LIVE CLIPS
EpisodeĀ 12-15-2025
Years. Right. So we're seeing these creators do this for the first time. Totally. Well, I think on this, like, on the point of Netflix, YouTube, Warner, like, we'll see what happens. But I do think it's interesting that Netflix, in order to compete, is it needs to acquire a bigger catalog. And maybe, maybe, maybe with what they're doing, they're starting to also offer deals to podcasters. Yeah. At what point do they just open up the platform and say, here's an upload button. 2026. Really? Yeah. Whoa. I think there's going to be a creator program of some sort. Because even if you think. That'S more curated, but. But I've heard you. Can do that on Amazon Prime. Yes. I'm pretty sure if you have a calculator, you can just upload. Say. You can do that. I think if you look what Netflix did with Mark Rober, massive YouTube creator. They took his top 10, basically greatest hits from YouTube, repackaged them. They're on Netflix right now, and it was a top ten show on Netflix. Number one kids show. Wow. They also signed a deal with him to make a new series, a reality show. He also did a Christmas special with Elmo, directed by another creator, Daniel Thrasher. So very cool. But that shows. Me that they have an openness to content that I don't think they would have been open to a couple of years ago that maybe they wouldn't have deemed premium. Sure. But I mean, at a certain point, Even just the AdSense is. If you just took the ad.
Years. Right. So we're seeing these creators do this for the first time. Totally. Well, I think on this, like, on the point of Netflix, YouTube, Warner, like, we'll see what happens. But I do think it's interesting that Netflix, in order to compete, is it needs to acquire a bigger catalog. And maybe, maybe, maybe with what they're doing, they're starting to also offer deals to podcasters. Yeah. At what point do they just open up the platform and say, here's an upload button. 2026. Really? Yeah. Whoa. I think there's going to be a creator program of some sort. Because even if you think. That'S more curated, but. But I've heard you. Can do that on Amazon Prime. Yes. I'm pretty sure if you have a calculator, you can just upload. Say. You can do that. I think if you look what Netflix did with Mark Rober, massive YouTube creator. They took his top 10, basically greatest hits from YouTube, repackaged them. They're on Netflix right now, and it was a top ten show on Netflix. Number one kids show. Wow. They also signed a deal with him to make a new series, a reality show. He also did a Christmas special with Elmo, directed by another creator, Daniel Thrasher. So very cool. But that shows. Me that they have an openness to content that I don't think they would have been open to a couple of years ago that maybe they wouldn't have deemed premium. Sure. But I mean, at a certain point, Even just the AdSense is. If you just took the ad.
You don't want to be all gimmick all the time, but you need a little bit of style and a little bit of substance. You need some. We got to talk about, we clearly needed an hour, 90 minutes. I know, I know. I don't know if you got to talk about some of your guys new releases, the chip and some of the new technical decisions that you guys are making around self driving. Sure, yeah, sure, yeah. So last week we actually finally announced it. Amazingly it didn't leak. We been working on this since 2020, early 2022. But we, we've re architected our whole self driving platform around we call an. End to end train model. So the model is using the millions and millions of miles are being accumulated through our deployed fleet of our Gen 2 vehicles which launched a little more. Than a year ago to build effectively. A neural net or a foundation model. For how to drive. And on our Gen 2 hardware stack. It'S, it's, it's using an Nvidia platform. It's got around 200 tops of compute. We'Ve got 55 megapixels of cameras, nice array of radars. But it's a great platform for building. A data flywheel and for delivering ultimately this will be able to deliver point to point, you know, autonomy. So you can put the address in the vehicle drive there. But to get to higher levels of autonomy we've developed an in house processor. Well yeah, the timing, you guys did that perfectly. But the in house processor is a significant step up. So it's A800 tops. It's got 35 billion transistors on the silicon. The neural net's capable of processing 5. Billion pixels per second. So this is like an incredibly powerful platform and we brought it in house just to given the importance of vision. This is a vision based robot the vehicle. But we have other things we're doing. In the vision based robotics space as well. And so we came to the view that having our own inference was going to be really valuable. But it took the better part of. Almost four years to build the team develop it. These are things that take a tremendous amount of capital. We're working with TSMC on making it. But yes, we're excited about that. And then we also include upgraded cameras. And we have a new LiDAR which is another sensing modality that sits at the top of the windshield which will. Help us raise the capability of the vehicle ultimately to what we call level. Four, which is think of it as. The vehicle can operate empty or without. Anybody in the driver's seat. So it could pick your kids up. From school or drop you at the airport. These kinds of things. Is lidar dangerous? I was telling Jordi that there was a lidar system that if you pointed your phone at it, it.
But what do you think American manufacturers can learn from what they're doing? Yeah, I do think it's important to sort of pull the curtain back on what sometimes I think gets presented as if it's magic, particularly on cost. So I think there's two things to take away from the Chinese electric vehicle space. First, there's over 100 different brands and manufacturers in China, and there's only a small fraction of those that I would consider to be in the category of what I'm going to talk about leading in technology and having really robust architectures. But the two things. With that said, if you say there's more than five, less than 10 manufacturers that fall into the categories I'm about to describe, you have an interesting phenomena where a lot of these newer companies in China, for the same reasons that Rivian or Tesla have very different software architectures, electronics architectures, and incumbent OEMs is they started with a clean sheet. And when you start with a clean sheet, you would very quickly arrive at a completely different technology topology than what evolved into cars over the last 50 or 60 years. And what I mean by that is prior to early 1960s, cars were 100% analog. So there are no computers in a car. And the first computer to make its way into a car was ironically was for the fuel injection system. And so for any of the car enthusiasts out there, this is like those original, like Bosch K tronic fuel injection systems that we started to see emerge 1960s and early 70s. And car companies at the time said, boy, we build engines, we design vehicle bodies, we assemble the cars, we don't need to make these little electronic modules. And so they pushed that work to suppliers. Companies like Bosch or Continental would make these little computers that would run the fuel injection system. Then subsequent to that, over the currents of the last 50 years, a bunch of other things started to have a need for computers. And your seats suddenly became smart. And there was a computer that went with the seat, your air conditioning became intelligent, there was a computer that went that your sunroof had a computer. And before you knew it, the vehicle architecture was this proliferation of in some cars, 100 to 150 little mini, what we call electronic control units or computers that run these specific domains like the domain of an engine or the domain of a door or the domain of a seat. And it's precisely the opposite of what you would architect. If you were thinking about it as a clean sheet, you would never say, I'm going to build a network architecture and Software platform. There's 150 different software code bases running 150 different little mini computers which communicate through this sort of klutzy can architecture. That might all be need to be independently updated at various points in the car's life cycle. It's just like it's a total disaster. So what you'd say is I'd have as few computers as possible doing as much as they can. And so the fancy way we describe that now is it's a zonal architecture, it's a computer that controls a whole zone. And so Tesla of course developed their architecture like that. We of course developed our architecture like that, and a couple of the Chinese did as well. And the real benefit of this, besides just taking a lot of cost and complexity out, is that you can make updates really easily. And so if I want to change, let's say, the sequence of events that occur when you unlock the car, I don't have to coordinate amongst 15 different suppliers. The supplier for the horn ECU, the supplier for the door lock ECU, the supplier for the, you know, the interior lighting issue. I can do all of that in like a matter of minutes internally because it's running on our own software platform and it's all of our own code. And so just the emergence of regular updates, improved features, features that respond to dynamic customer needs, I think is a really big shift. And in the west there's two companies that have that, Rivian and Tesla. And then in China there's, as I said, more than five, less than 10 companies that have that. And if you don't do a software defined architecture well, the ability to do like AI integrated into the vehicle or an AI defined vehicle is enormously hard. So you have to have all these ingredients to be able to do the broader architecture well. Anyway, so that's one big difference. And that actually underpins we did a $5.8 billion software licensing deal with Volkswagen. That technology I just described for network architecture, software OS is what we license as part of a big partnership with Volkswagen Group. But the other is that the Chinese companies have just fundamentally lower cost structure, they have much lower labor costs, their cost of capital's.
