The Exchanges

Q Given that, like, we shouldn't underestimate it, Do you think it's right that we have export controls on chips? You know, obviously Nvidia has had a lot of export controls back and forth. Do you think that's right or not?

A I think the export control on chips has largely backfired. Um, the way that the US and, and I think that the way the US first put restrictions on Huawei, uh, uh, and then later on, you know, exported NVIDIA and AMD and, and other semiconductors that really incentivized China. So before the export controls, semiconductor development in China, it was not, frankly, it wasn't moving that fast. You know, it was a nice area. There was some investment, But when America did that, then China really accelerated its semiconductor development, and so America incentivized China to do this, and it is paying off for China. Um, I think, uh, uh, your number of Chinese companies are building offerings that, um, the individual chips are less powerful, but maybe a much larger number of chips trying to build offerings competitive with certainly the last generation of Nvidia. Maybe increasingly the current generation. So I think, um, if I were to analyze just purely, you know, U.S. national self-interest, I think that caused China to accelerate a semiconductor industry in a way that may not be helpful to the U.S. long term.

AI assessment note: “I think the export control on chips has largely backfired.”

Q think about the question of, you mentioned earlier, brilliantly, that articulation of kind of horizontal and then the verticals beneath them, and Google and now OpenAI being the horizontal. How do you think about the question of a world of large monolithic models versus much smaller, much more efficient, much more specialized models? How do you think about that? And has your mindset changed around which will be more dominant?

A I think it's good. It'll be all of the above. We will have large models and mid-sized models and tiny small models. And the reason I'm confident about that is because, um, the nature of intelligence is diverse, right? Sometimes we do intellectually really easy tasks. Like if someone asked me, I don't know, when, when, like, all right, yesterday my daughter, um, well, she misspelled the word butterfly. So I need to tell her how to spell butterfly. It's a low, you know, it's an easy intellectual task. And sometimes I'm sitting down thinking for hours about some, you know, complex, like technical problem, right? And that's, that's really hard. And so intelligence has a range of things we want to do. And so the set of things we want AI to do too, has a huge range. If you want AI to do basic grammar checking and spell checking, you don't need a trillion parameter model. Use a tiny model, maybe running locally, just do that. But if you wanted to do complex reasoning, to write a piece of code, then yes, having a Powerful model is going to do better. And so, um, I'm actually very confident we'll end up with a huge range of models, small and large to do the huge range of tasks. Just, just like we have humans do a range of tasks of difficulty. Same with AI.

AI assessment note: “It'll be all of the above. We will have large models and mid-sized models”

Q What else does everyone think they know about AI and its adoption and implementation that they get wrong?

A Even earlier this year, we saw some senior business leaders advise people to not learn to code on the grounds that AI will automate it. We'll look back on that as some of the worst career advice ever given. As coding becomes easier with AI assisting us, a lot more people should learn to code, not fewer. And I'm already seeing, I mentioned the mock example just now with building an app for feedback, swiping. But I think, um, for a lot of job functions, people that know how to tell a computer exactly what they wanted to do, so the computer can do it for you, they'll just be more powerful. And for the foreseeable future, the language of precisely telling computers what you wanted to do is coding. It doesn't mean you should write code by hand. Writing code by hand is becoming obsolete, right? Oh, really don't do that. Not, but they get AI to write code for you. And people can do that would be more effective and more powerful and have more fun.

AI assessment note: “advise people to not learn to code on the grounds that AI will automate it”

Q Uh, and I think- Why do you think China is wanting an open AI world?

A It turns out that openness is great for a country's, um, development. So it turns out that, uh, when a team releases open source software, circulation of knowledge is much faster to the close by community. And so what I see is when, when a team in China releases an open way model, then yes, of course, American can take advantage of it, but the China economy benefits even more from it because Once something is open, it's easier for teams to call each other and say, hey buddy, how does this really work? I'm having trouble with this model. It's just that circulation of knowledge is really valuable for innovation, and when the US, um, has more closed models, and when, you know, teams are trying to pay these hundred million dollar salaries to extract talent, then that circulation of knowledge becomes very slow, and it slows down the rate of American and European innovation.

