Weekly Recap: Tesla's Crazy Plan, USA vs. Google, Palantir CEO on Conspiracy Theories

You're watching, you're watching, you're watching TVPA. Cracks are forming in Meta's partnership with Scale AI, says TechCrunch. It's only been since. It's only been since June that Meta invested 14.3 billion in the data labeling vendor software.

Invested?

Yeah, invested. So it's not an acquisition, it's not an equi hire. It's a trade deal. It's basically a trade deal. That's how I think about it. 14.3 billion for less than half the business equity. A big dividend.

49%.

Some of the scale went over.

That was a nice, nice number.

Yeah, they just pull it out of a hat. Random.

It's random.

It's just random. And so several of the top executives left Scale to go run meta superintelligence labs. That's MSL. But, says TechCrunch, the relationship between the two companies is already showing signs of fraying. At least one of the executives being brought over to help run MSL scale AIs former senior vice president of Gen AI product and operations, Ruben Mayer, has departed Meta after just two months with the company. Two people familiar with the matter told TechCrunch. Mayer spent roughly five years with Scalia across two stints in his short time at Meta. According to these sources, Mayer oversaw AI data operations teams, but wasn't part of the company's TBD labs, the core unit with in Meta tasked with building AI.

Superintelligence where top they made MSL and TBD and like, the.

Just to confuse TechCrunch reporters.

Just to confuse.

Yes, yes.

Journeys.

Yeah, yeah, it's entirely the 40 chest that's going on. And so after this, after this article was published, Mayer reached out to TechCrunch and was like, hey, you got the story wrong. My job was to set up the lab with whatever was needed rather than data, and I was part of TBD labs from day one rather than being excluded from the core AI unit. Mayer also clarified that he did not report directly to Wang and was very happy with his Meta experience and was leaving for a personal matter.

So, again, it's hard to read too much into this article, but the bigger.

The bigger question is like, like, TechCrunch has their angle and then the timeline was kind of in turmoil. I feel like the timeline's always been rooting against Alex Wang and there's a few reasons. So I wanted to go through, like, the bear case for just the question of, like, was Meta buying Scale AI the right move? Like, did they overpay? Will we look back on this as a great deal, because when we look back on Instagram, we're like, that's one of the greatest acquisitions of all time. WhatsApp, also great acquisition. Very expensive at the time. Both very expensive at the time. Even Oculus. I was thinking about it and I was like, oculus VR, super expensive, multibillion dollar deal. And the level of headset adoption, I mean, they didn't even have like a consumer product at that time. It was like dev kit still. And obviously retention was super low. People would churn off of them and they still do, even what, a decade later, like.

But it was very important to Zuckerberg.

Yes. And VR is a tech wave that's going to happen at some point. We know that the technology is going to get there where Zuck's not going.

To miss the next platform.

He has his surfboard and when that wave comes, he's ready to surf. Did you get any surfing in this weekend?

I did, I did, I did.

Describe your surfing experience. Can you do like 360s off the back and stuff? Can you do a backflip?

No backflips, but I went surfing with a tvpn, a technology brother, Steve, founder of Clock Tower Capital, and he saw me do a couple errors.

You can do errors.

I didn't land either of them.

But you were telling Speeder like, he has to get barreled. Did you get barreled?

It wasn't barreling this weekend.

It wasn't barreling this weekend.

But it was fun. There was definitely swell. I got quite sunburned a couple days in a row.

It's fantastic. We should one day stream your Stream your surfing endeavors and we should do it on Restream. One livestream, 30 plus destinations, multi stream. Reach your audience wherever they are. People have been asking, will TVPN expand to other mediums? Will you do other things? And it's yes. Surfing live streams. That's next. And you, if you're surfing and you want to live stream, head over to Restream. There's a bunch of questions in the chat we should get to at some point, but let's keep talking about this scale AI stuff. So Zuck has a surfboard with Oculus VR. The question is, what's going on with scale AI? Because it's not necessarily an obvious compounder where like with Instagram and WhatsApp, it's like you have this network effect. It's just going to keep growing, keep growing, keep growing. You can grow the user base. The user base never goes down. The business never gets Smaller, you just run more and more ads and it just prints, prints, prints. Right. Buying a growing social network that has strong product market fit. Easy, easy to justify at any price. Basically.

Scalia, well, Elon wanted to get out at one point.

Anyway, for the most part.

But if you can buy it and then lever it up and then merge it with a foundation model lab.

So Scale AI does not have that obvious winner take all network effect, that.

Very real competition, extremely real compete Surge, which was the bootstrap scale AI that was printing money. Merkor, which was emerging.

Isn't there another one? Garrett at Handshake Handshake label box. There's a ton of these companies, there's.

A bunch of players.

And so the reason is because it is not a monopolistic market by default you might be able to make it one, but it's tough to justify. It's tough to think that, oh yeah, it'll just continue to compound. And then also if you're AGI pelled, you don't believe that basic data labeling tasks are going to be done by humans in the future. If you have a bunch of tasks that are like, oh yeah, like just, you know, what do you go to Scale AI for? Oh, RLHF this. Tell me if this is a good answer. If you're AGI pill, do you think that the next version will be able to do that level of task perfectly?

Yeah. And the expectation if Scale had stayed fully independent would be that they would keep having to bet the company on these new sort of eras.

And that's sort of the story. I mean Skale started as a data labeling company for self driving cars and then eventually that kind of hit takeoff where there was not that much more business for scale to do, I believe because Waymo had gotten all of the data and Cruise has gotten the data and Tesla had gotten like the base level data and then the RLHF boom and the LLM boom happened and scale was able to move over to that. And then all of a sudden they were having their best years ever. And so the business was kind of like up and down, very much like, oh, they have a second act, do they have a third act? Chunky Ye very high, high volatility. And so as the market shifts and more and more expert like, like as it shifts to more of these expert data collection processes, like what we see from Merkor, Scale potentially becomes less and less relevant. It's not like an obvious, just beneficiary of every next wave. You have to keep kind of reinventing the company. And so there's.

And they, at some point they stopped working with OpenAI. Correct?

Yes, yeah, yeah. They stopped working with OpenAI and then post Metadill deal. That's part of what created the opportunity for that. Post metadeal, it seemed like Microsoft and Microsoft and Google both pulled back from working with Scale. And so the core skale business doesn't seem like it's just endlessly compounding. So you really can't underwrite this $14.3 billion investment purely on the basis of Scale's business potential.

They're not trying to make money on the investment.

Yeah, maybe, yeah. It seems like they need a mad scientist for their lab and every other AI lab.

Or they need a deals guy.

Yeah, or a deals guy. Exactly. So OpenAI has Mark Chen, SSI has Ilya, DeepMind has Demis, Anthropic has Dario. Each leader has a different shape and style, but they're all capable of rallying top AI researchers and building teams of missionaries. Alex Wang is unproven here. So if you believe that the best talent magnet wins, it seems like a bad deal. And that's kind of the bear case. Now, the bull case is that, yes, metaboling scale AI is a bit pricey, but ultimately it was a good decision for the company. Here's why. Let's review the landscape of big tech's AI efforts. Google has DeepMind firmly on the frontier. Microsoft has GPT5, also frontier. Amazon's a bit behind, but the core business doesn't seem very threatened by lems. Nvidia benefits from basically every outcome. Right now, Apple acts as a window into AI, probably not too threatened. Meta feels like it could benefit Lee, benefit hugely from getting to the frontier, but it doesn't have an obvious dance partner. So what do you try and do? You go down the list and you try and buy every company or hire every researcher you can. Hence the rumors that Zuck tried to buy SSI, tried to hire Mark Chen, etc. Etc. Right. Because it's super high. I mean, we saw that image of like some Wall street investment bankers, like did like a kind of some of the parts valuation of Google, and just DeepMind was worth like $150 billion. Right. And so if you're thinking like, okay, if I have my lab and it's adding all this value all over the place, like, is that worth $200 billion to my market cap? Like, absolutely. Right. And so you try and do that. So at the top of the list you have something like, you know, assemble a dream team, get Ilya Mark, Chen, Demis, get everyone just put the OG OpenAI team and the DeepMind team together at Meta and like, you win. Right. But that's obviously not on the table. There's a bunch of reasons why you can't make that happen. There's economic reasons, there's interpersonal reasons, there's some ideological reasons. But Alex Wang isn't that far down list. And so yes, he hasn't led a real AI lab that's trained a popular big foundation model, but if you look at his trajectory, all of a sudden it becomes I can be a lot more optimistic about. So he's 28 years old. He's a fantastic communicator. You've seen him on every podcast and he clearly communicates very well about the Avon.

One of the few AI heavy hitters that's been on Theo Vaughan.

Obviously you joke, but he was on Theo Vaughan really early and he tells a very convincing story. And he's actually able to communicate to like to both insiders and outsiders, I think. And he's genuinely been at the center of the AI boom for his entire career. But he wants to go bigger. He's built a great company that easily could have cash flowed hundreds of millions of dollars over time and continued to serve the training data market. But getting further into the action that's happening at the big labs was probably not in the cards. If you stayed at scale and so teaming up. Yeah.

And people, people would push back on that and say that that scale was losing real market share to surge and other players who had a reputation for having higher quality data, losing market share.

