The Exchanges

Q Got it. So you're an investor in it, and you donated to it. Actually, let's start there. What's up with that corporate structure? How do we make sense of that? Something's a non-profit, you donated to it, and then you invested in it, and everybody's making money and selling in secondary at a hundred billion. How does that work in the world?

A So, um, so it's, it's a five one C three is the governor thing. That's what started. And, you know, when, you know, kind of Elon and Sam were starting this and said, look, we need, you know, philanthropic support and we're, we're trying to make sure that there's like open access to AI, which is going to be an instrumental technology. And we've got some great technologists who want to come do this. We started as a, as a, you know, kind of as a five or one C three for doing it. That, that, that persists as far as I know till today. Then, you know, one of Sam Altman's, you know, pieces of genius was that he kind of said, look, we're going to need scale capital, and I'm trying to go out to raise, and the commercial round was six hundred million, I'm trying to raise six hundred million philanthropy, and it's not working, right? So, so I have this idea, which is the five one C three, which is doing this kind of research mission of, you know, AGI for humanity, is also producing commercial benefits, and we can create initially an LP, which has a Uh, kind of a, a, a revenue right on the commercial things that investors can invest in. And, you know, um, you know, Reed, it'd be really helpful if you led this, um, because, you know, and I was like, well, but you don't have a go-to-market plan, you don't have a product plan, you don't have a business plan.

AI assessment note: “we can create initially an LP, which has a Uh, kind of a, a, a revenue right”

Q I, I saw you on CNN where they were like, are you telling Kamala and Biden what they have to do?

A And I'm like, no, because I don't believe in that kind of corruption of, of, of politics. The only way she's gonna learn about it is either she asked me or she watched this television show. Um, and, um, and so, so she's done a good job on the price cartel. She did a good job on the anti-competes, both of which I think are very, very good for, you know, Competitive markets. The problem is I think she has a misunderstanding of these, uh, large tech companies. And, um, for example, on the M&A thing, you know, her theory is you gotta prevent the aggregation of power, so you gotta, you gotta fight every acquisition of note. And the problem, of course, is that actually quells venture capital investment. It's just like, okay, um, part of the, the returns is if I'm gonna invest in something that might be competing with, you know, One or more of the large tech companies. I need to have acquisition exits as part of being able to fund enough capital to really make that acquisition, you know, that, that, that, that investment possible work, because if it doesn't work, I want to be able to at least return, recover my capital by an investment. So the right way to look at it is, is there competition amongst the top tech companies? Because, you know, if one of them is like squashing all the other ones, that's a problem. If we're, if we're Five large tech companies heading to three, then I'm mu…

AI assessment note: “And I'm like, no, because I don't believe in that kind of corruption”

Q Do you, what did you think of Elon's first lawsuit and then he dropped it and then he refiled? Where, where do you think he's coming from?

A Well, um, what, what is the, uh, you know, I, I'm not very charitable about those lawsuits. Um, the, you know, I would like to be because, you know, Elon's one of the entrepreneurial heroes of our time and generation, but I think it's the, I think it's, you know, frankly, I think probably the most charitable thing to say is sour grapes, um, because, you know, for example, I know Sam offered him as much of the investment round as he wanted. Right? Like, he could have done the whole thing. He could have let it. You know, it was kind of like, hey, look, we still love you. And he was like, no, it's not a company that I control. It's gonna fail. So I'm not interested in investing. And I was like, okay.

AI assessment note: “I think probably the most charitable thing to say is sour grapes”

Q ago, you said something to the effect very publicly that you had had a one hour or multi-hour lunch with Biden, and he just seemed like super on his game, and then he was kind of dumped. Was that just a moment in time where he was really great with you? Or how do you reconcile that with Pelosi and all of these other folks and what happened to Biden?