AI assessment note: “China economy benefits even more from it because Once something is open”

Q David Kahn from Sequoia said, hey, a really useful barometer for effectiveness is, can AI replace the bottom five percent of capabilities of what workforce does? Joel from Cohere said, no, that's crap. The real question is, can it 10 x people's ability? Forget the bottom five percent, can it 10 x? How do you think about a barometer for success of the workforce with AI? With those in mind.

A In the case of software engineering, it is accelerating the writing of code. Uh, there are so many projects that used to take, you know, six engineers, half a year to build, that today I or one of my engineers can build in the weekend. I hope that we never have to go back to coding with our AI systems again, because the acceleration, the productivity boost is, is, is incredible. I mean, for example, one weekend I thought, oh, I want to You know, uh, my daughter, I wanted flashcards for my daughter to practice multiplication. She wanted to practice multiplication. She wanted flashcards. So I thought I could either drive to the store and buy a bunch of flashcards for her, or I could just, you know, use AI to write code for me, to generate and print out a bunch of flashcards. And so I did the latter. And so this is a very low economic value task. But with AI-sensitive coding, I could get that done, uh, uh, very quickly.

AI assessment note: “In the case of software engineering, it is accelerating the writing of code.”

Q Given that, like, we shouldn't underestimate it, Do you think it's right that we have export controls on chips? You know, obviously Nvidia has had a lot of export controls back and forth. Do you think that's right or not?

A I think the export control on chips has largely backfired. Um, the way that the US and, and I think that the way the US first put restrictions on Huawei, uh, uh, and then later on, you know, exported NVIDIA and AMD and, and other semiconductors that really incentivized China. So before the export controls, semiconductor development in China, it was not, frankly, it wasn't moving that fast. You know, it was a nice area. There was some investment, But when America did that, then China really accelerated its semiconductor development, and so America incentivized China to do this, and it is paying off for China. Um, I think, uh, uh, your number of Chinese companies are building offerings that, um, the individual chips are less powerful, but maybe a much larger number of chips trying to build offerings competitive with certainly the last generation of Nvidia. Maybe increasingly the current generation. So I think, um, if I were to analyze just purely, you know, U.S. national self-interest, I think that caused China to accelerate a semiconductor industry in a way that may not be helpful to the U.S. long term.

AI assessment note: “I think the export control on chips has largely backfired.”

Q Do you get annoyed by the bubble discussion?

A I don't get annoyed by the bubble discussion. Um, I do get annoyed by the hype. I don't know, when regulators are calling me up and saying, hey, we heard AI could lead to human extinction. Thankfully, much less of that now than a couple years ago. Then kind of like, you know, and then instead the conversation should be, how can we upscale the workforce? Where can we invest? You know, not like, how do we slow this thing down? I think the hype has really distorted public perception of AI. Oh, and, and, and one downside to the hype too is, um, without public support of AI, things slow down. So for example, actually, uh, one of my friends, uh, friends works a lot with high school students, and he told me that he was talking to a girl, uh, a high school student that was, uh, that he was talking to her about maybe pursuing a Korean AI, and she said, you know what? I heard AI could have something to do with human extinction. I don't want to have anything to do with that. And so this hype turned a high school girl away from working on AI at a time where it'd be so promising for them to leap into AI. And I think this really causes people to make weird decisions, both at the individual school student level, as well as at the community level, where, um, when the community, you know, shuts down building all the data center, they could be good for the community and good for the world. I thi…

AI assessment note: “I don't get annoyed by the bubble discussion.”

Q think about the question of, you mentioned earlier, brilliantly, that articulation of kind of horizontal and then the verticals beneath them, and Google and now OpenAI being the horizontal. How do you think about the question of a world of large monolithic models versus much smaller, much more efficient, much more specialized models? How do you think about that? And has your mindset changed around which will be more dominant?

