The Exchanges

Q And yes, you can go even with, with, with cars, right? It's, um, A car is worth more if the prior, if the previous owner had, was significant in some way.

A Yes, yes, yes. A Steve McQueen is gonna, is gonna sell for more. And I think that, that for certain storytelling elements, for certain stories, that lore of like who this person is and their likeness, and then, and then he also makes a really good point that it's like, yes, you can, I mean, you can obviously generate something that looks exactly like a Van Gogh right now. You can paint it with someone who knows exactly how to paint like Van Gogh. It's not gonna have anywhere near the value, and when he gets to the idea of, ah, just AI generating a Dwayne Johnson movie, it's like, well, he'll sue you immediately, and that's very easy, and you've been able to Photoshop Dwayne Johnson into marketing materials for years, and we've had a system for counteracting that, like, you have to pay licensing fees, and that will happen. Ah, so he has a good, he has a good balance.

AI assessment note: “Yes, yes, yes. A Steve McQueen is gonna, is gonna sell for more.”

Q How did you conclude? What was your takeaway?

A My question was, so I don't see anyone who's a writer online, who's on sub stack, uh, being like, okay, Now's the moment. A million dollars on the table. I'm shutting down my sub stack and I'm going over to X exclusively. But a lot of people cross post. That's what we do. I think a lot of people, if you have a free, if you have a sub stack, that's part free, part paid, post the free part on X. Why not? It's all upside. It's just more attention that can actually get you into funnel. I believe that the posting of the articles on X has been additive to our sub stack list. I think this has been net growth. I don't think people have been like, oh, I'm unsubscribing from TBPN's newsletter because I'm getting it. On John's random Twitter feed that may or may not surface to me in the algorithm. So I don't think X needs writers to go all in. I just think that they want more copies of the best writing to make their way to X. And so if you have a great YouTube video, they want you to share it on X as well. So then they can surface that to people. If you have a great piece of writing, put it on X as well. And I think this does that. And I think they'll be successful there.

AI assessment note: “So I don't think X needs writers to go all in.”

Q How much did you know exactly what you wanted to build and then just execute it against that plan versus iterate to get where you are today?

A Um, I would say very little. So, so the story there is we, we just wanted to come and do something that we could do well. And even doing one thing well is very, very hard. So we were, we were laser focused at first. We just kind of looked around and thought like, we kind of, we weren't afraid to dream big, but we were kind of practical about it. We just thought like, okay, like what's one really, really big thing in crypto that we think could benefit from a fully permissionless platform. And purpose trading was the obvious one I think it was probably doing. More than 50% of the revenue in all of crypto back then, sort of back of the envelope calculations. And so we tackled that initially. And I think we, we, one big thing is like, we weren't willing to compromise even at the very beginning on How the thing ought to work.

AI assessment note: “I would say very little.”

Q is going, you know, negative free cash flow for the hyperscalers and really, you know, levering up in order to just win, right? Everybody just wants to win, and so, yeah, I was just curious at any kind of point. Obviously, the telecom was very debt-fueled, and we saw what happened there, uh, but it doesn't feel like in the modern era we've, you know, the hyperscalers have ever said,

A Let's really lever up our, well, and there's, there's, uh, the event that we were at this week together, uh, you guys missed it yesterday. There was more discussion. There's, you could define leverage more broadly. Like leverage isn't necessarily just debt capital. Like there's a lot of leverage in the system. If you look at the contracts and company, like look at opening at Microsoft, right? Like, you know, or, or any of these deal, like how much of the capital, whether it's, I mean, a lot of these contracts literally live on the balance sheet as liabilities. Totally. You've got the hyperscalers, you've got the media, you've got, like, the money's all just going around and around in a circle, and that builds leverage in the system.

AI assessment note: “leverage isn't necessarily just debt capital. Like there's a lot of leverage in the system.”

Q Uh, what, what, what's next for you at Meta? What, what, what are you focused on? What do you want to deliver this year?

A Yeah, for us, there's, look, the really big, big arcs, obviously are continued towards full AR, so we're really excited that we're supporting the entire strata we talked about before. Full AR, display glasses, regular glasses, and, and even other, other exciting fun factors that I can't yet tease. Um, on the VR side, we're advancing the hardware. We have multiple fronts that we're advancing the hardware on, and also on the software side, supporting creators. Gen AI. Don't sleep on it. That is the real unlock. You've got Roblox, you've got Minecraft, they're awesome tools, you've got tremendous communities there of creators, but there is a ceiling on how good the rendering can be in those platforms, which I know, and it's also, you have to be a certain level of creator to be able to produce good content in those platforms. Um, and with Gen.ai, you can lower the floor, so it's so much to do.

