenAI looked like. I love the unpopular ideas. I have always liked ideas that are big if they work, difficult to get to work, and unpopular. My career as an investor was basically betting on those for whatever reason that people think.
[0:14:34] One thing is fun and not another. This is something that has always seemed particularly fun to me. So even if it weren't that I love AI so much, I would have thought the fact that this seems really undervalued and promising was cool. And that was basically the pitch. The pitch was something important is happening with deep learning and scale. We don't know what we're going to do with it. We don't know where it's going to go. We had a bunch of ideas that turned out to be wrong on the hops along the way.
[0:15:02] But we're going to push on this and something good is going to come out of it. And other people are undervaluing this. So it was like, that was really the pitch. It was not a great pitch. Well, it worked. How did you meet your co-founders? I had known Greg forever. I was an early investor in Stripe. I think he was the first employee and I was brought in to help close him. And we had stayed in touch for the sort of decade or whatever in between.
[0:15:34] Ilja is the sort of genius and godfather of the field, and I stalked him at a conference. And then later we started talking about doing a startup. Wojciech I met through kind of the like weird AGI nerds network of the Bay Area at the time. And John, one of the others connected me to, and Elon.
[0:16:03] I guess, yeah. So how did founding feel on this time with OpenAI compared to your experience with Looped before? I mean, this was really a weird one. The degree to which OpenAI was not meant to be a company is very hard to overstate. We had no idea that we were going to ever do a product. We thought we were just going to do research papers.
[0:16:32] As we got going, it became clear that A, there was going to be a lot more economic value along the way than we thought. We kind of had this conception that you were going to build this super intelligence and you're going to have this crazy thing happen all at once. And we learned that was not quite the right picture. We also realized that instead of needing the few billion dollars we thought we might need, you're going to need hundreds of billions of dollars. And so it evolved. But at the start, it was like, we're going to be a research lab. Actually, one of my most vivid memories of
[0:17:01] OpenAI. Very first day. Been all this drama to get it started. The very first day, we're together in Greg Brockman's apartment, which was where we worked until we got our first office a few weeks or maybe a month later. And it's like, you know, 9.30, 10 in the morning, everybody rolls in. We're sitting on Greg's couch. Maybe there's 10 or 12 of us. Maybe not even that many, maybe 8 or 10. And we look around and say, like, okay, what are we going to do? Yeah. And someone said, we should write some papers.
[0:17:33] And someone said, OK, we need a whiteboard to start talking. And someone orders a whiteboard on Amazon Instant or something. And I had this real moment of like, what? This is not good. I don't know what this is, but this is not a startup or any kind of thing that's going to be effective. But I'm a big believer that in moments like that, you just breathe, trust that if it's
[0:18:01] If you have good people around, it's going to work out. And we just kind of went with it. Sometime in that first week, most of the ideas that turned out to be the core ideas for the first four years of OpenAI were written down on that whiteboard. Even at the time, we didn't realize that. And we're kind of just like, eh, this is probably not in. We went in some other directions. And we said, we need to get some GPUs. We need to do this. We need to do that. We need to pick some projects to see if this is really working.
[0:18:32] And stuff just kind of kept going. So you were in this place where you're like, okay, what do we do? Let's start doing some research papers. When was the moment after that where things started to click for the first time and you're like, things started to feel like you guys really had discovered something or something big was about to happen? There were a few moments. I think it was not just one. But the felt sense was it was like miracle after miracle.
[0:19:10] And so at some point, more than any one thing that worked, it was like, man, something about the underlying approach here feels like we've discovered like a new law of physics or something. And, you know, it was like great to get Dota working or the robot hand working and then eventually the first language model. But the biggest takeaway was just like, there is something about this underlying approach that really feels like we're on a trajectory to the degree that I had like a single moment
[0:19:40] It was the GPT series. By GPT-2, it was clear to me, man, something important is about to happen. Do you remember that day, the GPT-2 launch? I do. I don't remember the launch day, but I remember the day that I got to first talk to the model. I was just like, wow, this is great. Actually, I think it was technically a model that was a little bit beyond what we released as GPT-2.