A I think it's good. It'll be all of the above. We will have large models and mid-sized models and tiny small models. And the reason I'm confident about that is because, um, the nature of intelligence is diverse, right? Sometimes we do intellectually really easy tasks. Like if someone asked me, I don't know, when, when, like, all right, yesterday my daughter, um, well, she misspelled the word butterfly. So I need to tell her how to spell butterfly. It's a low, you know, it's an easy intellectual task. And sometimes I'm sitting down thinking for hours about some, you know, complex, like technical problem, right? And that's, that's really hard. And so intelligence has a range of things we want to do. And so the set of things we want AI to do too, has a huge range. If you want AI to do basic grammar checking and spell checking, you don't need a trillion parameter model. Use a tiny model, maybe running locally, just do that. But if you wanted to do complex reasoning, to write a piece of code, then yes, having a Powerful model is going to do better. And so, um, I'm actually very confident we'll end up with a huge range of models, small and large to do the huge range of tasks. Just, just like we have humans do a range of tasks of difficulty. Same with AI.

AI assessment note: “It'll be all of the above. We will have large models and mid-sized models”

Q What else does everyone think they know about AI and its adoption and implementation that they get wrong?

A Even earlier this year, we saw some senior business leaders advise people to not learn to code on the grounds that AI will automate it. We'll look back on that as some of the worst career advice ever given. As coding becomes easier with AI assisting us, a lot more people should learn to code, not fewer. And I'm already seeing, I mentioned the mock example just now with building an app for feedback, swiping. But I think, um, for a lot of job functions, people that know how to tell a computer exactly what they wanted to do, so the computer can do it for you, they'll just be more powerful. And for the foreseeable future, the language of precisely telling computers what you wanted to do is coding. It doesn't mean you should write code by hand. Writing code by hand is becoming obsolete, right? Oh, really don't do that. Not, but they get AI to write code for you. And people can do that would be more effective and more powerful and have more fun.

AI assessment note: “we saw some senior business leaders advise people to not learn to code”

Q I had Joel Pinot from Cohere and formerly of Facebook on the show recently, and she said that AI coding assistants are in the same place that maybe image generation was in 2016, 2017 in terms of maturity. Do you think that's a fair state of the environment today, or do you not think so?

A I don't know. I, I think it's further along. I think in 2016 image generation wasn't super valuable. Uh, I don't remember it being that valuable back then, but I, I, I think today AI coding assistance is really, actually at AI fund, my, uh, head of, uh, engineering recently, I would say, hey, let's think about standardizing on tools. And, you know, basically he said, you know, I need these tools and you have to pry them out of my code dead hands. Right. I think, I think our developers feel really strongly. I, I myself, I don't ever want to have to code again without AI coding assistance. So I think the tools are really working well, but still with a lot of, uh, head room for how much better it can get.

AI assessment note: “I think it's further along. I think in 2016 image generation wasn't super valuable”

Q Do you think vibe coding is an enduring market? Like, do you think everyone will want to code, and accessibility is important, or do you think it bluntly just allows builders to build better and more efficiently?

A I think we need all of the above. Um, you know, I've had mixed feelings about the term vibe coding, but nitpicking terminology aside, I think everyone should learn to code. Uh, what I'm seeing is for a lot of job roles that aren't just software engineering, people that can code, Can get more done than people that can't code. For example, I think, um, my, my, um, my marketer wanted to run a user survey once, and, uh, she wanted, you know, something for people to give live feedback, and she looked at the app store, couldn't find anything, so she said, you know what, I'm gonna spend two days to code up. It did take two days, but my, uh, marketer then built a little mobile app where users could swipe, you know, left or right, To give feedback on some marketing messages. We want to use a test, and because of that, we're able to run user experiments, get feedback, and so hope her do her job better as a marketer. Whereas in contrast, a marketer that could include a little app to, you know, let people swipe around and get feedback. They would just not have been able to do this, would not have gotten the feedback, would not have been able to move forward. Today, my best recruiters, Not only do they screen resumes by hand, they are writing prompts to get AI to help them screen resumes. Um, and it's been interesting.

AI assessment note: “I think we need all of the above. Um, you know, I've had mixed feelings”

Q democratization of knowledge there and the benefits that come from it. You said a word before, which was open, about the kind of the open weights ecosystem we've seen. We've seen this reversion back to like a closed world in a lot of cases. How do you feel about the reversion back to a closed? And how do you analyze the state of play today in that open versus closed?

