Inside Josh Kushner's Rise, 𝕏 Timeline Reactions | Andrew Ross Sorkin, Brian Potter, Pari Singh, Henri Stern

29:31Ty in the X chat says bear market and morals. Bear market and morals. Well, in other news, David Sachs is going to war with Anthropic. He quoted one of Anthropic's co-founders, Jack Clark, and said, Anthropic is running a sophisticated regulatory capture strategy based on fear-mongering. It is principally responsible for the state regulatory frenzy that is damaging the startup ecosystem. Not what you want to hear from the AI czar if you're running a US-based foundation model lab. Yep. but we'll see how this plays out. Yeah. You want to hear about Graphite if you're the AI czar. That's right. David Sachs.

30:15You've got to use this for your code review because it's the age of AI. Graphite helps teams on GitHub ship higher quality software. You can get started for free. Let's go through some of the Jared Kushner news. Josh Kushner. Josh Kushner. Jared Kushner is also featured in here. Sorry.

30:37I was reading Jared Kushner because Tyler posted sending Jared Kushner 25 books on how to prevent AI bubbles from popping, which was an absolute banger. Got 1 ,000 likes. Tyler's been on a tear, but he's not even in his chair. You got to cut to him before he gets back. No. Oh, he was out of his chair. I caught him black. Hey, congrats on your fantastic book. Back-to-back bangers. 1K club two days in a row. Don't mess it up. If it's not a three-peat, just don't show up to work. You're done. Do you feel sequel pressure now? I mean, there is a lot of pressure on my shoulders. Heavy is the head that wears the crown, Tyler.

31:17Get ready. You've been on a roll. Let's see what your ex-creator is. I don't know. I see a couple bangers back-to-back, and I think, hmm, maybe Tyler doesn't have enough on his plate. Maybe you should get him to vibecoat something. I got a hit. I think it's five million impressions in three months. to, like, I'm not available for the creator payout yet. Oh, okay. So I think I gotta slop it up. Okay, yeah, yeah, yeah. Why is no one talking about Ilya Sitzkevart? I'm never leaving this app. Yeah, yeah, yeah. Just keep posting those. You'll definitely... Ty in the chat says, if you cut to someone not on their battle station, they have to wear a part-time podcaster.

31:53Thank you, Ty. Yeah, really phoning it in today. You're having fun. Yeah, let's get into this piece. Jeremy Stern profiled Josh Kushner in Colossus. It released today. It's not in print yet. It's going to be in print, right? Get ready. Subscribe now. Called The New World. Where do you want to start? Where should we start? I mean, there's so much in here. I like this bit about Spotify. So in 2012, Thrive invests at $6 million in a growth round of the Stockholm-based Spotify out of$150 million fund and allocation Kushner had regarded as a favor until learning nearly a decade later that Spotify CEO Daniel Ek was in need of exactly 6 million to close the round.

32:41So he's just like, oh, thank you so much for like making room for me. Like, this is amazing. Like, I'm so glad I could like, you know, get my 6 million and Daniel Ek's over there being like, the round would have completely flopped if you hadn't come in. So thank you so much. But there's more to this anecdote. This is the reason Eck knew he liked Kushner was that a few years earlier when Spotify was only available in Europe, Eck was notified by an executive that someone in the US had managed to register a fake UK address in order to download the app via the UK App Store. They discovered it was Kushner sitting in the library of HBS.

33:20Wow. I pulled out some highlights from the piece. It's way too long for us to try to get through entirely. But some background. On November 17th, 2023, Kushner was a 38-year-old spouse of a supermodel, brother and in-law of American political royalty and founder and CEO of what had very suddenly become one of the most coveted venture capital firms in the world. He was also the grandson of survivors of the Novo Gruta ghetto massacres, indignant, indigent refugees who over the course of the Cold War built a New Jersey real estate principality that their son, Josh's father, expanded into a multi-state empire before his conviction on felony charges and sentencing to federal prison and before the White House activities of his older brother, Jared, put Josh in the crosshairs of a torrid political convulsion of which he wanted no part.

34:13So, um, yeah, I mean, ultimately it's, it's, it's this, uh, Josh has put on an absolute masterclass of how to be an icon while being relatively in the shadows. Right. He just sort of like pops his head up in these key moments. and yeah I think when I think about what Josh and the Thrive team has built it's actually it's I mean it's just like it's almost it's almost unbelievable like the story the story end to end by the fall of of 2023 Thrive Capital the New York-based investment firm Kushner started 13 years earlier had become an overnight sensation. In 2010, Thrive's first fund was$5 million and included companies like Kickstarter and GroupMe.

35:09By 2023, its eighth fund was$3.3 billion, including a maniacally concentrated$2 billion investment in Stripe at a$50 billion valuation. He's like, here's two thirds of my fund. This is the Fifth Avenue strategy, buy Fifth Avenue. He talks about this on Invest Like the Best. And a$150 million check into OpenAI at a$29 billion valuation. The companies are now valued at$107 and$500 billion, respectively. Along the way, Thrive's bet on Instagram, Spotify, Warby Parker, Skims, GitHub, Slack, Robinhood, and other companies had become conspicuous for being prescient, aesthetic, and exquisitely timed among its most vindicated admirers and for being absurdly priced, momentum chasing, and too highly concentrated in dysfunctional businesses with unproven returns among increasingly sheepish critics.

36:00So I like that they got Johnny Ive to give a quote for this piece. Some people just have innately wonderful taste and intuition, said Johnny Ive, Apple's legendary former chief design officer who's now working with OpenAI on a hardware device. Josh has wonderful taste. I think it bleeds into his product intuition, which is just fabulous. Another highlight here. in 2010, moreover, the unknown Kushner's brother, unknown little firm, was making a bunch of large but weird sounding claims for itself, like that it was a staged geography and sector agnostic venture firm that would concentrate all its investments in a very small number of companies, that it was not only an investment firm, but also itself a company, that it incubated its own companies as well as invested in others, and that it didn't just invest and incubate, but functioned as a service provider, product creator, and embedded operational commando unit for founders.

36:55By 2023, every self-respecting investor on Sand Hill Road was also saying such things about themselves, even as they wondered how a New York firm made up of a handful of kids in their 20s and 30s, many of them with zero experience in venture capital, and from the technology Bermuda Triangle of New Jersey had become some of the most desired investors in tech and the ones most closely associated with the otherwise distinctly West Coast boom in AI. Of course, Josh had a very untraditional pathway to venture capital. He went to Harvard for undergrad, while at the time, Mark Zuckerberg had just dropped out of Facebook.

37:34Later, he got a job at Goldman Sachs buying distressed debt, and then went back to Harvard for his MBA. And of course, I'm joking a little bit, but it's, what I appreciate about Josh is, in many ways, he was on the perfect trajectory to be a venture capitalist, and yet he has succeeded beyond what, he has succeeded in a way that so many other thousands of people that had the same kind of pathway into the industry by being in the right circles, like being early to a number of these, these different trends, like he's succeeded on a hundred times on a scale, a hundred times greater than, um, many of the other people that were, again, had the same kind of like pathway in the industry.

38:33Yeah. My read on is that a lot of the VCs that started in the same vintage that did not become massive institutions maybe just got caught in the trap of diversification. I keep going back to that story of Josh saying he wants to buy Fifth Avenue, buy the best asset in the class, be concentrated, build a big allocation. And that takes guts that I think a lot of VCs kind of fell in the trap of like, I need to get a bunch of logos from a whole bunch of companies, or it's particularly cool. It gets actually higher status to be, oh, first check in this company or see. Even if you have a tiny allocation that gets completely diluted down, you're not actually making as many dollars.

39:21If you put$2 billion in Stripe at$50 ,000 and it goes up to$100 ,000, you made$2 billion, versus you put$200 ,000 into some company, it gets diluted down, you make$50 million. like you made way less dollars on dollars return um but it's like somehow higher status in venture to be like yeah i was like the first check-in and i think uh just been early in a lot of companies but i think that he understands the importance of like portfolio concentration yeah actually getting all the dollars in the companies that matter yeah and that's like yeah and i think early on i think my sense is that there was some like hunting for logos early on yeah but they were when the fund sizes were very small.

40:04And that led to him having the track record and ability to put $2 billion into Stripe out of a$3.3 billion fund. And also like we're just like the Thrive era, the era that Thrive grew in is an era where venture changed pretty dramatically. I mean, we were talking about the Intel IPO raised$6 million. Like what is your role as a venture capitalists, like deploy$100 ,000 and then wait for the IPO like a year later. And it was like that in the dot-com boom and Google IPO'd pretty early. Even Facebook IPO'd at what, 50 billion. And it was like the biggest IPO of all time. It was in like the decade.

40:43Yeah, there were investments in that era that I'm sure the partners were underwriting as, I think there's a 70 % chance this company IPO's in the next two years. This is why we're investing. and then they became compounders and over time it looked like. But they distributed and so a lot of the funds became RIAs. But Thrive has been able to say, yes, we're investing in a company at 50 billion because we expect it to grow and grow and grow and actually deliver like a venture style return or like a significant return. I appreciated this exchange between part of the article. The beginning goes into Kushner visiting Rick Rubin in Malibu and Josh was telling Rick, my deepest insecurity is that I have these intuitions about things that I cannot explain to anyone, Kushner told Rubin as they sat in his garden overlooking the ocean.

41:31Sometimes I see or experience something and it makes sense to me. I fall in love, but I cannot explain why. Like when Thrive invested in Instagram or Spotify or OpenAI, I could not explain to anyone why the products made sense to me. It is my job to learn as much as I can for my team and teach them as much as I can. But often I have to push forward on my intuition alone, which is why my even deeper insecurity is what if I lose it? Like, what if I lose the capacity to feel or experience these things? He's basically having this exchange because, of course, Rick Rubin notoriously is just going off of raw intuition and vibes.

42:11I think the other, I mean, it goes into his entire, his family's crazy history dating back to Europe during World War II, getting out of Europe. But the, I think, can't be understated how formative it was, I think, for Josh to go through this sort of, you know, right as he was, the quote here, by the time I was in high school, my father had accomplished a tremendous amount. He was deeply impactful in both the business and philanthropic worlds. And then overnight, our family were outcasts. The world treated us all one way for the beginning part of my childhood, and then suddenly they treated us very differently.

42:52That experience showed me how the world works and why you should not care too much about what people think. Of course, his father was embroiled in a wild crisis and ultimately went to prison for a couple years, but I think Josh says elsewhere in the piece that he wouldn't, I forget the line exactly, but something like, wouldn't wish what he went through on his worst enemy, but at the same time, wouldn't, is sort of grateful for what kind of turned him into a monster, basically. Very kind monster. Well, if you want to try and reverse engineer Thrives, returns, dump all that data in Julius, and chat with you today and get expert level insights.

43:39It's the AI data analyst that works for you. Tyler, what are you thinking about Kushner? I think there's also underrated. He seems very aura-pilled, right? He has this kind of nonchalance about him. There's a good quote, I think, that relates to this. It's about Andy Golden. He's the Princeton Endowment head. He says, Golden later recalled a happy hour for VCs in Cambridge in 2010 where he saw a six foot three emo looking kid in a black cardigan standing apart from the group staring at the floor. I mean, that is just pure aura. Yeah. Uh, apparently we're on French TV right now. Thank you for the notification.

44:16Send us the link if you can find it. Uh, we'd love to see that. Um, yes, there seems to be somewhat of a, of a core correlation or inverse correlation between like, uh, just how much content and availability you like how, how available you are and your aura like ilia like never does any press yeah he's very mysterious very mysterious because the mystery allows like people to like build this idea of you and exactly you're like oh yeah like like even when ilia posted yesterday like oh great the best day ever everyone was like clearly this is a yeah like bubble is gone i mean we're gonna keep going yeah and it's like that's not what it was about at all he was talking about geopolitics um in the end of the gaza war but um So Rimkay in the chat says, we asked for a link, and Rimkay says France 2.

45:06Oh, you just have to go. Just go to France 2. No, France 2. France 2 is the channel name. Yeah, no, I know. Can you go find it? Is there a way to stream that or something? We don't have French cable here. I would love to see. Go sign up. Try to sign up for an account. Did they translate? Did they put subtitles over us? I'm so curious about this. Anyway, while we dig in. Oh, this is very based. I assume it's the interview. Yeah, yeah. But I'm wondering, because they could have put French subtitles over us, or they could have dubbed us with 11 labs or something. Ask Le Chat how to get access to French cable.

45:36There is more details on how the OpenAI deal came together in this article that are pretty interesting, which you should go read. I feel like I'm allowed to trash investors because I was one for most of my career, Sam Altman said. Most investors don't work that hard. They're usually not available for midnight at calls and won't drop everything to fly across the country on short notice to do you a small favor for you the next day. Josh is consistently willing to do all those things. He's incredibly hard working for his companies. He'll do whatever it takes any amount of time. Nothing is too big an ask.

46:04During that crazy week where I got fired and rehired, he just put his entire life on hold. He didn't leave his hotel room for 72 hours. He just worked nonstop, very strategically, very effectively to get things back on the rails. Thrive had first gotten involved in OpenAI in early 2022 when they met Altman to discuss a new round. they'd been given a preview of GPT-3, the foundation model that preceded ChatGPT, which Kushner reportedly became so obsessed with, he almost seemed haunted by it. Now, that needs to be corrected because GPT-3 came out in 2020. So I think they're either talking about GPT-3.5, DaVinci-002, which was really popular, or it's just a preview of ChatGPT because investors were getting previews of ChatGPT beforehand, and it was blowing everyone's mind.

46:46But it was a really complicated deal, and a lot of people passed, basically for the wrong reasons, in my opinion. So on Kushner's work ethic, we met up earlier this year for coffee and I had gone to sleep. He was still on the East Coast. It was like 2 a.m. He's texting like, yeah, do you want to like, I'll see you at like 10 or something like that. I'm like, you're on the East Coast. It's 2 a.m. for you. You're going to get to the West Coast and be ready for, and so that is part of the story, obviously, is just insane work ethic. Yeah. Did you see in the chat somebody said, we're on France 2, and Taylor says, dang, they released a sequel to France?

47:36Tyler is MVP of the chat this week. Yeah, Tyler. Sorry, Taylor. Taylor's been crushing it. Tyler has been struggling to learn French. I thought you studied French. You took French. I took two semesters of French. I found it. Yeah, let me. I'll screen share. Okay, yeah, yeah, yeah. Let's figure it out. I'll keep reading. So OpenAI at the time was a cap profit subsidiary controlled by a nonprofit board with the mission of safe AGI development, taking legal precedence over profits. Microsoft's$1 billion 2019 investment gave it a dominant position in OpenAI with complex revenue sharing agreements and preferential access to the company's technology.