[0:20:11] But I knew GPT-3 would be amazing and GPT-4 would be amazing, and they were, and they were amazing in new ways, but that first night that I got to play with the model, and it was like doing something I never had seen a computer do before, I was like, okay. And we had the built-up confidence of deep learning working. Yeah, and you initially withheld that model, GPT-2, right? At what point did you feel like we were ready, or what made you decide that the world was ready for these models?
[0:20:41] Look, clearly that looked like silly in retrospect. I think the principle of precaution with each new level of AI capability is reasonable. Alignment has been easier to make progress on than many of us in the field thought, but the societal impacts have been, or are getting big. And if we bias like slightly too much towards the cautionary side, I think that's good.
[0:21:15] Too much would be bad. I'm also worried about like a mega AI capability overhang building up. Like the way I would explain this is if companies don't adopt AI fast enough, then they will lose to fully autonomous, completely AI run and started companies. And that will be a crazy destabilizing thing for society. So you can't be that cautious either. But you know, like we'll go a little bit on that side of the line. So what do you think the next chat GPT moment will be?
[0:21:49] Well, I think the next Chai CPT moment already was coding agents. And I think the next one after this that has not come yet will be that same level of ability to get stuff done for all knowledge work, which I don't think is that far away. So, I mean, we're all sophomores here. There's a lot that's going to change in the next few years, I mean, before we graduate. I mean, do you think AGI will be ready by the time? Yeah, I think if you are a sophomore now, you will graduate to a world with AGI in it.
[0:22:21] And in some ways, it'll be the same. You'll go move and get a job and think about having a family and whatever else. You never bet against the human drives and what we care about staying the same. But in some other sense, it's just going to be wildly different. I think we'll have AI improving at an incredibly rapid clip. Society will be coming along at a slower clip, but still very fast.
[0:22:51] Science will be getting automated. What it means to start a startup will be totally different. What it means to go work at a company will be totally different. There's going to be a lot of change in the next couple of years. You are lucky to be a student right now because you can go do kind of whatever and you can try a lot of things quickly. And you are also lucky in that this is probably the most interesting time to be starting off adult life or a career.
[0:23:25] But I think a lot of the traditional advice is not quite going to work. So speaking of execution, in your first ever blog post on blog.samaltman.com, you say that knowing is easy and it's the doing that's difficult. Is there a truth that you think you know now about the future where the missing piece is execution? Well, I think that's just generally going to continue to be true.
[0:24:01] Knowing is easier and easier. You get the help of these AI tools. Getting stuff done is easier too, but it's easier for everybody. And so you have to compete with everybody else trying to get stuff done in the world. It is for sure true that what one programmer can do now is mind-bendingly different than what you could do a year or two years ago. A year ago, let's say. It is also true that
[0:24:31] The people that I know that are using these tools the most effectively say it has never been harder to do their work. And I suspect this trend is going to keep going. But man, watching these people who are really good at using AI coding tools have 12 agents running all the time and these complicated custom harnesses that they're constantly evolving. The tools have gotten unbelievably good. And the ability to use them.
[0:25:00] To be at the top of the game, I think, has never been harder. And you think this will apply across all verticals, not just coding? I mean, look, eventually, you will just say like, hey, Mr. AI, please make me my whole startup and do everything and do a great job. I'll help you with whatever you need. But between now and then, yeah, I think it will apply to all verticals. Given that.
[0:25:22] These models are becoming more and more powerful then. We have people that are running this many coding agents. There's definitely some of y'all in the audience here. What should people be working on and focusing on then? What do you think is the most important problem in the world right now? The degree to which it feels like everything is up for a rethink. I'm not that interested. People are like, oh, all SaaS is dead. I'm going to make a new startup to make...
