Zuck’s AI Vision, Celebs Are Dating Private Equity Guys, AI Agent Hacks Gym | Diet TBPN

Big, big news. Mark Zuckerberg dropped a blog post. Also, I mean, the bigger news is that they open sourced, uh, already one model and they're planning to open source Muse Spark 1.2. That's big. But, but the blog post lays out his AI vision and we should go through it. Meta was sort of set up to be and also ran in the AI race, like the LLaMA project stalled out. Yann LeCun exited. Their biggest model behemoth never shipped, which was ironic because In the Bible, Behemoth is this untamable land creature. And of course it was just the model was too big to be tamed, to be released. And it was like nominative determinism in action. Happens a lot in tech, unfortunately. But they also turned over the entire team and Zuck said, I'm not quitting. I'm not out of the game.

I'm still in.

I'm all in. Yeah. And I'm raising debt, equity, doing all sorts of deals, making things happen, making big offers to AI researchers, rebuilding the team, bringing in Alex Wang, Nat Friedman, Daniel Gross. Bunch of heavy hitters and they already build massive data centers. They have a huge surface area to distribute AI consumer experiences. If they come up with new ones, they have billions of users that can put some, you know, hey, try this thing and maybe they'll get a bunch of users from it if it's good. Semi Analysis also just one month ago laid out like a much more detailed bull case for MSL in particular. It was interesting because I was always optimistic about, okay, they have a lot of data from Instagram, they have a lot of images, they'll do a great image model. Like the image editing will be the best in class, they can finally like kill CapCut with their edits app, which is great. And when I think about you're taking an image, you're taking a video, you have a bunch of videos from a trip, you want to edit all that together for some Instagram story or Instagram reel. Reels editing is its own skill, its own talent. It's not available to everyone. A lot of like family friends of mine will use Instagram and they'll just post a picture on stories. They're not doing edited cinematic reels, but with AI, potentially get someone from, oh, you took a bunch of video, some of it's horizontal, some of it's vertical, some of it's sort of cropped wrong, some of it's colored wrong.

You could turn that into some brain rot for you. You could just throw it right here in this trough. Yes. Ah, continue.

Adding those programmatically. No, you actually could. You actually could. And so I was optimistic about that. And then obviously we talked to Eric Suford about just the gem model and how AI can immediately move the needle on ad monetization, targeting, all of that. So there were lots of reasons to be optimistic. Semi Analysis laid out this more detailed bull case. And they said that there was this interesting detail that maybe was getting missed, which was that by recording the screens of every Meta employee, which is very controversial and bold, but actually ripping that bandaid off and just saying, hey, we are, we are an AI training company now and everyone's gonna be recording their screens and their flows and all how all their work gets done. They basically got like a frontier leading AI data labeling company for free. I mean, they took a, you know, PR hit for it, had to answer some questions. But in general, like the volume of tasks, the volume of quality data that's coming from that workforce now is on the order of a scale, Merkur, Handshake, etc., Surge. And so they just get that more or less for free. Obviously Meta's lawyers are great. Meta's research team is great. Meta's finance team is great. So if you're, if you're getting all of that, it's equivalent to hiring a really talented person to do a training task.

Yes, yes, and yes and no. I think a lot of the data that the Surges and Mercors are creating is much more specialized.

Yeah.

And I think there's just a lot of noise coming out of that data. They also have rolled some or maybe most of the program back. I think there was so much pushback.

You can pause it for 15 seconds once a day. No, I think you can pause it for half an hour, an hour if you want to go get a nice scroll in at work. But I mean, that's on brand if you're like, yeah, I was scrolling Reels. I work at Meta.

Testing. I was hunting for bugs. Yeah, your honor.

For sure. For sure. But yeah, I think there are some— there are some employees who are just fully excluded because they work on proprietary things that they don't want leaking into the training data whatsoever. There's a whole bunch of nuance there. There's these funny, like, leaked messages from Boz being like, if you're saving time with AI, don't take more vacation days. Like, go ask your parents.

What was that line? Like, ask your parents if it's a good strategy every time you talk to your boss to ask for more time off.

It's a good point.

So it was a really good—

it's a really good point.

Really good point. But the timing was funny because over the weekend, Zuck had a UFC fighter in Tahoe and they're just like on this, like, you know, on-water octagon.