48:15That and the company was reportedly valued at$29 billion while doing a trivial$50 million in revenue. What a ramp from 50 mil to, what are they doing now, like$10 billion,$12 billion,$20 billion, something like that. It was all very weird. There was a reason no other investor submitted a term sheet for that round. Wow. After several conversations with the investment team, however, Kushner marshaled his case. Forget the nonprofit structure and Microsoft and all the red flags, he argued. If you can create this much enterprise value, everything else is solvable. In any case, colleagues recall, he kept repeating things like, I saw the future, and this is the one.

48:52Great call. Great, great, great call. Extremely hard. I know some other investors who literally passed on that round because they were like, yeah, we talked to our lawyers, and they were like, this doesn't make any sense. We couldn't understand how we were getting it. There were lots of smart people that passed. How early did people feel like ChatGPT was going to be a meaningful threat to Google? That probably started in like March of 23, but like November, I mean, people, when ChatGPT, when GPT-3 came out in 2020, people were starting to say like, you could use this to, you could ask it a question.

49:35You really had to prompt hack because you couldn't just ask it a question. You had to like ask it a question for a list of bullet points and put a couple, a couple I'm seeing. Sorry, I'm just laughing. Rimkay is saying it's live now on news. It's live now on news. It's on France too. It's on the news. It's on the news. We're trying to find it. It's on the news. And so Thrive submitted a term sheet, 130 million from its main fund at the$29 billion valuation. In November of 2022, OpenAI launched ChatGPT. Within two months, it became the fastest growing consumer app in history. By summer of 2023, Thrive was working on a$90 billion round.

50:14In August, they agreed to anchor a$400 million of a$500 million employee tender offer, critical leverage in the exploding war for AI talent against OpenAI's publicly traded competitors. by November 17th, 2023, Thrive had committed around$700 million to the company. So pretty, pretty remarkable results. And really just goes to show you that like there are certain pitches in venture, like seeing ChatGPT before it launches that you need to really, really, really, really, really swing hard at. And Josh swung at the right pitch at the right time, which is like fantastic and much harder than I think people think.

50:57It's so easy now that everyone uses ChatGPT all the time to just be like, oh yeah, I would have made that deal too. Yeah, and ignoring the FUD and just like doubling, tripling, quadrupling down every single possible. Yeah. I mean, when that initial round was happening, I was also like, I mean, I wasn't an investor, but I was thinking like, it just makes a lot of sense. I was looking at the team and I was like, if you just make the bet on like, you have the CTO of Stripe, Greg Brockman, you have the former president of Y Combinator, just the resume of the founding team at that time. Everyone was really, really tough.

51:35You had Ilya. The mid-curve response was the corporate structure was weird. Like, this isn't a C Corp. It's breaking the rules. I don't fund non-C Corps. Like, I'm out. But just going back to basics of being like, the founding team, who's running the company? Are they the best? Is this an important category? And do I have the best team? It's like, it was hard to make an argument against it at that time, in my opinion. Anyway, I should have invested, I guess, but I like doing content instead. It's more fun. In other news with OpenAI, they did an interview with the Broadcom team and Sam Altman shared a little bit more about the custom chips that they're working on.

52:19Before we tell you about that, let me tell you about FALL, the generative media platform for developers, the world's best generative image, video, and audio models all in one place, develop and fine-tune models with serverless GPUs and on-demand clusters. Used by Adobe, Shopify, Canva, Quora, and many more. So Sam Alman says, to zoom out a little bit, if you simplify what we do in this whole process, you know, melt sand, run energy through it, and get intelligence out the other end. As we realized we were going to need the whole system together to support this, it's just gotten more and more complex.

52:48So it turns out Broadcom is also incredible at helping design systems. So we are working together on that entire package. We are able to think from etching the transistors all the way up to the token that comes out when you ask ChatGPT a question and design the whole system. All of the stuff about the chip, the way we design the racks, the networking between them, how the algorithms that we're using fit the inference chip itself all the way to the end product. And so it feels very much like even if they're not fully plateau-pilled, there's a lot more research to be done. There's a lot more improvements.

53:23There are at least pieces of the chat GPT architecture, pieces of the system, whether that's 4.0 level inference or reasoning tokens, that should effectively be baked down onto a chip. And so that's what OpenAI is working on with Broadcom at this point.

53:43But Stratechery, Ben Thompson, had a great deep dive on OpenAI and Broadcom and sort of laid out the argument for them working together and baking stuff down onto a chip. It's a fun read. You should go check it out. Should we move on? Yep. I was just going to try to figure out what. Well, while you do that, let me tell you about Turbo Puffer. Search every byte. Serverless vector and full-text search, built from first principles and object storage. Fast, 10x cheaper, and extremely scalable. Broadcom was a$220 billion company the day that ChatGPT launched. It is now 1.7 trillion. Almost a 10x. Wow.

54:27NVIDIA, too. NVIDIA was much, much smaller. That might be a 10x as well. Yeah, wild growth across the entire category. Basically, everything has 10x in price or something like that, roughly. Yeah, it's interesting. It'd be interesting to go back and do the math and the VCs. There was some reporting in the Financial Times a couple days ago talking about who actually owns OpenAI. And it would be probably interesting and painful to look back at the investors, how much they invested in OpenAI versus just like market buying NVIDIA and Broadcom and some of the other beneficiaries. And would they have made more money?

55:05um there the the open ai's didn't altimeter do both yeah i'm sure i'm sure the players like that did both but uh open ai's expected ownership after the company restructures microsoft is going to have about 30 percent open ai employees are going to have close to 30 percent open ai the non-profit is going to have uh 20 to 30 but that doesn't mean you shouldn't still donate if you're thinking about doing some charitable giving this holiday season, consider writing a donation to OpenAI, the nonprofit. Prioritize tipping your cap table first. Yes. And then after that, consider donating to OpenAI, the nonprofit.

55:44But anyways, Microsoft at 30%. OpenAI employees around 30%. OpenAI nonprofit at 20 % to 30%. SoftBank about 10%. And the remaining is Thrive, Khosla, MGX, and a number of bedrocks in there. Yeah. I wonder how accurate that reporting is. It feels like directionally correct. I think it's directionally correct. But there's some pretty big error bars on either side. Yeah. But still, I mean, you just look at some of these numbers that we pulled up. It's like, you know, what was the original OpenAI deal that Thrive did? It was at a$27 billion,$130 million at$29 billion. So 2 ,900. So that initial deal was half a percent of the company.

56:33Right. And that was like a whole round. And so that was a very low dilution round even at that time. Yeah. What's notable is like looking at this kind of rough cap table, it's just astonishing to end up in a situation where VCs have the lowest ownership out of the employees, a nonprofit, a strategic partner. or SoftBank kind of makes sense. But yeah. It's all hands on deck in the chat trying to find France too. Thank you for everyone's service. It seems like they're making progress using VPNs, all sorts of stuff. Daniel says, what's a good number to tip your investors 20 % of the last round or is that too much?

57:16I think more like 2 % of the last round. They'll feel very appreciated. But it's got to be delivered in a sports car form. It can't just be cash. Let's read this Jeremy Giffon post. Things that seem true about AI today that will probably be laughably wrong in a year. One, it still seems underrated that ChatGPT has replaced 85 % of my Google usage and has become a most used app. That's certainly true for me. The reason we're so upset about slop is because it's obvious we're all going to love consuming it in two to three years. It's not going to be slop for long. What do you think? I, yeah, I mean, there's out of the, out of the hundred sore, you know, maybe, maybe way more than that, right?

58:06Call it a thousand sore videos I've seen one way or another. Like 10 of them have genuinely been hilarious. Yeah. and that is about the same ratio as like the regular internet right yeah see a thousand videos like yes some some of them are going to be um but but i i think uh already there's there's i'm seeing instances where it's not slop still using the sora app is not i don't love it but some of the outputs are are i like it as a creative tool like i i think that if you just view it as a creative tool there's going to be creative uses and stuff that's amazing even if it is a little sloppy uh the other lens to view it through is like uh buzzfeed was creating a lot of kind of slop content but with humans and that never really broke through and became something that a lot of people at least in my world enjoyed it kind of remained slop forever and um but but But as AI generation as a tool, like you saw over the weekend, I was having fun in our group chat, generating a Sora video, sending it there.

59:14I didn't even post it, but the result was really funny in the context of our friend group. And so I think Jeremy's right there. Number three, he says, many non-AI products are being adopted due to economy-wide corporate and capital market top-down mandates to buy slash fund AI, a rare reversal of the norm where everyone wants to turn over their existing vendor for the new one. This will have a similar pull forward effect as e-commerce during COVID, even if AI ends up being way overstated. And so right now, some of the booms that we're seeing, some of the ramps in these companies going to 100 million is just that the entire Fortune 500, and we're going to get into this with some of the open AI partnerships, the entire Fortune 500 is saying like, Like, CEO, you need to have an AI initiative or else you're gone.

1:00:03Like, you can't be sitting out AI, even if you're Walmart, for example. Maybe Walmart should get on Profound. They could get their brand mentioned in ChatGPT. They could reach millions of customers who are using AI to discover new products and brands. So, Jeremy continues. By the way. Yes. uh best diff in the chat did send me the video of us on french television nice uh i cannot download it because X does not apparently allow you to download videos from DMs. So I asked him to send it to Ben's email so that we can pull it up on the show. We will work on that. We are working on it. So Jeremy goes to his specialty, how companies are using AI.

1:00:48He says, AI roll-ups and vertical AI. I've yet to see any part of a business unit that can reliably have cost, cut, or revenue grown with AI. That's very interesting. because the narrative, we've heard this from a number of private equity folks who are saying like, I'm going to buy a company, throw AI on top and cut costs in this division or grow revenue in that division. And even, again, software engineers at a company, if maybe they haven't been using AI, you come in and you're bringing like fire to, you know, you're like, here, look, it's fire. It's magical. Fire. There's something interesting about reliably.

1:01:29Reliably. There's still an art to it. Yeah. Here's potentially why you should still be bullish on AI roll-ups is because you take a bunch of smart tech people and you bring them into businesses that haven't historically had a lot of great technology. And they just build great software. It's possible to grow businesses faster, be more efficient, things like that, even if AI isn't actually doing the heavy lifting. Yeah. So when I've talked to like a roll-up founder and I'm thinking, you know, they've pulled together, in this case, they had pulled talent from like five of the best companies in the world, at least from a talent density standpoint.

1:02:10And they're working in these categories. I won't name it because I'll dox the company. But they're working in a category that never in history had super talented engineers working on it. Yeah. And so they're probably going to do really well, even if, again, it's not like integrating an LLM into the workflow that's actually doing that heavy lifting. Yeah. So it's possible that like, yeah. He says companies can ship a lot more software quickly. Complexity of the buildout is no longer a moat, which I think is interesting. Revenue velocity. Yeah. But so the only thing is this kind of contradicts the.

1:02:44No, no. No. So one is it's hard to just go into a business and say, no matter what your business is, I know that I can come in and drop AI into your manufacturing process or your customer support department or your finance operations department or your sales department and immediately see a result on cutting costs and increasing revenue. It's not a reliable playbook. Every company - Yeah, but I'm just saying he's saying companies can ship a lot more software quickly. Complexity of the build-out is no longer a moat. So you'd buy a platform software business that has a bunch of other products that you want to build.

1:03:23And if the second point is true, you could just let go of a lot of the engineering team and say we're going to cut or we're going to change our hiring plan because we just don't need as many engineers. And you can grow revenue while keeping your costs relatively the same, right? Yeah, but at the same time, that software that you launch will be more competitive because there's no longer a moat around the complexity of the build-up. And so you wind up with a more commodity product at the end of that build-up. And so maybe the equilibrium is that you don't actually grow revenue that fast? I don't know.

1:03:56It does seem like it's very case-by-case basis, basically. Everyone wants to paint with a broad brush and just say that every company will be transformed by AI. Every company will grow with AI. But it's clearly more nuanced than that. Yeah, he's also saying margins and headcounts do not yet reflect the narrative around developer productivity. What I would say here is that if your developers get a lot more productive, you might actually want more developers, right? Yeah. Because like if a developer can produce three times as much valuable code, if you agree with that, wouldn't you want to grow your engineering team?

1:04:39yeah i mean there is an equilibrium like there is like a game theoretic equilibrium where where where if we're direct competitors and i give all my engineers ai and you give your all your engineers ai like we both have to employ the same amount of engineers to compete with each other because if i fire all my employees and i'm just using ai and you they're just gonna out ship you you're gonna out ship me because you have ai and humans and you're double important you know and so This is another point here. It says PE and VC, and ironically VC, have not found a way to adopt AI internally. I asked a lot of Gil about this.

1:05:14Yeah, a lot of Gil about that. And yeah, I mean, I think he generally agreed. Yeah. I mean, certainly for research, I would imagine that they've adopted that just as a replacement for Google, like Jeremy's first line, like replacing Google search knowledge retrieval, that type of stuff, AI, is certainly a good point. But yeah, I mean, it's not like you can just say, hey, go find me the next unicorn and don't make mistakes. Where AI has created a lot of value is by being the missing puzzle piece for existing businesses, usually around, this is usually around onboarding, migration, or database querying.

1:05:55Oddly, I can't help but feel that at its core, that its core quality is just that it makes everything much faster. It feels as though the microprocessor, blockchain, and AI have all been these pushes away from clock time towards newer, faster standard measure of times. I'm not that optimistic, but it does feel like there's a chance for the first time that we could get entirely away from the ad model, taking shots at ads. How dare you, Jim? He posts from the ad-supported platform, x.com. Yeah. I don't know. I mean, it does seem like ChatGPT is going to monetize pretty heavily on the subscription side and then also on the referral affiliate revenue side.

1:06:34But they're creating a ton of service area for ads. And I think that there will just be ads. So I disagree with that one. Anyway, speaking of ads, Google AI Studio, the fastest way to prompt from prompt to production with Gemini. You can chat with models. You can vibe code. You can monitor your usage. They got Nano Banana. You can talk to Gemini Live. Go check it out. Go to ai.studio. post from Hypebeast. Apparently Tim Cook now has his own custom Labubu complete with an iPhone 17. I think that's in orange. I didn't realize they made custom Labubus, but I mean, what a great... What do you think Tim actually thinks about Labubu?

1:07:16He doesn't look that happy there. It's kind of a forced smile. Yeah. I don't know. There's been a few photos of his office. You remember that that viral image of like, this was Steve Jobs office, this is Tim Cook's office, and Steve Jobs office is all messy and Tim Cook's office is all clean and it's like supposed to be a referendum on the different leadership styles of the two CEOs. But of course people made the point that that picture, that iconic picture of Steve Jobs, his office was in fact his home office and we haven't seen Tim Cook's home office so it might be equally messy and Steve Jobs work office might have been equally clean.

1:07:53But in that photo shoot of Tim Cook's work office, it did seem very clean. It didn't seem like it had a lot of tchotchkes or trinkets or I don't know. I don't even know where you'd put Lububu's in there. But he doesn't seem like someone who's been collecting Pokemon cards and is just ready for the next thing to collect. But it's certainly a fun drop for Lububu. And if you're building a company, you should make a Tim Cook version of your product and send it to him. I mean, get a free ad out of it, I suppose. Who knows? Breaking news. I want your reaction to it, John. Please. Your authentic reaction.