[0:25:57] SaaS company X better. Like, you know what? That will probably work. But I think that's missing the point of what's possible now. The degree to which things are just completely, fundamentally re-imaginable, and the space feels super open on that. Like, you know, just to pick a random example, social products have not changed that much in the last, and I'm not even counting like,
[0:26:29] Video scrolling apps, just like an actual connect to people and look at information about them style apps have not changed that much. I think MultBook points to the first interesting new thing where you can imagine a bunch of agents on behalf of their users off in a virtual space talking to each other, exchanging news information, completely changing what it means to interact with other people on the internet. That was just not possible at all.
[0:26:57] Until very recently. And now it feels like the space of social for the first time is totally open for some startup to do something completely new. A lot of people are like bolting AI into some productivity software. And I don't really need something in Google Docs trying to summarize it or write a sentence for me. I mean, that's fine. But probably the whole thing of what it means to write a document is going to completely change. And there's a whole new kind of thing to imagine there.
[0:27:27] And this will apply not just to traditional software startups, but most companies in the world pretty soon. You know, robots won't be that far off, whatever. I love times like this. They're rare. The iPhone, when the iPhone launched was one of them. I don't think we've really had one since. And the fact that this time, not only can you use the technology to do these things that you could never do before, but the fact that you can use technology to like make the product or make the company so quickly and with so few people,
[0:28:00] I think it's going to mean a lot more startups getting super successful really fast. And a lot of startups that have like one person or six people or something like that, that get as impactful and as much revenue as some of the, you know, a medium-sized company does today. Do you think then that, given you're talking about like Malt Book, these agents,
[0:28:26] That like once AGI happens, it will then surprise us and just all of a sudden, like all of everything we've done, all these new companies are popping up and creating themselves and integrating into our world and competing with things in a way like that? Or like how will it be introduced into our world then? So I think that what it's going to feel like in the next few years, let's say two years, is stuff's just going to happen faster. You know, startups are going to do new things faster. Existing companies are going to ramp their revenue faster.
[0:29:00] We've broken the laws of traditional speed and product development and business growth rates and things like that. And companies will start to hire AI coworkers and do amazing things with fewer people. I think we'll continue to see AI research go much faster and we'll see scientific research happen much faster. But I don't think it'll feel like totally autonomous companies transacting with each other in an environment sort of mold book style.
[0:29:31] Eventually that probably does happen. And, you know, one of the sort of like central observations I've had in life is that people are unbelievably adaptable. Society is unbelievably adaptable. So that sounds like a totally crazy destabilizing thing. And I bet it's not going to be as weird to live through as it sounds or only for a day. So I kind of wanted to go back to that thing that you'd mentioned about societal impacts. I mean, how do you think this is going to impact like politics, like religion?
[0:30:03] Just expanding more just outside of the startup space, how is society going to be impacted by this, do you feel? Everyone's got a different definition for AGI, a different definition for superintelligence. I kind of think the most important definitions are something about the ability to self-improve, but one of the ones I like as a sort of example for a way to think about this is if we had an AI system that could do a better job at making decisions,
[0:30:34] Would we, and like confidently, it's much better, would we politically or societally want that? It's very unclear. You know, most people I think today at least would say probably not. Maybe as things evolve more, they do. Or, you know, like an AI CEO of a big company. Would you rather have, I think it'd be awesome for OpenAI to have an AI CEO.
[0:31:03] But I think a lot of people would be like, ah, that's a little scary. If, though, AI is better at doing some of these things, I think there will be many things that people are better at than AI long term, or at least people really care about other people doing. But I have no delusions that an AI will not be a better CEO of OpenAI than me at some reasonably soon time. If that's true,
[0:31:32] And if some companies or countries don't do that and others do, and AI is much better, that will have a real impact in economic competitiveness, national competitiveness, things like that. And so I think one of the things that will happen about how this is going to impact us economically, societally, politically is if the AI is just much better, I think people will feel increasingly like we either do worse or we put the AI in more and more.
