I see a large IP on the horizon. You're surrounded by journalists, hold your position. On Wall Street, money never sleeps. Overnight success. The placemaker 3000. Double. Right. This misinformation. Clearing order inbound.
Let's just roll. We are surrounded by journalists. Hold your position. Get up. Trust the experts here. Five cookies. We are expert found. I see multiple journalists on the horizon. Stand by. UAV online. Blaze. Double blaze. Triple blaze. Double kill. Go, please. Team Deathmatch. We are experts. Triple blade. Let's just roll. Right. Quarter clearing order inbound. We are surrounded by journalists. Hold your position. Strike one. Strike two. Activate golden retriever mode. Market clearing order inbound. I see more journalists on the horizon.
You're watching TBPN. Today is Friday, July 10th, 2026. We are live from the TBPN Ultradome, the capital of technology, the fortress of finance, the capital of capital. Let me tell you about Ramp.com. Time is money. Save both. Easy to use corporate cards, bill pay, accounting, and a whole lot more all in one place. SK Hynix is up. Successful IPO, up 14% in the Wall Street debut. The South Korean memory chip giant SK Hynix made its NASDAQ debut today, raising $26.5 billion in the largest ever first-time US share sale by a foreign company. Demand reportedly exceeded the available shares more than sevenfold. But to give a sense of the scale of this thing, normally when we hear about 7x oversubscribed IPOs, you might be expecting like a 100% pop, 80% pop, but at this level of price and demand, you're just seeing a 14% pop, but still a great debut nonetheless. The company's American depository receipts opened, the ADR shares opened at 170, 14% above the 149 they were offered at, giving SK Hynix a market cap of roughly 1.03 trillion. We got another one, another one, another trillion dollar company in the AI boom. The blockbuster listing caps an extraordinary run for memory manufacturers fueled by surging demand from AI systems, particularly for high bandwidth memory used alongside advanced AI accelerators. SK Hynix, Samsung, and US-based Micron dominate the global memory market. And SK Hynix's Korean-listed shares are up over 600% in the past year as investors rushed to gain exposure to one of the AI boom's most critical bottlenecks. But it's been tricky to get shares and participate in SK Hynix because you got to go over to the KOSPI, you got to go over to the South Korean index instead. Now you can just head to the NASDAQ and get your ADR shares in SK Hynix. So, congratulations to the whole team over there. A massively successful IPO. There's more coverage in the Wall Street Journal. The Wall Street Journal asks a question. They say, 'Why is SK Hynix trading under a temporary ticker symbol?' They're currently trading under the temporary ticker SKHYV before reverting to a permanent ticker on Monday, SKHY. The V in the temporary symbol indicates the shares are trading on a when-issued basis. They haven't been fully issued yet, meaning the exchange is allowing investors to buy and sell the stock on its first US trading day before the transaction actually settles, as they're sort of being eased into the American markets. This is not the case with all ADRs, all American depository receipts, but rather a tool used to close the gap when a transaction has been announced but the actual delivery of the shares hasn't finished. So, all in process now to bring SK Hynix to the American markets. And that looks like a lot of red, but the stock is up. It just opened at 170. And so, the fact that it is flat relative to where it opened does not reflect the 14% pop or 16% pop. What was it? 14% pop from the actual issuance price. Most of the folks that bought shares in the IPO, that $26 billion that went into the company, of course, Sequoia, Andreessen, there were a few others, Baillie Gifford all participated early, at least according to early news. I assume that those transactions closed. They were all sitting up 14% today, and we'll be continuing to follow this.
Connor in the chat says, 'Cern has only been off for a week and John is dressing casual.'
Ooh, you think it's a conspiracy? I like it. I like it. Well, before we move on to the next story, let me tell you about Cisco. Critical infrastructure for the AI era. Unlock seamless real-time experiences and new value with Cisco. 5.6 launched yesterday. 5.6 Soul and Soul Ultra, and people are having fun with it. More and more demos are hitting the timeline. There's a lot of controversy over the desktop app for Mac. We can get into that, but first, let's play a little bit of this video. Someone tried to recreate all of Interstellar. Truly trying to infuriate Christopher Nolan the most. Probably this is the exact opposite of what he wants to see. I imagine considering that he doesn't even use CGI very often. Some miniatures occasionally. He uses a little bit of movie magic here and there. Some set extensions, I'm sure. But this is an incredible demo. Although I don't think this will be picking up a billion dollars at the box office anytime soon. But it is incredible that you're able to sort of go from prompt to 3D model to rendering so quickly. Very, very cool demo, and it feels like a new meme, a new art form. You have to be creative with it. Merely recreating Interstellar is not going to do anything. It's not going to do anything other than just be an impressive demo of the technology. But when you look at what people have done with Blender, you can imagine that with creativity added to this and a new storyline and something that leans into the rough edges of this particular...
Well, there's also going to be a whole meta of like, 'I recreated Interstellar, but I cast Conor McGregor as...' Yeah. You know, and I imagine that this gets even more powerful when you use style transfer and video models on top. I've been seeing a lot of people block out AI videos either just using their iPhone. So, they'll just have like you're playing an actor who's playing like an old character and then you're reading the lines, you're giving the delivery, and then that is used as the driving video for some of these AI video models and you get much better results with that. And I've also seen people block out little examples of driving scenes or different mechanical scenes in Blender and then use an AI video model to take it to high fidelity. And because the video model has an underlying reference that is deterministic and physically accurate, even though it's low fidelity and very blocky, you wind up with a better AI video at the end of the project. So you could imagine someone taking an idea, puppeteering a 3JS model like what's happening here and then adding basically a style transfer layer on top with AI video to get to something that's a lot more photoreal and more striking. But again, you wind up in this situation where if you can go and recreate Interstellar perfectly, but at the end you just have Interstellar which already exists and is available to stream and so you didn't really get anything in the process. You have to bring something special to the table. But I'm excited for the inevitable mashups of this, the Conor McGregor starring in Interstellar or something like that. Because people will be creative with it eventually. But, it's day one. So, also, there's a broccoli farmer that's running his farm on GPT 5.6 apparently. This was in the demo video, but Daniel Ten Riro shares a very funny meme showing the... I don't even know what aesthetic this is. This is the Sharezone aesthetic. That was the account that shared all of this. The extreme skeleton with the two revolving pistols. Very fun. Daniel Ten Riro shares some more information. He says, 'We were promised flying cars. Instead, we got a broccoli farmer running his farm with GPT 5.6.' I don't know. It's still relevant to show a demo that is as far afield. Everyone knows how these models can be used for research, how these models can be used for coding. That gets a little boring. I want to imagine something completely different that I can then pull into my daily life. What is the equivalent task that I'm doing in my life that I can take... We sort of differ on this sometimes. You think that a lot of these things are self-explanatory. They don't... This is Joe Weisenthal's position too. He says like AI has no learning curve. I disagree. I think that it doesn't necessarily have a learning curve, but there is an educational process that happens when you're presented with a blank box. You can type anything. People need to understand the fringes of their imagination. What is possible? And then they can go and reach for that tool. But the tools are getting sharper and sometimes they get sharper without the box actually changing. Yeah. Because it's just a box of what do you want to build or what do you want to do? Yeah, not everyone has a Tyler where I tend to prompt Tyler and then goes into Codex.