AI assessment note: “the really big, big arcs, obviously are continued towards full AR”

Q So what's the bull case for, for software? What's the bull case for SAS? What's the bull case for AI slopcos?

A Look, so to, to quote the godfather Neil, Neil Mehta himself, the laws of great businesses are the laws of great businesses. The job of a business in a capital society is to maximize and find the efficiency frontier for three things. ROIC, AKA return on invested capital, the amount of capital you can actually deploy, and how long you can deploy that amount of capital at and above market ROIC. There's a lot of different framings for the paths to do this and how companies can actually do this. The one that people like in tech circles is Hamilton Hemler's Seven Powers. A company accumulates power in the form of scale economies, network effects, uh, whatever, whatever power you want to take, and then uses that power to produce above market ROICs for as long as possible, and with as much capital invested in the business as possible. There are great atoms-based businesses that do this. There are terrible atom-based businesses that don't do this. There are great digital businesses that do this. There's great, or there's terrible digital businesses that don't do this. Um, I mean, you want to hear of a great atoms-based business that does this, listen to the Acquired Pod on Costco. Like, it's certainly not like a atoms versus SaaS thing necessarily. The advantage that digital businesses have is that in, like, in this process of producing above-market ROIC for a long time is that their p…

AI assessment note: “forms of power that naturally lend themselves to digital products”

Q No, it's very economically rational. Uh, what about the, uh, conjecture that, that you had, uh, a, a, a team of junior developers underneath you helping you actually accomplish the tasks? Is there any truth to that?

A I wish I had the money and I wish that was true, but that is not true. Um, any of the founders that, uh, you know, uh, I worked with can watch with that. I have multiple locations like pair program with people, you know, I've written every single inch of the court. Uh, you know, anti-metals, one of the companies I worked at, uh, you know, they have pretty high bar for design, as you might know, uh, they did a launch recently. Um, you know, there were two occasions where, you know, just before the launch, like, You know, the, the contractor who was going to do the front end for us essentially had bailed out and, you know, we as a team had to figure out and do everything, which was a team of only three engineers, right?

AI assessment note: “I wish that was true, but that is not true.”

Q have it all be driven by, uh, the, like kind of the same volume, uh, uh, like the volume algorithm that you see on, on, uh, on YouTube or Spotify. Uh, does that seem like kind of what the longterm, um, outcome might look like, or do we need to also get to some sort of like ad driven marketplace? Because that's ultimately where YouTube, how YouTube prices these things.

A Yeah, you know, I, I think it's going to be, um, I, I think that we, there's, there's a lot that we still have to figure out. I, I, I, last week, um, traveled up to Stockholm to meet with Daniel Ek, the, uh, the founder of, of Spotify. And, and it was, you know, it was really, um, it was, it was interesting. He said, you know, that the ultimate model that Spotify settled on, uh, and I think it's, I don't want to put words in his mouth, but, but it was basically like, it was, it was sort of the most naive of all of the different options. That he had come up with and it is works as it is a pool of funds and then they divide it up based on basically how many minutes of, of content you, you consume and there are different rates for music versus podcasts versus audio books. And, um, and it was fascinating to see. So you could imagine a version where Cloudflare effectively negotiates on behalf of all of the content that sits behind us with the big AI companies. People can opt in or opt out. Of being a part of that pool, the more content that's part of that, the more valuable it is to the AI companies, and then we would look at how we could split that up across how much crawling is actually done across that. I think one of the real keys will be, how can we give hints to the AI companies on what is the content which is the most valuable to them? So for instance, if Taylor Swift release…

AI assessment note: “the ultimate model that Spotify settled on... you could imagine a version where Cloudflare effectively negotiates”

Q the timeline there? What, what are the different players? What do people want out of it? What, what is the core, you know, kind of, uh, crypto industry, and both on the investor side and, and company side want out of it? And, and what are, what, uh, who, who would, who would not want it to go through, At least how the, the crypto side has it in mind.

