Chamath Palihapitiya: Money, Success, Startups, Energy, Poker & Happiness | Lex Fridman Podcast #338

Chamath Palihapitiya 0:00 In terms of your mistakes, society tells you, "Don't make them because we will judge you and we will look down on you." And I think the really successful people realize that actually, no, it's the cycle time of mistakes that gets you to success. Because your error rate will diminish the more mistakes that you make. You observe them, you figure out where it's coming from. Is it a psychological thing? Is it a, you know, cognitive thing? And then you fix it.

Lex Fridman 0:30 The following is a conversation with Chamath Palihapitiya, a venture capitalist and engineer, founder and CEO of Social Capital, previously an early senior executive at Facebook, and is, uh, the co-host of the All-In podcast, a podcast that I highly recommend for the wisdom and the camaraderie of the four co-hosts, also known as besties. This is the Lex Fridman podcast. To support it, please check out our sponsors in the description. And now, dear friends, here's Chamath Palihapitiya.

Lex Fridman 1:05 You grew up in a dysfunctional household, on welfare. You've talked about this before. What were, for you, personally, psychologically, some difficult moments in your childhood?

Chamath Palihapitiya 1:16 I'll answer that question in a slightly different way, which is that I think when you grow up in a household that's defined by physical abuse and psychological abuse, you're hypervigilant all the time. And so it's actually easier for me to point to moments where I was happy or I felt compassion or I felt safe. Otherwise, every moment... I'll give you a couple of examples, like, you know, I was thinking about this a while ago. There was a tree outside of my apartment where we lived when I was growing up, and my father would sometimes- would make me go outside to take the tree branch that he would hit me with. Um, and so you can imagine if you're a 10, 11-year-old kid and you have to deal with that, what do you do? Well, a hypervigilant child learns how to basically estimate the strength of these branches, right? How far can he go before it breaks? You have to estimate his anger and estimate the effective strength of, you know, branches and bring back something because, you know, th- I remember these moments where if it was- he would look at it and then he would make me go out again and get it, right? Get a different one. Um, or, you know, there was a certain belt that he wore that had this kind of, um, belt buckle that stuck out. And you just wanted to make sure if that- if that was the thing that you were gonna get hit by, that it wasn't the buckle facing out, because that really hurt. And so you became hyper-aware of which part of the buckle was facing out versus facing in in those moments. And there are, like, hundreds of these little examples, which essentially I would- I would say the through line is that you're just so on edge, right? And you walk into this house and you're just basically trying to get to the point where you leave the house. Um, and so in that microcosm of growing up, any moment that's not like that is seared in my memory in a way that I just can't, um, describe to a person. I'll give you an example. I volunteered when I was in grade five or six, I th- can't remember which it was, in the kindergarten of my school. And I would just go and the teacher would, you know, ask you to clean things up. And at the end of that grade five year, she took me and two other kids to Dairy Queen. And I'd never been- I've never- I'd never gone to a restaurant, literally, um, because we just- we didn't have the money. And I remember the first time I tasted this, you know, this Dairy Queen meal, it was like a- a hamburger, fries, a Coke, and a- a Blizzard. And I was like, "What is this?" And I felt so special, you know, because you're getting something that most people would take for granted. "Oh, it's a Sunday." Or it's a, you know, or, "I'm really busy, let me go take my kid to- to fast food." I think that, you know, until I left high school, I think- and this is not just specific to me but a lot of other people, it's you're in this hypervigilant loop punctuated with these incredibly visceral moments of compassion by other people. You know, a different example. Um, we had such a strict budget and we didn't have a car. And so, you know, I was responsible with my mom to always go shopping and so I learned very early on how to, you know, look for coupons, how to buy things that were on sale or special. And we had a very basic diet because you have to budget this thing really precisely. But the end of every year where I lived, there was a large grocery chain called Loblaws, and Loblaws would discount, um, a cheesecake from 7.99 to 4.99. And my parents would buy that once a year, and we probably did that six or seven times. And you can't imagine how special we felt, myself, my two sisters. We would sit there, we would watch the, you know, the New Year's Eve celebration on TV. We would cut this cheesecake into, you know, five pieces. It felt like everything. Um, so that's sort of how, you know, my- my existence when I was at that age is, for better or for worse, that's how I remember it.

Lex Fridman 5:54 The hypervigilance loop, is that still with you today? What are the echoes of that that's still with you today, the good and the bad?

Chamath Palihapitiya 6:02 If you put yourself...... in the mind of a young child. The thing that that does to you is, at a very core basic level, it says you're worthless. Right? Because if you can step outside of that and you think about any child in the world, they don't deserve to go through that. And at some point, by the way, I should tell you, like, I don't blame my parents anymore. It- it was a process to get there, but I feel like they did the best they could, and they suffered their own issues and enormous pressures and stresses. And so, you know, I've really, for the most part, forgiven them.

Lex Fridman 6:49 How did you, sorry to interrupt, let go of that blame?

Chamath Palihapitiya 6:53 That was a really long process where for, I would say the first 35 years of my life, I compartmentalized and I avoided all of those memories. And I sought external validation, right? Going back to this self-worth idea, if you're taught as a child that you're worthless, because why would somebody do these things to you? It's not because you're worth something, you think to yourself very viscerally, "You're worth nothing." And so then you go out and you seek external validation. Maybe you try to go and get into a great college, you try to get a good job, you try to make a lot of money. You try to, you know, demonstrate in superficial ways with the car you drive or the clothes you wear, that you deserve people to care about you, to try to make up for that really deep hole. But at some point you- it doesn't get filled in. And so you have a choice. And so for me, what happened was in- in the course of a six-month period, I lost my best friend and I lost my father. And it was really like the dam broke loose, because I- the compartmentalization stopped working because the reminder of why I was compartmentalizing was gone. And so I had to go through this period of disharmony to really understand and steel man his perspective. And can you imagine trying to do that, to go through all of the things where you have to now look at it from his perspective and find compassion and empathy for what he went through? And then I shift, you know, the focus to my mom, and I said, "Well, you were not the victim actually. You were somewhat complicit as well because you were of sound mind and body and you were in the room when it happened." So then I had to go through that process with her and steel man her perspective. And at the end of it, I never justified what they did, but I've been able to forgive what they did. Um, I think they did the best they could. And at the end of the day, they did the most important thing, which is they gave me and my sisters a shot by emigrating, by giving up everything, by staying in Canada and doing whatever it took between the two of them to sort of claw and scrape together enough money to live so that my sisters and I could have a shot. And I'm very thankful for them. Could they have done better? Obviously. But I'm okay with what has taken place. But it's- it's been a long process of- of that steel manning so that you can develop some empathy and compassion and forgive.

Lex Fridman 9:35 Do you think if you talked to your dad shortly after he died and you went through that process, or today, you'd be able to have the same st- strength to forgive him?

Chamath Palihapitiya 9:52 I think it would be a very complicated journey. I think I've learned to be incredibly open about what has happened and all of the mistakes I've made. I think it's- it would require him to be pretty radically honest about confirming what I think he went through, because otherwise it just wouldn't work. Otherwise, I would say, "Let's keep things where they are," which is, "I did the work," you know, with- with people that have helped me, obviously, but, you know, um, it's better for him to just, you know, kind of hopefully he's looking from some place and he's thinking, "It was worth it." I think he deserves to think that. All of this-

Lex Fridman 10:36 Yeah.

Chamath Palihapitiya 10:37 ... because, you know, I think the immigrant challenge, or not even the immigrant challenge, the lower middle class challenge-

Lex Fridman 10:43 Mm-hmm.

Chamath Palihapitiya 10:43 ... anybody who really wants better for their kids and doesn't have a good toolkit to give it to them, some of them just they choke up on the bat. They just get so agitated about this idea that all the sacrifice will not be worth it, that it spills out in really unproductive ways. And I would put him in that category.

Lex Fridman 11:05 And their self-evaluation, introspection, they- they have tunnel vision, so they're not able to often see the damage they did. I mean, I- I know, uh, like yourself, a few successful people that had very difficult relationships with their dad, and when you take the perspective of the dad, they're completely in denial about any of it. So if you actually have a conversation, there would not be a deep honesty there. Uh, and that, I think that's maybe in part the way of life.

Chamath Palihapitiya 11:39 Yeah, and you know, I remember pretty distinctly after I left and in this, you know, in my middle 30s where, you know, by all measure I had roughly become reasonably successful. And my dad didn't particularly care about that, which was so odd because I had to confront the fact that, you know, whether it was a title or money or press clippings, he never really cared. He moved on to a different set of goals, which was more about my character and, you know, being a good person to my family and really-... preparing me to lead our family when he wasn't there, and that bothered me because I thought, I thought I got to the finish line, and I thought there was going to be a, a medal. You know, meaning, like, I can tell you, Lex, you know, he never told me that he loved me. I'm not sure if that's normal or not. It was my normality, and I thought there's going to be something, some gold star-

Lex Fridman 12:36 Mm-hmm.