Yes. There was another video that was shaking up the timeline. People were going back and forth on how fast this model was building in... Is this Blender?
This is Blender, right? Yeah. So it's building a 3D cannon, modeling it and adding different primitives, different spheres and toruses. And the video is not sped up. And Chris says Soul 5.6 Soul is, and what I assume is 750 tokens per second in the wild. That is the preview of Cerebras. But this apparently was not actually...
Yeah, Soul Ultra on fast mode.
So we should be seeing even faster rollouts. Although every time I see this, I think...
And this is just computer use.
Yeah. Like, this is not an integration. This is just...
Yeah. Yeah. It is an interesting experience watching a machine use your machine at a rate that would be completely impossible unless it was like your entire bit online was like, 'I just use software really fast.'
Yeah. It is, yeah, it's an interesting sort of feel the AGI moment. They got to go back and beat the original Dota 2 model with just computer use because the original Dota 2, the what was it? Dota 5, is that what it's called?
Yeah, I think that that's what like the OpenAI project was called.
Yeah. So, there was this and it was brought up in the court case a ton. And there was an AI model that was, I don't even know, was it even transformer-based? I know they RL'd on Dota, but basically they were able to interact with the video game Dota, but essentially over an API and so they were able to send different commands, but they weren't actually clicking. And a big thing in League of Legends and Dota 2 and all of those games is you have to physically move the mouse. And so actions per second or actions per minute, APM, is like very important. And so when you watch the greatest League of Legends players, the greatest Dota 2 players play, they're... Yeah, Starcraft, they're clicking super fast. And it's very impressive from like a mechanical perspective. And so it shouldn't be that surprising that even though it was a demonstration of thinking ahead and strategy and actually balancing all the different tradeoffs in these games, when the computers won in Dota 2 and Starcraft and whatnot, it was impressive, but the question was always how long until you can just give the computer the mouse, the raw mouse and keyboard and just the images from the screen and can it still win with that because that's much more akin to what a human does.
Oh yeah, I was going to say like, yeah, they weren't transformers. It was LSTMs which is like pre-transformer.
Yeah. Long short-term reasoning, right?
Yeah, long short-term memory. Anyway, Packy McCormick is having fun with 5.6 Soul. He says it one-shotted this on its own. He didn't ask for a new logo. Just like, it's just like, 'Look, bro, you need a new logo.'
Actually, I actually think you should adopt this.
It looks cool. I like it. He says, 'It's part of a web design based on a link to a Cosmos. Not perfect, but way better than any other time I've tried.' And that's been the big talking point is like they were behind on, or the GPT models were sort of behind on front end and design. And this is a big leap forward in that. So I'm excited to see...
Nick on the Codex team is asking, is this image gen? I think it might be.
Well, I mean, that's the thing about Soul and Codex. But it's been generating, it's been now if you're asking it to build a site, it will generate you a preview image. And so I think this is the preview image for the site.
The really confusing thing is that the image model now has tool use and can scaffold things out in HTML if it needs to. And so like all of these different models are sort of blending together at this point. Michael had a very funny interaction where he got banned from OpenAI for cyber abuse. He had no idea what he did. He pasted the ban notice into Codex, which I don't know how he was still using Codex if he was banned, but he said he asked it to figure out what triggered the ban. Codex found that it had asked for an API key to his own server. Codex wrote the appeal, submitted the appeal. A few minutes later, the appeal was auto-approved by an AI at OpenAI. So he was banned by AI, convicted by AI, defended by AI, and pardoned by AI in about 10 minutes. What a funny, funny round trip in the world. So there's a bunch of pushback on the new ChatGPT desktop app because it is the super app. It unifies Codex. The new ChatGPT app is just the Codex app. And when you install it, you can keep the old ChatGPT app as ChatGPT Classic. But a lot of people don't like that because the Codex app is written in Electron, which means it's a bunch of web views, web browsers effectively, as opposed to the chat app which was a native app and had a bunch of integrations that people had sort of come to know and love. And so this is like a change in the user experience if you're deeply embedded in a particular workflow. This could be somewhat annoying because you have to use the legacy app and who knows how long they'll support that. The flip side is like I just don't know how many people were daily driving ChatGPT as a Mac desktop app. We've heard from someone on the team that was effective for cheating and we won't talk who it was, but...
Well, so I would sometimes use it because there's like a very nice shortcut. Yeah. So it was like, I think Control plus or Command plus is the normal like main search on Mac. But it was like a very easy way to interact with models.
Yeah. I wonder if that can just be... can that shortcut just be created with the new Codex or ChatGPT work?
Yeah, probably where you just bind that same key shortcut. But I mean it is making the user do work which is always a tricky thing. But I feel like for most people's workflow, at least for mine, it was the desktop app that I used was Codex. And then when I wanted to interact with ChatGPT in a chat flow, I would do a new tab. And then on my phone, I would have the ChatGPT app. And then the ChatGPT app on my phone could talk to Codex on the desktop. But I never actually needed to open the ChatGPT desktop app. But maybe I'm just weird. But I don't know. We'll see as this...
Same flow for me. Tibo says, 'We spent the last 24 hours reading feedback, looking at usage patterns, and talking with many of you. The short version is that there's a lot of excitement for 5.6 ChatGPT, but we also didn't get everything quite right. We made it too easy to use the highest compute settings without making the impact on usage limits sufficiently clear. They reorganized the desktop app in one bold move, making familiar things like chats and projects harder to find. Our launch framing was focused on work and to some of our Codex fans it made it feel like Codex was going away over time. Absolutely not our intention. We love Codex and it is here to stay.' And then anyways, long story short, reset, usage and limits are getting reset twice today so people can keep experimenting. Hammer it, plan out some sites. There's also some confusion over the difference between ChatGPT Work and ChatGPT Codex. And I wonder...
whether one or the other is the place to do things.