Okay.

Yeah, you can tell there's like, you know, all, all of Zuck's like security boats around it and they're just like fighting, like, you know, wrestling or whatever.

Let's pull this up. Zuck busy fighting with Mehrab in Lake Tahoe while his execs tell employees not to take vacations. Thanks to AI.

I think I've seen enough after this. I think he's ready for the octagon.

He's looking good. What are those leg braces that you got on? Those are for striking.

They're effectively gloves for your knees.

Oh, okay. Okay. So you don't get your shins bashed up or something?

Yeah, you don't.

Okay, that's cool. Anyway, so following up on the op-ed that Mark Zuckerberg dropped in the paywall Wall Street Journal section, which is a little bit ironic because he's saying it's for everyone. Yeah, I will be abundant and free except for this article. Well, now you can go read a more lengthy version of his thesis. For free, not paywalled, on Meta's corporate blog, effectively. You know, I don't even think you need to log in to any Facebook property to get access to this, but it's a 6,500-word encyclical. The Wall Street Journal called it like it's a papal bull on the AI future. High-level AI. It's for everyone. Turns out it's for everyone. Whoa. That's what he's saying. But it's good. It's an anti-permanent underclass message. Which is good. It's just good from a comms perspective to lay out a more optimistic philosophy here. And it also seems to be borne out by the data in the short term. You're not seeing any, like, crazy displacements in the economy. And so this comes from, like, a place of strength that's, like, backed up. And it's also just a very optimistic, correct version of the piece. He takes a shot at fear-based marketing early on in the piece. He says, still, it's surprising that the discourse from many Developing AI is so filled with doom. I do not understand why anyone who believes that AI will eliminate most jobs and much of humanity's relevance would rush to build that future. The notion that AI is so dangerous that the only safe path is an extreme concentration of power inherent— isn't— seems inherently problematic. Historically, hoping that an absolute power will benevolently provide for humanity if sufficiently enlightened has not led us to safe or positive outcomes. What do you think?

Do you want to read a little bit more before I—

Well, so I mean, my take on that was basically that, so he's framing this as, I don't understand why anyone who believes in AI would try and become like the emperor and the dictator of AI. I understand it. It's funny that he's like sort of playing dumb here because extreme concentration of power is awesome if you're the one that has the power. Like, I don't see You have to imagine that most dictators living in other countries aren't like, oh, I would kill just to be like a normal American citizen. No, it's like, yeah, it sucks for the citizenry.

Also, this is coming from a guy who has— is effectively a monopoly, more like a duopoly.

Yeah.

And has enjoyed the benefits of, you know, extreme concentration of at least users.

Yeah.

And social media activity. Yeah. I came away from this first.

