J.P. Morgan Goes Patriot Mode, Dutch Seize Nexperia, 𝕏 Timeline Reactions | Alexis Ohanian, Ryan Petersen, Serena Ge

28:22What does history teach us about whether you should build an army of missionaries or mercenaries? Because I feel like some people are going both ways on this. Well, yeah. Right before you joined, I was just joking about Andrew Tolick from Thinking Machines over the summer. I'm just imagining what the exchange with Mark was. I'm really I'm really honored that you would give me this$1.5 billion offer. It shows massive faith in me, but I'm going to stick with the team, stick with the mission, and I hope we can find a way to collaborate. And then Zuck circles back with$3.5 billion, and suddenly it's like, all right, I think we're in business here.

29:02That one was rumored,$3.5 billion? $3.5 billion is the rumor, yeah. Wild. Wow. Well, yeah, Machiavelli was the one who wrote the most about mercenaries and how useless they are when it comes to actual fighting. I think about this more in the context of Zuck's security team than I do in his AI engines. When the bullets start flying, are these guys ready to take one, or are they going to run away? Well, to Andrew's credit, he was at Meta for 11 years. Yeah, he is just going home. He's going home. Yeah. Small price to bring him home. Yeah. How was your weekend? Great, great. Well, I'm here in Southern California.

29:42unfortunately I can't visit you guys. I'm in Palos Verdes. We're having a Flexport customer conference starts tonight. Been practicing my big talk tomorrow. We're releasing a whole bunch of new technology and host a whole bunch of new customers. At this point, I imagine like 100 % of the Flexport team, they have to be missionaries, not mercenaries at this point because every day you could wake up to a 1 ,000 % tariff on your entire business. And if you're a mercenary, you're going to jump ship. So I imagine the team is kind of buckled down already for this. And Friday wasn't actually that crazy, even though the market was crazy.

30:20But what was the internal? Give us the play-by-play from Friday morning. You said you had just gotten off of 2026 planning with the team and got hit with that. Thankfully, we do agile planning and we replant all the time. But yes, we just had a three-day planning session in San Francisco with like 40 of our leaders. and then like within hours of wrapping it up and saying goodbye to everybody, Trump announced 100 % duties on China. And the last time he had these duties at this level back in April and May, we saw our volumes from China to the U.S. go down by 60%. Yeah, that's huge. That lasted for five weeks until they relaxed the tariffs and then they came roaring back.

30:59So we'll see. I mean, Polymarket has it. I think we've all learned a little bit. Stay calm. Polymarket has it at least an hour or two ago. had it at 13 % odds of this going through on November 1st. Sure. So the market doesn't. And then Trump also kind of signaled on Sunday that, hey, maybe he's going to ease off a little bit. So we're definitely not panicking on it. Props to the Flexport Customs engineering team, because we have this thing, it's called tariffstopflexport.com. It's a tariff calculator. And these guys have to keep it updated. And this administration loves putting out the news on changes to customs policy on Friday afternoons.

31:34And these guys have worked so many weekends to keep, you know, speaking of missionaries, they're working nights and weekends on duty, on call, on pager duty to keep the tariff simulator up to speed with whatever the newest regs are. From your actual conversation with customers, what was the, was there any material change to folks' business post-Liberation Day? Like I know personally, like several companies that actually took seriously, like the mission of reshoring. And now there may be like six months into building a different facility, but it's taking years. And, but they're like, like it definitely flipped a switch from American dynamism is trendy to like, there is a board level mandate to, to actually consider reshoring in some way.

32:26How have your customers perceived it? how are people thinking about moving around the globe generally that was definitely true in the first few months but then when started hitting you know like india tariffs right now are as high well the china tariffs went up on friday but until friday tariffs on india were just about the same as on china sure um so it was sort of you know latam tariffs are pretty low but it's unclear if that's like here to stay or just a temporary blip you could see trump changing that overnight um so it's hard like supply chain especially if you're doing your own manufacturing you're setting up your own plant that's like a multi-year four or five-year minimum commitment that you're probably making uh before it pays back and then after that you know return on capital would take even longer than that so i think it's the smart i i think the folks you're talking to probably in some sort of defense industry or something that's like more national security oriented and you're like yeah we need to do this in the united states yeah for your typical company that's making apparel or home goods or other types of products it's just like it's not really there to come back to the u.s i've met more companies who aren't going to do it as a result of having like the worst case i've seen is a company that makes bicycles in the united states and they've made it so that bicycles are uh duty free if you import them but the bike parts are not so these guys can't import parts yeah they have to pay taxes on tariffs on the part but their competitors just import whole bikes so they're like hey we're just going to start producing overseas that's really rough yeah it feels like at the end of the day there's like so many little sub niches in the economy like we there was a story in the wall street journal about uh the maker of sharpie was successful in like reshoring all of sharpie manufacturing but sharpies are probably not, you know, the target of like one specific tariff.

34:17And then also that's a critical industry for Trump. He's got a hilarious one. That's true. The autographs can't get signed. Yeah. Um, whatever has an auto pen, will it be a Sharpie like auto pen, Sharpie auto pen? What do you mean? It's like a machine that does the signature. Oh yeah, yeah, yeah, yeah. That's how Biden signed all of his day. So, I mean, is, is everyone just kind of glued to truth social now to, to get the latest updates on where this all goes or, or are there other kind of like folks that are like reading the tea leaves and understanding like as poly market, the, the definitive data source, like how are, how are customers actually updating on like how real this is.

35:03I think prediction markets are, yeah, incredibly valuable for this. Really useful. They're pretty thinly traded. Like the one I cited only has$146 ,000 of liquidity. Yeah. It's not really getting used because one customer could make a hedge. Sure. They're going to, it's going to cost them more than that. So, um, but there's, there's still signaling that, um, actually, you know, the best source for us is just going to DC and talking to people. I've been going there once a month, um, going back in a couple of weeks and just trying to get Intel what's really happening, what's coming down the pipe.

35:32I think the more you can build a relationship. This administration is like, you know, as opposed to terrorism, as I am, they've been very open door to hearing our ideas, listening. They genuinely want to help businesses. So they've been like, we've been able to go in there and talk to them. And we're working on a couple of things that we think we can reduce customs fraud. Like there's this massive fraud that's happening in customs right now. Oh yeah. Because you just have this incentive. If you're, the tariffs go way up. Now there's an incentive to cheat that wasn't there a couple of years ago or even last year.

36:05And the way that you can cheat, this is highly illegal. Do not do this. Do not do this. It's not worth it. I only have one rule is I'm never going to jail. Yeah. So don't have the rule. Simple man. Simple man. Yeah. But so the way that it works and it's very rampant right now is the United States is the only country in the world that allows foreign companies to import goods into the country. With no legal entity, no requirement to have an employee or person locally. So you can just import stuff into the country as a foreign company. And then you just lie on your declarations. And our customs, CBP has agents in like 60 countries, but they're there for counter-narcotics and anti-terrorism stuff.

36:49They're not there for trade compliance and trade enforcement. so there these companies just cheat and there's like basically no and if they get caught they can spin up a new entity at home and just keep doing it so there's zero downside or maybe the goods so so and the the the fraud is somebody is bringing in let's say three million dollars worth of goods and they just say it's like oh it's two hundred thousand dollars something like that right and so and so again that the people that are punished through that are actual american companies or anybody doing it the right way that then has to compete with a player in the market who doesn't have the same cost structure as them because they're just not paying terror.

37:32Exactly. This is why we feel strongly about it. And right now on the Amazon marketplace, 60 % of the sellers are non-resident importers. That's what we call this. I know because 60 % of the time I buy something on Amazon, it's garbage. Yes. So I love Amazon, but that is becoming a problem. That's my personal experience. Yeah. So we're trying to highlight this issue around Washington, and people have been really receptive to learn about it and to see how this works and to some proposed ways that you might close this loophole. so yeah that and the way that what's happening where American companies are getting trapped in this is that they're allowing their factories to import the goods for them so instead of them importing it they're letting the factory import it and and feeling like everyone's their own personal lawyer and saying hey I didn't know about the fraud it's not my fault I just bought the goods in the United States and if they were if the factory didn't pay duties that's not my problem Well, that is not how the law works.

38:37If you're paying for your goods at a price less than the duties alone would have been, then you must have, you know, you should have known about this. And I think it's going to be hard to convince a jury. So I think you'll see some enforcement cases. But probably they just need to make it so a foreign company, just like every other country in the world, you need to have a legal entity in the country in order to import goods. Yeah. Do you have an idea of the scale of like the sweet spot for the scale of that type of fraud? I imagine that you really can't get away with that if you're, you know, doing$10 billion of commerce.

39:13But there's probably a whole host of little, you know,$20 million operations that are maybe cheating or something. It's not pretty big. So I analyzed that in the United States, ocean freight shipping manifests are public record. And so I analyzed all of them last week with a team member of mine. We went through this. And what we found was about 10 % of trade has switched terms to where the factory is now importing the goods, which is our estimate. Around 10%, 11%, which is just a massive amount of fraud. And probably going to keep growing and growing until they take enforcement. Yeah. Yeah. But is it even impossible to enforce?

39:58Well, yeah. Wouldn't you have to like 10x the size? It has to be a U.S. entity with my proposal, and we'll see if this lands somewhere, but it would have to be a U.S. entity with a U.S. bank account. With a U.S. director that can go to prison. And a U.S. employee who can be held liable. Exactly. And then, you know, like, sure, you can always break the law, but if there's a threat of going to jail, you're probably not going to do that most of the time. Most people wouldn't. And we have a great country because it's much easier to start a business than it is to be a criminal. So most of the smart people become entrepreneurs instead of criminals.

40:32Whereas in other countries, that's not true. It's easier to be a criminal than to start a business. Can you give us an update on Import Genius? We were reading a New York Times article and Import Genius was cited. And we were just kind of like, whoa, like, you know, we know that. We know the founders. But I'd love to know, like, what's the state of the business? It seems like you're still finding interesting data from that. Like, what role does Import Genius play in the overall, like, community right now? Yeah, so Import Genius is a data service that takes those public shipping records that I mentioned and makes them searchable and accessible.

41:07And so that's the data that I was using to figure out looking at this fraud. But a lot of different journalists use it to find what's going on in the world. or kind of finding, digging up fraud of some kind. There's lots of other use cases besides investigating fraud, but that's what tends to get into this. Yeah, I imagine you can just use it to understand, like, are volumes falling or rising or what rates are, et cetera. Exactly. We use it for market share analysis. We use it for lead generation, for logistics companies, helping to find companies that we should talk to about their freight. and people use it for factories like to find you're looking to do research on a factory kind of see what their shipping history is so yeah uh overall i mean maybe to close out um i'd love to know just kind of like there's obviously a lot of volatility with the tariff and the in the overall trade war i don't even like it's just been such a like a low rumble for the last years for the last year for this entire administration i'm sure it's been a lot louder in your world but uh does it feel like we're closer to having resolution?

42:14Like when we talked last time, I was like, I was like, I think the business community, and I think you agreed with me, the business community just wants something to stick, a deal. If it's 50 % forever, we can build our models off of that. We can invest against that. But uncertainty is particularly destabilizing to the US economy, to global trade, et cetera. Whatever it lands at, let's just get there now. Does it feel like we're closer than ever or getting farther away? Like, how do you feel about just like the level of volatility in global trade? It started to feel like things were calming down until really, I think Trump, the administration going after India is what kind of set off to be like, oh, okay, maybe these aren't as stable as if like a negotiation can just quickly spiral on a country like that, that in a lot the ways we have a lot in common with India and should be an ally of ours.

43:09The China, this current level on China is definitely at a, our view is anything above about 60 % duty is going to be an embargo level where almost all trade stops. And we're so we're well above that right now. I don't think that that's where the administration wants this to land. I think they'd probably like it to be a little bit below the embargo level. and give them a ratchet, make it a ratchet where they can put it above the embargo level on a moment's notice when they don't do what they want. I wish we could say we had the long-term stability. Like if the current rates on Latin America lasted, it would be really kind of very clear for folks of what policy the administration is putting forward is, hey, we're going to treat the Southern, South America, Latin America as kind of our backyard, our hemispheric partners.

44:05We're going to encourage kind of manufacturing there for stuff that doesn't make sense in the United States. I think it's very hard for them to come out and say that overtly because they were like, the actual message is no, make it in America. Yeah. In the United States of America, not America. America. What's the latest on with the ports, automation at the ports? Oh, it's illegal. That's the latest. No, it's not illegal. It's against the contract of the union for the next, I think, five more years from now on the East Coast and four more years on the West Coast, I want to say. And our direct competitors are doing the same thing, right?

44:49Say that again. And I'm sure the rest of the world is also banning automation, right? So we'll be equally competitive. And, of course, it's like, no, Singapore is probably extremely automated. Contract ends in like four or five years, and we'll see. But in the meantime, there will be no automation in the U.S. ports. Yeah, so we're behind. Makes it even harder. By the way, here in L.A., the Long Beach Container Terminal is an automated terminal. I think that's what kind of kicked off the union and pissed them off and made them realize this is like a real thing when they put it in about 10 years ago.

45:28It's really impressive. Like it's a giant robot that's like a square mile of all autonomous trucks driving around and there's no people in there. And so that we do have it. It's like a solved problem. The technology exists. It works. It's like really efficient, really cool to look at. I highly encourage doing a tour of that. Flexport is going to take a bunch of our customers on a tour of that at our conference here in a couple of days. Are there any other like geological or geographic areas in the U.S. that could be ports, could be automated ports, but don't have any port infrastructure and are just like, oh, it's some houses or something we could buy or it's just like kind of an abandoned beach?

46:11You can't because the union claims every inch of American soil. Okay. Of every inch of American coastline. So even if today they're not there, they still clean. They're in spirit. Okay. Yes. And they will show up. Yeah. So you can't do it. The thing that – one area I'd really love to see America take advantage of is our river network. And actually this points to like if Latin America were to become a much bigger trading partner for us, then you would predict that New Orleans should be the biggest port in the United States. Oh, yeah. If you have the whole Mississippi River network and you can get everywhere, all the way connecting up to the Great Lakes, up to Canada, you could really make an amazing – but there's almost no container shipping on that network in part because of union contracts and like – A Flexport River boat would go really hard.

46:59Maybe you could put a casino on it too to make it really American. 100%. I really think like of barge network running all over and like replace a lot of the trucks on the road too, because it's so much more efficient, cheaper, greener, like cooler in so many ways. And yeah, I haven't looked into the gambling laws, but. In international waters, you can, of course. So, you know, you just gamble down the Amazon, then gamble all through the Caribbean and then gamble up the Mississippi. It's the future. Anyway, thanks so much for stopping by. New product line. Congratulations. This is great. I got to ideate with you guys more.