Chamath Palihapitiya 12:36 ... which never appeared, and so that's, like, a hard thing to kind of confront because you're like, "Well, now what is this, what is this all about? Um, was this all just kind of a, a ruse?" But then I realized, "Well, hold on a second, there were these moments where in his way..." Again, putting yourself in his shoes, I think he was trying to say he was sorry. He would hold my hand, you know, and he would interlock the fingers, which I felt is, that's a really intimate way of holding somebody's hand, I think. Um, so I remember those things.

Lex Fridman 13:09 Mm-hmm.

Chamath Palihapitiya 13:10 So, you know, these are the things that are just etched in, at least in my mind, and at the end of it, you know, I think, uh, I've done a decent job in repairing my relationship with him, even though, you know, it was posthumous.

Lex Fridman 13:24 It does make me wonder in which way you and I, we might be broken and not see it and might be hurting others and not see it.

Chamath Palihapitiya 13:34 Well, I think that when you grow up in those kinds of environments, and they're all different kinds of this kind of dysfunction, but if what you get from that is that you're not worthwhile, you're not, you're less than many, many other people, when you enter adulthood or, you know, semi-adulthood, in your early 20s, you will be in a cycle where you are hurting other people. You may not know it. Hopefully, you find somebody who holds you accountable and tells you and loves you enough through that, um, but you are going to take all of that disharmony in your childhood, and you're going to inject that disharmony into whether it's your professional relationships or your personal relationships or both until you get to some form of rock bottom, and you start to repair. And I think there's a lot of people that resonate with that because they have each suffered their own things that, at some point in their lives, have told them that they're less than. And, um, and then they go and cope, and when you cope, eventually those coping mechanisms escalate, and at some point, it'll be unhealthy either for you, but oftentimes it's for the people around you.

Lex Fridman 14:49 Well, from those humble beginnings, you are now a billionaire. How has money changed your life or maybe the landscape of experience in your life? Does it buy happiness?

Chamath Palihapitiya 15:05 It doesn't buy happiness, but it buys you a level of comfort for you to really amplify what happiness is. I, I kind of think about it in the following way. Let's just say that there's a, a hundred things on a table, and the table says, "Find happiness here," and there are different prices. The way that the world works is that many of these experiences are cordoned off, a little bit behind a velvet rope, where you think that there's more happiness as the prices of things escalate.

Lex Fridman 15:42 Mm-hmm.

Chamath Palihapitiya 15:42 Right? If you live in an apartment, you admire the person with the house. If you live in a house, you admire the person with the bigger house. That person admires the person with, you know, um, an island, right?

Lex Fridman 15:56 Mm-hmm.

Chamath Palihapitiya 15:56 Um, some person drives their car admires the person who flies, who admires the person who flies business class, who admires the person who flies first, you know, to private. There's all of these escalations on this table, and most people get to the first five or six, and so they just naturally assume that items, you know, seven through a hundred is really where happiness is found. And, the, just to, you know, tell you the, the finish line, I've tried 100 and back, and then... (laughs) I've tried to- to move 400 to it, uh, and happiness isn't there, um, but it does give you a level of comfort. I read a study, and I don't know if it's true or not, but it said that, um, the absolute sort of, like, maximal link between money and happiness is around $50 million. And there was a, it was just like a social studies kind of thing that I think one of the Ivy Leagues put out, and underneath it, the way that they explained it was because you could have a home, you could have all kinds of the creature comforts. You could take care of your family, um, and then you were left to ponder what it is that you really want. I think the challenge for most people is to realize that this escalating arms race of, you know, more things will solve your problems is not true.

Lex Fridman 17:20 (sighs)

Chamath Palihapitiya 17:21 Um, more and better is not the solution. It's, it's this idea that you are on a very precise journey that's unique to yourself. You are playing a game of which only you are the player. Everybody else is an interloper, and you have a responsibility to design the gameplay. And I think a lot of people don't realize that because if they did, I think they would make a lot of different decisions about how they live their life, and I still do the same thing. I mean revert to basically running around asking other people, "What will make you like me more?"

Lex Fridman 18:02 Hmm.

Chamath Palihapitiya 18:03 You know, "What will make me more popular in your eyes?" And I try to do it, and it never works. Um, it is just a complete dead end.

Lex Fridman 18:13 Is there negative aspects to money? Like, for example-... it becoming harder to find people you can trust?

Chamath Palihapitiya 18:20 I think the most negative aspect is that it amplifies a 360-degree view of your personality. Because there are a lot of people, and society tells you, that more money is actually better, you are a better person somehow, and you're factually more worthwhile than some other people that have less money. That's also a lie. But when you're given that kind of attention, it's very easy for you to become a caricature of yourself. Um, that's probably the single worst thing that happens to you. But I say it in the opposite way. I think all I've ever seen in Silicon Valley, as an example, um, is that when somebody gets ahold of a lot of money, it tends to cause them to become exactly who they were meant to be. They're either a kind person, they're either a, a curious person, they're either a jerk, you know, they're either cheap. And they can use all kinds of masks, but now that there's no expectations and society gives you a get out of jail free card, you start to behave the way that's most comfortable to you. So you see somebody's innate personality. And that's really interesting thing to observe because then you can very quickly bucket sort where do you want to spend time and who is really, you know, additive to your game play and who is really a negative detractor to your game play.

Lex Fridman 19:38 You're an investor, but you're also a kind of philosopher. Um, you analyze the world and all its different, uh, perspectives on All-In podcast, on Twitter, everywhere. Uh, do you worry that money makes, puts you out of touch from being able to truly empathize with the experience of the general population, which in part, first of all, on a human level, that could be limiting, but also as an analyst of human civilization, that could be limiting?

Chamath Palihapitiya 20:13 I think it definitely can for a lot of people because it's just a, a, it's an abstraction for you to stop caring.

Lex Fridman 20:19 Right.

Chamath Palihapitiya 20:20 Um, I also think the other thing is that you can very quickly, um, especially in today's world, become the scapegoat, just to use a Girardian, like Rene Girard. If you look, if you think about like memetic theory in a nutshell, you know, we're all competing for these very scarce resources that we are told is worthwhile. And if you view the world through that Girardian lens, what are we really doing? We are all fighting for scarce resources, whether that's Twitter followers, money, acclaim, notoriety, and we all compete with each other. And in that competition, you know, Girard writes, like, the only way you escape that loop is by scapegoating something or somebody. And I think we are in that loop right now where just the fact of being successful is a thing that one should scapegoat to end all of this, you know, tension that we have in the world. I, I think that it's a little misguided because I don't think it solves the fundamental problem. Um, and we can talk about what the solution to some of these problems are, but, um, that's, I think the loop that we're all living. And so if you become a caricature- caricature and you feed yourself into it, I mean, you're not doing anything to, to really advance things.

Lex Fridman 21:40 Your nickname is The Dictator. How'd you get the nickname? Since we're talking about the corrupting nature of money.

Chamath Palihapitiya 21:46 That came from poker. In a poker game, you know, when you sit down, it's chaos, especially like in, in our home game. There's a ton of big egos. There's people always watching, you know, rail birding the game, all kinds of interesting folks. And in that, somebody needs to establish hygiene and rules.

Lex Fridman 22:04 Mm-hmm.

Chamath Palihapitiya 22:05 And I really care about the integrity of the game. And it would just require somebody to just say, "Okay, enough." Uh, and so, and then people were just like, "Okay, stop dictating." And that's where, that's where that nickname

Lex Fridman 22:16 So who to you, speaking of which, is the greatest poker of, uh, player of all time, and why is it Phil Hellmuth?

Chamath Palihapitiya 22:23 Exactly. You know, Muth probably knew this question was coming.

Lex Fridman 22:26 Yeah.

Chamath Palihapitiya 22:27 Um, here's what I'll say. I think Hellmuth is the antidote to computers, more than any other player playing today. And when you see him in a heads-up situation, so I think like he's played nine or ten heads-up tournaments in a row, and he's played like basically call it 10 of the top 20 people so far, and he's beaten all but one of them. When you're playing heads-up, you know, 1v1, that is the most, um, GTO understandable spot, meaning game theory optimal position. That's where computers can give you an enormous edge. The minute you add even a third player, the value of computers and the value of their recommendations basically falls off a cliff. Okay? So one way to think about it is Hellmuth is forced to play against people that are essentially trained like AIs. And so to be able to beat, you know, eight out of nine of them means that you are playing so orthogonally to what is considered game theory optimal. And you're overlaying human reasoning. The judgment to say, "Well, in this spot, I should do X, but I'm going to do Y." It's not dissimilar in chess, like what makes, you know, Magnus Carlsen so good. You know, sometimes he takes these weird lines, he'll sacrifice positions, you know, he'll overplay certain positions for, for certain, you know, bishops versus knights and all of these spots that are very confusing. And what it does is it throws people off their game. I think he just won a recent online tournament, and it's like by move six, there is no GTO move for his opponent to make because it's like out of the rule book. Maybe he read some game, you know, I read the quote, it was like he probably read some game in some bar in Russia in 1954, memorized it-

Lex Fridman 24:16 Mm-hmm.