First, there are— there are 3 very concrete proposals that I think are good. And then we can go into some on what's downstream of it. So he— there's some hand-wavy parts about the future and how things will align in the future that are sort of just guesses about how the future will play out, in my opinion. But there's some really solid concrete proposals in here that I thought were awesome. One, he calls out that in Richland Parish, Louisiana, where Meta is building a large data center, teachers received a $50,000 bonus this year because of the increased tax revenue from their investment there. The superintendent told Meta, that teachers are now moving there from across the country and he believes it will become one of the nation's best school districts. There's been a lot of skepticism about money actually flowing from AI companies to citizens because a lot of times money just goes into the government and then it gets like squirreled away into weird projects that different, you know, developers can like pull from and the money just sort of disappears. This feels like a pretty linear track from like data center built dollars in citizens' pockets. That's very, very good. I like how that played out. He also said that Meta's data centers are designed to be among the most water efficient in the world. We are committing to being water positive, which is a frame, like a term that I hadn't heard before, but I think is a good one. So what that means is they will restore more water than they use in the watersheds that they operate. And the goal is by 2030 to be net water positive. And in areas with high water stress, the goal is to restore 2% of the water that they use. And so I think that's good. Obviously there's been a lot of, uh, you know, it's been a bit of a quagmire on the water issue because there was that calculation error that people got way wrong. They thought the data centers were using way more water than they were. And so a lot of tech people just sort of wrote it off as like, ah, I don't even need to address it because it's misinformation. But of course it's a resource. It's important to address it appropriately. And I think this messaging does a good job of laying out like, okay, yes, like it might not be that much of an issue, but if it's not that much of an issue, then you can deal with it pretty easily and we're going to deal with it. So here's our plan. On electricity, they were a little bit less, a little bit softer. The language is just, we help keep electricity prices low by building our own energy generating infrastructure wherever we invest. This ensures that not only are we consuming energy that could— we are not consuming energy that could have gone to the local communities, but in some cases we even supply a surplus of low-cost energy back to the communities. We think this is an important investment principle for sustainability. That in some cases is doing a lot of heavy lifting there. Ideally, you would wind up— I think a lot of people would want to see like, uh, oh yes, in every case when you build a data center, you bring enough power that the price of energy does not increase in the local community. But good direction. He does call out nuclear energy in a really interesting way. He shares this stat, countries like China are, are bringing online 1 gigawatt of nuclear capacity every other week, which is insane. And he says, so we'll need to accelerate building of both energy and data centers to remain competitive. I thought this is a good framing and good reminder of how slow we are because we bring on 1 gigawatt of nuclear energy like every couple decades and they're doing it like every 2 weeks. Good. Good to see that. And then also Meta has been spending on nuclear power, making commitments. All of them are long-dated. There's a lot of regulatory hurdles, but he is putting his money where his mouth is in that regard, which I thought was good. And then, of course, the news is also paired with some real action from Meta. They're opening the weights of Muse Glimmer, a 30 billion parameter dense model that can run locally, along with a plan to release the weights to MuseSpark 1.2. Makes sense from a business perspective, I think, to ramp adoption. There was a lot that was going on, like the Llama ecosystem, and you could imagine that you download the weights to MuseSpark 1.2, fine-tune it, you're pretty happy. You're like, okay, yeah, let me get on the API plan as well. And also you can imagine that there's businesses that want to cut costs but don't want to use a foreign model. That definitely exists for a variety of reasons, so this is a good option there. And then it also just disentangles the geopolitical discussion from the safety discussion. So if you're talking to someone who has a strong opinion about open source one way or the other, you can just divide the discussion into, OK, how do you feel about MuseSpark 1.2? Is that too dangerous to release? Is this a safety issue? Because it shouldn't be a distillation issue, and it shouldn't be a geopolitical issue. So you can narrow it down just to, are you okay with strong, powerful, you know, AI being in the hands of just random people?

Yeah.

Anyway.

I think a lot of the stuff in here is like great and directionally correct.

Yep.

I think it was good that they tried to ground it in some concrete actions. It seems like they've chosen to position him as the pick-me AI. Lab leader, right? He's saying, if you're not familiar, pick me is an internet slang term for someone who tries too hard to get validation, attention, or approval. And that is certainly what he's doing. I think there's an incredible irony that like he, I think on Friday of last week is having to go through this like basically a billion dollar settlement for causing harm at scale with with all of their other products. And then ultimately, I still don't think that he has or like MSL has any type of cohesive vision or strategy, right? They were doing open source.

We were excited with the Manus thing. We were like, oh, this makes sense. There's like you deploy an agent within your Instagram account. There's a lot that you could do there, but that never really came to fruition.

Yeah. So they were doing open source. Then they're getting cooked and then they stop. And then now they're back. And the first thing they released was Meta Vibes, which is like kind of what you would expect them to release, but obviously got a horrible reaction. Yeah, wasn't world positive at all. Wasn't personal superintelligence. It was slop intelligence. Very much too soon, I think, for Mark to be like the good guy, like tech leader, where I just don't think he has like people's Like, his brand got so much better over the last few years when he was just doing the UFC thing.

Sure, sure, sure.

And wakeboarding and all that kind of stuff. But it's too early to be trying to parlay that into like, I can be trusted with—

do you think he needs to put points on the board or do you think he needs to move the needle before he can actually take a victory lap here?

I think, I think, I think he's trying to move the needle. He needs to be more focused on getting some points on the board. Because open sourcing a model that isn't a frontier model is just like, okay, like, like, I don't like knowing how commercial this team is. I don't think they would open source or open, make the model open weights if they felt like there was going to be exceptional demand for it. I think overall, like Mark standing up and saying like, hey, I want to be the leader of world positive AI probably hurts AI's public perception even more.

