KPop Demon Hunters breaks records, Grok speeds up, and an AI-powered browser deep dive | E2171

25:57What we would think of as online comics, uh, that are a huge source of IP for Korea and Japan. Like they're constantly making animes or TV adaptations or movie adaptations of stories that got popular on Naver as a webtoon and then got turned into this next thing. Like a lot of the Netflix anime originals or new anime shows you'll see out of Japan started life or like even Korean TV shows. A lot of them started life as these, you know, basically manga or comics, but they were not published on paper ever, just exclusively from the internet. Before we jump back into this, Jason, there's a company doing this.

26:36It's called MicroCo. And this is an article from earlier this month. They want to take the micro drama phrase, infuse it with AI and avoid somehow the Queeby comparison, which is the obvious thing. If you don't recall folks, Queeby was a short-lived, short-form video project. right? Quibi, not Quibi. Quibi. It was a terrible name, but it was Katzenberg. Katzenberg comes up as a young assistant in the Barry Diller. He was, I think, Barry Diller's assistant for a little bit, or one of the heads of Paramount Pictures or Gulf Western. And he's a friend of mine. I play cards with him sometimes, and we chat once in a while.

27:11Katzenberg's like one of my favorite people. And he had a really good idea there. I think the problem was too much money going too big. Remember they had a$2 billion valuation before they started? Yes. You can't have a$2 billion valuation. You got to have a$50 million valuation. You have to raise 10 million. You have to go slow. You have to learn. And you have to make mistakes. And the thing that I got out of the Barry Diller biography was that he just liked to fail fast and learn by doing. And that's how he did his entire career. Barry Diller created the movie of the week. Lon, you can explain to people what this kind of cultural significant was.

27:48But he, when he took over CBS, decided to do the movie of the week. It was ABC, wasn't it? I think it was ABC. It was ABC. Yeah. The ABC movie of the week. I think so, right? I think it was like third place. Because that was like Duel. And he started doing these. Duel is the most, Steven Spielberg got his start doing these. Duel, that first Spielberg film was an ABC movie of the week, famously. And it turned out that the movies of the week, they had had all the different movie studios pitch them. And he said, you know what? We're just going to do it ourselves. ourselves, we're not going to give the creatives the ability to like just go off for two years and make something.

28:22We're going to actually have a process here where they pitch, we iterate, we read the scripts and they have like tighter controls on it. And they made so many amazing films. Some of them had like six or seven sequels to them. And it was the number one show. The other interesting thing I was going to say IP wise was they would do, they would make a lot of pilots as movies of the week. And then the popular ones would get TV shows. So like Starsky and Hutch, Six Million Dollar Man, Kung Fu. Those all started as ABC movies of the week that then took on a life of their own as shows. Yeah. This was like a crazy idea.

29:00So there's always, even though the media space feels like it is the toughest business in the world, and it is in some ways, there are opportunities here to take the new mediums, whether it's AI, generative AI, specifically TikTok, you know, shorts, vertical video, or panels on Instagram. And you could figure this out. So if somebody out there is like, got a lot of energy for this, I would love to hear some pitches on it. Jason at calacanis.com for life. All right. Thank you, editorial hair director. My pleasure. I think that's the proper way to say it, hair director. Yeah, I'll take that. Sure.

29:39Yeah, sure. Hair director. Okay. Thank you. Yeah, yeah. All right, we'll drop him off and continue going through the docket because I'm sure there's a lot of tech news that we haven't talked about since Wednesday. Yes, very, very much so. Although it's not going to be as fun, everybody, as K-pop Demon Hunters. We're now going to go eat our vegetables. But Jason, there's quite a lot going on. There's your dad joke, folks. Dad joke number two. Doom, I need my stinger. Dang it. I love it. Keep doing them. I want to have like a stinger that like everybody groans and then there's a layer of cricket.