A Look, I think, I think everyone in the crypto industry wants the stable coin bill to go through and become law. I don't think there's really any question about that. The question is, do you go for both? Um, someone just described it to me as a, in a sports analogy, and I'll probably fumble that one, but it was like, do you go for the field goal or do you go for the touchdown? And the thing about after a field goal, you know, after a field goal, the other side gets the ball back. And I think Congress really operates in kind of six to 10 year windows. And, you know, it's on their mind, reform regulatory clarity for crypto right now. And I think the transformational bill for crypto right now is very much that market structure bill. We also want that stable coin bill. And we have this unique moment in time where we have bipartisan support for both bills. So personally, I say go for both of them. And worst case, you end up getting the stable coin bill only. Um, but I do think that's not the Most desirable outcome for the crypto industry. I think the crypto industry deserves, especially after years of uncertainty, both a clear message from Congress. I think Congress ought to do its job and pass both bills. And I especially think that because after the Supreme Court last summer, uh, in a case I said that this was the most important case for technology policy, not for just crypto, but …

AI assessment note: “I think everyone in the crypto industry wants the stable coin bill to go through”

Q Do they ever have, like, Predator drones, like the Global War on Terror type of drone, or do they jump straight to quadcopter and kind of, like, leapfrog the technology?

A So, you know, you've had this, this, uh, Russian aggression war in Ukraine since 2014. Obviously the full-scale invasion was 2022, but even during that, that period before the full-scale invasion, there was some usage of drones for surveillance and, uh, dropping explosives. These are primarily still, like, small drones like what you're seeing now, uh, but this was not a, a proliferated technology. Then, when the full-scale invasion happened, uh, within a few months, the Ukrainians started Thinking about all these ways that they could use, you know, inexpensive drone technology to get a, an asymmetric advantage. And that is where FPV drones started becoming a really, really big deal. So they pioneered the, uh, really this idea of, you know, putting an explosive on a racing drone and using that as a precision strike weapon. There were instances of this happening in other places, but they really scaled it and they've really refined it. And then. Russia was, was much slower to take it seriously. Although now they're, they tend to, uh, in some ways outproduce Ukraine and they have a much, you know, more direct line to China, um, where most of these components are coming from. But since. 2022 and FPV is just starting to get used. Now it's reached an unbelievable scale. Uh, it's estimated Ukraine is going to produce four and a half million FPV drones this year. And those are ranging f…

AI assessment note: “These are primarily still, like, small drones... where FPV drones started becoming a really, really big deal”

Q Because, yeah, I mean, it's like if, if the reasoning tokens are inference tokens and, and, but they're what lead to higher intelligent, more intelligent models, like it's almost back in the training bucket again. Um, what bucket should we be thinking about and, and, uh, and, or, or are we, how firmly are we in the, the, the, uh, the, uh, the applied AI era versus the research era?

A Well, I think research is here to stay, and it's for all the reasons I mentioned above, right? It's such a, like, a rich time to be doing research, but I do think, you know, inference is going to be increasingly important as well, right? It's such a core part of RL, um, that you're doing rollouts, and I think, you know, we see 25 as this year of agents, right? Um, we think of it as a year where models are going to do a lot more autonomous work. You can let them Kind of be unsupervised for much longer periods of time. And, um, that is just going to put big demands on inference, right? When you think about kind of our overall vision, right? We, we lay it out as a series of steps and levels on the way to AGI, right? And I think the pinnacle really that last level is organizational AI, right? Like you can imagine a bunch of AIs all interacting. Um, and yeah, I think that's just going to put huge demands on inference, right?

AI assessment note: “Well, I think research is here to stay, and it's for all the reasons”

Q new Harry Potter. Everyone has to read it. It's amazing. And it's fully aged and it's fully generated or, uh, or this image, the images, they do go viral, but they go viral because they're AI move. 37 in the context of go did not go viral because it was AI. It felt like it was Actual innovation. So, uh, is that the right frame? Does that make any sense?

A Yeah. Um, I think it's not the wrong frame. So I think some, some quick thoughts on, on, on that. Um, I, I think kind of, um, when you have something that's, you know, very measurable, like win or lose, right? Something like, uh, like go, um, yeah, it's like very easy for us to kind of just judge, right? Like did, did the model do something right here? Um, and I think the more fuzzy you get, um, you know, it, it is just harder, right? Like, um, when it comes to, you know, is this the next Harry Potter, right? Like, you know, it's not a universally loved book. I think it's fairly universal, but you know, there's, there's some haters. Um, and yeah, I, I think it's, it is just kind of hard when it comes to these human subjective things where, um, it's really hard to put down in words, like what makes you like Harry Potter, right? And, um, and so, Um, I think those are always gonna lag a little bit, but, you know, I, I think, you know, we're, we're developing more and more techniques to attack kind of these more open-ended, um, uh, domains. And I don't know, I, I wouldn't say that we're not at an innovative stage today. So, um, I think my biggest touch with this was when we had the models compete on the IOI last year. So, uh, IOI, it's like the, the international, basically Olympics for, for computer science, um, Basically, the, the top four kids from, from each country go and comp…

AI assessment note: “I think it's not the wrong frame... when you have something that's... very measurable”

Q Jordy? Where, what are you expecting to see out of the kind of real world asset category over the next couple of years?

