Host 1 (1:46)

I think it was potentially Amazon just turning it off for a second just to show everyone that, hey, cloud still pretty important. Yeah, forgotten about us.

Host 2 (1:55)

That makes sense.

Host 1 (1:56)

You're still dependent on us.

Host 2 (1:57)

That makes sense. Well, if you've been living under a data center. Andrej Karpathy made waves this week after he dropped a massive bombshell of a podcast with none other than Dwarkesh Patel. On the Dwarkesh Patel podcast, he said a lot of good things about AI. He said a lot of negative things about AI. He called AI slop. He also said that the models are amazing. He said that autocomplete is his sweet spot. He had a bunch of positive things to say, but overall, people did not react Positively everyone says the AI bubble is over. It's popped. You gotta rotate out of AI stocks. My take is that if you're looking for something safe, you gotta rotate into food, water, shelter and guns, because that's all you're gonna have. AI is fake. The Internet's fake, computers are fake, the steam engines fake. Railroads are fake, electricity's fake. All of industrialization is a bubble. We're going back to monkey, we're returning to monkey. We're going to be living in huts because technology is a bubble, technology is fake, and all we have left is to homestead and live in a shelter. So get ready to go back to sticks and stones. That's my take.

Host 1 (3:09)

Good take.

Host 2 (3:09)

But of course it's not that bad. But the general tech community is having whiplash from this for sure.

Host 1 (3:16)

Baze over on X says, so Happy, it's contrarian to believe in AI progress again.

Host 2 (3:21)

Yes. Oh, that's true. Yeah. Yeah, it is. It is a good thing. And so basically, Andrej Karpathy went on, and Andre is in an interesting position because Andrej Karpathy was a co founding member of OpenAI. He was there from day one. He also was at Tesla and built the self driving car program there. He is one of the most respected AI practitioners, researchers. He's implemented Nano Chat and Nano GPT, these repos that are handcrafted by him to explain to anyone who can program at a high level how to build an advanced chatbot, an advanced AI model, an LLM, train it from scratch. And so he's always been extremely respected, but he doesn't have crazy bags right now because he's kind of out of Tesla, out of OpenAI. He was at OpenAI for a little bit. He went back and then left. He's working on an education product right now, which was very, very interesting and very cool, called Eureka. I'm actually very excited for it and would recommend anyone who's getting into programming stay on top of that. Because I feel like if you can power through the Andrej Karpathy pedagogy or all of his different coursework that he puts out, you will probably have a very, very rewarding job very quickly. But people are bearish because he's basically saying that AGI is like 10 years away. Not one year, not two years. He's kind of debunking the idea, or at least he's just sharing the vibes that he doesn't feel like the AI 2027 fast takeoff scenario is probable. And what's interesting is that this aligns with what a bunch of people have been saying. But it's one like Sam Altman's been saying, curing cancer and teaching people free education. Free education, right. But then also I went back to over A year ago, September 23, 2024, Sam Altman posted a blog post, said the intelligence age. And this is what he said in there. He said, it is possible that we will have superintelligence in a few thousand days. I've always read a few is three, three or four, right? What's 3,000 days? What's 4,000 days?

Host 1 (5:31)

About 10 years.

Host 2 (5:31)

It's about 10 years. And so there's two reads on that. One is that, let's say that he knows for some reason. I don't think he does. I don't think anyone does, but let's say he knows. Well, he's still setting expectations to be like a decade away, which is exactly what Andre said. But when Sam said it, everyone's like, let's go super long. And when Andre says, oh, it's bad, which is kind of an interesting tweak. But also there is the more kind of critical take, which is just that when you, whenever you talk to a technologist and you say like, okay, you're working on getting us to Mars, you're working on fusion reactors, you're working on quantum computing. When will this actually change my life? When will this be? When will this technology be ready for primetime? Everyone says it's just a decade away because a decade is so long that if you make a decade prediction, like, but like if we were to say, you know, oh, where will the show be in a decade? Or what will we do in a decade? We could say crazy stuff because it's going to be so long that everyone will have forgotten in a decade.

Host 1 (6:25)

Basically it's somewhat straightforward to make 1, 2 year, 3 year prediction. Totally beyond that.

Host 2 (6:29)

It gets, it just gets so fuzzy. There's so many other things that could happen. It's really complicated. And I mean people are making long term predictions about AI for a long time. Called the Singularity near by Ray Kurzweil. I believe he published that in like the late 90s, maybe early 2000s. Tyler, have you read the Singularity Is Near? It's a tome.