30:12Somebody in the audience make like a great stinger for us. Dad jokes from Alex. I'll pay you like 500 bucks to make me, whoever makes the best stinger for this will pay you 500 bucks. I'll pay it valid to you personally. If you wanna run a successful company or maintain a profitable portfolio like I'm trying to do, you need reliable data. And without the most accurate and up-to-date information, you're flying blind. That's why you and your company need a partner like AlphaSense. They're the world's number one most trusted AI platform for market intelligence. That means they can help you and your team track the latest economic trends before your competitors even know they exist.

30:55Hey, listen, hear a twist? We eat our own dog food. We only pick advertisers and partners here on the show that we love their products. Alpha Sense has a massive searchable database of public companies, and that lets us quickly dig up the most relevant facts and context. And their equity funding screener includes private companies, not just the public ones. They save us time and they make us superhuman. So start your AI search and market intelligence journey today and get deeper insights to power your business. We're even going to start you off with a free trial. How great is that? Alpha-Sense.com slash twist to get started.

31:32make sure you use that URL so they know we sent you and you can get this great deal. But let's get started here. I know XAI has a coding model and that people are kind of losing their mind over it, yeah? Yeah, it's fantastic. So everyone knows XAI is Elon Musk's foundation AI model company competing with Anthropic and OpenAI. And we talked about them on the show when they launched the Grok 4 family of models, which as we said at the time were quite good. They ranked very high. People quite liked them. Now, there are also something new. It's called Grok Code Fast 1 with dashes in between. This was, quote, built from scratch, starting with a brand new model architecture, and they built it alongside some partners, Jason.

32:13So as they were building it, they were talking to Cursor and GitHub Copilot and Windsurf and others to get feedback to make it better. Now, what's really interesting here, because we have other models in the market that are pretty good at coding, from Anthropic in particular, this one is fast and cheap. For use, it's only 20 cents per million input tokens and only$1.50 per million output tokens. And for reference, Jason, that's cheaper than GPT-5 Mini. And it has a really, really quick token throughput. So it's just quite impressive. And here you go. Here's the model performance on the XAI site.

32:46As you can see here, you have the output price for 1 million tokens and tokens per second. Tokens are, you can think of them as words or chunks of knowledge. and grok code fast one is the cheapest and the highest tokens per second and then if you go look at you know grok 4 and gemini gpt5 claud sonic sonic very expensive one of the great things about what we're seeing in a competition like this is how um what's the economic term deflationary I guess this technology can be, if you want to win, you can change the world by making something that was previously extremely expensive, extremely affordable.

33:34And you see this in everything from tickets for airplanes, right? That's like an attack vector. We're going to make flights cheaper. You can see it with food, cloud kitchens, delivery of stuff. You can see it in hotels with Airbnb and hospitality, right? And so here we're seeing it as the, I guess, way for XAI to catch up. When you're behind, when you have a couple of players ahead of you, what are strategies for startups? You know, one of them is to make something free or to make it dirt cheap. That gets people's attention. And then there's a certain group of people, usually the vanguard, the early adopters, who will notice something like that.

34:14So you can make it better, that's always good. Another attack vector is to make it cheaper. And another attack vector is faster. So if you want to beat people, better, faster, cheaper are the three things you're looking at. And if you were to look at Elon's success at SpaceX, cheaper and better, right? Right. And then you look at his electric cars. They were faster, but more expensive and better. Actually, it wasn't even better. They were just faster when the Roadster came out. They were very fast. Yeah. Then when the Model S came out, it was faster and better. And then when the Model 3 came out, it was faster, better, cheaper.

34:56All three. All three. And then you look at BYD being, you know, what is it? Your dreams? Build your dreams? Build your dreams. I think it's build your dreams. Yeah. And so you look at BYD, they're cheaper, not faster. They're certainly not better, but they're going after cheaper. And the Chinese government is subsidizing them and they are ripping through some markets, you know, and it's quite unfair when the government price dumps like that. But hey, that's the game on the field right now. So it is what it is. Jason, I want to illustrate exactly what you're saying in the form of a chart. one thing that I track a lot is how AI models rank and then how much they get used.

