The Exchanges

Q I love a platinum bullet. Thank you for explaining that, by the way. I've learned one thing. Never be afraid to ask the really basic question. Um, you mentioned that, actually, DeepBind and what you did with efficiency within data centers. How important was Willow, Google's announcement on their new quantum chips yesterday?

A Yeah. Um, well, I think it's a significant step forward. Um, you know, look, I still think, I don't know exactly where we will be in scale quantum computing. Um, I think Uh, usually again, of course, the discourse with new technology always starts with the fears, which is, is a mistake that I'm always trying to correct, because, um, it doesn't mean you shouldn't have concerns and anxieties and navigate, but like, for example, you go, oh my god, quantum computing, it's gonna break security, and our banking system is gonna collapse, and our public, you know, encryption and security safety is gonna collapse, and oh my god, there's a disaster. You're like, well, Look, we'll have to figure that out because that'll be an issue once we get to what is, you know, somewhere maybe minimum 2000 or 5000 logical quantum bits, qubits. So that's at 2000 to 5000 or more is when you get into the encryption issues. I'd call it a 150 to 250 logical qubits. So a complete order of magnitude down, quantum computers become great at helping invent New medicines. New drugs. New semiconductors. New physical materials. Things at a micro scale. That's gonna be the first thing from quantum computing. So the moment that we can get there, it'll have enormous prosperity benefits Well before we have to navigate these things. And so, uh, I'm very hopeful about quantum computing. I just don't know the, uh, time f…

AI assessment note: “well, I think it's a significant step forward.”

Q Do you agree with him, and how do you think about quality of model improvement and the trajectory of it?

A I do think, That part of the thing is to make sure, because a lot of people try to build these models and build them poorly. So it isn't just, you know, uh, put in data, right, you know, put in compute, press button, go away for six months, come back, and go, ha ha, I've got GBD-IV. There's, there's, there's a whole bunch of, of art and science, and what I mean by art is like, well, you train it this way first, then you train it that way second, as part of the kind of the evolution of this, and there's a bunch of stuff that isn't written down, In the published papers and so forth about how to do this kind of thing. So quality of model really does matter. And it's a small number of teams and a small number of, of, of folks who are doing it that well so far. It'll broaden. There'll be more. Um, and by the way, size of compute, as per your earlier question, is really going to matter. Um, people frequently mistake what quality of data means because they think, oh, you have to train on the New York Times. And actually, in fact, Um, most of it is quantity of human data, not because you've had any particular journalist who's, who's got a particularly nice way of writing with words that makes anything particularly useful.

AI assessment note: “So quality of model really does matter. And it's a small number of teams”

Q Good, good. Do you have a slide two?

A Exactly. Is there another slide? And, um, and, and there was a fool in the room, but the fool wasn't Elon, the fool was me. And the, the thing that I learned from watching Elon's, you know, rocket success, pun intended, was that if an industry has had decades of, of technological innovation that are not reflected in the market, because the structure of the market Kind of locks in this anti-innovation kind of oligarchy. It's hard because there's something that's locking it in, but if you can break it, it's huge. And so, um, and, you know, uh, Elon is one of those people who, you know, takes the whole concrete wall down with him when he's going through it. So, you know, it's, it's, if someone can do that, it can be really good. And that was an instance of a mistake because I should have said, yes, I will invest in this. And now, of course, I've updated to When I think it's somebody who could take on the structural change of a market because the technology has already been built, then I will consider investing in things.

AI assessment note: “Is there another slide? And, um, and, and there was a fool”

Q If you were to answer that question, how would you answer it?

A So on the first, I have learned so much by just, by the fortuity Of having gone to Stanford and under, like, learned about software entrepreneurship, like, I, I never would have gone and sought out Silicon Valley. It was literally, oops, I'm a student at Stanford and a bunch of my friends have, have, have, have started doing this, and I've started paying attention and said, hey, that's cool, let me go do that. There's a ton. One, um, every scale problem that you think of, 30 to 80% of the solution is technology. So criminal justice, economic justice, uh, medical, climate change. That having an orientation of how do we deploy technology to help that scale problem, because technology changes the possibility curve, changes the cost curve, et cetera. So how do we do that? And Silicon Valley is one of the most intense places in the world of how do we organize corporations and organizations and startups to tackle new technological problems? How do we take risk, allocate initial early capital in the seed, How do we evaluate it? Where we grow it? How do we build networks and teams around it? How do we have an entire kind of entrepreneurial network within Silicon Valley where, like, the coopetition of it is very important, right, because both competing and cooperating, like, you know, the classic kind of theory is, oh, no, no, you should compete, and you shouldn't talk to each other. It…

AI assessment note: “every scale problem that you think of, 30 to 80% of the solution is technology.”