Tyler (6:51)

I actually have not. I need to read that.

Host 2 (6:53)

Oh yeah, you should.

Tyler (6:54)

That's like a foundational text.

Host 2 (6:55)

And the crazy thing about that is that he makes a whole bunch of predictions about how scaling will happen in terms of like compute power and he bounds everything in flops. So he just says like let's sum up the total amount of computing power in the world right now and just comp that to how many flops the human brain has or all human brains have. And based on that, he predicted that we'd pass the Turing test in, like, 2022 or 2023. He, like, nailed it, like, 20 years earlier. Completely nailed it. Although we. We do debate, like, are we actually past the Turing Test? Because you can just ask a model, like, are you a model? And it'll say yes. And so. But in general, it feels like we passed the Turing Test, and it feels like we're hitting Kurzweil timelines. And so I've always kind of like. Like, leaned back. Whenever somebody presses me on my AGI timelines, I always just say, like, I kind of. I kind of stick with Kurzweil. 2045 is what he says. And that sounds bearish now, but when he published the Singularity Is near, people were like, this is never going to happen. Like, this is just insanity. Maybe in 10,000 years, bro. But, like, it's not.

Host 1 (7:53)

I thought it was interesting. You know, some people listen to the Karpathy interview and thought, Great, I have 10 years to escape the permanent undercover.

Host 2 (8:01)

Yeah, totally.

Host 1 (8:02)

Which is exactly how OpenAI is also acting.

Host 2 (8:05)

Right?

Host 1 (8:05)

They're acting like a big tech company, they're launching a social media app, they're working on ads, they're trying to scale paid users. Right? They're just acting like a traditional hyperscaler.

Host 2 (8:17)

Hyperscaler, Yeah. I think that's the correct model. I think that's 100% the right move. Did you get the time?

Tyler (8:22)

Yeah, I was just gonna say, like, the fact that 10 years AGI is, like a bearish take is crazy. That's so bullish.

Host 2 (8:29)

That's extremely bullish.

Host 1 (8:31)

Yeah.

Host 2 (8:32)

Yes, yes, yes, yes. Bearish takes are bullish. I agree. I agree. Yeah. No, no, no, it is true. But. But, I mean, a bunch of people have been saying stuff like this. I mean, when we talk to Tyler Cowen, he says, you know, I think progress will be very incremental, that we won't just see some sort of fast takeoff, because there's so many different areas of the economy that you can't. It's not just a. Like, a single input output. Text in, text out. That's true for small programming tasks, for sure. That's true for customer support, for sure. But it's not. It's not true for medicine. It's not true for so many other things. And Andre Karpathy does a great job Breaking down that prediction about radiologists, that the, the idea that radiologists would be put out of a job by AI because AI is really good at looking at images. Well, it turns out that radiologists do like 25 other things that are really important and it's not just a job of looking at a picture.

Host 1 (9:25)

And did you see a lot of pushback from anyone at the big labs on the interview?

Host 2 (9:31)

No, I didn't see anything.

Host 1 (9:33)

I expected to see more people saying Carpathy's been out of the game for a few years. He doesn't have the scale. He's not really at the frontier. He's not seeing what we're seeing. And I didn't see any of that.

Host 2 (9:45)

Just bounce off Carpathy for sure.

Host 1 (9:46)

I don't know. Is that bearish, Tyler?

Tyler (9:49)

All right, maybe that's a little bearish. I'm looking at the broad strokes here.

Host 2 (9:52)

Yeah, yeah, yeah. But I mean I think that the.

Host 1 (9:57)

I saw a lot of researchers being like, yeah, RL is pretty middle.

Host 2 (10:01)

Yep. Yeah, it's like the best. It's the best we have, I guess. But I mean the big question here is what does this mean for big tech companies? What does this mean for the hyperscalers for Oracle? Those big tech companies can't exactly turn on a dime on the basis of the Latest so funny.

Host 1 (10:18)

AWS up 1.3% today.

Host 2 (10:21)

You mean Amazon as a whole?

Host 1 (10:22)

Sorry, sorry. Amazon as a whole up 1.3% because everyone is like, oh, this is actually pretty important. We kind of forgot about us.

Host 2 (10:29)

Yep.