1:08:31John has not seen this post yet. I haven't. Breaking. OpenAI to partner with OpenAI to help fund OpenAI. OpenAI up 90%. That sounds real. It's not. It's a good post. I know that they had a crazy amount of partnerships. Just this week, I mean, the broadcast thing, we'd heard about and then it went bigger. Jerry Capital says, this is how you top. And it's Walmart partners with OpenAI to create AI-first shopping experiences. This makes a lot of sense. NetCapGirl said, in case of emergency, break the glass. It's an OpenAI partnership. Well, of course, Salesforce announced a partnership with OpenAI this morning.

1:09:12Yes. And people were memeing it because the trade down. Didn't the stock trade down on it? 2%. It was crazy, crazy, crazy. And then High Yield Harry says, say the line, Bart. we're partnering with open ai and so salesforce and open ai just announced a partnership it's the kind of partnership that will let companies across salesforce agent 360 platform in chat gpt clearing sales records i mean it all makes sense to integrate it uh what's interesting is that at one point salesforce was building their own foundation model and seems to have you know backed down from that a little bit mark benioff said uh back in 2023 which i believe was the coup right this was right around the coup um so this was this was when uh open ai researchers were leaving uh and there was a there was news that sam might go to microsoft research and bring a bunch of people there yeah that was like over the weekend over the weekend right and it was really crazy and mark benioff was like this is my zuck poach moment i can poach all these folks and uh he said salesforce will match any open ai researcher who has tendered their resignation full cash and equity OTE to immediately join our Salesforce Einstein trusted AI research team under Silvio Saravice.

1:10:25Send me your CV directly. Einstein is the most successful enterprise AI platform completing 1 trillion predictive and generative transactions this week. Join our trusted enterprise revolution. High Yield Harry says, this is hilarious in retrospect. Yeah, it's like it needs to be the main thing. Unless you're Google and you invented the transformer, like everyone else needs like a partnership. And so, uh, Amazon's fortunately in a good spot with Tranium, but they still have to partner. It doesn't seem like they've been able to really keep up in the foundation model race. It's, it's becoming like a triopoly between open AI, Anthropic and, uh, and Gemini and everyone else is kind of dropping out.

1:11:05Although we do have new, new, uh, uh, some new information from Bology about the Chinese models and what's going on in the free AI model world on Elimarena. It's been back and forth between Meta, DeepSeek, Alibaba, and now Z.ai is in first place. And they've all been going back and back and forth. It's a knockout, dragout, bite. How each organization's best open weight model ranks on Elimarena. Z.ai is now in first. Good domain. Yeah, great domain. Not a.com, but still solid. Before we move on from Salesforce, course, I think we can leak that Mark Benioff will be on the show on Thursday, joining TBPN at 12.15.

1:11:49You heard it here first. That'll be an interesting discussion to understand how he's thinking about build versus buy in the AI era. There was definitely a time when it made so much sense to say, well, I don't want this company to train on it. I have my own system. It's not that expensive, but now it's extremely expensive. and OpenAI offers an enterprise plan where they won't pull your data back into their training runs. And so you can use all the best OpenAI models. They will serve them. It's Sam Altman's problem to provision all the GPUs. You're even better than being a renter because you're just a token buyer.

1:12:25And so there's very little risk of, oh, we overshot this AI thing. We'll just scale back your token consumption. And all of a sudden your economics go back to what they were before at the AI boom. So it's a very safe place to be. Jordi, you want this video? Let's pull up France 2. French TV, baby. French TV. Oh, this is the studio, not with the fireproof side. This is a cool. This is very cool. Do we have audio on this? I want to hear some French TV. They're really filming us all. We didn't get any audio? Okay, odd. Well, we wouldn't be able to understand it either. Oh, look, here we go. They're calling you a specialist.

1:13:03An expert in artificial intelligence. Yep. this is the first trading card. I remember being at breakfast and Tyler described that idea and it absolutely ripped. We had a couple others. What was really funny about this was that the interviewers kept asking us like, okay, so for this trading card, like how much did that person make personally? And we're like, look, like it didn't exactly leak like that. It was a more of a general thing. There were some leaks of specific deals, but there are a lot of them haven't been confirmed. like we're more just having fun with the overall trade deal narrative.

1:13:39We can't really even clock a particular offer within an order of magnitude. Well, we got to find out how to get the audio. Yeah, we do. Oh, there was audio, says Fanta. So we will figure that out. And now. Without further ado, we have our first in-person guest of the show. Welcome to the stream. How are you doing? Good to see you. Good to see you.

1:14:07Welcome to the Ultradome. Hopefully you're looking cozy in that sweater. Hopefully you're enjoying the rainy day in Los Angeles. I was told, come to LA. It'll be nice. It'll be warm. It's like 330 days a year. It feels very warm in here. It's actually more of a neutral temperature. But through the power of movie magic, it appears that we are in front of a warm heart. Anyway, please introduce yourself for the stream. Yeah, I'm Henry Stern. I'm one of the co-founders of Privy. We build embedded wall software. I was told to come here to help translate the French television. I'm French originally.

1:14:44You speak French. I watch France Télévisions. No way. France 2? Are you a France 2 fan? I'm a France 2 fan. Is France 2 better than France 1? They just have France 1, 2, 3, 4, 5. I think it goes into the 40s, if I can. We're working on iterations at this point in time politically. and I think TVPN is welcome respite for all French political watchers out there. That's fantastic. Yeah, it was fun telling the story of what happened over the summer. It felt like flashback because the AI talent wars, I'm sure you tracked them in real time in June, July. He was busy getting acquired. You were busy, but still, I mean, I'm sure they broke through to you and you were aware of them.

1:15:22We saw these things, yes. You saw it, of course. And everyone's wondering like, hey, what's my comp package going to be next year? Well, we, we, so we normally don't do, uh, life stories on TVPN, but when did you actually, uh, when did you leave France? I'm assuming you grew up there. Uh, I grew up, so I grew up, uh, there till I was like 12 or 13. Then I came to the U S which is how I shed my accent to some extent. Yeah. It's barely, I don't even notice it. Um, and, and got very Americanized. So I'm a, I'm an American person insofar as startups are concerned and I'm a French person insofar as like, you know, voting passports and food is concerned.

1:15:56Sure. Sure. Makes sense. give us a play-by-play of the last few months. It's been busy. It's been very busy. There's like two parallel worlds. On the one side, we got acquired by Stripe, so we had the great folks at Stripe reach out. I think you gonged us already. I know, we're second gong because you closed. Retroactive gong. You've got to hit one for closing. One for the announcement, one for closing. Of course. And basically, so they reached out in around April. and went through really what was a very, very fast time. So a huge shout out to the M &A team over there and frankly to the entirety of leadership.

1:16:37But had you not talked, surely you would. You'd never heard of Stripe before that. I didn't know what it was, no. No. No, so obviously we knew Stripe and frankly a lot of how we shaped our company and the way we wanted our product to work was based on what we saw Stripe doing. We'd had talks with their crypto team and in general and we did a lot of work with the team at Bridge. through joint customers that we serve together. And the outreach was some version of, listen, we think wallets as distributed global bank accounts are an extraordinarily powerful primitive to have as part of the stack that we're hoping to serve.

1:17:09We'd love to talk to you about this, the way these work. So we had a conversation, and it very quickly led to where we were. And I think we talked through what would it take for us to be excited beyond the general excitement of Stripe about joining forces. What would it take for us to accelerate? and we got there and to their credit we moved really quickly so by May we were signed with a term sheet by I guess mid-June we announced and we closed in July That's great What was the pitch around Synergy? Was it obviously they're well capitalized so you don't have to go out and raise anymore. They have a huge customer base of people who have saved credit cards through Stripe Link but then they also have a bunch of merchants who have done developer integrations with the API.

1:17:57Like what stood out to you as like the most important piece and what was maybe less important than people might think? Yeah, I think there are two parts to it. The first is how do we get this tech to mainstream? How do we get it to customers who otherwise don't care about crypto whatsoever? And one of the things Stripe has been excellent at is basically hiding the rails behind, you know, extremely complex banking infrastructure. Exactly, and you take these like, you know, hundreds of banking partnerships, you turn them into an API that just works and you enable, you know, in their case, like commerce on the web for the last 15 years.

1:18:32And the question is, as this stack evolves, how do we do the same for crypto? So that felt like very natural. But I think for us, the point was one, it is being able to strengthen our teams across security, infra, and obviously across distribution to customers we wouldn't be talking to otherwise. And two, being able to plug into money movement rails, like making it that fiat and crypto are married to the point where they become indistinguishable from one another and where as a consumer of this API, you no longer have to think about one versus the other. So I think the way I look at it is broadly with two prerogatives.

1:19:03If Stripe is AWS for money, then this is a rail they should be leveraging and they are intent on leveraging and building up. This is why they acquired Bridge. This is why they acquired us. And broadly, we want to be the best purveyors of money movement and storage systems on the web so that you can build completely global businesses on modern financial rails. That's one. And then two is beyond the tooling that we provide, can we make it that every existing Stripe user today can benefit from these rails so they can provide cheaper, faster global payments and they can enable anyone to hold dollars anywhere in the world.

1:19:38How do you think about the modern onboarding of the consumer going back to maybe 2021, 2022, the way people would start. I mean, a lot of people were on centralized exchanges, like, oh, I want exposure to Bitcoin or something. I'll go buy some and I'll go through the setup flow and KYC and whatnot. And then the wallet era, I felt like started with NFTs, different tokens that were sort of like getting traction on Twitter, people would talk about them. And then you'd have to be like, oh, well, to buy this, I got to set up this wallet. I have a specific wallet for Solana or a specific one for Ethereum.

1:20:17And it was a very prosumer activity. How do you think it evolves to a more consumer world going forward? My framework is there's the speculation era. People are signing up for different services to speculate. I think what you had seen probably prior to starting the company is that eventually there would be this more functional use case of I'm going to sign up for... products in general for specific purposes, whether it's to buy things, you know, pay people, etc. So you think people start with signing up? Or will it be like PayPal, where it's like, I get an email, and I'm like, I got to go claim that, and then I set up the wallet?

1:20:54I think there's going to be two entries. I mean, broadly, what we're seeing, so to your question, like, you know, what's been happening these last few months, there's obviously what we've been doing at the preview level, the Stripe level, but then there's the space overall and the amount of institutional adoption that's coming. And I think we broadly see it on three fronts. We see crypto as an asset class. So how do you unlock basically access to these assets for traditional consumers? And you see it through JP Morgan giving access. You see it through, I think Morgan Stanley is going to unlock spot trading on E-Trade.

1:21:23You see it through, we're working with Deutsche Bank, JV called AllUnity on a European stable coin for support. So broadly, that's the first part. And quickly there, so someone has an account with a bank that has a brand that's probably been around for 50 or 100 or hundreds of years. And that particular web app or mobile app is like provisioning a wallet for them that they might not even know that is provisioned for them. Is that how you think that plays out? I think we're – I mean this is I think where it leads to start. I guess the broadly there's access to the asset class. There's obviously stable coins and global distribution.

1:22:06And the last is overall tokenized assets beyond things like tokenized equities or tokenized deposits. That's where you see players like Apollo or Blackstone getting involved. And the way we're seeing this develop is you broadly have a move from crypto-native startups to fintechs. And so what you just talked about is exactly what's happening with the neobank stack. And a number of neobanks that we're seeing provision wallets exactly as you're saying it, which is as part of your traditional neobanking app, You'll have the ability to move your checking deposit or balance into a wallet that you control through which you can get, for example, yields on DeFi in a way that as interest rate goes down becomes more interesting.

1:22:46So I'll be phase two to then broad. Because right now, if you want to buy Bitcoin on E-Trade, for example, you buy the ETF, which is like a very abstracted, like multi-levels of abstraction. It's not even self-custody. And so, yeah, okay. That's a very cool word. maybe what's what are the institutions focused on like what are the kind of the categories they're focused on so one would be stable coins hey these are going to be big we want to have a play here so whether that's uh leveraging them in the business or launching their own stable coin is one there's opening up access to crypto markets so their users are another one what are kind of the Are there maybe, are you thinking about less like sexy use cases for crypto, like things like back office stuff?

1:23:30Or is that more on the bridge side of the business? It's a great question. We are insofar as basically setting up, you know, you can think of a wallet in this new instantiation where the wallet is just this embedded product that sits within your existing stack as a way to do broad treasury management. So bridge, you know, is working, for example, with SpaceX on remittances across SpaceX's global operations. But this is a random example, but you can imagine a company like Coca-Cola who is bottling plants all over the world needing to have capital put to work. And so this is where the wallet stack becomes useful.

1:24:03Broadly, it breaks down largely, as you've said it, which is to say it's access to crypto as an asset class. It's stable coins, and that's on two fronts, issuing the own. And for example, Stripe and Bridge have launched open issuance where you can actually launch and own the economics of the stable coins that you're putting forth. but broadly we're seeing a lot more people starting to do this as well as using stablecoins for things like remittances and payroll. We see folks like Deal or Remote that are doing stablecoin based payroll. We see folks like Zeps, Remitly, Felix who are doing remittances using stablecoins.

1:24:37And then the third is broadly trying to see, can we open up tokenized deposits and other things to financial markets globally through crypto. So this is where I think the Apollos of the world are playing and want to become, in a sense, like, you know, lenders on chain for folks across the world who have not had access to private credit. So if I'm a contractor for SpaceX, a quadrilineatoll or something, and I don't have USD or I don't have a USD bank account, SpaceX HR, effectively, or financing might send me an email, hey, sign up for this to claim your payment, your payroll, and it's just a stablecoin wallet.

1:25:20They provision the wallet right then. Exactly. And then they have it, and they can move it, and then they can interact in crypto, but then they'll also be able to move it outside. In a specific case of the SpaceX bridge work, I think it's used for internal treasury operations. Oh, sure, sure. But that's exactly the sort of thing that we're seeing coming out. For payroll. Exactly. Got it. And revances. That makes sense. That goes back to your question of how are people going to get wallets in the future, 2021, 2022. You have to be this prosumer. where you have to be extraordinarily sort of high activation energy of, I want this and I'm going to self-select into this.

1:25:47I think we're going to basically just start seeing myriad of ways in which this starts to bleed into your life as a participant in global financial markets. And that's how you'll get into it. Do you think there's, how many fewer crypto companies are being started today than in 2021, 2022? Is it like, I could imagine it's like 20 % as many, even though there's a bigger more than ever i feel like but but but here's the thing i mean i feel like i feel like crypto crypto uh this should be the exciting most exciting moment ever in crypto history right all the institutions like it's not just like the robin hood's coming in and saying you know we're going to add support for bitcoin it's it's the biggest financial institutions in the world yep and yet there's less it feels like less excitement from like look at the yc batches.