Q Do you think more value accrues to existing incumbent brands given the importance of verification in the new AI world? So like the New York Times becomes more valuable because it is a validated source of truth versus content creator called Sarah on Twitter who is not a validated source of truth.

A I would To some degree, hope so. I mean, one of the things that I think that the kind of the libertarian kickoff of the internet has underplayed is that we need to have shared media be how we learn together. And so some of that is you need to say it isn't just, you know, propositions in the wild that were, you know, us, a thing is said that you can, like, for example, get better from COVID by taking hydrochloroquine, you know, like, like you just, Like, it's like, no, no, no, just because that, that proposition is said and said by the then president of the United States doesn't mean that it has any scientific validity and isn't actually in fact destructive to your health. And so, um, uh, and so I think that the, like, we need to have better sources of how are we collectively learning and what, and what, like for example, when, when we, When we as human beings try to figure out how to get to truth, we use panels of human beings to do it. It's like a scientific panel, you know, kind of observing a study or deciding on a, on a, on a paper. It's a blue ribbon commission and trying to figure out what the facts are in a public circumstance. It's a jury and a criminal trial. We, we bring a group of people together and that brings a set of balanced perspectives. We need to have that kind of thing and we need to have the brands and institutions that reflect that kind of collective appro…

AI assessment note: “I would To some degree, hope so.”

Q Do you think teams can keep up with the progression of the LMS? Because I, I speak to founders often, they're like, God, we almost need to hire someone full time to do an analysis of the different changes in LLM providers just to keep up.

A That's not necessarily a bad thing, but part of the thing is to be choiceful about what your analyses are relative to what's important that you solve and understand that everything is changing at a fast pace. So if you say, for example, like what I've heard from some people who've done analysis, Gemini is better at producing fiction, you know, OpenAI is better at the kind of Wikipedia-like reports, like, okay, like, you know, but by the way, three months from now, it may be different, right? And so it's kind of like, like, that doesn't, Me as a user, that might say if I'm trying to solve a certain kind of thing, like if I'm trying to write a science fiction novel, maybe I'm going to go to Gemini and, and work on that mostly. But by the way, of course, one of the things for products, for individual users, using multiple is good, because that's where I was going with the, is it commoditized? Is like, actually, in fact, I think there will still be differences, but I think our behavior, both in creating products and as individuals, will be almost more like kind of a conductor in an orchestra. It'll be like, well, you know, a little bit, you know, louder on the bass, a little bit softer on the cello, you know, as kind of a, as a, as, as a way of navigating. It won't be that they're all just vanilla, vanilla each other. Now, I do think because we have multiple ones that are competing…

AI assessment note: “part of the thing is to be choiceful about what your analyses are”

Q Do you think acquihires are inevitable for LLM companies? We see the likes of Mistral in France. They just don't have enough cash. It's spade to spade. Are acquihires inevitable?

A I would, in a well-functioning society, um, hopefully it's acquisitions rather than acquihires. Um, part of the thing that Um, there's a, there's a threat of antitrust that misunderstands their own game. So, for example, the threat of antitrust roughly looks like this. It says, we should stop the aggregation of power in the large companies, so we should block acquisitions. And in some cases, limited specific cases, that makes sense. But in general, if you want competition with those large companies from startups, Investors, say for example, I'm an investor who's gonna put a billion dollars into a company that might compete with one of the large tech companies. I'm only gonna put the billion dollars in if I have a chance of an acquisition exit, because I'm gonna need the billion dollars back, possibly. That's a huge loss. So if I am blocked, if you go, if you have a regulatory authority that says, we are never gonna allow that, then I'm never going to do the investment That allows a company to potentially compete with those large tech companies. And, and so their theory, which is we're stopping the aggregation of big power, is they're creating more aggregation of big power because they're stopping the financing of competition.