A It's still very dynamic. Um, I think it is, uh, so for a lot of American companies, the leading frontier model is often kept close, and then the one tiered down model, not quite as good as release as open. I think it's much better than nothing. I'm actually grateful for all the teams that are releasing open source, open weight models. And then the other dynamic is, uh, uh, China especially has been really taking the lead, uh, or, well, Taking a lead or getting up there in terms of releasing tons of really good, um, open way models. So I would say if not, it is kind of, um, not whatever predicted, you know, a decade ago that China AI would end up being more open than America AI.

AI assessment note: “I'm actually grateful for all the teams that are releasing open source, open weight models.”

Q Uh, and I think- Why do you think China is wanting an open AI world?

A It turns out that openness is great for a country's, um, development. So it turns out that, uh, when a team releases open source software, circulation of knowledge is much faster to the close by community. And so what I see is when, when a team in China releases an open way model, then yes, of course, American can take advantage of it, but the China economy benefits even more from it because Once something is open, it's easier for teams to call each other and say, hey buddy, how does this really work? I'm having trouble with this model. It's just that circulation of knowledge is really valuable for innovation, and when the US, um, has more closed models, and when, you know, teams are trying to pay these hundred million dollar salaries to extract talent, then that circulation of knowledge becomes very slow, and it slows down the rate of American and European innovation.

AI assessment note: “when a team in China releases an open way model... China economy benefits even more”

Q You have the most fascinating perspective having obviously spent many years at Google and then obviously Baidu as well, and so having been on both sides of the table in certain respects, we have this kind of strange binary polarization of The AI race, China versus the US. Do you agree with that positioning of China versus the US in an AI race?

A I think there's a lot of room for, um, Corporation, and then also some places that will be competitive. So first, um, while people sometimes, even me, talk about the AI race, there's no single finish line. It's not one race. It's AI is a general purpose technology, and you could be better aware of coding, better aware of answering questions, better aware of helping with, you know, molecules and finance and so on. So AI has many different capabilities, and there's no one finish line. And even though it's one capability, I think we're going to keep on improving for a long time. So I, I feel like because of PR goals, AGI has been hyped up as a finish line, but I don't think it's a finish line. It's just, we'll have continually improving capabilities for, you know, decades to come. Having said that, nations with stronger AI capabilities are going to be more powerful. Uh, the citizens will be more prosperous. The economies will grow faster. Uh, so I think there is, so to the extent that different nations' incentives are not aligned, nations with more powerful AI capabilities will be able to do more. Just like, you know, if, if a country has a fantastic electricity grid and another country, you know, has power outages and so on, well, one country can just use the electricity grid to do more manufacturing, more industrial work, just do a lot more that way.

AI assessment note: “talk about the AI race, there's no single finish line. It's not one race.”

Q When we think about a core of a business, it's margins, and most of these business don't have margins. Do you care about margins when investing today, or with absolute respect, and it sounds disrespectful, do you take the kind of utopian view that it will just correct itself with time and with efficiency gains?

A At some point, uh, the laws of physics, I think, or the laws of finance or something, margins do matter. But one of the tricky things about AI is, um, we know the technology is going to change. So we don't build assuming the technology will be stagnant. We do build assuming the technology will evolve. So one, one obvious one, um, um, token prices have been rapidly falling, right? Uh, depending on who you believe, falling a percent year or whatever. Kind of, frankly, when we build prototype, We routinely just not worry about token costs because the first most important thing is this build a product that users love. Um, and then what we find is after we build some, this actually happened to me a few times now. We'll build something and not worry about the cost. And then, you know, users start to use it. And then our API build starts climbing. And then it is really like kind of, you look at this every few weeks and go, whoa, this getting really expensive. This costing me. Saturday of one engineer cost me more than two engineers cost me more than a whole bunch of engineers. All right. But fortunately, when that has happened, um, almost every time so far, we've been able to use techniques to bend the cost curve back down even faster than the rate at which token prices are falling in the market. And so I find that, uh, absolute margins are important, but when you have a view for wher…

AI assessment note: “we don't take a blind utopian, you know, AGI blah blah blah view either.”

Q How do you think about defensibility in the AI world? A lot of people suggest that the time to copy is reduced significantly, um, that defensibility itself is questioned in AI. Do you agree with that and the questioning of defensibility today or not?