This scale to support all this R. And D. Let's start with R1T. Electric truck felt very contrarian. It feels still a little contrarian. Truck buyers, not to generalize, but it's very much like, I like my truck reliable with gas. I don't like this new thing. Did it seem like a risk to you at the time when you put. When you said, okay, we're going to do the truck first, we're going to do that so early, what gave you confidence that it would deliver and you could actually win that market over? You mean the truck? The truck market specifically. It's such a large market. And I think we often think of it as like one singular type of buyer, which. Is like the person who's loading up. 2,000 pounds of concrete in the back. Yep. Vast majority of trucks used in the United States are actually used more like cars. So they're, you know, they're used as a daily driver. The, the, the bed is used more for lifestyle activities. So, you know, put a dirt bike in the back or got it, you. Know, kids toys in the back, that kind of thing. And so we said we're not, as a company trying to address the contractor use case. So if you're, if you want to. Load concrete and cement blocks in the. Back of the truck, this is probably. Not the vehicle you want to do it in. It could do that. It's capable of, but that's not the brand aspiration. And it was one of the reasons. We also made it really clear that. We had both the truck and then the sibling vehicle, the suv. And there's so much shared content between the two. But the other thing we wanted to do is just eliminate any questions of whether or not it was capable. You know, you'll laugh, but in 20, I guess, 2019, I would get questions. Like, well, what if it gets wet? Can it drive through deep puddles? And like. So it's just like a very, very. Low level of understanding, I'd say, generally around electric electrification, what that can do in terms of off road and capabilities. And so with R1, we made it really capable. So it's, you know, it can go extreme rock crawling. It can go drive through three feet of water. It can, you know, it can accelerate faster than, you know, most hypercars. So our current quad does 0 to. 60 in, like and a half seconds. It. It's sort of unnecessarily silly in terms of its capabilities, but it's a flagship. Product, so it's there to make a statement. It's there for building the brand? Yeah, on the SUV side. Have you been surprised by how.
AI as a moderation tool sort of under the hood. So, first of all, I do think AI moderation is already like very widespread. I think that's already best practice. I think virtually everyone's already doing that, at least, obviously we've embraced that at snap. I think it's very important. And I also think it's a really powerful tool, again, for customers if they want to fact check something, even if their friend told them. The ability to fact check using AI today is pretty compelling compared to what it was in the past. So when people are talking about concerns about misinformation and this sort of thing, I think AI is a pretty powerful counterbalance to that. What I'm more interested in though, is the use of sequence models to find bad actors and to stop them before something bad happens. So that, to me, is what I think the frontier of Internet safety is currently. It's certainly an area of research for us and something that we're really excited about because we have a lot of information about what bad actors try to do on our platform. We catch a lot of them, but with almost a billion people, it's hard to catch all of them. But we're trying. But what's really interesting is we can train sequence models based on those patterns. What sequence of events are bad actors engaging in? How are they using the platform so that we can predict when someone is about to do something that is against our terms? And so to me, I think that's a lot of like the paradigm that we're going to see in Internet safety over the next couple of years. I think right now there's a big focus on features and trying to tease apart what sort of different features different platforms have. I think there's just going to be a massive shift to being able to identify bad actors even before bad things happen using tools like sequence models. Yeah, yeah. Are you seeing the level of bot activity or at least.
Conversations. Being able to have those conversations, not just with your best friend, but also with, you know, agents like Sparky I think is an interesting opportunity. That's very cool. There's a whole bunch of news. OpenAI Today shortened their vesting period. How do you think about aligning talent, Aligning incentives for talent? You've, you know, this transition, now you're a public company. How do you think about whether or not a vesting cliff makes sense? We were going back and forth on it. I was saying that I'm so used to a one year clef growing up. Coming from the private markets. Private markets. At the same time, there's a lot of people that show up and on day one, they're creating a lot of value. So how have you thought about incentivizing employees with stock? Our perspective on this has really evolved over the years and we learned a ton. Actually, in the early days we did something really different. We had a 10, 20, 30, 40 vesting schedule. So you would only vest 10% of your grant the first year, 20% the second, 30%, 40%. And the idea, and we were in LA at the time, the idea was to really make sure if you were joining Snap that you were joining it because you loved it it, you wanted to be there for the long term. It was very popular at the time. I think it's still very popular in the Valley to kind of hop, you know, spend one year. Yeah. Why do four years when you can build a basket of. Exactly. Yeah. And I think, you know, I think that can be real, that can put a lot of pressure on the culture, especially in a startup where you really, you know, I think it's so important that the team is all in on what you're, what you're building. But you know, I think over the time our view has just been like, to take as much friction out of the compensation conversation, you know, as possible. I think like a lot of this stuff is really frankly, like irrelevant. Right. Like what really matters? Obviously if you look at retention and building a team, it's your leader. Right. The people that you're working with, obviously what you're working on, the culture and that long term orientation. So I think a lot of this compensation stuff is sort of like details and the less complicated it is and the less time you spend talking about it or thinking about it, the better. Yeah. What about other. Just lessons from becoming a public company. We like to joke with Eric Gliman at Ramp One.
Incredibly talented folks with a very, very singular and focused vision. And we've done that over the. Is that part of the pitch? I mean, over the summer, we saw insane talent wars. Snap is in a unique position, like, somewhat isolated down here in Southern California, which I think helps, but is part of your pitch to talent. Like, hey, this is not a program that we just started up and we're going to work on for three years and then the budget might get cut by xyz. Is, like, durability and longevity a part of your pitch? I think that's definitely a part of it. But the most talented folks, they want to create the future, not copy the future. Right. And I think we have a reputation now for leading in terms of innovation in this space. And the folks who are really deeply in this space and who are experts in this space know that what we're doing is leading. Right. I mean, what you experience with specs today, there's no other product that you can. That a developer can just sign up for today and have that type of experience doesn't exist. So I think folks who are really deep in this space understand our leadership here, and that's really helped. Yeah, the pure depth of different experiences in the product, too. I mean, we only.
Community of people who have wanted this product for a super long time are going to be the most excited about it. And that's really who we. How do you and the team think about timing, timing out these bets? Because obviously you've been at. You've been at it for 11 years. And to me, it seems from the outside that it's like trying to pull the future into the present, but at the same time being patient and not betting the company on, like, hey, we need to make, you know, 10x the progress in the next 12 months, otherwise we're killing this product. Like, it feels like more, quite a bit more, like steady and being, being open to again, like, not being, being impatient, but patient at the same time. Is that accurate? I really love that question because I think that is exactly what has killed a lot of Glass's efforts to date. Like, we're going to raise $200 million, spend it all in two years, and then you're sitting there and being like, okay, we made progress, but it's not commercial. It's not a commercial. It feels like we're in an age of research for vision, broadly, VR, AR, the entire category. There's just a lot of. There's a lot of experiments that need to be run and there's no guarantee that it's going to be, oh, next year's the year, or five year. We don't know. This is one of the reasons why we thought it was so interesting to play in this space because we have this huge advantage of being able to invest consistently over the long term. Right? And that is a unique advantage, especially when we're talking about this amount of capital. Right? We've invested more than 3 billion doll in glasses over 11 years, and we've been able to do that very, very consistently. And so at the same time, we've seen companies both large and small, start a Glass initiative, shut it down, start a new one, fire the team, hire a new team. And just the whole time, we have built a team of incredibly talented folks with a very, very singular and focused vision. And we've done that over the last. Is that part of the pitch? I mean, over the summer, we saw insane.
Maybe enterprise is aware all this stuff goes to die. But I'm wondering if there's any place. But I guess the question is, like, I think people assume that VR eventually will be good enough, that a huge swath of the planet will be engaging in it a lot. At least my personal theory. And maybe you believe differently. It sounds like you're even okay with a scenario where even if it does work, you're like, other people can kind of play in that, and we're happy to kind of exist in this hybrid world. Yeah. Just philosophically, like, that's not a world that we want to support in any way, shape or form. You know, I think what's really important is actually trying to create products that foster that human connection. Right. And in doing so, in person, I think we're in a moment where people are really valuing that. Right. And they want more of that, and they're thinking about how to get more of that. And I think AR is coming about at a really important time when people are saying, actually spending a lot of time in front of a screen. I'd like to spend more time together with my friends. I'd love to do that in a way that's more fun or entertaining. Our kids, even. You know, our kids are. At least three of them are still pretty little. But, like, when I see them playing around with those specs, even though they're clunky and heavy or whatever, they're outside running around, playing together, and you're like, that's a different vision for computing. Yeah. I mean, you should. You should see us doing the demo this morning. I was running.