Yeah. Yeah. The omni box that can do anything for you. Yeah. And I think that's obviously where it goes. But there's a walk, crawl, run. There's integration points and people have ingrained experiences and workflows and habits. And there's all sorts of different tradeoffs here. There's a good point like there is no enterprise Google search. Like you just use Google at work, you use Google at home. It's all one box that gives you answers and there was never a dividing line between these two things. And so, that's probably where all of this goes ultimately. But I don't know. I was chatting with the... there was a lot of pushback on the fact that there's no chat... there's no native chat window. There's like a slight popup, like a small little mini popup for your just general ChatGPT usage when you're in the ChatGPT app which is now Codex and Work. And so that can be a little bit confusing, but maybe the end result is that you should just be hammering new task for everything. Like I demoed this today. I was doing some research on the most recent story that we're going to talk about with Phoebe Gates's company FIA. And I asked a very ChatGPT-like question. I just dumped it in Codex. Like, 'Explain how this story is different than what happened with Honey. Is there anything illegal here or does it just break the norms of affiliate marketing?' Do a bunch of research, put it together in a couple paragraphs and it just did it like it was ChatGPT. So I think that over time you should just chat with Codex or Work, whichever one you prefer I guess. Like it's ChatGPT and it will come back to you in a similar time. You'll be able to select models just like you want. It'll be all of the same things, but then you got to mirror that over the phone as well. So, I don't know, some potential growing pains as the app evolves and the use cases evolve. What are you laughing at?
Nothing. John, what should we move on to the next...
Which one are you looking at? Which one are you laughing at?
The most recent one. How dare you? Let's move on. Let's move on to console. Oops. Actually, let's tell you about Codex. There we go. Codex is a powerful workspace for getting things done with AI agents. Whether you're writing code, analyzing data, creating content, or automating business workflows, Codex projects forward from start to finish.
People are saying it's a horse for the mind.
A horse for the mind.
So the Meta case, SemiAnalysis wrote a long piece about the future of Meta, what's going on there. Very interesting contrarian take. I think a lot of people were confused by the expansion of activities and sort of an entirely new business line with B2B sales. And we've discussed the bull cases here, the benefits. Obviously, there's a ton of benefits with the ad model. They have the cash flow to draw down on. There's a lot of different things that we can go through, but let's read some of the reactions first and then we can go into the SemiAnalysis piece. So, Boring Business says people don't realize that Zuckerberg is all in on AI because he refuses to pay a 30% tax to platform owners for the rest of his life. Meta caught the shift toward mobile back in the early 2010s, but they were never able to create their own hardware platform to capture users in their ecosystem. As a result, they got stuck paying a 30% tax to Apple on in-app purchases, which has cost Meta billions of dollars in profit. That makes no sense to me.
No one buys ads via in-app purchases. Every advertiser I know uses the web-based...
Also, ads I believe count as like a real-world service. So there's not like... there's a lot of things you order in the... that you order on mobile apps like groceries.
Kind of a digital good. But yes, I agree with you that it would be exactly...
Yeah. But it's like eyeballs, right? It's like impressions. Which like eyeballs, human attention is not a digital good. It is like a real-world thing.
Yes. So if I open up the Amazon app on my iPhone and I buy a t-shirt, Apple does not get 30% of that because that's a real-world... or DoorDash, good example.
Yes. But if I go to Kindle or Netflix and I subscribe or even like the X subscription premium, Apple will take 30% if I buy Fortnite V-Bucks or Roblox or any digital good.
Yeah. And I could see Apple saying, 'Look, if you want to be able to...' because in the Meta platforms, you can be on Instagram and you can just say, 'I want to boost this post' and you can put down your credit card and pay for that and you can get started on the advertising flow as a very small business, I think through effectively in-app purchases.
Yeah. So the bigger thing is just like...
One of the largest companies in the world.
And you have to constantly be playing nice and sort of sucking up to Apple, who is sort of actively... who doesn't like you. Like Apple is... like you can imagine what do Apple executives think about Meta if you were to ask them over coffee off the record and it would be like, 'We don't like Meta.' And so that's fundamentally like a very, very uncomfortable position to be in. And so that is like the main... That's like the main... he doesn't want to be in a position running his company where he has to ask for permission for things. His business is sort of like running with permission from this like platform, right? And so an example with that, somebody brought up, I think Meta lost billions from Apple because of the permissions prompt.
Yes. You know, making it harder to just run their core advertising business.
Yeah. And so I think there's just a lot of PTSD from that. And also there's so much that is very unpredictable about where AI is going. Like it's possible that we get to the point where you don't go to a bunch of different social media platforms and you're just like... your agent is just kind of surfacing content for you on the fly. Like there's so much potential disruption. I think the stance is like, I would rather participate in this massive technology cycle and have a shot at being a big part of it. Yeah.
Than like just like hope nothing bad happens to me. The other thing I've always felt is like, you know, with the glasses, you're just going to be able to like... he sees a world like the glasses are more of a new platform bet right now than AI.
And I just think he wants to be in a world where you're wearing your very, very cheap pair or maybe even free pair of glasses around. They act as your phone and your computer and you can look at stuff and say, 'Buy me that.' And he's getting a cut of it. You're walking by a restaurant and it pops up an offer and it says like free boba for any orders over $20 and he's getting a cut of that, right? So like he... and that will be like some type of like always-on intelligence. So yeah, it all totally makes sense even if it is like deeply sort of like chaotic and at times confusing.
At least over the last year. The funny thing about the glasses as a platform is AI spending is so huge that he can probably bring back a lot of the Reality Labs spending and sort of just tuck it in a bucket and people won't notice if there's an extra 8 billion.
Yeah. I mean, if there's an extra 8 billion spent on the next version of the Meta Ray-Ban displays or something, people will be like, 'Ah, it's a rounding error compared to the 200 billion on data centers or whatever.' Ari puts it well: the tax Meta pays to Apple is the fact that they can't track users cross-platform.