Chamath Palihapitiya 24:16 ... and all of a sudden, by six moves in, the computer AI is worthless. So that's what makes Hellmuth great.... the- there is one person that I think is superior.

Lex Fridman 24:28 Mm-hmm.

Chamath Palihapitiya 24:29 Um, and I- and I think it's what Daniel also said, and I would echo that because I played Phil as well, but Phil Ivey is, um, the most well-rounded, cold-blooded-

Lex Fridman 24:42 (laughs)

Chamath Palihapitiya 24:43 ... bloodthirsty animal. (laughs) 'Cause he is- he's just so- and he l- he sees into your soul, Lex, in a way where you're just like, "Oh my God, stop looking at me."

Lex Fridman 24:53 Have you ever played him?

Chamath Palihapitiya 24:54 Yeah. Yeah, we've played. We've played and, you know, he crushes the games. Crushes the games.

Lex Fridman 24:59 So what- what does feeling crushed mean and feel like in poker? Is it like that you just can't read at all, you're being

Lex Fridman 25:07 constantly pressured, you feel off balance, you try to bluff and the person reads you perfectly, that kind of stuff?

Chamath Palihapitiya 25:14 It's, um, it's- it's a- this is a really, really excellent question because I think this has parallels to a bunch of other things. Okay, let's just use poker as a microcosm to explain a bunch of other systems or games. Maybe it's, um, running a company or investing, okay? So let's use those three examples, but we use poker to explain it. What does success look like? Well, success looks like you have positive expected value, right? In poker, the simple way to summarize that is your opponent, let's just say you and I are playing-

Lex Fridman 25:46 Mm-hmm.

Chamath Palihapitiya 25:46 ... are gonna make a bunch of mistakes. There's a bunch of it that's gonna be absolutely perfect, and then there's a few spots where you make mistakes. And then there's a bunch of places in the poker game where I play perfectly, and I make a few mistakes. Basically, your mistakes minus my mistakes is the edge, right?

Lex Fridman 26:07 Mm-hmm.

Chamath Palihapitiya 26:08 That's- that's pure- that's how poker works. If I make fewer mistakes than you make, I will make money and I will win. That is the objective of the game. Translate that into business. You're running a company. You have a team of employees, you have a pool of human capital that's capable of being productive in the world and creating something. But you are going to make mistakes in making that. Maybe it doesn't completely fit the market, maybe it's mispriced, maybe it actually doesn't require all of the people that you need so the margins are wrong. And then there's the competitive set of all the other alternatives that customer has. Their mistakes minus your mistakes is the expected value of Google, Facebook, Apple, et cetera. Okay? Now take investing. Every time you buy something, somebody else on the other side is selling it to you. Is that their mistake? We don't know yet. But their mistakes minus your mistakes is how you make a lot of money over long periods of time as an investor. Somebody sold you Google at $40 a share. You bought it and you kept it. Huge mistake on their part, minimal mistakes on your part. The difference of that is the money that you made. So life can be summarized in many ways in that way, so the question is, what can you do about other people's mistakes? And the answer is nothing. That is somebody else's game. You can try to influence them, you could try to subvert them, maybe you plant a spy inside of that other person's company to sabotage them. I guess there are things at the edges that you can do, but my firm belief is that life success really boils down to how do you control your mistakes? Now this is a bit counterintuitive. The way you control your mistakes is by making a lot of mistakes.

Lex Fridman 28:04 So taking risks is-

Chamath Palihapitiya 28:05 You have to.

Lex Fridman 28:06 ... is somehow- (laughs)

Chamath Palihapitiya 28:07 You have to. You have to.

Lex Fridman 28:08 ... a way to minimize the number of mistakes.

Chamath Palihapitiya 28:10 L- let's just say you want to find love.

Lex Fridman 28:12 Yeah.

Chamath Palihapitiya 28:12 You know, you want to find someone that you're-

Lex Fridman 28:13 Go on. (laughs)

Chamath Palihapitiya 28:14 ... deeply connected with.

Lex Fridman 28:15 Yeah.

Chamath Palihapitiya 28:17 Are you- do you do that by not going out on dates and-

Lex Fridman 28:21 Yes.

Chamath Palihapitiya 28:21 (laughs)

Lex Fridman 28:22 Sorry. Sorry. I guess so.

Chamath Palihapitiya 28:24 Okay. Yeah. You're gonna be the only person that thinks that's the answer to that question.

Lex Fridman 28:25 No, I'm- I'm joking. I'm joking. That's-

Chamath Palihapitiya 28:27 No, but you know what I mean? Like, you have to date people, you have to open yourself up, you have to be authentic and, like, you put- you- you give yourself a chance to get hurt.

Lex Fridman 28:36 Yes.

Chamath Palihapitiya 28:36 But you're a good person, so you know what happens when you get hurt? That is actually their mistake. Okay? And if you are inauthentic, that's your mistake. That's a controllable thing in you. You can tell them the truth, who you are, and say, "Here's my pluses and minuses." My point is there are very few things in life that you can't break down, I think, into that very simple idea. And in terms of your mistakes, society tells you, "Don't make them because we will judge you and we will look down on you." And I think the really successful people realize that actually, no, it's the cycle time of mistakes that gets you to success. Because your error rate will diminish the more mistakes that you make. You observe them, you figure out where it's coming from. Is it a psychological thing? Is it a, you know, cognitive thing? And then you fix it.

Lex Fridman 29:30 So the implied thing there is that there is, um, uh, in- in business, in investing, in poker, in dating, in life is that there's this platonic GTO, game theory optimal thing out there, and so when you say mistakes, you're always comparing to that optimal path you could have taken.

Chamath Palihapitiya 29:49 I think, uh, slightly different I would say mistake is maybe a bad proxy, but it's the best proxy I have for learning. But I'm using the language of what society tells you.

Lex Fridman 30:00 Sure. Got it.

Chamath Palihapitiya 30:01 Society tells you that when you try something and it doesn't work, it's a mistake. So I just use that word because it's the word that resonates most with most people.

Lex Fridman 30:11 Got it.

Chamath Palihapitiya 30:12 The real thing that it is is learning.

Lex Fridman 30:14 Yeah. It's like in neural networks, it's loss. (laughs)

Chamath Palihapitiya 30:16 It's a neural network, it's loss, exactly.

Lex Fridman 30:18 Uh, yeah, right, so you're using the mistake that is most, uh, the- the word that is most understandable es- especially by the way people experience it. I guess most of life is a sequence of mistakes. The problem is when you use the word mistake and you think about mistakes, it actually has a counterproductive effect of you becoming conservative and just be- being risk averse.So that's, if you fo- if you re- if you flip it and say, try to maximize the number of successes, somehow that leads you to take more risk. Um, mis- mistake scares people.

Chamath Palihapitiya 30:55 I think mistakes scare people because society likes these very simplified boundaries of who is winning and who is losing, and they want to reward people who make traditional choices and succeed. But the thing is, what's so corrosive about that is that they're actually not even being put in a position to actually make a quote unquote "mistake and fail." So I'll give you a, if you look at, like, getting into an elite school, right? Society rewards you for being in the Ivy Leagues in a way that, you know, in my opinion, incorrectly, doesn't reward you for being in a non-Ivy League school. There's a certain level of status and presumption of intellect and capability that comes with being there. Um, but that system doesn't really have a counterfactual because it's not as if you both go to MIT and Ohio State, and then we can see two versions of Lex Fridman so that we can figure out that the jig is up and there was no difference.

Lex Fridman 32:03 Mm-hmm.

Chamath Palihapitiya 32:03 (laughs) Right? And so instead, it reinforces this idea that there is no, uh, truth-seeking function. There is no way to actually make this thing whole. And so it tells you, "You have to get in here, and if you don't, your life is over, you've made a huge mistake." You know, or "You've failed completely." And so you have to find different unique ways of dismantling this. This is why, you know, part of what I've realized where I got very lucky is I had no friends in high school. I had a few cohort of acquaintances, but part of being so hypervigilant when I grew up was I was so ashamed of that world that I had to live in. I didn't want to bring anyone into it. I could not see myself that anybody would accept me. But the thing with that is that I had no definition of what expectations should be, so they were not guided by the people around me, and so I would escape to define my expectations.

Lex Fridman 33:06 It's interesting, but you didn't feel, um, like, your dad didn't put you in a prison of expectation? Or we- 'cause, like, that's, if you don't have a friend, like, th- so the flip side of that, you don't have any other signals, it's very easy to believe, like y- when you're in a cult that, like-

Chamath Palihapitiya 33:24 Well, he, you know, he was angry. He pushed me. He used me as a mechanism to alleviate his own frustration. And this may sound very crazy, but he also believed in me. And so that's what created this weird duality where you were just, I was always confused about-

Lex Fridman 33:42 That you could be somebody great? He believed that you could be-

Chamath Palihapitiya 33:46 He did. I didn't believe him-

Lex Fridman 33:47 ... somebody truly special?