But still like so many big contracts that if they just went like, like to like weaker and weaker clients and just like held on and just had high margins like I built to die, basically. Like, I do think like the, like the run out the clock value on that company is definitely like hundreds of millions of dollars every year. It's just such a big market. But that's clearly not what, what Alex wants to do. He's 28. He wants to go bigger. He loves being at the center of AI and wants to work on interesting huge problems.

Now he will have, does or will have more compute than pretty much, pretty much everyone.

Like the latest cluster that Zuck's trying to build is supposed to be just a couple percent over the next biggest cluster. So he will have the biggest. And so I think that when you look at Alex Wang, you see someone who's been through like the Gartner hype cycle of training data. Right. It's like wow, we are teaching cars to drive. This is incredible. Then oh wait, like they actually don't need that much more data. And then like, oh wait, like labs need incredible amounts of RLHF data. And then like, oh wait, so he's been on the up, up and down. He's. So there's, there's a bunch of different takes here, but let me continue. So there's also the rumor that Scale AI isn't fully delivering all the data that MSL to train their next model. But the reporting here is a bit questionable. I don't think that the scale acquisition was ever fully underwritten against the value of the training data business as we discussed. So the, and the AI race is so aggressive that every company is grabbing every possible resource. Not only is Metta using other data providers, they also just signed a $10 billion cloud deal with Google. Yeah, so this idea of like, oh.

Demand is outstripping supply.

Oh, they did one deal with Scale that that means that they shouldn't do a deal with Mercour or they shouldn't do a deal with, with Surge. Like no, they're going to do deals with everyone.

I think. Yeah. All it says is that it was primarily an act like an acqui hire.

Yeah, an acqui hire of the team and a bunch of the people. And mostly it's a bet on Alex Wang. And so I think that the FUD over the departures is overstated right now. It doesn't seem like it's an exodus. They hired a ton of people. There's been rumors that like one person was thinking about leaving but then wound up staying. And then, and then one person left, but they said they were like never really planning to stay. And then another person left, but clearly to start a company. So it doesn't seem like there's some sort of massive exodus. And basically it just comes down to the value of developing an in house AI team that's like DeepMind. It's that that team, if it works and if they build it out, the value of that team is immense, probably in the hundreds of billions of dollars. And so there's inevitably going to be bumps in the road. But at the end of the day, Zuck is just betting on the most successful entrepreneur that Gen Z has produced thus far. And it seems still reasonable that even if he's not entirely a researcher, he's deals guy. You have him and then you have Nat Friedman who's able to, who's worked with Ilya and you have the pieces of the team to put Together the right amount of researchers and engineers to actually go and build out a frontier capability or near frontier.

It's an All Star team.

It's close to an all star team. It's not the All Star team. The all Star team is Ilya and Demis and Mark Chen. But that's not happening. It's just never going to happen. What is the big announcement from today? Are you trying to tell more of a story around Enterprise with this?

You know, we're kind of not, we're, I think we're just, it's more like we're crushing it.

Yeah.

Everyone tells us to be super modest. About 93% growth in the US and 94 rule of 40. They may be redefining the rule to like make sure the other people don't like have to live in shame. I keep seeing these articles like in the Wall Street Journal. It's like, Rule 40 isn't real. Isn't edge is real? Because we're like crushing everyone.

You were forced to be humble for a really long time.

I was forced. Well, people were showering me with humble nuggets all day. It didn't really exactly work. But you know, I do think you have to judge humility by the delta between performance and ego. And I would say somewhat ill, modestly, I'm the most humble I've ever been. And, and, and, and, and now. And I just, I think it's like so what we try to accomplish with the. We've been doing these kind of conferences forever basically because everything we've done at Palantir is like completely, it's antithetical or at least orthogonal to what you would, how you would build a business. You guys look at a lot of businesses. You would never build a software downstream from value creation. It's all basically how do I make the client feel like they're getting laid when they're getting fucked. That's the whole way you build a software business. In our business, we began in the beginning, I used to tell people, you know, this is a, we're a mutually servicing business. Both sides should like be happy. And, and the way we built the business was it basically underlying metric I always thought was, you know, the logic of software should be we charge you something downstream of value creation. That sum is a percentage of the value you create. It's better for both sides because it's, it's, it's, it's significantly less than the value create. It's good for us because there's a multiple on the value. The flaw in the logic was always FDE model would basically mean that you'd get a one multiple. So we were structurally misaligned with everyone in finance, everyone not at the Founders Fund, but basically everybody else because of that. Now what we've proven with ontology, FDA structures, where FDA are actually technical and internal orchestration, which is largely artistic, basically was now we got very lucky because without large language models this would not be hypercharged. So it still didn't exactly make sense. But lo and behold, we have large language models, it hypercharges everything. So downstream value creation is an enormous amount of money. And because of our unit economics now, which some people believe are the best in the world, we actually get fairly valued. And what are we doing actually downstairs is we're saying America's central advantage is the plasticity of how we approach the pragmatism, right? So businesses have to move from businesses where it made sense to have parasitic software products that are like basically helping you. It's like one of these things. It's like you believe you're learning to sell. They're selling you on something that is that you can't get rid of. You then run to Wall street and say our clients all, we have 50,000 clients that all hate us. They're like, great, that's a software business because the hating means they can't rid of it.

But a platform business means that you're creating more value than you capture.

Well, the way we do, the way we sell is like. And this is why it just all is like all these things are hugely contradictory. Our revenue is going up, our sales orders going down. The number of people we plan to have in the is less than now. We are very focused on, you know, everybody's like high volume. The volume makes up for, you know, the fact that revenue decreases per client. We're not focused on that at all. We believe we're going to make more from people in the future than in the past. Sizably more. Because it's like why should we not capture part of the value that we help create? Actually it doesn't have to be the majority. In fact, it's usually the minority of the value you create. We also believe that if from more kind of like kind of architectural implementation, technical perspective, the value is in high fidelity data captured in ontology with FDEs and where there's an enhancing factor with LLMs and that that's going to be very, very hard to replicate. But, but, but again all of this is kind of very non traditional. And so what we're really doing in These conferences is saying the same thing we say on the outside. Don't believe anything we're saying. Talk to other people have done it. We're not, we don't chaperone the people here. So they're like, you can talk about things you like, things you don't like. People are on stage. But learn how to build the business of the future. What does the business of the future look like? Actually the interesting thing is workers become more valuable. Like actually trained workers become more valuable. This is exactly the opposite of what people are saying. But it's true. The person at the top is actually crazy valuable. People with technical expertise are crazy valuable. And everything else is going to be done in foundry ontology and something like an fda. So like the orchestration of the business is completely different.

Where are Fortune 500 companies getting screwed by these AI pilots? We saw this stat. Like 95% of AI trials in the enterprise aren't converting. Like what's. Yeah, what does it look like when somebody sells someone?

Well, I mean that there's a technical reason. LLMs are probabilistic, they're not precise. The, the value of LLM is when it's essentially in an ontology wrapper. Because to, to, to actually create value, you have to be able to take the output, serialize it and deserialize it in the context of the business. So the logic, actions and security of the business and its tribal knowledge and what it's trying to accomplish, LLMs are vertically crucial. But the, but, but the error bound is very, very, very narrow. And the way you actually do LLMs in the real world, not in theory, not, is like, is that you essentially put them in a concatenated chain where each single thing has to be done as a street unit because otherwise the underlying math is 95 times 100 separate chains. It's like totally unreliable. And if you do it any other way, you're getting a steak dinner. And that steak dinner is super tasty. It's not going to work. And even worse than the steak dinner, honestly, is that you're being taught how to do something incorrectly. It's like, it's like, okay, I'm going to learn how to learn from a Wokester. Great, Great. The damage that Wokester's doing mostly on the left, but occasionally on the right. The real damage they're doing is they're teaching you how not to learn. Like, and if you just pick your favorite person, right, left, center, who's just selling complete garbage, it's all conspiracy. The whole thing. Yeah, it's like, it's like. It's like, there's no such thing as building. There's no such thing as agency. You can get away with eps. Well, if you want to. Like, Palantir is lifted. One of the things I'm proudest about in the world is we've lifted people from their mom's garage to their own house. Millions of people. You want to stay in that garage, you listen to those people. And it's the same thing happens in enterprise. They're selling you something where you think you're getting laid and you're getting fucked. And once you're fucked like that, it's very hard to undo it. And, like, yeah, you know, the crazy thing about my life is I'm like this wacky dyslexic. It's actually much harder to be dyslexic, but it's also much harder to get fucked because you don't believe. You don't. But you don't believe in any of this bs. It's like.

So speaking. Speaking of sales, there was the CEO, founder, CEO of a CRM company that was making some comments yesterday. Did you. Did you catch.

I. Look, Palantir, we structurally mind our own business. And I love that everyone minds our business. But I would say that what I. We constantly have people on tv. It always sounds like, you know, the guy in high school who's like, but I'm so nice, why don't I get laid? It's like. It's literally like, it's the same thing. I'm so nice. I'm so nice. I create all the value, and I'm so nice, I'm begging to get laid. And no one. It's like, I have such a big this, I have such a big that. And we're like, yeah, we're not trying, dude. We're here, you know, and, yeah, I.

Don'T think about you at all.