A Well, like most of us, um, I was pretty dismayed by the debate performance, um, because when I talked to him, like, detailed, thoughtful analysis with no notes on Gaza, questions about AI, you know, and what kinds of things, you know, what did I think about what the progress, you know, the thing they were doing with the voluntary commitments, the executive order, and, you know, what kinds of things should happen in the future, and all that kind of stuff being on the game. A little slower, Right, then, then, you know, a, a fifty-year-old would be, but, but, you know, like, cogent and totally worth it. And then you kind of, uh, looked at the debate. Oh my gosh, this is, this is, this is a disaster. Um, and so, um, it was like, look, is the debate a one-off thing? Is it, did you, were you ill? You know, like trying to reconcile the two and, you know, spend a little bit of time trying to figure that out, uh, to, to what was going on because it was, it was the first time I'd seen something like that. And, you know, um, you know, I don't, you know, I'm not enough of a DC insider to know exactly what the set of conclusions were, other than I, you know, I plotted, you know, Biden for having the kind of integrity to go, look, I'm, maybe I'm ill, maybe I'm old, maybe I'm slower, but, you know, the, it's about the country more than it's about me, because I'm, you know, not, um, you know, …

AI assessment note: “Pelosi couldn't force it. Anyone else. It's ultimately his decision.”

Q Or language translation, right? If I just want to do language translation, I don't need a massive model. I just need a model that's really good at language translation.

A Exactly. And so what I think you're going to see is, is, is networks of models and like kind of traffic control and escalation and all the rest. And agents are not going to be one model. They're going to be blends of models. And that's one of the reasons why you say, well, there's actually, in fact, a lot of, a lot of room for startups, because it's not like we say, well, we, we take GPT- seven. And we just serve it for everything. It's like, well, it's gonna be super expensive. And there's a whole bunch of things about like serving it more cheaply. And like, for example, one of the, the, the really great technical papers that I love from Microsoft is, you know, all you need is textbooks. It's like, you can, you can train very specific models on kind of like high quality data, um, along with, by the way, the larger model helping train it, that all of a sudden you have a functional smaller model. And you know, the, the question will be a blend of these things. So I think the, the, The, the multi-model, uh, model approach is, I think, gonna be, you know, quickly universal.

AI assessment note: “Exactly. And so what I think you're going to see is networks of models”

Q Meta is obviously far behind OpenAI, far behind Google, Microsoft, so they went open source. When you're behind, you go open source, I guess is the idea here. Um, but they're making some big progress. Who do you think is going to win this ultimately? An open source, uh, provider of LLMs or proprietary closed source like OpenAI is? And it's confounding to say OpenAI is closed, but closed AI.

A Yeah, um, look, from the very founding of OpenAI, I was never claiming it was gonna be open source. I was claiming it was gonna be once safety, open access, um, and not differential or controlling access, uh, for that, and I think that's, they stayed true to that principle, which is, I think, what the genesis of the word open is there, and, um, look, I think the key thing is there's gonna be winners all over the place. I think there's gonna be winners in the open source side, um, and, you know, I don't I don't know if Llama's gonna win from its open source thing as much as it's just trying to say, hey, we're training these models, so we're gonna, you know, uh, put them out there because our closed system, closed loop, you know, doesn't require selling for tokens and so forth. But there's also, you know, Mistral and other folks who are doing competent models. Um, and then I think that the, uh, the, but, you know, there'll be wins in different ways. So it's not like, I think, like, for example, you know, I think there's gonna be a bunch of different startups they're gonna win, whether it's coding agents or, You know, kind of very specific applications within medical or other kinds of things, and I think they will, you know, generate big companies, and I think large companies like, you know, the hyperscalers are gonna, are gonna succeed as well. Now, in the pure model competition,…

AI assessment note: “there's gonna be winners all over the place. I think there's gonna be winners in”

Q Reid, give that welcome to the program and give us a little story. What is your fondest memory or the most quirky memory? David Sachs and that all those weirdos, I'm sorry, not supposed to use the word weird anymore. I get banned on X. All of those unique personalities at PayPal. Tell us about that moment in time, and do you remember the first time you met David Sack?