A Yeah, I mean, I think of stable coins as sort of the initial real world asset, right? And so you have dollars, bank account, and then the next, and then, and then the great thing about that, the kind of adoption there, and hopefully all the regulatory clarity we're getting is that will be a natural stepping stone to You know, you have Robin Hood talking about stocks. You have BlackRock talking about, you know, treasury bills. Um, wouldn't, I think it would be nice if anyone in the world, I think it'd be good for the US and I think it'd be good for the world if anyone in the world could buy, you know, four and a half percent, uh, treasury, you know, uh, interest treasury bill, right? And have it secure and easy and permissionless and low fee. Uh, I think it's, you know, it's good for our, you know, the US, the popularity of the dollar. And I think it's, you know, it'd be nice for somebody who has a volatile currency to have access to those, um, Um, so, you know, stocks, stocks, bonds, and then you can imagine, you know, um, all of these sorts of so-called dark pools. So, you know, still a large part of finance is people using Bloomberg and calling each other. And a lot of that, you know, like trading bonds and muni bonds and corporate bonds and things like that. There's a lot of interest from the banks to think about ways to use blockchains, uh, to create essentially marketplace…

AI assessment note: “natural stepping stone to You know, you have Robin Hood talking about stocks.”

Q Where, what are you expecting to see out of the kind of real world asset category over the next couple of years? It feels like it's, this year, at least to me, it's felt like we're days or weeks away from some pretty high profile assets coming on chain. Uh, I'm curious how you think about the category broadly.

A Yeah. Um, I mean, I think of stablecoins as sort of the initial real world asset, right? So you have dollars in the bank account, and then the next, and then, and then the great thing about that, the kind of adoption there, and hopefully all the regulatory clarity we're getting is that will be a natural stepping stone to You know, you have Robin hood talking about stocks, you have black rock talking about, you know, treasury bills. Um, wouldn't, I think it would be nice if anyone in the world, uh, I think it'd be good for the U S and I think it'd be good for the world. If anyone in the world could buy, you know, four and a half percent, uh, treasury, you know, uh, interest treasury bill, right. And have it secure and easy and permissionless and low fee. Uh, I think it's, you know, it's good for our, you know, the U S the popularity of the dollar. And I think it's, you know, it'd be nice for somebody who has a volatile currency to have access to those, um, Um, so, you know, stocks, stocks, bonds, and then you can imagine, you know, um, all of these sorts of so-called dark pools. So, you know, still a large part of finance is people using Bloomberg and calling each other. And a lot of that, you know, like trading bonds and muni bonds and corporate bonds and things like that. There's a lot of interest from the banks to think about ways to use blockchains, uh, to create essentially…

AI assessment note: “I think of stablecoins as sort of the initial real world asset”

Q I have one last question then we'll let you get out of here. Um, uh, and feel free to just, uh, take it as like meta commentary on the nature of the question, but who is Satoshi Nakamoto?

A Um, So, the official answer is, it doesn't really matter, because, like, the idea stands on its own, and so if anybody ever comes out and we find out who it is, like, in their, their will, or, you know, whatever. But, but the, the unofficial answer, I think, is It's probably some combination of, like, Hal Finney and Nick Szabo and, um, maybe one other person, but, um, that's my guess, is that it was a collaboration between those people, um, and, you know, but ultimately it doesn't matter. And Hal Finney is frozen at Alcor, but he's, you know, he's no longer living, so we don't, you know, but maybe if Alcor actually works, which is a big if, in the future we can bring him back to life, we'll find out.

AI assessment note: “It's probably some combination of, like, Hal Finney and Nick Szabo”

Q There's hills, there's random people on the street, there's bikes, and, uh, I think, like, the telling that some people would get would be, like, if you can solve San Francisco, you can solve kind of anything, Maybe not Boston in the winter, but, uh, is that the story you tell, or is there something else, and, and what are you pulling from their strategy overall, seeing what Waymo's done?