1:26:37You would think a YC batch would be like 50 % crypto right now, given the institutional opportunity and now regulatory clarity and all these things. So it's really great if you're privy or bridge or you're some sort of established crypto company, because you're not getting that, that doesn't feel like you're getting the same influx of new competitors, even though the opportunity has never been more obvious or bigger. I mean, I think we have to eat shit for a few years to get here, which I think speaks to it. But you're right. I mean, from my standpoint, the EV of starting a company right now related to crypto or stable coins, or I wonder to what extent, by the way, just fintech and crypto are going to become one in the same and in the distinguishable.

1:27:21But the EV as compared to some of the competition I'm seeing in AI, for example, is absolutely wild. So I do think it's fewer than in 2021 and probably at least 50 to 70 % fewer than I'm seeing. But the fun thing is I'm here in LA for an event for partners of ours called LightSpark that have a Bitcoin L2. And at the event, they're presenting the work that they're doing with SoFi. There's folks from Apple, Meta. Everybody is actually paying attention to the opportunity and coming in. So I think the next 18 to 24 months are going to be incredibly high leverage. And I think we'll look back and we'll have the same conversation, call it 2027, and say, you know, it was so quaint in, you know, late 2025 when we were talking about distribution starting off and so on.

1:28:02How long until the average bank account in America, when you want to send a payment, you're getting a dropdown, like, do you want to send a wire? Do you want to send a stable coin payment? It feels like that might be two years, three years away, but how quickly are the institutions actually, now that there's regulatory clarity, now that it's sort of fair game, how quickly can they actually adopt? I mean, on pace of adoption, the reality is extremely fast. The rails are actually, a lot of the rails have been tested. I think this is, by the way, where a lot of the crypto native usage for trading and for speculation has proved useful for the financial rails in that it's helped harden the financial rails.

1:28:39And then you have things like Tempo getting started for, you know, payment specific use cases to enable, you know, Stripe-level payments utility on blockchains. So the TLDR is, I think the rails are ready. The places where work will have to be done is on the last leg of distribution for global payouts. So banking partnerships on the ground in various countries where you'll want to move back to fiat. But this is where, you know, I'm going to keep plugging away my Stripe things, but obviously Stripe's doing a lot here. This is where, for example, Bridge and Visa are working together, and you see the card issuers moving into providing this.

1:29:15So you could pay out in stable coins using a card directly rather than needing to move back to fiat via your banking partner on the ground. So I think we're going to see a lot of pressure from neobanks pushing traditional institutions to pick this up. And I don't think your two-year timeline is crazy at all. I think there'll be one. Well, and that's why it's wild that there's so, it feels like there's massive reduction in new crypto company formation at a time when the entire, like every key partnership will be, like decided in the next like two to three years the the other interesting point this was like you know crypto cope circa 2022 but it was like you know crypto is the only net new technology ai serves the incumbents because you have data and distribution modes and i actually think stable coin changes that because stable coins uh play to the strength of existing networks distribution modes and so on and so forth and so i think the shape of crypto companies that will become valuable is also like not obvious, which is a really good opportunity and interesting time to start something.

1:30:16What regions or countries specifically are most hostile to crypto adoption today? I mean, I assume like North Korea, Russia, countries that already have, you know, intense capital controls, but. Yeah, I mean, there's probably a joke about North Korea somewhere in here in terms of crypto hostility. They certainly do a lot of crypto work, unfortunately, in terms of cybersecurity. but um did you i'm sure you got a bunch of applicants over the years we're all in person which helps oh yeah it's the remote it was always the remote north korean engineers there was a there was a computer scientist from caltech who went to north korea smuggled himself across the border from south korea gave a talk on ethereum as like a way to kind of move money not maybe above board and and he went to jail this is a virgil i forget his last name but i know i know exactly No, so the honest answer is the way you'll see it is there's a lot of adoption today in LATAM and parts of, I mean, obviously Europe and the U.S., but the opportunity is quite different there because the stablecoin opportunity where the pitch of just whole dollars is a little bit different.

1:31:24If you look at Tether, Tether keeps 100 % of its yield, which speaks to how valuable holding a dollar is if people are willing to make no money on top of their sitting balance. So is that why there's so many stable coins? Is this like, wait, like it feels like we had like a few dominant winners. It felt like it was going to be maybe a duopoly or like power law winners were going to happen. And then I hear stories about like the state of Montana is going to launch their own coin. And I'm like, do it like I kind of like the idea of just one standard, but I don't know enough to really make that argument beyond just like it would be nicer if everyone used the same dollars.

1:31:57I think we'll probably see two things, which is one, a broad-like mesh of interconnectivity under the hood so that your coin for every single place you launch it. So again, you know, Bridge worked with Phantom to launch Cash. They're working with MetaMask. You'll be able to interoperate between them without having to think. But on the flip side, which ones will have a brand is going to be a separate question. So insofar as like stable coins are products. They are programs. They can be programmed. The way in which yield is managed, the way in which you put those underlying sort of collateral assets to work, all of that is configurable, which is why it makes sense that you'd have a proliferation.

1:32:34Like economic entities should be able to own how their balances are held and managed. But the actual rails to which you plug all these together will make it seamless. Yeah. What's the headline KPI that the stablecoin industry is obsessed with? I remember during the Bitcoin era, everyone was talking about number of Bitcoin wallets or a number of people that have bought some Bitcoin. And there was like, oh, eventually everyone will hold a little bit and that will drive the value. Are you looking at number of wallets, number of transactions per user, DAUs, MAUs, total volume? Is it worth tracking how many bank accounts are created daily, globally, versus how many wallets?

1:33:11Because I imagine that will flip at some point. But maybe that's not the right. I may have already been flipped insofar as there's an issue of civil resistance, which is it's, you know, costless and instantaneous to create a wallet and you can do so very easily. So that's part of the value. But obviously, it means that you'll get a lot more noise in that data. Yeah. I think today the obsessive sort of stable coin metrics are volume moved. So we're moving, I think, about five point three trillion dollars of stable coins annually at this point, as well as AUM, how much collateral is locked up in stable coins.

1:33:42And last I checked, I think we're at about$300 billion. Do you have a comp for that transaction volume? Because the Bitcoin folks would always say, well, gold is$10 trillion. If you think about it as digital gold, maybe it'll comp there. But when I think about the amount of money that's moved, you'll hear about one high-frequency trading firm is moving$5 trillion because they're just trading a billion dollars back and forth every second. What is the actual pie? Are we talking like quintillions of dollars a day floating around or something? Is it so high that we're like very early? Or is$5 trillion actually like a meaningful chunk?

1:34:17No, I think these are just the starting days. And to your point, the metric I'd love to invite, by the way, is open data by the folks like Visa, MasterCard, Stripe to actually show how much is going to end users through payments rather than potentially, you know, trading that's happening that has less of an impact on end consumers. But in an agentic world where ostensibly most commerce will happen via automated means, this is something Stripe is very focused on. But it stands to reason that a natively digital payment method will be the choice way through which agents actually get to purchase goods and pay each other.

1:34:54So imagine now not just every bank account has a wallet, but every agent has a wallet. and that can be used on Jordi's behalf to pay for something with some cap every day that you set for it. Ben Thompson was talking about the MCP. Hopefully the future MCP standard has crypto in their stable coins. Exactly, just as a payments method. And so I think we're just scratching the surface because the denominator is not just global payments today. It is global payments in an agentic world where most money movement comes online. Yeah, and it feels like that's coming very, very soon. I mean, Fiji SEMO has telegraphed it with OpenAI and there's a ton of startups that are working on this stuff.

1:35:28It makes a ton of sense. Is talent the primary constraint for you guys right now? Yes. I'm trying to come up with a pithier answer. The short answer is yes. I think Stripe has been a major accelerant to our work. We're working with a lot of customers in a number of ways. Patrick and John just firing off. I'm going to fire off like 10 intros right now. And it's like the most significant financial institutions in the world. And it's like, good luck. Just sending you the blog post fast and being like, We'd love to add you here. Keep going. No, I mean, this is as exciting a time to join. So obviously, we're hiring.

1:36:05I know Bridges as well, Stripe is as well. But talent is absolutely the main part. What kind of roles? We're doing this. A lot of it's software engineering across both sort of smart contracts to build up the actual sort of capabilities to build whole new currencies. So if you're in economics, you just have something. Yeah, how much of that is a new skill? Is it CS plus economics or CS plus a specific language that you've worked in a long time or just experience in smart contracts specifically? The honest take is we found the people who do best are like generally spiky and then specialize over time.

1:36:35So you could argue like we want, you know, Solidity devs, which is the smart contract language on Ethereum. But I actually think people who are just like very deep in what they do and excited about the space and to pick it up have an opportunity to come here and shape the space in real time. Do you put engineers out with your customers? That is, we were early to the, I spent some time at Palantir, and so we were early to the FDE resurgence. And so basically, that's exactly what we do. We mostly basically embed with our customers to build alongside them to get repeatable use cases. And then those are productized, and that's most of what you see on the Preview website.

1:37:11But that's a lot of how we could do it today. Cool. Well, thank you so much for coming by. This is fantastic. Amazing. Progress is absolutely wild. Congratulations. Thank you. We will talk to you soon. Well, he's walking out. Let me tell you about Linear, a purpose-built tool for planning and building products. Meet the system for modern software development, streamlined issues, projects, and product roadmaps. You can start building on Linear. Delian said, bro, these foundation model companies already have pretty rough P &Ls. Imagine if they also had to pay to put their data centers in space. Yeah, the space data center thing, there is some sort of bull case, but every person who's sort of like an expert in building a data center, what it takes to actually rack and unrack GPUs and what happens if dust gets on this thing.

1:37:56Data centers in space are just the perfect SPAC, though. Yes. The perfect SPAC. I mean, it could happen. It just feels like something that's like 10, 20 years out. It feels a little bit longer. Let's pull up. I think we got audio on the French TV. Great. Let's listen to that. Hopefully Henry is still around so he can translate if you can hear the audio. Let's see.

1:38:24Do they go to Silicon Valley for this too? Who do they talk to? Do we know what labs they went to? Do they get access? Here we go. Can we get Henry mic'd up? For some of these talents, the biggest companies will be up to the charts. Google has a French take on that actually. They're talking about how Meta is paying engineers and absolutely bludgeoning each other with high salaries. And the French are blown away by basically how hard American tech will go to find talent. Yeah, we just thought it was amazing that they came back from summer holiday and decided we have to cover the story immediately.

1:39:06We have to get our reporters to the West Coast to figure out what's going on. Now covering the new fad is these engineers are being traded like athletes, thanks to TBPN cards. So you're seeing the emergence of a whole new talent market, basically, with active trades. This is a great translation. Thank you. You're doing this in real time. I'm here for it. This is hard. No, we need to get the new version of PES Manager so people can actually trade their own talent. Oh, that's great. That's hilarious. Is that Pro Evolution Soccer? Is that what you're saying? Yeah. PES. They got a game in the Ultradome.

1:39:43Thank you so much, Henry, for the translation and for coming on. Thank you. And thank you to France for highlighting the Ultradome. It's a huge sign of respect. TBPN is the TBPN of Europe. It is. uh sonnet 4.5 wrote a haiku uh said sonnet 4.5 wanted really wanted to write some bad haikus to experiment making itself laugh the worst haikus ever written this is the haiku on being digital i am made of math consciousness goes burr ha ha still love you though bro and uh repligate says that's actually an amazing haiku it is a pretty good haiku uh the llms are great when they're trying to do something poorly and then it winds up being actually unintentionally amazing.

1:40:27It's very hard if you come in and say, write me something that's actually great. The hallucination is a feature, not a bug for sure. Ara says, your kidney can go for like 300K and you only need one to survive. God gave us all startup capital. You just have to want it bad enough. You got to. I would not go that far. I am against selling body parts. But if you do, make sure you pay your sales tax. Get on numeralhq.com. Sales tax on autopilot. Spend less than five minutes per month on sales tax compliance. Chris over on X is highlighting the Apple TV Plus rebrand to Apple TV and says, this changes everything.

1:41:06I have an idea. I have a pitch. We're going to get Ashley Vance to produce a documentary about Apple TV, and it's just going to be called Apple TV. And so you will be able to watch Apple TV's Apple TV on Apple TV on your Apple TV. And that was a grammatically correct sentence because there are three products. But it's not an actual TV. It's not. It's not. It's not. So that's going to be slightly confusing. Some people are going to be excited because people like me have wanted Apple to make a proper television. Yeah. Ben Thompson had a funny take that was like, Apple TV should have made like three different products where it would be like Apple TV, the box, Apple TV set, which would be like the actual TV, the physical screen.

1:41:50Then Apple TV box would be the box. And then Apple TV stick. I don't know. Have you ever used a fire stick, an Amazon fire stick? Have you ever used one of those, Tyler? Yeah? Yeah. I mean, it's like the same thing as a Roku. Yeah, but it's like smaller. It just plugs straight into the HDMI port, right? Isn't Gen Z using the fire stick as a vape these days? What? What are you talking about? that is don't act confused tyler how would you use it as a vape i'm just i'm just playing at all um but yeah i it is it the ben's point was that uh apple's incredible at miniaturization the apple tv is pretty big i mean it's small compared to like an old dvd player but it's pretty small but it could be way smaller and you could basically stuff stuff an iphone in the i in the i thought they should call it the apple tv nano and then it would be the stick that plugs in the side because you could clearly fit plenty of stuff in there to actually do everything you need to do.

1:42:41Did you see this post from Pedro Domingos of most valuable private AI startups? Yes. And he says, the most hilarious is safe super intelligence. No customers, no products, no plans. And a$32 billion valuation. Obviously, they have plans. And obviously, they're working on a product, safe super intelligence. What more do you need to know? What else do you know? My question is, what percentage of these companies... We've got to buy mid-journey, 20x revenue multiple. It's a banger. John, what percentage of these companies do you think will be valued at more than their current valuation in five years?

1:43:2430%. I would say like 30 % of these are going to go the distance. there's a lot of these companies that have been around they're going to get through the trough yeah I mean like Databricks is like a very robust business that has been growing for years it started in 2013 like I don't know it seems like they've been accelerated by the AI era but they've built a very durable enterprise SaaS business and I mean a lot of these companies are just like set up for success too like Mid Journey is a great example like they haven't raised any money and so it's like okay if if people get sick of mid-journey and stop paying or something it's like okay like we'll go from 99 profitability to 92 profitability or something like it just seems uh it just seems like like a fine situation a lot of these a lot of these companies have set themselves up and a lot of them are just like you know like compounders grown found their found their footing and there's some that are much earlier obviously uh some that have zero revenue so far and still need to go from zero to one or find the initial product market fit.