It's so much stickier. And it's because they've developed this services monopoly, essentially, where they have aggregated, you know, it's the Ben Thompson aggregation. It's a toll road for your life. It's not a toll road. That's a great thing. There is a difference between a toll and a tax. And a toll is something that you pay that is directly linked to the service that you're getting. Okay, so your toll road would be more like a toll road now? Yes, now it will be a toll road. And we should celebrate that. Or developers should celebrate that. Tim Sweeney should be celebrating that. Because a toll road is something where it's like. It's like, I'm paying for $5 to drive down this road. That money goes directly towards this road. Yeah. The tax structure applied to a toll road is like, what economic value are you creating a road? Exactly. We're 30% of that. Exactly. Oh, you're. Oh, you're. You're. You're transporting a shipment of televisions on this road. Those are high margin. Or, oh, you're, you know, oh, you're a rich person driving. You have some GPUs. You're hiding your truck. You have some GPUs on your truck, are you not? Yeah, yeah. Oh, oh. You can certainly break us off more as opposed to saying, you know, every time someone drives on this road, it takes a dollar of depreciation. We need a million dollars to repave it every year. And so we need to link the cost of using the service with the actual underlying cost of OPER operating that service. My question with these changes is what is the consumer experience going to be when trying to cancel subscriptions?
Product that I'm like, hm, maybe I want that. But if it's on Tik Tok Shop, I'm like, I don't rough. Yeah. Blanket dis. Endorsement. Is that not. But, like, do you guys not. I've not even. I don't even have Tik Tok installed. These guys are deep. Negotiation with Tik Tok, I've never been met with. It's the opposite. I actually. Hardest dig for this year. No, we're. We're bad creators. We don't have Tik Tok on our phones. We get some there. We have one friend that sends us Tik Tok links, and I've never watched any of them, not a single one.
Serious guys. Yeah, yeah, yeah. So you're bridging the gap. But anyway, give us an update on what's new in your world. Yeah, so we just announced a really exciting new program today that I'd love to tell you and your listeners about. Of course, it's called U.S. tech Force. Yes. So this is all right, this is a two year program where we are recruiting a thousand engineers, product managers, data scientists, a AI specialists into government. You'll work in government for two years. Literally every agency in the government basically is participating in this. So if you want to work at Department of War or Health and Human Services or State Department or irs, whatever you want to do, like we've got opportunities and the whole idea is how do we modernize the entire federal government infrastructure? So we've got a real challenge in government. Number one is just obviously we need more smart people who've got like modern software development, modern AI expertise. And then we're also really have not done a good job of recruiting early career people. So if you look at kind of people earlier in their career, only about 7% of the federal workforce is early career. And at all the companies that you guys talk to on a daily basis, I bet you that number is like 25 or 30%. So we are by at least a factor of three to one in a real world of hurt in terms of being able to recruit and retain early career people. So this is a two year program. We're doing this in partnership with about 25 of the tech companies that you all know and love. So, you know, Coinbase, Robinhood, Databricks, Snowflake, Nvidia, Xai, OpenAI. And what those companies are doing is they're going to help us kind of create a program around this. So in addition to working in your day job in government, we will have a speaker series with, you know, we'll get Sam Altman to come, you know, talk to you and tell you about what it's like to work at OpenAI. We're going to do career development. And then at the end of the two years, these private companies have all agreed to kind of participate in a job fair where we're going to showcase all the work that these guys are done. And you know what, if you want to go in the private sector, God bless you, go do that. If you want to stay in government, we'll find a job for you. But we're not asking you to make a 40 year commitment. We're asking people to do, you know, do good for their country for two years, solve some of the world's biggest and toughest problems, and then we will gladly help you in terms of your private sector career opportunities. A thousand people? How.
You're watching TVPN. Today is Monday, December 15, 2025. We are live from the TBPN Ultradome. The temple of technology, the fortress of finance, the capital of capital. It's Christmas week, baby. And the Christmas decorations continue to grow in the TVPN UltraDome. Our good friend Tyler Cosgrove over there is feeling the holiday spirit. He's looking fantastic. And if you're wondering if he has shoes and socks and pants, put the shoes up on the table. But the shoes. It's a full costume. Yes, that is fantastic. Almost as fantastic as ramp time is. Money save both easy use, corporate cards, bill pay, accounting, a whole lot more all in one place.
Stuff is sort of like details, and the less complicated it is and the less time you spend talking about or thinking about it, the better. Yeah. What about other. Just lessons from becoming a public company. We like to joke with Eric Lyman at Ramp, one of our buddies, that he's maybe about to go out. What is he preparing to do? What are your recommendations for CEOs that you counsel about. The process and the. Transition that the company goes through in the IPO process? Just culturally. I mean, the most substantive change is that people's compensation changes on like a minute by minute basis. Right. And like that can be very distracting for people, especially in the early days. And when companies are newly public, there tends to be a lot of volatility. Right. And so I think it's just really important again to create a company culture that's not focused on that, especially if you want to build something meaningful over the long run. I remember hearing a story about Enron pump putting the stock price in the elevator. So like the exact opposite of what you said, like literally orient the entire company culture around the stock price. So you come in, stock price is up, everyone's having a good day, stock price is down. How did that work out for them? I don't think it worked out well. I don't think. What about, what about setting? What about setting? You seem like, you know, through.
To feel the music. You're watching TVPN. Today is Monday, December 15, 2025. We are live from the TBPN Ultradome. The temple of technology, the fortress of finance, the capital of capital. It's Christmas week, baby. And the Christmas decorations continue to grow in the TVPN UltraDome. Our good friend Tyler Cosgrove over there is feeling the holiday spirit. He's looking fantastic. And if you're wondering if he has shoes and socks and puts Tom Dice. Put the shoes up on the table. Put the shoes on the table. It's a full costume. Yes, that is fantastic. Almost as fantastic as ramp time is. Money save, both easy use, corporate cards, bill pay, accounting, a whole lot more all in one place. Get started@ramp.com so Tim Sweeney was taking a victory lap. I sort of missed this. This was last Thursday. Basically, the ninth Circuit struck down Apple trying to do something else in the Apple tax battle with Epic Games. So a quick, quick refresher on this. Tim Sweeney says the Apple tax is dead in the United States. This particular nail in the coffin comes from the ninth Circuit. These things are never fully over, I've learned. Like, it's just, it just, there's, there's a class action lawsuit, then there's another lawsuit, then there's this one, then there's appeals, then they go to the Supreme Court, then they go back to the Supreme Court. It's always up and down. Like, that's just the nature of these things because the stakes are so, so high. Exposure seems a little hot on that. What's going on there? But basically, the court had said that Apple could not charge 30% if a developer routed an app customer to their own payment page. And so Apple was like, yeah, totally, we're cool with that. How about 27% plus 3% for payments and 27% for like, IP licensing. And so the end result was exactly the same. It was literally 30%. It was just like structured slightly differently. They just changed the language. And so they were, they were, you know, this was contempt. Like, you know, it's like we told you not to do this and you're still doing it. Apple. And so now they're not supposed to. Of course, like, the weird takeaway here is like, these big momentous things happen and then the stock moves, like, not at all. Because, of course, my conclusion from digging into this was that the fundamentally, like, consumer behavior has built up over almost two decades now. I mean, the App Store launched, I believe, in 2007. This 30% fee, this initial, this whole saga starts in 2011, this guy, Robert Pepper, he sues Apple along with three other plaintiffs. Mr. Pepper, Mr. Pepper. Alleging that he was overcharged for iOS apps. Imagine, I mean, that's not exactly what happened because it's like a class action lawsuit and some much bigger stakes. But it's funny to imagine a guy just saying, I'm coming for you, Apple. You know what, flappy bird? I was charged $2 for. It should have been $1.70 or something like that, a $20 and taking it all the way to the Supreme Court. I mean, if he's a flapping of whale and he spent millions, tens of millions of dollars in the app. But yeah, the economics are a bit, you know, you wouldn't normally see somebody like that suing Apple because they're like, you overcharged me by $2,000 across the lifetime and I'm suing you for damages. And it's like a lawsuit that, you. Know, with class action lawsuit, obviously you get a couple plaintiffs who are exemplary of the