A Most are changing. Um, so I find that most tend to be a function of the industry rather than the function of the technology. So AI as a technology Doesn't really offer an answer to the moats for most businesses. So if you're building AI, you know, for drones or legal or for whatever, the moat is more of a function of that industry. Um, but one thing that is changing with regard to moats is previously software used to be a moat, right? If you had, you know, invested 10 years to build a software, it's really hard to replicate that. That one moat is much weaker than before, but Other modes, like, are you trying to use AI to accelerate, to build a two-sided marketplace, which can be very defensible, or, you know, are you building for a consumer, more for consumer and enterprise, are there on brand and reputational effects, right, that can help you build defensibility there. So I find that, um, the software mode has changed, but other modes tend to be analysis based on the industry.

AI assessment note: “moats tend to be a function of the industry rather than the function of the technology”

Q You said economic to do so. Do you think it's crucial that we see vertical ownership in terms of see Nvidia own models as well as chip player? And we're seeing Facebook build out data centers more than anyone. We're seeing everyone build out data centers. Is it important that we own every layer of the stack? Or actually will we see individual participants own horizontal layers of the stack?

A I think this will evolve over time. Um, I'm going to make an analogy. In the early days of, um, say the computing industry, it was the vertical players that won. Because, you know, if you want to connect the keyboard to your computer motherboard, which is a CPU, is it okay if your keyboard has a, you know, plus minus five volts and your CPU has some other voltage? Is that okay or not? So, We didn't know where the API boundaries or, um, if your, uh, CPU has, uh, you know, memory laid out a certain way, a compute and your, you know, math accelerator, they needed to interoperate with each other. So before we wound up having a clear conception of where the draw lines and whether the API boundaries, the integrated players, IBM back in the day, could solve all the problems and, and, you know, build, build valuable working But as the industry mature, we started to have standards. Like for example, now we have a USB standard before there were other standards. So now you make a computer, someone else makes a keyboard, we'll plug them together, and it all works. So when an industry is immature, it turns out where to draw those boundaries to let different participants do their part and have it still interoperate. That's less clear. But then as the industry matures in this You know, more standards for, uh, if I want to publish a compressed LN model on the internet, what's the file format f…

AI assessment note: “I think this will evolve over time. Um, I'm going to make an analogy.”

Q Can you give me an example? I'm fascinated.

A Over a year ago, we thought that, uh, this is actually, um, uh, after, uh, after one of the Biden-Trump debates, you know, for better or worse, we thought that tariff compliance may become an issue. Maybe, unfortunately, we turned out to be right. But so last year, I think it was around August, we started building, started exploring building technology to help with tariff compliance. And by the way, I don't know if you've seen these tariff compliance docs, but frankly, When I look at what it takes to follow these, this paperwork is like, it, it makes me want to, oh my God, what is this? So, you know, you, you say import a bicycle, you know, then you look at the specs of the bicycle. Um, how much does it cost? The size of the wheels? There are all of these rules and regulations to like import a bicycle. It just, it just makes me go, oh my God, are humans really doing this? So we built agentic workflows, um, to read the tarot compliance documents carefully, Get the spec for what someone wants to import carefully, try to match, make suggestions, and so this is now one of our portfolio companies called Gaia, Gaia Dynamics, that, you know, because of the increased complexity in tariff compliance, has been doing pretty well, right? And so I find that we just could not have done this without agentic workflows. With medical assistants, we have, um, different startups, AI fund for fello…

AI assessment note: “we started exploring building technology to help with tariff compliance”

Q I sit in Europe. I obviously live in London. You told me you were born in London before this. My question to you is, it transparently feels like we are very far behind, and people say, oh, you've already lost. How do you feel about Europe's position in a very new world, and what can Europe do to regain some semblance of equality between the US and China?

A If I had one wish for the European regulators, I spoke with quite a few European regulators, I was hearing things like, we want to be leaders in regulating AI, and that's a competitive advantage. And with all due respect, that's not a competitive advantage. So my one wish for Europe is, uh, stop regulating so much and just focus on investing and building. The thing is, it's still early in the days of AI, um, it's still early in, in, in, in, in the game. And Europe has plenty of smart people, uh, Let people work hard. Don't force them to not work hard. Let people that want to work hard, work hard and stop over-regulating and just go and invest and build stuff.