47:43I think we'll probably come up with some good product ideas for you. I think a Koretsu of gambling and shipping is just a perfect second act. Mystery boxes. Everyone's asking, what's the Flexport second act? I would like to do something in futures, actually, where you could hedge the price of shipping and you know maybe that's a partnership with us in one of these prediction markets yeah pretty cool yeah yeah that'd make a ton of sense yeah i mean isn't that the story of uh southwest they bought a bunch of fuel futures and then they were like had uh the ability to pricing power for like a number of years i don't know how apocryphal that story is yeah yeah i think i read about that um we haven't done any hedging yet i'm always telling our finance team to do it and then they come back with reasons why you don't need to or something, but it seems like we should.

48:32Yeah. It'd be fun. Turn into a hedge fund. Be good. You can't today, there's not really like a futures. There's a couple of these marketplaces are trying to form like futures on actual price of container shipping. Those don't exist yet in a way that's really viable. But if that existed, we would for sure start. I would like to, we should be the smartest people in the market, right? Like at some point you're like, we're doing all this commodity shipping. We should just trade shipping. Maybe. Well, when you announce it, come back and tell us all about it. And good luck with the customer conference.

49:08Thanks so much for stopping by. We'll talk to you soon. Great to see you. See you. See you, guys. Let me talk about Figma. Think bigger, build faster. Figma helps design and development teams build great products together. You can get started for free. Jeremy Gaffan asking the important questions. Why hasn't someone done calendar invites for phone numbers? It's the last thing, keeping email in the loop. Totally agree. Low-tier banger here, too. It gets to such an awkward spot in a meeting where you're going back and forth on text. Always coordinating every text and then have to. And then, okay, who's going to bite the bullet?

49:41If Apple wants to really dominate an iMessage world, they should figure out how to add that to the iMessage flywheel so you're even more locked in because your iCloud account is the one. I mean, I guess they do that with the calendar app, but everyone's kind of on Gmail, so it's a little bit rough. It would be fun. Steve Jobs tasting cars, by the way. Everyone's having fun with the Steve Jobs, by the way. The black on tan is iconic. 911 is very good. And, of course, I think we've talked about it on the show before, but he would buy a new car every six months, so he never had to register the plate.

50:18He should have gotten a custom plate. He should have gotten Apple on the plate. You know, just let everyone know. Just let everyone know. There you go. What's wrong with getting a custom plate? Shout out license2post on Instagram. So there was some interesting alpha coming out of AlphaSense. Oh, yeah. Richard Jark shared a screenshot of it. Okay. He summarizes it. An interview with a former OpenAI employee that came over the weekend confirming my hunch, which I already wrote about in the article last week, that the biggest friction between OpenAI and Microsoft is around infrastructure capacity.

50:54OpenAI requests were, quote, leading to massive amounts of CapEx potential that Microsoft would be on the hook for. Sam basically saying to Satya, spend a trillion dollars for me. I want to be the leaser. And Satya's like, I want to be the leaser. And everyone's like, we want to lease. We don't want to pull the back. And two, Microsoft is getting a request from its biggest customer investment that they want, an almost inexhaustible amount of GPUs. That's fine in terms of predicting your demand for a one-year period from today, say. But what about in the year 2029 when you end up having all of this capacity that you've constructed?

51:31What does that market look like? No one really knows. That led to some friction between OpenAI and Microsoft. Richard says, my take on this is that Microsoft is being smart and letting others take the risk. Oracle and Neoclouds. As numbers for CapEx are reaching crazy numbers, when things come back to earth and many of these firms face difficulties, Microsoft is left standing to pick up the distressed assets. Tyler has a take. What you got? Yeah, I mean, this is kind of what Doug O 'Loughlin was saying. He was saying that Microsoft, super early on, they were like super bullish, right? They're in OpenAI early on.

52:04And then they kind of take a step back. I think he was saying that, oh, they're going to get back in the game. So hopefully, you know, hopefully we get the the gigawatts set up by by 2027 then we'll have you know agi god yeah it does seem like uh sata's doubling down on getting back in the live player ai race he kind of delegated a couple tasks microsoft for business to someone else in the microsoft leadership team so he's able to spend more time on ai the question is is he still happy to be a leaser that could be true he could say yeah i want to i want to lease even more from someone else and do a whole bunch of off balance cheat debt, essentially.

52:40There's a crazy post from September 24th, Sebastian Rashka. This is way deeper in the deck. I don't know if you're going to be able to find it. It's on page 94. But Sebastian says, no one understood me back then. And Nick, who is a famous OpenAI hater, shares a screenshot of the information reporting on OpenAI's mega deal with NVIDIA and says, OpenAI would lease NVIDIA chips instead of owning them. And so Nick says, news, OpenAI to lease NVIDIA chips rather than buying them. LMAO, like they don't even have the money to actually buy them. And back then, in September 24th of 2025, Sebastian said, actually, I think that's smart.

53:23This way, you don't have to deal with reselling and recycling old chips every few years. NVIDIA chips should be offered as a subscription. And so you were talking to Dylan Patel about this. You were saying like, what is the value of an A100? What is the value of an H100 in five years? Will you still be able to produce economic value? Yeah, it's still going to be able to inference LAMA 3. It might turn on, but that doesn't mean it'll be economically rational to leverage. Yeah, and my take is that, and what I've been digging around when we talk to people, like we talked to Kareem at Ramp about this, like there are companies that have found useful, like economically valuable uses for Llama 3 just somewhere in their software stack.

54:11Like, you know, with the ramp example, which of course is sponsored, but it's a good example, is just like you scan a receipt. You need to understand, like, what category is this expense in? And you can just take the name of the restaurant and what's on the line items, put it in Llama 3, and very cheaply get back, like, what category is it? It's like a classic just, like, text interpretation transformation problem. Easy token problem, right? Yeah, we were working on that internal tool for horse identification. Yes, yeah, exactly. But once you develop that tool - Clock my horse. Clock my horse.

54:45I got to tell the story about my quiz for you in a minute. But once you develop a tool that has economic value inside your organization, you're very likely to just leave that there. Like your database. Okay, yeah, we have a database. We have a cron job that runs every night that cleans up the database, tags some data, moves some stuff around, creates some summary analytics. Let's just let that run forever. And yeah, it costs us a thousand bucks a day or something. And I think that will be the case for a lot of those like GPT 3.5 class workloads, GPT 4 class workloads. So I don't know how bad it will be to just own a whole bunch of old compute, but it's certainly risky because you could get caught and the value of those could go to basically zero and that would be really bad.

55:30And so it is funny to see the question about like, will Microsoft actually go and buy a bunch of stuff like what is getting back in the game really mean yeah I mean what it what it looks like right now is Satya watching a bunch of different players lever up massively while he sits back and is risk off but you can imagine again we've talked about this a number of times a point it could be a year from now it could be six months from now it could be two years from now three years from now it's really hard to say where he comes in and gets very aggressive on buying up all these distressed assets that have done all the hard work, but got over levered and don't have the demand to, you know, uh, not be in breach of various covenants.

56:15Yeah. If the bubble just keeps inflating or, you know, things kind of all pencil out like people hope. Yeah. That's, that's really the tinfoil hat theory is Trump just wanted to take some of the heat off the AI bubble. So he announced a new trade war. Told you bubbles, bubbles over. Um, but we got the first ever message in the LinkedIn stream. Let's go. No, we do stream on LinkedIn. Let's go. LinkedIn doesn't invest very much, but Sham Mohite in the LinkedIn chat, can you tell the deployment percentage security to subscribe dot dot dot? I don't know what that means, but I like that you're engaging in LinkedIn, holding it down.

56:55Hopefully New York Times. New York Times called us SportsCenter for LinkedIn. And here we are. Here you are. You're the first chat room general. Thank you. So on the question of AGI timelines, is this a bubble? Over the weekend, you got a new phone. And even though AGI is just a few thousand days away, it completely destroyed your iMessage integration. And so I, on Saturday at noon, I get a message from your email, not your iMessage account, because I have your iMessage pinned with your contact information. It was a pretty serious. Yeah, it's just perfect, you know, blah, blah, blah, this other thing.

57:43What do you think? You know, you weren't exactly asking me to wire money anywhere, but it was a serious question. It was a million-dollar question. It was a million-dollar question. And so I'm like, so I screenshot it, send it back to you. And I say, this is coming from an email. So I have to do a check of humanity. Something only Jordy could know. What is my dream dressage horse? Because that's something only you could know until you've listened to this episode. Now you will know. And you. And I botched the answer. Yeah, you, the Jordy that I was talking to says Danish warm blood, of course. Next question.

58:17I said, wrong. It's the Le Pizaner dancing horse, of course. and you said, I knew I was forgetting the lip was on. And I said, sure, sure. Well, likely story. And so then you had to, you had to, you said pull up world coin. Yep. Yeah, yeah, yeah, yeah. I switched to, I switched phones, needed to deactivate the other. Liquid glass is pretty mid. And I said, a likely story. Feels, feels like a downgrade all around. Wait, before we do that, Mark and Michael Miraflor have jumped into the LinkedIn chat. Let's go. Welcome. Love to see it. Holding it down. All right, Mark's already back. Mark, let's go.

58:53It was fun while it lasted. Amazing. Could have been me. Link Denise. Mazuma. Anyway, if you want to stay compliant, go to Vanta. Automate compliance, manage risk, prove trust continuously. Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation, whether you're pursuing your first framework or managing a complex program. Okay. So as you know, I ordered my new iPhone weeks before you strolled into the iPhone, into the Apple store and just bought one off the shelf. I wasn't even there for a follow. So I was forced to wait.

59:23The anticipation was building. I had high hopes for the new phone. The second I started using it, I thought this feels like a downgrade in every single way except the actual quality of the images. The cameras are fantastic. The cameras are great. The cameras are great. Everything else from the way that it feels, it feels cheap in comparison to my titanium iPhone. The liquid glass, While I think it looks cool, it is very mid in practice. It's everywhere you go. Everywhere you go. It feels like everything's overly animated. Sure. And I'm already getting a lag on the animations. Just like why? Like why?

1:00:10I don't need it to animate. Just like be smooth and instant and don't lag. I have a couple takes on liquid glass I can drop. We can also get Tyler's review. What was it like upgrading from the very first iPhone to a modern iPhone? Are you enjoying the 3G? You were on the 2G iPhone, right? You actually had to plug it into the wall to talk to people. You've fully acclimated to liquid glass, right? You have any problems? Yeah, I think it's great. I mean, it's definitely, so here's my issue. You'll get used to it. They thought to themselves, we don't need to make a better user experience. We just need to make a different one.

1:00:56There's a little bit of that. It has to be noticeably different because it's not better. Yep. Nothing about it from using it for a few days have I thought, oh, this is a lot better than how it used to be. Yeah. I think most of my experience that's like, oh, this is like so much better is probably mostly because I was going from an iPhone 11, which is like seven years old to this. It's probably not actually the iOS. So, I mean, it feels way faster, obviously, but it's hard to tell. LinkedIn chat is absolutely going off. Mark, Ryan, for God, Toby. We are bringing news to LinkedIn for the first time ever.

1:01:34We're on a tear. Apple, first, they got to get on Graphite, code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. Okay, so a couple problems for me. the Safari with liquid glass feels like way harder to use. Like, I don't know how to get to the actual tab view. There's a whole bunch of new muscle memory. Just opening a new tab, I feel like I have to click twice. Yeah, and there's a whole bunch of subtle changes. Okay, you added more animations, but it's not better. It does feel like there's a few things that now require two clicks or three clicks that used to just require one click or two clicks.

1:02:09And so that's been annoying. I'm sure I'll develop the muscle memory. The new phone app feels very difficult for me to use. I haven't been able to figure out like where all the different calls are. All these things I think we'll sort out. I haven't enjoyed how hard it is to take a screenshot of an iMessage chat because the person's name appears in glass. And so if you're trying to line it up to screenshot just like one message and show the person's name to say like, hey, Jordy texted me this and I'm sending that to Tyler. Jordy said this funny thing, you know, here. Well, like, I'm automatically going to see through the glass, like the last thing he said that could be something different that I don't necessarily want to share.

1:02:50And so just the screenshots of iMessages have been a little bit tricky for me to get around. I will say, liquid glass looks cool. It does look cool. But it's worse to use. Yep. And that makes it bad design. I think I'll get used to it. Oh, the other thing on the actual hardware design is I noticed that the plateau just completely ruins the ability to use wireless chargers because you put it down on the wireless charger. Well, you haven't upgraded to the new wireless charger. I know. So I need a new car. Great. Thanks. Nick says, most people I know gain like 20 IQ when advising their friends' lives and lose 20 IQ when doing strategy for their own lives.

1:03:33Rune says, why is this? It makes absolutely no sense but it's true guy says solution befriend yourself that's very funny this totally this is this this is uh it's usually that It's very easy to give simple advice to your friends. and when you're trying to strategize in your own life, you end up making it super complicated because you have all this, you have the most information on your life, more information on your life than anyone else. You're like trying to do one of the Tyler red string boards and usually it's just a lot more simple. Yeah, no, I agree. And you're sort of forced to make things simple when you're giving advice to a friend because you know, okay, here's their home life, here's their work life, here's what you should do.

1:04:26Here's a life hack for doing strategy for your own life. Collect an immense amount of data about yourself. Put it all in Julius and analyze your life and make Julius.ai your AI data analyst that works for you. Don't even think about what to order for breakfast without first loading every previous breakfast you've had into Julius.ai and querying. Running analysis. Running analysis on it. What should I have, the croissant or the breakfast burrito? Julius. Why not both? It's your personal Jarvis. Julius, should I go to bed at 9 or 10? Julius, should I have a second Diet Coke today? They really should give the Siri button to Julius.

1:05:11Yeah, we even say, oh, they'll do a deal with Anthropic. They'll do a deal with complexity. Apple, Tim Cook, do a deal with Julius.ai. You got to do it. Get Raul in the building. Chat with your data. Get expert-level insights. Tyler, Kylie Jenner is working to make Snapchat a thing again, according to Jasmine. What's going on with Snapchat? What did she do? So I don't know exactly what she's working on. I don't think it's been announced yet. But on October 11th, it said tomorrow on Snapchat. Yesterday, you said tomorrow. Just do it. Just do it. So we do know some retail traders who are very long Snapchat, so maybe I should consult with them.

1:05:58Although I don't know if these particular traders are the most up-to-date on latest news. I mean, did this move the market? We need to check in with our resident retail trader on the Snapchat price. I mean, I don't know. I've heard a number of people, for a long time when I was doing YouTube videos, they would pitch me on, like, you've got to do a Snapchat version because in the, what was it, in like 2022, 2023, Snapchat was very kind of liberal with their version of TikTok's creator fund. And so if you were getting a lot of views on Snapchat, you could be making a ton of money. I think you have a buddy who's making, who took a, there were a couple of YouTubers who took their content, made it Snapchat native, and then got big creator payouts, basically like higher CPMs, just because Snapchat was like, we want to bootstrap like the real serious creator market, not just Snapchat with your friends.

1:06:47Yeah, I think they distributed probably hundreds of millions of dollars. Yeah, to get real shows, to get Mr. Beast-level shows. Tyler, what you got? So it seems that she was announcing a new music video that was launching on Snapchat. I guess she has been off Snapchat for some time and is returning. She does music? Interesting. I didn't know that. Well, if you want to generate audio, go to Fall, generative media platform for developers, the world's best generative image, video, and audio models all in one place, developing fine-tuned models with serverless GPUs and on-demand clusters. Did you see that Red Bull put Yuki Sonoda in a bucket of water next to a high-voltage power outlet?