[0:32:02] Influential positions. I think people will always like, you know, if you think of like an AI CEO of a company, I think people will always want human board members and that's probably a really great idea. But I kind of think that people may have not quite, I certainly have not really wrestled with like, is there a way that this doesn't happen and what all of the political, economic, societal impacts are.
[0:32:34] I still believe that although it sounds like a completely crazy thing to say, we're very adaptable. We really care about other people. Maybe there's like a AI CEO, but you just care about like the company's spokesperson speaking to you in these like eight second video clips or whatever. I think I'm not a jobs doomer at all. I mean, a lot of the current jobs are gonna go away or be heavily changed, but human desire for being useful to each other.
[0:33:10] To compete with each other, to create differential status, to do something fulfilling, to express creativity. That is not going to go anywhere. And maybe the jobs of 100 years from now look nothing like the jobs of today. Nor do many of the jobs of today look much like the jobs of a few hundred years ago. But I'm confident we'll have stuff to do. I'm confident we'll really care about other people and less about the machines. And then beyond that,
[0:33:40] Yeah, I think AI is going to get ever more competent and we're going to want to use it for more and more things. Awesome. All right, so we're going to go into a quick Q&A segment now with questions from the audience. I would love hard questions I've never heard before. We have a Slack channel, Sam Altman Q&A. Sam Altman Q&A, guys. And a lot of you have put in questions.
[0:34:05] So we're going to start off by reading some of the questions that were already submitted, I think, since we posted it yesterday. But we encourage you guys to submit more questions. And then if you really, really like a certain question, we want you to just reply with some sort of emoji so that we can see it and we can ask Sam. All right, do you want to start off? To get started, you're talking about OpenAI potentially having an AI CEO. So if you ask a question that's kind of related to this about, say, OpenAI becomes this
[0:34:33] And I'll be all company that can do anything because CEO is an AI. How do you compete with OpenAI then in the future? When we started OpenAI, everyone was like, you can't do this. You can't compete with Google. It's impossible. And there will someday be much bigger and more successful companies than OpenAI. They will not do exactly what OpenAI does. You may steal our users. You may do some other thing. You may have a replacement product. It will not very unlikely to be exactly what OpenAI does.
[0:35:07] Google screwed up, didn't react enough to our approach early enough. They're doing a very nice job now, but had Google had their act together, we would not have been able to get off the ground at all. And they have very smart people. It's just big companies have big company problems. Someone in this room, I hope, will do the same thing to OpenAI. We sell intelligence for what I think of as a very low price. You have...
[0:35:38] Probably the biggest option space of any startup, of startups that have at any time in history. If you try to make an exact copy of ChatGPT or Codex or Frontier models in an API, probably not gonna work. There's so much other stuff to do. And the best thing of all is to figure out, we were talking about this earlier, things that are just gonna totally change now.
[0:36:10] Take your pick. We're not going to be able to go do all those things. And neither is any other company that is not a startup and that has the freedom to be a little bit weird and decisive and pull together a band of rebels. Just go do that. It always sounds really scary. I remember how many very wise people said, you can't start an AI company. Google is just going to crush you. Maybe they still do. I don't know. But just go at it.
[0:36:38] Okay, well, I can ask the next question. This might be a burning question, let me see. All right, news reports say OpenAI is burning cash very fast and could face liquidity pressure in the future. How do you plan to handle this? And is the situation really as serious as you think the media reports it to be? We are burning cash very fast. It is not a serious situation. Like the, you know, if we can spend a billion dollars this year to make $3 billion next year,
[0:37:07] There's a lot of capital in the world that wants to do that. So we will raise a lot of money, equity and debt. We will spend a lot of money. We'll make a lot of money. It's weird. The economy hasn't seen something quite like this before. But given what we see about revenue growth and model growth and how much model growth correlates with future revenue growth, we're super confident in where things are going to go. So we've got another question here.