And, well, it's like, we are very focused on value creation, and we ask to be modestly compensated for that value. And, you know, if you disagree, like, you don't like us as a client, or you love us as a client, but you think it's, like, great, we're doing our thing, you know, in Palantir right now. In the US Is the market cap that counts. We don't have the people. We don't have the time. We. Orchestrating completely perfectly at Palantir, which of course we don't do, is we're like an artist colony, right? We don't have a time to, like, Actually focus on like what we need to like extending certain components of ontology. We have to do extending maven for the sake of the West. Building things in classified environments, extending things with high value things like yeah, we're focused on that and we don't have the time. Like when you're growing 93% off of a very serious base with a de facto de minimis. Yeah, yeah, it's the 93 and that's not even our best number. It's 94 rule of 40. It's like, and then people, that people are like, oh yeah, yeah, well but we have all the skills, we have all the motion. But, but like somehow our ocean isn't working. It's so big, but it's not, it's like yeah, great, you have problems to. You have time to focus on us. We got things to focus on here that are crucial.

You guys are, it feels like you're reacting to the changing world and actual like customer needs, whereas other players are reacting.

Let me give you, let me give you a more kind of slightly philosophical economic thing. What the large language model does it models do in combination with ontology and FTEs and, and knowing what you're doing is it creates period of optimality over time. We're not there exactly, but every single tech company in the world is going to be paid based on value creation. Maybe that's not completely true today, it will be true tomorrow. So when any company is saying something, you really have to ask. Given that the aspiration of LLMs are transparency and competence broadly defined, they've actually the big cultural shift on enterprises. People running enterprises believe that this thing should work. I should know the cost of the components in my business to the second. I should know how to rebuild things. If there's a macroeconomic I should be able to put the bomb on your head and not on his head. Okay, so that basically means every conversation in the future is going to be you create X value, I'm going to pay you. Why? And the central problem a lot of the larger kind of less agile sclerotic companies have is it's like they can't. It's very hard to move from I get paid because you can't get rid of me to I get paid because you could get rid of me, but you don't want to because you're creating so much value. But that's where the future is going. And like people talk about like, you know, how are we going to you know, get do 10x in revenue, blah blah blah, with the same or Less people. It's like, yes, but the whole market's going to have to move to value creation and we're in the business of that and try to do it, you know, it's not.

Yeah. Do you think long term that the gross margins of software companies will change materially because of like LLM inference costs, like token factory costs, that type of thing?

Well, you mean like enterprise software companies.

Or if I, if I look at like the Fortune 500 right now, there's like a set number of gross margins that's out there. Should we expect like gross margin compression based on well, I bills basically?

Well, first of all I think, let me just give you the trends. I think first of all skilled workers are going to become more valuable.

Sure.

You're going to be paying them more, they're going to be happier. It's exact downstream. Politically it's very hard to argue for anything but high end immigration. So like why do you need more people? Like we got to make the people we have here work. So like politically it's like, like you know, I'm an unhappy Democrat but running around saying oh crime isn't an issue when everyone knows crime is an issue is like, it's like suicidal BS and no one believes it. And now that wokeism is luckily mostly, at least in that way, you know, not as punishing. We can all just admit the obvious. So like transparency is going to be like, so the people are like, workers are going to become more expensive, the overhead is going to become less truly basically artist shaped people are going to be incredibly valuable and they're going to demand to be very highly paid. So but the aggregate cost structure will come down. But more importantly the products you build are going to be much closer to what the market wants in real time. And then again just an obvious thing, this is happening like we have 10x growth in America compared to Europe. Same people, same product, same everything. So it's like. And then the other thing, the point that's a little less obvious that I think people ignore is time is not time. We always assume a minute of time is a minute of time. It's not like it's like from the time you want to do something to the time it happens, if that's 10% of the time, you just graduate, you just got a 10x. So it's like, you know, it's like Palantir is not these kind of atrophied companies, they really, they every, it takes them three years, five years to get a year, it takes us a week to get A year. So it's like, you know, it's like that. That's actually what, what explains the numbers in a weird way is. Yes, but what if five years represents 40 years? What if I'm saying in the next five years. Not. We're actually. It's like the whole problem with the DCF model actually that experts love is A, they don't understand products said, then B, they kind of extend the DCF if they like you. So it's like, oh, I like the person. The DCF is super.

Extra decades.

They give them an extra decade of steak dinners. But the real problem that they somehow don't understand in the DCF mount is a year is not a year for Palantir. Like a year is like, we don't do holidays. I'm working all the time. I'm working. Honestly, I, I sometimes hate the enemies of Palantir, but God, do they get me to go back to orchestration Because I'm like, I'm gonna these people and like, like, you know, and the basic way I'm gonna do it is, you know, going back to like dyslexic, you know, like organization, orchestration of we're gonna have the best products, the best people. I'm going to recruit those people. I'm going to make sure they're the most valuable and I'm going to put them in enterprises that value us. And if you don't value us, go, go work. Go, go work with the people that hate us. Try them out.

Yeah. Do you have advice for young people? I mean, you said like artists, like people. Not literally artists.

You said, you said the company is like an artist colony.

Yeah. Can you just become an artist?

Well, people underestimate, like their artistry because, like, from a young age you get huge benefits for conforming. And you can say, well, I don't. I mean, the central advantage of being dyslexic, we can't conform. So that was. That ends up being a huge. Because you just can't. So you're gonna have to. So your basic thing you've emerged. Do not conform. And by the way, the people who are telling you simplistic bullshit, that means, you know, like meritocracy isn't going to matter. You're not going to judge all these conspiracies. It's. You can't do wealth accumulation if you're in this country, country, like in America, that I think actually a lot of these things are true in other countries. But in this country, they're teaching you how not to learn how to be complacent, how to give up your agency, how to fail and how to blame it on anyone else. And if you're so you have to say, it's like, reject that. Yeah, reject that. This kind of. And then you have to really, really look at people and judge them by their fruits. The best way to learn is to look at somebody and say, okay, well, you know, it's like, you know, you work with somebody like the co founding team at Palantir. So you have Peter, Joe, Stefan, Nathan. Like, part of made us so good is it's like, okay, you can measure yourself. It's like, you know, when I started at Palantir, I actually, just because I just wanted to be left alone, I was like, yeah, I'm gonna make some money. I'm gonna move to Berlin. I'm gonna live a debaucherous life. That was my goal. Like, I'm moving to Berlin. I. I thought I needed 250k. I was like, at 250k is a minimum, a million dollars in maximum. I'm moving to Berlin. I'm gonna like, debauchery forever.

Bergen.

Yeah, yeah. Like, yeah. So it's.

And then set up a remote office.

But like, you then measure yourself and it's like, okay, well I'm highly differentiated on measure, on managing complicated people who have to believe their opinion is their opinion, but still have to build a product that actually delivers value. That's my differentiation. And so like you surround yourself and then remember, you have to remember the persuasion, being persuasive and being right are not correlated. So you have to really look at people who are historically right, rebuttably give them the rebuttable presumption that they are right, and work back to discover if they're right or wrong. Not just. And like, and all these things. And like, for example, on the Palantir thing is a great lesson. Go listen to our critics, whatever critic you love. We're a conspiracy theory. So like you could take the left wing version, which is like Palantir is stripping you of your civil liberties with. Some people on the right believe Palantir is a Jewish conspiracy run by a. A much somehow, okay, whatever. You know, it's like, okay, well go. Actually, how does the product work? Does the product protect data? How does it protect it? Is it better than any other company in the world of doing this? How do you build a company? Do you think it's just like an allocation based on a conspiracy?

Why did we just pick your conspiracy? And that's the Strategy.

And then. And then. But then unpack it and learn for yourself. Like, did this work? How did this work? How did they do it? Assume that at every single decision, if it was a decision anyone else would have made, you would not have worked, because that's a commodity. Commodities aren't valuable. And then apply that to your life. What part of this do you understand? Like, you know, what part do you not understand? What part do you understand better than them? What part could you do better than them? And the weird thing about LLM Ontology Foundry is this actually will work for anyone watching this podcast.

Yeah.

If you're watching this podcast and you enjoy this, you've already passed the test. I don't care whether you're a welder, a plumber, a carpenter, an astrophysicist, or a somebody who'd like to build a business or just want to get rich or you want to get enough money and move somewhere and do what I want. Germany, it's not the right place anymore. But any case.

But.

But you've already passed that test. Now go out and pass the test for life.

Yeah, you said Germany's not the right place anymore. Like, what is your current mental model for the state of the world order? Like, is. Is. Is America in decline? Do we need to bring things back? Like, who are the power players?

America is power pair number one right now. And like, all this media bs, it's like, you know, you got to compare America to. And you can't compare America to some thing you're pretending in your head could be America. Compare it to Europe.

Yeah, compare.

I don't know what you want to compare it to China. Like, you want to have no rights, you know, I mean, again, I'm actually not anti Chinese culture, but ccp, you know, it's like, compare it to Europe. Like, no tech industry. Yeah, everyone rich was born rich, basically, or with almost no exceptions, the most important Germanic company. I hope someone from Germany is listening to this compt Alpalo Alto is Peter Thiel. Undish. This like, the only German company since SAP that's real. Like, and they won't listen to us. Like, just think about that. You have Peter Thiel, like the most important venture person maybe that's ever lived. Co founder of Palantir. And you have me was like, somewhat, you know, basic, partially dramatics. Did my PhD in German. And you have no tech industry. Wouldn't you have us on speed dial?