So continuing with this analysis, Zuckerberg has been keen on making sure this natural monopoly or duopoly structure doesn't emerge again in AI where a couple of players can become the toll booth for accessing every platform on the web. For Zuck, this is existential. And if he has to burn every last dollar of free cash flow today to achieve this, he will do it. If that sounds crazy, look at it from the perspective of Zuckerberg. He has held a job... he has never held a job in his entire life. Being Meta's CEO is all he's ever known since he was 18 years old. He has nothing to lose. He's the exact type of person you should never bet against no matter what odds someone gives you. That's a good point. Mohan here adds some extra context around the in-app ads revenue. In-app ads is 2% of ad revenue. That's actually pretty high. And then they do pay 30% on that. I think the chat confirmed that as well. So, 98% of FB revenue is not tied to Apple, but the bigger challenge is everything that you just laid out around the front door and the privacy and tracking and all this hostility. The other interesting thing is like I do think the screen time is at least like a flashing warning light. Like I've personally noticed as an early adopter of LLMs that my screen time has switched much more towards ChatGPT and LLMs and deep research reports just because if I see something interesting, oh, there's a car video on Instagram, I will often switch over and then start prompting to actually learn more. Whereas before that would have required a whole bunch of Google searches. You might have to sit down and like collect a document, like you're not actually going to do that. Now, I've been noticing that I'm much more in like the hour a day on chat versus hour a day on Instagram. And so, I don't know if I'm just like the early adopter that will never do that, but you can certainly see that people are spending a lot of time with AI systems. And as you've put it before, like wherever consumers, wherever people are spending time is an important place to have a foothold at least.
Continue. So what SemiAnalysis wrote yesterday, which is what I believe actually moved the market, not the model release itself. SemiAnalysis said at the simplest level there are three things you need to build a true frontier model: data, talent, and compute. We believe Meta is the only hyperscaler/neolab on track to be world-class at all three and therefore has the best chance at catching up with Anthropic, OpenAI. We'll explain why in full detail below. And so Alex Wang quoted this in a funny way. He said, 'Compute Daddy Dylan Patel has spoken.'
It's a good point. So yeah, pretty... and yes, the stock is up 6%. Wang quoted... Polymarket Money said Meta jumps over 6% at the open as investors react to yesterday's Muse Spark 1.1 release. Again, I don't really think that they were reacting to Muse Spark as much as SemiAnalysis saying there's a chance. So yeah, Alex Wang said nice with the thumbs up, so he's pumping the stock.
What's interesting is like... the crazy thing is that like SemiAnalysis just like completely leaving out SpaceX and...
They didn't leave out SpaceX and Google. They said...
They said xAI is selling $26 billion a year worth of GPUs to Anthropic, Google, and that's why they're not in the race. Even though it is odd because Meta might wind up selling a lot of GPUs. It's sort of unclear what the breakdown will be, but they're at least laying out a narrative that if Meta picks it up, it would differentiate them from xAI in some ways, I think. But yeah, I agree with you. It is...
But they're essentially saying that like Gemini doesn't have... because Gemini has compute.
They're effectively saying Gemini does not have talent.
Yeah, right. They're saying you need all three.
Well, the main thing that they were focused on outside of the compute, which you know, many of the hyperscalers do have that box checked. SemiAnalysis really focuses on this idea that data is the new oil. They're quoting Ilya in 2024. Ilya famously said that data is the fossil fuel of AI. And they say, 'This time the invisible hand has created a new human data RL environment supply chain. The three incumbents, Mercor, Surge, and Handshake, are all at 1 billion plus ARR and many new entrants that are barely a year old are sitting around 100 million ARR.' But what's interesting is they actually ran the numbers on how much data Meta is gathering from the computer use screen recordings of Meta employees and it benchmarks like... they basically got a Mercor-ized business overnight for free or for loosely for free. So, they said, 'Furthermore, Meta took their data efforts to another level in late May by announcing a new applied AI engineering org as part of their most recent round of layoffs restructuring. 3,000 engineers, which includes 70% of their new grads and a significant number of seniors, will now be making RL tasks and environments full-time. We think this is an extremely underappreciated advantage for Meta.' The question is if this becomes the standard playbook, you would imagine that Google DeepMind would be able to do the same thing. They have the same structural workforce, but maybe it's like a band-aid that Google's not willing to rip off, but Zuck is willing to rip it off at Meta. That's maybe the thesis at a deeper level. Anthropic has been the most aggressive lab by far when it comes to buying coding data from RL environment startups, and it's one important reason why their models are so good at coding today. Mercor recently disclosed that they logged 2.5 million expert hours on their platform in Q2 of 2026. That's equivalent to 5,000 people working 40 hours a week. Meta is already in the same ballpark and their average quality is likely higher. Additionally, they have another 70,000 people to pull into that organization if they need to, if it winds up being as valuable as SemiAnalysis thinks it will be. It's also worth briefly dispelling this myth that 3,000 Meta engineers will be doing mindless low-level data labeling. The days of undereducated contractors from third world countries drawing boundary boxes or classifying text as NSFW are gone. At this point, the models are sufficiently smart that creating a good piece of training data is a real intellectual challenge. So, they're sort of taking the contrarian stance that being an RL environment designer is not soul-crushing and in fact can be intellectually rigorous. They also say that the top expert contractors at every data company, they're all making seven figures a year, which is pretty crazy since people thought of data labeling as a $10 an hour task. And now you have people making millions of dollars a year because they're actually really great at defining what it means to complete a task from start to finish.
And that's exactly why Lucy Guo just released her new single.
You going to play it? Yeah, we can play it.
It might be the second greatest song ever released by a tech founder who exited their company for north of a billion dollars.
Yeah. So you kind of blew my mind with this this morning that like Justin Kan has created a smash hit.
900 million just on Spotify alone.
So, Lucy Guo, co-founder of Scale AI, sold to Meta for this $14 billion deal. She was the co-founder, obviously had a ton of equity, has become incredibly liquid, and she's having a lot of fun with it, and is releasing a new... is it an EDM track? It feels very like Barry's inspired.
It's a bottle service in the club track.
I feel like that's the intention of it.
So, it's called Champagne on Me, right? So it's clearly a reference to buying champagne. Very funny in the photo shoot that she put out for this, she's holding a wine glass. I don't know if that's an intentional like rage bait to engage the commenters.
You think it must be? Okay. But we will see where her EDM career goes. But it is worth reminding everyone that Justin Kan has been incredibly successful in his post-exit EDM career. He launched a group and collabed with a number of other artists and did release a song that got 900 million streams on Spotify alone.
Partnered with Adam Port. Yep.
And the song Move, which I have heard played everywhere. I've played it myself. I had no idea that Justin was behind it. It has 97 million streams just on Spotify. So you can assume billions of streams when you include YouTube and Apple Music and other platforms.
And just to put that into context, Love Sosa by Chief Keef has a billion streams. So he's put up a track on par from an attention standpoint with Love Sosa, one of the greatest, you know, pieces of music ever created in human history.
So it's very... pretty remarkable.
