Diet TBPN: October 21st, 2025

0:28You're watching TBPN. closed today. Oh, I'd love to play the video from open AI announcing their browser Atlas.

0:46I like the different sound cues. This one, that sounds kind of like typing on an iPhone. I mean, it's, it sounds exactly like, is that what it sounds like? Okay. I turn, it's funny. I like the sound cue in videos, but I turn it, I certainly turn the sound off on my actual phone. So I don't actually, uh, hear it very regularly. So this is, you're on all trails. Is that the website? So you're in Coursera and you're saying, cheat on my homework. Uh, wall street journal summarizes for me. Interestingly, the wall street journal actually has AI summaries now in every article, just a couple of bullet points.

1:17Uh, there, I don't go to them a lot though. Um, Um, so it does some online shopping. Tyler, were you, uh, were you successful in your prompt? Give me your review. People are saying, you know, it's new browser. It's not just a browser. I think it's kind of a whole new way to really use the internet. Browse the web to browse the web. Will this be successful? Uh, I believe that chat GPT, especially once it had like some close to live search was was five to 10 times better than Google search for a lot of searches. Yep. And my prediction is that Atlas in its current form might be 1.1 times better than Chrome.

2:04Yeah. And that will not be enough to get large-scale consumer adoption. I'm not super bullish, but I think it's great that they're taking a big shot on goal, and it makes sense strategically. Well, if you're looking for a product that is 10 times better, go to ramp.com, time is money, save both, easy to use, corporate cards, bill payments, accounting, and a whole lot more all in one place. Me, my daily driver in a peaceful corner of the internet. Safari power user. Safari power user. First of all, let's say thank you to Sam Altman for constantly creating content. It really is like the bull market in tech news.

2:35There's always something to talk about. There's always a Samma story. Wall Street Journal's OpenAI piece makes you realize Sam Altman must succeed or he could be a real problem for an otherwise sterling industry. Dot, dot, dot. Sam Altman is kind of becoming the preeminent deals guy of the modern tech era. He's like the deals are so big and they're coming so quickly that a lot of people are asking, is Sam just so good at deals that all the counterparties are actually making mistakes by tying their fortunes to OpenAI? If OpenAI doesn't deliver and Oracle does all the build out, but then OpenAI doesn't have the revenue, is Oracle in trouble?

3:14That's the narrative right now. So the question is, like, why are these deals happening? I don't think Sam's gotten any foxier now that he's 40. He's going to be a silver fox. But he's always been a great deal maker. Even going back to his very first company where he got looped, acquired when the company wasn't doing that well. It was sort of one of the first aqua hires that turned out really well. But now everything has three, four, five extra zeros after it. And I think that the reason isn't just that he's become a better dealmaker. It's that he actually has more leverage now. When you have the breakout consumer product of a generational tech trend, you have a lot of leverage.

3:50If you're running AMD or NVIDIA or Oracle or any hyperscaler computing company, and you have to go in front of shareholders and they're asking you, are we doing anything with open AI? the company that's scaled to over a billion users. The thing that will be the next Google, the next Amazon, the next Microsoft. And if you have to go in front of them and say no, they're like, okay, what exactly do you do here? Why are you missing this? OpenAI could swap out the underlying model and still accrue the vast majority of value. So hypothetically, you could have chat GPT, chat.com, powered not by GPT-5, but instead powered by Gemini or Claude and still see positive user growth, monetization, because at the end of the day, most of the users aren't there for the model.

4:41They're there for the app. They're there for the ecosystem. They're there for the platform now. Joel Spolsky in 2002 coined this phrase, commoditize your compliments. And so anything that complements your product, you want to be a commodity so that it can bring more content onto the platform. This was the thesis behind meta open sourcing llama. If it's free to generate content, there will be more content on the platform, just like the Sony camera commoditized the video production, made more content for YouTube. All of their partners are looking at them, whether they're Google with the CPU program, Amazon with Tranium, OpenAI, say AMD is pretty good.