35:40And one place that we look at that usage data is on OpenRouter. It's a service that lets people kind of pick and choose models for their inference needs. And I don't bring it up too much because it is a partial data set. I'm trying to avoid those. But sometimes something moves so quickly that I think it is worth our time to look at. Yeah, it could be directionally correct. It might not be the perfect data, but it could be directionally correct. Yeah, hit us with it. Especially when the direction is this clear. So what I'm showing on the screen for folks listening to the audio is the combined tokens process on OpenRouter, which is a venture-backed startup, for all of XAI's models.

36:11And Jason, as you can see here, the teal bars are when Grok4 came out and their usage did go up by maybe 30, 40%. Then they started letting people use Grok code fast one and it tripled to up to 80 billion tokens. I think that's a day. So it's been a dramatic increase in XAI's, I would say, in market share. And I think that's critical because they've been building good models, but I don't think they've been commercializing them or getting revenue out of them as effectively as I might've expected. But here they have a demonstrable straight up hit. Amazing. And Microsoft, which has a very deep relationship with OpenAI, but let's call it perhaps tenuous, at times conflicted, because they own 49 % of the company until it hits some amount of revenue, this whole convoluted deal structure that, you know, is a headwind for open AI, all this convoluted nonprofit to for-profit, the Microsoft relationship, the nuance of that relationship, which has them making back a certain amount of money, and then it becomes public demand.

37:22It's just all very weird. But it's a real testament to how great their product is and how the utilization of it and their product cadence, how often they release new stuff, that despite all of the hair on those deals and the corporate structure, they're able to keep this thing growing and the revenue growing. If you have revenue growing, it can make up for a lot of mistakes, but you shouldn't make these kind of organizational mistakes. You should avoid those. That's why you got to have good attorneys, good board of directors, et cetera, which they did not. Maybe they had good attorneys, but they didn't certainly have a good corporate governance.

37:57But Microsoft, I'm guessing, has seen enough to know that that relationship is not reliable. And they're doing what exactly, Alex? So Microsoft's AI division, named Microsoft AI, or MAI, if you will, has put out two homegrown AI models this week. Now, Microsoft has released some models in the past, but these are more flagship. These are larger, more general use. And so the one that we need to talk about is called MA1 Preview. It's a mixture of experts model trained on 15 ,000 H100s. So they put a lot of GPU power behind this. And it is a model built for kind of a purpose inside of the Microsoft domain.

38:39So it's for following instructions and providing helpful responses to everyday queries. Think about things like Office Copilot, Jason, is my guess. What's interesting to me is that they immediately charted with this one. So I was over on LM Arena, which runs a lot of side-by-side tests to track out which models are best across which metrics. And this one shot up to the 13th best model in the world, which I think for a preview from Microsoft's AI division for their first model of this sort is quite good, but it's still very, very far behind what OpenAI has cooked up. Because right now, if you look at the top of the charts, it's Anthropic and it's Google and it's OpenAI and then Microsoft's far down below.

39:15But this de-risks them, Jason, because if they have their own model, they don't have to pay anyone else's margins. So in time, I can see Microsoft moving away from OpenAI's models for stuff inside of the Microsoft world and still offering them, of course, on Azure and so forth. But this strikes me as Microsoft saying, we don't only have OpenAI in our pocket. We do have our own smart nerds who are cooking up their own models. Yeah, that's an important hedge. And they had hired, was it Ilya that they hired? No, it's Mustafa, I believe. Oh, Mustafa, right. Sorry. Mustafa got hired by Microsoft to become the CEO of Microsoft AI.

39:51So What's interesting about this is we're now starting to see a couple of different cohorts. You got the non-Mag 7, I'll call native models, OpenAI, XAI, Anthropic. Then you got the Mac 7, in some cases, pursuing models. I understand Amazon has a model now. Apple was doing something in vision and images. This is Microsoft, obviously. We know Gemini doing very well at Alphabet. And then Meta has Llama. The llama, yeah. And then you got the Chinese companies coming in with DeepSeek and Moonshot, Baidu. And I guess Mistro in Europe would be, I would put them in that first group. So you kind of have these three groups, the Mag-7, the non-Mag-7, and then the Chinese ones.