AI assessment note: “stop over-regulating and just go and invest and build stuff”

Q I sit in Europe. I obviously live in London. You told me you were born in London before this. My question to you is, it transparently feels like we are very far behind, and people say, oh, you've already lost. How do you feel about Europe's position in a very new world, and what can Europe do to regain some semblance of equality between the US and China?

A If I had one wish for the European regulators, I spoke with quite a few European regulators, I was hearing things like, we want to be leaders in regulating AI, and that's a competitive advantage. And with all due respect, that's not a competitive advantage. So my one wish for Europe is, uh, stop regulating so much and just focus on investing and building. The thing is, it's still early in the days of AI, um, it's still early in, in, in, in, in the game. And Europe has plenty of smart people, uh, Let people work hard. Don't force them to not work hard. Let people that want to work hard, work hard and stop over-regulating and just go and invest and build stuff.

AI assessment note: “stop regulating so much and just focus on investing and building.”

Q democratization of knowledge there and the benefits that come from it. You said a word before, which was open, about the kind of the open weights ecosystem we've seen. We've seen this reversion back to like a closed world in a lot of cases. How do you feel about the reversion back to a closed? And how do you analyze the state of play today in that open versus closed?

A It's still very dynamic. Um, I think it is, uh, so for a lot of American companies, the leading frontier model is often kept close, and then the one tiered down model, not quite as good as release as open. I think it's much better than nothing. I'm actually grateful for all the teams that are releasing open source, open weight models. And then the other dynamic is, uh, uh, China especially has been really taking the lead, uh, or, well, Taking a lead or getting up there in terms of releasing tons of really good, um, open way models. So I would say if not, it is kind of, um, not whatever predicted, you know, a decade ago that China AI would end up being more open than America AI.

AI assessment note: “I think it's much better than nothing. I'm actually grateful for all the teams”

Q When we think about a core of a business, it's margins, and most of these business don't have margins. Do you care about margins when investing today, or with absolute respect, and it sounds disrespectful, do you take the kind of utopian view that it will just correct itself with time and with efficiency gains?

A At some point, uh, the laws of physics, I think, or the laws of finance or something, margins do matter. But one of the tricky things about AI is, um, we know the technology is going to change. So we don't build assuming the technology will be stagnant. We do build assuming the technology will evolve. So one, one obvious one, um, um, token prices have been rapidly falling, right? Uh, depending on who you believe, falling a percent year or whatever. Kind of, frankly, when we build prototype, We routinely just not worry about token costs because the first most important thing is this build a product that users love. Um, and then what we find is after we build some, this actually happened to me a few times now. We'll build something and not worry about the cost. And then, you know, users start to use it. And then our API build starts climbing. And then it is really like kind of, you look at this every few weeks and go, whoa, this getting really expensive. This costing me. Saturday of one engineer cost me more than two engineers cost me more than a whole bunch of engineers. All right. But fortunately, when that has happened, um, almost every time so far, we've been able to use techniques to bend the cost curve back down even faster than the rate at which token prices are falling in the market. And so I find that, uh, absolute margins are important, but when you have a view for wher…

AI assessment note: “we don't take a blind utopian, you know, AGI blah blah blah view either.”

Q Do you think Anthropic will beat OpenAI in the coding wars?

A Really hard to say. So OpenAI has a very strong consumer brand, and that's very defensible. In contrast, uh, developers are more likely to switch coding tools on the dime. So I love Cloud Code, I think it's fantastic, but I find myself using OpenAI codecs much more over the last month. Um, I think OpenAI codecs has actually gained real momentum, and then I'm also keeping an eye on Gemini CLI, which I think is also, uh, getting better, maybe the faster rate than people have given them credit for. So the coding dev tools and API tools market, the mode there's weaker than having a strong consumer brand. So I think that's something that, uh, uh, you know, companies have to sort out.

AI assessment note: “Really hard to say. So OpenAI has a very strong consumer brand”

Q David Kahn from Sequoia said, hey, a really useful barometer for effectiveness is, can AI replace the bottom five percent of capabilities of what workforce does? Joel from Cohere said, no, that's crap. The real question is, can it 10 x people's ability? Forget the bottom five percent, can it 10 x? How do you think about a barometer for success of the workforce with AI? With those in mind.