Chamath Palihapitiya 33:49 ... because I couldn't reconcile then the other half of the day, you know, those behaviors. But what it allowed me to do was I escaped in my mind and I found these archetypes around me that were saviors to me. So, you know, I grew up in Ottawa, Ontario, Canada. I grew up right at the point where the telecom boom was happening. Companies like Nortel and Newbridge Networks and Mitel, Bell Northern Research, these were all built in- in- in the suburbs of Ottawa. And so there were these larger than life figures, entrepreneurs, Terry Matthews, Michael Copeland, and so I thought, "I'm gonna be like them." I would read Forbes Magazine, I would read Fortune Magazine. I would look at the rich people on that list and say, "I would be like them," not knowing that maybe that's not who you wanted to be, but it was a lifeline, and it kept my mind relatively whole because I could direct my ambition in a direction. And so why that's so important, just circling back to this, is I didn't have a group of friends who were like, "I'm gonna go to community college." You know, I didn't have a group of friends that said, "Well, you know, the goal is just to go to university, get a simple job and like, you know, join the public service, have a good life." And so because I had no expectations and I was so afraid to venture out of my own house, I never saw what middle class life was like, and so I never aspired to it. Now, if I was close to it, I probably would have aspired to it because I, my parents in their best year made 32,000 Canadian together. And if you're trying to raise a family of five people on $32,000, it's a complicated job. And most of the time they were probably making 20-something thousand, and I was working since I was 14, so I knew that our station in life was not the destination (laughs) . We had to get out. Um, but because I didn't have an obvious place, it's not like I had a best friend whose house I was going to and I saw some normal functional home. If I had had that, in this weird way, I would have aspired to that.

Lex Fridman 35:58 What was the worst job you had to do?

Chamath Palihapitiya 36:00 The best job, but the worst job, was I worked at Burger King when I was 14 years old and I would do the closing shift, and that was from like 6:00 PM till about 2:00 in the morning. And in Ontario where I lived, Ottawa borders Quebec. In Ontario, the drinking age is 19. You can see where I'm going with this. The drinking age in Quebec is 18, and that year made all the difference to all these kids.

Lex Fridman 36:27 Mm-hmm.

Chamath Palihapitiya 36:28 And so they would go get completely drunk, they would come back, they would come to the Burger King. You know, you would see all these kids you went to high school with. Can you imagine how mortifying it is? You know, you're working there in this get up. And they would light that place on fire, vomit everywhere, puking, pooing, peeing.And when the thing shuts down at one o'clock, you know, you gotta clean that all up. All of it. Changing the garbage, taking it out. It was, um, a grind. And it really teaches you, okay, I do not want this job. (laughs) I don't want it.

Lex Fridman 37:08 But it's funny that that didn't push you towards the stability and the security of the middle class, like, life.

Chamath Palihapitiya 37:13 I didn't have any good examples of that. I didn't have those around me. I was so ashamed. I could have never built a relationship where I could have seen those interactions to want that. And so my desires were framed by these two random rich people that lived in my town who I'd never met, and what I read in magazines about people like Bill Gates and Warren Buffett.

Lex Fridman 37:36 You were an early senior executive at Facebook, uh, during a period of a lot of scaling in the company history. I mean, it's actually a fascinating period of human history in terms of technology. Well, in terms of human civilization, honestly. Uh, what did you learn from that time about what it takes to build and scale a successful tech company, a company that has almost immeasurable impact on the world?

Chamath Palihapitiya 38:06 That was an incredible moment in time because everything was so new. To your point, like even how the standards of Web 2.0 at that time were being defined, we were defining them, you know. I mean, I think if you, if you look in sort of the, if you search in the patents, um, patent library, there's a bunch of these patents that, like, me and Zuck have for, like, random things like cookies-

Lex Fridman 38:28 Mm-hmm.

Chamath Palihapitiya 38:28 ... you know, or, like, cross-site JavaScript. Like, all these crazy things that are just like these duh kind of ideas in 2023, we had to invent our way around. How do websites communicate with each other, you know? How do we build in the cloud versus in a data center? How do we actually have high performance systems?

Lex Fridman 38:46 You mentioned data science, the term and the idea.

Chamath Palihapitiya 38:49 We had to, we invented this, I invented this thing called data scientist because we had a PhD from Google that refused to join unless, because he got a job offer that says data analyst.

Lex Fridman 38:58 Yeah.

Chamath Palihapitiya 38:59 Um, and so we said, "Call him a scientist." 'Cause he was a PhD in particle physics. So he really, you know, he was a scientist. And I said, "Great, you're a scientist here." Yeah.

Lex Fridman 39:05 And that launched a discipline.

Chamath Palihapitiya 39:08 That launched a discipline.

Lex Fridman 39:08 I mean, a term, you know, what's, uh, a rose by any other name. But yeah, like, you know, sometimes words like this can launch entire fields, and it did in that case. And you d- didn't, I mean, I guess at that time, you didn't anticipate the impact of machine learning on the, the entirety of this whole process, 'cause you need machine learning to have both ads and recommender systems to have the feed for the social networks, all that good stuff.

Chamath Palihapitiya 39:33 Exactly right. The first real scaled version of machine learning, not AI, but machine learning, was this thing that Facebook introduced called PYMK, which is People You May Know. And the simple idea was that, "Can we initiate a viral mechanic inside the application, where you log in, we grab your credentials, we go to your email inbox, we harvest your address book, we do a compare, we make some guesses, and we start to present other people that you may actually know that may not be in your address book?" Really simple, you know, couple joins of some tables, whatever. And it started to just go crazy. And the number of people that you were, you were creating this density and entropy inside this social graph with what was some really simple basic math. And that was eye-opening for us. And what it, what it led us down this path of is really understanding the power of, like, all this machine learning. And so that infused itself into News Feed, you know, and how the content that you saw could be tailored to who you were and the type of person that you were. So, there was a moment in time that all of this stuff was so new. Um, how did you translate the app to multiple languages? How do you launch the company in all of these countries?

Lex Fridman 40:46 How much of it is just kind of stumbling into things using your best, like, first principles, gut thinking? And how much is it, like, five, 10, 15, 20-year vision? Like, how much was thinking about the future of the internet and the metaverse and the humanity and all that kind of stuff? Like, 'cause the News Feed-

Chamath Palihapitiya 41:07 I'll say, I'll-

Lex Fridman 41:07 ... sounds trivial-

Chamath Palihapitiya 41:08 I'll say something very-

Lex Fridman 41:09 ... but that's, like, changes everything.

Chamath Palihapitiya 41:11 Well, you have to remember, like, you know, News Feed was named, and we had this thing where we would just name things what they were. And at the time, all of these other companies, and if you go back into the Wayback Machine you can see this, people would vent, would invent, you know, an i-, you know, an MP3 player, and they would come up with some crazy name. Or they would invent a software product and come up with a crazy name, right? And it sounded like the pharma industry, you know, blocazamab-

Lex Fridman 41:45 (laughs)

Chamath Palihapitiya 41:45 ... you know, tag your best friends. (laughs)

Lex Fridman 41:47 Yeah.

Chamath Palihapitiya 41:47 And you think, "What is this? This makes no sense." And, you know, this was Zuck's thing. He was like, "Well, this is a feed of news, so we're gonna call it News Feed. This is where you tag your photos, so we're gonna call that Photo Tagging." (laughs) I mean, literally, you know, pretty obvious stuff. Um, so the thing, the way that those things came about though was very experimentally. And this is where I think it's really important for people to understand. I think Bezos explains this the best. There is a tendency after things work to create a narrative fallacy because it feeds your ego.

Lex Fridman 42:23 Yeah.

Chamath Palihapitiya 42:23 And you want to have been the person that saw it coming. And I think it's much more honest to say, "We were very good probabilistic thinkers that tried to learn as quickly as possible, meaning to make as many mistakes as possible." You know, I mean, if you look at this very famous placard that Facebook had from back in the day, what did it say? It said, "Move fast and break things."

Lex Fridman 42:50 Mm-hmm.

Chamath Palihapitiya 42:50 In societal language, that's saying make mistakes as quickly as you can....because the minute you break something, that's, you don't do that by design. It's not a feature. Theoretically it's a bug. But he understood that. And we embraced that idea. Um, I used to run this meeting once a week where the whole goal was, I want to see that there was a thousand experiments that were run, and show me them all, from the dumbest to the most impactful. And we would go through that loop. And what did it train people? Not that you got celebrated for the right answer, but you got celebrated for trying. I ran twel- twelve experiments, twelve failed. And we'd be like, "You're the best."

Lex Fridman 43:30 C- can I just take a small tangent on that, is that move fast and break things has become as, uh, like a catchphrase of the thing that embodies the toxic culture of Silicon Valley in today's, uh, discourse, which, uh, confuses me. Of course, words and phrases get c- sort of captured and, and so on.

Chamath Palihapitiya 43:52 It becomes very reductive. You know, that's a very loaded set of words that together can be ... Many years later, c- people can view very reductive.