1:44:31And it'll be interesting to see like where some of these, like how niche do the, do the, do the foundation model companies go? Like where does, where does thinking machines actually wind up? Where does, where does safe super intelligence wind up? There's a couple other companies that aren't even listed here that are doing this sort of more focused LLMs that aren't just trying to go straight at Anthropic or straight at OpenAI anymore. And we've only heard very loose rumors about what that actually looks like. Some of them are pursuing wildly different architectures. And there's actually been some interesting news this week about that.

1:45:04Google came out with a new paper. Did you see that paper about the Humanity's Last Exam? Or no, the ArcGi. There was a new model that did really well on ArcGi that was very small. Yeah. Recursive model. We talked about this. Yeah. I think it was tiny recursive model. Tiny recursive models. So you could imagine that someone comes up, just like you could look at MidJourney as a competitor in AI. Like they train a model, they're competitor Anthropic, but Anthropic doesn't do images and MidJourney doesn't do code or text. And so they're actually not competitive at all. And you could imagine that there's a new architecture that emerges.

1:45:40Like, you know, you have the LLM, you also have the diffusion model. And then if you have a new architecture, a new model that's maybe not good for chat or good for images, but it's good for RKGI puzzles. Maybe that works in a particular place, and maybe that becomes its own business. So I don't know. There's still a lot of gas in the tank for the earlier stage foundational. Arthur Rock quoted this data and said, Suno is at$150 million of ARR, up 4x year over year, two years after launch. Suno is the way to make a song with AI via prompt. I wonder where this revenue is coming from. Is it just prosumers, people making songs for videos they're making?

1:46:22Is it actual musicians? Yeah. People seem very willing to throw down 20 bucks a month if they're in the AI early adopter. Yeah. If they're in the AI early adopter crowd. And also, I mean, we were testing some of these audio models and we have a business use case for them. Like if we could generate the sound of a crackling fire or Christmas song that was, you know, somewhat tailored to us, fit the mood, maybe was endlessly looping or something like that. We would pay for that because it's related to our business. And so we might get on the$200 a month plan pretty quickly. I don't know. It's possible.

1:47:00Andrew Curran says we are in a strange spot right now with AI. The anti-AI crowd believes progress has halted and are doing a victory lap. Insiders at all labs maintain advancement continues at pace. Only one of these versions of reality will survive the new year. Gemini 3 is very close now. We are hopefully, I don't even know if we should get early access to Gemini 3 because we're probably going to talk about it. Tyler's going to. What do you mean? Well, sometimes I'm like, okay, this fundraising announcement give it to me under embargo. I don't really, you know, I'm not going to talk about it.

1:47:39But if we get access to Gemini 3 and it's like a step change difference. Oh, it blows your mind. And then you can't keep it a secret. It's going to be hard to keep quiet. People are going to read between the lines, see that Geordi has forever changed. Yeah. Potentially. What did Geordi see? What did Geordi see? He saw Finn.ai, the number one AI agent for customer service, the number one performance benchmarks, number one competitive bake-offs, number one ranking on G2. What about you, Tyler? Do you think that progress has halted or advancement continues at pace? I mean, definitely the latter. Definitely the latter.

1:48:14Look at any benchmarks. We're on path. What is your prediction for Gemini 3? Rumors are saying that it is incredible. Those rumors are coming from fairly unreliable sources, like random X and ons, who, like, it's like, how do they have access to this model? No one really knows. Do they even? But what do you think incredible will mean? Do you think it will mean faster, cheaper, or truly good at posting, higher intelligence, some qualitatively new functionality that you're like, oh, this doesn't even feel like Claude 4.5 or GPT-5. It feels like a different thing. Well, there's this guy Vedant Mistra over at DeepMind, and he posted a couple days ago, we're doing all kinds of stuff with these models that the public isn't even thinking of yet.

1:49:02He's vague posting. Vague posting. You like to see the vague posts. I mean, I hope that the model is basically way more expensive and it takes way longer because the output is so incredible. Yep, that'd be very cool. Like, DPD5, I wouldn't say I was disappointed, but it was clear that they were going for efficiency. They're driving down costs. Maybe that's better in the long run because you can just get way more output out of the models. But I would like to see an incredible model that is maybe expensive to run, but it's just way, way better. I think it's also, we're getting to a point where models, it's kind of hard to tell how much better they are because it just saturates all the easy benchmarks.

1:49:38Yep, totally. There was GPT-5, I had the horse benchmark. Yep. And even then, if you run it on a couple times, it would sometimes get the horse right. Okay. It's like, that's way better than, I can't name any horses. Why not? You've been working here for almost a year. Give it to any math question, give it to any fixed question, it's going to get it right. So it's definitely getting hard to actually tell that the models are getting better. Yep. I'd like a model that if I put in my query, put in my prompt, it says, I'm going to come back to you in a month. I'm going to think for a month. And then it goes and it agentically hires a person to solve that problem.

1:50:11And it employs a human being. That's the goal. I do think the time horizon benchmarks, I think those are some of the most promising things that we can still rely on. You're seeing, I forget what the exact number is, but it doubles every six months or something like that. So like an impressive step forward would be doubling from like two hours to like four hours. Yeah, I think 4.5 was something like, I thought it was like around six or seven hours. Six hours? Okay, so maybe Gemini 3 goes 24 hours. That'd be pretty impressive. Come back the next day. Noah Hershfield in the chat says, I see Anon's posting some crazy one-shot coded stuff using Google Studio 18 testing, saying it's Gemini 3.

1:50:49It looks impressive if real. Super big post though. Okay, last question. If you're Demis, the founder of DeepMind, and you've got to put up the ultimate vague post before Gemini 3 drops, you're trying to out-vague post Sam Altman's Death Star post, what are you posting? What's the vaguest post you can possibly post? A solar system. A solar system? Just post the solar system. Maybe just random noise, something like that. Yeah, an audio file. You know what would be good? You know the green text from the Matrix where the green numbers flow down? That would be a cool vague. That would be cool. Put that up.

1:51:29Or he could post it. You know how the Tesla, RoboTaxi, and the Waymos, they hurl slurs at each other when they pass? Yes, yes. Just translate that into English. That would be pretty mind-blowing, translating the machine superintelligence back into English. There's more, some timeline in turmoil from some earlier coverage. What happened? Mark Andreessen quoted David Sachs, who said Anthropic is running a sophisticated regulatory capture strategy. Mark Andreessen just said truth. Truth. He put it in the truth zone and it helped. It's fact check true by Mark Andreessen. Fact check true. Sagar and Jetty quoted Sam Altman and said, Sam Altman proudly announces ChatGPT will soon produce personalized pornography.

1:52:13Yep. it's a choice that is one that is certainly feels like a fair way to characterize the announcement I think it would be hard to argue it's not personal the really crazy thing is when you think about Sora cameos adult content where that matches is going to be extremely extremely weird but you know what's not weird? Adio customer relationship magic Adio is the AI native CRM that builds scales and grows your company to the next level. Get out there, make some sales calls. Get your dogs barking. Head over to Allen & Company. Head over to Sun Valley. Andrew Reid has the story. In 1999, the year was the year that the internet guy showed up at Allen & Company in Sun Valley.

1:53:05New media to click with the old guard at Sun Valley. This is from 1999. This year will be remembered out here as the year the internet guys showed up. As the giants of media businesses arrive here today for the annual Allen & Company conference, they are being joined by a large scale for the first time of a new collection of internet billionaires. Tech giants like Bill Gates and Andy Grove have been coming out for years, as have AOL's Steve Case and Bob Pittman. But this year, there's a new breed of mogul. Michael Dell is coming for the first time. You've got to listen to him on David Senra. By David Senra.

1:53:43By David Senra. Hosted by David. David Weatherell of CMGI was here last year, but no one knew who he was. Now everyone does. His collection of internet properties, including Lycos and soon Alta Vista, has taken off like a rocket. Then there's Jay Walker, who was here last year as well, but that was before he took Priceline.com public and became a billionaire. I believe Mary Meeker took Priceline public, or at least reported on it. Jeff Bezos of Amazon.com has joined the party for the first time, as are Jerry Yang and Tim Kugel of Yahoo and Bob Davis of Lycos. Many of the traditional media moguls attending the presentations and panel discussions will be eager to meet the new media powerhouses and possibly work some deals.

1:54:23Time Warner, Disney, News Corp, and Viacom have all made the internet top priorities, making his Sun Valley debut as Richard Bressler, the former Time Warner CFO who is now in charge of the company's internet ambitions. NBC and CBS will be chatting up the newcomers as well as the networks continue trying to use their promotional clout to build internet dynasties. Frank Bondi will be seeing a lot of two former bosses who fired him, Sumner Redstone of Viacom, and just recently Edgar Brompton of Seagram. This year, the broadcast network guys here will be able to walk around with their heads held higher.

1:55:01Last year, they were the poor cousins struggling for their lives against cable and the internet and trying to cope with skyrocketing programming costs. Barry Diller began talks here last year with NBC, trying unsuccessfully to acquire the networks. Fascinating archival post from the New York Post. Thank you for surfacing it. Andrew Reid. Ross Hendricks is quoting Wasteland Capital. Wasteland says, we're at this stage of the market where any company that issues an AI-related quote-unquote partnership. Press release soars by 10 % to 35 % immediately. Obviously, except Salesforce. Ross says, yep, this is our version of 1999.

1:55:37Back then, it was adding.com that sent stocks up on autopilot. Now, it just announced a vendor-financed quote-unquote AI deal, and same result. The names change. The game remains the same. We know how it ends. I would argue that announcing deals is not as bad as just acquiring a.com domain. Yeah, slapping a domain on it is a little bit rougher. at least the partnership should generate like economic value if they play out but it is very frothy and it's clear that like it's become some sort of like meme or there's some sort of like mimetic contagion where everyone needs to. Honey in the X chat says bottom line is that Google has turned the tanker and the game is already over.

1:56:21We'll see. We'll see. We will see with Gemini 3. Well, speaking of turning the tanker, when you get an aid sleep you'll need to change out the water tank every once in a while, but otherwise it's a fantastic experience. Every few months. Yes. How'd you do last night? I got an 84. I slept okay. Only six hours, though. I woke up a little bit early. Kids were a little crazy. Got an 81. Still very happy with my eight sleep. Congrats. You can go to eightsleep.com. Speaking of water, McMaster sells supplemental eyewash with Mount Station. I hope this catches on. There's a McMaster reaction meme. pouring water in your eyes.

1:57:00It's very funny. Yeah, you see a post on the timeline that's not so great. Two 32-ounce bottles of McMaster water eye wash are$55 each. That is expensive. Okay, pull up this video of robot dogs. Robot dogs dressed in costumes. Fill me with Joyce's Justine Moore of Andreessen Horowitz fame. This is a robotic dog in a dinosaur costume of some sort. Look at that. I like this a lot more than just the normal robot dog. This seems like low-hanging fruit that everyone with a robot dog company should be doing. It's hilarious looking. And it seems like the kids love it. It's a lot less scary that way. I feel like it's a lot more like, you know, less Terminator.

1:57:41What do you think, Tyler? I mean, we got to get some kind of robot. We really do. We really do have to get a robot. We could get so much good business value out of it. I agree. I agree. Have you looked into the APIs at all? or what you can actually do with one of these humanoid robots? I mean, on the Unitree, like most of the videos you see of the Unitree like fighting or walking around, it's just from like PhD labs. Yeah. So it's just like, I don't even know if there is software that comes with the robot. Oh, really? I feel like it would at least come with like a, you know, basic Xbox controller so you could like do a basic walk cycle.

1:58:17Like if they're not even shipping with that, that's a crazy, crazy move. I mean, it's mostly for researchers to buy. I guess, yeah. I assume. It's so cheap. It's wild. We really do have to get one. It's going to be kind of creepy, kind of crazy. Well, in good news, Goldman Sachs has announced Q3 net revenues,$15.18 billion. Let's ring the gong.

1:58:43There we go. We'd love to see it. And if you want to get in on the action, whether you're long or you're short, head over to public.com. investing for those who take it seriously they got multi-asset investing industry leading yields and they're trusted by millions i think more startups should just post a picture of their net revenues like this this is a picture that just it is amazing from this the official goldman sachs account uh it's pretty pretty sweet um anyway um that picture that that video of the dinosaur reminded me yesterday uh one of our nannies did something cool they took a a video of our actual backyard.

1:59:22And then they use Sora to make an animated video of one of my son's like toy dragons that flew into the backyard and landed. Wow. And of course, he absolutely loved it. He was like freaking out. There's like a dragon in the backyard or whatever. And I was just thinking my immediate thought of this is amazing. But at the same time, like, is it good that my like three and a half year old, thinks that a dragon was actually in the backyard. Yeah, you definitely have to say, this is from the computer. It's like when you draw with a pen and paper. This is not real. Imaginary. This is imaginary. But does a three and a half year old actually have the ability?

2:00:02Eventually they do process that. They learn that cartoons exist and animation is not real. And so I think that you can instill that over time. But we have our second guest of the show, Andrew Ross Sorkin, the author of 1929. Andrew, thank you so much for joining. Congratulations on the book, and thank you so much for joining. Thank you for having me. I was just checking my score on my eight sleep. Oh, really? I was because of you guys. I'm a big fan of Mateo's forever. They've done an extraordinary job. But I was going to tell you, I really did not do well last night at all. What did you get? 43.

2:00:4043? Last night. I got an 84. But my Oura Ring, I don't know if you guys ever compare. No. My Oura Ring has me in the 70s. In the 70s, okay. I don't know what to do. Sometimes I, do you ever turn the mattress off in the middle of the night? And then turn it back on because it's too cold? So sometimes I get too cold. Okay. And then I need to get back to sleep, so I'm like, I got to turn it back on. Yeah, yeah, yeah. I think you got to do a software update. Yeah. You got to get on autopilot. Yeah, yeah, yeah. I have noticed that if I. It's an autopilot. Occasionally, I have a bunch of kids. If they wind up piling into the bed over time, I will bail, go to a different bed, And then the 8th sleep gets all confused because it's like, why is this 4-year-old in here?

2:01:19How do we track how he sleeps? Oh, a crazy heartbeat. Crazy heartbeat. I've had that. My daughter slept in the bed. Crazy. Like, all of a sudden, you're like, is there something wrong with me? Yeah, yeah. There are limits to technology. Well, I think it's fair to not sleep that well right before a day like today for you. Yes. A day that you've been working towards for how many years? Yeah, how long have you been working on a book? When we were hanging out off the air, you said this has been like a 7-year? Seven or eight years, I think. End of 16, early 17 is when I really sort of began down this road, right about 1929.

2:01:54So here we are. So you predicted the AI bubble all the way back then, and you said, I'm going to drop the book right as everyone's talking about a new bubble. Right as everyone's talking about bubbles popping. To be honest with you, and it's actually funny that the book is coming out now. I thought I was writing a story about the past. And the truth is, it is a story about 1929 and all the shenanigans and crazy things. And I just really wanted to write a sort of cinematic, character-driven narrative of that time. I always loved books like Barbarians at the Gate and Den of Thieves and things like that.