40:38I think it's starting to come down to who has the wherewithal and the deepest pockets to stay in this game. This is not a cheap game. This is like being in like the F1 or in like the, what were those sailing races that Oracle would always win? America's Cup. America's Cup. Like it's expensive to feel the team, I guess, and also like professional soccer too, right? Like it's not cheap to have Arsenal or Newcastle or whatever the teams are there. I'm glad you started with the best team in the Premier League. Newcastle. Oh, Arsenal. Okay. Just to point it in perspective though, Jason. And so the F1 racing world is famously expensive.

41:22They did put a cost cap in. So now it's about 140, 150 million a year. That would get you, what, seven GPUs and a cooling rack? I mean, it's funny how much more expensive this stuff is. Yeah, and then I guess the question is, we're now in a game of chicken. With these companies, who can keep building data centers, keep raising money, and then race towards the cliff, hoping that the cliff turns into a bridge with the bridge being revenue and customers willing to take out their wallets and give 20 to$200 a month,$250 a year to$2 ,000 a year to use these platforms. And there probably don't need to be 20 of them.

42:12It's probably going to whittle down there. It would be great if there were a dozen available, but it will eventually be six. So we're going to see six seems like a good number. Yeah. I think six, you know, three from the U S two from China, maybe one from Mina, Europe area, you know, somewhere that something like that. I can see that. Okay. Uh, wow. Uh, I guess maybe we should talk. I know you love revenue and I saw my friend Brad Gerstner talking about, um, Nvidia and there is a, you know, pursuant to what we just discussed in this game of chicken, who's willing to build the most data centers and fill them with, uh, Nvidia gear.

42:55My Lord, six customers are now 85 % of Nvidia's revenue in the last quarter. Break it down for us. Yeah. All right. Uh, Nvidia is a company that if you are a gamer, you have known for a long time because they make graphics cards for computers. It turns out those little computers are great in mass scale to handle parallel calculations. And so as the AI boom has expanded, Jason, people that run AI compute are buying mountains of GPUs. Now, there's only so many companies out there that are building what we call hyperscaled compute, which is your Microsoft Azure, your Google Clouds, your Amazon AWS, et cetera.

43:31What this means is the number of companies that are really buying enormous sums of GPUs from NVIDIA, which is the leading GPU company in the world, isn't that many. There's only, like, as you said, you know, maybe six future AI companies. There's only probably five or six hyperscalers in the world today. So unsurprisingly, they buy a lot of stuff from NVIDIA. And unsurprisingly, that yields a really funny customer revenue count. In startups, Jason, if you saw a company come in that had one customer that was 80 % of their revenue, big red flag. You don't want that level of customer concentration.

44:02So people are a little worried in NVIDIA's case. So how big is their number one customer 23 of revenue in their last quarter and that's probably xai or open ai i think have the biggest because colossus was the biggest so it's got to be xai is number one i mean it's got to be top three it's got to be yeah one and two are open ai and xai you got to think and then after that meta which is spending tens of billions as well and doesn't have its own in-house chip division the same way that Alphabet does, for example, with their TPUs or the training in chips over at Amazon's AWS group. So basically there's one customer, 23%, another one, 16.

44:45Maybe it's Microsoft because they have Azure and Azure powers OpenAI. Actually, that's probably what's going on here is that OpenAI is buying them, but they also have a deal where they get compute from Microsoft. That was part of that original hairy deal we were just talking about. So maybe it's Microsoft because they represent two of the language models or two of the big players. Yeah, absolutely. But 23, 16, 14, 11, 11, 10. That's, you know, most of NVIDIA's revenue from just a handful of companies. And this is why people are so worried about NVIDIA's earnings, Jason. Like if you think about, you know, when they dropped, everyone was really holding on tight because if they miss, it implied that the hyperskillers weren't spending, it implied that AI demand was slowing, it implied that et cetera, et cetera.

45:28So it was kind of a linchpin earnings report. You know, this could be a stop and start kind of a situation where it takes time to digest the models. It takes time to finish the products and get them to market. And then the software and the developers become more efficient at utilizing the infrastructure. So there's a confluence of factors here that are going to trend towards growth for NVIDIA. But what if somebody like Alphabet says, hey, we're going to be, we have our own silicon, Amazon's working on their own silicon. I think we have enough H100s and whatever, you know, and we're going to just try to build on our own.