A Yeah, I mean, just on the commentary of kind of hard cities to drive, I would love to see Waymo's in Riyadh, Saudi Arabia. There is, it is bananas to take car rides there. Um, but what, what I think is interesting is we've always compared the kind of Waymo and the Tesla strategy, right? The Waymo Cause I think, yes, they operate in San Francisco, which is, you know, lived there for many years, not, not trivial to drive them, but the cost structure of, you know, how much does that vehicle cost, right? The amortizing that CapEx against all the vehicles, right? If you have this really expensive Spencer suite, you really like cost optimized, uh, the vehicle. Um, and it's hard to, when you're kind of reliant on a lot of these sensors and kind of advanced computing, um, it makes it difficult to really get it to be substantially cheaper over time. Um, and to deploy a bunch of, you want to deploy a bunch of these, right? Like the, the, the question I've always asked is why doesn't Waymo deploy, you know, you have Google's budget, right? Why wouldn't you go deploy, you know, a thousand of these tens of thousands of these, right? Um, we've always viewed it more in the Tesla approach, right? Of like, if you can make kind of camera based, uh, autonomy and kind of driving work, um, you can have a substantial simpler cost structure, right? Um, you're able to, and with that simpler cost struc…

AI assessment note: “we've always compared the kind of Waymo and the Tesla strategy”

Q down to temperature. Um, out of all, you know, when you look at manufacturing broadly or I'm assuming you have some type of framework for evaluating whether something like can be automated to the degree that people would like to see out of manufacturing or areas that, you know, are basically shouldn't be touched, right? Like something like, like shoes, which have infinite sizes and a bunch of different factors.

A Yeah, I think it's like a combination of like three things, like the market size opportunity and how, how technology ready it is to, to, to be, to be kind of disrupted, right? So, um, uh, but also I think there's a lot of conversation around automation that, that is in the previous paradigm. I think for now, for the first time, we have this concept of element, this concept of We can actually reason very complex sequence of operations as long as we can train the robots on that sequence of operations. So it comes down to what data do we have available to train the robots? We already figured out, okay, you know, neural networks, LLMs, uh, these, uh, transformers, if you give it enough data, it can actually learn a very complicated, complicated task. The real key was, okay, what, where do we generate enough data? Where do we have enough data to train it? Um, so it means, and unfortunately for a lot of manufacturing tasks, the data is not out there, right? You cannot train a very complicated model on it. For, you know, for ChatGPT, the internet had trove of free data that you could use to create a very complex, um, kind of chatbot. For us, coming up with right sequence to make car doors, um, or shoes doesn't exist. So the key is, can you actually provide a solution That can scale with limited amount of data, with human intervention and limited amount of data. So you can deploy it in…

AI assessment note: “combination of like three things, like the market size opportunity and how, how technology ready”

Q done today, they're selling servers for businesses. That's about forty billion dollars revenue. And then, uh, laptops for often computers, personal computers, but for business workers, another forty billion or something in revenue. Uh, he seems much less focused on product design, brand, all the different things that Steve Jobs is known for. Um, is he more of like a private equity financial mastermind? Is that what makes him special?

A No, he's definitely, he's, this is the weird thing too, where again, not really comparable to any other people. He was upset. He was a computer nerd. Like he'd get in trouble in high school because he'd just sit in the back of class and read like bite magazine and like, which is the stuff it was computers, but his parents were successful in business. And so every single conversation they'd have when he was a kid, like this dude was reading fortune and force magazine at 12 years old. And then we didn't even get to the point where, like, he does, he takes Dell private, by the time they, they, they go off on this crazy run. By the time, uh, they start hitting a lot of headwinds, like, 2005. By the time they get to 2012, 2013, they're in real big trouble. And so he takes this huge risk of doing the lever, the, the, the, uh, the biggest technology leverage buyout in history with Silver Lake Partners. So they take Dell private in 2013. Okay. The story gets even crazier. So he understands he was actually preparing for the age of AI like 10 years ago. And so Jensen, I talked about this in the, um, in the episode where I found an interview with Jensen Wong and Michael Dell and Jensen's like, well, there's only one company in the world that can actually build these data centers at scale and give, do every single thing from the storage to compute everything else. And that's a Dell. And th…

AI assessment note: “No, he's definitely, he's, this is the weird thing too, where again, not really comparable”

Q had in your careers, uh, Google and Stripe, uh, in some ways, big companies that maybe move slower than a small startup, but at the same time, both of those organizations, I feel like the time from, Hey, we're starting the company to, we have a product that it wasn't a research organization for years. Um, what have you learned from those organizations? What are you taking into this experience?