problem. And then when the settlement happens, it's like billions of dollars paid to everyone who ever purchased an app. Right. And you see these things before, like, did you use Facebook between 2016, 2017, you may be entitled to like 5 cents. Right. Class action settlements happen all the time. Obviously, massive economic incentives for the, for the lawyers who fight them. Anyway, before I continue, let me tell you about linear Meet the software. Meet the system for modern software development. Linear streamlines work across the entire development cycle from roadmap to release. So basically the, you know, Apple's gonna work to. So now the court ruled that Apple can't do that anymore. They can only collect fees that are in line with actual costs of facilitating links. Links to out to other, other payment processors. Costs for one link. I know, I know, it's crazy. And the intellectual property, we need to use AI for this and we need to. No, no, no. That's actually what's going to happen. So basically right now it's like the, like if I'm, I'm the, I'm the App Store, I'm Apple, I have the App Store. You have your own app and you have your own Stripe account and you want to accept payments your way. Apple says, well, even though, yes, you are checking out on your payment Rails, it's your app. Like, I created the link and the technology that creates links within iOS and that is helping you. So you got to pay me an IP licensing fee. And typically it was like 27% of whatever you make, which doesn't make any sense. Because obviously my, as Apple, my, my costs don't scale proportionally to your revenue. It's. It's linear with regard to my cost. So like, yes, if Apple probably. These are pricey links, John. These are pricey links. Like, you're laughing, but like, like, realistically, like the iOS team that has been working on just, just links like, there's, there's code there, there's code there. Like they get paid a lot. It's probably in the millions. It might be, you know, across all of the different amortized. I know. But if you're, if you're a. Is a lot. If you're a successful mo or you've been paying millions of dollars to Apple forever and you're talking about millions, like relatively fixed opex for Apple that again doesn't. And they make like 30 billion a year or something. So it's like they have paid for the R and D fully. But basically the idea is like justifiable costs should be like 10 bucks, maybe like a hundred bucks for reviewing an app and just saying, okay, we ran our software. We understand, is this violating any rules? We maybe had a human and pop by and look at it for a couple minutes, make sure that this is compliant with the App Store. And then there will obviously be other costs. But should it be millions of dollars? Should it be proportional to revenue? Should it be 30% of revenue? A lot of people have been arguing no, but of course this will go back and forth and Apple will probably try and make the fee as high as possible, of course, because they have every incentive to. But what else is interesting about this? So. Oh, the other interesting thing is in Pepper versus Apple, there was this question of where in the chain is the monopoly pricing having an effect? Where is it increasing the price of the good? And so this all goes back. Ben Thompson's like the best chronicler of this whole saga, of course, and he goes back to this. Illinois Brick Co. Vs. Illinois, the state of Illinois. And so this was in 1977, and the Supreme Court held that only direct purchases of direct purchasers of illegally priced goods had standing to sue. So the Illinois Brick case, this is pretty interesting. He says the value chain was very straightforward. Concrete block makers, including the eponymous Illinois Brick Company. Great name for a company that makes concrete blocks. Illinois. They were accused of colluding, colluding to fixed prices of concrete blocks, which were bought by masonry contractors. Masonry contractors in turn submitted bids to general contractors for construction projects, which were ultimately paid for by the state of Illinois. And so the state of Illinois sued for damages, alleging that the higher prices resulting from the price fixing had been passed through to the state of Illinois. So even though the masonry contractors were like, okay, yeah, like the Illinois brick company is with all the other brick companies, they're jacking up prices, it doesn't really matter because they just passed that through. And so there's a question about like, well, you didn't actually pay the higher price directly, state of Illinois, but it was passed through. And so that that harm gets passed through. And so, you know, he says, in this value chain is obvious who the direct purchasers were. Masonry contractors. To the extent the state of Illinois suffered harm, it was indirect pass through harm. Thus, the supreme court ruled that the state of Illinois did not have standing. So the state of Illinois could not sue. Then if every party in the value chain were to sue, the infringing party could be the subject of duplicative recovery for damages. And parsing out the share of damages would be extremely difficult. So it's the masonry contractors that have to sue. Now in Apple vs Pepper, there's this question of who is harmed by Apple's alleged monopolistic practices. According to the plaintiffs, the value chain looks the same as the concrete block manufacturers. Basically, there's developers who sell their apps to Apple who sell those apps to consumers. But Apple said, whoa, whoa, no, we're not a retail store. Even though it's called the app Store, it's not a real store. We don't buy inventory and sell them. Exactly. We're more. We're an agent. And so the company argued, Apple does not buy and resell apps. Instead, Apple acts as an agent for developers and says, as respondents note, this is from the actual court case, the developer agreement confirms that Apple acts as an agent for app providers in providing the app store and is not a party to the sales contract or user agreement between the user and the app provider. Thus, respondents concede that the direct sale is actually between developers and consumers, facilitated by Apple as an agent and conduit. And that sort of makes sense. You know, you go to a grocery store, they buy the apples from the farmer, they sell them to the customer, they take possession. A real estate agent facilitates a transaction. Doesn'T fee, takes a fee, takes a. Fee, takes a fee. If Apple, if Apple was actually going and like buying licenses and then reselling them, you know, they'd be negotiating like crazy too, and be like, we'll give you a dollar and then we're going to sell for $10. You could argue it'd be even worse for developers in that situation. Yeah. And so the, the, the interesting thing about, about this, this whole case is just, just how much the reality of the shape of the App Store and the business of the services narrative changed the, the entire financial story of Apple over the last, I guess, decade and a half. Yeah, the. If you want to look at Apple stock, you should head over to public.com, investing. For those who take it seriously. They got multi asset investing. They're trusted by millions of. But what you should do if you go and you look at Apple is look at the price to earnings ratio because back in 2011 it was 9.7, let's call it 10x price to earnings. Today it's 37x. So obviously the business has grown, but the value of the earnings is so much higher. Why is that? It's because it's so much stickier and it's because they've developed this, this services monopoly essentially where they have aggregated, you know, it's the Ben Thompson aggregation toll. Road for your life. It's not a toll road. That's a good, that's a great, that's a great thing. There is a difference between a toll and a tax. And a toll is something that you pay that is directly linked to the, to the service that you're getting. Okay, so you're toll road be more like a toll road now. Yes, now it will be a toll road and we should celebrate that. Or developers should celebrate that. Tim Sweeney should be celebrating that. Because a toll road is something where it's like, it's like I'm paying $5 to drive down this road. That money goes directly towards this road. Yeah. The tax structure applied to a toll road is like what economic value are you creating by driving? Exactly. We're going to charge you 30% of that. Exactly. Oh, you're, oh, you're, you're transporting a shipment of televisions on this road. Those are high margin or oh, you're, you know, oh, you're a rich person. Driving, you have some GPUs, you're hiding your trucks on your truck, are you not? Yeah, yeah. Oh, you can certainly break us off more as opposed to saying, you know, every time someone drives on this road, it takes a dollar of depreciation. We need a million dollars to repave it every year. And so we need to link the cost of using the service with the actual underlying cost of operating that service. My question with these changes is what is the, what is the consumer experience going to be when trying to cancel subscriptions? Because the one aspect of the App Store that I've always appreciated is the ability to one click cancel from within the App Store and not having to go to individual apps or services and cancel. Yeah. And so I do wonder as soon as you let payments live outside of the App Store, everyone has experienced any type of software or retailer making it hard to figure out how to cancel a subscription. And so will Apple keep that kind of one click cancellation ability within the app or will they actually have to? No, I mean this already exists because like you can go put your credit card down in Fortnite. The thing is that I think the subscription revenue is not as much as you think. It's not as much of a driver like the in app payments. The one off clicks like those are much bigger driver of overall economic activity. But I don't know on the subscription side it would be interesting if there's like some sort of re aggregation at like the stripe level or something like that. I don't really know. But in general the