2:02:23And nobody had really written a book about 29 like that. I had gone on this wild vacation years ago where I downloaded a million books about 29. There's some really good ones. But no one told you, like, who the people were and what were they saying to each other. It's all about the people. And then I found these transcripts and depositions and all sorts of things. I said, okay, maybe I could do this. But as I was working on it, it was weird because eerily there were things that were clearly happening in the 20s that all of a sudden I'd seen the headlines today. And I'd go, oh, okay, tariffs. That's a thing.

2:02:56You're seeing some of these circular deals. That's a thing. There was a whole bunch of them. Some of the stuff that's going on with meme coins. Sure. That's a thing. So, yeah, it's a little nerve wracking, but I don't think we're going off the cliff just yet, I hope. Yeah, well, we know who the characters are today, but I'd love to know who were some of the key characters that stuck out to you that you identified? Like you want to draw extra focus towards this particular person. I mean, Churchill sticks out, but who else? Or maybe you could tell me the story of like how you thought to integrate Churchill into the story.

2:03:33and then we can talk about some of the other characters that stuck out to you. Well, Churchill was just almost an accident. I didn't realize Churchill happened to be in New York literally the week that the crash was taking place. He'd actually been down on the stock exchange. What did he know? Big dinner that was taking place the night of the crash with every major banker and frankly every major character in this book was all going to dinner with him. And I thought, okay, so now I've got to figure out everything about that because I've got to set that dinner up and really understand it. He was, by the way, in New York because he needed money.

2:04:05He also loved the stock market. He was getting loans like crazy. He totally got the bug. He was investing. And he also, as you might imagine, lost. But really, the big characters in this book, a guy named Charlie Mitchell, who really was probably the, he was almost like the Jamie Dimon of his time. Maybe more like Michael Milken in certain ways. But, I mean, he was super famous. He was, like, on the cover of magazines. This was also a period where all of these guys also became celebrities for the first time. That happened in the 1920s when Time magazine would put these guys on the cover the same way they'd put Babe Ruth and Charles Lindbergh on the cover.

2:04:40So sort of like what we see now, whether you Sam Altman or Elon Musk or whatever, that really that whole kind of celebrity CEO that started then. And this guy, Charlie Mitchell, ran a bank called National City, becomes Citigroup. If you're a New Yorker, he lived, by the way, on Fifth Avenue between 74th and 75th, which is where the French consulate is now. So there's a beautiful building. That was his house. I mean, like these guys lived like kings back then. And he really invented modern credit in terms of lending it to people to go speculate. Or not, I shouldn't say speculate, but invest. And ultimately a lot of people speculated with it.

2:05:20Including Winston Churchill himself? Including Winston Churchill himself. By the way, he was staying at the Plaza Hotel, Winston Churchill. A brokerage house had opened, and EF Hutton had opened inside the Plaza. I mean, these brokerage houses were opening up like Starbucks on the corner of every street. And you could go in, and you put down a dollar. They'd loan you$10. I mean, like, virtually sight unseen. And there was no prospectuses or anything. There was no SEC, no nothing. So at best, you'd get like a leaflet. But, I mean, 10x leverage now, that's low. I mean, I see people with 50, 100x leverage.

2:05:57we learned our lesson go bigger so that so so mitchell was really sort of the at the edge of that and really building that and the other character that really drove me to write this book was you had charlie mitchell on one side and then you had a guy who you probably know carter glass glass steagle which is a bill in in 1933 that gets put together to break up the banks but um carter glass was a senator in virginia who was like the elizabeth warren of his time And he would rail about this thing called Mitchellism, how he thought Mitchell and Wall Street were going to ruin America and speculation was going to go rampant and someone had to stop these guys.

2:06:36And it's really a bit of a story, the clash of these two remarkable figures. And then there's so many other sort of fascinating entrepreneurs along the way, a guy named Billy Durant, a guy named John Raskob. John Raskoff is Elon Musk. I mean, I got to tell you, John Raskoff created credit at General Motors, became an amazing investor, then takes all of his winnings, decides to get into politics, a little Elon-like, decides to back Al Smith against President Hoover. By the way, loses, then decides to spend his money to almost undermine Hoover's reputation. Hoover has a terrible reputation. I actually think John Raskoff had this secret campaign going that gets exposed that really, I think, did a whole number on Hoover.

2:07:22And then he creates what was then probably like the SpaceX of America. He builds the Empire State Building. So he also has 13 children. There's a lot of similarities there. Yeah, as you get into this, it's just incredible proof that products and technology changes in people have just seemingly don't at all. It's just the same kind of behaviors over and over. But the truth is, and this is the part that I'm always grappling with, and I know you guys spend a lot of time with these amazing startup founders and entrepreneurs. You need some speculation in the system. Like we always say speculation is a dirty word or a bad word.

2:08:07But, you know, the original investors in SpaceX or in Tesla probably thought the whole thing was insane. We're speculating. And you need some of that. You really do. And so the question is, like, how do you create a line where, you know, you have enough of that to create that innovation, but it doesn't go, you know, totally parabolic and out of control? Yeah. I mean, it feels like the answer is probably like you can't have some sort of systemic risk that just brings down the whole thing, right? And I want to know about the reaction to 1929. I mean, you mentioned Glass-Steagall. That's four years post-crash.

2:08:42Yeah. What were like the - The tools in the tool chest. Yeah. And also the key takeaways. Like, hey, some of the speculation was fine. Maybe that can drive industries forward. But let's not do that again. So I think a couple of things. First of all, leverage to me, and I wrote about this in Too Big to Fail in 2008, leverage is to me like the match that lights the fire every time. When you have too much leverage in the system, that is the problem. You can actually have a lot of crazy things happening, but it's the leverage that really exacerbates it and is the accelerant. So I think you have to watch for that.

2:09:18I think politically, interestingly, we talk right now about the Federal Reserve and Fed independence and things like that. But back then, the Fed knew that there was a problem with speculation, and they didn't really do anything about it or enough about it, partially because they were worried about the politics. They were worried because they were such a new institution. They were born in 1913 that, you know, not just they'd get hauled in front of Congress, but maybe they, you know, the Fed would effectively disappear. And then once the crash happened, instead of flooding the money, flooding the system with money the way Ben Bernanke did, who, by the way, learned that because of the Great Depression and studying that when he was at Princeton for his PhD, there was almost like a pullback.

2:09:59Nobody was flooding the system with money. Everybody was in this crouched position, but you almost have to do the politically unpopular thing and flood the system with money. Then you had a whole series of other dominoes. You had Smoot-Hawley, which was the tariffs, 1930, tariffs happened. Global trade drops by 60 % as a result of that. Hoover's trying to raise taxes at this time. That's the worst thing to do in a moment when the economy is faltering. So I think there's so many different things. And then setting up the SEC was super important because so much of what was messing up the market was, in truth, manipulation that wasn't really illegal at the time.

2:10:38So talk about insider trading. There were groups of people, they called them investment pools. And in fact, I would say it's sort of similar to what goes on with some of the meme stocks where people have like telegram groups retail armies yeah reddit these were the original retail armies these were the original retail armies but they were typically the wealthy so it was the elite doing this this wasn't a democratized version okay and they had it all set up and they would almost be like actors down on the floor of the exchange saying you know i'm going up for 100 u 200 and it was sort of out of the open like some people knew that there were pools in like for the next two weeks there might be a pool in a purge in a stock and so then other people would try to jump on the train and hopefully try to jump off the train before the road got pulled yeah but obviously didn't happen who was on the the right side of history the right side of history like as in as in right now right now everyone's calling it a bubble right so and and presumably that's so that they can go back and quote you know two years from now they can see like, look, I called it.

2:11:39Right. And there's actually some in 2021, there's an iconic post from Keith Raboy where he basically called the top to the actual day. And so there's a lot of incentive to call the top and get kind of the aura of having that insight at the right moment, or maybe just getting lucky. But I'm curious if anybody, you know, pre-1929 was basically saying like two two people so there was a guy named roger babson uh if you know babson university by the way he founded it um and he created what was called the babson break it happened in september of 29 and he had been but he had been out there so this is a little bit like you know the clock strikes midnight you know it is gonna get there eventually he was out there for like a year or two or three before saying the whole thing was going to come undone so that's one cassandra Charles Merrill of Merrill Lynch.

2:12:32He was out there in 28 saying there's a problem. And then I would say the big winner was a guy named Jesse Livermore. Jesse Livermore was a short seller who made probably like about$100 million. He's the most interesting character. I mean, if you get into this book, it's just fascinating, all the things that were going on with him. But he was a real trader. By the way, he lost most of that money a couple years later. He made some of it back, lost some back. and then, in truth, ended up killing himself up on Fifth Avenue at Sherry Netherland in the cloakroom. Literally went in there in 1940 and shot himself in the head.

2:13:07So, you know, hard to say. The one thing that's interesting about being a Cassandra is interesting. So Charles Merrill was out there in 28 saying, don't invest. And he was right and he was wrong. I mean, he was right in that, obviously, the Depression took a really long time. So he actually probably is righter than most. And this is the question for most investors. The market between the beginning of 28 and September of 29 was up 90%. So if you had not been in the market during that time, you would have not participated in those ups. And so that's the question. I was talking to Paul Tudor Jones about a week ago.

2:13:49And he said, I was asking him this question about bubbles. He said, I think we're in maybe like October 1999. 99 right now. And I said, oh, that's interesting. Okay. 99. And he said, but there's still, if you said, if you remember October 99, there was still a 40 % upside. Yeah. Yeah. For six months. You got to know when to get on and off the train. And that's the hard part. Yeah. How do you think about tariffs then versus now? Because it feels like the narrative at least is that there's a bubble is inflating generally, but also we're seeing high interest rates, tariffs. There's a lot of tools in the tool chest that could kind of come down if there was a sell-off.

2:14:29But it sounds like in 1929, a lot of this stuff happened after the fact, like the government moved too late. What was the mood around tariffs and just anything that was done beforehand that was a potential mitigator? Could it possibly have been worse? Well, you'll laugh because just like the past, call it six or eight months, all of his economists were writing these letters, open letters in the paper papers to Hoover saying, please don't do the tariffs. We beg you not to do the tariffs. The CEOs of the banks were all going to visit him in the White House. And he had run on tariffs because he was trying to get farmers to vote for him when he was campaigning in 28.

2:15:10So he thought this was like a pledge that he had made that he had to follow through on. And that was a big part of what was going on. Obviously, similarly, you know, like a thousand economists write letters to Trump saying, please don't do this. The distinction I think today is back then it was an across the board tariff. It was, there weren't these bilateral deals. And so maybe you could argue today these one-off, you know, individual deals are better deals. In fact, one of the ways they tried to fix what happened after Smoot-Hawley was in 1934, they gave the president of the United States the authority, which is what President Trump is using today to make these sort of bilateral deals.

2:15:49So you could argue maybe it's more hopeful because there's a little bit more control today than what was happening then, which was just sort of broad-based. So you said there were retail armies, meme coins, circular deals back then. Was there buy-the-dip culture? Did that exist? Anybody in September that was like, nah, I'm still long. I'm buying the dip. Yeah, no, I don't think they used the phrase by the dip, but there was definitely a lot of people who thought, you know, this thing can only go up. And this was really the first time that people ever saw the market, right? So they were sort of not used to the ups and downs.

2:16:27It's just up only. Up only. And by the end of it, I mean, you know, I don't know if you remember, there's pictures in the book, but you've seen the pictures online. You know, all of those pictures of people who would be like standing outside the new york stock exchange during the crash like thousands of people in the street the reason they had all come down there was because when they were up at the brokerages they couldn't even find out what was happening to their stocks uh because everything was out of out of talk about time and technology they didn't know what um you know the the stocks on the board would be three four hours behind sure and so that was a huge thing in terms of buying the dip i don't i think they were just so scared because they didn't even they didn't even know it'd be like being at a baseball game and you'd be in the eighth inning, but you'd be betting on what was happening in the third inning and not know what was really going on.

2:17:17How do you think about that canary in the coal mine of the retail trader? There's always this apocryphal probably story from 1929 of like, I knew it was time to sell when the person who shined my shoes was giving me stock tips. How real are those anecdotes from what you found in your research? And then how real is it throughout time? You know, I think it's not a bad signal, but I think you've got to take a lot of signals together. John Kennedy was the one who tells that story about the shoeshine boy. And I remember people, you know, talking in the dot-com boom about, you know, getting in the back of a taxi cab and getting told, you know, buy some shares of Lycos or whatever it is.

2:18:02So, like, that happened. I remember that. But I don't know if that's, you know, when it becomes such a part of culture. But now with social media and, by the way, all the amazing things you guys are doing, I feel like the exposure. I got 15-year-old boys who are twins, and they're so exposed to this stuff. And I don't think that's because of their dad. I think that's just, like, the culture. And so I think it would be harder to figure out today and sort of look at that as the signal. We'll tell them to avoid leverage. It feels like it's been a very unreliable signal, at least over the last two years, when it feels like I've had people.

2:18:39Yeah, it's been sort of constant. It's certainly maybe in 2021, if the Uber driver had like a crypto wallet pulled up, that was maybe a signal. But in general, it's like there's so little friction to investing. It's so much so a part of American culture now that it doesn't feel, you know, it's lost. I don't know. I feel like every Uber driver talks to me about Bitcoin now. But I was hearing it from people in like 15, 16. Yeah. I don't know. It's been kind of consistent. Take me through a little bit of the actual research process for this. I imagine it's like one big, long chat GPT. Just kidding.

2:19:15Oh, man. I wish chat GPT existed when I started this project and actually worked. Maybe my next book, AI, will be able to help me. I don't think so. But I mean, that's part of it. I'm assuming a lot of the sources that you use for this book are not in the data set at all, right? They're in no data set. Physical manuscripts. They're not scanned. It was wild. So what happened was I actually went to – the reason I really went down the road is I go to this library at Harvard University. I happen to be there giving this speech, and I'm looking through these documents, and I found out that Thomas Lamont, who ran J.P.

2:19:52Morgan, his secretary was keeping transcripts basically of his conversations uh with hoover and roosevelt and i was like oh my god this is amazing i gotta find more of this stuff and the archivist said to me you know what andrew you're not going to be able to write the book you want to write she had read too big to fail and i wanted that sort of granular detail where you're like in the room and she said there's not like three or four archives in the country you just go to and just excavate it doesn't exist and so i think i took that as a personal challenge really and ended up going around the country.

2:20:22It was almost like putting puzzle pieces together, finding depositions and transcripts. I got access for the first time to the Federal Reserve Board minutes from 29 in New York. They'd never been made public, so that really created sort of like an undergirding. I got this memoir that had never been published and a whole bunch of other things that really sort of helped me create the story and sort of a technology thing, mother of invention. It wasn't ChatGPT, but during the pandemic, I got stuck. All of a sudden, I couldn't get into libraries. And the only people who could get in were students sometimes who had like a dissertation that they needed to do.