A Stripe really taught me everything I need to know about building a robotics research lab. Uh, it's just lessons galore. I'd never worked in a research environment. Uh, so I don't have priors. I don't know what a research environment actually looks like. I just know what our research environment looks like. I feel like we Whatever it looks like, uh, maybe it operates exactly like startups, like maybe grad programs are exactly like startups, but we feel like every other company I've worked with that moves extremely quickly and has a clear set of goals and direction and just has a bunch of people that work behind it and work extremely hard to solve whatever it is that, that we're setting our, our minds towards and, and, uh, there's so much I learned from Stripe that informed that, but a lot of it's just the, the obvious stuff, right? It's like, Hire exceptional people. Set a really high bar for it. Don't compromise. Set a very clear set of goals for everyone. Really align people. Actually, I think that's one thing I really took from Stripe, is you want an extremely aligned set of people. And I've never seen more alignment than I've seen at Pi. We talk about the alignment tax. Like, when we're bringing someone on, how much work is there to align them around our mission, our way of seeing the world, and almost everyone that joins, there's like basically nothing. Most of the people t…

AI assessment note: “Actually, I think that's one thing I really took from Stripe, is you want an extremely aligned set of people.”

Q get a post-mortem on that, but then you're also not afraid to go and you did this sort of Squid Games campaign that I saw getting a lot of attention as well. Um, how do you, how do you think going forward sort of balancing these more like scrappy kind of organic activations versus, you know, going big and, and working with, you know, global celebrities to build the brand?

A I think we should keep being scrappy. Uh, the thing, it's, it's, by the way, like, uh, he's a pretty global star, like, well-known star, uh, like, people recognize his face pretty quickly, but he's also not, like, as expensive as Hollywood people, and so, uh, that's why we decided to work with him. And, uh, the other, other things, like, it's the concept that matters at the end, right? Like, coming up with the right concept is more important, uh, than, like, who you work with. Um, and we, we will still try to keep doing these one-off viral moments. It's all, it's about taking bets, not all of them land. Um, the Super Bowl was good, like, in fact, like, the retention from people who came to the Super Bowl exceeded my expectations. Uh, and, um, again, like, whether it's better than doing Instagram ads, it's not clear to me yet. Uh, we are, we have to explore all these different platforms. Um, but I do think there's, like, you know, two things you benefit from, right? Like, we did this thing with Ben Shapirover in his podcast. Like, he pulls up perplexity and asks questions. Um, I think, like, uh, that doesn't actually convert. Like, you cannot track performance, but, uh, it, it, it does lead to more brand awareness. Like, if they're, like, he has a lot of listeners, and, like, it's a different way of doing ads on a podcast, where you're not actually, like, Having him say perplexi…

AI assessment note: “I think we should keep being scrappy.”

Q like the, the failure or like the, the, the setbacks to Apple intelligence have kind of raised questions about like, Hey, should that Siri button be remappable to other things? Um, how are you seeing the, the landscape around hardware evolve? Because that feels really, really important to you. If you don't want to go build it yourself, burn a hundred million dollars, trying to build the next Apple product.

A Yeah, I, I, I don't actually think that, uh, we'll, I, I, I doubt that we'll ever be integrated into the iPhone. I don't actually know that that is the best modality, even if we could. Um, I think, I think most people, I mean, friend.com only like, like everyone is sort of betting on the idea that there will be like this new hardware element that, that, that, that is like a companion. Um, I think that's, if we were to delve into hardware, that would be most realistic, um, like the next step. But what I really would like to do is sort of like, Beat out Neuralink and just sort of just directly skip all that and get to the chip inside your brain. That is the true end state is what we want anyways.

AI assessment note: “I doubt that we'll ever be integrated into the iPhone.”

Q wrong, but do you need to build equipment from scratch? Is this a project where you're going to be building, like, A, a gigafactory like what we've seen for the Starlink units at some point, or is it more about, um, assembling different components and then being really strategic about placing them and then building a network on top of that and really like the services side of the business?

A Great question. Something that our head of manufacturing Thomas thinks about a lot. Um, we're definitely going to be leveraging outsourcing in early days. We, we are a vertically integrated company. Like we, we design all of the different, uh, components. Um, we have those outsourced and then, um, you know, brought in. We're not, we're not, um, you know, like making our own circuit boards at this point in time. Um, there's certain things that are like more efficient, uh, efficient to insource versus outsource. Um, so largely, um, leveraging outsourcing. Um, initially I think, you know, we're really focusing on modular units and making our, our units designed for manufacturing. So, um, making sure that we can parallelize development, making sure that, um, we can have things ready to be integrated at like kind of the final hour is, is the thought process just to accelerate our manufacturing, uh, capability. And then gradually over time, when we figure out like what is actually cost efficient, um, and time efficient, we bring it in house entirely.

AI assessment note: “we're definitely going to be leveraging outsourcing in early days”

Q an essay, and I'm struggling with something in particular, and theoretically somebody, you know, a child could have the experience of their teacher just sitting, you know, with them, writing, and being able to talk out loud, and I don't think we've seen voice Uh, adoption as an interface as much as some people would have thought 10 years ago, but it feels like maybe now is the moment, right?