interesting thing is that like so Apple's price to earnings goes from 10 to 40. Basically like massive run up. And this is all on the back of the services narrative that Luca Maestri, the CFO sort of outlined in I think 2016. He said each quarter we report for our services category, which includes revenue from itunes, the App Store, Apple Care, icloud, Apple Pay, licensing and some other items. Today we would like to highlight the major drivers of growth in this category which we have summarized on page three of our supplemental materials. The vast majority of the services that we provide to our customers, for instance apps, movies, TV shows, are tied to our install base of devices rather than to current quarter sales. And so he's saying like you need to stop thinking about our financial performance as driven by how many phones did we sell this quarter. You need to think about just how many users we have broadly and start valuing us more like Google, more like Facebook. Like, you know, if you're an iPhone user, it doesn't really matter if you bought a new one this quarter. Obviously that's sort of continuing. But the big, the big game is better monetization on top of the user base. We should try to get Justin Kahn on to talk about his company Stash, because Stash is effectively payments built for game developers. Interesting circumnavigate the App Store and so they're very well positioned here. Yeah. So Luca Maestri went on to say for some of these services such as content, we recognize revenue based on transaction value. For some of these services, such as App Store, we share a portion of the value of each transaction with the app developer and we only recognize revenue on the portion that we keep. To fully comprehend the scale of the services that we are delivering to our installed base and how fast this business is growing, we look at purchases in addition to revenue. When we aggregate the purchase value of all the services tied to our install base during fiscal 2015, it adds up to more than 31 billion. That's an increase of 23% over 2014. So he's like. And so Ben Thompson has some funny quotes here. He says like, first off, it's striking that when Apple was facing one of the most challenging years in the stock market, its first response was basically to make the plaintiffs point in Apple versus Pepper. Suddenly the company wanted to recognize all of the app revenue, a portion of which is shared with developers. Sounds like a company in the middle, sounds like a tax. And then secondly, we repeat that original line. You said they've grown 34%, 31 billion. They increased 12, 23% over fiscal 2012. 2014 was the historical growth. So basically what's going on is Luca Maestri is the CFO of Apple and he's, he's having trouble in the market because it's 2015. And what's happening? Well, you're eight years into the iPhone. 2007, the iPhone comes out, it's expensive, it doesn't have 3G, it's got a. Lot of rough edges, doesn't have copy and paste. But it's cool and it's interesting and there's lines out the door for them. And people at the higher end, like a cell phone back then was like 100 bucks. Or you'd get it for free. You'd get it as part of a similar bundle. Yeah, yeah, yeah. Every two years you'd get, you get a new phone and it was free. Basically. Then the iPhone comes in, it's $600, very expensive, very upmarket. Then a year or two in, they start figuring out how to bring down the price. It comes down to like 300, 400. There's these incentives for signing up for a year long plan. There's a whole variety of things that make it. The App Store comes out, there's just more functionality. You don't, you no longer are like, well, my BlackBerry still does this, but Apple doesn't. It's like, no, they do the enterprise stuff. They checked all the boxes and so it's growing, growing, growing. But eight years in, everyone who has one, they won the game. And so Device, like actual. The device install base device sales are starting to flatline. And so then they need sort of a new narrative for the stock because it's sort of getting beat up because Apple's basically winning the sold everyone an iPhone, but it's over. Like the trade's over, right? It's like, yeah, we get it. Everyone has smartphones now. So the, the, the iPhone iPhone revenue is basically or unit sales are slowing significantly. Of course they're able to raise prices still because people are locked in like it's a good business, but it's not this like incredibly high growth thing anymore. So Luca Maestri needs to come in with a new narrative and that's the services narrative. And so the services narrative is saying, hey, for a long time you've been looking at this bucket of basically like other revenue. We have device sales, which you've been obsessed with as the investor, as the Wall Street. You've been obsessed with how many iPhones we're selling, how much we're selling them for our margins on those iPhones, how many we can make, all of that. And then we've had this other bucket which is like itunes, App Store, Apple Care, icloud, Apple Pay license. There's a bunch of other stuff. They were treating it like this storage feature on the iPhone where it's like, hey, you have like ph photos and these other things that are taking in apps. And then like, don't worry about other. Yeah, don't worry about other. Don't worry about other. We're just going to throw everything in there. Don't worry about other. Because we don't really know how it's growing. Is it that high margin? We don't know. Then all of a sudden it became like, don't just not worry about other. Like in fact focus entirely on, focus on it exclusively because it is the best revenue. That's super high margin and it's growing really fast. And oh, by the way, if you zoom out and you look at the economic activity that is driving on top of the App Store, that's, you know, yes, our take rate's 30%, but on top of it, just in 2015, it's $31 billion of economic activity in that ecosystem. And so this is like an admission of like the economic power that they have. But Ben Thompson was not a fan of it. I mean, I mean he was a fan from the stock price perspective, but he had some really, really harsh words he said at the time. It seems incredibly worrisome to me anytime a company predicates its growth story on rent seeking. It's not that the growth isn't real, but rather the pursuit is corrosive on whatever it was that made the company great in the first place. It's like, whoa. It's sort of like okay, they're going like private equity mode. Like the beautiful art. The creativity is gone at this point and that's a large. Tim Cook has effectively done exactly that. Yeah. And he says that is a particular. Again, this is from 2018. That is a particularly large concern for Apple. The company has always succeeded by being the best. How does the company maintain that edge when its executives are more concerned with harvesting the profits from other companies innovations? So going forward the growth story of Apple has been someone else innovates, someone else creates an app and we'll take 30% and we don't need to do the innovate. The innovation. That doesn't need to happen here. Yeah. And that's a big shift in the narrative. Whereas before, all through 70s, 80s, 90s, 2000s, it was like the innovation comes from app. They're like we're going to make the iPhone 1700, it's going to be newer, lighter, better, faster, stronger and they're going to buy it and then we're going to, then we're going to take our cut from everything on top of it. Exactly, exactly. Anyways, should we. Yeah, we continue? Yeah, we can move on. Let me tell you about Apple's newest partner, Gemini. Gemini 3 Pro, Google's most intelligent model yet. State of the art reasoning, next level vibe coding and deep multimodal understanding. Megan. Megan in the Wall Street Journal has a scoop. OpenAI ended a policy earlier this week that required employees to work at the company for six months before their equity vested. They only had a six month cliff. Yeah, it was already, already short. A few months ago. XAI shortened their waiting period, known as a vesting cliff, from 12 months to six. And then there's been. I don't know if you want to start reading through it at all but. Change to the vesting cliff announced by applications chief Fiji Simo came on our show last week. Is designed to encourage new employees to take risks without fear of being let go before accessing the first chunk of equity. Interesting. So. So they had a problem where people would come in and say okay, I gotta just do politics for the first six months because I don't want to get fired before. That's like, that's a crazy culture. I feel like I think this has to be more reactionary to meta. Yeah. And I don't know. Yeah, let's. And obviously some of the other. OpenAI has shortened its vesting period for new employees to six months from the industry standard to 12 months in April. We have to, we have to ask Evan Spiegel about how he thinks about vesting periods and running a public company where your, your employees can, you know, sell the stock that they get. He's the perfect person to ask about this, this exact article. Let's see. Elon Musk's X AI and OpenAI competitor made a similar change in late summer. People familiar with the matter with the matter said XAI didn't respond to a request for comment. Of course the decision to loosen or do away with restrictions meant, meant to ensure new hires stick around reflects the frenzied competition for top tier technical talent within AI. Within the AI industry, tech companies typically have a one year vesting cliff. That's what I'm certainly familiar with for new employees, preventing them from having to give away stock to hires who leave quickly or don't work out. But with AI companies, including meta platforms, Google anthropic wooing top researchers with pay packages that can be worth $100 million or more, researchers and engineers have been able to hold out for the most attractive terms and in many cases have been quick to leave jobs they have found not to their liking. Yeah, I think that this is just like a sort of a red queen's race, just if