Lex Fridman 44:00 Can you steel man each side of that? So-

Chamath Palihapitiya 44:03 Yeah.

Lex Fridman 44:04 ... p- pro move fast and break things-

Chamath Palihapitiya 44:06 Yeah.

Lex Fridman 44:06 ... and against-

Chamath Palihapitiya 44:07 Yeah. Absolutely.

Lex Fridman 44:07 ... move fast and break things. (laughs)

Chamath Palihapitiya 44:09 Um, so I think the pro of move fast and break things is saying the following. There's a space of things we know and a massive space of things we don't know. And there's a rate of growth of the things we know, but the rate of growth of the things we don't know is actually, we have to assume, growing faster. So the most important thing is to move into the space of the things we don't know as quickly as possible. And so in order to acquire knowledge, we're going to assume that the failure mode is the nominal state. And so we just need to move as quickly as we can, break as many things as possible, which means, like, things are breaking in code, do the, you know, root cause analysis, figure out how to make things better, and then rapidly move into the space. And he or she who moves fastest into that space will win.

Lex Fridman 45:04 It doesn't imply carelessness, right? It doesn't imply moving f- fast without also aggressively picking up the lessons from the mistakes you make.

Chamath Palihapitiya 45:16 Well, again, that, let's, that's steel manning the pro, which is-

Lex Fridman 45:19 Yeah.

Chamath Palihapitiya 45:19 ... it's a thoughtful, um, movement around velocity and acquisition of knowledge. Now let's steel man the, the con case. When these systems become big enough, there is no more room to experiment in an open-ended way because the implications have broad societal impacts-

Lex Fridman 45:44 Yeah.

Chamath Palihapitiya 45:44 ... that are not clear upfront. So let's take a different, less controversial example. If we said, you know, Lipitor, um, worked well for all people except South Asians, and there's a specific immuno response that we can iterate to, and if we move quickly enough, we can run 10,000 experiments and we think the answer is in that space. Well, the problem is that those 10,000 experiments may kill 10 million people. So you have to move methodically. When that drug was experimental, and it wasn't being given to 500 million people in the world, moving fast made sense because you could have a pig model, a mouse model, a monkey model. You could figure out toxicity. But we picked all that low-hanging fruit. And so now these small iterations have huge impacts that need to be measured and implemented. Different example is like, you know, if you work at Boeing and you have an implementation that gives you a 2% efficiency by reshaping the wing or adding winglets, there needs to be a methodical move slow, be right process, because mistakes when they compound when it's already implemented and at scale have huge externalities that are impossible to measure until after the fact. And you see this in the 737 Max. So that's how one would steel man the, the con case, which is that when an industry becomes critical, you got to slow down.

Lex Fridman 47:15 This makes me sad because some industries, like Twitter and Facebook are a good example, they achieve scale very quickly before really exploring the big area of things to learn. So you basically pick one low-hanging fruit (laughs) and that became your huge success. And now you're sitting there with that stupid fruit-

Chamath Palihapitiya 47:40 Completely. Well, so you're, you're, I think, so as an example, like, you know, if you had to, you know, if, if, if I was running Facebook for a day, you know, the big opportunity, in my opinion, was really not the metaverse, but w- it was actually getting the closest that anybody could get to AGI. And if I had to steel man that product case, here's what, how I would have pitched it to the board and to Zuck. I would have said, "Listen, there are three and a half billion people monthly using this thing. If we think about human intelligence very reductively, we would say that there is a large portion of it which is cognitive, and then there's a large portion of it which is emotional. We have the best ability to build a multimodal model that basically takes all of these massive inputs together to try to intuit how a system would react to all kinds of stimuli." That, to me, would have been a profound leap forward for humanity.

Lex Fridman 48:40 Can you dig into that a little bit more? So in terms of, uh, now this is a board meeting, how would that make Facebook money?

Chamath Palihapitiya 48:48 I think that you have all of these systems over time that, that we don't know could benefit from, um, some layer of reasoning to make it better. Um-What does Spotify look like when, instead of just a very simple recommendation engine, it actually understands sort of your emotional context and your mood and can move you to a body of music that you would like? What does it look like if, you know, your television, instead of having to go and channel surf, you know, 50,000 shows on a horrible UI, you know, instead just has a sense of what you're into and shows it to you? Um, what does it mean when you get in your car and it actually drives you to a place because you should actually eat there even though you don't know it? These are all random things that make no sense a priori, but it starts to make the person or the provider of that service the critical reasoning layer for all these everyday products that today would look very flat without that reasoning. And I think you license that and you make a lot of money. So in many ways, instead of becoming more of the pixels that you see, you become more of the bare metal that actually creates that experience. And if you, and if you look at the companies that are multi-decade legacy kinds of businesses, the thing that they have done is quietly and surreptitiously move down the stack. You never move up the stack to survive. You need to move down the stack. So if you take that OSI reference stack, right? These layers of how you build an app from the physical layer to the transport layer all the way up to the app layer, you can map from the 1980s all the big companies that have been created, right? All the way from Fairchild Semiconductor and Nat Semi to Intel to Cisco to 3Com, you know, Oracle, Netscape at one point, all the way up to the Googles and Facebooks of the world. But if you look at where all the lock-in happened, it's by companies like Apple who used to make software saying, "Mm, I'm going to get one close. I'm going to make the bare metal and I'm going to become the platform." Or Google, same thing, "I'm going to create this dominant platform and I'm going to create a substrate that organizes all this information that's just omnipresent and everywhere." So the key is, if you are lucky enough to be one of these apps that are in front of people, you better start digging quickly (laughs) and moving your way down and get out of the way and disappear. But by disappearing, you will become much, much bigger and it's impossible to usurp you.

Lex Fridman 51:33 Yeah, I 100% agree with you. Um, that's why you're so smart. This is, uh, the deep personalization and the algorithms that enable deep personalization almost like a operating system layer, so pushing away from the interface and the actual system that does the personalization.

Chamath Palihapitiya 51:52 Yeah.

Lex Fridman 51:52 I think the challenges there, there's obviously technical challenges, but there's also societal challenges that... It's like in a relationship. If you have an intimate algorithmic connection with individual humans, you can do both good and bad, and so there's risks that you're taking.

Chamath Palihapitiya 52:12 Yeah.

Lex Fridman 52:13 You can... So if you're making a lot of money now as Twitter and Facebook with ads, surface layer ads, what is the incentive to take the risk of guiding people more?

Chamath Palihapitiya 52:25 Well-

Lex Fridman 52:25 Because you can hurt people, you can piss off people, you can, um... I mean, it... (laughs) There is a cost to forming a more intimate relationship with the users (laughs) in the short term, I think.

Chamath Palihapitiya 52:37 You said a really, really key thing which is, um, which was a really great emotional, instinctive reaction which is, when I said the AGI thing, you said, "Well, how would you ever make money from that?" And that is the key. The presumption is that this thing would not be an important thing at the beginning. And I think what that allows you to do if you were Ti- Twitter or Google or Apple or Facebook, anybody, Microsoft, embarking on building something like this, is that you can actually have it off the critical path and you can experiment with this for years if that's what it takes to find a version one that is special enough where it's worth showcasing. And so in many ways you get the free option. You're going to be spending, any of these companies will be spending tens of billions of dollars in OpEx and CapEx every year on all kinds of stuff. It is not a thing that money, um, actually makes more likely to succeed. In fact, you actually don't need to give these kinds of things a lot of money at all because starting in 2023, or right now, you know, you have the two most important tectonic shifts that have ever happened in our lifetime in technology. They're not talked about, but these things allow AGI, I think, to emerge over the next 10 or 15 years where it wasn't possible for. The first thing is that the marginal cost of energy is zero. You're not gonna pay for anything anymore, right? And we can double click into why that, why that is and the second is the marginal cost of compute is zero. And so when you, when you take the multiplication or, you know, if you want to get really fancy mathematically, the convolution of these two things together, it's going to change everything. So think about what a billion dollars gets today, and we can use OpenAI as an example. A billion dollars gets OpenAI a handful of functional models and a pretty fast iterative loop, right? But imagine what, um, OpenAI had to overcome. They had to overcome a compute challenge, they had to strip together a whole bunch of GPUs, they had to build all kinds of scaffolding software, they had to find data center support. That consumes all kinds of money so that billion dollars didn't go that far. So it's a testament to how clever that OpenAI team is. But in four years from now when energy costs zero and basically GPUs are like, you know, they're falling off a truck and y- and you can use them for effectively for free-... now all of a sudden, a billion dollars gives you some amount of teraflops of compute that is probably the total number of teraflops available today in the world. Like, that's how gargantuan this move is when you take these two variables to zero.

Lex Fridman 55:21 There's, like, a million things t- to ask. I almost don't want to get distracted by the marginal cost of energy going to zero 'cause I- I have no idea what you're talking about there. It's fascinating.

Chamath Palihapitiya 55:31 Can I give you the 30 seconds?

Lex Fridman 55:31 Sure. (laughs) Okay.

Chamath Palihapitiya 55:32 Okay.