A Yeah, I met, um, David, uh, because Peter had known him from Stanford and hired him in, um, and, you know, David, uh, very quickly, because he, you know, uh, has a, he has a strong learning curve as he plays these things, kind of got the instinct of what the game we were playing with PayPal was, and it's part of the reason why I think, you know, each of the execs have had, you know, kind of key contributions to making, um, You know, kind of PayPal successful, and, and David's was this, kind of, like, um, maniacal focus on the, kind of, the, the cycle of how the product worked on eBay. Um, and, like, like, there was just a whole bunch of stuff I learned from him. It's part of how I track, you know, kind of. You know, people I respect is what do I learn from them? And that was one of the things that I would say I learned from, uh, David at PayPal.

AI assessment note: “I met, um, David, uh, because Peter had known him from Stanford and hired him”

Q I think as a candidate you're allowed to do that, and those same groups that once fought To keep him off the ballot or trying to keep his name on the ballot. Cause now they perceive. Yeah. Because now they perceive the political calculation to be a little different. So Mike, so I don't see any of this as being democratic. This to me is just partisan hackery, isn't it?

A So look, yes. Fundamentally from a viewpoint of like, look, for example, my, my direct actions and yes, there was, you fund a whole bunch of different groups and you have different groups doing different things, but you funded them to do this thing that you were, Thinking of, and things happen. It's just like companies. You know, my thing was actually, in fact, making people aware of, of RFK's anti-vax, you know, statement, his anti-science stuff, because I thought that would be relevant in the polls in November. That, that was the, the actual strategy that, that I believe, and I think that would differentially, you know, hit Trump more, and so therefore would be a spoiler Right, as these independents are.

AI assessment note: “So look, yes.”

Q Do you think that they should have run an open primary after that? And would Kamala have won an open primary?

A Well, it's hard to know. I mean, I think they were definitely leaning towards an open primary, and then all the people who would be the most natural contenders all endorsed Kamala. So, um, and by the way, you say, well, kind of democratic process. It was like, well, there was a democratic process that picked the Biden-Harris ticket, which turned into the Harris-Waltz ticket, and so that's not anti-democratic, but I think if you look at the sequence of events, it was kind of like, well, you know, we're, we're gonna sort out, you know, what, what we're gonna do, and then, you know, all of the key folks, you know, Shapiro and, and Whitmer and everyone else all endorsed Kamala. I was like, okay, let's just, let's get back to, you know, kind of the, the choice of two candidates. And so I, you know.

AI assessment note: “Well, it's hard to know. I mean, I think they were definitely leaning”

Q There's three. There's Alvin Bragg was convicted, uh, and then the, um, Trump organization with the CFO, uh, committing tax fraud. He was convicted in that one as well. Or the Trump organization was convicted. And people say that's lawfare by Letitia James. So guilty, guilty, guilty in those three of five. So what's your take on the four that we haven't discussed yet and heard your opinion on Hoffman?

A So, look, I think it's, you know, it's definitely possible to have some versions of lawfare, although I think most people use the term when it's the legal process and the law enforcement that they don't like. Um, you know, I think that in the Bragg case you had, you know, um, you know, indictment and 12 jurors. There's jurors, I think, as I recall, one of the jurors said he got, uh, that juror got their principal news from Truth Social. It was a unanimous conviction. Um, I think that, you know, you know, you have Vice President Pence, you know, comes out and says, you know, Trump asked me to overturn the election illegally, right? That's your own vice president. So I don't think that that kind of suggests that there's this rampant political persecution, that there's a lot of fire where there's all this smoke. Doesn't mean that every single thing, uh, you know, kind of.

AI assessment note: “I don't think that that kind of suggests that there's this rampant political persecution”

Q I guess, yeah, all cloud computing platform companies, how are, yeah, what's your sense on how they're thinking about it?