46:11And the long tail of demand here is also significant. There are startups who are using Azure, Google Cloud, et cetera, for their GPUs. So those are long-term gonna keep buying, you would think. Absolutely. Also, OpenAI's Sarah Fryer, I think we mentioned this on the show a week or two ago. She did an interview with CNBC. The CFO, thank you, Jason. And she said that they're still constantly under compute. And so at least for today, there remains a shortage of free GPUs to use at scale, but we are seeing some other changes as well. So a story that we're not gonna focus on today, but I'll just mention it.

46:48Alibaba is mentioning over in China that they're working on their own chips that's going to help out with the intra-China GPU shortage. So everyone wants NVIDIA's market cap. They are the number one most valuable company in the world. You're not worth$4 trillion plus without having everybody taking shots at you. But, you know, to quote The Wire, if you come at the king, you best not miss. And thus far, no one's really managed to slow down Jensen and his team. Yeah, and we have to get our hands on this Claude browser. We had a great experience with the comic browser. It's become - The browser though, Jason, is a way to let Claude run Chrome for you versus a distinct whole cloth browser, just so people are sure.

47:27Yeah, everybody basically uses Chromium, I believe is the name of the open source project, or Mozilla, which is the Firefox project. Both of those are open source projects. You would never write a modern browser from scratch. you would use those basic architectures and then have those, you would wrap them basically. And I use the Brave browser and Comet now. Brave I use because of the protections and getting rid of ads and privacy. And it's fast. I think it's also fast because it's not letting people cookie you and all that's cruft slows you down in terms of web speed. um but tell me about comet i'm curious you've been using it now for a couple weeks i think a month yeah so a month yeah an ai browser browser the future jason has it changed your productivity your workflows how you approach the internet yeah it has um you know perplexity's got a pretty good search product um and the google one is kind of two different products they haven't moved to the AI native search on Google, you can use it.

48:37But Perplexity does a nice job with like a comprehensive search, which was pioneered by Mahalo, my company when humans were doing it, and Naver and Daum in Korea, both of those Korean companies. And so they do a good job at that comprehensive search, not for everything, but pretty solid. And they're taking a Yahoo-like approach now where they have shopping and finance, I think. And so Yahoo Finance for me isn't as good as Perplexity's finance product. So when I'm checking stocks, I find myself using perplexity stock checker instead of Yahoo finance and Google now, um, because it does an interesting thing.

49:12It goes and searches and summarizes the news with AI as opposed to just natively listing the stories. Yeah. And so they're kind of intercepting my clicks from going to market watch or to fortune or whoever's covering what's going on at Robinhood or Uber or Korean defense or whatever I'm in at the moment. So I do find that, what's that? So bad for Yahoo. Yeah. I mean, Jim Lanzone's got to decide if they are going to super invest in AI to summarize their partner's stories. But I think they have a prisoner's dilemma there because I think they make money by sending traffic to their partners. I think their partners pay for placement or because you see some of them say sponsored on them and their news stories.

49:58Yeah. So if you were to summarize them, well, then you don't get the advertising there. All these AI companies don't have advertising products. So their products are kind of like Uber or Lyft in the early days where they're being subsidized by venture capital. And if you could get a summary of all the stories without any advertising or cruft in it, and then Yahoo needs to have the cruft in the advertising, man, that's a huge advantage for making a better product than Google's, which is filled with ads as well. But I do like using it for multi-step processes. So I will pull up the assistant on a page and I'll say, hey, summarize the themes here in this story, or please go find these individuals, LinkedIn's for me and go follow them on LinkedIn.