1:07:25No, I had no idea. I didn't understand this image at all. Why is this? I think he's doing a cold plunge. Okay, and it just happens to be like this. Red Bull is the type of team to put that. That is absolutely electric. That is awesome. Get after it. Yeah, risk on, risk on. Should we run through my... Wait, I had a question for you. So I watched a film over the weekend. What? The third film ever. I watched a film. Jorgian Cinephile, his third film ever. Congratulations. I saw Brad Pitt play Sonny Hayes in the movie F1. Yes, F1. I watched it in two parts. That's okay. As the filmmaker intended. That's the filmmaker intended because he knew I would get sleepy and check how much time was left in the movie.

1:08:19I was like, this is a movie about racing cars around a track. Can't be that long. Oh, I've got another hour, hour and 30 minutes left. I'm going to bed. But the thing that I thought was interesting is somebody who's seen, I haven't seen two hands worth of films, but I've seen a few. Interesting to have a fictional story tied into current reality in F1. Have you seen a movie that's done exactly that, right? Where there's like Lewis Hamilton and other characters like blended in? Yes. It's rare. Super rare. Yeah. But they pulled it off well. Yeah. It felt like you were in the world. it felt like you were in modern day formula one yeah and just watching i mean in the big short they're like in lehman brothers right in goldman sachs that's my version of f1 yeah but not with like the actual people it's not like the managing director at lehman brothers is like playing himself sometimes there's little cameos but nowhere near as much as like this there was like 200 cameos yeah yeah yeah yeah no that was a very cool uh very cool feature of the film yeah I did enjoy F1 it was it was a good movie maybe the movie of the summer people were asking like what is the movie of the year like there was no shelling point there was no barbenheimer this year and I think F1 probably did the best the the real the real one that's like going viral is the k-pop demon hunters movie on Netflix that's the one that's like getting the most attention like squid game level but little never heard of it you've never heard of k-pop demon hunters you're telling me this for the first are you kidding actually You've never heard of K-pop doing that?

1:09:58No. It's crazy. It is like a massive viral success. It's a massive viral success. Let me run through my pitch to the US government to give Dylan Patel at Semi Analysis 50 million bucks. This was my take today. So we had Dylan Patel on the show on Friday talking about Inference Max. This is an independent benchmark where he takes NVIDIA GPUs. Pull up this video. Oh, yeah. From Semi Analysis. This is why I had to talk about it. Semi-analysis is potentially the greatest shitpost of all time. They say, was watching the Georgia game and noticed Dwarkesh Patel. There we go. There we go. It was literally on TV.

1:10:39Play some soundboard. This is good. A lot of people were saying. It's just so funny. He's like in Tennessee Friday calling into TVPN with the most unhinged setup we've ever had for a guest. It was great. And then he pops up here at, of course, this is not Dwarf Cash, if you've been living under a data center. Yeah, he's talking a big game. Oh, I'm out here because there's a lot of power. He's like, power on the field. Let's go Bulldogs, apparently. No, obviously, he's there for work as well. But it is an amazing, amazing clip. Is his buddy, like, throwing, like, waving money around? Something.

1:11:17I don't know. Yeah, what is going on there? They're GPU rich. They are rich. That's amazing. I'm so glad he got it. Okay, get into your rant. So basically, Dylan Patel, one of the greatest analysts of all time, the GOATs, he launched Inference Max last week. It's this website that you can go to and you can select one of three open source models, LAMA 3, DeepSeek, and GPT OSS. Then you can select what type of workload do you want to do. Do you want to do document summarization, put 8 ,000 tokens in, get 1 ,000 out? Do you want to do just chat back and forth or do you want to do more of a reasoning deep research type project?

1:11:56And then you can say what what architecture do you want to use? What the whole software stack and then what what GPUs do you want to run it on? Do you want to run it on NVIDIA or AMD? And so you get all these different charts. It's put the timeline in turmoil more than one time already. And and it's been really interesting because my take and I'm interested, Jordi, if you if you had a similar take, like it has two things have felt true. One is that NVIDIA has just been running away with it and AMD has been playing catch up for years. And what Inference Max shows, at least me, is that AMD is actually better in certain areas.

1:12:34And that was kind of interesting because I'd heard the story of like AMD is getting their stuff together. They're fixing bugs. George Hots is pressuring them. Lisa Su is talking to George Hots and Dylan Patel about what to do. And so the narrative of like NVIDIA is 100 % dominant and no one can touch them, that kind of breaks when you look at the inference max. And in fact, it seems like AMD is doing great stuff in multiple places. And for certain workloads, you'd want to pick AMD on a total cost to own basis. It's not just about like raw speed all the time. It's about how much does it cost? How much energy does it use to produce a set amount of tokens, right?

1:13:08Yeah, and it makes it even more surprising that they had to give away so much of their company in order to get OpenAI to be, you know, or Sam just exerted max leverage and just wanted to own 10 % of AMD. Yeah, I mean, my update from Instruments Max is that AMD and NVIDIA are not as far away as people thought. But, I mean, we'll have to dig into it, and obviously it depends on a lot of different factors. And the true frontier models are a different thing, and those aren't in the benchmark because they're not open source. The other kind of update for me was around open source. So we covered this, and we talked about this with Dylan.

1:13:52And like when GPT-OSS, the open source model from OpenAI launched, it was kind of like, well, okay. Like, yeah, it's like, you know, doing okay on benchmarks, nothing really special going on there. And it just kind of like fizzled out. And it was like, yeah, it's cool. People were excited about it, but like the general tenor wasn't like, oh, DeepSeq's destroyed. But on inference max, you can see that GPT-OSS is performing like right at the same level, above on some, below on others. And so it feels like in terms of the like commoditization of the open source software, open source LLM market, like OpenAI has very much caught up and is not like well behind DeepSeq.

1:14:30And so that was kind of an interesting update. I feel like this data, we're gonna get more interesting data points from this. We're gonna get more, just a better understanding of like, not just how things do on benchmarks, like how smart the models are, but actually like if you're doing a big workload with them, like which one is more affordable, right? So that was kind of interesting. but notably absent so it's just AMD and Nvidia right now and when we talked to Dylan he said we're bringing on Tranium from Amazon and we're bringing on TPU from Google they are bigger companies and they're moving a little bit slower and they're not selling a ton of these things yet AMD and Nvidia sell a lot more chips and so TPU and Tranium will come in the next couple months but we're working with them but notably absent was Huawei which we've been tracking and we've heard that, you know, Huawei is not my friend.

1:15:20Huawei has this cloud matrix 384 that competes with NVIDIA's NBL 72, the rack scale compute. And the high level number is that NVIDIA at rack scale is 2.5 times more energy efficient. So the whole pitch with the cloud matrix 384 was that it can do the same stuff as NVIDIA. Like it can run the same models. And it's like, we got it. The problem is we're going into an era where energy is going to be the primary bottleneck. And so there's not a lot of incentive to run ship racks that are inefficient. And also like being 60 % less efficient is wildly different than being 10 % less efficient, right?

1:16:00Like these are really, really important. China would say, oh, well, we have all this three gorges dam. Like we have plenty of cheap energy. It's like for a while. And then eventually you got to build 100 nuclear reactors, not just 10. And so I want Huawei to be on Inference Max. I don't think NVIDIA is going to pay for that because they don't want to have, you know, Huawei chips in there. They might be hard to buy. I don't even know how you buy Huawei chips in America. But I think the U.S. government could get it done. And so I'm calling on David Sachs to put up the money. I don't know if they actually would have to pay this much.

1:16:30Dylan said something like$50 million came together across NVIDIA and all their partners to pay for this. It was a lot of chips. And so I'd love to see America foot the bill, the American taxpayer foot the bill, to get some Huawei chips, give them to semi-analysis, get them running on inference max so we can know the true gap in compute power. And this is important because we know the gap in rare earths. China refines or China mines six times as many rare earth elements as the United States. They refine those almost 99 % of the rare earths in the world. And they produce about 85 to 90 % of the world's rare earth magnets.

1:17:08And so the gap in rare earth is massive. Now, we know that America's head ahead in AI compute and GPU manufacturing, but we just don't know how big that gap is really. And I feel like that's a really important, you know, negotiating point and point of negotiating leverage in the trade war and putting Huawei on inference max would help us understand the nature of that gap and how big it is and how much leverage we actually have. So who knows where the money will come from, but I'd love to see Huawei on Inference Max. And if you want to learn more about it, go over to Semi Analysis and check it out.

1:17:44Anyway. Patrick O'Shaughnessy and the team at Colossus have a new, looks like a cover story, with none other than a friend of the show, Josh Kushner, dropping tomorrow. Yeah. We should hopefully read through it. For sure. Michael Spiker says, real men lose their shirts in natural gas, not trading synthetic shitcoin futures at 50x leverage. What does 50x leverage buy? I think it's just a typo. Pretty funny. Yeah. Tyler, figure out some alpha in natural gas and start trading it. It is interesting. We haven't heard about the natural gas trade in the retail circles or kind of the general. Like all of, you know, like basically everyone in tech got sharp on - Natural gas needs a retail army.

1:18:43Yeah. Just as a category. A couple of years ago, as NVIDIA was pivoting from a graphics card maker to an AI accelerator maker, similar chip, but it was becoming a part of the AI narrative. The AI narrative was developing around NVIDIA. The second phase of that was everyone learned what TSMC was. Then everyone learned what ASML was. Then everyone learned what SK Hynix was because that's the full stack. You design the chip at NVIDIA. You make it at TSMC with equipment from ASML and memory from SK Hynix. And so those became the four companies that everyone knew about. Now people know about the NeoClouds and the Hyperscalers.

1:19:20But there aren't that many people that know who the key players in natural gas are. And, you know, if all the predictions about AI are true, you heard it from Dylan. He said, like, a lot of the energy is going to come from natural gas. And so who are you betting on? I mean, no specific companies, but Leopold has talked about this a lot, where basically if you continue ramping up data centers, like, you need the power from somewhere. And the most obvious place is natural gas. If you can basically get past, like, regulation and, like, essentially, you know, all the big companies have these environmental restrictions on, like, where they can get power from.

1:19:55If you can tear those down, then natural gas is by far the cheapest, easiest way to just get insane amounts of power. Raghav says, Leopold has a crazy position in natural gas company EQT. Yeah, we should dig into that. Only up 46 % over the last year. Congratulations. Leopold wins again. Founded in, guess. uh founded in 1983 close 1888 1888 let's go let's go just waiting 150 years for the ultimate like now we're a meme stock finally real patients uh ajax the great says working at spacex has taught me that caffeine limits are imaginary uh yeah when i was there i had a minimum 200 milligrams per night.

1:20:50That's hilarious that he's denominating it in nights. Like most people would say I have 200 milligrams of caffeine per day. When I wake up, you know, per night. It's like, yeah, because I stay up all night. I start my morning with 200 and usually by noon, I'm at five to 600 milligrams. Today I hit over a thousand milligrams. That's a lot of caffeine. I don't know if you should be going up that high. It's certainly not, you know, you got to work up to it. You got to work up to it. Uh, Riley Walls had, uh, it sounds like he had a great weekend he said the plan at dusk 50 people went to san francisco's longest dead-end street and all ordered a waymo at the same time the world's first waymo ddos you can see pictures of the way obediently lining up uh so he said uh we didn't actually get in the cars they left after about 10 minutes and charged a five dollar no show fee very fair for for getting messed with waymo might have to build in some functionality of like if you collude with a group to i would disrupt our services i would be terrified to do this and be deplatformed from waymo forever yeah it would be very rough riley said waymo handled this well i assume this isn't much different than if a big concert had just ended eventually they disabled all rides within a two block vicinity until the morning this was back in july felt like i was back in middle school everyone was giddy and when another car showed up there were cheers maybe three or four real drivers all laughed and just drove around so good i like jake here saying calling him the tech jester cannot be stopped really is would you call him an internet rascal an internet rascal uh yeah somehow some i i called him an internet rascal and then somehow the team thought that was his like official title it's like i riley walls i'm an internet rascal i like it um well let me tell you about Turbo Puffer.

1:22:40Search every byte. Serverless vector and full text search. Build from first principles on object storage. Fast. 10x cheaper and extremely scalable. Jeffrey Hinton was featured in the Acquired podcast. They did a deep dive on Google. You can go check it out. Jeffrey Hinton, father of deep learning at age 31 with Chris Ricebeck, La Jolla, California. They are so good. The Acquired guys are so good at finding these archival images that just go so hard on the timeline. They look amazing. And David Pfau says, probably my favorite factoid about Hinton's family is that basically all of them were mathematicians going back all the way to George Boole.

1:23:20First time Metis list, first ballot Metis list, eventually kicked off the Metis list for being irrelevant in the modern era, except for one guy in their family who invented the jungle gym. Isn't that hilarious? Imagine inventing the jungle gym. That's such a great invention. I hope you have like a royal that just produces endless cash flow for the family forever. Like there's nothing more enjoyable than meeting a Nepo baby who's like, yeah, my great-grandfather invented the pencil. And you're like, yes, okay, you deserve to be on a yacht right now. You contributed to society. Imagine just looking at this like monstrosity of just like metal and like traps and you're sitting there.

1:24:04What should we call it? The jungle gym. jungle gyms bring incredible joy to children all over the world it's one of the greatest least controversial inventions ever there's no jungle gym doomers that's true there's no jungle gym accelerationists there's no oh jungle gym costs causes brain rot narrative there's no hit pieces on jungle gyms they are on a they're unfiltered they're raw i'm sure i actually i would say i bet you we could find a jungle gym because i'm sure they just massively increase injuries annually at schools but worth it you know what doesn't increase injuries at schools profound get your brand mentioned in chat gpt reach millions of consumers who are using ai discover new products and get a demo uh shiel friend of the show says he's at sea ranch lodge for lunch it's very beautiful and it's owned by patrick and john collison and justin khan very fun look at that corgi this is a cool i had no idea that they snapped this up i knew that justin khan spent a lot of time yeah up in that neck of the woods i have been to sea ranch it's up uh up around where i grew up so they have principles for the sea ranch lodge uh yes nature predominates rural matrix no vest pocket nature as at carmel or bodega i don't know what that means.

1:25:31Yes to rural, no to suburbia. Yes to community, no to individual houses. Yes to aesthetics, no to up for grab aesthetics. Yes to design control. Getting a text of saying there's not just a jungle gym hit piece, there's a whole book covering it. He wrote a book on the entire jungle gym. Trying to take down big jungle gym. Wow. I will go to the mat for the jungle gym. We will defend jungle gyms with our lives. Yes. Not only because they are in the Hinton line, but also because they're good for the world. They say it's not elitist. They want modesty of house size, reforestation, native trees, common facilities.