5:16Sam's looking at that margin profile, just being like, Jensen, I'm so happy. The whole industry is losing money. And they're like, we're trying to find the one guy that's making money. Exactly. And they see Jensen just absolutely printing. If only there were another plate that we could eat off of, and Jensen's just there with like the massive Thanksgiving dinner on his plate. And so everyone's kind of incentivized to like work against the video. Obviously, Jensen would have been incredibly frustrated by this for a number of reasons. His reaction, very presidential, was, I saw the deal. It's imaginative.

5:48It's unique and surprising, considering they were so excited about their next generation product. The Pew says, enjoy the vague posting while it lasts, my friends. And soon all the AGI companies go public and ruin an eye shitposts are also called securities violations. Martin Shkreli says jail is not so bad. It's funny because OpenAI is somebody at OpenAI Legal is going to probably send this to Will, his post to him and be like, hey, look, like a lot of people are going to read into this that going to go public soon. Genuine question. For VC firms invested in both Frontier Model Labs and coding application layer companies, how do you handle the discrepancy between the two, since coding is number one on the roadmap of the labs versus other products which are more adjacent?

6:35If you're a company, I don't think there's huge downside here. Like VCs tend to give immense support for their winners. I think something similar might have happened with Palantir and Databricks, which had some same investors actually, but there was this thesis of like, well, if Databricks is the database layer and they allow you to do AI on top of that, like what is Palantir's role? And they, and they, they kind of wound up partnering. Venture investors want to become, want to invest in companies that become so successful that they launch products that ultimately compete in a bunch of different adjacent categories and ultimately compete with a range of their portfolio companies.

7:13Like that is ultimately what success looks like. Yeah. Something I've been thinking about a lot is this dynamic between Anthropic, where all of their revenue comes from having the best coding model, which is very different than having a lot of revenue because you have a hit consumer product because you have to keep training the next model. Pets.com is the go-to slur of the dot-com era. Let's see. Pets.com burned$182 million of total investor capital before going bust. Today, just U.S. online pet goods sales is$38 billion a year, $70 billion total market cap, including portions of Amazon and Walmart.

7:53Lesson, should have invested more in pets.com. Like, makes sense to take a really big swing at what became a$38 billion market. U.S. East is down from Syriac. and it's, what exactly is this image from? Is this like Rome falling? Is this the fall of Rome? This is very - Do we know what caused it yet? Wasn't it database migration or something? Some data, oh no, DNS configuration, I think was the root, something like that. The outage began around 3 a.m. Eastern time when Amazon made a technical update to a widely used AWS database service, DynamoDB, the update which included an incorrect domain name service or DNS information for DynamoDB, kicked the database offline in Amazon's critically important Northern Virginia data centers.

8:42Really exposed to insane dependency. Yeah. So we're taking it on-prem. This was my take yesterday. We're going back to sticks and stones. We're going back to pen and paper, baby. The computer revolution is over. AI is dead. So Tuxedo Sam says, meet James Hamilton. He's Amazon's literal, very top engineer, the brain behind all the data centers. He lives in a custom yacht and does not give an F about any RTO. Wow. So it claims that his yacht made landfall, and this is like in AWS update. Fake news. But it was digitally altered. It was fake news. Moving on. Coinbase and Kobe announced that Coinbase is acquiring Echo.

9:20Yes. Echo is an AngelList-like platform for crypto, so they help groups of people invest in different projects. Yep. They paid$375 million, And so you can think of this as effectively a$400 million acquisition. It's such a creative way to get a little bit more attention on an acquisition that would have been cool, but it wouldn't have had a narrative to it where people were like, this is crazy. Oh, it actually all makes sense now. So it was a really good Mission Impossible rip off the mask. Coinbase isn't actually crazy. They're not actually paying$25 million just for a podcast. One of the most iconic quotes from a tech CEO this century from Chesky said, I was basically going room to room, just pouring out this stream of consciousness manifesto like Jack Kerouac writing on the road.