A In the case of software engineering, it is accelerating the writing of code. Uh, there are so many projects that used to take, you know, six engineers, half a year to build, that today I or one of my engineers can build in the weekend. I hope that we never have to go back to coding with our AI systems again, because the acceleration, the productivity boost is, is, is incredible. I mean, for example, one weekend I thought, oh, I want to You know, uh, my daughter, I wanted flashcards for my daughter to practice multiplication. She wanted to practice multiplication. She wanted flashcards. So I thought I could either drive to the store and buy a bunch of flashcards for her, or I could just, you know, use AI to write code for me, to generate and print out a bunch of flashcards. And so I did the latter. And so this is a very low economic value task. But with AI-sensitive coding, I could get that done, uh, uh, very quickly.

AI assessment note: “because the acceleration, the productivity boost is, is, is incredible.”

Q Can you give me an example? I'm fascinated.

A Over a year ago, we thought that, uh, this is actually, um, uh, after, uh, after one of the Biden-Trump debates, you know, for better or worse, we thought that tariff compliance may become an issue. Maybe, unfortunately, we turned out to be right. But so last year, I think it was around August, we started building, started exploring building technology to help with tariff compliance. And by the way, I don't know if you've seen these tariff compliance docs, but frankly, When I look at what it takes to follow these, this paperwork is like, it, it makes me want to, oh my God, what is this? So, you know, you, you say import a bicycle, you know, then you look at the specs of the bicycle. Um, how much does it cost? The size of the wheels? There are all of these rules and regulations to like import a bicycle. It just, it just makes me go, oh my God, are humans really doing this? So we built agentic workflows, um, to read the tarot compliance documents carefully, Get the spec for what someone wants to import carefully, try to match, make suggestions, and so this is now one of our portfolio companies called Gaia, Gaia Dynamics, that, you know, because of the increased complexity in tariff compliance, has been doing pretty well, right? And so I find that we just could not have done this without agentic workflows. With medical assistants, we have, um, different startups, AI fund for fello…

AI assessment note: “we started exploring building technology to help with tariff compliance”

Q to start with something that you've said before. You said AI is the new electricity, and when I think about electricity and where we are today, I want to understand the bottlenecks. And everyone seems to suggest that it really is about data, compute, and algorithms. Is that the three parameters to which we should think about bottlenecks? And if so, which one do you think is the biggest bottleneck?

A I would say the two biggest bottlenecks right now, um, it may be, I think electricity is one of them. Uh, so in the US, I am honestly worried that many data center operators were stuck in kind of permitting and, You know, and, and I know that local community support is important and some people don't want a data center there. But, um, once we build roads and railways as the infrastructure for a certain generation, data centers are the critical infrastructure for building the digital economy. And so lack of electricity in, in America and in number of Western countries is a problem. And in contrast, I see China building power plants left and right, including nuclear. So that would be interesting dynamic. And then semiconductors is another bottleneck. Um, but AI is so complicated. I think we also need more data. We also need more, um, better algorithms. You know, all of it is worth working on, but in the short term, some constraints with electricity and, and, and semiconductors.

AI assessment note: “in the short term, some constraints with electricity and, and, and semiconductors.”

Q to software budgets. And if we have that, then holy grail, me and you will make a lot of money with our funds and fantastic news because the TAMs have massively increased or the spends massively increased. If we're like, hey, we're not going to actually lose any people. Then actually we don't see that transition from human labor budget to software. Do you think we won't see that transition?

A So to me, the question is, um, is AI mostly for cost savings or is it for growth? And I know that, you know, it's difficult to change workflows. A lot of companies tend to think cost savings, but maybe here's, here's the problem. There's actually one pattern I see. Let's say I have a work task that has, you know, like Five steps, right? And let's say each step takes 20% of my effort. Like, uh, maybe I'm, um, underwriting approvals, you know, do I approve this loan or not, right? So let's say, let's say for simplicity, the five steps each takes 20% of my effort. If you can ultimate one of those steps, it's a 20% cost savings, which is really nice. You know, it could be great if you're a low margin business, but it doesn't feel like a game changer. So what I find is that the more valuable users of AI, It actually requires, it often requires rethinking that workflow. And the pattern I see is instead of taking a 20% cost savings, which you could do, that's fine, nothing wrong with that. The, the two patterns to then getting growth is, um, is either do more or do it faster. So in the case of underwriting, making loans, um, if instead of saving 20% of my human labor, if I can now rework the workflow, To turn around my decision making time. So instead of someone needing to wait, you know, two weeks before a loan officer looks at it, but we can just give you an initial answer in 10 min…

AI assessment note: “So to me, the question is, um, is AI mostly for cost savings or is it for growth?”