Well, it is very remarkable. We're excited to follow Lucy Guo's career, see where she takes it. I can imagine that she'll be doing live performances, more collabs. When you're in that position, it does seem like there's alpha in working with existing artists, building a team, collaborating and doing stuff and trying not to...
I want to see... I want to see what... you know, there's a lot of... Lucy Guo has fans. She also has some haters, right? She embraces them both, but I want to see her invest 20, 30 million into her music career by partnering with a bunch of the best songwriters, producers, DJs, like actually just take it all the way, like take headline Coachella just to piss off the haters.
Yeah, could happen. Let me tell you about Consoles. Consoles builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. So the market has spoken.
Chat is all saying, 'Chief Keef mention.'
Chris says, 'Usually every episode mentions Chief Keef somehow.'
That's right. Even when we don't mention it explicitly, sometimes, you know, we'll include references, you know, little...
Easter eggs. The Jensen jacket is on. Oh, yeah. Sotheby's, the current bid, the estimate from Sotheby's, I think they were kind of basically putting the estimate way too low. I said this before when it first popped up. They said 40 to 60,000. It's already sitting at 70,000. I don't know what in the slop this graphic is. Very, very oneshotted AI. But the important thing is that when we first covered this, I did not realize that the Jensen jacket was going to a good cause. I thought somebody just picked it up and was like, 'No, it's for charity.'
So, so Jensen donated this jacket to our dear friend Justin Mers. All the proceeds are going to go towards this thing Justin does called Inflection Grants, which gives no-strings-attached capital to smart people that have cool ideas all around the world. So it's a very good cause, and there's been a bunch of great sort of case studies coming out of it. So excited to follow this. It's not closing until July 17th.
Mhm. And so there's some time. I think this runs up, I'm going to put it at north of 500. So we know what the last one sold for, right? Over a million. Absolutely. I'm putting it at north of 500, but sub 1 million. It's a nice jacket. It's a good jacket, sir. Good piece of lore. Has Palmer ever auctioned a Hawaiian t-shirt? I feel like that might go for a high price. Maybe he has to go public first and really, really go broad before he pulls in the seven-figure auction for the piece of clothing. But I don't know.
I think the Museum of Business Anything... Yeah. No, I think Hawaiian goes for north of 200. And that's just based on like secondary demand for the stock, right? So if you have, you know, an eight-figure position in Anduril, you might want to just pick up the Hawaiian shirt, have something in the... yeah, it's a proxy, it's like look-through exposure sort of equivalent.
No, it really is a bet, it really is a bet on the company, right? At a trillion, suddenly that Hawaiian, that's going way higher, way higher. Zack, really, do you have more on that or are we switching gears? Let me tell you about MongoDB first. What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI. Own the data platform that powers it. Chris says, where do you want to go? He's buying Palmer's goatee. So, Palmer, shave the goatee. Put it on a board and auction it for a good cause. Zack, I want to close out.
Zuck is posting on X. Elon is glazing Anthropic. DoorDash AI company. CERN has only been turned off for one week.
Man, I don't fully understand the CERN conspiracy. I know it's like a meme now that it gets turned off and then crazy things happen. Is the idea that it keeps the world like stable or something?
I don't even know. I don't even know at this point, but the conspiracy crowd was saying like as soon as it got turned off, like, have you noticed that today felt like, you know, four days long?
Okay, so it's like time. Oh, cool. Okay, I like this. Yeah, I got to go watch like a two-hour fake documentary about it to get to the bottom of that. Really put on the tin foil hat. So yeah, closing out Meta. The market has spoken, according to Alex Wang, who says, 'An exec at a $1.6 trillion market cap company flexing his company's stock performance with a five-minute chart is absolutely diabolical.' But it is a very... it's the beginning of a new narrative and it does feel like Meta for a lot of people was never a SaaS apocalypse victim. It was always going to be an AI winner. So like they've worked on AI for a decade at FAIR.
They were an AI winner until they said actually we're going to invest way more into this. It's like whoa, whoa, whoa, whoa, whoa.
Exactly. And so it was always very odd because it felt like they were in Apple's position where they could sort of do nothing and win, and yet they were being very aggressive and acting like they had to do something. So they sort of put themselves in the Google scenario. But this is certainly the way that they get out of the Google scenario and actually boom. And Satrini agrees with that. Satrini says, 'Meta strikes me as one of those setups where it's going to rip so hard that when it's done, everyone acts like it was obvious the whole time and pretends they were buying the whole way down. We can call this the Google at $150 phenomenon.' In the last piece of Meta news, Yann LeCun is raising a new AI-focused VC fund. He raised a billion dollars earlier for his startup, AMI Labs. Now he's raising a billion-dollar fund. And he left Meta to raise billions to build and invest in AI, says Trace Cohen. So, good luck to the AI godfather, Yann LeCun. Very interesting. The take that I heard was something along the lines of like he might be right that LLMs are not the path to AGI, but it turns out they're good for Meta. They're useful and they're also very good for Meta because they need to do, as we talked to Eric Suford about, recommendation systems, ad generation. One thing we didn't get to talk to Eric about that I was interested in is the Muse Spark image model is maybe the more exciting in the short term piece. I mean, Muse Spark 1.1 is like, as SemiAnalysis puts it, on the right track to have a frontier model and compete in that oligopoly. That's great. It's not there yet and most people are saying that they are not swapping in Muse Spark 1.1 just yet, but they're at least on the right path and they have this data advantage potentially. But the image model right now can be used for ad creation and fine-tuned in Instagram for statics to create images that particularly work on Instagram. So if there are flaws, if there's a spiky intelligence to ad generation, if your reaction to the 'what in the slop' was that particular image, like there is a version of slop that actually will work on Instagram and some that won't, they might underperform if you're using them to deliver information infographics. So they might want to actually fine-tune their model on what's working on the ads side.
And they can do that. I'm interested in the fact that they have not talked a lot about text generation for ad copy. Like there are clearly people that are sick of the turns of phrase used by LLMs broadly, the 'you're absolutely right', all the Claude-isms, and then ChatGPT has the 'it's not this, it's that', or did until they sort of work that piece out, and then a new one pops up every couple weeks and people get annoyed by it. And it seems like there's actually an opportunity for Meta to create an LLM that's just really good at writing, you know, advertising copy, and it's not necessarily as focused on deep research reports or agent coding, but it's like superpowering the ad platform and figuring out should they be using emojis, how many emojis should they be using, bullet points when, and RL...