10:08I basically said, we're not just a vacation app. We're going to, uh, M dash, we're going to be a platform, a community. What's funny is like, I think of Airbnb as a community from like day one. What does community mean now? Like I met my co-founders on Airbnb. I met in YC. My pushback would be that maybe the last 10 times I've used Airbnb, it was to book a vacation rental. Yeah. Yeah. So, so, so, so want to meet the host. Yeah. I mean, also once I was traveling and I was staying in an Airbnb in New York and it was very clear that the, uh, the owner of the house, like wanted to meet me and like wanted to like make friends.

10:45And in that, in that context, I was like, uh, I don't, I don't, I don't know. Like I, I, I'm here on a business trip now. I kind of don't want to make a new friend. Um, you seem cool, but like I'm good. Apparently, Anthropic spent$2.66 billion on Amazon Web Services in the first three quarters of 2025, around 100 % of their estimated revenue. This article is implying that Anthropic has basically 0 % margins, but it's possible that some of, like, they're paying to serve and run the models for customers, for API clients, but they're also training. You can now fundraise in Stripe Atlas. It's very cool.

11:26Stripe Atlas generates a significant amount. I forget the exact percentage, but a meaningful, I think, double-digit percentage of all C-Corps created are created with Stripe Atlas. I built this product at Party Round, and it's very useful. It's not a great business, but it makes sense to, you know, tie it into the Stripe ecosystem. Let me tell you about TurboPuffer. Search every byte. Serverless vector and full-text search. Built from first principles and object storage. Fast, 10x cheaper, and extremely scalable. The world is changed, subtle and obvious. For the first time in living memory, the scope of live player action expands.

11:59Thinking of tech companies as nations, right? If you start thinking of OpenAI in the context of like a nation state, and the nation state is saying like, we're going to need trillions of dollars, we're going to need a huge amount of energy, and yes, it's going to be distributed all over. It's not that crazy for a country to be saying like, we need to spend hundreds of billions of dollars on it's like a digital nation state. It still feels like the U.S. government is at the top of the stack and the tech companies answer to the U.S. government. But I do wonder if there's some element of flipping that will happen at some point or has already happened.

12:37Like, did you know that Walmart spends more money than the United Nations? Isn't that remarkable? Did you see this picture? it looks like somebody found them filming the open AI. Is this real or is this just someone dressed up as Ilya? ChatGPT21 says this movie is going to be sick and it looks like It looks like Joseph Gordon-Levitt. Oh, you think so? He's supposedly playing Ilya, right? Yeah, maybe. Oh, it does look like him. He's playing Ilya? It would make sense that they would film in San Francisco. Anyway, Wild. Wild. Going to be a good movie. Hopefully it pumps some people up. We'll see.

13:14It's probably going to be a brutal hit piece. I wonder if we're going to see any releases of who's playing Rune. I would love to know who's playing Rune. Okay, let's pull up this video. Every story you love. Is this a new MacBook Pro? Is that where we're getting? I don't know. It seems just like a pro-Apple creativity vibe reel. It's a Mac ad, John. It's a Mac ad. It hits extremely emotionally because that ad is voiced by Jane Goodall, who passed away on October 1st of this year. Uh, she was of course the british primatologist. She studied monkeys, uh, studied chimpanzees. Warren Buffett missed out on 50 billion dollars in profits by selling apple too early.

13:59Uh, Barron's calculated that he sold 650 million dollars, uh, 650 million shares at an estimated price of 185 dollars. Now they're at 263 dollars. He might've made more money than any investment manager ever with that trade. And it was sort of a narrative violation because he's often bought these low PE ratio stocks and just generally stayed out of technology. Of course, Apple kind of violated them both, but it was a fantastic success. Crazy because I remember when the trade war hit, the tariffs, Liberation Day, everyone was like, oh, Buffett is a genius, right? He looked So smart for selling when he did.

14:41And then, of course, ripped back up because nothing ever happens. Nothing ever happens. Absolutely remarkable day. Well, thank you to everyone in the chat. Thank you to everyone who's listening and supporting us. Can't wait for tomorrow. We will see you tomorrow. Have a great evening.