A It's just one of the principles in the Microsoft thing is the company speaks for itself, board members don't, don't, don't speak for them. Um, but, you know, uh, I think Satya is like the best public market CEO of our, uh, generation. Um, I think he is stunning, um, in kind of blending combination of strategic insight with also kind of being, you know, kind of return on capital, you know, sensible risk-taking, etc. And so the actual thing between your guys' questions in terms of, because I can comment on how Satya thinks with this stuff, is he's both thinking about, like, it's a platform change, and you have to be there for the platform change, for productivity, for cloud, etc. And Ok, let's rationalize the capital to when are we expecting revenue? How do we get revenue sooner to, to have that as a good productive cycle? How do we, um, you know, be not trying to, you know, just, uh, spend like drunken sailors, which is easy to do, right? But, but to, to be targeting, you know, kind of business outcomes. And it's part of the reason why, you know, they're like, like, you know, he's very focused on what are we doing with office? What are we doing? Cloud. What are we doing with, you know, as opposed to, like, you, you, you rarely hear him talking about AGI or never digital gods, because it's kind of the question of, I am, I am focused on this in a business sense, and I think that's…

AI assessment note: “I can comment on how Satya thinks with this stuff, is he's both thinking about”

Q companies have too much power or agrees With Lena Khan on that. And I guess specifically, do you think any big tech companies should be broken up? If so, which ones? I mean, I, I would actually entertain that idea of, of, of deconglomerating or breaking up some of these big tech companies. Do you, do you think, do you agree that big tech is too much power or not?

A Um, I think it's TBD, um, but the reason I'd say, look, take the opposite point of view and say no, uh, the thesis for no is that they are very strong American companies that get, you know, in the most cases over half the revenue from overseas. They create, uh, technology platforms that beneficially differentiate the US, um, versus, you know, many other, uh, countries in kind of global circumstances like the internet and other kinds of things. Um, I think that they are competing ferociously with each other. I mean, uh, you know, Jason just mentioned that, you know, it's kind of like, look, we've already got, like, ChatGPT competing with Google Search and other kinds of things, and I think it's, Competitive pressure, right? This is, I think, what capitalism's about is competitive pressure that essentially creates the thing, and that's the reason why, like, if it's, if, if we were shrinking, like it was Google, Uber Alice, or I actually, frankly, think that the, you know, everyone likes to talk about, you know, Google, you're like, I think that the prime candidate is likely to be, and I'm speaking as an individual and as a venture capitalist here, is Apple with the App Store, right?

AI assessment note: “the prime candidate is likely to be... is Apple with the App Store”

Q I get that. And then I get the answer, and I don't need my New York Times subscription. I don't visit the New York Times anymore. Um, feels pretty clear cut to me, but how do you think about IP? Should, uh, an LLM be able to ingest whatever they please, or should they be required to get permission in advance and pay a royalty to content creators? Reid Hoffman.

A Well, as a content creator too. Look, I think, I think that it tends to be a little bit of a We, we do want content creators to benefit economically from the work. It's part of the reason why we have copyright. It's part of the reason why we have payrolls, you know, other kinds of things that I think are, are very important. And I think it's, it's a complicated thing that needs to be sorted out. Now, that being said, I think we also want to say that we can train these models, like, you know, like training is like reading and like reading things is, you know, like when something's available to be read and you've engaged in the right economic thing for reading it, I think that's a, Kind of a reasonable fair use thing. Now maybe we update the terms of service. Maybe we update, you know, copyright law and other things to say, okay, that now changes. You know, I think we don't want to forbid changes in the future. You know, um, this is like one of the problems we get with it blocks innovation when we do that. It blocks innovation and, you know, kind of Hollywood. It blocks innovation in music. It blocks innovation. So you want to allow some new chain, you know, changing landscape. And I think this is a changing landscape that arguably is reading. So I think that, Like, both of those things are true in terms of what do we, what do we want to sort through? I think that one of the reas…

AI assessment note: “don't try to hold out for money on the training side of things”

Q recently with RFK Jr. and the lawsuits against him in being on the ballot in different states, it makes it very difficult maybe for the people to have their choice. And is that a bad way for democracy to work? And I just love your philosophical view on this. I'm like, what's the best way for democracy in the United States to work? So, for the president, for the president.