A Voice is so exciting. I, I do think that, um, when we, when we're able to make voice a main mode of communication digitally, we're going to get so much more information and it is the information that continues to power and tailor more personalized and engaging experiences. So the, the, the amount people share If they have a voice, you know, a speaking opportunity versus a typing opportunity is tenfold. So I do think, I do think that's a big unlock. And I think, you know, this is a good point because this is how this is going to unfold. Like we're just in the early stages of seeing what's possible. And, um, you know, there is some very exciting voice technology out there and, you know, everyone's been buzzing about the company Sesame. What they have is pretty amazing.

AI assessment note: “I do think that's a big unlock... we're just in the early stages”

Q Yeah. So last question is a question I think you had for us. Do you want to close out with that?

A Well, I mean, this is the question that I've just been thinking a lot about and it gets down to like the nature of the soul and the nature of what it means to be a human. You know, there's that famous, um, idea of the map and the territory. And what I think is basically happening is that when it comes to writing the map and the territory are going to be the same. And when the map and the territory are the same, what is just, what is, what comes from that? Um, is, like, if you have such a good simulation of humor, and such a good simulation of consciousness, and such a good simulation of care, is it those things, or is it not those things? And I don't know the answer. It's just the question I've been asking all week. And, um.

AI assessment note: “if you have such a good simulation of humor... is it those things”

Q That's very cool. Can you talk a little bit about logistics and, uh, just how, how can we increase resiliency there and what startups are kind of tackling that or how, how should they even think about tackling that?

A Yeah, great, great question too. Um, let me start framing it by, I look at the defense, national security space, and again, we at Shield, we think pretty broadly when we, when it comes to national security, right? So that is anything that touches autonomy, AI, cyber, and space that has an application to both. When I think about the national security context, it's how do you sense, make sense, affect, and sustain? Sense is easier, right? That's, that's satellites, like Topher came on, one of our portfolio companies, right? He's sensing the environment and making sense of it. How do you affect it? So what are you actually going to do about it? And that could be something physical that could be Brandon and Chile drones, or that could be non-physical digital in nature. And there's this, and then there's a sustainment part. And I think this is just such a huge opportunity. And there's a lot of companies working on it and so much more to do. And I specifically call it sustainment, not logistics for a couple of reasons. One, this term of contested logistics, while important, doesn't get to the root of the issues, which is how do you both operate in a constrained environment? So how do you do things when you know you're going to run out of fuel or you're going to have these certain constraints? Um, uh, and then what are you going to, how are you going to sustain it more broadly, not ju…

AI assessment note: “I specifically call it sustainment, not logistics for a couple of reasons.”

Q Um, what, what, what immediately comes to mind when somebody says, uh, Five simple words, short term pain, uh, sorry, six, short term pain, long term gain. How does that make you feel?

A Nothing's good. Nothing's good is going to come from that. Wasn't it Xi Jinping just like a couple of months ago with all the youth unemployment? I think the phrase he used was eat bitterness. That's what, that's what they use should do is like, you should just eat the bitterness that you have and enjoy it. Nothing's going to good from is nothing good is going to come from that. Of course, there is logic to the idea of like invest in the short term, sacrifice in the short term for long-term gain. I'm not sure that's what's going on. Today with what's happening today. I don't think this is seven dimensional chess trying to get some sort of long-term plan. I think this is, this is, that's, that's a different topic. And I, I think there's a difference between, you know, investing for the long run and like foregoing consumption so that you can grow your money and like, you know, hitting your kids with a belt to instilled grit in them. That's, that's not short-term pain for long-term gain. So there's, there's, there's a balance that needs to be struck here.

AI assessment note: “Nothing's good. Nothing's good is going to come from that.”

Q Can you talk just a little bit about how you're personally using AI? What, what, what, what tools and models are working their way into your daily stack, if anything?

A Mostly it's just kind of fun and amusement. Um, and in terms of work, I I've, I've tried to do a few things. I've uploaded book chapters or articles into chat GPT and, and, and several other models and just said, Hey, give me some feedback on this. And I found that it's pretty good for what I would call micro feedback of just like, Hey, you probably needed a comma here. It's not that good for macro feedback of like, Hey, was this chapter good? Like, did it make any sense to you? It's like, so far it's not that great for that. I'm sure that'll change, and there's probably other models that are better at that. Um, and so it's, it's, it's, it's, it's pretty good for, for, for writing, for, for my profession. You know, it's going to differ for everybody. Um, I, I, I use it quite a bit as, like, a, as, like, a thesaurus of, like, I'll, I'll be, I'll, I'll have writer's block on a sentence, and I'll, I'll put in my partial sentence and say, finish this for me, and that is actually pretty good at that. So it's, it's good at all these, like, little things for writing. It's obviously going to be, this is such a, a, A dumb take, like an obvious take that the people who are going to do the best are the ones who use their personal skill and then use AI as a, as a tool to leverage it rather than the ones who are using it to replace it. So use, if you're writing and you're writing your own s…