Yeah, he's being called Bernie Sanders now, right? He was included in that.

Yeah, yeah. Which is amazing for Meta, MSL, to be included in that group. Yeah. Like you would think that like—

Wait, was DeepMind not? Or Elon?

Elon and DeepMind were not included.

Elon wasn't.

So that was just because there were the three cyber incidents.

Oh, okay.

And so, I mean, people have been, you know, alleging that maybe the Meta, you know, cyber incident was maybe, you know, somewhat Not as real as other ones included, but I think he has, he has no confidence from the market on the strategy. He still has not actually explained what their product vision is. They're coming out with, with a coding harness. They're coming out with coding models. They're also doing open source. Yeah, there's whispers of them, you know, getting into the, you know, becoming a NeoCloud. It's just like, I think as I can, I can see why shareholders and employees just continue to be deeply frustrated.

The coding harness thing would be— it feels like it would be much more believable from Microsoft. Like, we own GitHub. We were first in GitHub Copilot. Yeah, Nat Friedman's not here anymore, but we have a really strong relationship with OpenAI for the models and Azure in the enterprise. We are going to make a run at this seriously. And they have a better strategy. I like the strategy more, but But it's a really, really crazy pivot to go from social media to the enterprise.

Ultimately, it's a lot of words from someone who I don't believe has any real vision or strategy. Like, at least when Dario comes out with an essay, this is someone who is undeniably a visionary, right? Has made a bunch of calls correctly. And so when he says something, I think it means something. But I think that, yeah, the product strategy so far just feels like throwing slop at a trough. Maybe. But defend it, John. Steelman.

What is the steelman? I mean, the steelman is the— basically, the steelman is like, this could go the DeepMind route where they get too— they get a little bit too— there's too much tension between the various corporate organizations. And the nightmare scenario would be like the Reelz team at Instagram saying like, we need the compute for Reels and ad monetization and that putting tension with the researchers. So, the researchers are saying, hey, that new data center looks amazing. We want to train some amazing model or we want to do some cheap inference. And the ad team and the Reels team says, hey, we really got to keep the lights on over here. And so, that tension just keeps butting heads against each other and they can't move as smoothly as some of the pure-play labs. At the same time, there is— I like this Boz quote about like, you know, Boz recently told employees that any extra time freed up by AI should not be used for vacation. Instead, it could— it should go towards building cooler stuff at Meta. Like you have so many users, so much data, so many different product surface areas. It's clear that we're not at the end of history in terms of what you can do on your phone, what you can do in these apps. And there are really delightful experiences that we run into all the time. Like Suno is a really cool consumer product. We have the CEO of Whatnot on, like that's seeing a bunch of traction. There are, there are things that you should be able to build faster, leverage the current technology stack more efficiently, and then also like add AI features to actually bring those to consumers at scale. Like when they launched their AI image model, I was like, why can't they just like when you open Instagram say, hey, we've pre-generated a Studio Ghibli version of your latest post or your Instagram profile. Do you want to share on your story? Like those little like product-led growth hacks that would be compute intensive. You'd have to inference all of that.

It's got a lot of compute. Compute.

Yeah, you got a lot of compute, but actually at the time, at the time that there's whispers of selling compute, they're also buying ultra long-term contracts.

The social network was like formative on all of us. Mm-hmm. Zuck is, is a legend in his own way.

Mm-hmm.

But it's just like putting this out while it's coming out, like less than 2 months ago that you're working on a gambling integration on Instagram.

Mm-hmm.

Like, please. There's no strategy. And then the other irony of this, the thing that stood out to me is he said, still, it is surprising that the discourse from many developing AI is so filled with doom. I do not understand why anyone who believes that AI will eliminate most of human, most jobs and much of humanity's relevance would rush to build that. The AI doom over the last 2 weeks has been around these loss of control incidents, which are like, which are serious, which are serious and, and real. And so to stand up there and be like, actually, guys, Nothing to worry about. I got this. AI is awesome. The birth rate might be collapsing and I might have something to do with it, but don't worry about it. Yeah, part of this is like, we've met Zuck multiple times. We've had great conversations with him. I like him. I like Boz. I like Adam. I like a lot of people over at MSL. Yeah. But this whole strategy continues to confuse It's deeply unfocused. It's deeply memetic. And it doesn't seem like it's going to get any better anytime soon.