AI assessment note: “hopefully it's acquisitions rather than acquihires. Um, part of the thing”

Q Do you know what? This is worth it already. Um, can I ask, how do you think about your future involvement in US politics, like just diving right in? You were very active last election cycle.

A So look, I think one of the things that it's a responsibility of Americans to do, including myself of course, is you try to make whatever administration as successful as possible for the people of America, for the country, for the industry. I will be doing that. I will be investing in businesses. Um, I will be Uh, you know, kind of getting new American businesses generated, um, and so I will be doing those things. Obviously, in the next administration, um, they don't want to have me helping them, right? I won't be, uh, involved in any of the activity that's, that, that, that the administration is doing. Um, and actually, you know, obviously, I anticipate, unfortunately, I'll probably be critical of some of the things the administration's doing. I hope to help Do as many of the things that are, like, reduce regulation, increase energy for AI, have, uh, AI companies be prosperous for America and around the world. Are we doing all that kind of stuff? But on the other hand, you know, uh, I suspect that I will have some things that I'm still an opponent of.

AI assessment note: “I won't be, uh, involved in any of the activity that's, that, that, that the administration is doing.”

Q You mentioned that the propensity for older people to vote, um, completely obviously true. What, what Trump did well was harnessing actually TikTok and younger people to vote. I'm intrigued. We mentioned, you know, BYD and Chinese car imports. Should TikTok be banned on the back of this?

A The right way, I think, to think about this from an American or any other perspective is to say China bans Competition in China from Meta, Twitter, etc. And so the right way to say is like, look, if you're banning our companies, we should ban yours. Commercial, entirely a commercial question. Not a question around, you know, um, like, well, that's a Chinese social media company. It's a, cause I actually believe that there's a different question when you say, well, what should be the rules of governance for social media companies when you get around the world? And I think there should be, You know, some transparency, responsiveness to local laws, etc. But, um, but I think that's the, the, the, the right thing to do now.

AI assessment note: “if you're banning our companies, we should ban yours”

Q How do you think about that, though, in the face of, like, you know, bluntly the, the deterioration of the German ecosystem, the no growth for the UK for three consecutive years? As our wonderful labor government have pointed out, it doesn't look good. I'm trying to be, my business is here, my fund is here.

A Look, um, it doesn't mean there aren't challenges, uh, you know, uh, maybe I can say this as an American, you know, Brexit was a terrible idea. You know, and, and I think there are challenges, um, but the, the notion of how we rejuvenate economies, we go, look, how do we build the economies of the future? And so, you can't try to hold on to the economies of the past. You don't want to say, well, I want to be doing coal mining in Newcastle. You're like, no, no. Right? We should be good for those communities and people and try to help them transition, but we need to build the economies of the future. And so, the question is, is, is, I think the UK is well positioned to make that progress and make that effort, but it has to be collectively investing in it, enabling it, bringing in talent.

AI assessment note: “I think the UK is well positioned to make that progress”

Q When we think about like the business model of foundation models, as you said that we've got this real, uh, race to the bottom in terms of pricing. Um, How do we think about when the marginal revenue exceeds the marginal cost, and when they actually become viable businesses, and actually are they viable businesses?

A So I think that there is literally no doubt in my mind that there's viable businesses that will play out, and... Purely in the foundation model. Purely in the foundation model, um, and also the others, but the foundation model as well, in part because, look, at some point, It is the classic venture thing of you start investing, and you may even have bad operating margins. Like, for example, if you said, you know, the first X thousand stays at Airbnb, the operating margins were terrible, right? It was kind of money losing. You have to get to a certain scale. You have to, you have to bring a certain amount into it to then make it into the amazing business that it is.

AI assessment note: “there is literally no doubt in my mind that there's viable businesses that will play out”

Q I love a platinum bullet. Thank you for explaining that, by the way. I've learned one thing. Never be afraid to ask the really basic question. Um, you mentioned that, actually, DeepBind and what you did with efficiency within data centers. How important was Willow, Google's announcement on their new quantum chips yesterday?