50:44And it doesn't quite work perfectly yet. And it's all based on Claude, by the way. I use Comet as the browser, but I use Comet as the language model. So very interesting that the browser, you can kind of like, you can pick a search engine, DuckDuckGo or Google or Yahoo, whatever. You can actually pick what language model you use. And I don't think I can use a non-AI browser ever again. Ah, see that's interesting. So it has changed your behavior patterns for life, which means that we actually have turned a new page in browser technology. okay i think so yeah when you have the assistant there at the side when you have the one click summary of the page when you can fire off like a i'll call it i'll get a light agent to do something sure um it works relatively well not good enough yet but if i want to like when i was having my mexico city trip i took i was like go to these websites figure out the consensus of all the best restaurants, you know, near this hotel, uh, in this neighborhood, which was Roma where I stayed, um, and then put it in a table and then give me the Yelp link, the Instagram link, you know, and the trip advisor link or whatever.

51:55And, um, and the Google local link. And it kind of did a decent job at that almost 60 % of what my Athena assistant would do. So I had my Athena assistants do it. I did it. Now I've trained my Athena assistants to use Comet to do this. So I'm using a combination of Athena assistants and the emerging technology. And that's seems to be a winning combination for me for getting more done faster. And if you want to know what it's like to have an Athena assistant, I have two of them. They work seven days a week. They have a split schedule. Some days they overlap by half an hour and I have them in a iMessage group.

52:32I have them in a Slack and I was like, give me the restaurants, rank them by Michelin stars, rank them by the reviews or whatever. They do all that, a little bit of AI, a little bit of their manual stuff. And then I gave them my open table account. And I said, get me two restaurants per night based on this list, one reservation at five 31 reservation at seven 30. Then I said, okay, I will tell you which reservation I'm going to use around four o 'clock. So the restaurant can give it up. They don't mind you doing that. Um, and the ones that are hoity toity, sometimes we'll give you like a$10 per person cancellation fee, which is if that happens to me, I'm happy to give a restaurateur because my dad was one and I know it's a hard business.

53:15I'm happy to give them the 20 bucks if I don't cancel within an hour or two, whatever their policy is. And man, that makes your life real simple. You have options and it really works well. Also, you're making me incredibly hungry because I'm a huge fan of Mexican food. But this brings up a really important polymarket, Jason. and people are trying to figure out what's gonna happen with OpenAI launching its own browser. We've seen reporting that this is going to come. Now this is a relatively small market and I wouldn't normally bring up a smaller one, but there's something very interesting that happens if you click a button, watch this.

53:47If you change the timeframe here on this Polymarket, this one is OpenAI browser by September 30th, not that high of a chance, about a third. If you click on change the date to October 31st, suddenly people think there's a two thirds chance. So it does appear that some people are really wagering that this OpenAI browser is going to come either in the first month of Q4 or around there. So here's hoping they're right, because if Comet is good, then I would love to see more browsers do the same thing competing. Yeah. Shout out to our friends at Polymarket, because one of the great things here, and I don't know what the rules are on this, but if you have inside information, are you, like, if you worked at OpenAI or you were, let's say you're a recruiter and I'm a recruiter and open AI had me recruit 10 people to work on a browser six months ago.

54:40And then they said, we don't need any more. There's no more open recs for that. I might be able to discern, okay, they're getting ready to launch it. And if those people are still working there and I check their socials now, is that proprietary, is that inside information? It's not like I wonder how that works for I mean, I know how it works in sports leagues. Like you're not allowed to like miss your free throws to, you know, get the under. OK, but how would that work on a poly market or any prediction market? So insider trading laws apply to securities. And I know as far as I understand it, we're trying to narrowly define securities to leave crypto alone.

55:17So I'm going to presume that poly market doesn't land there. There are some states that have insider betting laws, but that's not as broad across the nation. And when we're talking about what Polymarket does, it's, is it technically betting? Is it wagering? Or is it investing? Hedging, investing in an outcome is a way to say it. Here's a funny thing. Some experts argue that allowing insiders to trade may increase market accuracy. Yes. Though this conflicts with fairness. Yes, it does. And yes, it would. So they don't, yeah, I think that's a feature here. I want the insiders doing this because it means the data will be more accurate.