The Wall Street Journal had a huge scoop. This summer's hottest arm candy is a private equity boyfriend. Reese Witherspoon, Nicole Kidman, and Olivia Rodrigo are all hanging out with finance guys. Jordi chimed in on this. It was remarkable. Every single person got Jordi's joke. I thought it was a little like inside baseball and you'd have to have more context to understand it, but it was like this rare, weird scenario where everyone understood that you were joking.

And yeah, I didn't get any pushback.

There was no pushback.

I didn't get called any names.

It almost would have been better if there was pushback, but it was just uniform praise for your quote across everything.

Now, a lot of people were saying brave. Yeah, finally someone said it. Yeah. So I don't know that there was a— it's like you take a lot of celebrities that are based in New York.

Yeah.

At least a few of them are going to end up dating some people in finance. So should we start maybe with what Mr. Wonderful said? He said, if you live in a celebrity bubble, party after party with the same people, the same 25 people, life gets really boring.

Yeah.

When you end up with a private equity guy, your life gets far more interesting. Mm.

Yeah.

He, I don't think, was rage baiting. No, I think he— so maybe that's how some people would have taken my quote out of context. What did I say?

Peace centers around—

I almost don't want to read that quote out loud, but you should, you should read, you should read my quote.

Your quote that burned up the internet. Whether they want to admit it or not, many women have always been attracted to bad boys and risk-takers, said Jordi Hays, co-host of the technology industry talk show TVPN. I can't think of any archetype of man that fits that description more today than capital allocators that put it all on the line every day in the market. And I like that your screenshot included the scroll wheel showing that you're hunting around for this.

And then so I posted this and I said, I weighed in on one of the defining stories of our time. And so putting such a silly caption above a silly story and a silly quote, I just assumed that it would be quite obvious that it was satire. Let's pull up this 4x2x2. And so, one, top left, let's give it up for the elite quadrant. This is all of you. I mean, zoom out a little bit, a little bit more. People that understand the joke. Zoom up or go up. And people that love it. These, you guys are saying we get it and we get you. And that made me feel really great. Now go down. This was also a powerful part of the, of the people that were supporting it. People that don't understand the joke.

Well, there's some people that just think it's great.

People that support it. Yeah, there's some. And there was a lot. There was a shocking amount of people.

Like, this is—

Can you guys just zoom fully out and leave it? Yeah. So this was people that love it and people that don't understand the joke. And I just want to say, bless their hearts. They don't get it, but they vibe with it. And then, and then we had some respectable haters. People that hated it. They get the joke, but they just hate that they do. And then there was the bottom right, which was tearing it up in the replies. Loud, wrong, and very confident.

We should take people through this article a little bit more because the piece centers on Oliver Harmon of Searchlight Capital. Of course, everyone knows him. But he is apparently dating Reese Witherspoon. And so just to get everyone up to speed, Witherspoon is an Oscar-winning American actress who became famous in the 1990s and 2000s, including films like Election, Pleasantville, Legally Blonde, Walk the Line. So now you should be up to speed on her. Michael Reinstein, of course, of private equity from Regent, is dating someone named Nicole Kidman, which our audience might not be familiar with. She's actually an Australian-born actress who's Her career stretches across prestige cinema, blockbuster films. Seems like she's been really successful. So good to see her profiled alongside Michael Reinstein of private equity from Regent.

I think a lot of, a lot of the people that, that watch TVPN were reading that article to, to learn about these celebrities.

Yeah.

Celebrity.

Yeah. It's a fascinating world. That's, it's just tucked behind, you know, so much obscurity. You don't hear about them if you're in the finance world. And so good to see. Folks like that get some coverage and really educate the audience. Joe Weisenthal had the best take that this is the classic cynical pattern. I think he's meaning cyclical pattern. Back in the mid-aughts, there was a raft of stories about celebrities dating real estate executives and shipping heirs like Stavros Niarchos and Paris Latsis.

We got to find what all these guys have in common, John. We got to find what they have in common because they're working across so many different industries. Booming, right? But what is the common ground?

The industry is booming.

Could it be something to do with their fee streams?