A Yeah. Um, well, I think it's a significant step forward. Um, you know, look, I still think, I don't know exactly where we will be in scale quantum computing. Um, I think Uh, usually again, of course, the discourse with new technology always starts with the fears, which is, is a mistake that I'm always trying to correct, because, um, it doesn't mean you shouldn't have concerns and anxieties and navigate, but like, for example, you go, oh my god, quantum computing, it's gonna break security, and our banking system is gonna collapse, and our public, you know, encryption and security safety is gonna collapse, and oh my god, there's a disaster. You're like, well, Look, we'll have to figure that out because that'll be an issue once we get to what is, you know, somewhere maybe minimum 2000 or 5000 logical quantum bits, qubits. So that's at 2000 to 5000 or more is when you get into the encryption issues. I'd call it a 150 to 250 logical qubits. So a complete order of magnitude down, quantum computers become great at helping invent New medicines. New drugs. New semiconductors. New physical materials. Things at a micro scale. That's gonna be the first thing from quantum computing. So the moment that we can get there, it'll have enormous prosperity benefits Well before we have to navigate these things. And so, uh, I'm very hopeful about quantum computing. I just don't know the, uh, time f…

AI assessment note: “I think it's a significant step forward.”

Q Tell me, what have you changed your mind on in the last 12 months?

A Like, a lot of the changes, part of the reason why it's taking me a little bit of time to answer this is a lot of the question changes are micro changes. Like, it's kind of the, you know, um, what is the right way for governments to interface productively with cryptocurrency, right? And there's micro things. I still, I'm a believer in cryptocurrency. I, um, bought Bitcoin back in 2013. I've invested in, in cryptocurrency companies. I think it can be very good for society, but the way that it, you know, the kind of the ways that it can integrate well in society, like there's evolving beliefs there. Like, like what should happen, what should not happen, but they're all like micro things, not macro things. Maybe the most, um, macro thing, it, it's not a change of my belief as much as it's kind of like an open question. Um, I watched, and this is like the real, like I'm going to the totally crazy thing, so I don't, I'm not saying I believe in this, but I watched some congressional testimony from people from the national security, um, establishment. Government labs were studying, uh, material from extraterrestrial spacecraft, and this was congressional testimony, and so I was like, Huh? Like, you know, was it, is it an asteroid, or is it a, is it a robot ship, or what, like, or is there nothing? Is it just craziness? But if you, before I had watched that testimony, if you had said, …

AI assessment note: “before I had watched that testimony... what do I think about UAPs, the answer is conspiracy theory”

Q Did you see the news of, like, we don't actually know where your taxes go?

A Well, that's generally true for very large Financial institution. And by the way, I think it's an important thing to, like, that's the reason why, like you say, bull bear case on, on Doge as well. You go, look, I actually think kind of going through, um, balancing the books, figuring out the economics, applying kind of fiscal discipline in the way that we do in companies is a very good thing. What I think the, is, is a false promise is to say, well, the real important thing is if we lay off X percentage of the federal workforce, and all of a sudden, our federal budget will be so much more trim. It's like, well, no, no, actually, that's not, that's, you just have to do the math, and that's actually not where, where the, where the numbers that you're claiming you'll get to is.

AI assessment note: “Well, that's generally true for very large Financial institution.”

Q Tell me, what have you changed your mind on in the last 12 months?

A Like, a lot of the changes, part of the reason why it's taking me a little bit of time to answer this is a lot of the question changes are micro changes. Like, it's kind of the, you know, um, what is the right way for governments to interface productively with cryptocurrency, right? And there's micro things. I still, I'm a believer in cryptocurrency. I, um, bought Bitcoin back in 2013. I've invested in, in cryptocurrency companies. I think it can be very good for society, but the way that it, you know, the kind of the ways that it can integrate well in society, like there's evolving beliefs there. Like, like what should happen, what should not happen, but they're all like micro things, not macro things. Maybe the most, um, macro thing, it, it's not a change of my belief as much as it's kind of like an open question. Um, I watched, and this is like the real, like I'm going to the totally crazy thing, so I don't, I'm not saying I believe in this, but I watched some congressional testimony from people from the national security, um, establishment. Government labs were studying, uh, material from extraterrestrial spacecraft, and this was congressional testimony, and so I was like, Huh? Like, you know, was it, is it an asteroid, or is it a, is it a robot ship, or what, like, or is there nothing? Is it just craziness? But if you, before I had watched that testimony, if you had said, …

AI assessment note: “before I had watched that testimony... what do I think about UAPs, the answer is conspiracy theory.”