Yeah, yeah.

I mean, like on speed dial. Like, you don't have to listen to what we're saying. You don't have to agree with what we're saying. Who are you talking to?

Conversation.

Who are you talking to? You're talking to your like, I don't know, expert that came here and studied us.

Trust the experts.

Trust the experts. It's like so it's. Yeah, it's like energy. Like we're.

They will. Do you think that there's, there's optimism around the idea?

You still pick up the phone call, right?

Oh, no, no, I just, I mean I pick up. It's crazy who calls me? It's like, it's honestly like I, I can't talk out of school. Calls me. You'd be surprised number People come in. I begin every call with don't listen to me. Very few people have. I'm going to give you the freak show answer. You probably want to ignore it. This is what I think. And they're like, huh? Okay. Yeah, yeah, okay. Some callback, some don't. But yeah, of course I would. I mean I have a lot of. I mean like, honestly, we have a huge retail. Crazy thing about Germany is a huge retail investor. Sure they don't admit it in public.

But like keep going, keep going.

But yeah, no, I'm just saying the point, I'm saying is, you know, it's like, oh, so then it's like energy, technical talent, understanding how to manage the technical talent. That's an art. Like we have the right venture people, the right entrepreneurs, the right spirit. We have generations of people who are entrepreneurial here.

It's like tall poppy syndrome. Yeah. Yeah.

Well, it's funny you mentioned that. That's like, like, yeah, like you, we were very. Well, this is the thing, we have to fight for this.

Yeah.

Because that. No tall pop. What that basically means. And in every. People may not realize this, but in any, every other culture I know of and like, and I lived abroad in Germany, Europe, incredible cultures. But if you, you, your head sticks above the line, it gets cut off. There's one culture where that doesn't happens here. The only thing is we have to fight for that because the thing that unifies the woke left and the woke right is they don't like the consequences of meritocracy. They want to work back to the inputs. So. And that, that like just will screw society. It's like you've got to be able to allow people to succeed wherever they go. Now I was kind of still progressive. Even though it believes it. I super would like the inputs to be fair.

Yeah.

But the outputs, those are the outputs. My friends of Freedom.

Okay, last. We got to get you out of here. I walked by your office, there were some kettlebells. What. What are the kettlebells for?

Oh, okay, well, this is slightly longer. I'll give you a short version. So to be a cross country skier, you've got to train year round.

Yeah.

So you need substantial VO2 max. And actually you need to be strong per unit of weight. So as an example, I do three days a week of kind of above and below lactate threshold running, but mostly pretty far. And then once a week kind of at. And then I do two days of strength, one day of like endurance strength. And currently the thing I'm actually really proud of is I. I just started doing hanging from a bar, so dead hang like four months ago, and I. I hit 4 minutes and 36 seconds. 4 minutes and 36 seconds.

What's the goal for the end of the year?

What do we do?

Well, actually my goal for the.

Yeah.

This isn't just money. This is. No, I mean my goal for the year was for actually the next next 12 months was. Was four minutes.

Okay.

But then there's the numbers.

We got to get those numbers up.

Yeah, yeah. Well, no, but the number two, the second best mountain climber in Norway. I don't know if we know his name, but he. I have a picture. He did 4 minutes and 22 seconds. What can I do?

Thank you for having us. Appreciate your work.

We'll talk to you soon. Have a great rest of your day. Congrats. My question is we're going to go through the timeline. Tyler Cosgrove has written out the full United States versus Google case timeline. We're going to dig through. We're also going to go through Ben Thompson's. But my question that I want to answer is what do these pay for default deals look like going forward? Specifically in artificial intelligence, there is going to be a particularly odd dynamic around how LLM queries route on the iPhone by default. Like the last possible moment to start taking AI seriously and build like a serious AI foundation model lab. The last moment that you could credibly do that I feel like was earlier this year. Like that was when the last train left the station. And who was driving that train? Mark Zuckerberg. And what did he do? He went and tried to buy everything that was buyable and hire everyone who.

Is hirable that weren't viable.

Yeah, for sure. And what he wound up with was not exactly a dream team. He didn't get Ilya, he didn't get Dario, he didn't get Demis, but he got a really solid crew. Right.

And some great business guys.

Some great business guys, some great researchers. He also just filled out the researcher tiers. Like he's got a ton of researchers.

So I feel like wasn't worried about salary caps.

No, no, no, not at all. And so, and so we were debating yesterday whether Alex Wang was a good pickup for Meta, and my conclusion is like he is the best possible option and I think it'll pencil out. I think it'll be a good deal. It is a big acquisition and it'll be interesting to see how we look on that in a decade. But I think the opportunity is so big it makes sense. At the same time, like, what would Apple even do if they wanted to compete in AI? Like, everyone is taking off all the, all the pieces are off the board at this point. They don't have the DNA for mega acquisitions. So they're not going to go and try and buy anthropic. Like that's just not how they work. They don't like writing hundred million dollar checks for Tal. Tim Cook only makes $74.6 million a year. They're not going to pay some AI researcher $100 million. It's just not going to happen. And so everyone else sort of has a dance partner at this point. Apple wouldn't just be like trying to turn the cruise ship that is Apple. They're building an entirely new cruise ship. It's just not going to happen. So I don't think Apple is going to try and build a serious frontier lab. I think they're going to partner on this. And the question is, where does it leave them? They have something incredibly valuable. Do you know how many active iPhone users there are worldwide right now?

One and a half billion.

One point four billion. Isn't that a ton? That just. I feel like I was, if I had to just guess, I'd be like, job's not finished. Truly. Truly. But, but. So like, that is incredibly valuable because it's not just, it's not just over a billion users.

It's over a billion users that can charge 9.99amonth. Forever. Random.

Yeah, yeah. So many different monetization. But it's also, it's like, it's the top 1 billion really basically of like earners because it's the most expensive phone usually. And so. And they also have that button on the side that right now activates something that should feel like AI but is very clearly far from the frontier. And so the logical outcome feels like a partnership Here. But what will the scale and structure of that partnership be if OpenAI really nails agentic commerce as semi analysis suggests? It wouldn't be crazy for OpenAI to pay Apple billions of dollars to be the default. And so right now we're thinking about if you push the Siri button, you trigger an LLM inference query that's very expensive. Every time I hit ChatGPT, it's a couple cents. If I do some crazy reasoning things, some deep research report might be a dollar, I don't even know it is expensive. Right. If I'm sending off some agent to go and like research every single type.

Of drained an aquifer somewhere.

Yeah, that's what it feels like, right? It feels like every time you hit a query it's expensive, but that's going to flip. And I think each query is actually going to be monetizable and they're all going to be, they're basically all going to be profitable like Google searches. And so if they are profitable, if every time someone hits an LLM, it's not a cost center, but it's actually a profit center, well then OpenAI can pay to get more search volume, some more query volume. And so that means that maybe OpenAI will wind up paying Apple billions of dollars to be the default. And of course they're going to try and build their own device and there's a lot of other dynamics. But would they go with Gemini? Because they're already on Google for the search default, but would they go, would they do that? Or is that too, Are they never going to be able to compete there because Gemini is so deeply integrated with Android and the Android phones are over here kind of doing their own thing. So you can imagine.

And over this year there's been rumors bubbling up of Apple in conversations with Anthropic, with chatgpt, with Gemini.

Right.

And a lot of these companies have been throwing around some very big numbers with Tim and the Apple team and I think they had, at least back then, they seem to have some stickers.

So right now it feels like the foundation labs are going to Apple and saying every time we run a query it costs us one cent, ten cents, a dollar, whatever. You got to pay us to use our amazing intelligence.

Yeah. To bring value to your users.

But I think it might flip and I think that each query might actually be profitable just at the query level because there's going to be commerce that's triggered from those. So in the future you'll pick up your phone, you'll press the button on the side and you can instantly fire off a best in class OpenAI agent to do your bidding. So you'll say order me some creatine and it'll just go do it. And that will be valuable and that will actually drive. That will be a profitable query. And so OpenAI could potentially be paying Apple for the right to do that. So right now OpenAI and Apple do have a partnership in place, but the reporting suggests that no money is directly changing hands. But I don't necessarily expect it to stay that way. So I don't know what's your take? How do you, do you think that this is reasonable, that OpenAI could be paying Apple billions of dollars within the next, I don't know, five years? Or do you think it flips the other way and Apple is the one shelling out billions of dollars for access to frontier AI models from maybe anthropic, maybe, maybe OpenAI, maybe someone else.