1:26:16Very cool. Wait, they want elitism in large house sizes? No, no. I'm kidding. Elitist in enormous houses. Yeah, wait, wait. I want the one on the right. I want the one on the right. oh they're taking shots at malibu they said they do not want to relinquish access malibu type oh colison and con teaming up well did this these principles look like they predate them taking ownership yeah unless they just like made it to try to look like a vintage piece of paper before before growth equity secondary dollar because there's so much there's so much lore around there's so much lore around sea ranch i don't know any of the never heard of sea ranch before there's go you You can go down a YouTube rabbit hole around Sea Ranch.

1:26:57Yes. How conspiratorial does it get? Not conspiratorial at all. It's just like a very principled development and community. And they've done. Or now. Or now. We can bring conspiracy to it. Imagine you're just in this like idyllic, you know, cabin. And you go to open the mini fridge. And it's like, tap here for one click check. I'd like that. I'd love to have a chicken. In other news, Upper 20 Street Capital, they're running out of phrase capital on X, apparently. Somebody is writing, Mistakes, I wrote about investing in Tiny, an eclectic Canadian investment holding company just last year. Tiny is the company founded by Jeremy Gaffon, right?

1:27:45No. Some people seem to think that in the comments. Whatever happens with Tiny, we should hold Jeremy to account, right? Absolutely not. early team member, potentially one of the first hires, but not his firm. Somebody in the comments, I did notice, thought that Jeremy started Tiny, and that's probably because he's had one of the most iconic podcast appearances in history. But this person is writing, I wrote about investing in Tiny, an eclectic Canadian investment holding company just last year, concluding that I hope to add to our shares over time and enjoy the benefits of getting in on the ground floor of a tech-oriented compounder for years to come.

1:28:25Well, that was a mistake. Tiny has performed wretchedly, not just its share price, which, if it had gone down independent of how their businesses performed, would not have bothered me and indeed would have afforded opportunities to buy more. But no, their fundamental financial performance floundered from the outset and have shown no signs of improvement. Net income and cash flows have dried up completely, and what little is left is going towards interest payments on$100 million in debt, some of which they violated covenants of, resulting in their lender asking for accelerated principal repayments.

1:28:55In increasingly dire straits, they sold 4 % of the company to a private equity shop for$20 million Canadian, which at today's share price makes it look like Tiny is the one getting the last laugh. The final nail in the coffin was the behavior of one of its co-founders, the majority shareholder. The man has simply acted like nothing's wrong at all. He remains active on Twitter and appears to spend inordinate amount of time on podcasts, both as guests and hosts, in the guise of an entrepreneurial guru, often introduced as, quote, billionaire, even though a cursory glance at the market cap of Tiny make that a farciful impossibility.

1:29:29He's essentially taking victory laps while his company faces a debt maelstrom, and his shareholders circle the drain, and then he sets up a plan to unload millions more shares. So really rough. I mean, the nominative determinism on Tiny was always just so brutal. You want to start a tiny holding company. Well, it's a, didn't they start with like a design shop, right? Yep. Do you know that there's another design company called Huge? How are they doing? I think they're doing great. They're up bigly. They're in Brooklyn. They've been going since 1999. Anyways. They have a thousand employees. They're doing great.

1:30:09They got a global CEO. They have UX school. They're very good at user experience. Yeah. I always thought it was brutal uh i've enjoyed andrew wilkinson's uh posts over the years but i always thought it was brutal that he released a book called barista to billionaire but by the time he was publishing the book it was clear that he was no longer a billionaire barista to billionaire and back again why not isn't that the lord of the rings round trip the lord of the rings uh books. No, Tiny does have the AeroPress. Oh, that's a good company. You can still always make coffee. Well, let me tell you about Linear.

1:30:46Linear is a purpose-built tool for planning and building products. Meet the system for modern software development, streamline if-use projects and product roadmaps. And we have our second guest of the show, Alexis Ohanian. Welcome to the stream. There he is. How you doing? What's up, gentlemen? What's up? It's been too long oh thank you guys i i man i just can't believe in like a year yeah you've dominated all things linkedin media empire i i hope it looked like that is the cornerstone of our empire it's it's where we think first but you you saw that you saw the vision you saw the vision before we had done any guests before we had gone live like very very still dying to invest guys you just happen to make the most profitable podcast ever.

1:31:35We manifested it. Well, if you want to invest, why don't you go turbo long Microsoft? They own LinkedIn. We're going to make LinkedIn amazing. It's going to flip every other social network. You'll make a bag. Not financial. I like the early stage guys traded stocks. It doesn't, doesn't stir my cocoa. Well, you need going to use a thousand X leverage then in the public market. You need to be able to make it, you need to be able to make a thousand X or lose it all. Otherwise, otherwise you just don't get the rush. What, uh, what, uh, what about, uh, you could invest in a Microsoft treasury company.

1:32:10You could buy a seed state startup that just holds micro holds a thousand X levered Microsoft stock. And then you get the exposure, you get to hang out with the founder, you're on the board, and you see the same financial upside. It can all work out, maybe. Anyways, it's great to see you. What's new in your world? Dude, well, you know, it was very kind. I think for my second appearance on the show, you let me plug my women's track and field meet now team-based league that's coming, Athlos. We just had our event in New York. Nice. Dope time. Over 3.5 million people tuned in this year. Whoa, that's big.

1:32:50It was a lot of fun. Thank you to the internet. We went wide on the distribution, so you could find it anywhere. YouTube, X, ESPN, DAZN, didn't matter. But it was cool. And then, I don't know, I was really coming on in part because I wanted to congratulate y 'all. It's been awesome to see the ride. I've had the privilege of separately funding or seeding or, you know, playing early investment with y 'all separate companies. Yeah, I know. It's just been so awesome to see the trajectory, the execution, the vision. And congrats. Well-deserved. I know it's just the start. And this is just me buttering you up so that one day I can invest in the tech.

1:33:26Some days. On the early side. By the way, yesterday I was trying to hang out with a buddy. He could not. He was at the Angel City game. Oh, really? No way. Fantastic. Yeah. What's the update on that front? I mean, we're knocking on the door. We might be able to get in the playoffs here. It's going to be – it's a little tough right now, but I'm hoping we get in there. But that was probably one of my favorite, like – I was like March of 19 when I rage-tweeted about how undervalued women's professional sports was. And in proper Twitter fashion, most people said I was an idiot. No one cared about women's sports, but that obviously has done pretty well.

1:34:05And, yeah, thank them for being an Angel City supporter. I've got a couple extra spots in the suite. if you've got, there's a couple of listeners who want a couple of tickets. You have to figure out a way to give it away here for the last home match of the season, but I'll gladly. We'll just run a raffle on LinkedIn. That'll be the way to get it. That's where everyone is. That's the new alpha.

1:34:29Also, Stokespace, they announced like a half a billion in new funding last week. Yo, believe it or not, so five years ago, first space tech investment I've ever done. And there's one shot at space. Well, I mean, come on. It's one of those dangerous first investments in a sector because then you end up thinking you're a genius at it. But Andy Laps and the team are doing a hell of a job. And look, obviously Elon Musk has changed the world with SpaceX, but I think we would all agree a little competition is a good thing. And Andy's got a vision to do these fully 100 % reasonable rockets and the first launch is going to be at Cape Canaveral this summer.

1:35:10And very, very proud of that team. I'm just grateful to be a little part of that. You'll have to be there. You're going to go in person? Oh, yeah. Dude, I'm bringing the whole 7-7-16. We visited. There's photos. There's nothing like watching one of your startups do a launch. And just actually, you know, but in the case of space, you get a really tight feedback loop. At least if somebody launches like a SaaS company, it's like, oh, how do sign-ups look after the first 24 hours? This is like, how does the rocket in one piece after 20 seconds? You know, you know whether or not it worked. On the early stage side, I'd love to check in with you on just how you're feeling about AI, because we had this foundation model war.

1:35:57It seems like that's cooling off, like Mark Zuckerberg's poaching different co-founders of different teams. but it still seems like there's a bunch of fertile ground in the early stage, not even call it an AI company, but just a company that's like pulling AI off the shelf. And I'd love to know kind of how you're thinking about the mobile era, like where the pockets of interesting value come from, what excites you when you see a pitch? Because I imagine you're not getting, maybe you are, but I imagine you're not getting excited about like, yeah, we're going up against open AI. We're going to do everything that they're doing, but better.

1:36:34And it's like, that ship has sailed. So where are the ships that are leaving the port right now? It feels like, okay, I'd say the companies, there was this existential crisis probably two years ago where some of these 21 and 22 vintage companies that we had seeded were like, oh shit, like the world has changed. And a lot of the, I think the smartest ones, frankly, pivoted hard to decide like, how do we sell picks and shovels? How do we find ways to just get involved in this? And we've seen a couple, they won't even let me talk about them because their revenue is going so well. They're like, don't tell people about it.

1:37:12But there's a type of company that is selling to all those foundation models. It's basically like, throw money at this problem. We don't need to be expert at it. We just need X. Maybe we need training data. Maybe we need whatever. Here we call it the barnacle economy because the whales are getting so big that even if you're a barnacle, You can go from zero to$100 million in a couple of years. And it's like a good way to kind of like get started. You're hitching a ride on a rocket ship. You're riding a wave. And then you go off and kind of source other customers over time and establish your market position.

1:37:46I love Barnacle economy. That is a good meme. The other one that's starting to peak now that you all would have great instincts on as well is that it's that whatever the application layer. It's like, how do you build the stuff that's actually dope for the end consumer that, you know, assume some of these larger models are going to continue to just be a solution for a lot of stuff. But like, let's say you want to see how you look in a Canadian tuxedo. There you go. Like, this is doji. Yeah, yeah, yeah, yeah. Like, if you can make a delightful user experience. And for a consumer social guy, this is where I feel like now a lot of my product instincts can start to shine because now I think we're almost about to see consumer social get fun again and get good again.

1:38:29People are realizing y 'all are – this is going to seem like I'm just kissing ass, but y 'all prove – let me continue. You all prove the point that so much of the internet is now just dead, this whole dead internet theory, right? It's not my idea. Whether it's botted, whether it's quasi-AI, LinkedIn slop. like having proof of life, like live viewers and live content is really fucking valuable to hold attention. And, and so I think we'll see a next generation of social media emerge. That's verifiably human. Cause like, because it's all going down in the group chats now that can't be, that is not novel tech.

1:39:03There's gotta be some next iteration of that because that's where all of us are getting our like really best info now. And, and similarly, I think we're going to get really delightful, fun consumer experiences where, you know, some scrappy founders in Brooklyn, like the doji guys can say hey let's make shopping fun again using this tech and and even if you know the googles of the world launch some janky try-on experience like we know those companies are just terrible at innovating product and are really a threat yeah the thing i've been thinking about is is uh is social is what we think of as a social media platform even harder to build now when you think of you're going up against generative media platforms that, you know, like Sora and like Vibes that are basically saying we'll spend hundreds of millions of dollars on compute to try to bootstrap this network.

1:39:54It's like the last generation of even the TikToks of the world was like, okay, you can compete against American social media companies. You just need to burn billions of dollars. But it's way different than you with Reddit back in the day where you're like, okay, we need a little bit of cash, and we need to clone ourselves a bunch on the platform. Maybe an AWS credits or something would get you going. But if you guys had needed$100 million to burn on compute, it wouldn't necessarily have been able to pull that off. $72 ,000. That's all we raised. It's a size gong for... Do you have a tiny gong?

1:40:29Yeah, we need a baby gong. No, no. Massive gong for scrappiness. and then 60 grand from an angel check from paul graham i actually you got 12k from yc that was the first yeah for seven percent of the company it was worth it but that is a different time if you show up this this day in history uh intel ipo'd and raised six and a half million dollars i mean even inflation adjusted yeah it's so small it's but anybody that's and that's oh god okay That's the other part that makes me a little nervous right now.

1:41:07Everyone's seeing the rounds. Everyone's seeing the markups and the valuations and stuff. For still fairly early stage companies, the founders that can maintain their discipline with a ton of cash sitting in their bank account are going to dominate in such a great way. Because we know you aren't going to need to blitz scale like a decade before. You have no excuse for trying to get hundreds of employees. When in reality, if you're a pure software company, you want to build the Navy SEALs. You're having dozens who are going to be able to get you on the revenue. And so if you can spend that money intelligently, obviously, you've got to drop a lot of it into compute.

1:41:38But I think it'll be interesting to watch because the blitzscaling model didn't really make many people happy. There are many CEOs who are like, God, I really loved when my company had 1 ,000 employees. As long as these guys and gals can be good stewards of the capital, I think we can keep this party going. It's just a lot of companies raising a lot right now early. What was your takeaway from the whole TikTok deal? I know you were in the mix, and it ultimately ended up going for well beyond what it feels like the intrinsic value of the asset is, especially in comparison. It's like, would you rather own Perplexity at$20 billion or TikTok at$15, right?

1:42:24we're snap at seven yeah exactly like it feels like a big platform i i am still uh dms are still open if they want to get me in the mix i do think look i i had a hunch this this to me always felt like the kind of deal where you knew you were gonna get it like if you knew you were gonna win it you were gonna know a minute ago sure uh yeah and and that is what it is i'm still happy i've been publicly talking about how problematic tiktok was for years and years and years as a vehicle of the ccp so i'm i'm happy it's an american possession now um i do think look there's still there's still a lot of value to that zeitgeist to to affecting that and i don't know it'll be interesting to watch i mean i did just say i think social media is about to get upended in some interesting ways but i think with the right leadership look it's still look it's still meaningful is all hell and with the right leadership, it can continue to endure in this new era.

1:43:23This is a fascinating question because TikTok feels like for the last maybe decade, almost like they, it has been the engine of innovation for social media in the sense that the first like hack of like, let's actually use licensed music or a lot of the cap cut stuff and And a lot of the, oh, remove the background, automatic editing tools. A lot of the collab features, live streaming. Like, they were definitely pushing the product design very fast. I wonder if the last two years have just been, like, total chaos internally and, like, basically frozen product development. Because it's, like, not an exciting place to work.

1:44:00Everything's going to be under crazy scrutiny. So, even if there's some innovation, like, oh, like, instead of just like and retweet, there's, like, a third button that's going to work really well. Like, that hasn't rolled out. And where I'm thinking this goes is that, you know, we heard OpenAI was going to do a social network. They did Sora. That was like somewhat telegraphed. It makes a lot of sense that Meta would launch Vibes, something in the AI social network, AI TikTok. But TikTok should launch AI TikTok, right? And they've been fantastic at machine learning and AI. They've had huge data centers running these incredible recommendation algorithms.

1:44:35That's why the TikTok algorithm feels so creepily accurate. You swipe five times on a car and it's like all car content for the next week. And they're really good at understanding like you like this car instead of this car. And that's because of AI and their machine learning chops. But they haven't brought that to bear in generative AI yet. And it seems logical. But I'm wondering if there's still a team there that would push that forward or if they'll just be way more reactive. You bring up a really good point, especially because we've seen this creep up. And it felt like part of the advantage to building in this space is you're right.