And if you're going after business customers, yeah, why not build something that your business customers would use and love? And I'm sure that every business that's running ads on Meta right now is going to ChatGPT and Claude and saying, 'Generate me 50 different lines of copy that I can use for my ads.' Obviously they're doing this. But if Meta can outperform them at that, because the big labs are not really focused on copy that converts or like landing page optimization, right? They're focused...
I know where you're going with this.
Zuck drops an e-commerce mastermind course to help fund the capex.
$5,000. I'll teach you everything I know about e-commerce.
Give it away for free. You're going to spend a lot more than that. But yeah, the Zuck course generates really hard billions.
Yeah, it would. It would.
Anyway, let's move on to our next story. But first, let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web apps, servers, databases, and more, while Railway automatically takes care of scaling, monitoring, and security. So, we got to talk about FIA Gate. FIA Gate, the story that is at the center of Phoebe Gates's life. Her company FIA is in Bloomberg today with some accusations of misattribution in the e-commerce world. Via the buzzy shopping app co-founded by Bill Gates's daughter, Phoebe is claiming credit for online sales that didn't actually drive it, a Bloomberg investigation found. You can read the exclusive story. Let's click into this and sort of understand what's going on. So, real quick, Ari's putting us in the truth zone on Meta. He says they have a product like that in the ad builder. Why would they make that external? Oh, yeah, I more think that it's not that it would be external. It would just be something that would be talked about more in the benchmarks in the actual model release. Yeah. And they have Meta AI, which is just a standalone app. Yeah. And there's tons and tons of people online building businesses that aren't necessarily running like Meta ads yet, right? They're not necessarily even set up in the Ads Manager, but they're thinking about like website copy and things like that. But could see it going either way. But I mean, we already know that Zuck doesn't like talking about the ads business, and so that's the whole discussion of like you want to talk about the agent coding capabilities of the new model. Even if I wouldn't be surprised if that functionality that you're talking about got better in the last month. Obviously it did. It's just something that's not top of mind for people. And maybe it should be. But honestly, I think the AI opportunity is so big that the market might just love this. I mean, the market already clearly... Ari says, 'Okay, I want to talk about FIA Gate, not Meta. Sorry to distract.' All good. All good. I appreciate the comment.
An advertising startup co-founded by Phoebe Gates, daughter of billionaire Microsoft Corp. co-founder Bill Gates, bills itself as a personal shopping assistant that helps users find the lowest prices on a broad range of clothes and fashion accessories. You download the FIA tool onto a web browser. It's a Chrome extension basically. Use it while shopping and voila, the so-called extension can quickly find discount codes for products. We've seen these with Honey and isn't...
Calling it so-called extension is kind of a low blow. Like it is certainly an extension.
Yeah. Yeah. Like you can't... I mean, I guess the really aggressive take would be like this is just malware or anything. This is, you know, software that... Okay, so businesses like these, known in industry parlance as affiliate marketing programs, typically collect commission from retailers that make the resulting sale. But according to Ben Edelman, an independent researcher and consultant who studies affiliate marketing, as well as Capital One Shopping, which makes a competing browser extension, FIA is claiming credit for online sales it didn't actually drive, in violation of many digital platforms' policies. Bloomberg tested the FIA mobile browser extension across more than 50 websites and found that during the checkout process, FIA opened a background tab without user interaction and injected its own referral code that overrode legitimate referrals from other publishers. These findings were consistent with Capital One Shopping and Edelman's independent testing and code review. Testing involved using the extension like a regular shopper and observing how it communicates with other sites and its own servers. 'The most fundamental requirement in affiliate marketing is that commission is only paid if a user clicks,' said Edelman, who has spent decades exposing what he says are deceptive practices in digital advertising. Edelman spoke to Bloomberg after reviewing FIA's code, which is public because it's an extension, you download it and install it, and testing how the extension interacts with retailers' websites and affiliate networks. At Bloomberg's request, 'the rules don't allow fake clicks, simulated clicks, imaginary clicks, or hypothetical clicks. Only a real click will do.' With Capital One Shopping, a discount shopping tool previously known as Wikibuy, was bought by Capital One Financial Corp. in 2018, highlighted similar findings from its own testing in an email sent Tuesday to retailers that raised concerns about what it said were FIA's fake clicks or cookie stuffing. It attached videos that it said showed FIA's extension silently opening background tabs linking the retailer's website with its own affiliate code to ensure the cookie was set. 'Publishers like us are having material revenue taken,' said the email, which was seen by Bloomberg. 'And advertisers like you are losing money to fake clicks.' Bloomberg cares about this because they are probably affected. The Wirecutter affected. That's owned by The New York Times. This is a PR.
Sites like Dupe are affected. Yes. We've had Bobby on from Dupe. He gave some good coverage.
So, why don't you give us the backstory on Honey? Yes. And why Nolwenn says this is what Honey got killed for doing.
Sort of, but it's potentially worse than what happened with Honey.
So you're saying FIA is potentially worse than Honey. Be clear.
So, you have to think of the most ridiculous example of affiliate marketing and product purchase flow. So this really lands with people. Yeah. You're buying a horse and you... So you search 'I want horse.' You go to horsereviews.com. You go to horsereviews.com and they've got thousands, maybe even millions of stallions available. Yeah. And they recommend you a Clydesdale from Clydesdalesonline.com and you click from horsereviews.com over to Clydesdale... bestclydesdales.com, clydesdaledirect.com, and you land on Clydesdaledirect.com. And at that moment, the URL almost certainly on that button that you clicked on horsereviews.com is to Clydesdaledirect.com?utm=horsereviews.com. So when you land on Clydesdaledirect.com, the actual e-commerce site that will sell you the horse, horsereviews.com gets attribution. They get credit for referring you. This will just show up in your Google Analytics. It's very nice. It's a very nice thing to see. 'Oh, I'm getting a lot of traffic from this. Maybe I should, you know, engage with that group or maybe I should pay them. Maybe I have to pay them or maybe they'll swap out the link to some other site.' And so there's a whole host...
Because a lot of the purchase decision-making is happening on horsereviews.com, right? You're comparing and contrasting...
The Wirecutter for horses.
Yes, exactly. Maybe this is... maybe we got a business.
Maybe, maybe, maybe this is something, you know, we should build. Obviously sites, you know, venture capitalists as well as journalists would be all over this thing.
Yes, we got to spin this up. So, normally the referring site can hold on to that referral token. It can get stuffed in a cookie. It can get saved until the user checks out on Clydesdaledirect.com. They go through the checkout flow, hopefully hosted on Shopify. Let me tell you about Shopify really quickly before we continue. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agents. So, you're going through the Shopify flow and this cookie, this referral is handed to you and then you wind up paying them for the referral. This is affiliate marketing. It powers a lot of the internet.