A Yes, we do live in a republic. All right. And there, there is various, like, you know, some people have much more influence than others, um, whether it's media platforms, whether it's, um, you know, economics and ability to spend, whether it's, you know, a history and a brand and, and, and other things. And so, you know, this melee and, and, and kind of, you know, whole integration set of things. Now, ultimately, you know, Voters are gonna decide in November, right? So, you know, people do have a, you know, and I think that, that, that staying to our democratic process is what's really key, like, you know, people going to the polls, you know, I think we should want to live in a country where everyone does, you know, everyone who is, who is legally allowed to vote does vote, um, and I think that that's, you know, ultimately a good thing. Now, you know, are there things that I would like to change? Sure, I'd like to change I'd like to have ranked choice voting. You know, I'd like to have open primaries. There's, there's a set of things like, like actually my principal frustration in all of this stuff is, you know, what's, what's one of the fundamental things that the two parties agree on that, that, that, that shouldn't be is that there should be only two parties.

AI assessment note: “I'd like to have ranked choice voting. You know, I'd like to have open primaries.”

Q some have argued that the anti-Semitism arises from these principles and that the Jews are considered a privileged and powerful, uh, cultural class. Is that, is that not being observed? Do you not, do you not think that there's some tendencies that are emerging in the Democratic Party and maybe influenced by a louder far left and that far left is becoming more loud and better represented in the party?

A Look, I think we should speak out against both the far left and the far right. Um, I think it's important to do both. And so, you know, since, you know, I'm, I'm, I'm playing the Democrat here on this conversation, I'll ask you guys to play the, the, the, or especially Sachs, play the Republican and, and speak out against the, the far right too. Um, but the, the, the short answer is yes, there are, amongst the extreme left, that's not everybody in the Democratic Party, but the extreme left, there is, um, some, like, you know, misunderstandings about, you know, why it's important to do, defend, you know, kind of anti-genocide, like, from the river to the sea. It's like, yeah, that's a genocidal statement. Don't use that one, right? Understand what language you're using. And, and to be like, look the, you know, we've had a great genocidal moment with, you know, World War II, and we're still trying to recover from it. Um, two, you know, questions around like, like what I think is a, a, a, a foolish wealth tax, even though it's, by the way, narrow to like 80%. And then on like the, the price gouging stuff, um, you know, one of the things is I started scratching at it, you know, it was interesting. I think this week Kroger said, yes, we did actually artificially raise prices. To profit from the pandemic, you know, and yeah, you should stop price gouging. It's not quite the same thin…

AI assessment note: “the short answer is yes, there are, amongst the extreme left”

Q Reid, were you surprised that then the first target where there was like some successful antitrust pushback was against Google versus Apple? And then second, do you think that there's a chance, like a meaningful chance that the government tries to break Google up? Or do you think it looks something maybe more similar to what happened to Microsoft?

A So I think in mandating breakups, you know, like I think is a, um, look, I think we should operate through competitive networks and competitive ecosystems. I think it's part of what's smart about capitalism. And I think mandating breakups is only when essentially capitalism is failing, uh, on specific things. You want to do the least, the least you can to get back to competitive networks in terms of how you're operating. And so, you know, you say, hey, look, I, iOS has this kind of monopoly, and you say there's no side loading, you have to use the app store, you have to use the payment mechanism, you know, et cetera, et cetera. It's like, well, that quells a ton of startup innovation. We all know this as investors, because we look at anyone who's prospectively doing a business like this and say, no chance. It's, you know, you're, you're not going to succeed. And so, so then you say, well, what's the least thing that we can do, right? And, you know, a classic anti-trust, like, well, let's break off the app store from Apple as well.

AI assessment note: “mandating breakups is only when essentially capitalism is failing”