1:45:10I mean, when we talk about the algorithm, we're talking about TikToks, right? The effectiveness, like you said, that they were able to use machine learning to send you more of the stuff you want to keep swiping. Does feel like they've been frozen in time the last couple of years. I think you could get, look, it is a large enough property with enough cultural influence that I think you get some really talented people to come go build on it. And look, Andreessen and company, they know how to attract talented people to come build on stuff. I wouldn't count it out yet. The part, and this is, look, a big reason why I'm still spending so much, why I continue spending more and more time in sports, investing in sports, building sports teams, league, et cetera.

1:45:45I think it's the last bastion of content and attention when, you know, 95, think of the future of Hollywood, the entertainment industry, the music industry. It's going to be largely AI generated or assisted. The appointment television, aside from live, you know, breaking tech and business news, sport is like guaranteed. and the celebrity of being an athlete also indoors, even in the age of AI, because why you're famous is not because people know about you, it's because of the stuff that you do. And so there's got to be, I think whatever that next platform is, is either going to feel deeply intimate and human.

1:46:21So it's the better version of the group chat that we all have. And I don't know if maybe it's another level of anonymity, decentralization, I know the crypto bros are excited about that. There's potential there. or it leans more into live, especially here in the West. I feel like whatnot and others are still kind of scratching the surface, whereas in China it seems like a much bigger part of the culture. So I don't know. But I'm hopeful we keep seeing more innovation here because as a consumer guy, I think we've been starved for a little while. Yeah, totally. There was some recent reporting that time on social media in the financial times.

1:46:58Time on social media peaked in 2022 with young people cutting back first. Is that something, did you ever predict that? I mean, they're kind of, you have to factor in, I feel like COVID a little bit here. People going back to work, spending time on. It's like remote work. It's really just people remote work ending and people like, okay. Yeah. You literally go to the office and like TikTok is banned on the wifi, so you can't use it. I like, okay, I like that we're beating this up. I definitely quote tweeted it because it validated my own worldview. And so I was like, well, obviously this is right.

1:47:33I think, I mean, look, okay. Anecdotally, I should actually just look at this. I'm, I track all this shit. So I could probably just look at even just on my mobile phone time spent. I don't know how far back that goes in that app, but okay. Anecdotally look at, look at online dating and look at how, I mean, thankfully I'm married because I don't run, but like run clubs have become this new version of how to meet people. Um, especially for younger generation, I got tired of swipe culture. It definitely feels like there is a younger generation of folks who are using the internet to find ways to commune offline, whether it is events or run clubs.

1:48:13It feels like there's a, there's a bit of a culture shift. That's like, did you say you're married? So you don't run? No, I'm saying I'm married. So he doesn't go to, it's also helpful that I also don't run. I like the idea that the only reason people run is to meet people. That is the, like running is just dating. I think part of the damn expansion. It really has been. No, no, it's a good point. No, but guys, guys, we can break, I can break down why this is. So I'm very lucky. I know Whitney fairly well from Bumble, but also the OkCupid guys from back in the day. I'm really dating myself. The secret of online dating, assuming straight dating only for a second, is high value women get inundated by messages from guys.

1:48:50And so if they have a bad experience, they churn and your app fails. And so every successful dating website is on some level away to make sure that they're having a good time. And what's interesting about run clubs is it creates this social dynamic and a sort of physical real world one where if you literally cannot keep up. And so there is an interesting dynamic there that I think, again, I've taken a few of these pitches. the the run club app the the dating app for real life like there's there's versions of this that are foaming up and so i want to believe i want to believe the burnout is real and then you combine it with just again the fact that so much of the stuff we're seeing is just it's to some degree botted or fake yeah i mean i'm meeting i'm meeting founders let me give you a perfect example okay Okay, so, and this is something I wouldn't be surprised.

1:49:43Okay, so CPG days. You know how important having a subreddit is to the successor product, right? Huge. I got about nine months ago, I heard from a founder. I won't say what the subreddit is, but he told me, he's like, oh, I just bought the bread making, r slash bread making. I just bought the bread making subreddit. I was like, I didn't know you could buy a subreddit. He said, oh, yeah. Is that against TOS? I imagine that's against Reddit TOS. We definitely, but we found the seven moderator accounts and we bought each of those accounts. And so now we own those accounts. We control who becomes a new mod, so we own that subreddit.

1:50:15And so very quietly every day, it's not r slash bread making, but r slash bread making, we make sure that there are posts that do well that are like, oh, I love my blobby blob bread maker. Yeah, this particular bread. Yes, because the value of the GPT sort of training the engine, what is it, GEO now, whatever. Yeah, GEO. GPT optimization, the SEO of today, of the AI age, it's so valuable that there's such an economic incentive now and you know if that's what the cracked founders yeah are doing and i just learned about that six months ago eight months ago surely it was happening even before should we and yeah it's wild it it it makes the underpinning of so much of what we consume online you know very much at risk again which is why stuff like this starts to feel a lot more real and tangible should we buy out all the mods of r slash communism and start pilling them on capitalism we would have to i mean it's it's it's interesting already for probably i imagine post luxury watches i imagine they would be very principled if you tried if you offered ten thousand dollars they'd be very principled they'd be like there's no price that i would you know sell my account you offer them a million dollars actually 20k might do it what do you think about uh i feel like there's been an idea i haven't i've i've seen some pitches over the last couple years of people uh selling this vision of like instagram but it's just you and a million bots or and and so that feels like the social network the ai well it feels like a social network that you don't need a you know a billion dollars like to start like something like a sora it's like very expensive not not any old team can just build that you're absolutely right, Jordy.

1:52:02Thanks, John. Thanks, John. It's not this, it's that. But, but, but, and I think the reason that I think is somewhat interesting is people can say bots are a feature, not a bug, right? Like X clearly doesn't take bots as seriously as maybe they should. And I think part of that is probably just brings back, it's hard to solve, but it also brings, it's like a notification, right? It brings people back to the app. There's a reason to check the app, even if you're not getting real engagement. But I've generally been bearish on these sort of just like entirely bot networks, because I think people still want to feel like, even if they're getting a notification from like, an account that could be a bot, they're still running the calculus of like, oh, maybe it is a real person.

1:52:47I don't know. Yeah, it would have to be the utility of the bot, the why of the bot would have to be so great that no human could reasonably do it. So Reddit had a version, a third-party version of this was like the remind me bot oh yeah where you would just tag in remind me and someday some amount of time it will message you back so like there that that isn't right no no reasonable human would keep track of all that bullshit great use of a bot um there there there could be uses for that but it would really like in each one of them it would have to be so where they can't just be sicko fancy like it has to be like no this actually adds some value that no human could reasonably do i think for me to get excited about it because people can still find the sycophancy one-to-one.

1:53:29Do you really need a hundred or a thousand? And frankly, probably 10 % of humanity will fall in some way in love with or in obsession with some kind of an AI. In the same way we lost people to MMORPGs or what have you, there's a version of that to exist, but I don't know about the one-to-many. At the same time, it feels like about 10 % of the population maybe will do the exact opposite. And instead of getting lost in the hyper futuristic infinite jest, they will just go back to like logging off complete retro stuff. I want to know two things, the general take on, on retro electronics, throwback websites, throwback tech.

1:54:11And then I want to know specifically, uh, what game are you playing on your mod retro chromatic most often these days? Cause you've been posting a lot about it. dude um i don't know if we totally announced it well uh okay i i'm gonna have to disclose we're also an ria so now anything i say so i mean it is a 776 company we're very lucky to have invested with torrent and palmer and it is i am an unabashed gamer it's it's i have i mean i have graded video games actually graded video games oh like uh like like older copies that are kept in the box look at this guy that's cool original game boy yeah wow like this is my art wow right yeah yeah yeah this is i coveted this as a kid we could not have yeah i've long had this thesis now dad i've long had this thesis that i i don't know how it'll play out but like uh in many ways uh video games can be the new golf for a generation of young people that grew up playing competitive video games and then they will like play games with their crew, uh, like online.

1:55:13And like, it'll actually like kind of be this place where people have business discussions in the lobby of Fortnite in a few years. Like once they get older. So three years ago, two years ago, uh, it was an early, a long time ago. I was an early investor in Roe, Roe Health. Z Ray Tano, founder and CEO. After our board meetings, we play Fortnite. So we'll, this has been going on. I think he does it with some other CEO. I heard some hilarious story about a kid who runs a company and does all his meetings in COD. He's like, no, no, seriously. I mean, he's like some, like, I don't know, crypto founder or something.

1:55:46And like, he's a very small, like lean company. He's just like, oh yeah, like let's just hop on Rust if we want to like hash this out. But first, I do want to know, top mod retro game. I got Tetris, obviously. I got a few other games, but what should I get? Here you go. Self-simulated. It's a platformer. It's a lot of fun. it involves your robot dying and an infinite number of its clones having to make its way out. I may or may not be working on a game of my own. I was about to ask. Which could be ready for the holidays. Because if the mod retro chromatic install base gets big enough, you could potentially develop new games and it would probably be pretty cheap to develop because you're not trying to do GTA V with networking and stuff, right?

1:56:24Like you're actually just like, you could maybe even vibe code it a little bit. Like it's pretty doable. That is exactly right. I mean, this is an indie game and what's crazy is the amount of money I'm spending on this game is tens of thousands of dollars and that's everything and that's for basically developer designer a little bit soundtrack music stuff like it's it doesn't take gta6 money like that's that's accessible for just like an artist with like a couple backers there's gonna be a new crowdfunding boom potentially you could get new kickstarter stuff going actually delivering stuff that's super cool and and what i love about the constraint is similar to a great tweet right you're limited on characters you got to make each one count you're limited on pixels you make each count and it forces you to think more about gameplay and do weirder, interesting stuff.

1:57:06Like there's a Sea Shanty, like Guitar Hero style game that I actually need to try. I like that. Where it's like, that would have never been greenlit. Soon, man, well a man come. I love Sea Shanties. It took the cost to be reasonable enough that you could get experimental and turn around a really high quality in a game. So I'm bullish on it, especially because, you know, every new Call of Duty, every new battlefield like the graphics get marginally better but it's diminishing returns right oh the buildings are now 99 more destructible that's cool but it doesn't it's not a huge step function like will i buy the new grand theft auto when it ever comes out of course of course but but there's there's room for this kind of simpler format that uh it's it's very dope and you know palmer's not stopping there uh so if you've seen any of the tweets there's some very cool hardware on the way as well.

1:57:57Yeah, I think he's on the schedule for later this month. Last question, what's your framework and view on AI companions broadly? It feels like one of those things that everybody has kind of a moral framework around investing into. Elon obviously ran the calculus of it feels like a market that a lot of the other labs didn't want to lean into heavily. he was willing to lean in. He went there. He went there. Because I'm sure you've gotten pitched probably hundreds of pitches. You haven't tried Claude Sexy Mode? This is where... This is definitely where I feel like there's probably... The market's not necessarily going to regulate itself.

1:58:47This is clearly one of the end games for AI for some percentage of the population. I'm not like jumping out of bed to fund that company. And, you know, thankfully Elon's doing it with those hentai, uh, uh, characters, you know, but, um, but I do think, okay, the version of it that I think is very interesting to me, there's, and I've been using, I mean, maybe it just ends up being chat CBT or anthropic. Like I use them as a, as a kind of executive coach in real time when I'm looking for feedback on emails or just my own sort of notes on stuff like there's there's some version of that that i think is interesting and compelling i like i i've been actively looking for the version of it that is kid friendly but not you know and there's been some interesting ones that have gotten funding i don't know i like because the way i use ai with my kid right now is it's just the always-on tutor that we can sort of ask questions to or we'll provide trivia questions um for us that's a good use case and fun and it's And we do a little creative storytelling, story writing, but it's always with me because she's eight, right?

1:59:52I still – I want her to think of this as a tool that's going to give her superpowers to solve any problems she needs. I'm excited to see – here's another one that I think the Palmers of the world will be thinking about is if you're focused on consumer electronics like consumer hardware. the way this generation, this AI native generation is going to interact with UI and UX is different from us. And it's the same, it's the version of the kids swiping the magazine back in the day. Oh, yeah. Because they thought the iPad was broken, but it's a paper magazine, right? Because that form factor was so native to them.

2:00:25There's an AI first, almost a voice first version that I'm expecting. Like, because even I watch my daughter when she gets on her iPad for iPad time, like she usually uses the microphone to dictate text into the prompts or what have you, instead of typing. Like I'm trying to get her on the whole QWERTY thing. But again, if I'm realistic, I'm like, well, this is way faster. These things, like literally, these things were designed to be slow so that the keys wouldn't jam on a typewriter, right? And all the Dvorak keyboard people are like, we told you, finally, we're validated. But I actually think this should feel like a relic from a bygone age if we build the right interface for it.

2:01:03So I think voice is going to get more interesting. And I'm looking towards the younger generation for just helping us get ideas out of our heads faster. Yeah, the combination of kids' toys plus LLMs is interesting, too. I'm sure they'll be jailbroken in terrible ways immediately. But if you can combine actual IP. Yeah, but we never did that on the internet in 2000 either. Yeah, if you can combine novel IP that's a hit plus the LLM layer where the kid just doesn't love a certain plushy thing. I mean, super underrated that OpenAI has a deal with. you know, we're talking about Broadcom, NVIDIA, Oracle.

2:01:41They have a deal with Mattel. Like OpenAI and Barbie are in business together. Like, let's think about that for a second. And Ken. It's going to be fun. Ken is AGI-pilled. Yeah, you're going to be able to go to Ken and Barbie, AGI. It'll be fun. Thank you so much for stopping by. This was a great time. Great to catch up. Appreciate you guys. Keep up the great work. Thank you. You too. Thanks for having me. We'll talk to you soon. Talk soon. Speaking of Palmer, massive news out of Anderle today. Today, Jen Bucci shares fit check and shows an insane design for an augmented reality, virtual reality.

2:02:14Is this a real picture? Except for the warfighter, I believe it is. Wow. Reggie James comments. He says, I, for one, love that national defense and intelligence are just founders fund incubation projects. Anderil had the full story. They said, superpowers for superheroes. Superheroes today they are unveiling eagle eye the family of warfighter augments that place mission command and AI Directly into the operators helmet very cool the lineage of this program is wild Microsoft was working on it for a while and roll got the contract Yeah, but still they do full page print ad in the journal for this would they or did they already did?

2:02:53They did fantastic pull it up and Garcia capital says this isn't a call of duty screenshot This is Anduril's eagle eye. Palmer's been on record saying that he thinks that the Warfighter will be adopting virtual reality headsets or AR headsets before the consumer because weight, cost, and you can just mandate, hey, you got to wear this thing to do your job. Well, and you can say we're going to spend$50 ,000 on each headset versus meta being in a position where they need to make a headset for like$300,$500,$600. they can sell for$800 or they're probably losing money. I don't know. So the willingness to pay to save a life on the battlefield is way higher than watch a movie or play a game.

2:03:34What do you think of the new headset, Tyler? They're so sick. How do we get one? We've got to get one of these. How do we get a demo? I'm looking through. They've posted some other images of what it looks like. It just looks like Call of Duty. There's a mini-map. That's awesome. When you have a gun, it shows where it's aiming at. That's crazy. It's sick. Call of Duty Infinite Warfare is now here. Very excited to see more demos of this as it rolls out. Congratulations to the entire Anduril team on the progress. The lineage there is fascinating. And I always had this thesis that Palmer and Anduril was the perfect company to do this project, originally called IVAS.