Lex Fridman 55:32 Yes. Okay. (laughs)

Chamath Palihapitiya 55:33 So if you look inside of the two most progressive states, the three most progressive states, New York, California, and Massachusetts, a lot of left-leaning folks, a lot of people who believe in climate science and climate change, the energy costs in those three states are the worst they are in the entire country. And energy is compounding at 3 to 4% per annum. So every decade to 15 years, energy costs in these states double. In some cases and in some months, our energy costs are increasing by 11% a month. But the ability to actually generate energy is now effectively zero. The cost per kilowatt hour to put a solar panel on your roof and a battery wall inside your garage, it's the cheapest it's ever been. They're- these things are the most efficient they- they've ever been. And so to acquire energy from the sun and store it for your use later on literally is a zero-cost proposition.

Lex Fridman 56:29 So what's- how do you explain the gap between the cost going up?

Chamath Palihapitiya 56:31 Great question. So this is the other side of regulatory capture, right? You know, we all fight to build monopolies. Well, there are monopolies hiding in plain sight. The utilities are a perfect example. There are 100 million homes in America. There are about 1,700 utilities in America. So they have captive markets. But in return for that captive market, the law says need to invest a certain amount per year in upgrading that power line, in changing out that turbine, in making sure you transition from coal to wind or whatever. Just as an example, upgrading power lines in the United States over the next decade is a $2 trillion proposition. These 1,700 organizations have to spend, I think it's a quarter of a trillion dollars a year just to change the power lines. That is why even though it costs nothing to make energy, you are paying double every five or- every seven or eight years. It's CapEx and OpEx of a very brittle old infrastructure. It's like you trying to build an app and being forced to build your own data center. And you say, "But wait, I just want to write to AWS. I just want to use GCP. I just want to move on. All that complexity is solved for me." And some law says, "No, you can't. You gotta use it." So that's what consumers are dealing with, but it's also what industrial and manufacturing organizations... It's what we all deal with.

Lex Fridman 57:56 So how do we get rid of ourselves of this old infrastructure that we're paying for?

Chamath Palihapitiya 58:01 So the thing that's happening today which I think is... This is why I think it's the most important trend right now in the world, uh, is that 100 million homeowners are each going to become their own little power plant and compete with these 1,700 utilities. And that is a great-

Lex Fridman 58:20 Uh, in the United States or globally?

Chamath Palihapitiya 58:21 No, just- just deal with the United States for a second, because I think it's easier to see here. 100 million homes, solar panel on the roof, and by the way, just to make it clear, the sun doesn't need to shine, right? These- these panels now work where you have these UV bands that can actually extrapolate beyond the visible spectrum, so they're usable in all weather conditions. And a simple system can support you collecting enough power to not just run your functional day-to-day life, but then to contribute what's left over back into the grid for Google's data center or Facebook's data center where you get a small check. The cost is going to zero.

Lex Fridman 59:00 How obvious is this to people? You're making it sound-

Chamath Palihapitiya 59:03 Not obvious.

Lex Fridman 59:03 Okay. So, because this is a pretty profound prediction. If the cost is- is going to zero, that, I mean, the compute, the cost of compute going to zero, I can-

Chamath Palihapitiya 59:14 So the cost of compute going to zero is simple to see. It's bas-

Lex Fridman 59:16 ... can kind of understand. But the energy seems like a radical prediction of yours.

Chamath Palihapitiya 59:20 Well, it- it's just, it's just naturally what's happening, right? Now- now, let me, let me give you a diff- a different way of explaining this. If you look at any system, there's a really important thing that happens. It's what, uh, Clay Christensen calls crossing the chasm. If you explained it numerically, here's how I'd explain it to you, Lex. If you introduce a disruptive product, typically what happens is the first 3 to 5% of people are these zealous believers.

Lex Fridman 59:46 Mm-hmm.

Chamath Palihapitiya 59:47 And they ignore all the logical reasons why this product doesn't make any sense because they believe in the proposition of the future and they buy it. The problem is at 5%. If you want a product to get to mass market, you have one of two choices, which is you either bring the cost down low enough or the feature set becomes so compelling that even at a high price point. An example of the latter is the iPhone. The iPhone today, the 14 iPhone, costs more than the original iPhone. It's probably doubled in price over the last 14 or 15 years. But we view it as an essential element of what we need in our daily lives. It turns out that battery EVs and solar panels are an example of the former, because people like President Biden with all of these subsidies have now introduced so much money for people to just do this, where it is a money-making proposition for 100 million homes. And what you're seeing as a result are all of these companies who want to get in front of that trend. Why? Because they want to own the relationship with 100 million homeowners, they want to manage the power infrastructure, Amazon, Home Depot, Lowe's, you know, you c- just name the company. So if you do that and you control that relationship, they're gonna show you, they're gonna... You know, for example, Amazon will probably say, "If you're a member of Prime..."... we'll stick the panels on your house for free. We'll do all the work for you for free. And it's just a feature of being a member of Prime. And we'll manage all that energy for you. It makes so much sense, and it is mathematically accretive for Amazon to do that. It's not accretive for the existing energy industry because they get blown up. It's extremely accretive for peace and prosperity. If you think the number of wars we fight over natural resources, take them all off the table if we don't need energy from abroad. There's no reason to fight. (laughs) You know, there's, you'd have to find a reason to fight. Um, meaning, sorry, there, there'd be a moral reason to fight. But the last number of wars that we fought, uh, were not as much rooted in morality as they were rooted in-

Lex Fridman 1:02:00 Yeah, it feels like they are very much rooted in, uh, conflict-

Chamath Palihapitiya 1:02:03 Resource scarcity.

Lex Fridman 1:02:04 ... over, over resources-

Chamath Palihapitiya 1:02:06 Yeah.

Lex Fridman 1:02:06 ... e- energy specifically.

Chamath Palihapitiya 1:02:07 And then, sorry, just the last thing I want to say, I keep interrupting you, apologies.

Lex Fridman 1:02:09 No.

Chamath Palihapitiya 1:02:09 But the chips, all, what, what people want to say is that, you know, now that we're at two and three nanometer scale for typical kind of like transistor fab, we're done. And, you know, forget about transistor density, forget about Moore's Law, it's over. And I would just say, no, look at teraflops. And really, teraflops is the combination of CPUs, but much, much less important, and really is the combination of ASIC, so application-specific ICs, and GPUs. And so you put the two together, I mean, if I gave you a billion dollars five years from now, the amount of damage you could do, damage in a good way-

Lex Fridman 1:02:47 Mm-hmm.

Chamath Palihapitiya 1:02:48 ... in terms of, you know, building racks and racks of GPUs, the kind of models that you could build, the training sets and the data that you could consume to solve a problem, it's, it's enough to, to do something really powerful, whereas today, it's not yet quite

Chamath Palihapitiya 1:03:01 enough.

Lex Fridman 1:03:02 So there's this really interesting idea that you talk about i- in terms of Facebook and Twitter that's connected to this, that if you were running sort of Twitter or Facebook, that you would move them all to, like, AWS. So you would, uh, ha- have somebody else do com- the compute, the infrastructure. It probably, if you could explain that reasoning, means that you believe in this idea of energy going to zero, compute going to zero, so let people that are optimizing that do the best job.

Chamath Palihapitiya 1:03:33 And, and I think that's a, you know, the... Initially, in the early 2000s and the beginning of the 2010s, if you were big enough scale, oh sorry, everybody was building their own stuff. Then between 2010 through 2020, really, the, the idea was everybody should be on AWS except the biggest of the biggest folks. I think in the 2020s and 30s, I think the answer is actually everybody should be in these public clouds. And the reason is the engineering velocity of the guts. So, you know, take a simple example which is, you know, we have not seen a massive iteration in database design until Snowflake, right? I think maybe PostgreSQL was, like, the last big turn of the dial. Why is that? I don't exactly know except that everybody that's on AWS and everybody that's on GCP and Azure gets to now benefit from 100 plus billion dollars of aggregate market cap rapidly iterating, making mistakes, fixing, solving, learning, and that is a best-in-class industry now, right? Um, then there's going to be all these AI layers around analytics, so that app companies can make better decisions. All of these things will allow you to build more nimble organizations because you'll have this federated model of development. I'll take these things off the shelf, maybe I'll roll my own stitching over here, because the thing that where you make money is still for most people in how the apps provision an experience to a user. And everybody else can make a lot of money just servicing that. So they work in a really, um, they, they play well together in the sandbox. So in the future, everybody just should be there. It doesn't make sense for anybody, I don't think, because, you know, if you were to roll your own data centers, you know, for example, like, Google for a long time had these massive leaps where they had GFS and Bigtable. Those were really good in the 2000s and 2010s, and this is not to throw shade at Google. It's very hard for whatever exists that is a, that is the progeny of GFS and Bigtable to be anywhere near as good as $100 billion industries' attempt to build that stack. And, and you're putting your organization under enormous pressure to be that good.