AI assessment note: “I've uploaded book chapters or articles into chat GPT”

Q reasoning model driven, but fine-tuned to help with the execution of ARK prizes? Is that, is that breaking the, like the, the rules, or is that something that is actually encouraged and fine? How do you think about, like, I guess it all, it all, um, boils down to, like, overfitting on this problem, but, uh, it seems like Even with the incentive to overfit, it hasn't really happened yet.

A Yeah. Um, arc, this is one of the reasons arc B one lasted five years was because it was, it had very good design, original design built into it with a public test set and a private test set. And that private test set really prevented folks from being able to sort of overfit on it. Um, and, um, so we, there's actually two tracks for the arc prize foundation. We have the, the contest track. This is on Kaggle. This is the like big grand prize. All the prize money is attached to this. Uh, Kaggle graciously donates a lot of compute to allow us to host there. And on, uh, the grand prize is basically to get to that 85% within Kaggle's efficiency limits. So this is about, get about 50 dollars worth of compute per submission that you send in. Um, and, uh, this is like a pretty high bar for efficiency. And in fact, this is not an arbitrary bar. We do think that efficiency is actually a really, really important aspect of intelligence. You know, you can brute force your way up to intelligence, but we really do want to be shooting for like human targets and efficiency for this stuff. That's the contest. When we launched last year, uh, there was a lot of demand From just the community of the world on like, Hey, I, okay, I get that. Like, I can't run my big language model on this, you know, in Kaggle, but I really want to know how to frontier AI systems do. And so we launched a second track …

AI assessment note: “private test set really prevented folks from being able to sort of overfit on it.”

Q What are these, uh, Is there anybody besides you, uh, fighting to save, uh, you know, uh, the, the private and public, uh, just private and public groups from being able to do, uh, cloud seeding? Like, is it, is, are, are you the last man standing, uh, or do you have a team?

A So, well, I've got a great team at Rainmaker, and I'm really grateful for that. And we've got a bunch of farmers, um, that are Beneficiaries of our program go to bat for us because like they previously had to tear up their pistachio ranches or their pistachio orchards because they didn't have enough water for them, and now they're better able to farm because of the production Of water that we're, we're doing. Um, but you know that, that part where, or that like famous Stanford speech where Teal says, um, you know, you should go after a really small market. Like if there's lots of people involved and it's already too late. Well, I'm not like the last man standing. I'm kind of like the first man standing. Um, so it's other than the farmers that we're helping and the other great team members. I had a rainmaker running around the country. Um, it's, it's, uh, Augustus DeRica of the, uh, a couple States.

AI assessment note: “Well, I'm not like the last man standing. I'm kind of like the first man standing.”

Q that, the golden retriever's running outside, you're taking the horse down to the local, uh, golden retriever mode. Yeah, to the local, uh, saloon, and it just seems like, Like, it's the, it's the West, right? Again, you have new land that people can go live in, and that's just, like, a land of opportunity, and I think that'd be so cool, but, uh, yeah, what's your take on it?

A Dude, I mean, that's what the Central Valley was. The Central Valley used to be deserts and swamps, and now it is the most productive agricultural region on the planet. It produces, like, 30% of all of the fruits, and then the majority of, like, the red fruits in the United States. Um, we used to terraform all the time. Like the Hoover Dam was an attempt to terraform the West. Uh, I think that like people that, you know, like maybe it's a last man problem. Maybe it's Luddism. Um, maybe it's just like general, like, well, no, I won't say that. Um, but, uh, like we've just lost the desire to be great. I think in the United States we've lost a desire to like build the future and see the future through. Um, if I was just planting trees, Yeah, for sure. It would be easier. Um, you need the water though, so unless you're setting up enormous conveyance systems, which China largely is, and cloud seeding in addition to that, it's a non-starter. Um, there's a couple technologies, you know, I mentioned last time, like soil amendments that can get soil to retain more water, that would be good. Um, but, uh, you know, trees, they emit dust that creates like a majority of our clouds in some regions and induces precipitation. Cloud seeding Like, it mimics that natural process. It's just with the material that works a little bit better. Um, so Uh, yeah, man. Like, I, I plan to plant trees in th…

AI assessment note: “Dude, I mean, that's what the Central Valley was.”