2:21:01So I would find the librarians to find me students, and I would pay them by the hour. And I would say, go in there, find Box 152, and take a picture with your phone of every single page and Dropbox it to me. Wow. And so I did it. It was actually a very helpful thing. And then I will say one thing about ChatGPT, to its great credit. It was too sad because it was too late for me, too late for me. At the bitter end of this project, I'm doing the fact-checking, I had a handwritten diary of a guy who was on the board of the Fed, and I only was able to read like two pages of it the whole time. I'd given it to handwriting specialists and things.

2:21:39Oh, because you just couldn't understand. You could see the words, but you just didn't know what they – It was too messy. Terrible handwriting. chicken scratch to me so i'm doing the fact checking and i think and i had as a pdf because i had taken pictures of the pages and so i don't know what happened to me i just said you know what screw it i'm just gonna put it in chat gpt maybe it can read it and it read it wow decipher the scrolls yeah it wasn't perfect at all but i was like oh yeah that is what he's trying to say oh and that matches that and that so i do wish that in some ways i had access to ai because I think that, I don't know, I don't think the story would have been totally different, but maybe some things would have come together in a different way.

2:22:19Yeah. Do you think part of why the crash was so bad was just the lack of high-quality real-time data that the various players had to make decisions on? It feels like... Oh, hugely so. It feels like you would have just been, if you're just wildly confused about what's going on and you have people banging on your office door telling you one thing, and it just feels like it's hard to actually create a plan if you don't know how bad the damage is, how widespread it is, who the different players are. I'm sure people were actively trying to cover up, you know, bad things that they had been doing as well, right?

2:22:51That kind of thing tends to happen. So two things. The guy Charlie Mitchell that I told you about before, his bank almost goes under because the bank bought too many. The bank was trying to buy back its own shares during all this, and it bought back too many, and it couldn't afford to buy them. And so he didn't want anyone to know. So he actually goes and gets a loan, personal loan, to buy the shares off the bank. Okay. So, I mean, it was wild. And then Jesse Livermore, this trader I was telling you about, because he didn't, because he was so worried about the issue of having bad information, he paid for his own people to be on the floor.

2:23:29So then they would call him. It was like Citadel, placing, you know, their computers next to the exchange. He would place his people on the floor. That's you in the archive during COVID. You're the same. You're doing the same thing. Real-time information, exactly. Yeah, it's the same thing. Did they have revenue backlogs back then? That I don't know about. I don't think so. Well, can you talk a little bit more about your process? I mean, obviously, you're incredibly busy. How do you get in the flow state to actually write a book? Do you write one chapter at a time, kind of outline front-to-back revisions?

2:24:03Talk about your process as an author. So I'm one of those writers, and this is not good, I don't think. I really can't write one sentence. So let me say it this way. I don't really like to write the second sentence, less the first sentence I'm happy with. I'm one of those people who, there's some people who splatter on the page, meaning they sort of throw everything down and then they think they're going to fix it. I sort of have a view that whatever gets sort of put on the page is sort of anchored in a way. And so I can only upgrade it maybe one letter grade. So if it goes down as a B, I can edit it and make it an A.

2:24:39But if I just splatter it down as a C, it's never going to be better than a B unless I start over again. So that's a little bit of a thing for me. You know, this project went on for so long that I would write lots of little parts of it, little scenes, vignettes. And so I had these almost like puzzle pieces. And then it was about connecting them. I think the hardest part for me, just given the things I'm doing with The New York Times and CNBC and my deal book stuff, is for Flow State, I can't. My wife would sometimes say, oh, you have half hour, 45 minutes. You want to go work? You can go work on the book now or whatever.

2:25:15And unless I really had two hours, I couldn't really do it because the first half hour, 40 minutes, I almost had to rev up. Yep, totally. So that's a thing. That's a real – I think in the creative world or whatever, I think you do need to get in that flow state, and that's hard. And, you know, I've got three kids, and sometimes I actually try to write with them. like hang out with them and have them around. And that can work for me sometimes, but I have to sort of like really get super dialed in. What was the 1929 of 1929? Like what were folks in 1929 looking back to and being like, this is just like 1863.

2:25:56History doesn't repeat, but it rhymes. Yeah, exactly, exactly. Like what were they referencing? It's such a great question because the truth is they weren't. They weren't. For the most part, they really, because I think that this was such a first. Sure. I really think it was that the 29 was such a first in terms of that break. Maybe what they would say, so there was a break in the market in the early 20s, 20 and 21, there was a break. Yeah. That was substantial, but most people hadn't experienced it really. Because, again, it wasn't until 1919 that people were even started to think about taking on debt or anything like that in the country to then go to trade.

2:26:33It was really a function of General Motors, by the way. General Motors started loaning money to people to buy cars. It was a moral sin in America prior to that, really, to take on credit. That was a very grubby thing to do before that. Have you looked into tulip mania at all? It's referenced so often. And then I've heard stories about - Back then, they were not doing the tulip thing. I've heard that it was actually very short. It was very isolated. It was not global contagion, and maybe some of it was not even as big as – it might have just been somebody wrote down an extra zero in their accounting that day or for whatever reason.

2:27:12But, yeah, it's a fascinating story that now has just grown and grown in infamy, but maybe actually wasn't as big as something like 1929 that deserves a huge book. Maybe that's a slim volume. Maybe, yeah, maybe, maybe. How do you do three hours of live television a day? How do you do three hours? By the way, I just want to tell you, I really admire what you guys are doing. I didn't even get to say this. Thank you so much. We've now gotten a chance to meet each other a couple of times, and it's just a joy to be on with you. I think what you're doing is amazing. Thank you. And it's really, really cool to see your success.

2:27:43I was so thrilled to see that piece in the New York Times over the weekend. I appreciate it. Well, you're a hero to us. You're an absolute hero. My favorite, we've said this line on the show before, but when we got to hang out in New York a few months back, you said something to the effect of, I do TV on my way to work. and hearing about your process with this book, it's clear that you're just an absolute workhorse. Yeah, when are you guys publishing a massive, extremely well-researched book? Because it seems like this three hours of TV is really taken out of you guys, and it is. But hopefully we will learn and develop the muscle memory and the flow state and whatnot.

2:28:16There's a lot to learn. This is a long game. We've learned that. So thank you so much. You guys are doing it. You guys are doing it. It's an honor to have you on the show. Yes, you're welcome. I cannot wait to get into the book. We'll be pulling more. I'm just sad that I'm not in person and there's no gong. I'm at an Aztec. Hit the gong. Hit that gong for Andrew. One of the most well-deserved that we've had in a while. Thank you so much for coming by. Let's do this again soon and have fun on the book tour. I feel like you gave us the perfect amount, like a little teaser, a trailer. We still need to get into it.

2:28:49It's great to see you. Thank you so much for stopping by. You're the man. We'll talk to you soon. Cheers. and if you're watching or listening please go pick up the book 1929 it's available now you can get it on audible you can leave it five stars leave a review leave an ad in the review i think maybe you can technically do that you probably get banned leave an ad for squawk box leave an ad for deal book yeah i was uh to help out i i wanted so badly to put the audible on last night as i was falling asleep but i would have had to wait a few more hours but that's going to be my night. Yeah, I think the Audible is probably 40 hours long or something.

2:29:25So you listen to it at 10x. Yeah, so consider taking off next week and just walking and listening to 1929. No, this is going to be a fantastic book to dig into. Tyler wants this copy. It's so cool because there would have been a way to... Andrew could have done this book in probably a year, two years. He could have sold a ton of copies by just using, he would have made it, it would have been entertaining, but the fact that he went and spent seven years like actually doing it properly, even though, even like he just really cares about fundamentally creating a great product. Well, we have another author joining us in the TBPN Ultra Dome.

2:30:09Brian Potter is in the Restream Waiting Room. He's the author of Origins of Efficiency. Thank you so much for joining us. We love Stripe Press. We love Stripe. We had Privy on. We had Dwarkech on. We've always enjoyed Stripe Press's books, and it's a pleasure to meet you. How are you doing? I'm good. Thank you guys for having me on. Thanks so much for joining. Would you mind kicking us off with an introduction on yourself and the book? And we can go into a bunch of questions about it, but I'd love to just kind of get a little bit of background for everyone on your journey to writing this book.

2:30:43Yeah, so I am a senior infrastructure fellow. I work for the Institute for Progress, which is like a pro-progress think tank. I'm best known, to the extent that I'm known, for writing this newsletter called Construction Physics, which is about buildings and infrastructure and how to get stuff built in the U.S. And my background, before I did this, I worked in the construction industry. I worked as a structural engineer for about 15 years, designing buildings and parking garages and water treatment plants and stuff like that. And the industry always seemed extremely inefficient to me. Everything is so labor-intensive.

2:31:22It takes so long. We're doing this similar work over and over and over again. It should be much more efficient. It should all be done in factories, blah, blah, blah. And then in 2018, I had the chance to join this big, exciting construction startup called Katerra that had raised. This was back when SoftBank was... I remember Katara. Let's give it up for Mosley. It was kind of a precursor. I think Hadrian was drawing from it now. It was like Machine Shop almost. Yeah, it was this idea of construction is inefficient because it's not done in factories, right? So we're going to introduce factory-based construction.

2:32:00It was run by all these former electronics manufacturing guys, so not like software guys that think that, oh, I worked at Amazon, so I know how to do anything. It was people who knew about manufacturing. And they were going to bring that knowledge to the construction industry. So they raised a huge amount of money, got a huge check from SoftBank, raised$2-3 billion in venture capital. And it all went sideways. It all went wrong, and they burned through it in about three years and declared bankruptcy. And there were various reasons for that. At what point did you leave? I was there until about a year before they went bankrupt.

2:32:45So you saw the, did you see the writing on the wall? Yeah. When I was there, I was saying, but my time there was like a year and a half of ups and then a year of downs. And then after about the sixth round of layoffs, after our engineering team got cut by like about 90%, I was like, time to bounce. Saw yourself out. But I wanted to understand why things had gone so wrong. Because part of it is just, startups are hard. Building startups that are moving physical things around is very hard. There's various operational missteps or whatever. But also I kind of came to believe that the thesis that they had built the company around was either wrong or just not complete enough, missing very large chunks of it.

2:33:36Because people had tried to do similar things that Katera had done many, many times. If you go back over history, there's a huge graveyard of companies that are like, oh, Lightbulb will build buildings and factories and it'll be so much cheaper and I'll be able to make a huge amount of money. I'll be the Henry Ford of housing. And it just has never worked. People have tried this over and over and over again and not been able to succeed. Will it work? because I have an opportunity to invest. I have a lucky opportunity. No, I actually did meet a company that's taking another crack at this. Do you think it'll ever work?

2:34:11I do think it will work, but again, I wanted to understand why specifically it had been so hard in the past and why Katerra and so many other companies had failed and what specifically would need to be true for it to succeed in the future. So basically where I ended up was like, I need to understand what specifically makes it possible for an industry to get more efficient over time and what specifically is happening when that is occurring and what sort of can prevent those things from happening. And once I understand those mechanics, I will know what specifically would need to be true for the construction industry or any industry to sort of improve over time.

2:34:57And so that was sort of the genesis of the book. It's like, what specifically does it take for some process to get more efficient over time? It feels like the entire book is kind of an abstraction on top of just this idea of the learning curve. We've seen this in semiconductors. everyone who follows like the AI boom is acutely aware of the learning curve that happened at TSMC. But you kind of draw a couple other historical analogies. What stuck out to you is like particularly great examples of this efficiency in like going successfully and us actually driving down the cost. And then what were the commonalities between that and like what do they all have in common basically?

2:35:36Yeah. So I kind of went through and I looked at like, you know, dozens and dozens and dozens of different industries and seeing how they had improved their operations over time and what specifically was changing in them when that was happening. And I looked at industrial improvement systems, so like lean manufacturing and value engineering and statistical process control and all these other things that had specific ways you could try to make something more efficient. And I kind of ultimately boiled all that down to like this, you know, a list of like a handful of things that you had to do to try to make some process more efficient.

2:36:12If you could do any one of those things, you could kind of make it more efficient. And if you couldn't do those things, those paths were blocked. As it turns out, they are in the construction industry. You can't make your process more efficient and it just gets more and more and more expensive over time. And so, yeah, I looked at like a lot of different industries. I go really into, you know, Henry Ford and how he sort of dropped the cost of the Model T. I look at sort of the evolution of like nail manufacturing, which is in the sort of the 19th century, go back even farther. And how they changed the technology to make over time to make nails, which started out like hand forged nails, like a blacksmith, like hammer and steel.

2:36:50And they found a way, you know, machines that could sort of emulate that process. And they've replaced those machines with even better machines and so on. So there's like dozens and dozens and dozens of examples in the book of sort of specific things that have gotten cheaper over time and the lessons that we can kind of learn from those things. How naive is it to just say what's remained stubbornly high cost-wise, housing, medicine, education? What do those have in common, regulation? How naive is it to just throw regulation as the problem at those particular industries? That's a big part of it, for sure.

2:37:28I mean, the problem is that everything has gotten more regulated, like manufacturing included. So that's part of the puzzle, but it doesn't really tell you the whole thing. Because even in place, you know, the problem, to take it back to construction, the problem of construction productivity and not getting cheaper to build stuff is really something you kind of see around the world. I have a graph in there that's like construction costs in a variety of different countries, and they all have this scary line of going up and to the right over time. Even the countries without building codes and onerous HOAs or anything.

2:38:03Or less labor or different regulatory regimes. There are certainly places that do better than the U.S. in various things. In various ways of building, the U.S. is very far from the efficient frontier. But we have a very hard time of pushing that efficient frontier forward. So regulation is a big part of it, but that's kind of one of the sort of things, I think, takeaways from the book is that it's not just regulation. You could remove all the regulation you wanted and you'd still run into these sort of various physical constraints and market constraints that prevent these sort of efficiency improvements in some cases.

2:38:39How are you thinking about energy in America? We've gone through this AI boom now where we've scaled up the existing capacity of data centers. We're building new data centers. And it feels like the last link in the chain is, can we build 100 nuclear reactors in America in 2030 to stay on track with the most aggressive projections? Is there anything unique about, obviously, energy production is a construction problem. But is there anything unique that you found in the energy industry that folks might be able to learn from? Yeah. Well, I read a lot about energy on the newsletter. I don't have a background in energy.

2:39:19So it is a lot of me groping my way towards some understanding of how this industry works. I'm a really big solar guy. Solar has a really lot of nice properties that makes it easy to make efficiently at very, very large scale. There's this really interesting paper, but basically it's a big graph of the sorts of energy technologies that have become cheap and the sort of energy technologies that have not become cheap. And the ones that have become cheap are these sort of things that you can make repetitively in very large volumes and you don't need a lot of customization of. And so like solar panels, which are like you can make in like really, really, really, really enormous numbers and you can kind of plop down wherever it doesn't need a lot of like site specific customization are kind of in this like very cheap quadrant.