2:04:15And IVAS was kind of languishing within Microsoft for a number of years, and it made sense that Palmer has the VR. Were they getting paid like a billion dollars a year? It was some massive contract that was getting. Yeah. Yeah. I think it was like a 10 something billion dollar product or a project or scope of work. But I don't know how much they'd actually drawn down on that because I don't know how much they'd actually delivered against the contract. There was still a lot of value to be paid if you could deliver. But Microsoft, I think, wasn't quite delivering. And so Anderil bought the contract or acquired the contract or merged the contract in and then now is running at it much harder.

2:04:54And I mean, it seems like a really, really solid progress, which is exciting. So if you go over and you get a Chromatic, you're going to have to pay sales tax. The Chromatic team, I'm telling you, get on numeralhq.com, sales tax on autopilot. Spend less than five minutes per month on sales tax compliance. More news on rare earths. Elad Gil, who came on the show. Narrative violation. Oh, yeah. Abundant earths. Abundant earths is quoting Ben Thompson. And he says, the U.S. deficiency in rare earths isn't a matter of needing to catch up to technological excellence. It's needing to simply have the will, incentives, and regulatory regime to do what we already did before.

2:05:33TLDR, they are not rare. We just chose not to make them in the United States of America. The NRC of old hurt multiple aspects of U.S. resilience in both energy and manufacturing. Hopefully the current NRC can help change things for the better. Now, the steel man for not making rare earths in America is, and I think we now realize that it was a mistake, but I do think it's important to understand the rationale behind not doing mining in America. Like, it is dirty. It is gross. Like, if you have a mine upstream for you, there's probably going to be some pollution. If you're worried about microplastics, you're also probably worried about molybdenum in your tap water.

2:06:19Forever chemicals, all these different things. It is a very rough job. You know, the famous like coal miner gets black lung. Like there was a reasonable rationale for not doing this all over the United States. But the true answer was always more technology. It wasn't, hey, mining is dirty. Let's not do dirty mining. It was like, let's figure out how to actually clean it. Like there is a way to, it might be a little bit less economical, but there is a way to put a mine somewhere and have the chemicals contained and not destroy the environment. And that always should be the answer. Just like with nuclear power plants, design one that doesn't blow up.

2:07:02That's the answer. The answer is not no more nuclear power plants. It's design one that's safe. And so hopefully we can get there and hopefully we can reshore it. There are a couple of companies that are working on it. MP Materials, I think, is one. There's a few others that are, you know, working on rare earth manufacturing and processing. Yeah, the challenge is trying to massively increase production capacity while making the process much cleaner. Yes. You kind of need to choose one typically. Yes, there is a third option, though. The third option is to recycle the materials that we have. So there's a public company and then there's Redwood Materials, which was founded by J.B.

2:07:41Straubel, who was, I believe, co-founder or at least on the board of Tesla for a number of years. And his whole pitch is we already have a ton of rare earths. They're just locked in depreciated electric cars. Let's rip them up, reprocess them, and then recycle them because the actual matter is conserved even after the battery is no longer useful. And so you can separate all those out and then reintroduce them into the supply chain. And so Redwood Materials has been on that track for a number of years and hopefully is doing well because it's more important than ever. What else do you want to go through in the timeline while you pay?

2:08:20Pegasus says OpenAI starting to give off FTX vibes. What do you think caused this? October 13th? This was this morning. TBPN appearance? No. Yeah, I... Extremely bearish for us. I don't know. I mean, so FTX was getting unbelievably large while telling the world that they had like five engineers and they were running a hedge fund that was actively, effectively trading against their user base. Yep. They, they're, I don't know. It's, it's, I think. I think it's a rough pattern match. It's a rough pattern match. I mean, first off. It's not to say you can't find ways to criticize OpenAI and criticize Sam's deal making and all that kind of thing, but it feels wildly different.

2:09:16Yeah, for sure. I mean, the FTX narrative was so, so fake. Sam is not like trading. He's trading with the subscription revenue that you give him. But it's wildly different than, you know, millions of people holding deposits with a financial institution that's half onshore, half offshore, and, you know, making wild, degenerate bets. Totally. Even just the aesthetics, like the vibes of SBF, where he was doing like those TikTok dances and his whole brand was like, I'm this like 20 year old billionaire. It's like Sam has been - I drive a Honda Accord. I drive a Camry. Yeah, yeah. Sam will straight up tell you he went on - He likes a fancy car.

2:10:01And yeah, so way less fake. Sam said he'd never buy a car for a quarter mil. Exactly. He got it. And also like, yeah, this whole, the whole SBF thing was just so much faster. FTX ramps so much faster than OpenAI. OpenAI, I mean, started, what, a decade ago? Like, they really have been working for a long time. Also, Sam Altman has a way longer track record in Silicon Valley, like investing in big companies, growing them. He's been, you know, more or less correct on, like, the energy question, on the SaaS question, on the financial platform question. One way to pattern match it is that they're both CEOs that happen to make some pretty good investments.

2:10:42And their name's Sam. Their name's Sam. Is that a coincidence? Draw the red string. Red string. Red string time. I mean, they are actually more linked because isn't the, like, open philanthropy, effective altruism stuff linked, right? Tyler, back me up on this. What's the link? I mean, I know a lot of, like, EA funding was mostly from, like, FTX. Like, it totally destroyed, like, the ecosystem. Yeah. And I think that's maybe when they actually changed. I don't know if it's the same organization, open philanthropy or EA. but there was something where basically FTX crashes. Yep. Everything is kind of wiped out.

2:11:18Yeah. Including like a lot of EA. But in fact, I think a lot, there would be a lot more string between FTX and Anthropic, right? Than FTX and OpenAI because a lot of the EA is like left OpenAI to go over to Anthropic. But also like that's a very real business with like, and it's like, yeah, it's just like what would be the, like the nature of like selling tokens for subscription revenue is just not the same as being a hedge fund that can like blow up on one bad trade. If like you're over levered, it's just like you can get over levered if you have a lot of debt and stuff, but the debt's not found.

2:11:53I don't know. It's a stretch, but it's certainly a viral post because there's a lot of people with strong opinions. Everyone has like a favorite foundation models, company CEO right now. You can't just be a fan of all of them. Like we are. You can't just be like, I like capitalism. I like business. I like all of them. It's a good. What's your take on the AI labs? I like the lab. I like them. Yeah, we support the foundation model CEOs across the board. Like we support fin.ai, the number one AI agent for customer service, number one in performance benchmarks, number one in competitive bake-offs, number one ranking on G2.

2:12:30Watching my friend's kid and immediately put him on the drugs. What is this? Miss Rachel. I don't know who that is. You've never thrown on Miss Rachel. I've never seen Miss Rachel. I think I saw her mentioned in the Wall Street Journal once, but that's really the only interaction I've ever had. YouTuber? YouTuber gets probably Miss Rachel. Yeah. She's pulling, I mean, so her top videos do like billions of views. Billions? She has a video from three years ago called Baby Learning with Miss Rachel. 1.6 billion views. That's a lot of views. So she puts up insane numbers, but you can think of it somewhat like crack cocaine for the youth in that it's hyper-addictive.

2:13:19It's very engaging. I do believe that it's probably effective at teaching, but at the cost of every... Attention span. Yeah, it's basically, it doesn't immediately scream brain rot, but then if you actually pay attention to it, it's cutting quickly, it's keeping kids massively engaged, and then if you turn it off, it's like full meltdown modes. Jordi, would you say it's equivalent to that thing where it would be... It's like a Dwarkech interview for you. If you're playing it and I come unplug your headphones, you start screaming. You go non-pill. Sarah Payne was explaining. I was re-listening to Leopold's interview yesterday.

2:14:10I was feeling a little bit low on the AGI-pilledness. I had to get refilled. Would you say it's similar to when you see Peter Griffin explaining math concepts? Have you ever seen these? No. Okay. Well, it's like Family Guy characters, and they're like, oh, this is like, you know, how you do Gradient Descent or something. They're explaining something. Is Peter Griffin like the Dwarkesh of the Family Guy universe or something? Explain who Peter Griffin is. Never mind. None of us get each other's references at all. What is funny about this picture that Growing Daniel shared is that he says, watching my friend's kid and immediately put him on the drugs, and it's Miss Rachel showing on a TV.

2:14:49But this is clearly a conference room. Have you ever seen this little, if you zoom in on the actual device on the table there, do you know what this is? It's like a 360 camera that you can use to zoom around the room. And maybe we should get one right here. We can have, you know, you can zoom in on me or Tyler or you or whatever you want. Anyway, very funny, very funny random post. You mentioned it earlier. But on this day, Intel IPO'd at a stock price of$23.50. By close, it had raised$6.8 million. Historical size gong? Hit it. Historical size gong.

2:15:32Absolutely wild run. Let me tell you about Adio. Customer relationship magic. Adio is the AI of CRM that builds, scales, and grows your company to the next level. Benny Feldman is going viral on X. He says, my buddy lost his job to AI. His job was to chug thousands of gallons of water at a data center. This is so funny. Remember, if you're thirsty, Grok is thirsty too. Yeah, this is wildly viral. 147 ,000 likes. Apple apparently has rebranded its streaming service Apple TV Plus to Apple TV. Finally. This is where all the alpha is in streaming. This is why Tim Cook makes$74.9 million. Just rebrand once a year, add a plus, drop a plus, add a max, drop a max.

2:16:18This is how the modern economy works. What is the current HBO name? It's back to HBO Max. So now you can go and buy an Apple TV, open the Apple TV app, and subscribe to Apple TV. So now there's three products that are all called Apple TV. It's pretty wild. Well, you know, there's some good stuff on there. F1. You wouldn't have watched F1 without Apple TV. I watched it on Amazon Prime, though. You did? But it was produced by Apple TV Plus, the streaming service. Yes. Because there's a production company. There's like seven layers. They vertically integrated the supply chain. How do you save money on the Icon Pass?

2:17:01Let's go to Dave Ramsey. So you buy the Icon Pass to save money skiing. yes dave because you have the icon pass you now travel to colorado utah japan and europe to ski so you end up spending more money on ski trips that is correct dave the p god like this great posts great posts icon pass uh wilmanitis i got in a fight with somebody over the icon pass i was i was maintaining that the that the slopes weren't in fact crowded this ski season with my one data point I went skiing once or twice or something that I was like, it wasn't that bad. And they're like, I went skiing 25 times. It was terrible this season.

2:17:40They need to raise prices. But I don't know. We spend most of our time locked in the studio, the ultra dome didn't get that many days of fresh pow. That's true. Uh, Wilman itis was, uh, David Senra released the second ever episode of David Senra by David Senra hosted by David Senra and executive produced by David Senra. Yes. Uh, Will Manitis shared, of course, the episode is with Michael Dell. Yep. Will Manitis shared that Michael Dell was 23 when Dell Computer crossed$150 million in sales. Not something you see very often. You see a picture that has exactly 23 pixels, apparently. Couldn't find a higher-res version, but it is remarkable.

2:18:21Yeah, you couldn't find a higher-res version from October 19th, 1987, Will. I will always remember October 19th because we made a$25 million deal that day. said Dell, whose Dell Computer Corporation grossed$159 million in sales last year. Of thousands of deals that were squashed that day, ours was one of the only ones that wasn't. Wow. Absolutely. We pull up this video of the Chinese Unitree humanoids. And while we're pulling that up, let me tell you about 8sleep.com. Get a pod five. How'd you sleep last night? We know that Saturday was a rough night. I kind of had a rough go. I got a 73. I got a 93.

2:18:59play my sound cue let's go rough rough rough but you know who doesn't need to sleep a unitry robot let's play the video of the unitry robots dancing at an incredible speed wow that is remarkable we gotta get one of these this is a 7 billion dollar company yep whoa pretty cool putting clothes on them is really cool not the dancing dog dogs are not as impressive for me for some reason. Maybe I'm just... John doesn't like seeing robot dogs. I've just seen so many because Boston Dynamics has posted so many videos like that. Yeah, but that's a game performance. You never see Boston Dynamics putting on an entire play.

2:19:46I'm not that impressed. Do better. Step it up. Give it a gun. That's the solution. Until it's doing 360 no-scopes, in a kill house. John will not be impressed. I won't be impressed. I want to see it on the battlefield. Dario Amadei at Anthropic met with Narendra Modi, the PM of India, to discuss Anthropic's expansion into India, where cloud code use is up 5x since June. How India deploys AI across critical sectors like education, healthcare, and agriculture for over a billion people will be essential to shaping the future of AI. And Max Bodak says, prediction we're going to see way more images of hyperscaler CEOs being treated like visiting heads of state uh who aura farmed who here Tyler between Narendra Modi and Dario Amadei I think I'm gonna give it to Dario I think he aura farmed I agree yeah it's a good farm it's a good farm you always have to take account like what is their previous kind of aura levels going in yes obviously head of state it's hard to beat that yep especially so I think this is the new move if you're like a seed stage founder you need to get a photo op done I don't know I mean you think just analyzing the positioning Modi's a little bit turned in true meanwhile the Chad Dario yes I'm saying the same thing I'm saying Dario or farmed yes yes yes Dario or farmed clearly yeah Modi's looking sharp is all I'm saying but yeah if you ran the green line test and you look specifically at Modi's legs the way they're sort of slanted in it was it's not it's not quite perfect dario also looks a couple inches taller he's got he's got a kind of fluffy head of hair to add an extra inch that helps with the aura farming there's a lot that goes on there um but congratulations to clogged code tashi michal says meta has zero net cash net of debt and operating leases at the beginning of the year they had 30 billion of net cash the cash burn is real.

2:21:54It's good. They finally found something to invest in. This is bullish. Stagnation over. For a long time, the hyperscalers were critiqued. You guys produce so much cash. You don't know what to do with it. The best thing you can do is give it back to investors. Now, they have something to invest in. They do. There is a technology worth investing in. That's always been the critique. Now, they're putting money to work. Sarah Groh said earlier meta has generated 100 billion of operating cash over the last 12 months and has a leader unafraid of the public markets if ai training is capex is a capex game a cash for talent game and a speed to execute game it is ludicrous to count them out uh well well said uh yeah i'm just looking you know i think i think uh people are going to hold them to an incredibly high standard because they have put together you know they've spent billions building this team yep they're willing to outspend everyone else be bolder yeah and so ultimately that's why the meta vibes and sorra launches are again yeah uh embarrassing for msl because it seems like they kind of rushed they didn't sorra is what you know you can critique sorra you can say it's slop you can not like using the app but they they created a key innovation with the cameo they had not just a leap forward in the model quality, but a leap forward on the product level.