55:56So like when I was saying Biden is going to be hot swapped and I made that incredible call as Nostracanis on all in and here, you know, it turned out I had heard from people like, you know, yeah, I wouldn't say, as I said, call them insiders, but in and around the political arena that they felt that was going to happen. And I felt it certainly was going to happen because I thought it gave them a free second chance at winning the election. Right. If you knew you were going to lose and you got this free second chance, why wouldn't you take it? Exactly. Right. And as I explained, like, why would you do a June debate with Biden if the debates normally start in September?

56:33That when I heard that they had done that, like, I guess in May they announced it or April, I was like, wait a second. And I'm not an insider, but I had an insight, which is, oh, the Democratic Party is setting up and stress testing Biden's ability to win this race so they have enough time to hot swap. And that's actually how I made my decision. Good news. Producer Claude just got back to us. And traditional insider trading laws don't apply to prediction markets because they fall under the CTFC's jurisdiction. And the Commodities Futures Trading Commission, I don't think, deals as much with insider regulation versus just keeping tabs on the commodity market.

57:10So it should be OK. Actually, let me rephrase that. It may be OK. This show is not legal advice. And if you do get in trouble, you're not allowed to call us. Yes. We're just fans of wagering and predictions and figuring out what's going to happen next. That's all we are. I do think if you are the principal in this, like if you were Sam Altman and you get to pick when you launch, you're not allowed to participate based on the terms of service at Polymarket. That would be fair. Also, by the way, producer Claude is brought to us by our dear friends over at Anthropic. And we talk about them all the time on the show, but we talk about them most of the time in a non-sponsored format.

57:44Yes. They're a partner on the program, as is Polymarket. Yeah. Lots of friends. We are so lucky to have the top brands partnering to bring this amazing episode. It's not cheap to have someone like Alex here. It's not like we don't. This is an expensive talent you have on this program. Ruinously costly. But if you do want to dig out more about Claude, you can go check it out at claude.ai slash twist. And now I'm going to get five internal points. All right, listen, another amazing episode of This Week in Startups is in the can, as my dear friend Leo Laporte would say. Another twit is in the can.

58:19Shout out to my guy. Shout out to all the people who made podcasting happen. My friend Dave Weiner created RSS. Big part of my career. I sent him a little note just thanking him for the impact his technology had on my career. So thank you to my friend Adam Curry, who created podcasting along with Dave Weiner. He's got a great show, No Agenda. He's doing this like podcasting 2.0 thing. Oliver, I want to book him for the show, please. Adam Curry would love to have Dave Weiner on and hear what he's up to. He's working on some content management systems using WordPress, I understand. And Leo Laporte for being a pioneer in the space and teaching a lot of us how to be broadcasters.

58:59So thanks to all those legends. They don't get any credit. Like Theo Vaughn doesn't know who Dave Weiner is. Theo Vaughn, you know, Joe Rogan knows who Adam Curry is because he has him on his podcast. If you want to see Adam Curry on a podcast, he does Joe Rogan once in a while. Adam Curry had convinced Steve Jobs to in iTunes. iTunes was where music used to exist on your laptop and would put the music on your iPod before there was. Is our audience that young? Yeah, before there was, yes. Before there was an iPhone, there was an iPod. And he convinced them to have in the menu, add RSS feed. Oh, wow.

59:37And you would cut and paste the RSS feed from somebody's blog or podcast and put it into iTunes, and then it would sync it to your iPod. If that hadn't happened and Adam Curry and Dave Weiner didn't do it, there would be no Joe Rogan. There would be no all in. There would be no. Ah, so they're responsible for the second Trump administration. I see. Oh, God. Please don't. It's Dave Weiner. You want to know who's responsible for the second Trump administration? It's pretty simple. The Democrats did not field a qualified team. Joke. Joke. joke we don't even i said dad joke no i just i hate the fact that people think that like this like podcasters created this you have to feel the team yeah this is my message to the democrats like feel the team you can't put rec players you can't put a retired player against kobe bryant you got to have like a sharp player if you're going to be up against like a media savant like trump I mean, a Democratic Party is just so frustrating.

1:00:39Ay, ay, ay. Bring me the America Party. That's what I want. We'll see you all next time. Have a great weekend. Enjoy the long weekend. Enjoy the long weekend. We'll see you guys on Tuesday.