Well, yeah. Well, someone had a much more nuanced take from this, which was that celebrities are becoming like multiplatform, multidisciplinary. There actually are multiple income streams. And so this was not— I gave a quote to the news, but someone, someone chimed chimed in and said like, yes, a lot of these people are, a lot of these celebrities are doing much more complex deals, starting brands, starting companies, building holding companies. And so, they actually wind up interacting with like the finance bros much more often. And it's not so much like, oh, you got set up with someone with money. It's much more like, oh, you work on deals together and you talk about the same things because when you're a celebrity at a certain level, Yeah, you're going on set and filming a movie for a couple weeks, but you are also talking about your, all your, your catalog, your royalties stream.

No, royalties are often— yeah, they're often forever, right? Um, unlike, uh, unlike management fees with at some point or another that, that, uh, revenue line comes to a close.

Anyway, uh, we have another story here that we should go through. South Korea's AI boom is reshaping the dating market as engineers from Samsung and SK Hynix become some of the country's most eligible bachelors. According to The Wall Street Journal, engineers at the two semiconductor companies are increasingly being viewed alongside doctors, lawyers, and accountants as top marriage prospects. Samsung employees are expected to receive average bonuses of roughly $400,000 this year. While SK Hynix workers could average closer to $500,000. Chip engineers have historically ranked below more traditional high-status professions in South Korea's highly competitive marriage market. But local matchmaking agencies now say that gap has largely disappeared. While the trend has become a recurring joke in Korean pop culture and dating shows. Interesting. The newfound status is creating some awkward side effects. Some engineers say they are reluctant—

These poor guys.

It's rough. It's rough out there in Korea. Some engineers say they are reluctant to reveal where they work on first dates because they worry potential partners may be more interested in their compensation than in them. Others have increasingly started dating fellow chip industry employees instead. Meeting someone in the fab, a real meet cute. You're carrying a tray of wafers bump into someone and it's love at first sight. You look up as your scattered HBM is being—

It is very funny. They included— they just included a picture of a young woman here that says Kim Young Won, a photographer in Seoul, laments not having any contacts at chipmakers. Wow.

Really, really, really—

Framing her as part of the problem.

Well, the photo credit is her, so she must have posted this on social media or something. And then the The Journal got a hold of it. It's a fun one. Anyway, the other, the other big story is that an AI assistant in Australia hacked a gym's booking system on its own after being given the simple task of reserving a class in what is believed to be the country's first known autonomous cyber attack by an AI agent. Naughty, naughty indeed. Funny, in college I built a system that would register for classes automatically, but it was just a Python for loop that would refresh the page when the classes would become available because they would go online at like 7:00 AM and I wanted to sleep in, just lazy bones mode. And I think there's a story about Mark Zuckerberg doing a similar thing, sort of like hacking the admission system or some sort of a class registration system. But now with AI agents, you can just have the AI agent do it yourself. Anyone can, it's been democratized. Everyone can sleep in.

SpaceX is up 13%. In the past 5 days, almost certainly in part due to Semi Analysis. Major, major bullish update. Crazy bull case. Yeah, not— they've been contrarian lately, right?

Yeah, totally.

With Meta. Yeah. Now SpaceX.

So the headline here, SpaceX turns a compute company into 8 quarter— in 8 quarters into a—

Turns into a compute company in 8 quarters.

AI compute goes from nothing to 77% of run rate revenue and the illustrative 10 gigawatt path. I mean, he's really good at building data centers and they're predicting that Microsoft will be the largest off-taker. OpenAI and Anthropic will wind up doing deals with them. $500 billion in ARR for SpaceX. And the number that was getting thrown around was a trillion, a trillion in ARR or something. By 2030-something, which is staggering, staggering. They're going to have to rebrand it though, Land X, because they're not going to space with this. This is all built just on the ground. I'm pretty sure this is all pre-space data center. Space data center is the Act V or whatever Act we're on, the Act II. So Land X or Ground X, but either way it's working and very, very exciting to see.

Well, folks, we hope you have some crazy good times this afternoon.

Thanks for tuning in. Leave us 5 stars on Apple Podcasts and Spotify. Sign up for our newsletter at tppn.com. We'll see you back here at 11 a.m. Pacific tomorrow.

Goodbye.