Lex Fridman 1:05:59 I guess the implied risk taken there is that you could become the next AWS. Like, um, Tesla doing some of the compute in-house. I guess the bet there is that you can become the next, the next AWS for the new wave of computation if that level, if that kind of computation is different. So if it's machine learning, I don't know if anyone's won that battle yet, which is machine learning-centric compute.

Chamath Palihapitiya 1:06:28 Well, I think that software has a very powerful property in that there's a lot of things that can happen asynchronously so that real-time inference can be actually really lightweight code deployment.

Lex Fridman 1:06:41 Mm-hmm.

Chamath Palihapitiya 1:06:41 And that's why I think you can have a very federated ecosystem inside of, inside of all of these places. Tesla is very different because in order to build the best car, it's kind of like trying to build the best iPhone, which is that you need to control it all the way down to the bare metal in order to do it well. And that's just not possible if you're trying to be a systems integrator, which is what everybody other than this modern generation of car companies have been. And they've done a very good job of that. But it won't be the experience that allows you to win in, in the next 20 years.

Lex Fridman 1:07:17 So let's linger on this social media thing. So if you, you said if you ran Facebook for a day. Let's, let's, let's extend that. If you were to build a new soc- social network today, how would you fix Twitter? How would you fix social media? If you want to answer a different question is if you were Elon Musk, somebody you know, and you were taking over Twitter, what would you fix?

Chamath Palihapitiya 1:07:45 I've thought about this a little bit. Um... First of all, let me give you a backdrop. I wouldn't actually build a social media company at all. And the answer is... The, the reasoning is the following. Um, I really tend to believe, as you've probably gotten a sense of sort of patterns and probabilities. And if you said to me, "Chamath, probabilistically answer where, where are we going in apps and social experiences?" What I would say is, "Lex, we spent the first decade building platforms and getting them to scale." And if you want to think about it, again, back to sort of this poker analogy, others' mistakes minus your mistakes is the value. Well, the value that was captured was trillions of dollars, essentially to Apple and to Google. And they did that by basically, um, attracting billions of monthly active users to their platform. Then this next wave were the apps, Facebook, QQ, Tencent, TikTok, Twitter, Snapchat, that whole panoply of, of apps. And interestingly, they were in many ways an atomized version of the platforms.

Lex Fridman 1:08:59 Mm-hmm.

Chamath Palihapitiya 1:09:00 Right? They sat on top of them. They were an ecosystem participant, but the value they created was the same, trillions of dollars of enterprise value, billions of monthly active users.

Lex Fridman 1:09:13 Mm-hmm.

Chamath Palihapitiya 1:09:14 Well, there's an interesting phenomenon that's kind of hiding in plain sight, which is that the next most obvious atomic unit are content creators. Now, let me give you two examples. Lex Fridman, this random crazy guy.

Lex Fridman 1:09:28 Yeah.

Chamath Palihapitiya 1:09:29 Uh, MrBeast.

Lex Fridman 1:09:30 Mm-hmm.

Chamath Palihapitiya 1:09:31 You know, Jimmy Donaldson. Just the two of you alone, add it, add it up, okay?

Lex Fridman 1:09:35 Mm-hmm.

Chamath Palihapitiya 1:09:35 And you guys are going to approach, in the next five years, a billion people. The only thing that you guys haven't figured out yet is how to capture trillions of dollars of value. Now, maybe you don't want to and maybe that's not your stated mission.

Lex Fridman 1:09:46 Well, yeah. Right, right. But let's just look at MrBeast alone because he is trying to do exactly that probably.

Chamath Palihapitiya 1:09:51 Yeah, and I think Jimmy is going to build an enormous business. But if you take Jimmy and all of the other content creators, right? You guys are atomizing what the apps have done. You're providing your own curated news feeds. You're providing your own curated communities. You're allowed... You let people move in and out of these things in a very lightweight way and value is accruing to you. So the honest answer to your question is I would focus on the content creator side of things because I believe that's where the puck is going. That's a much more important shift in how we all consume information content and are entertained. It's through brands like you, individual people that we can humanize and understand are the filter.

Lex Fridman 1:10:34 But aren't you just arguing against the point you made earlier, which is what you would recommend is the invest in the AGI, the, the depersonalization? 'Cause what-

Chamath Palihapitiya 1:10:44 No, because they, they, they could still be a participant. In that, in that end state, if that happens, you have the option value of being an enabler of that, right? You can help improve what they do. Again, you can be this bare metal service provider where you can be a tax.

Lex Fridman 1:11:00 Yeah.

Chamath Palihapitiya 1:11:00 Right? You can participate in everything that you do, every question that's asked, every comment that's curated. If you could have more intelligence as you provide a service to your fans and your audience, you would probably pay a small percentage of that revenue. I suspect all content creators would. And so it's that stack of services that is like a smart human being. It's like, you know, how do you help produce this information? You would pay a producer for that. I m- I mean, maybe you wouldn't, but... So back to your question. So what would I do? I think that you have to move into that world pretty aggressively. Um, I think that right now, you first have to solve what is broken inside of these social networks. And I don't think it's a technical problem. So just to put it out there, I don't think it's, um, you know, it's one where there are these nefarious organizations. That happens. Brigading XYZ, that happens. But the real problem is a psychological one that we're dealing with, which is people through a whole set of situations have lost belief in themselves. And I think that that comes up as this very virulent form of rejection that they try to put into these social networks. So if you look inside the comments on anything, like you could have a com- Like you could have a person that says on Twitter, "I saved this dog from a fiery building." And there would be negative commenters. And you're like, well, again, put yourself in their shoes. What do you... How do I steel man their case? I do this all the time. You know, I get people throw shade at me. I'm like, "Okay, let me steel man their point of view." And the best that I can come up with is, "You know, I'm working really hard over here. I'm trying. I played by all the rules that were told to me. I've played well. I've played fairly. And I am not being rewarded in a system of value that you recognize, and that is making me mad. And now I need to cope and I need to vent." So back in the day, my dad used to drink. He would make me go get things to hit me with. Today, you go to Twitter, you spot off, you try to deal with the latent anger that you feel. So a social network has to be designed, in my opinion...... to solve that psychological corner case because it is what makes a network unusable. To get real density, you have to find a way of moving away from that toxicity because it ruins a product experience. You could have the best pixels in the world, but if people are virulently spitting into their keyboards, other people are just gonna say, "You know what? I'm done with this. It doesn't make me feel good." So, the social network has to have a social cost. You can do it in a couple of ways. One is where you have real world identity. So then, there's a cost to being virulent, and there's a cost to being caustic. A second way is to actually just overlay an economic framework so that there's a more pertinent economic value that you assign to basically spouting off. And the more you want to spend, the more you can say. And I think both have a lot of value. I don't know what the right answer is. I tend to like the latter. I think real world identity shuts down a lot of debate because there's still too much, um, you know, there's a sensation that there, that there'll be some retribution. Um, so I think there's more free speech over error, but it cannot be costless, because in that, there's a level of toxicity that just makes these products unusable.

Lex Fridman 1:14:38 A third option. And by the way, all of these can work together. If we look at this, what you call the corner case, which is hilarious, what I would call the human condition, uh-

Chamath Palihapitiya 1:14:50 (laughs)

Narrator 1:14:51 (laughs)

Lex Fridman 1:14:53 (laughs) Which, which is, uh, you know, that anger is rooted with the, the challenges of life. And what about having a, um, an algorithm that shows you what you see, that's personalized to you, and helps you maximize your personal growth in the long term such that you're challenging yourself, you're improving, you're learning? There's just enough of criticism to, uh, keep you on your toes, but just enough of, like, the dopamine rush to keep you, uh, entertained and finding that balance for each individual person.

Chamath Palihapitiya 1:15:33 But you, you just described an AGI of a very empathetic, well-rounded friend.

Lex Fridman 1:15:38 Yes, exactly. And a- and then, you can throw that person, even anonymous, into a pool of discourse-

Chamath Palihapitiya 1:15:44 100%.

Lex Fridman 1:15:45 ... and they would be better.

Chamath Palihapitiya 1:15:46 I think you're absolutely right. And it's a-

Lex Fridman 1:15:48 O-

Chamath Palihapitiya 1:15:48 ... very, very, very elegant way of stating it. You're absolutely right.

Lex Fridman 1:15:50 But like you said, the AGI might be a few years away, so that's a huge investment. Like, my concern, my gut feeling is this age, thing we're calling AGI, is actually not that diff- difficult to build technically. But it requires a certain culture, and it requires a certain, certain risks to be taken.

Chamath Palihapitiya 1:16:11 I think you could reductively boil down the human intellect into cognition and emotion. And, you know, depending on who you are and depending on the moment-

Lex Fridman 1:16:24 Mm-hmm.

Chamath Palihapitiya 1:16:24 ... they're weighted very differently, obviously. (laughs) Um, cognition is so easily done by computers that we should assume that that's a solved problem. So, our differentiation is the reasoning part. It's the emotional overlay. It's that, it's the empathy. It's the ability to steel man the opposite person's case and f- and feel why that person, you know, you can forgive them e- without excusing what they did, as an example. Um, th- that is a very difficult thing, I think, to capture in software. But I think it's a matter of when, not if.