Q Where do we most need to be investing where we are not investing enough?

A There's tons of capital going into data centers and infra. Uh, we can debate, is there a bubble or not? We definitely need a lot of investments. Are we, you know, getting to the point where People are using such esoteric financial instruments to find cash for it, that there'll be a bubble. We could debate that, right? So we definitely a lot of investments, but when does it become over investment? That, that's a, that's an interesting question. The other place that I think we need to invest in a lot is not just the infra data center foundation model layer, but the application layer because of others having spent, you know, billions of dollars to train these AI models. We can now access them for, you know, hundreds of dollars, or thousands of dollars, or whatever, for tens of dollars. So, it's wonderful to build tons of applications that just were not possible before. Now, from a VC investment perspective, I've heard from multiple VCs, is, um, bizarrely, the cost of trying something out is so low that there are fewer ideas. It's not quite sure where to put massive amounts of capital to work at the application layer. In fact, if you look a lot of the, um, Uh, uh, application layer investments. Sometimes it feels like, you know, firms are putting in a hundred million dollars so that they can pay Open Ananthropic. So the Open Ananthropic can pay NVIDIA, which is where all the money …

AI assessment note: “The other place that I think we need to invest in a lot is... the application layer”

Q to software budgets. And if we have that, then holy grail, me and you will make a lot of money with our funds and fantastic news because the TAMs have massively increased or the spends massively increased. If we're like, hey, we're not going to actually lose any people. Then actually we don't see that transition from human labor budget to software. Do you think we won't see that transition?

A So to me, the question is, um, is AI mostly for cost savings or is it for growth? And I know that, you know, it's difficult to change workflows. A lot of companies tend to think cost savings, but maybe here's, here's the problem. There's actually one pattern I see. Let's say I have a work task that has, you know, like Five steps, right? And let's say each step takes 20% of my effort. Like, uh, maybe I'm, um, underwriting approvals, you know, do I approve this loan or not, right? So let's say, let's say for simplicity, the five steps each takes 20% of my effort. If you can ultimate one of those steps, it's a 20% cost savings, which is really nice. You know, it could be great if you're a low margin business, but it doesn't feel like a game changer. So what I find is that the more valuable users of AI, It actually requires, it often requires rethinking that workflow. And the pattern I see is instead of taking a 20% cost savings, which you could do, that's fine, nothing wrong with that. The, the two patterns to then getting growth is, um, is either do more or do it faster. So in the case of underwriting, making loans, um, if instead of saving 20% of my human labor, if I can now rework the workflow, To turn around my decision making time. So instead of someone needing to wait, you know, two weeks before a loan officer looks at it, but we can just give you an initial answer in 10 min…

AI assessment note: “to me, the question is, um, is AI mostly for cost savings or is it for growth?”

Q Can you talk to me about the constraints around semiconductors that you think are most pressing that most people don't realize?

A First, in my career working in AI, I have yet to meet a single AI person that ever felt like they had enough compute. So, um, you know, get us any amount of compute, we will use it all up and say we still don't have enough. So this is a constraint for the last 20 years or so. But what I'm seeing is, um, with the rise of Gen AI, there are very valuable workloads. For example, AI assisted coding, you know, it's fantastic. It's making us so much more productive. But if you use, Cloud code, enough. Sometimes you get rate limited, and I find that many companies have, really have excess demand, which is a very rare problem to have, but so many people want more OM inference, want more tokens generated, and we just don't have the semiconductors and data centers and electricity to meet the demand. But, you know, there's a lot we could do with AI, um, token generation, uh, and it's frustrating when we can't, when the supply side, we can't supply enough. To people that want it, on the demand side, you know, you, you get very limited if you, if you use too much.

AI assessment note: “we just don't have the semiconductors and data centers and electricity to meet the demand.”