Host 1 (10:30)

And Oracle of course is down almost 5%.

Host 2 (10:34)

Well, whatever you are thinking of doing in the public markets, go to public.com investing for those that take it seriously. They got multi asset investing, industry leading yields. They're trusted by millions. And so Oracle's now on the hook for something like $300 billion in infrastructure over the next five years. That's a lot. Very quickly. They're hoping that OpenAI revenue will pay for that. And there's the hyperscaler narrative that you laid where if the ads product really ramps. If the soar product ramps my view.

Host 1 (11:03)

Oracle basically has to bet that OpenAI can get almost every Internet user to be an OpenAI user.

Host 2 (11:11)

Yes. And I mean I'm super fascinated in how fast the commerce ramp will happen. Just over the weekend. I mean, I'm on the furthest edge of early adoption here. And I forget what I was searching for. I was searching for a product. Oh. I was searching for gray printer paper that would look like a newspaper. So if we printed a Tweet, we could make it look like newspaper. For some reason I just wanted to know if that was the thing. It found me some options and when I got through to the product, it took me to a landing page and I wound up just sending it to one of our team members and saying like, hey, can you order this? Because the one click checkout, I wasn't at a point in ChatGPT where I could just say, okay, buy it and ship it to the alter dome. Like that was not an option for me. So I. So like they need to obviously improve like features there, but then there's also like customer adoption and actually training the user. Because if 700 million of those 800 million are still in the natural like muscle memory workflow of when I find a product on ChatGPT, I Google it and then I buy it there. Well, that's not going to monetize very well. They have to train people into thinking, yes, I trust that If I tell ChatGPT order this and deliver it to my house, it's going to buy it with the right credit card, it's going to ship, it's the right address, I'm going to be able to get a refund if I need to. That's a whole new user experience workflow that they will have to train people on. This is important.

Host 1 (12:38)

It is very notable that OpenAI has announced partnerships with Etsy and Walmart and not Amazon and ebay.

Host 2 (12:47)

Sort of the Temu Amazon and the Temu eBay. You can think of it that way. That's why Walmart is kind of like TEMU Amazon.

Host 1 (12:54)

It really is. No, but it's true. I mean obviously OpenAI I'm sure would have liked to already announce partnerships with Amazon. Would have liked to already announce partnerships with ebay.

Host 2 (13:04)

Yes.

Host 1 (13:04)

And certainly those conversations would have happened.

Host 2 (13:07)

Yes.

Host 1 (13:08)

Nobody either management team is asleep at the wheel.

Host 2 (13:11)

Yes.

Host 1 (13:11)

But I would expect that Amazon's management team, eBay's management team aren't so quick to let Sam the Fox into the henhouse.

Host 2 (13:22)

We love when foxes get in henhouses. We also love ramp.com, time is money save, both easy to use, corporate cards, bill pay, accounting and a whole lot more all in one place.

Host 1 (13:31)

Pedro.

Host 2 (13:32)

Hold on, hold on. One thing on that. So the question about ebay and Walmart, eBay and Amazon staying out of the OpenAI partnerships and OpenAI having to go to the tier twos, Walmart and Etsy. Right. I think that's like the correct framing that Amazon and ebay want to say, hey, I Think we still have leverage here, but it can flip if OpenAI becomes a true aggregator. Amazon does have to run ads on meta properties, they do have to run ads on Google. And these things can get out of control where like the iPhone got so dominant that Google had to pay Apple to be the default search. Like, there's just a certain amount of like, if everyone winds up using one thing, then you do have leverage in that conversation. And I wouldn't be surprised if in, in, if in 2, 3, 4, 5 years OpenAI has enough leverage that Amazon and, and ebay capitulate and say, yep, we're partnering because we, we, we, we just have to do it. And so I mean the last thing we could talk about is like how Oracle is positioning their decision making. I had a little bit of take on this. You want me, run me through it.

Host 1 (14:39)

Hit me.