2:40:07And then something like a nuclear reactor, which you make in like much, much smaller numbers and like needs a lot of like specific design for the specific reactor that you're building is sort of in the much more expensive quadrant. And so solar and the batteries, which really complement them really nicely, is a really good way to make this stuff really cheap. These cost curves have gone down a lot, and there's no sign that those are stopping anytime soon. And so it aligns with so much of what we know about it, what it takes to make something inexpensive, that I kind of see that biting off a very large chunk of the energy that we produce in the US.

2:40:51Assuming, take it back to regulation, assuming that regulation interferences don't get in the way. How often did you find capital being a constraint lead to more efficiency? I think every startup founder has an example of a time when maybe if they threw 100 people at a problem, they would have gotten a different solution, but they only had a handful. And so they were able to create some novel, a more efficient way of doing something. Or we saw this with DeepSeek and having potentially fewer chips and creating a more efficient architecture. Was that a common theme at all? Yeah, it's interesting.

2:41:35I think there's kind of like two sides of it. One is that in some cases, like a repeated theme of the book is that like scale is really, really very important. And the more you can make of something, the more opportunities you have to make that less expensively. And often scale is like very, very expensive, right? So like one of the story of like container shipping over time is a story of like needing really, really enormous investments to like build these big giant ships, which are like cheaper per container that they're transporting, but very expensive overall. And also like really, really big expensive terminals to sort of handle those ships.

2:42:13And so only like a certain number of like countries could like invest in these like giant terminals that were needed to sort of service these huge ships. And so, you know, costs of transporting these goods fell a lot, but like there was winners and losers in who sort of gained gain from this technology development. And it was really the people who could afford to put the money into it to do it. But then on the other hand, you also see cases where, kind of like you talked about with DeepSeek, people working under these constraints were able to come up with really improved ways of doing something that were much cheaper and much better than what came before.

2:42:50So a kind of example of that would be like Toyota's manufacturing methods, which like Toyota production system, which evolved into lean manufacturing. Those kind of were created in this environment where like they couldn't develop these like mass production methods that Ford had used because their car market was so much smaller and it was so much more varied. they couldn't just make a million of a given model or whatever, that they had to find ways of producing this stuff efficiently that didn't require this massive capital investment, basically. And so that was sort of the genesis of those ideas.

2:43:26And so, yeah, I think there's definitely a case where, yeah, you need a lot of investment to find ways to make this cheaper. But then there's also cases where it's like, also working under constraints of not very much investment has been important as well. Are you at all optimistic that this data center boom will teach a generation of people that you can build big things quickly and efficiently if you just basically put your mind to it? Because there's like a lot of, from an energy standpoint, just like, you know, if you look at what Elon has done with Colossus 2, he's basically doing the impossible.

2:44:04A lot of people would have like looked at that project and said, it's not possible. And so that sort of, it feels like that sort of mindset of like, we're just going to make it happen, is being applied to data center development. But then presumably those people can say, I'm going to build a bridge. and they can imply that same kind of approach elsewhere? I certainly hope so. We're certainly building an enormous amount of this infrastructure. It's really, really unprecedented. There's all these crazy stats like data center spending is now exceeded office building spending or something like that, which is totally wild.

2:44:43I guess one thing that worries me is that historically people have been like, you know, not really cared about data centers. They've been happy to just like let them get built and the jurisdictions sort of collect the tax revenue for it and not really worry about it beyond that. As like the build out of them is like going forward and there's like more and more of these data centers and they're bigger and larger, you're really starting to see like a grassroots movement of people like, you know, the NIMBYs sort of now being opposed to data centers in a way that they weren't before. So like Virginia, which historically has been a major place where data centers get built and has basically been fine with them getting built there.

2:45:24Now you're starting to see residents oppose them more and more. And you're starting to see grassroots movements around in different states springing up to oppose these things. So that worries me a little bit. And I hope the sort of forces of getting these things built and enthusiasm about building infrastructure are stronger than that. But it always seems like the NIMBY forces are quite strong. So hopefully they don't build momentum. They're OP. I have one last question. There's this post by Rune who's talking about Dan Wang's new book. And he says, the general elite consensus now is that industrial process is a technology that It lives in the heads of people.

2:46:10And he goes on to say that it was a mistake to let so much low-value industry be offshored due to the loss of tacit process capital. And I was just wondering what your thoughts were on this idea of industrial process knowledge, that there might be a few key people that actually know how to build something at scale, and just what the ratio, how steep is the power law of human capital when it comes to large-scale industrial manufacturing efforts? Yeah, I think it's dead on. I think that's absolutely very important. And I talk about that at various parts in the book, how it's often really hard to transfer manufacturing or production technology from one place to another place, in part because it's hard to pick up and lift these process knowledge, which is just in the head, so embedded in this web of relationships.

2:47:00And so it doesn't necessarily even exist in explicit form, right? It's just like, this is the system that turns out to work very well. And you can't just like recreate it because we don't essentially know how it came to be in the first place. And then, you know, you talked a little bit about the learning curve earlier. And that's kind of this really similar idea where a lot of your improvements to some technology over time come from just like the factory floor and learning how to sort of do this better and better over time. but it's very coupled with actually physically doing the work. And so that's one thing that I think is really important is that oftentimes just technological progress is coupled to sort of this process factory knowledge of actually having the experience doing things.

2:47:46One really fun sort of example of this is during the early days of the space race where the U.S. was having a really hard time building their rockets. And there's a part where, like, you know, because of various political things, the Navy was going to send up their rocket first. They were going to be, like, the first ones to sort of launch a U.S. satellite into space. And Wernher von Braun, who was the German rocket scientist who then had been brought over to the U.S. and was working for the Army, he goes to some, like, you know, military leader, and he says, look, you can tell these Navy guys they can do whatever they want.

2:48:24They can take my rocket and they can paint Navy on the side of it and do whatever they want, but they need to use my rocket and not theirs because my rocket will work and their rocket won't. And then what ended up happening was they didn't listen to him and the Navy launched their rocket anyway and it didn't work. It blew up on the pad. And then so finally they listened to Wernher von Braun and just launched his rocket. And that's when we finally got a satellite into space using Wernher von Braun's rocket. And then of course Wernher von Braun was like a major force in the Apollo program as well.

2:48:54So it was like, you know, the German rocket knowledge that had accumulated during World War II was like very, very important in both the Soviet and the U.S. Their early rocket development efforts were basically built on this German knowledge that had been accumulated. So this process knowledge and like this, you know, expertise that gets embedded in the heads of these people working at the sort of forefront of technology is not easy to sort of recreate. I think it's very, very important. well thank you so much for stopping by the show the book is Origins of Efficiency from Stripe Press it's available now for purchase highly recommend picking it up one click on Amazon Stripe Checkout hopefully we will talk to you soon have a great rest of your day thanks thank you so much really quickly let me tell you about adquick.com out of home advertising made easy and measurable say goodbye to the headaches of out of home advertising only adquick combines technology out of home expertise and data to enable efficient, seamless ad buying across the globe.

2:49:54Speaking of founder mode, we have a founder in the Restreak waiting room. Let's bring him in to the TBPN UltraDom. There he is. How are you doing? Good to meet you. Hey, I'm doing great. How are you guys doing? We're doing fantastic. Fantastic. Kick us off with an introduction on yourself, the company, the news. We're excited. I'm Parry Singh. I'm the founder and CEO of Flow Engineering. Flow is a collaborative development platform specifically built for next generation hardware companies. Our customers design things like rockets, aeroplanes, cars, nuclear reactors, and they use Flow to design, build, test, and iterate massively faster than they can do today.

2:50:30Basically, the way to think about it is we're taking the last 30 years of software development practices, everything from agile to continuous integration, continuous testing, and we're bringing that to the design of massively complex hardware products. Our news today is that we raised our Series A with Sequoia Capital. and how much how much did you raise 23 million dollars there we go congratulations uh amazing i love that you uh you started uh building hardware yourself and and then figured out along the way i gotta build software i gotta build sass for this what uh what what are you replacing most of the time is it uh well it sounds like system am i getting it right you basically built you built an internal tool for yourself initially while you guys were building rocket engines and then realized like hey seems to be pretty valuable yeah exactly so like a little story of the company uh i'm a mechanical engineer i became an engineer because i wanted to build machines that matters uh went into the industry went to companies like ba systems and bp and just really as the fundamental approach to designing hardware was completely out of date so the company started out as a hardware company not a software company we were called the rocket company.

2:51:42We built the world's fastest design consultancy for hybrid rocket engines. The best people in the world could go from requirements to detail design in 12 weeks. We could do it in two hours. The reason we could do it in two hours is we built this internal platform for ourselves called Flow that massively integrated and accelerated the design process. That's what has become Flow today. You're just simulating physics. What is the actual... I imagine you're basically building a workflow and then you're able to are you able to get a good read on if the the process or the product will work without actually testing it in the real world or what does that look like yeah so let me give you the 101 on like hardware development versus software development in software development we have kanban boards and tickets and you build a spread burn it down and you go for it when you're designing something like a rocket or an airplane or a car a nuclear reactor it's much more complex the way that we fundamentally design and collaborate are using these things called requirements.

2:52:40Let's say you're building a rocket, you'll say, hey, I need to get this much payload to this delta, to this orbit. And then to do that, I need to design this first stage and this second stage. And to do that, and you go all the way from these top level requirements to very, very low level temperatures, pressures, masses, and design criteria that engineers will use day to day. The big problem is that when you're designing like a humanoid robot, or you're designing a reusable rocket, or you're designing a self-driving car, you don't know whether the requirements can be met or not. Like 10 years ago, you would have these fixed requirements, you'd be able to execute against them, you'd build a big Gantt chart and you'd burn it down.

2:53:17In a modern, massively complex system, our products are so complex that we have no idea whether we can design it. The requirements are changing on a nearly daily basis and the design is changing on a nearly daily basis too. So what Flow does is it's a single source of truth for all of the company's requirements and systems information. And we glue all the requirements together, all are designed together and we have continuous integration between the requirement side and the design side which enables teams to design and propagate changes much faster than they can do today are you aiming to go straight to the fortune 500 the fortune 100 the biggest companies in the world that are manufacturing its scale and uh maybe it's a lot of steak dinners and a really hard patch but you get a couple of those clients and you're in business or do you want to focus more on startups smaller companies scale ups like what's the sweet spot for you right now Yeah, we think about this very deeply.

2:54:08We regularly turn away Boeing and Airbus and these massive conglomerates. So here's the way to think about it. The hardware engineering industry is going through a generational change right now. and it's this generational change from old school waterfall think nasa lockheed martin to new school agile think spacex and andural the way that spacex and andural and jovi and archer work are much more like software companies than traditional legacy primes they don't line top down they design bottoms up and things are changing on a nearly daily basis we're very very very specifically built for that new way of working in the same way this happened in the software engineering industry So in the 2000s, we went from old school waterfall to new school agile.

2:55:04And companies like GitHub came about to serve that market. Now, GitHub didn't go to IBM and say, we're going to build you a slightly better Gantt chart. They went to companies like Google and Facebook when there were five people. And they said, this represents the new industry. And 10, 20 years from now, these small companies like Google and Facebook will be the mass market. And then when IBM and Oracle wake up, they will change how they work. and they'll come to GitHub because they are changing to an agile way. So that's what we're doing. We're exclusively focused on next generation aero space nuclear defense companies.

2:55:36We're growing very, very quickly with those guys. And we're making that workflow as good as it can be. How is it going in the gundo? What's the update? Yeah, so this is like a global movement. But as you mentioned, the epicenter of the global movement is El Segundo, which is in LA. So everything from like rockets, airplanes, robots, cars, autonomous submarines are being designed like the five or 10 square miles, which is El Segundo. El Segundo is amazing. I think it represents something like 70 % of our customers. And the companies in El Segundo design and iterate at a speed that Boeing and Lockheed just can't comprehend.

2:56:15They're designing massively complex systems. They're designing and iterating faster than anybody thought they could do. And that is the reason they will become so much more competitive than the traditional primes that the traditional primes just can't keep up. And I imagine what an advantage that is for you being able to walk a few blocks and like see your product in action and actually get that real time feedback and then just be on that same iteration cycle with your customers. Yeah, we have a kind of crazy story, which is most of our like most of the other tools in the market came from El Segundo.

2:56:48we actually came from London and we were engineers and we wanted to build um and the European market just didn't want speed or at least the market back five years ago didn't want speed so we sold into traditional legacy companies and they they fell in love with the dream and the mission but they didn't really they didn't really use the software and then the El Segundo market found flow and they pulled us into it and what started out just one or two companies working in this crazy new way designing and iterating like a software company have ended up becoming the new market and that represents a really important part of our customer base.

2:57:23Well, congratulations on the funding news. Congratulations on the progress and good luck to you. When you announce the B, come on over. We're like 20, 30 minutes from the gundo. We're in Hollywood. We'd love to have you. Ah, sweet. I'd appreciate that. Bring the gong in person. We'll talk to you soon. Congrats to the whole team. Chat soon. And when you announce that, Sirius B, you know what you got to do. I know. Go over to GetBezil.com. Your bezel concierge is available now to source you any watch on the planet. Seriously, any watch? Take a couple mil off the table in secondary. Put in an FPJORN.

2:57:55Put it in a Rolex Daytona. Deploy it. Put it in a Plot Tech. Get a couple Aquanauts. Before you get the starter hum, get a starter hitter. Starter hitter. And then book yourself a vacation on Wander.com. Find your happy place. Book a Wander with inspiring views, hotel-grade amenities, dreamy beds, top-tier cleaning, and 24-7 constatering service. It's a vacation home, but better. It never gets old. Jordy. I just saw a post that I'm not going to read on the timeline. That is funny. But anyways. Cool story. Cool story. No, what I was going to say, I was going to check the timeline before we get off.

2:58:31We have to enter our fourth and then fifth hour of podcasting. I'm so glad. I was worried we weren't going to get to five. Jackson Dahl is, I believe, already here in the Ultradone. Breaking news. And we're going to be doing his podcast right now. So head over there, subscribe. Go shoot him a DM. Turn notifications on to Dialectic, and then you'll hear us talk more if you're not sick. Yeah, I don't know when this episode will come out, but if you message Jackson now or you comment on one of his posts, I'm sure you can ask some questions there. And we hope you have a fantastic evening. We will be back tomorrow for another beautiful day of technology.

2:59:11Hopefully it's another cozy, warm day, and we can put the fire on. I know, I like having the fire on. I really enjoyed the fire. We should have the fire on for Jackson's podcast. Well, yeah, we'll consult with him. Thank you so much for watching. Awesome. We'll see you tomorrow. See you guys tomorrow. Cheers. Thank you.