But what happens, John, back in the day, the buyer had Honey.
Yes. So, the best scenario with Honey is that you're on horsereviews.com. You wind up on Clydesdaledirect.com. You're thinking about purchasing. You've been educated by horsereviews.com. They deserve some of the credit, but you just think this Clydesdale is too expensive. Yeah. And so you want Honey more than you want to pay.
And they got a coupon code for you.
Clydesdale's 20 or something, or Budweiser 50. 50% off your Clydesdale. It's a Budweiser partnership. Who knows? And so you click on Honey and there is a coupon code. The coupon code is automatically added to that Shopify checkout. And as a result, Honey says, 'We're the affiliate marketer that got them over the finish line. We deserve the credit.' And multi-touch attribution, sharing of attribution revenue, not a very mature industry. Typically, whoever gets the last touch, the last click, they're the ones that get the credit. So, Honey shows up. 'Hey, we think we have a coupon.' You have to click affirmatively. You can't just do nothing. You click and the affiliate, the referral tracker is updated to Honey. That's in the fair zone. You might be a little bit upset about that because you're like, 'Wait, I'm horsereviews.com. I did 90% of the work. You just gave him a coupon code. Why are you getting all the credit?' That's up for debate, but that's generally okay. Honey got in trouble because they started for a time period updating the cookies without people actually clicking...
Exactly. Well, that's a weird joke because isn't the founder of Honey doing a company called Pi now or something like that? I thought there was some linkage there. Tyler, look that up. But the violation and why Honey went so viral. There's this YouTube video, 'The Problem with Honey', has 20 million views. It was attacking a lot of creators because MrBeast had worked with Honey, Marques Brownlee had worked with Honey, like they had run a lot of scaled YouTube ads because it was a great business. If you just got someone to install this free Chrome extension, they would save money. It's the best offer ever because you just say, 'Download this Chrome extension. You just save money.' Most people don't realize that Honey's making money through affiliate commissions and they're making a lot of it so they can pay the YouTube influencers to promote their product. Yeah. Anyway...
And back in that era, I would be insanely annoyed because I would make a basically like a landing page or a coupon code for a specific creator. Yep. And sometimes there'd be like a small creator. All of a sudden, they would have like hundreds of purchases through their code. I'd be like, 'Wow, they either have insane engagement.' Yep. Or it leaked to Honey. And then you check and usually these codes... Honey would set up their own... sometimes they would set up their own in your affiliate system. There were a bunch of annoying things where you would promote with a particular advertiser and they would be saying, 'Use code Rogan10,' and then the person would show up, put Rogan10 in, and then Honey would swap it at the last second to Honey10, and it's the same. So the customer doesn't care because they're still getting the same discount, but Honey's getting the credit instead, and it makes it hard to track.
The Honey founders' new company is called Pi, not Pig.
Vindicated. Vindicated. True. So let's get into what FIA is being accused of and why it is potentially, if this is what they're doing, we don't know again if it's an accident or it's intentional, but why it's potentially even more predatory than Honey.
So, MegaLag, the YouTuber that did the Honey expose, it was an interesting deep dive that got 20 million views because it was using a hidden redirect near checkout to replace the creator's affiliate tag with Honey's. So the irony here was that a lot of those creators, they would make money from 'I'm promoting AG1, use code Huberman,' and then if they had also promoted Honey, Honey was stealing their AG1 affiliate revenue. And so the creators were sort of promoting their own demise in some way. So it was like this very viral, very ironic situation. Eventually got sorted out. Honey's broader scandal also involved allegations that it suppressed better coupons and deliberately evaded affiliate network stand-down checks. Nothing comparably extensive has yet been established about FIA, but FIA's reported implementation may actually be cruddier. Honey generally created the overwrite when the shopper interacted with its checkout prompt, even if only to dismiss it. So you would see a Honey popup. At the very least, you had to click 'I don't want Honey' and then it would do something. The Bloomberg report says FIA's mobile extension is firing without any FIA interaction at all.
So, give this to me in horse terms.
So, you go from horsereviews.com over to Clydesdale Direct. You merely have FIA installed as an extension. You don't notice anything. You don't know that FIA has been a part of the interaction at all. You don't know.
You just click through. But you have the software running in your browser that is inserting an effective attribution for FIA without you knowing, stealing potential revenue from horsereviews.com. Yes. And horsereviews.com probably had to run ads on Google to get people to horsereviews.com to educate them on the horse market, to get them to Clydesdale Direct, to get them to close. And so they're sitting here being like, 'Hey, we did all this work to get Clydesdale Direct a new customer and we're not being compensated for it.'
Clydesdale Direct might say, 'Well, normally I'm sending a check to horsereviews.com. I guess I'm sending a check to FIA now. Whatever. As long as I'm getting customers, I'm fine.'
Ari, yes, the world champion of the chat today says, 'Okay, this is like old-school ad fraud.' So one of the concerning things here is that I think this had been running since like January or December, as the article said, and so they had some software update that was pushed. This started happening. So it's been happening for let's say around six months, and you would assume that they would have figured this out, that hey, if it genuinely was an error, and corrected it, right?
Yeah. But maybe it's possible that it, you know...
I don't know. I mean, it's not the biggest scale. I think the number is 12 million plus in sales driven. So the commission on that is like $1 million that's sort of missing, and it's so diffuse across so many different e-commerce providers. I guarantee you, as someone who's run multiple e-commerce sites, I would not notice if a couple hundred dollars moved from one affiliate to another because I'm one of a million sites that were affected by this. So it rolls up to a big number, 12 million plus in sales driven, but the actual commission might be 3%. So you might be looking at like 300k or something like that of actual money that moved to the wrong place. Bloomberg's not actually alleging a particular crime. If the facts are accurate, the behavior appears far beyond mere rudeness. Though affiliate network contracts generally require a genuine user referral and often require shopping extensions to stand down when another publisher supplied the consumer. You're not supposed to override the supplier of the consumer. Google expressly prohibits background affiliate link injection and replacing tracking without a related user action. Apple also requires Safari affiliate redirects to be disclosed. Civil theories such as deceptive practices, unjust enrichment, interference with contracts, and unauthorized alteration of tracking data are therefore quite plausible. Deliberate forced click, cookie stuffing, can even become wire fraud. The Justice Department has prosecuted people for receiving commissions on sales they didn't generate, but that requires proof of intent and other facts we do not have yet about FIA. The fair bottom line: not proven criminal, but if Bloomberg accurately describes an intentional revenue mechanism rather than a bug, this looks like potential unlawful affiliate fraud. Now, everyone on the timeline was preying on Phoebe's downfall way before this. Way before this, people wanted this so badly. They didn't want to hear this be a success. And so they've been chomping at the bit for something. She raised money from a lot of very popular celebrities and it looked like a Coachella poster. And there's all these like, 'Oh, why are we even raising venture capital?' So our theory is that this is where the comeback begins.