you give a mouse a cookie sort of situation where competitive. Market employees end up benefiting. I also could see it them trying to kind of rehab their employer brand. You remember there was a bunch of. This was probably six, eight months ago at this point open. I had a bunch there. There were some articles surrounding they, they had some like really restrictive exit agreements. That's right. Right. Yeah. A lot of people pretty frustrated around a year ago. That was maybe, maybe like a year and a half ago. About a year and a half ago. Yeah. That story broke that if you left and did maybe didn't sign a non disparagement agreement or NDA. Yeah. If you didn't sign it they could claw back your equity. And then they were saying that was. Important during the whole like Ilia ousting thing. Yes. A lot of the employees left after that but then they couldn't talk about it. Yeah. That's interesting. I always thought that like the bull case for that was well if you're going to be whistleblowing on something that's like a true AI doom scenario, well then money doesn't matter. So you shouldn't matter about them clawing back your equity. But of course there are more like mundane reasons why, you know, you might want to talk about your previous competitor. Previous. There was another line that some people zoomed in on here. Connor Sen over at Bloomberg Opinion says the company expects to spend 6 billion this year on stock based comp, almost half of its projected revenue. And I saw this post going viral in a few different areas. But a company that creates however many hundreds of billions of dollars in value in a year and then has 6 billion of a non cash expense, it doesn't seem that crazy. Yeah, no, not at all. It seems like fairly low. I mean it's a $500 billion company, so it's 1% of the market cap that's going out to employees. And also they have to be in like an incredible talent war. They're growing. There's just so much to do there. The number that was shocking to me was that you remember last week how there's the rumored SpaceX IPO that is apparently continuing. So SpaceX schedules bank pitches for IPO is in the Wall Street Journal Journal. But the SpaceX IPO if they raise 30 billion will be lower than the 40 billion that OpenAI. Yeah, in this private, in the private markets. And so there's been this big question about are the private markets tapped? It appears it's the David Goggins thesis market. The David Goggins thesis for life. I mean if you went back like I don't know, five years and you were like, yeah, like do you think you could raise $40 billion in the private markets or would you have to go public for that? They're like, you know, the Saudi Aramco level funding, yes, you'd have to be public but turns out you do not have to be public because in fact 40 billion is available in the private markets if you're open AI. Let me tell you about numeral compliance. Handled numeral worries about sales tax and VAT compliance so you can focus on growth. Let me continue with this. Tech investors have privately complained. They're complaining. What do they have to complain about? We're in a bull market. They privately complained about the ballooning stock based compensation associated with fast growing AI startups, arguing that it eats into shareholder returns. Companies that are needing to be more competitive are dropping the traditional first year vesting. Cliff Co founder of Levels FYI says in August after Meta Chief Executive Mark Zuckerberg launched that full scale raid on OpenAI staff and offered giant pay packages, OpenAI gave some of its top researchers and engineers. A one time bonus with some employees receiving millions of dollars, the Journal previously reported. Yeah, I was thinking about that with the. Mark Chen went on Ashley Vance and was like, I don't compete with meta dollar to dollar. Like I won't match their offers. He won't go ban for ban. He won't go ban for ban, but it doesn't mean that he won't fight cliff for cliff. He'll go Clifford Cliff for sure. And also he'll go like one time bonus during a crazy raid if the time calls for it. So it's not like he's not participating in the war at all. He is a little bit, but I think it's the right move. Yeah, I do think that there's been this, I don't know, is it hedonic treadmill. What is the correct metaphor for what's going on with these offers? But it's very clear that like, you know, whoever, like the details of the packages are leaking at least through internal like back channels. And so the next person that gets hired wants something a little sweeter, like something a little bit extra, little cherry on top. And so you go from, you know, four year vest one year cliff to three year six months to no cliff to, you know, add more zeros on the pay package, more guaranteed, more upfront. But at the same time there is something to, I mean the reason I would argue in favor of no vesting cliffs here is that it does feel like the value that you can get from a new engineer joining from a, from a competitor at least is on day one. Like day one. Okay, you were, you just left Anthropic. We're going to plug your brain. We're going to. You just left open AI. You just left Meta. What's going on there? Just take us through it. Right? Like that's going to be the day one orientation. And so are they adding value on day one, like basically right? I would imagine so to be capturing a little bit of that value. That seems like a win for the AI researcher. At the same time, it is just another sign of the froth in the AI market. Finbarr says. Okay, but seriously, why is everyone leaving Meta? People are speculating in the comments. Jane says vest. Finn says, but if you stay longer, you vest more. True. Someone else says no direction for the company. Several major unannounced RL project canceled. The 100 million boys reportedly do zero work because they know that if Zuck fires them, he'll look foolish. I don't know. I think some of these $100 million men and women just that do they are about that life. They do want to pursue greatness. They do want to do the biggest run. So who knows? I'm sure there's some instances of that Andrew says behind state of the art on their models. Of course that's why Zuck, you know, overpaid for talent is to try to catch up no PMF on their consumer AI products even though they forced everyone on Instagram to see it. I think we got to wait a little bit for the Christmas season to come and go and see how these things are selling. They don't even use llama models internally to code. That makes sense. It is a tough position to be in because they don't have a public cloud. They don't have a cloud service even though they are a hyperscaler. So. With Google, like even if they. Even if Google gets completely smoked by OpenAI and ChatGPT and ChatGPT winds up being like truly like, you know, the, the Facebook of chat apps and it's like 99% people use it captures 99% of like the consumer AI value. It's like, well like having a Gemini API is still extremely valuable because like you have cloud services. It would be odd for Meta to wind up like trying to build a public cloud service like around the Llama models. Like, I mean even if they went. Cloud did say at one point though that they would be open to reselling capacity. But then you're going up against Anthropic directly. It feels very, very tricky. I don't know. Anyway, let me tell you about Cognition, the team behind the AI software engineer Devin Crush your backlog with your personal AI engineering team. What else is going on? VCs funded OpenAI and in return OpenAI automated them. This is GPT 5.2 thinking is able to construct a cap table accurately in spreadsheets. This is what Fiji SIMA was talking to us about about emerging capabilities. Seems like they might have RL'd on some spreadsheets. That's pretty cool. Truly everyone is going after spreadsheets at this point. There's a everyone from our sponsor ramp to Microsoft to a number of startups that have been on this show talking about Excel agents, Excel plugins, Google Docs plugins. I'm so interested to see how that market actually shapes out. Who is there's so many AI for Excel plays, some full stack, some plugins. Yeah and it's not that Harvey was pitched as a word plugin like even though lawyers do a lot of work in word. That was not the. That was not the instantiation of that product. It was much more just like, you know, full service product. It was interesting. Anyway, there's a clip of Sergey making a joke about automating away vc. Should we play this video? Pull it up? Have the ability to pull this up? Let's do it. What does he say? No audio on our side yet, folks. Any luck? Well, in the meantime, let me tell you about. Here we go. Can do that Human majority of jobs. That we know today, like more than. 50% might be replaced by machines that can do that human judgment piece better. Well, we've been working on the venture investment machine learning. No, I'm just kidding. That's kind of true, actually. As long as I can buy one, I'm good. You think you'll do it? That is kind of what Google Ventures does. No, they started that way. I don't know if they're actually doing that. Like, I don't know. They keep hiring partners for whatever reason, so. This is actually hilarious. Where was this? Extremely candid. This is very funny. Love it. Anyway, let me tell you about Adeo, the AI native CRM. Adeo builds scales and grows your company to the next level. What is up with this? The CIA lost a nuclear device in the Himalayas. Did you read this article? Apparently this is not new. It's not a new story, but there's some new reporting in here. Okay. What happened? How do you lose nuclear device in the Himalayas? I want to go through this. My only concern is that it's going to potentially take like an hour to read through. Let's do it another day then. Yeah, we can do it. Let's do it tomorrow. Let's summarize it and put it. Let's put a link to it in our newsletter where you can get tech analysis and news. Get our daily op ed, top headlines and more. Sign up@tbpn.com oh yes. You want to go to someplace? Take us wherever. I was going to take us to Brett