Yeah, it's interesting. I mean the people that are paying for ChatGPT today probably have a pretty insane overlap with iPhone customers. Like, it probably looks something like this. I think that Sam recognizes, I mean every consumer tech entrepreneur has like realized how important it is to integrate at the hardware level, right? This is why Zuck is hell bent on winning in VR. Right? He's been sick of being at the app layer and I know that Sam does not want to be in that same. Why would he pay billions for Jony I've and that team. He understands the importance of the hardware layer and certainly will recognize the leverage that Apple is going to have over the entire ecosystem. I think the question is, a lot of this just comes down to, in my view, are consumers going to be paying for AI in the long run? I haven't been totally convinced that the everyday American is going to be spending certainly not $200 a month. I don't even know about $20 a month. It's going to ultimately flip. And so I can see there being a period where OpenAI is willing to pay to be like the default intelligence product within the Apple ecosystem. But again, there it is just going to be a really, it'll be, it'll be an interesting dynamic in partnership, right? Because it's going to, it's so there's so much tension there because Apple is going to want us. Apple is already selling intelligence as a reason to upgrade the iPhone, right? And they've gotten lawsuits over this because they sold Apple intelligence. And then people are like, this isn't very smart at all. What did I just buy? Right? And so are they going to be.

Incentivized to say even 70 IQ is a form of intelligence? It's just a low level of intelligence.

Yes.

They didn't say. They didn't say Apple High IQ intelligence. They didn't say Apple super intelligence. They just said Apple Intelligence. It has some level of intelligence. Apple room temp intelligence.

Room temp high Apple.

What do you think, Tyler? Do you think that OpenAI will be paying Apple or do you think Apple will be paying other foundation labs? Or where will the flow of money go in the next few years?

Yeah, I don't really see it flipping very soon to where, like, prompts or tokens are, like, actually super profitable. Because you need, like, both massive. You need, like, way better capabilities and it needs to be way more efficient because, like, if you just have more capabilities still, like, right now, most people are not paying for ChatGPT. They're just using the, you know, free plan, free tier, which is like, if you compare. I mean, yeah, if you compare for, like, look at like, five months ago, if you compare four oh to like, oh, three, yeah, it was like, massive, you know, difference.

And people still weren't paying in cost or in capability capabilities.

Yeah, yeah, and people still weren't paying. So you needed to be, like, way better for people to actually start paying or you need to be way cheaper.

Yeah, I don't know if I agree with that. I feel like 4o was definitely good enough. People were obsessed with it. People were, like, falling in love with it. It. Right. Even Meek Mill, like, I understand as a power user, like, didn't get that much out of it, but.

Yeah, but I think you, like, if you ask norm people what they think of, like, AI, they're like, oh, yeah, this is not going to take my job at all. When reality. If you used like, O3 and they, if they used O3 every day, if they were a power user, they'd be. I think they'd be way more bullish on AI generally.

Yeah.

So I think that might be evidence that people, like, I don't know, are not.

I think, I mean, one crazy dynamic is, is Apple says, I mean, if they were to do a deal with OpenAI, they could just say, like, great, like, we're going to, we're going to sign up every, like, how many, how many paying subs do you have? Great. We're going to start heavily. If Apple just starts heavily pushing ChatGPT, they will by default get their 30% from the app Store. Right. But they could work out. So, like, this could end up being like, even if, even if OpenAI is not directly paying Apple for it, it could end up generating.

I don't think it's enough.

You don't think it's enough?

I don't think it's enough money for the value that it'll bring on. If it was actually integrated into the Siri button at the low level, I can just press a button and say, order me creatine. Order me creatine. And it knows my address because it knows my home address from my contact in there. It has my payment information saved. It does all of that. I don't know if taking 30% of my ChatGPT subscription, if I subscribe and if I subscribed on mobile, but really most people are going to subscribe on desktop. And OpenAI is going to be constantly being like, hey, go over here and use our web view to subscribe this way and stuff. I don't know. It'll be interesting to see how it pencils out. I feel like there's going to be some new deal that's going to happen. Someone's going to pay, someone's going to bid, someone's going to pay out.

Yeah. The Google team says, hey, we're already paying 20, 20 billion. Why don't we double it to be Gemini, be the default?

Can you imagine, I mean if Gemini was the default on an iPhone, that would be bizarre. But I mean it does kind of like, it does kind of match with the search deal. Right. Anyway, let's read through Ben Thompson. Maybe we go through the history first to get us up to speed. Tyler, do you mind taking us through it?

Take us through it.

So this is the United States versus Google which started in 2020.

2020 starts in October 2020. So that's under Trump.

Okay.

So yeah, I think let's just go over like kind of the central claim of the actual case first. So it's basically that Google is violating the antitrust act of 1890.

It's wild to think about. And you know, before Donald Trump was a technology founder.

Yeah.

With True Social there was was a time where he was beefing with big tech.

Yeah, it's cool.

So long ago. Instead of just competing, oh, thanks for getting us up on VHS. We needed this.

Fantastic. Only possible on restream. One livestream 30 plus destinations, multi stream your VHS tapes and reach your audience wherever they are, head over to Restream.

It truly is. Technology is incredible.

Yes. Also. Yeah, Continue.

Okay, so Sherman antitrust Act of 1890. I also put some of the victims, previous victims of this act. So you see, Standard Oil, Standard oil, American Tobacco Company, US Steel, AT&T. This is what broke up Bell Labs. Probably set us back a couple decades.

You know who founded Bell Labs? Pop quiz. Who founded Bell Labs?

Is it like, oh, I don't know, is it Claude Shannon?

No, no. Who found it? Bell Labs.

Oh, Bell Labs. Alexander Graham Bell.

Alexander Graham Bell. Yeah, that's correct. Yep. Alexander Graham Bell started Bell Labs. Let's go.

We don't know how to name. We don't know how to name companies.

I know, I was looking at the name.

Standard oil, American Tobacco, US Steel.

What is the A&AT and T stand for? Is that America? American Technology.

And let me ask. Gemini. Okay, what does at.

Okay. Anyway, continue. Microsoft, of course. It was a headache for Bill Gates for his entire career until he retired.

Yeah. Oh, that's so good. American Telephone and Telegraph. Oh, that's so good.

When you pick a name like that and you actually, you understand the job and you're just like, yeah, Standard Oil. I'm going to Standardize Oil.

I need a name for my telephone and telegraph company. What about American?

American Telephone and Telegraph Company founder. Okay, okay. Anyway, sorry.

Okay, so basically the claim is that they're legally monopolizing the search engine and the advertising markets. Like in search.

Yep.

So they're locking up distribution. They're self reinforcing barriers to entry and they're.

And the evidence is that they paid billions to Apple and other vendors.

Exactly.

Probably hundreds of billions at this point. Because it's been like 10 billion for decades.

Yeah.

So wow, they're up.

So that is they filed the case DOJ in 2020.

Wait, should we start with the history of Google search in Safari and then run through this? That might be better.

So yeah. OK, so Google search and Safari. So 2002.

2002. This is pre iPhone.

Yeah, this is pre iPhone. This is just on Safari on the, I guess the Mac or whatever it was. So Google is the default there. It's just like they just, Apple just picked that because it's just the best one. They just put it in. Yeah, there's no dealer.

So Google's been the default search engine on Apple devices since 2002. And we don't know specifics. Most recently is $20 billion annually. But, but it's been an over 20 year partnership.

I mean that's a crazy, it's crazy to go. I mean Google was founded late September 1998 and then by 2002 you're the default on Apple like the previous era. Biggest Computing company. I mean, I guess Apple wasn't like, what it is today back in 2002, kind of pre. IPhone. It was certainly not like the biggest company in the world, but. But still, like, pre.

Tyler too.

Yeah.

This all pre.

I mean, the Reborn 2005.

2005.

Wow. Hey, if you want to feel old. Born in 2005.

Okay. So anyway, Sergey Brand. What does he do?

So 2005, Sergey Brand suggests some kind of revenue share. They always call it a resident of.

The goodness of his heart. What is he thinking?

That's crazy. I have to assume it's like, there's some competition. He's seeing it on the horizon.

Okay. Yeah.

But, yeah, I like that he. They always call it a revenue share. They don't. They're not paying the default. They're just helping a brother out, you know? And then 2007. Well, yeah, 2007, 2009, these are like the real negotiations. They're like. It's like Sergey Brent says, like, oh, yeah, we do revenue share. But then when it comes down to it, he's like, well, you know, let's slow down. Let's push the brakes a little bit. The numbers are getting a little high. And then basically, the actual, like, numbers are not public. They really only show up as a result of this case. Like, them being.

Oh. After the fact. So we learned this through the case. We didn't know for years. Wow.

Yeah. So it's not really till 2021.

There's like a full decade where people were like, yeah, like, Google's probably paying Apple something. Yeah.

But hiding a $30 million, $20 billion line item. And you're.

You can. Can do it. If I believe the disclosure rules and SEC filings are something like, if it's a division of the company that reports the CEO, then you have to break out the financials. This is like YouTube. This is the story of AWS when we first got the AWS financials. So if it's just something like some deal and it's just one deal, it's not even a whole division company.

It's a deal I did with Steve Eve.

Yeah, it's just like. Just like other. Other income or something like that. I don't know. So.

Yes. So 2021, it's. It's kind of revealed or estimated at least that payments are around 20 billion a year. And then 2023, this is also a result of the case.

Sure.

Apple is, like, confirmed to be taking 36% of the revenue from Safari that Google makes.

Okay. No. So it's similar to their app store revenue a little bit higher. 36%.

Yeah.

And that's probably how they got to the number. That's pretty crazy that Google's making $6 billion off of.

I never knew they got where they were getting 30, 36% for the big man. For the big man. That's crazy.

Yeah. And then 2025, we know you paid.

To be here, but I'm going to need a cut too.