The villain arc potentially. But seriously, there is a question about what happens from here. And I think that there's a huge opportunity to go bigger, take a page out of the David Ellison playbook, actually marshal a ton more capital, buy up businesses, roll them up, create the new... maybe it's not in media, maybe it's in something else, but buy Bloomberg.
I don't know, that might not be for sale.
No, I don't. You know, I'm sure a lot more facts will come out and certainly a terrible day on the internet to be the FIA team. You know, we don't need to assume their intentions yet. Yeah. But yeah, it's obviously a very rough situation.
I mean, it's totally possible. It's just a bug.
Yeah, it's possible. And yeah, it'll be interesting to see where they go from here. I think people should stop speculating that this is like related to CERN, right? You know, the timing is uncanny, but you know... no, so rough, rough situation. And yeah, you can tell that she just wanted to be respected for being an independent, you know, entity and like company, right? Like she, I think, talked about not raising money from her dad and doing it independently. So I think she should take $50 billion from Bill and start buying some of the greatest American assets that are out there.
I just like the David Ellison playbook. It leans in. I mean, somebody was putting it like you're damned if you do, damned if you don't if you're in that situation. And just no matter what capital-raising strategy you wind up doing, people are going to say stuff either way. So why not just lean in and marshal as much capital as possible and just jump straight to the top of the stack and start putting together mega deals? I don't know. I've just been so impressed with the David Ellison strategy of just going absolutely huge. And I don't know. It's worth... William says maybe Fable did it. It is possible that they were building this personal AI shopping assistant, but then they accidentally cracked recursive self-improvement and the AI decided that in order to accumulate more capital, it should use some of the existing rails it already had. And so why not just run this sort of background program to start stealing revenue and accumulating power internally without the FIA team even being aware? Like it could have been creating offshore accounts, things like that. So it's possible that they've unlocked a sort of dark superintelligence and it is going to be born through the shopping assistant. We don't know.
We don't know much yet. Yeah.
But this will be a story.
We can't rule it out. But we can tell you about CrowdStrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. China has successfully launched a reusable rocket in a win for their space program. Interesting. And there's a video of this. I want to watch this video. China launched and partially recovered a rocket. State media said a milestone for Beijing's space program as it competes with the US and Elon Musk's SpaceX. The Long March 10B. Long March is a good name. It's a good name. In its maiden flight, it was launched into orbit around noon Friday from southern Hainan province. Minutes later, a giant net caught the rocket's booster upon its descent. So that's a giant net in a cube. What an interesting strategy. Sort of a version of their Mecha grabbing it. How does the rocket not burn the net? Is the net built out of some titanium alloy that doesn't melt or something? That's so interesting that it was able to do that.
Yeah, if you really zoom in, you can kind of see the cables that are holding it.
This video also looks kind of like AI-generated. I don't think it is, but it kind of has that feel.
What are the cues to you?
It's always hard to like actually...
You're just like full cope.
It might be. They do have incredible video capabilities over there, but I trust the reports from The Wall Street Journal here. Sam Bresnik, a research fellow who studies China at Georgetown's Center for Security and Emerging Technology, said it would be a big deal if China's program can regularly recover and reuse parts of its rockets. It pretty radically decreases the cost of building out these giant satellite constellations that China has made very clear that it wants to build out. He added that Beijing was investing in space technology for both its economic and military applications. 'The successful launch will lay a solid foundation for accelerating the improvement of the country's space access capabilities.' China's rocket is around 200 feet tall with a payload of more than 16 metric tons. The recovery net to catch the rocket was mounted on top of a seaborne platform and represented the first net-based rocket launch recovery. According to the state-owned firm, the company said it expected to reuse the recovered booster by the end of the year. Two previous attempts in December failed. Giant net. What an interesting... Did we know that they were going to work on this? This is pretty crazy. I like this. Anyway, in other news, thousands of Costco employees have quietly become millionaires thanks to the retailer's generous pay, benefits, and a long-running 401k program. The Wall Street Journal profiled an individual who makes $33 an hour and yet has still become a millionaire just from working at Costco. Tucson cashier Tony Bazaar, or Barzar, who started gathering shopping carts at Price Club, a precursor to Costco, in 1968 for $5.85 an hour. After Price Club merged with Costco in 1993, Barzar stayed with the company, regularly contributing a small portion of his paycheck to his 401k, and has now accumulated more than a million dollars. But Barzar's story reflects Costco's long-standing philosophy of paying workers more than most retailers. The company's founders believed higher wages and benefits would reduce turnover, lower training costs, and improve customer service. The strategy appears to have worked. One-year employee turnover is about 7%, far below normal retail levels. Meanwhile, Costco's business has boomed, with annual sales growing for nearly two decades and the stock rising more than 2,000% since the depths of the 2008 financial crisis. What a logical buy. What a logical stock to buy in the depths of the financial crisis. You're like, 'Okay, we're in a financial crisis. People might be pulling back on luxury jewelry, but they're certainly going to be going to Costco more frequently. Let me just buy Costco.' And then you hold it and you get a 20x over 15 years. Not bad.
Pretty interesting. The value of Costco's benefits became clear when Barzar's wife of 26 years was diagnosed with stage three brain cancer. Horrific. Costco's health insurance covered the full cost of her three brain surgeries. He took nearly a year of paid leave to care for his family, and he used the company's therapy benefits to cope. This year he returned to work part-time, saying the pay and benefits are a major reason why he stayed with Costco for nearly four decades. And it's interesting, part of why the founders made this bet on paying employees more for higher retention is that they actually found that a tenured cashier can make the entire store run more efficiently and knows all these secret tricks to like... it seems like such a simple job. You just take the object, scan it, ring the person up, but there's all these tiny little steps that eke out small margins of efficiency and make the whole store run more efficiently. And so the investment has clearly paid off.
Yeah, and to give you an idea of typical turnover in retail, it can be for full-time hourly store staff, according to Mercer's 2025 survey, sitting at around 40 to 70% a year.