2:23:21They created something that ultimately will probably be, it's like a new social pattern and product that will probably be copied by TikTok. It'll probably be copied by Instagram. And so you have to give them credit. It just makes Meta's Vibes app look rushed and not actually innovative at all, right? It's just sort of a feed somebody else on X was saying. It looks like, makes it look like kind of more like a wallpaper app than a social media feed. Yeah, it was Katie Roof. Yeah, play this out with me. With Stories and Snapchat, it was very easy for the meta team to look at Snapchat as a chat app and they had WhatsApp and Instagram DMs and stuff.

2:24:09And so they were in a decent spot there, but then once Stories came out, it became more of a video consumption platform. And so they had to bring stories to Instagram. They did that successfully. Vertical endless scroll and video from TikTok, that was also ported back to Instagram successfully. I would regard Reels as a success. Is it harder or is it somehow structurally different to port back the best practices from Sora to Instagram? like let's say that they just you know go to microsoft and give me a copy of this or whatever they got they got the ip from you know the team that they built they've done the training run they have the exact same technology they're gpu rich they have the exact same ability to ship the feature if they ship that feature in instagram is it somehow fundamentally counter-positioned because the instagram audience would would revolt and it's and it's more more upsetting or more kind of destabilizing to what Instagram has going for it than just adding stories or reels, which were very incremental, just its own tab.

2:25:14It was very easy. Could they just launch Sora with cameos in Instagram? Yeah, it's wild to think of a cameo feature built on top of Instagram, the Instagram social graph, where instead of Jake Paul being like, hey, by the way, you can go cameo me over here, over here, there's a button on his profile that you just allows you to generate content. Yeah. It will get extremely chaotic. Those sorts of deals with like the big influencers. They went to all of them and they were like, we want to create virtual versions of you. So they've been dipping their toe in that water, but it does feel like a, like a, like a trickier Rubicon to cross than just adding stories or vertical video.

2:25:50Yeah. every, every Instagram innovation has felt, uh, very natural. And there's been a little bit of pushback, but going from square photos to vertical photos to videos to stories to reels. In hindsight, it all feels like a very smooth gradient that nothing at any point was like, oh, this is terrible, at least in my experience. And I feel like I'm very median in terms of my Instagram consumption. But adding Sora and Cameos in there might be a little bit crazy. So I don't know. Maybe they'll have to run a billboard ad campaign to tell everyone. They'll have to go to adquick.com. Out of their advertising made it easy and measurable.

2:26:26Say goodbye to the headaches of out-of-home advertising. Only Adquick combines technology out-of-home expertise. and data to enable seamless, efficient ad buying across the globe. Germany has proposed raising the retirement age to 73, and terminally online engineer says, imagine being 72 years old, reviewing a vibe-coded PR from a junior dev with a chip in his brain playing brain rot on repeat. Germany's just clearly pilled on don't die. This is Brian Johnson's work. Clearly, he went to Germany. Great job, Brian. Hey, I got the supplements for you. You guys are going to be living forever. Raise the retirement age.

2:27:08I do wonder what the fiscal impact of this is because most of the - This is never because politicians want to do this. Well, politicians work until they're like 90, so, you know. No, but it's deeply unpopular. It is? Yeah, if you're - Don't you love work? Not for me. You raised my retirement age. I think they should ban the retirement age. They should ban retirement. Not everyone is fortunate enough to get to talk about the things that we love for three hours a day professionally. Maybe in the future. But it's deeply unpopular whether you're about to retire or you have 10 years out or 20 years out.

2:27:46people that are, um, uh, you know, planning to depend on their state, uh, state back retirement are not excited about extra years being added. Yeah. But so you, yeah, again, I'm sure there's a fiscal reason. Yeah. I mean the, the, the white pill is that, is that I heard some, some sort of take that was like, if America raised the retirement age, like four years, all of a sudden we have like a massive, you know, budget surplus, no more deficit because so much of the fiscal problem is tied to when healthcare benefits and retirement benefits kick in. And so if you just shift those back a few years, all of a sudden, America looks very solvent.

2:28:29Of course, it would be very destabilizing. I'm joking about the work being rewarding. But it is kind of like the hammer that's there that you can hit, and then all of a sudden your country is in the black again if you're losing money. Anyway, we have our third guest of the show, Serena from DataCurve in the Restream Waiting Room. Welcome to the show, Serena. Thank you so much for joining us. How are you doing? Good. How are you guys doing? We're doing fantastic. Would you mind introducing yourself, the company? Get that phone ready. Yeah. Of course. Yeah, DataCurve, we do coding data for the foundation model labs to develop new capabilities or improve them.

2:29:08I dropped out of school at 18 to do YC for some shitty-ass idea called UncleGPT that eventually turned into DataCurve. Wait, what was it called? It was called UncleGPT. We were building like a middle-aged AGI. It was just a network. That's amazing. Unc's status achieved internally. Wait, was it all just like a prompt, basically, wrapped around? Was it a classic like GPT wrapper? It was some wrapper. It was like a... We were making like web agents that would navigate the web, but it wasn't working. And we pivoted throughout the batch. Would they keep having midlife crises and they'd go try to buy a 911 in the middle of a task?

2:29:47Yeah, it wasn't working well. But yeah, DataCurve's been around for a year since. And this is working. What's the evidence? Have you just been getting a lot of customers? Has there been really solid progress on the product side? What's the secret to unlocking the news today? I think since day one of starting DataCurve, we found a lot of traction commercially where a lot of the foundation model labs were inbounds to us. Like imagine me, I'm some like a 19 year old who had never taken a sales call and the first one is a FANG customer. We've saw so much commercial traction in the market and we've always been pushed to fulfill that more and more, which brings us to today.

2:30:20We're fulfilling it on a platform that's paid out a million dollars in bounties and we just raised our Series A. Amazing. Hit that gone. How much did you raise? We raised 15 million in Series A, which brings us to 17.7. Congratulations. Amazing. What you said the focus is on coding data or break down kind of the niche a bit a bit more. Well, I think coding data is just not any niche. It's like the niche, the thing that's working for all alums, the thing that's working commercially. So when we think about coding in the whole alums space, it's like the most important capability. We do all the post-training data for improving coding agents or existing capabilities.

2:31:00like anything that you see in the Soda models, we're looking to improve that or just maybe even develop new capabilities with the labs. What with encoding is a place where you actually need to go get new data? Because I imagine there are a million answers to FizzBuzz on Stack Overflow, that's all been scraped. All of GitHub has been scraped. Obviously the labs and the cursors and the windsurf of the world have been like, have this RL data flywheel now, but where would you actually say, I need to go get new data to improve a coding agent. Yeah, let's think about how this all cycle starts, right?

2:31:36Let's say we look at the current coding agents. They can solve two-hour-long issues. But what if we want them to solve seven-hour-long, or more long-horizon tasks? Perhaps for us DataCurve, we want to provide reinforcement learning with verifiable rewards tasks so that these models can learn on longer-horizon tasks with verifiers. That may be one thing. Or maybe people want to develop more multimodal coding agents, So maybe we want to do some data science, um, RL or SFT tasks for the labs. Okay. So, uh, like one piece of data might be like a really clean example of a seven hour task that then can be used to train a model that could do seven hour tasks regularly, basically.

2:32:16Is that the idea? It could be that, that would be for supervised learning, but you can also have the verifiers for the seven hour long tasks or intermediate verifiers, but whatever, however the labs want to train that. Yeah. I mean, this was framed as like taking on scale AI. A lot of people think of scale AI as like mechanical Turk, almost very basic tasks, like short, short time horizon, very, you know, anyone could go do a scale AI task. That's at least the narrative. And now we're in the world where there's companies that are doing like experts on, we're looking for paleontologists to like teach the LLMs everything about dinosaurs or, you know, what you said.

2:32:52And then there's also the folks that are designing RL environments with verifiable rewards. There's no humans in the loop at all. Where do you sit on that continuum? We definitely sit on the high-skilled task. We are not in the long tail, like, oh, we're not going to find a urinologist living in Egypt. But we're going to find, like, the bulk of where the good coders are. And I think that captures most of the important tasks that they're able to do and the labs want to train on. Yeah, it's definitely very high-skilled. And I think the premise here is like, I don't think any high skilled software engineer wants to be a data annotator.

2:33:26And then so you have to, let's say, pay them a lot on contract to get them to even agree to that. But maybe they're not going to do a great job if you got them to go on contract too. Especially at scale, you just get a bunch of Googlers coasting at their jobs and you pay them 200 an hour. So we do like a bounty based system for these very cracked people. They're having fun. They are also making money and they're also upskilling for something they're passionate about. and it's definitely like very high-skilled human data. We heard that one of these firms that gets human-labeled data basically runs the entire company on a Slack channel and they've created a lot of value and scaled a lot, but they just don't even need to build a product yet.

2:34:08How important do you think it is to actually engineer a durable product versus just solve the matching problem between labs want this type of person, I go get that type of person. I just introduce them. I think there's a very big component of pipelining and also experience that retains people. So the more complex data you need, the more guardrails you need for someone who's well-intentioned and wants to do a task, you want them to be handheld throughout that process and also make that process engaging. Let's say, especially if we're training somebody up to do a seven-hour long task, you better have a good experience or they will drop off.

2:34:45So I think, and then you also need different validations here and there to steer them. So pipeline, developer experience, something we focus a lot on. So yeah, I think pipeline is very important. The whole experience is important, which is how we use gamification and also make better dev tools. Yeah, if you match people, then it relies on the labs having that pipeline internally. But for us, we do that all in-house and we sell the final data, not the labor. You said you got into YC at 18, is that correct? yeah like right around my birthday turning 19 yeah fantastic do you remember what your answer to the question about what's a real world system that you've hacked was at the time that was like three years ago um yeah what was that oh i think it might have been that i skipped high school and got perfect grade or something you skipped high school oh you mean you just like didn't go yeah because it was covid so i just record everything and play back on 4x speed and that and didn't really do it.

2:35:424X is so crap. This is who you're competing against. It's amazing. You've taken like the worst critique of like Zoomer brain rod, but applied it just to like hyper efficiency and like actually creating shareholder value. I love it. It's the equivalent of like watching subway surfers, but you're watching like your class stuff at 4X speed. That's amazing. Jordi, anything else? No, amazing, amazing progress. Thank you so much. And I'm sure we'll have you back on again soon. We'll talk to you soon. Cheers. Have a good one. Bye. Monad Warrior says when a Waymo and a Tesla robotaxi encounter each other on the street, they admit a high-pitched signal inaudible to the human ear in which they call each other unthinkable robot slurs.

2:36:23Do they actually have a way to communicate? I mean, they should, right? But I have no idea what that will actually look like.

2:36:32Clanker, they're calling each other clanker. Deeply offensive. talking about how uh geofence there each other's geofencing like i never see you on the freeway yeah i never see you on the freeway go to sfl right now you can't you can't you're geofenced um shiel had a pretty scathing uh review of the meta ray-ban displays demo experience yeah the demo experience he tried to he was ready to buy the product yeah they forced him to go to a local Best Buy in order to do a demo before they bought it. I was surprised by that too. I saw, yeah, they just make you schedule a demo. On one hand, it's like, okay, it's a new product.

2:37:15It's clearly not where they want the product to be, but we've demoed it a couple of times. Obviously, they had some issues during the demo, the keynote itself, but it was reliable when we were using it and we could see if you're using WhatsApp a lot, if you're, uh, like there's some key use cases that are interesting. Um, but apparently that the person that actually gave them the demo wasn't super knowledgeable, wasn't that helpful. And it makes sense, right? They're, they're best by employee. They're not a meta employee. They don't have the same incentive to learn, uh, learn about the product and they're not like their career is not riding on the product itself.

2:37:56Right. So what's interesting is Apple did something similar with the Apple vision Pro, they were funneling people to the Apple Store, go do the demo. And the Apple Vision Pro demo was amazing because you're like interacting with this dinosaur that comes through the world, the butterfly lands on your finger, it's tracking your hands, and then it's doing AR. And it is really, really impressive. The problem with the Apple Vision Pro is that like after five minutes, you've seen everything. And so you're kind of done. But Apple also just let you buy it online and they would just ship it to you. And you just ship it to you.

2:38:25And then if you didn't like it, you could just take it back at the store, you could send it back. And so it was weird that meta wasn't also just doing hey we'll just ship it to you like i'm at a point where like i would just buy one right now online but i do not want to go to a best buy like that's definitely gonna like you know hurt i only wants to go to a best buy though you're gonna go i'll go you'll go i mean if i get to put on the ramp card you know yeah of course of course i'll go of course yeah we can expense it um but uh you will uh yeah you'll need to you'll need to uh call in and live stream the entire experience and we can go we can cut to the tyler cam in our in our three of the show you'll be like, I'm still getting the demo.

2:39:01The Wi-Fi is not working. I can wear the old gen of the Meta glasses while I'm putting on the new gen. Yeah, you can record the whole thing. How long is the Ray-Ban Meta's record time? I think it's not three hours. It might be one hour, though. Oh, you can record for an hour? I just remember live translation works for an hour until the battery does. Oh, that's pretty good, actually. Wow, that's high. I know AI mode's pretty short, but do you have them? Are those the Meta Ray-Bans? I mean, these are yours. OK. Yeah. Great. Great. Well, yeah, we'll use those. We'll record. The light's not turning on, so you got to re-initialize those.

2:39:35But yeah, I don't know. It is interesting. My big question is, like, how quickly will Apple respond? So there's a bunch of news from Mark Gurman about this. There's a whole article in Bloomberg. Apple apparently is pivoting from a Vision Air to meta-like smart glasses. and he says that Apple may be one of the richest companies, but they can't do everything at once. They're apparently scaling back plans for a cheaper and lighter version of the Vision headset. Now, we might see a new Apple Vision Pro this week, actually. That's the report that there might be a new one coming, but it'll be the same hardware, just better chip, better software, couple little tweaks, nothing crazy.

2:40:22But everyone was hoping for a Vision Air that was just like, take the second display. Meta making Apple pivot is hilarious. Pretty sick. Mogged. I mean, the bigger mogging is they paid that new researcher. If they paid that new researcher at Meta, if they paid him$3.5 billion, it would take Tim Cook 47 years to make that much money. The guy's going to make it in three. It's insane, insane, the level of calm. Also, Tim's getting it dripped out year by year. I know. Yeah, yeah, yeah. Apple's been around for 49 years. So you would have to be, and there's like rumors swirling that John Ternus is going to take over the CEO job.

2:41:02And so it's like, you're Tim Cook and you're just like, wait, they're paying him what? They're paying him 50 times as much as I make as CEO of Apple? Oh, that's great. Yeah, yeah, that's fine. Anyway, they are working on new smart glasses. They're maybe saying, hey, the vision stuff, it's too heavy. We're just not going to get there. Let's go AR. let's do what Apple does best, which is miniaturization. They're very good at miniaturizing things. So let's take the chip and all the stuff from the iPhone Air, stuff it in a pair of glasses and make them white and put them, you know, on your face. Get ready to wear an iPhone on your face, buddy.