Yeah. Wow.

Yeah.

Okay, so let's go through the full. The case. So September 2023, this is really the first, like, actual, like, trial. It's a bench trial. I don't really know what that is, but August 2024 is. Is the actual, like, initial judgment. So this is Google is in fact violating antitrust.

Yes.

Like, something needs to be done. We don't know what is to be.

Done, but we're going to do. But then we will eventually decide what the punishment is. Yeah, but they are guilty.

They are guilty.

They were found guilty.

There's no, you know, results. Like, what are they gonna do?

But yes.

Yeah, they're guilty. And then it's not till April of 2025. So this is five years after the, you know, it was actually filed. There's the remedies trial. I think that's what it's.

Sure. Yeah.

So this is like, what is actually to be done.

Sure.

Now, like, we know that they're a monopoly. What do we do about it? And then it wasn't until, I guess yesterday that we found out.

Got it.

So I think we already went through some of the verdict.

Yeah.

But it's. The Judge denied the DOJ's request for forced divestiture of Chrome and Android. They ordered Google to end exclusive distribution agreements. So that's like. I think we talked about this too with Apple. You can't, like, just. It can't be exclusive to Apple, like payments or whatever.

Yeah. What does that mean? In practice, it can't be exclusive, so they have to pay like, Xiaomi or something. Are we talking about other phone manufacturers? Are we talking about other search engines? Like, I know that, like, it's not like, if I go into my Safari settings, I can't pick Bing. Like, I can still pick other things. It's just the default that they're paying for. Dig into that. Jordy, do you want to look that up or something?

Gemini.

Okay, cool.

Yeah, it's kind of like legalese. And then, then there's mandating Google to share search index and user. Basically sharing data with.

So I don't Understand the exclusive. So Google pays Mozilla. Mozilla basically wouldn't survive without Google. Apparently it's like a very significant amount of their revenue. Which is funny because Google obviously competes with on Chrome. With Chrome and then they do this with device manufacturers like Samsung that just continue to make Chrome and Google search pre installed.

Ok, and then what's the last one? Search index. Sharing of the search index and user interaction data with rivals. So was that Bing?

Yeah.

So I guess this means these are basically direct quotes from the actual file, but I assume it means that they have to just share some of the actual index data from whatever page rank is. They have to put it out publicly.

So we gotta hammer the chief product officer of Microsoft today about Bing and what this means because yeah, I mean there's got to be a way to make money off this. If you're Microsoft, this is the way to do it. Okay. Anyway, anything else from the history of simple.

Okay, so just to be clear, they, they prevent exclusive contracts. So Google is barred from entering or maintaining exclusive contracts that give its search engine Chrome, Google Assistant or Gemini app a monopoly position on a device. Basically they can't go to a smartphone manufacturer and say you just can't host any other.

Sure, sure. In the chat Paraclete says Google's only going to share their search data with like other top three companies. It's still remaining a monopoly. Interesting. And people, people enjoy the vhs. So thank you for tuning in.

Also. So these rules also apply to Gen AI stuff. All this also applies to Gemini if they want to do the same, if they want to lease it to Apple or whatever. All the same rules apply about.

If you've been enjoying the whiteboard segments, we are thinking about getting a smart whiteboard where we can put up a figjam from figma. Of course we're sponsored by figma. Think bigger, build faster. FIGMA help design development team build great products together. You can get started for free. Figma.com let's run through Ben Thompson's take. So he kicks it off with Google's remedy decision. Alphabet Inc. Will be required to share online search data with rivals while avoiding harsher penalties including the for sale. We know this Google statement was short and sweet. Earlier today, U.S. court overseeing the DOJ's lawsuit over how we distribute search issued a decision on next steps. He Google closes says as always, we're continuing to focus on what matters. Building innovative products that people choose to love. Choose and love. That's a statement of a company that lost some battles but won the war. In my estimation, says Ben Thompson in Stratechary, which you should go subscribe to. There are some cursory objections to Judge Mehta's decision, but by and large, that statement exudes relief, and rightfully so. The company that is truly breathing. The company is truly breathing easier today, however, is Apple. Apple's actually breathing easier because they're getting that sweet, sweet job, sweet 20, 30%.

Off the top, 99% margin. 20 billion a year.

The first set of remedies were the ones that Google proposed and has already implemented, namely ending the exclusive agreements that were the foundation of Judge Mehta's original finding of liability. So let's see. I want to know what these actual remedies are. Here we go again. This is literally illegal behavior.

We should just highlight again. So Google will be barred from entering or maintaining any exclusive contract relating to the distribution of Search, Chrome Assistant and Gemini. Google shall not enter or maintain any agreement that conditions the licensing of the Play Store or any other Google application on the distribution, preloading or placement of Google Search Chrome.

This is the bundling thing. So a lot of the antitrust cases hinge on bundling. You say, oh, you can't have the Play Store on your Android phone unless you're defaulting to search, or you have to have Chrome pre installed if you want access to search. All of that stuff is seen as anti convenience.

So Ben says again, this is literally illegal behavior. So ending it with was the bare minimum. Antitrust precedent, however, dictates that Judge Mehta go further again from the opinion. This is from Meta.

Oh, this is good too. So they cannot enter or maintain an agreement that prohibits any partner, either Apple or Samsung, from simultaneously distributing any other search engine, browser or Genai product. So you so Google can't go to someone and say, and say, hey you, in order to have Search as your default or Google Search or whatever deal we're doing, or whatever we're paying you, you can't have Anthropic pre installed or you can't have Mozilla pre installed.

Like you can't be in the contract.

Yeah, can't be in the contract. And that is illegal behavior. And so ending it was the bare minimum, says Ben Thompson. Antitrust precedent, however, dictates that Judge Mehta go further again from the opinion. The question now is what to do about Google's violations. Precedent requires fashioning antitrust remedies that effectively pry open to competition a market that has been closed by a monopolist's illegal restraints. Denying the fruits of the violation is a valid objective and so too is ensuring that anti competitive behavior will not recurring in the same or related ways. The Court has broad discretion to impose remedies to accomplish those aims. Judge Mehta laid out four fruits of the violation. The Court found that the agreements had four main anti competitive effects. They one Foreclosed a substantial portion of the relevant markets, thus impairing rivals opportunities to compete 2 denied rivals access to user queries or scale needed to effectively compete 3 reduced the incentive to invest or innovate in search and four Enabled Google to increase text ad prices without any meaningful competitive restraint, thereby allowing Google to earn monopoly profits to secure the next iteration of exclusive deals through higher revenue share payments. These effects did not persist independently. Together they enabled Google to widen the moats and pull up drawbridges to ward off competition. Great analogy.

Judge Mehta attempts to address number two necessary scale by forcing Google to share various types of data including Google search index, but not the actual data from the web pages and the index or the output Google's PageRank algorithm competitors, which explicitly include Genai providers, will get a one time snapshot, not an ongoing one, and only need to pay Google's marginal cost for providing the information user click and query data showing what results users clicked on. Competitors will get this data at least twice, but the final number will be determined by the technical committee the court will set up and oversee. Do you fully understand like yeah, that's.

A very, very odd, odd like change and remedy. Like I, I, I guess it like lets people catch up to where things are. It feels like the remedy is like, like there has been this, this unfair advantage for Google for years. We're letting everyone catch up and then we're creating like a fair race but we're restarting the race and then everyone can go out and do whatever they want. Current thing is big numbers. $200 million for the free press Barry Weiss's media company to CBS people did not like the quiet in the chat. They said Scream $610 million for the browser company. We talked about that a little bit Yesterday. Ramp hit. $1 billion in ARR. Let's hear it.

Great hit, great hit. 600 billion in Meta CAPEX and 1.

Trillion potentially on the table for Elon Musk at Tesla. So we're going to go through this. So the Free Press is one of the most impressive media companies built in the last decade, says Austin Reiff from Morning Brew. Correct?

Yes.

Huge congrats to Barry Weiss. Nelly Snoozy Weiss. This is a funny name but for building something truly Different. This is, I think, still rumored. I don't think it's fully confirmed at this point, but it's all.

But yeah, it was hard to tell which side was leaking the news.

Yeah, there's this dynamic where media likes to talk about media and I feel that 100%, it's super interesting to hear about like the business of Joe Rogan or the business of Huberman or any of the experts. But so obviously also, you know, you're surrounded by journalists. Literally if you run a media company, so they all talk, there's a journalistic force headed towards you standing. I've heard that one before. But you're surrounded by journalists and so obviously everyone talks and the media leaks out and the news leaks out.

And it could be interesting ways for either side to put pressure on getting the finish line. Exactly what they want out of the deal.

Yeah, yeah. And so there's a series of takes. I was talking to a buddy who works in media who was coming at me with like, it's a crazy number. And so that's kind of level one. Level one of the take is like on a price per subscriber basis, seems high. I think they have about a million free subscribers. $200 per free subscriber feels like a lot, I guess. Kind of hard to tell if you think about long term value for cbs. But then others are complaining about the political implications of the deal. Is this something that's like a give to Trump to help get the, the demerger, the spin out approved? Something like that, yeah.

And I mean this is an $8 billion deal, this merger. And media companies are now entirely like media companies at their best are personality led and paying $200 million to bring on, you know, a face for CBS, one of the most important properties. Makes sense.