Adcock. Okay. What's going on with Brett Adcock? A party over the weekend with deadmau5. Okay. And let's play this video. So I. He posted earlier today that that figure is looking quite cheap. Yes. In comparison to SpaceX. Okay. Right. Because there's a. There's a new biggest line on the private markets chart drinking on here. What is he drinking? I think he's having a. I think he's cracking open some cold ones. This is crazy. So my theory is that they have done billions of Dollars in sales. Okay. And this is kind of their way of signaling. Signaling if you know, you know. Because of course it would be absolutely insane to have a party like this if you still hadn't shipped a product. If you still were kind of a sort of. I don't know if they're pre revenue but certainly being valued on Vibe. So there's no way they would throw. A party like this because SpaceX has 30 billion in revenue, right? Yeah. And it's going out at 1.5 trillion. So it's at like 50x revenue. So to be cheap he probably trading at like what, 20x revenue, something like that. That was my, that was my theory at least. And then so they're at around 40. That's their 40 billion that are at around 40 billion. So yeah, they could be doing like 2 or 2 to 4 billion in revenue potentially. Yeah. If it's cheap on a dollar per revenue. I think this is, this is very. Much a, on a price to sales. Ratio, a wink wink moment. Because seriously, you know. No, no, I don't think any CEO would throw this crazy of a party if you weren't, if you weren't really, really printing. So expect an announcement soon. I would, I would expect an announcement on, on the revenue side over figure anyway. That's wild. Astro Compute Pranav. Are we getting him on the show? Have we tacked? Yeah, I think he's coming on tomorrow. Okay, fantastic. So we had. Pranavish has done the math. He sat down and did the math on the data centers in space. He said everyone and their mother has something to say about Space Compute. But no one has comprehensively broken down the physics, energy, cooling, economics and the real work involved. So I built a first principles model to show you guys myself. Tyler, what was his conclusion? Yeah, so I mean are we code? You can go through it, you can change all the different parameters and stuff and then you can run the simulation then and kind of see where it lands. I think the conclusion was that there are a lot of scenarios where it does make sense today. Right now if we use current technology, probably not. But if you kind of use the trend line where things are going, I think it's pretty reasonable to say that there's a scenario where it does work. Okay. Yeah. What are you laughing at? Fiberglass is clearly an og. He says that was some OG dry humor, rage baiting like from day one. Tvpn we certainly. A year ago, a year ago that was certainly like 50% of the show's content. For sure. For sure. Viber it's great. It's great to have you here. Thank you for being a day one. Andrew McAliff who's building spacecraft over at Varda Space. He's gonna put data centers on a boat potentially. He has a boat project that's very fun. He says data centers in space. It might not be economically rational, but it might be physically possible. Oh, is he breaking rank from his. From the co founder of his company Varda Dalian. He says I'm trying to bring some quantitative structure to a conversation that's been mostly big number vibes. So we have sort of dueling math equations at this point. What is he. This is vibe coded from public from. So maybe we need to have a debate. Maybe we need to have them both on. But he says tl Dr. The analysis is actually far more favorable than I thought. It's a close thing. I desperately want a Kardashev level civilization but we've got a lot of work ahead of us. Interesting. So he. So I mean deleon's been saying like it's impossible. It is not going to happen anytime soon. It's not going to. It's not going to be a thing. But Andrew McCallop says there just might be a chance. Who else is talking about this post from Goth. Goth says my dealer. I got some straight gas. This train called Space data centers. You are going to effing fry. Lol me? Yeah, whatever. 20 minutes later dude, wtf? We just need to radiate the heat and then we can totally bypass terrestrial regulations, my friend. Pacing. I should buy magnet stocks to get ahead of all the rail guns. What is there's smoke in space data centers. John. I don't get it. This is all over the place. Theo was happy about the EPIC thing. Props to Epic for seeing it through to the end. This is an incredible win for developers. Quoting. EZ in the chat says data centers in a volcano that. That might. Nobody's thinking of that. Nobody's talking about it. That does seem like free energy. Isn't that just like geothermal? I feel like that's like. I know, but that doesn't spack. That doesn't spack. Yeah, that's terrible vibes. Data center in a volcano spacs immediately for sure. Well, let me tell you about Privy. Privy makes it easy to build on crypto rails, securely spin up white label wallets, sign transactions and integrate on chain infrastructure all through one simple API. Anthropic has ordered $21 billion worth of TPUs to train large Claude. Is that really what they're Calling it text is from yesterday's Broadcom's earnings call. The scale at which we see this happening could be significant. As you are aware, last quarter Q3 we received a $10 billion order to sell the latest TPU Ironwood racks to Anthropic. This was our fourth custom that was we mentioned in this quarter. We received an additional 11 billion order from the same customer for delivery in late 2026. But that does not mean our other two Kaiju fights are the best kind of fights. From October. This is from Andrew Curran. He says Anthropic is in discussions with Alphabets Google about a deal that would provide the artificial intelligence company with additional computing power valued in the high tens of billions of dollars, according to people familiar with the matter. The plan, which has not been finalized involves Google providing cloud computing services to Anthropic according to people familiar with the matter who asked not to be named. Google is a previous investor and cloud provider for Anthropic Large. Claude was just me having fun with words and is not anything official. I also considered fat Claude. I'm saving colossal Claude and gargantuan Claude for 2020 Giga Claude Fast Claude. In other news, Fermi, which is the energy company that went public at a $20 billion valuation, they, they talked about getting, getting their kind of facilities online I think in the2030s and people were a little bit bearish about that. They've. They've traded down almost 75% since their IPO. So it seems like yeah, we've basically seen this kind of rolling correction across every single AI. Pure play. You could even say that. But nobody's done volcano data centers. No one has done those. Where are you going? To where the chat's asking where do you put the heat? Just shoot it up into the air. That's what. Yeah. Volcano data centers is funny because immediately like. But don't you want the chips to be cold? You want the. You need cooling, not heating. Yeah. The heat dissipates with the lava. It would be great to just hard light but the heat's free. Heat's heat's free. Data centers are hot and in this one we're getting the heat for free. Now what happens after a correction? You're corrected. I think we're corrected. I think it's possible that the market is cracked. It's perfectly valued. I saw somebody saying that the BG2 interview with Sam they were claiming like it's very possibly helped us avoid a 2000, 2001 style scenario. Yeah. Things could have gotten a lot Crazier. So maybe that was Brad's play. The whole time he was like, hey, we just need a little reset. He's a hero. He's a hero. I'm gonna look like. I'm gonna look like I might, you know, I'm gonna look like a. Maybe a bad guy for two weeks. But now he looks like a hero. I mean, even. Even in the moment. He didn't look like a bad guy. No, no. He asked the question. If he didn't ask that question. Yeah. He would have been. It would have. It would have been. There would have been, I think, some rightful criticism. It was an important question to ask. Yeah. What I really like young macro. This guy's a great poster. Nick Land proclaims, may coldness be my God. What normies think he means, you know, techno accelerationism, Techno capitalism. I like that the circular AI deals made it in here as well. Just the stock charts. What really means. He means Christmas. He means Coca Cola. He means a nice wintry retreat. Yes, exactly. Play some music. This is. This is what may coldness be my God means. It means cozy up by the fire and get festive. I love it. Did you see this reporting on Jim Carrey in Grinch? No. Jim Carrey offered to return his $20 million Grinch salary and was going to quit the movie amid panic attacks over the makeup. Then a guy who trained the military on enduring torture was hired to help him. Whoa. Richard Marcinka was a gentleman that trained CIA officers and special ops people how to endure torture. Carrie told the Vulture, he gave me a litany of things that I could go, that I could do when I began to spiral like punch myself in the leg as hard as I can. Have a friend that I trust and punch him in the arm. Eat everything in sight. Changing patterns in the room. If there's a TV on when you start to spiral, turn it off and turn the radio on. Smoke cigarettes as much as possible. There are pictures of me as the Grinch sitting in a director's chair with a long cigarette holder. I had to have the holder because the yak hair would catch on fire if I got too close. Later on, I found out that gentleman had trained me to endure. The Grinch also founded SEAL Team. So the only. My only kind of question here is, I feel like all these things, if you're being tortured, they wouldn't exactly be like, let him turn the TV off and on again. Let him turn the radio on. Oh, yeah, that is interesting. So I think. I think these are probably great things to do. If you're not getting tortured and you're just developing anxiety