[0:37:40] Sam, are you really happy? I am actually one of the happiest people I know. But I didn't used to be. I used to be a very anxious person. A thing people say to me all the time is like, you have a miserable job. You have all the stress. You have all these people coming after you. You have like crazy bad shit. So many things happen to you a day. How do you not have a mental breakdown? And my answer to that is like,
[0:38:13] I just figured out what it was going to take for me. In my 20s, when I realized I was not particularly happy and certainly not particularly calm, I figured out what it was going to take for me to get happy and calm. I think people find their own paths there, but meditation was a big part of it for me. And it's kind of like a one-time thing. And now, even when stuff is really hard, really painful, a thing I kind of
[0:38:45] What I learned in meditation that always stuck with me is that everybody thinks the opposite of a bad experience is a good experience. So if I'm having a bad experience, then I'm not happy because I wish I were having a good experience. And a big mental shift for me was to reframe it as the opposite of a bad experience is no experience. And someday that'll happen to you and you'll miss the bad experiences too. And I have a lot of days where it's very stressful and very hard and a lot of things go wrong. I'm still grateful for them.
[0:39:16] Adding on to this, someone asks, what does a day in your life look like then? Absolute chaos. I wake up early. I work for a few hours. I kind of have control of my life until 8 o'clock in the morning. And I can get something done. I can try to drive whatever I want to get done. Then I play with my kid for an hour. Then I go into the office. And then it's just insane.
[0:39:46] Whatever is going to happen happens like crazy. I don't think any company has been as fast-moving, chaotic in everybody's crosshairs as us. And so it's just, yeah, unpredictable chaos all day. Yeah, a lot of us definitely don't wake up before 8 in the morning. I did not used to wake up early. This job is the thing that made me start doing it. Awesome. Should we do two more questions? Let's do two more questions, and then we'll wrap up. We are at time.
[0:40:19] I have another question from someone. What AI subfield today feels like OpenAI in 2016? A little undervalued or underestimated, but could shoot out of nowhere? That's a great question. This is the kind of question that I would spend a lot of time trying to think about. I'm not sure if you mean it as a research thing or a company thing. I'll answer both. And I'll do it quickly because I almost out of time.
[0:40:51] On the research perspective, I bet there is another new architecture to find that is gonna be as big of a gain as transformers were over LSTMs. And I think you finally have models that are smart enough to help do that kind of research. So I would go look for where can I find a mega breakthrough, and I would use the models to help me. On the product side,
[0:41:23] I guess we talked about this a little bit earlier, but I would just pick a big area and say, what is possible now with AI that wasn't possible at all before? Like, where can I totally redo something that's like AI is the absolute core to the interaction working? And look, I don't want to say this is like a once-in-a-lifetime opportunity because I hope the world keeps getting dramatically better every few years and that if we all came back here, or actually when some of you come back here and sit up here and talk to...
[0:41:53] The then current students. AGI will look like just a warmup for whatever the next important thing was and that'll keep going for the rest of history. But this is at least the best time ever so far. And what you can do now is so crazily exciting that I think you should probably try to figure out how to take advantage of this moment. And one last question.
[0:42:25] A lot of labs like XAI and Google have their own data centers and even custom silicon for inference. Does OpenAI have any interest in developing hardware as well or stick to external partners? Why or why not? We have a big custom silicon program. We're very excited about our inference chip and clearly the world wants a lot of cheap tokens and I think what we're working on there will be great.
[0:42:52] We have a very deep partnership with NVIDIA. We do all of our training on NVIDIA. We also use them for a lot of inference. But we are excited to be doing our chip in our system. Building data centers, I would love to not have to do that. If we have to at some point, we will. You can't do everything well as a company, and that's a big, intensive thing to do. If we can just design a great rack and have other people...
[0:43:23] Operate those and build the crazy buildings around them. That would be awesome. Thank you, Sam. Everyone, let's give it up for Sam.