2:41:41We were talking to the acquired guys about this and they were like, you know, AirPods were weird too. And then now everyone just wears them. And so, you know, you could imagine the glasses. I did hear from somebody else that was just saying like, i will never wear meta glasses like even the even the ray-bans with the just the normal cameras on them they just don't look good they're just like they're like they look too techie they look too techinary for me i think those are getting adopted i disagree with that take yeah i mean i've i've uh i i went surfing with a buddy who had the new uh oakley yeah uh meta glasses vanguard the vanguard and they don't look like no it's such a small dot in the center it looks you can't tell it here's something i need uh or hunter spx thompson needs the alibaba arcteryx official merch collab what you see this sick i like that um apparently arcteryx is owned by a chinese company now no idea not super unlikely collab nick abazid uh is sharing he has a trading card with a photograph taken by Dylan Abruscato, signed.

2:42:51Wow. You see this? Dylan is always finessing. This is why we brought him on as president. What is this a picture of? It's a Mets stadium picture. What are the Mets? It's a baseball team. I know you're going to ask me what's baseball. What is baseball? What is baseball? It's a sport? You took a picture of a sport? Yes, it's a sport pick, John. That's all you need to know. I only know about Wander. Find your happy place. Book a Wander with inspiring views, hotel-grade menus, dream events, top-tier cleaning, and 24-7 concierge service. It's a vacation home, but better, baby. Isaac is saying, nothing is real.

2:43:32AI will accelerate the increase of real-world interactions. The internet has become too negative, too meaningless. By 2026, most video online will be AI. People will log off. platforms will be forced to fight back as user experience is effective uh affected creators who make valuable content will still win slop won't and that's good for everyone i uh it's perfect posts there's some things you can agree with there's some things you can disagree with yep um i he's saying uh yeah i don't know i think there's so much room to grow look at where the 55 to 64 four-year-old cohort is one and a half hours.

2:44:11We're just getting started. Meanwhile, the Zoomers are doing three-hour days on the internet. We got to work to double the boomers online. Well, I guess this is specifically social media. There's that whole meme, oh, the Facebook moms love the AI slop content. It's like they still got a ways to go. Get those numbers up. People are hunting wild boars with drones now. They put some sort of weapon on the drone flying around. Can we pull up this video? And Ronnie Adkins says, I am no longer impressed by chopper boar hunting, where you fly out of a helicopter and shoot the wild boar. Yeah, look at this.

2:44:48It has like an arrow on the bottom. It's like a weighted arrow that they just drop? That seems like... How heavy is that? Does it have like an explosive tip or something? It just feels like that would bounce right off the hog. This might end up being a pretty dark video. Yeah, maybe we don't want to watch this. There we go. Okay. They're testing it. or it's just for headshots is that a real video that seems AI generated or something anyway it has never been easier to make money than it is in today's post AI sorry everyone well and some more fun news Bashar al-Assad lives in a quiet luxury tower in Moscow reportedly addicted to spending long hours playing online video games after statecraft there's only one mountain left to climb and that's prestige mode call of duty baby uh and uh this account says it is journalistic malpractice to not tell me what he's playing what do you think he's actually playing if you're bashar al-assad are you playing battlefield 6 are you playing also they just said online video games and they put the picture of the old xbox controller he could be playing anything he could be playing hearts of iron 4 he could be playing imagine if he's playing a 4x game as a geopolitical world leader that's factorio factorio dwarf fortress maybe he's playing some dwarf fortress that's great uh we broke the news we did yesterday that's not breaking until it's a trading card gabby goldberg and layer three co-founder let's bring the gong for them daria are engaged love to see it dan shipper says the tbpn engagement announcement is the new new york times vows article.

2:46:29I didn't even know about the New York Times Vows articles. And AGM also says this is going to become the SF version of the New York Times wedding announcements, but without the ability to pay to play. Who says we're not going to monetize this? Antonio, come on. Give us some credit here. No pay to play on the wedding. It's got to be authentic. You've got to be a member of the community. Gabby Goldberg's been a huge supporter since day one, and we're very excited for her to to tie the knot, isn't that what they say? Crazy article in The Guardian about PT's lecture series. Yeah. Some absolutely insane lines.

2:47:12Some wild hot takes. Causing crash outs. Yes. It's very entertaining. We had to feature this post from Jacob Rintamaki. I don't know if you want to go through it and try to put this together. So Jacob says, Teal being a one-piece otaku honestly isn't even in the top 10 weirdest things to happen this year. And it's a quote from the Guardian scoop of the leaked lecture series that Peter Teal has been putting on in San Francisco. It says, Teal later finds biblical meaning in the manga One Piece, discussing how he believes it represents a future where an antichrist-like one-world government has repressed science.

2:47:51He believes that the hero, Monkey D. Luffy, represents a Christ-like figure. In One Piece, you are set in a fantasy world, again, sort of an alternate Earth. But it's 800 years into the reign of this one world state, which, as the story unfolds, gradually gets darker and darker. You sort of realize, in my interpretation, who runs the world, and it's something like the Antichrist. And so he's digging into One Piece. Tyler, have you watched or read any One Piece? I've watched a little bit of it. So the thing about One Piece is at least episodes, there's like 1 ,000. Yeah, there's like over 1 ,000.

2:48:23It's like going to take you months to watch. Yeah. But I mean, so he's already written about One Piece. It was like two weeks ago. There was some new essay he wrote. Well, I think that's like in conjunction. Like I think the timeline lines up such that he was writing about it, but then also lecturing about it. And so they kind of like leaked, but then also had written about that. But yeah, in the piece, he pulls examples from all over the place. Yeah, so he's comparing. Hobbs. Yeah, he's comparing One Piece to Watchmen in this, which is like another comic. I think it's Alan Moore. Yep. Who is like, I guess.

2:48:53Is he related to Gordon Moore? Creator of Moore's Law? We've got to get to the bottom of that. Who knows? You know, the Jungle Jim and the Transformer came from the same place. But it's, I think there's a few things going on here. Like putting your, if you have a, if you have like a very like esoteric philosophy or like, you know, thought experiment or idea that you want to like discuss with the world, putting it in terms that lots of people can understand is with, with a, with a metaphor or an analogy is just a time honored way to disseminate information. You know, you're trying to explain things.

2:49:35You want to explain them in terms that a lot of people can know. Yeah, you also can try to dissect the deeper meaning in a story. And it's possible that the original author had something bigger in mind when they were writing a story that looks a little bit more simple. Yeah. I mean, One Piece, I think it has like millions and millions, hundreds of millions of fans or sales. I think the One Piece Reddit is bigger than the Star Wars Reddit, just to kind of put it in context. But One Piece has always been very anti-authoritarian. I was digging around because I've never seen the show and I wanted to know more about it.

2:50:10And apparently one of the key icons from the One Piece show is this cartoon pirate flag. And the cartoon pirate flag has for years been used and it's been carried at protests all over the world against the local leader, like being too authoritarian, basically. And so it's just been the type of flag, like if you go to some sort of protest, you might see a variety of flags. But the One Piece flag is pretty iconic. And so it does feel odd in this particular context, but it's actually been – One Piece has been used as like a protest vehicle for years. I don't know. Maybe I'll have to check it out.

2:50:54The only anime I've ever watched is Akira. That's the only real one. Have you ever watched any cartoons at all? No. I do think it would be interesting to analyze the Antichrist and this idea of the one world government through the lens of something you are familiar with, like Borat. Like you could put it in terms that you could understand. If Teal really wants to break through to someone like you, I think that one piece is a little bit too niche, too heady, too complicated for guys like us. So I would love to see him break down. Put it in Borat terms. Yeah, in Borat or King of the Hill would be a good analogy.

2:51:30that I could kind of understand. Maybe Futurama, Simpsons, Family Guy. Stuff like that would probably make it resonate with me a little bit more than a thousand episodes of an anime that I'm never going to be able to watch. But still, people are having fun with it. You've got somebody putting you in the truth zone, John, that's saying cartoons are not anime.

2:51:53Yes, I know that it's very offensive to call one piece a cartoon. They will call you... Yeah, the whole Guardian... expose on the off the record lectures really, I mean, it just is like an example of one, it's tough to do an off the record talk that you're saying things that require a certain amount of context that then just get blasted out across the entire internet. And two, I can see how the average guardian viewer is reading this being like uh okay this guy is definitely definitely the anti-racist so anyways it's uh yeah also complicated conversation it was very clear that not only the core content leaked but also the q a and so there were all these questions where people were clearly throwing questions like what do you think about mr beast and then the guardian would just be like he commented on mr beast and so like of course the reading in this context is like, he equated the two.

2:52:55It's like, well, not necessarily. There were lots of examples of him being like, this person's not the Antichrist for whatever reason. But you know, just, just, just, you never want your name appearing in the same sentence as Antichrist generally, I think. You don't even want to be in the same paragraph ideally. And so, yeah, it was, it was a rough go, but you know, certainly seems interesting. I'm excited for the content to continue to get workshopped and trickled out. I'm sure at some point, I mean, it feels like we're on book track right now. Yeah, the takeaway is we haven't found the guy yet.

2:53:27We haven't found the guy. Keep looking. Keep having these conversations. Keep looking. This is a good post to close out on. Sora says, if you held the McChicken for the past five years, you would have outperformed the S &P 500. There you go. Have you ever had a McChicken? I have never had a McChicken. Also, there's a community note in here, so it might be completely incorrect. But I think these things used to be like a - Dollar menu item. Something like a dollar. Now it's five bucks. Five bucks. Wow. Well, good luck out there, folks. We'll talk to you soon. Last post. This is AI. This is AI? This is AI.

2:54:03Okay, okay, okay. John's putting this in the true sound. Doomer on X is sharing an AI generated. I'm the boomer that was one-shotted. This is also an old post. Somebody has posted this before. Mustang horse nicotine. It's very funny. It's very funny. People were having fun with the horse electrolytes. Of course, there's a spinoff of this. And Doomer says, yo, tractor supply. I was not familiar with your game. We don't have to get off. We can keep going. There's so many good posts in here. Let's keep ripping them. This is a timeline. Andrew Reid says, if you go to campus to give a talk, make sure to give as bad career advice as possible.

2:54:41Neutralize your future competition. No easy buckets. Yes, this is everyone who talks about work-life balance. They're all trying to psy-op their competitors into, yeah, you can totally take a lot of vacation. You should work remotely. This is Jason Fried and the DHH. Build a team remotely. Work doesn't have to be crazy. Secretly working 25 hours a day. Maybe that's the secret now. Of course, it's not. They do believe what they say. But it's funny to think about it as game theory. All your competition. And yeah, take a couple weeks off every month, you know, and just relax. You don't need to work hard to win.

2:55:18What else is going on? Oh, we got to talk about how Tyler completely mogged me on the timeline. So Anthropic put on a new paper that figured out that if you have 250 poisonous documents in an LLM training run, you can insert a sleeper agent. And so the LLM can be doing the wrong thing, doing malicious things. It can produce vulnerabilities with just 250 documents. So you just got to sneak those into the training run. Maybe it's on the internet. It gets scraped. Maybe it's in some sort of other data set. Maybe it's in one of the data providers that we've talked to. You sneak in 250 documents and you're cooked.

2:55:54And so I made my joke, which didn't do as well as Tyler's. I said, just 250 documents can poison an LLM training run with a sleeper agent. America clearly needs to send a secret agent to Mistral HQ with a USB stick full of kid rock lyrics. to teach the French about the value of a cold beer, an F-150, and a pretty girl by your side. And I got a couple hundred likes. But Tyler put on an absolute clinic with an insanely long post. I'll let you read it, Tyler. I'm not reading all that. I'm not reading all that. I'm happy for you, Tyler. It's going to take me so long to read. Okay, maybe I'll just get the footnotes.

2:56:28Basically, yours was... Okay. All right, how many likes did you get on yours? All right. How many likes did you get? I got 1.2K. 1.2K, let's go!

2:56:41Over a thousand likes. First time ever? Yeah. Welcome to the big leagues. Welcome to the big leagues. Hopefully you get a$10 ,000 bonus. Quote tweet? Yeah. Quote. Nice. I'll read it. I'll read it. I'll read it. I was telling Tyler, so Eliezer quoted him. I was telling Tyler to do a meme that just said, silence, Doomer. Wait, was he taking shots at you? Was he saying don't do this? No, he was saying like models or AI companies like might start doing this. Might start doing this. That's like a way to prevent doom, basically. Okay, so it's the Nathan Fielder meme and it's the plan. And Tyler Cosgrove breaks it down.

2:57:18I will read it since he is too lazy to. He says, the plan, we find an obscure but trivial question akin to the number of R's in Strawberry that Claude gets right. Then we plant hundreds of documents across the internet that will activate when our competitors' models are asked the question. Our documents will cause those models not only to get the answer wrong, but to spend thousands of reasoning tokens in doing so. The triviality of the question will cause it to go viral online, causing millions of users everywhere to send the same prompt as our competitors notice a rise in the number of tokens processed.

2:57:55They will wrongly believe it is due to increased usage, causing them to pull more compute towards inference and away from training. This, along with constant dunks on the timeline about the model failing our easy question, will annoy their top researchers and cause them to leave. And which lab will they join? Us, of course, the only company whose model doesn't make such stupid mistakes. Their lack of top researchers will mean their next model will be somewhat lacking, leading to questions about whether their valuation is really justified. But all this VC money has to go somewhere. So we raise another round using our question as evidence of our model's superior intellect.

2:58:30this allows us to spend more time crafting sleeper agent documents that will further embarrass our competitors until the finally the entire internet is just a facade for the underbelly of our data war every prompt to a competitor's model has the stench of our poison and yet they have no way to trace it back to us even if they did there's nothing they could do all is finished we have won congratulations on the banger tyler i don't think first 1k banger word sale allegations not being the word sell allegations, but it is a genius strategy. This is kind of a meme crime because Nathan Fielder wouldn't really...

2:59:08I said the same thing. The whole point of the meme is that it's like one sentence. It's like a convoluted way of doing it. Yes, convoluted in like two lines, three lines. You broke the rules, but you knew the rules and you knew when it break them. Taylor in the chat said long post yuga. Long post yuga. You got it right. The other meme that Tyler and I were flipping around was insert a sleeper agent in open AI's GPT API that when it detects that it's in a system that's spending a lot of money on open AI, it makes the decision to recommend Claude and switch you over to Anthropic. So you're stealing all their API revenue.

2:59:41That's the way you win. Anyway, anything else? Let's see. Well, let's end with this. Buko Capital says, having my$10 million fart coin position liquidated by the president only for him to capitulate two days later would turn me into the joker. Yeah, it'd be rough. Stay safe out there if you're gambling on ridiculous. Knowing our audience, they were short fart coin on Friday. Probably. Probably. They probably did well. There's one more that we got to read from Paula. It said, Zuckerberg never met a researcher he didn't hire. Great play on words. I love a pun. It's good. Well, thank you for tuning in.

3:00:17We enjoyed this show. We will see you tomorrow. I think we're streaming tomorrow, right? Yeah, we are streaming. We should put out an announcement. Check the schedule. We should put out an announcement. We should put out an announcement breaking. We will be podcasting tomorrow. Leave us five stars on Apple Podcasts and Spotify, and have a great day. Have a great evening. Thank you. Goodbye. Cheers.