Q Do you think more value accrues to existing incumbent brands given the importance of verification in the new AI world? So like the New York Times becomes more valuable because it is a validated source of truth versus content creator called Sarah on Twitter who is not a validated source of truth.

A I would To some degree, hope so. I mean, one of the things that I think that the kind of the libertarian kickoff of the internet has underplayed is that we need to have shared media be how we learn together. And so some of that is you need to say it isn't just, you know, propositions in the wild that were, you know, us, a thing is said that you can, like, for example, get better from COVID by taking hydrochloroquine, you know, like, like you just, Like, it's like, no, no, no, just because that, that proposition is said and said by the then president of the United States doesn't mean that it has any scientific validity and isn't actually in fact destructive to your health. And so, um, uh, and so I think that the, like, we need to have better sources of how are we collectively learning and what, and what, like for example, when, when we, When we as human beings try to figure out how to get to truth, we use panels of human beings to do it. It's like a scientific panel, you know, kind of observing a study or deciding on a, on a, on a paper. It's a blue ribbon commission and trying to figure out what the facts are in a public circumstance. It's a jury and a criminal trial. We, we bring a group of people together and that brings a set of balanced perspectives. We need to have that kind of thing and we need to have the brands and institutions that reflect that kind of collective appro…

AI assessment note: “I would To some degree, hope so.”

Q How do you think about that, though, in the face of, like, you know, bluntly the, the deterioration of the German ecosystem, the no growth for the UK for three consecutive years? As our wonderful labor government have pointed out, it doesn't look good. I'm trying to be, my business is here, my fund is here.

A Look, um, it doesn't mean there aren't challenges, uh, you know, uh, maybe I can say this as an American, you know, Brexit was a terrible idea. You know, and, and I think there are challenges, um, but the, the notion of how we rejuvenate economies, we go, look, how do we build the economies of the future? And so, you can't try to hold on to the economies of the past. You don't want to say, well, I want to be doing coal mining in Newcastle. You're like, no, no. Right? We should be good for those communities and people and try to help them transition, but we need to build the economies of the future. And so, the question is, is, is, I think the UK is well positioned to make that progress and make that effort, but it has to be collectively investing in it, enabling it, bringing in talent.

AI assessment note: “I think the UK is well positioned to make that progress and make that effort”

Q Do you think teams can keep up with the progression of the LMS? Because I, I speak to founders often, they're like, God, we almost need to hire someone full time to do an analysis of the different changes in LLM providers just to keep up.

A That's not necessarily a bad thing, but part of the thing is to be choiceful about what your analyses are relative to what's important that you solve and understand that everything is changing at a fast pace. So if you say, for example, like what I've heard from some people who've done analysis, Gemini is better at producing fiction, you know, OpenAI is better at the kind of Wikipedia-like reports, like, okay, like, you know, but by the way, three months from now, it may be different, right? And so it's kind of like, like, that doesn't, Me as a user, that might say if I'm trying to solve a certain kind of thing, like if I'm trying to write a science fiction novel, maybe I'm going to go to Gemini and, and work on that mostly. But by the way, of course, one of the things for products, for individual users, using multiple is good, because that's where I was going with the, is it commoditized? Is like, actually, in fact, I think there will still be differences, but I think our behavior, both in creating products and as individuals, will be almost more like kind of a conductor in an orchestra. It'll be like, well, you know, a little bit, you know, louder on the bass, a little bit softer on the cello, you know, as kind of a, as a, as, as a way of navigating. It won't be that they're all just vanilla, vanilla each other. Now, I do think because we have multiple ones that are competing…

AI assessment note: “That's not necessarily a bad thing, but part of the thing is to be choiceful”