Yeah. That's my level three final take, which is that you sort of have to put aside the price per subscriber take and you have to put aside the political implications of the take. I just think about it from it's a big company. 8 billion. You said something around there. Does bringing a younger, more entrepreneurial talent into the organization move the market market cap by 1% or 2% over the next few years? Like it seems like the same math that we're doing for buying a very expensive AI researcher. You're paying a lot of money in terms of like, what, what, like a salary would count, but you just can't get that level of talent on any.

Sort of overall property.

Exactly, exactly. And so I was the, the person I was talking to who was like, this is a ridiculous price. I was like, yeah, but, dude, if you were at CNN, you could probably add $200 million of value to that organization pretty quickly. And so it's just a matter of time until, like, the market and the board of directors and the shareholders kind of wake up to that dynamic of how power law, certain people are and actually go and figure out how to get the deals done.

And David Ellison is doing deals, Right. He just did CDO.

100%. 100%.

There's a lot of big numbers flying around.

And there's an. And there's an interesting dynamic where the premium.

Right. We saw this yesterday with the browser.

Oh, yes, yes, yes.

You do get a big, fantastic outcome for having the. At the beginning of the day, maybe.

We should just be the business production network. Drop the technology. Technology's over. The hype cycle's peaked. We're just. The business production.

We're in the trough.

We're in the trough. We have some fun stuff coming in terms of the hype cycle soon. But it's interesting because many companies just cannot justify putting someone on staff for $100 million. W2. I mean, we've seen this as Apple with Tim Cook, you know, the CEO pay once you get into the. These eight, nine figures.

Thank you for bringing that up. No, thank you for. Seriously, thank you for bringing that up.

Oh, yes, thank you.

Apparently, Tim Cook. We'll try to pull. Pull up this video in a bit. Apparently he said thank you 12 times in two minutes when talking with President Trump last night at the dinner.

Yeah.

So he's very grateful.

So the. So, yeah, Meta is like the first company to really, like, break this trend of just like, yeah, we're willing to pay $100 million in salary. We will do the crazy acquisitions and the acqui hires to get really talented people in the org. But I don't think other companies are in that world where they could go out and hire somebody like a BarryWeiss just for $100 million. It's gotta be done through an acquisition. That's what gets board approval. That's what gets shareholder approval. And so you're kind of wrapping an individual, an influencer, someone who's incredibly talented and entrepreneurial around an organization just to actually get the cash flow to flow through.

It's making me think about CBS News for the first time in decades.

Yep.

So.

Yep, yep. Yeah, yeah, it's true. And even if the. Even if, like, the subscriber numbers are, like, a little low, I mean, honestly, a million's A ton. It's a great job. But even if they are a little low, it's like, well, what is it? What are they compounding at? What will the free press be able to be at in a decade? With the support of CBS and the. And the backing, financial backing of CBS and Barry Weiss's execution strategy, like could be 100 million. I don't know, it could be really, really big. But you don't have to raise money for it. You can just focus on growth. Anyway, similar story at the browser company. $610 million from Atlassian. Very interesting deal. The timeline was not a fan of.

The Land down under with a Brinks truck.

Yes. And so a lot of people are asking, is this some run around by the Australian government to try and influence the browsing habits of Brooklyn hipsters? Yeah, I don't think there's much.

Try to control the narrative.

Exactly.

As a way to say, you know what if we could get all of the data from Brooklyn hipsters, all their browsing information.

Yeah.

They're known communication.

Yeah. They're known for drinking pbr. But what if every time they went to PBR on the ARC browser, the DIA browser, it just automatically rerouted them to Fosters.

Yeah. Or the AI assistant on the side popping.

Oh, you're in dia. You might like asking, oh, what should I do this weekend? And just auto. Auto populates. Oh yeah, we're using A.I. throw it. Why don't you throw another shrimp on the barbie.

Yeah. And if you search is it true that kangaroos are extremely violent? It'll. It'll start spinning it and saying, well actually, you know, they're only violent when promote.

Promote provoked. Yeah. They're actually quite peaceful.

Yeah. Your dog actually tried to.

Yeah.

Barked at the kangaroo and then the kangaroo went after it.

Exactly, exactly.

I think this, I think this really smart from Australia.

From people pulling the strings in Australia. Yeah. To get a little bit more influence over the Brooklyn hipster and the ARK and the DIA user. Zeb did not like the acquisition, said, pull this up me uninstalling ARK as a certified Atlassian hater. There is a wild gap between the vibes and so some people had this take that like it was a vibe acquisition. They acquired good, better vibes with the next generation of founders ark.

I think this is funny, but I think that from what we've seen dia, they're going to continue to operate independently. I bet they're going to build some beautiful enterprise browsing experiences like they always have.

Yeah. And they should build in restream one livestream, 30 plus destinations, multi stream and reach your audience wherever they are. So you should be able to stream from your DO browser, from your ARC browser. So my, my take on this was there's. So basically right now the team is saying we're going to stay independent. We're going to go. But my question is are they going to stay. Are they going to stick with that narrative of like stay independent. And this is very rare for pre product market fit acquisitions. I haven't seen a ton of examples of that where a company was acquired that wasn't like clearly on a. You know, just remember, remember path.

Instagram was bought for a billion dollars.

Yes.

And almost everyone involved at the time thought that it was crazy.

They did. Yeah. That is true. At the same time.

And it was like a key threat to meta's business. Right. Like the user growth was insane.

I mean I would love to look at the, the retention metrics for DIA versus Instagram because it's totally possible. I don't know. Do you know how many DAUs or total downloads Instagram had at the time of acquisition? I know it was small, but was it sub. A million? I feel like it might have been a little bit bigger. Can someone look that up? But basically look that up.

Tyler. I'm also curious about YouTube as well. YouTube was bought for.

Yeah.

How many people were using YouTube three times. And again the dollar isn't what it used to be.

Yep.

So when Instagram was acquired, estimated around 860,000. So less than a million DAUs.

Okay, okay. So yeah, yeah. But still, I mean we're in the same ballpark here.

Very interesting Instagram people that have.

Yeah, yeah, yeah, yeah.

Daus is wildly different than total download users.

Okay, okay. Yeah, yeah, but still, I mean we're, we're in, we're in the rough ballpark where you could. Where at the very least you like. We have to take what Atlassian is saying and what the browser company is saying at face value. They are saying we're going to continue building this browser. We're in it to win the AI browser. We're going broad, we're going consumer. This is not. We're acquiring this browser. We're going to bake it into Trello and we're going to bake it into Jira.

No, I actually disagree with you here. They don't. All the messaging was that they're not going to be focused on consumer like from Atlassian CEO was that we're going to build great enterprise browsing experience.

Oh, okay, okay.

So there's nothing. There's nothing in there that I saw from the Atlassian side that said we want to win in the consumer.

Oh, really? Really. Okay. Interesting. Yeah.

They know their business, right?

Okay. Yeah, so it's very interesting. I can't imagine having, like a work browser, but maybe that makes sense. I mean, people have work laptops, but I feel like the trend has been you have like, one phone that you do work and life on, and then you have separate apps and maybe you have, you know, an enterprise software product installed on your phone. But really, like, the trend seems to be more people using, like, Gmail and Google email at work. And it seems like there's less and less like, oh, yeah, like, I have a separate work computer with a separate operating system.

Chrome is very set up. Chrome is already very set up to have, like, profiles, like personal profiles and work.

Yeah, I have too. I have personal and VPN one, so I don't know. Yeah, we'll have to see. I mean, the founder is saying, like, we're going to keep working on dia, but maybe that means in the enterprise context anyway, we'll have to keep an eye on, like, where they wind up going.

So Mike said, again, I said this yesterday with DIA browser. We're going to collectively redesign the browser to help knowledge workers kick. But in the AI era. And so that to me, kind of reads a little prosumer, right? I don't think they're. But. But it feels like, you know, very much oriented towards people that are working in the browser.

Oh, yeah. OTB says get Sagar on here to give pushback on TFP buy. He's going postal about it. I. I should text him and see if he wants to hop on. We will. Yeah, we'll definitely get him off when he's ready to go live and rip some takes. I also invited Bari Weiss. I'd love to hear her side of what's going on and what her plan is. It's a fascinating story. Of course, these are still leaks, so most people don't want to talk until there's finalized information anyway. If you're trying to design a browser, you got to do it in Figma. Think bigger, build faster. Figma helps design and development teams build great products together. Get started for free. And the third big number of the day, ramp has crossed 1 billion in revenue. Sheesh, Says Tyler Hodge. A key question is whether it's gross revenue. It's now revenue. I think it's gross revenue. He goes back and forth on this. But anyway, they've been on a Tear, as David Senra likes to say. It's a great example of taking a simple idea and doing it deadly seriously. Taking a simple idea deadly seriously. I think David was quoting someone, but I attribute everything to David's.

Now we should pull up these videos from the dinner at the White House last night.

I only have one question for Eric, which is, is the job finished? So we will figure that out today.

You're gonna call him?

He's busy. I'm gonna, I'm gonna.

All right. Play this video.

Oh, yeah, please. You've done an incredible job with Apple.

Little company called apple.

Thank you, Mr. President.

Very, very few people have been able to do what you've done. Congratulations.