attack. Yeah, I did resonate with this because when we had the Halloween makeup, I. Mean, to be fair, it could be like metaphorically being tortured in a foxhole as a member of SEAL Team 6. Like, if you have, you know, oh, you're staking out someplace and you're in a mud pit and it's a hole and you're not actually being directly tortured by, like an enemy, but you have to deal with a really hard situation. And so you sit there chain smoking. And I don't know why you have a TV in this scenario. Again, I didn't land. I didn't land on that one. But yeah, this just resonated because when we had. When we were three hours into our Halloween episode and I started to realize didn't resonate for me, I was like, I am fully. It felt like very suffocating having, you know, I was fine with it 2cm surrounding everywhere. And yeah, it was, it was a very weird feeling, I think. I mean, he was doing like nine hours a day of makeup. Can we get the, can we get the. Can we get the gigachad filter on. Tyler that makes us look like, you know, wimps compared to him. We were only doing three hours of makeup. Nine hours is really, really heroic. Wow. Well, totally worth it. If you aren't watching the Grinch this Christmas, go watch something that's being restreamed. Restream 1 livestream 30 plus destinations. If you want to multi stream, go. To restream.com Will Brown says ChatGPT is the only consumer app with regular pop ups asking if I want to downgrade my subscription. This was hilarious to 3,000 likes. I mean, it's, it's a testament to the. They're running this as a B test. They're like, we tell people they can pay less and they don't. They enjoy giving us money. Hey, maybe, maybe, maybe the pro plan results in more permanent churn. And so this is actually ltv. A downgrade is actually LTV positive. No way to prove that that's true. But it'd be very funny if that was the case. No, the polish around the product is potentially like the way you win consumer. I really think we're in the era of like, of productization more than the latest model. The product managers are the heroes now. I want to see a product manager getting a four year, $1 billion package. It's not completely over for the AI researchers. The models are important. They do get better. And there's functionality under the hood, that's research driven, that's valuable. But if you believe in the Ilia, we're in the age of research, then what is the AI researcher doing? Like they are doing experiments that you have no idea the timeline. Like you're not just doing engineering. You can't just put one foot in front of the other and get easy wins. It's actually going and doing science and discovering new ideas, new ways to create new capabilities. And so in the age of research, the researchers should be off doing research and the user experience designers are the heroes basically. I mean the product managers are going to be more important as the apps fight for market share. What do you think Tyler? You agree with that? Yeah, that makes sense. It's fun to think about. Like Elon and Ilia are like perfectly counter positioned. Right. Like Elon is like we don't have any researchers, they're all engineers. They're going to make small improvements. And then you see like Grok, like Grok is like it's largely undifferentiated from all the other LLMs. They don't have like a, like anthropic has code. OpenAI has like the actual consumer. Grok is like still kind of undetermined. Yeah, I wouldn't say that they're counter position. They actually agree like they're consistent. I think they are opposites but they both embrace a fundamental reality about the world which is that you should either be doing pure research or pure engineering. And it's sort of hard to to mix the two or just that research won't immediately have, won't translate to faster times on the racetrack, which is what's happening right now in the race. Yeah, but I mean Elon is very AGI pilled. Right. But he still doesn't think about pursuing AGI as like a research thing. Oh, interesting. Okay. Okay. Yeah, maybe they are a little bit more different than I considered. I don't know. I mean how do you interpret the fact that the Arc AGI scores are going through the roof and yet qualitatively people are not shocked by the interaction of talking to 5.2 or the latest Claude 4.5 or. Yeah, I think it's just like, it's just about like the spikiness thing. Yeah. Like they just created a new spike. I mean the spike just got longer. Right. It's like the coding ability of the model I'm sure is much better. Yeah. But the actual like ability to write prose is not that much better because it's just like not what they're pursuing but it is bullish. Like this has been my take for a while where like if they can, even if it's just benchmaxing on or. If you can create a spike in any direction, you can solve any economic problem. Yeah. Then the real problem just create every. Possible spike and eventually it no longer looks like spiky intelligence, it looks like smooth. Yeah. Like the engineering task is to find the RL environment that makes the model better at writing, which is like what they're doing. Like you see all these RL environment startups that basically just make these things that models that labs train on that. Specifically do things suspended cap says OpenAI could probably save 25% on their compute if they just taught the model to give me the answer and then shut the F up a bit wild. This one resonated. Figma Think bigger, build faster. Figma helps design and development teams build great products together. Get started. Figma.com of course Jason Fried has a note here. He says one of the best upgrades to ChatGPT is simply changing the bass style and tone to efficient and giving it a simple get right to the point. Be practical. Above all instruction. No more praise, no more flourishes, just the answers. Everyone appreciates Tyler Jolly Maxing. He really is in the Christmas spirit. It's fantastic. Hey, are you a miner? No, I'm here to sell shovels. Everyone's selling shovels. 20,000 likes just for different shovels. Some people are mining, some people are mining, some people are in the trenches, but it is just a few people that are not selling. Big change from a year ago when everyone was just talking about how wrappers were cooked and people were talking somewhat negatively about wrappers. Yeah. It's interesting. Are wrappers selling shovels during. No, no, they're. They're mining. Mining. It's actually mining in like a very small side mine. Not the main mine but like a smaller side mine, I guess. What do you think? I would think of mining as being like if you are actually like using the models to do like a roll up or something. I think that's more much in the. AI shovels mining analogy. The mining I feel like it's like it's Elon Musk and it's Sam and Demis that are doing the mining. Right. They're doing the core thing and then everyone else is like. Is like I'm drafting off of them. I'm supporting them with shovels. Like there is a gold mine that's being mined and if I'm just like riding the coattails of that then I'm coming along. I think of the gold as being the output that you can get with the models when you use the models. The models are the shovels. The models are the shovels. Yeah. So then, so that is, that is actually funny. Like if you're, if you're doing a roll up of, of some, you know. Yeah, whatever companies and you're using AI to just. Yeah, you're not like it's on AI company but you're using AI a lot. To typically, typically the way, typically the way the whole shovels analogy is levied at startups is oh, you don't have it in you to create a great mobile app so you're going to build a developer tool that helps people ship mobile apps. That was the original like the shovels critique and it's the same thing with AI is the idea of like, like why don't you solve a cool problem with AI? Actually create some sort of like, you know, system that solves a particular problem versus saying like we're a platform for running AI agents on top of it. Like we don't know what the agents will do, but we know that they need, you know, a database or something like that. That critique has never been, it's never hit that hard with me because like they're, they're both valuable. We need all of that. Yeah. That being said though, yc, I think it was the spring YC batch. There was a lot of people selling shovels. A lot of like very specific. I'm creating a full stack solution for end customers or as I'm creating AI agents to help you manage your AI agent infrastructure company. No. And it was like, well, what about. I think we need more actual end state agent companies that are selling to everyday consumer. Everyone talks about selling shovels during a gold rush. No one talks about selling just the wood shaft that goes into the shovel. I don't even sell the shovel. I'm a particular slice of the supply chain in the shovel. In the shovel supply chain. I'm so deep in the shovel supply chain. I sell the screwdriver that makes the, the screws that go, that assemble the shovel. That's my business. I'm seven layers of abstraction away from the goal. I sell the shovel making machine. Yes, the shovel making machine. That's where you want to be wrappers. I wonder if anybody's building a wrapper company to help people wrap presents. I'm assuming that all the different LLMs can do this pretty well. You take a picture of an item. They can, they do that. That's sort of like your Rubik's Cube benchmark, Mark, because it requires shape rotation, which the models have been notoriously bad at. I wonder, can they actually give you instructions? An arbitrary size present the best way to wrap the present. You should. That would be an excellent. Well, speaking of holiday gifts, we did a Snap. We did a Snap demo. We did today. I was wearing the specs. We're proud to have Evan on. And I asked my specs, I said, can you count the presents under the Christmas tree? And nailed it. Yes. Fantastic. These are the important problems. I think it's about time. Yes. Let's bring in Evan from Snapchat. Welcome to the Stream. We're going to walk him in. Let me tell you about Fall Gener.