Lex Fridman 1:16:59 If done crudely, it takes a form of censorship, just banning people off the platform.

Chamath Palihapitiya 1:17:06 Absolutely.

Lex Fridman 1:17:06 Let, let me ask you

Lex Fridman 1:17:08 some tricky questions.

Chamath Palihapitiya 1:17:10 Okay.

Lex Fridman 1:17:10 Uh, do you think Trump should have been removed from Twitter?

Chamath Palihapitiya 1:17:13 No.

Lex Fridman 1:17:13 What's the s- what's the pro case? Can you... I'm having fun here.

Chamath Palihapitiya 1:17:18 Steel man.

Lex Fridman 1:17:19 Can you steel man each side?

Chamath Palihapitiya 1:17:20 Yeah, yeah. Um, let's steel man the get him off the platform. Here we have a guy who, um, is virulent in all ways. He promotes confrontation. Um, he lacks decorum. He incites the fervent believers of his cause to act up and push the boundaries, bordering on and potentially even including breaking the law. He does not observe the social norms of a society that keep us well functioning, including an orderly transition of power. If he is left in a moment where he feels trapped and cornered, he could behave in ways that will confuse the people that believe in him, to act in ways that they so regret that, um, it could bring our democracy to an end, or create so much damage or create a wound that's so deep, it will take years of conflict and years of confrontation to heal it. We need to remove him, and we need to do it now. It's been too long. We've let it go on too long. The other side of the argument would be, he was a duly elected person whose views have been run over for way too long, and he uses the ability to say extreme things in order to showcase how corrupt these systems have become and how insular these organizations are in protecting their own class. And so, if you really want to prevent class warfare, and if you really want to keep the American dream alive for everybody, we need to show that the First Amendment, the Constitution, the Second Amendment, all of this infrastructure is actually bigger than any partisan view, no matter how bad it is. And that people, um, will make their own decisions.... and there are a lot of people that can see past the words he uses and focus on the substance of what he's trying to get across, and more generally agree than disagree. And so when you silence that voice, what you're effectively saying is, "This is a rigged game, and all of those things that we told, we were told were not true are actually true."

Lex Fridman 1:19:46 If you were to look at the crude algorithms of Twitter, of course, I don't have any insider knowledge, but I could imagine that they saw the s- s- let's say there's a metric that measures how negative the experience is of the platform, and they probably saw, uh, in s- several ways you could look at this, but the presence of Donald Trump on the platform was consistently increasing how shitty people are feeling, uh, short term and long term. Because they're probably yelling at each other, having worse and worse and worse experience. If you even do a survey of, "How do you feel about using this platform over the last week?" they would say horrible, relative to maybe a year ago when, uh, Donald Trump was not actively tweeting or so on. And so here, you're sitting at Twitter and saying, "Okay. I- I- and- and I know everyone's talking about speech and all that kind of stuff, but I kind of want to build a platform where the users are happy." (laughs) And they're becoming more and more unhappy. How do I solve this happiness problem? Well, let's ban... L- let's- let's, uh, yeah. Let's ban the sources of the unhappiness. Now we can't just say, "You're a source of unhappiness, so we'll ban you." Let's wait until that source says something that we can claim, uh, breaks our rules, like incites violence or s- so on.

Chamath Palihapitiya 1:21:10 That would work if you could measure your construct of happiness properly. The problem is, I think what Twitter looked at were active commenters and got it confused for overall system happiness.

Lex Fridman 1:21:20 Yes.

Chamath Palihapitiya 1:21:20 Because for every piece of content that's created on the internet, of the 100 people that consume it, maybe one or two people comment on it. And so by over-amplifying that signal and assuming that it was the plurality of people, that's where they actually made a huge blunder. Because there was no scientific method, I think, to get to the answer of deplatforming him, and it did expose this idea that it's a bit of a rigged game, and that there are these deep biases, um, that some of these organizations have to opinions that are counter to theirs and to their orthodox view of the world.

Lex Fridman 1:21:59 (sighs) So in general, you lean towards keeping, um, first of all, presidents on the platform, but also, uh, controversial voices.

Chamath Palihapitiya 1:22:10 A- a- all the time. I think it's really important to

Chamath Palihapitiya 1:22:13 keep them there.

Lex Fridman 1:22:13 Let me ask you a trick- tricky one in the recent news that's become especially relevant, uh, for me. What do you think about, if you've been paying attention to Ye, Kanye West, a recent controversial outburst on social media about, um, Jews, Black people, racism in general, slavery, Holocaust, all these topics that he touched on in dif- in different ways on different platforms, but including Twitter. What do you... What- what do you do with that, uh, and like what do you do... What do you do with that at from a platform perspective, and what do you do from a humanity perspective of how to add love to the world?

Chamath Palihapitiya 1:22:59 Let's, um... Should we take both sides of that?

Lex Fridman 1:23:01 Sure.

Chamath Palihapitiya 1:23:02 Option one is he is completely out of line, and option two is he's not, just to simplify.

Lex Fridman 1:23:09 Sure. Right. Yeah.

Chamath Palihapitiya 1:23:10 So the path one is he is an incredibly important tastemaker in the world that defines the belief system for a lot of people, and there just is no room for any form of racism or bias or antisemitism in today's day and age, particularly by people whose words and comments will be amplified around the world. We've already paid a large price for that, and then the expectation of success is some amount of societal decorum that keeps moving the ball forward. The other side would say life, I think, goes from harmony to disharmony to repair, and anybody who has gone through a very complicated divorce will tell you that in that moment, your life is extremely disharmonious, and you are struggling to cope. And because he is famous, we are seeing a person really struggling in a moment that may need help, and we owe it to him, not for what he said, 'cause that stuff isn't excusable, but we owe it to him to help him in a way, and particularly his friends. And if he has real friends, hopefully what they see is that. What I see on the outside looking in is a person that is clearly struggling.

Lex Fridman 1:24:49 Can- Can I ask you, like, a human question? And I know it's outside looking in, but there's several questions I want to, uh, ask. So one is about the pain of going through a divorce and having kids and all that kind of stuff.

Chamath Palihapitiya 1:25:02 Yeah.

Lex Fridman 1:25:03 And two, when you're rich and powerful and famous, I don't know, maybe you can enlighten me to which is the most corruptive, um, but how do you know who are the friends to trust? So a lot of the world is calling Kanye insane and if, or like has mental illness, all that kind of stuff. And so how do you have friends close to you that say...... that say something like that message, but from a place of, uh, love and where they actually care for you as opposed to trying to get you to shut up. The reason I ask all those questions, I think, if you care about the guy, how do you help him?

Chamath Palihapitiya 1:25:47 Right. I've been through a divorce. It's gut-wrenching. Um, the most horrible part is having to tell your kids. I can't even describe to you, um, how proud I am of, and how resilient these three beautiful little creatures were when, when my ex-wife and I had to sit them down and talk through it. Um, and for that thing, I'll be just so protective of them and so proud of them, and, um, it's hard. Now, I don't know that that's what he went through, um, but it doesn't matter. In that moment, there's no fame, there's no money, there's nothing. There's just the raw intimacy of a nuclear family breaking up. In that, there is a death, and it's the death of that idea. And that is extremely, extremely profound in its impact, especially in your children. Um, it is really hard. Really hard.

Lex Fridman 1:26:46 Could you have seen yourself in the way you see the world being clouded during, uh, es- especially at first, to where you would make poor decisions outside of your, outside of that nuclear family? So like bus- poor business decisions-

Chamath Palihapitiya 1:27:02 Yeah. Absolutely.

Lex Fridman 1:27:02 ... poor tweeting decisions-

Chamath Palihapitiya 1:27:03 Absolutely.

Lex Fridman 1:27:04 ... poor, uh-

Chamath Palihapitiya 1:27:04 I think that-

Lex Fridman 1:27:04 ... writing decisions.

Chamath Palihapitiya 1:27:05 ... if I had to boil down a lot of those, what I would say is that there are moments in my life, Lex, where I have felt meaningfully less than, and in those moments, the loop that I would fall into is I would look to cope and be seen by other people. So I would throw away all of the work I was doing around my own internal validation, and I would try to say something or do something that would get the attention of others. And oftentimes, you know, when that loop was, um, was unproductive, it's because those things had really crappy consequences. So, you know, that was, that was, yeah, so yeah, I, I went through that, uh, as well. So I had to go through, you know, this disharmonious phase in my life and then to repair it. You know, I had the benefit of meeting someone and building a relationship, um, block by block, where there was just enormous accountability, where my partner, Nat, had, has, just incredible empathy, um, but accountability. And so she can put herself in my shoes sometimes when I'm a really tough person to be around, but then she doesn't let me off the hook. She can forgive me, but it doesn't make, you know, what I may have said or whatever, you know, uh, excusable. And that's been really healthy for me, and it's helped me repair my relationships, um, be a better parent, you know, be a better friend to my ex-wife who's a beautiful woman, who,