Please.

Thank you, sir.

That means a lot to me. I want to thank you for including me this evening. It's incredible to be among everyone here, particularly you and the First Lady. I've always enjoyed having dinner and interacting. I want to thank you for setting the tone such that we could make a major investment in the United States and have some key manufacturing, advanced manufacturing here. I think that says a lot about your focus and your leadership and your focus on innovation. I also want to thank you for helping American companies around the world. This is a very key, key thing. And I really enjoy working with your administration on those topics as well because I think they're so important to the country. I want to thank the first lady for focusing on education. Wow. There's nothing more important than education. It is the great equalizer and always will be. And so thank you so much for, for including me when we are all different in some ways, but we all believe in the power of technology to improve people's lives. And that, that is the thing that binds us all together. And Tim, how much money will Apple.

Be investing in the United States? Because I know it's a very lot and it's, you know, you were elsewhere and now you're really coming home in a big way. How much money would you be investing?

600 billion.

600 billion.

We're very proud to do it.

That's great.

Thank you very much.

Thank you.

Appreciate you, sir.

We just had 10. Thank you.

I think it was 12. Somebody else had counted as 12. I was around 10.

Crazy.

So we should.

Why wasn't Christina from Vanta there? I feel like that's key to the US's tech strategy.

I agree.

Automate compliance, manage risk, prove trust continuously. Vantage trust management platform takes the manual work out of your security and compliance process.

We got to pull up this video of.

With continuous automation of Zuckerberg this was our fourth big number of the day. Zuckerberg says he will invest around $600 billion by 2028. Zuckerberg says Meta will invest 600 billion in AI infrastructure by 2028. Meta's already guided CapEx of 70 billion in 2025 and 100 billion in 2026. So that's a big ramp to go from 70 to 100 and then they still got 430 billion to do in 27 and 28. To actually reach 600 billion, spending would need to jump to 200 billion in 2027 and 300 billion in 2028. That's a lot of dollars.

Video in the timeline.

Let's play it.

This is quite a group to get together and you know, I think, you.

Know, all of the companies here are.

Building just making huge investments in the.

Country in order to build out data.

Centers and infrastructure to power the next wave of innovation. So it's, you know, we don't often get together as the CEOs of the different companies, but it's good to see everyone.

How much are you spending would you say over the next few years?

Oh gosh, I mean I think it's probably going to be, be something like $6016-006000-00000,000 in the U.S. no, it's a significant.

That's a lot. Thank you, Mark. It's great to have you. Thank you. Well, thanks for, for hosting us.

I mean so does anybody in the chat know if who, which, which of them went first? Yeah, because that number, that number does not align with. That number does not align with. Meta's stated.

Yes. Yeah, it doesn't appear in the SEC filings yet, but it is like very forward looking.

No, it was 2028.

2028, which is something that they might not actually put out or might not be demanded by the by Wall street. But that is hilarious. Anyway, whatever they do, they're going to have to do it on graphite. Graphite.dev Code review for the Age of AI. Graphite helps teams on GitHub ship higher quality software faster. Get started for free.

Yeah, so meta guided capex of 70 billion in 2025 and 100 billion in 2026. So to get to 600 billion, spending would need to jump to 200 billion in 2027 and 300 billion in 2028.

Yeah, what's interesting is that so these numbers are like insanely big and obviously it looks like a big acceleration relative to past CapEx numbers. But I'm sorry, inured to big numbers now because of the fast takeoff crowd that was like, oh yeah, like trillion dollar build out data center. Like next year for sure. Like AI 2027, it's going to be 10 trillion, Stargate, 500 billion. And so when I hear 600 billion, I'm like, yeah, yeah, that sounds like totally reasonable. I think they'll definitely hit that and I don't know, they probably will. It doesn't seem like there's many barriers in the West. Like it's just continue to grow the core business, run more ads, dump the profits into capex, build some more data centers. Semi analysis. Doug o' Laughlin over at Fabricated Knowledge has a good deep dive today on the glut of lagging edge chips that I think we might have a chance to get to in a little bit. But it was a fun dinner. Are there any other videos that we should play from the, from the White House dinner? I like the idea of this is like this is kind of new podcast format. Just have a dinner with everyone, have some cameras there and just. Yeah, just. Yeah, yeah, it's good. What's up, Tyler?

I don't have another video, but there is, there was a photo that came out. I'll send it to the chat. But it was like the full dinner and right at the very edge. I'm pretty sure it's Dylan Field.

It is. Yeah, yeah, yeah, yeah. So someone had the full breakdown here. It's in the deck, much deeper. Let me, let me pull it up. Where is it? The full list of people who attended the dinner is. Wow, there's a lot of stuff. Okay, so Ed Ludlow has it. The full list of attendees at President Trump's dinner with tech leaders, the President, the First Lady, Susie Wiles, Sergey Brin, Geralyn Gilbert Soto, Sam Altman, Greg Brockman, Anna Brockman, Safra Katz at Oracle, Gal Tarrasch, Jason Chang, Meredith o', Rourke, Natalie Dompe, Tony Fabrizio, Dylan Field, John Herring, Jared Isaacman. Jared Isaacman. Oh, I guess he was missing one. Yeah, Elon didn't go, but Chamath. Oh, Chamath is in here. So Sunny Madra Satya, Nadella, Chamath Palihapitiya, Sundar Pichai, Mark Pincus, Vivek Ranadive who owns the Golden State warriors. And I believe Vivek Ranadive is the like the guy who worked with Chamath in the first SPACs. And Vivek Rana dive kind of like pioneered that format almost.

Did you say Lisa Su from amd?

Lisa Su made it. David Sacks, Sham Sankar from Palantir we missed him yesterday. He was at this dinner in D.C. jamie Siminoff, Alex Wang, Sanjay Mehrota, Tim Cook, David Limp, Mark Zuckerberg, and Bill Gates. What a crew. Jensen didn't make it. So the fact that Jensen and Elon are not there, I feel like you shouldn't read too much into it. It's kind of just like Trump catching up with a bunch of people. But I don't know. There's also, like, this. Yeah, I don't know. They had, like, falling out, but it seems like they've kind of patched it up. I don't know. Always, always hard to read too much into this stuff. But speaking of Elon Musk, he didn't need to be at the White House because he's setting up a $1 trillion pay package.

975 billion.

Oh, it's not a trillion, just a little bit shy. Okay, so the update is Tesla's board has greenlit a fresh pay package for CEO Elon Musk. If he hits 400 billion in adjusted EBITDA, he deploys 1 million robo taxis for commercial use, ships out 1 million energy storage units, produces and delivers 20 million vehicles, secures 1 million of ongoing FSD subscriptions. Aim for a $8.5 trillion market cap. The golden retrievers are bargaining 7 1/2 to 10 years in the CEO role. The starting share price is $334.09. The package includes 12% of company stock divided into 12 segments. And so Signal says this is a masterclass in how to design incentives. This deeply ties Elon's upside on the world, radically improving. If he actually hits those milestones, the value unleashed for society dwarfs his payout. Most CEO comp is rent seeking. This is moonshot seeking. I completely agree. Every pay package for executives should be this absurd on both sides for the company and the individual. And this was exactly my take. It's like, bet on.

Yeah. And he's done this. He's done this before.

He's done this before. And Dan Primack, when the pay package, the most recent pay package, came into under legal scrutiny, Dan Primack was like, I like this when it was announced. And I thought it was crazy. And everyone thought it was impossible that he would pull it off, and he did, so he deserves it. And so it was this crazy legal battle, but it was actually aligned with exactly what a CEO should be doing, creating shareholder value. This should be a model for CEO comp going forward. In my opinion. There's another side of this, which is increasing a founder CEO stake by doing a graph stock buybacks. But this is a way cooler because it incentivizes bold investments and huge capex on factories, moonshot technology gambles and generally thinking in decades. So I'm super stoked about it. I think it's good. I hope it holds.

It's a good way to signal. I mean, they're very much signaling to the base Tesla army that there's a lot of potential upside. Chris Camilo said earlier, been tracking xai's accelerating cash burn, flattening user growth, monetization and fundraising struggles since the beginning of this year. And I don't see how they last another 12 months without a Tesla or Elon bailout. I mean, saying, I don't see how they last another 12 months without the founder bailing them out. It's like that's kind of the founder's job, bail out the company. But I thought this was relevant because I do think that part of this, part of setting up this pay package could set up a scenario where XAI gets just merged into Tesla and it becomes a standalone. It just all gets rolled in.

Yeah, yeah. I mean, Tesla has. I mean, Carpathia worked there. Like, it's. It's a fantastic AI organization and it makes sense that there would be a lot of overlap in. In what they do. I guess there's probably.

And I mean, if te. If X the Everything app was part of a public company again, how does that get valued at less than 8.5 trillion? Right. So with. With Elon having that ace up his sleeve fact that we're all addicted to this app.

Yeah, I mean, I mean, I guess it's good that there's other milestones in there that. That you can't just, like, merge everything together. Because if you merge it is. It is.

It is funny to think about some of the different milestones, though. It's like. Like shipping out 1 million energy storage units feels like wildly like, does this even need to be included if you're going to deploy 1 million robotaxis for commercial use?

Yeah, that one seems a little bit, I don't know, easy sandbag. Who knows?