Host 2 (14:40)

So in the. So Oracle has two new CEOs, Software Cats is out. There's Clay McGork, he's the co CEO and he did an interview with CNBC and he said of OpenAI, look at their financials, their growth and what's being built with this technology. This isn't a typical company trajectory. They've reached nearly a billion users faster than anyone. Everything about this is unprecedented, but in a good way. And I don't like this framing. I think that it's not the right framing to say like we've never seen anything like this. It's breaking all of our patterns. So we're just completely operating off playbook. I think it's much better to just say that like, like this is a new hyperscaler. We all saw what happened with Google and Amazon. They were started before the dot com bubble popped and they scaled very fast. OpenAI scaling faster on users and revenue and so we are investing more, but it's all proportionate to the growth. And so it's the exact same playbook as you want to build. Yeah, the streaming Google in 1999. You want to build Google for, you want to build infrastructure for OpenAI in 2025. You underwrite it the same way. There's just, it's just a faster growth rate. So you're investing slightly more and it's all proportional and it's actually the same playbook.

Host 1 (15:52)

Do you think he's trying to frame the growth as so unprecedented? We're just trying to meet it so that people don't see Oracle directly as just this massive bet on OpenAI. Because if he just came out and said we're betting that chat GPT will Need to serve billions of users. Yeah, that feels like just a better kind of read on the situation overall than oh, we can't predict.

Host 2 (16:14)

Yeah.

Host 1 (16:14)

And at all. It's just running away from us. We're just trying to, trying to meet their capacity demand. So I don't know. That's because either way Oracle is a bet on OpenAI.

Host 2 (16:23)

Yeah, yeah, I, I think I prefer the world where they are saying like sure, even if we're in the dot com bubble, we think we found the Amazon, we think we found the Google, they're going to make it through and these are going to be multitrillion dollar companies. So we are going really hard on that to partner and draft off of that. The problem is, is that I'm sure Oracle has a million other partners that they can't like completely like frustrate by saying, hey, you know, we're not interested. Before we move on, I want to tell you about Restream 1 livestream 30 plus destinations, multistream to reach your audience wherever they are. Sorry Jordy, continue.

Host 1 (16:57)

At the same time as Oracle is obviously bullish on OpenAI, there is reporting, there was reporting in TechCrunch that was based on a report by Apptopia that was using App Store data to look at ChatGPT overall growth and user minutes and things like that. And this somewhat aligned to the reporting from Deutsche bank that.

Host 2 (17:25)

We got Nick on the camera somehow. Sorry everyone.

Host 1 (17:29)

AWS goes down, doesn't go down entirely, but it's closed.

Host 2 (17:34)

But welcome and please enjoy the beautiful view of one of our team members, Nick.

Host 1 (17:40)

So anyways, hopefully the audio is still coming through but we're back. Both of these reports showed that, yeah, OpenAI, OpenAI growth, ChatGPT growth is seemingly slowing a little bit and that makes sense if you just based entirely on the scale that they're at right now.

Host 2 (18:04)

Yeah, I mean you do saturate at some point. This is DAU is this, this is global DAU and it's around 72 million. What's interesting here is that for a long time people have been spinning the narrative that like oh, growth will plateau during the summer. Right. Didn't we look at some data that showed that August was particularly flat for some other reason? Who was that?

Host 1 (18:28)

Growth did plateau over summer at least.

Host 2 (18:30)

That'S not this data. At least this data shows very consistent. Oh, that was Europe. That was Europe. Okay. Yeah, it's interesting.

Host 1 (18:38)

Europe.

Host 2 (18:39)

America grinds during the summer and then plateaus and yeah, we burned out. The great lock in of September was like, okay, I'm not installing any new apps or something like that. I don't know, I don't know how much to read into this apptopia. Like some of these app tracking things like I don't know how accurate they are because they don't necessarily have a pixel in the exact app. It might be more like Nielsen ratings where they're just polling people. But I mean if you believe the beginning of the curve, what happened last year because there was actually a plateau back in it looks like back in January of What was that? 2023 or 2024? Because January. Oh no, that was the beginning of this year. That was the beginning of this year. So the beginning of this year was pretty flat. And then once April 1 came out, everyone went on for April Fool's day I guess and then it started ripping upwards and then, and then, and then September and October have been sort of flat. I don't know. We'll see. There was. Yeah, there's a lot more we could go through here. But first let me tell you about Privy Wallet infrastructure for every bank. Privy makes it easy to build on crypto rail, securely spin up white label wallets, sign transactions and integrate on chain infrastructure all through one simple API.

Host 1 (19:56)

Buco Capital was sharing another data point over from similar web. It shows that Gemini has been gaining market share from gaining on OpenAI. Obviously the overall market is growing, so OpenAI is still growing on a per user basis, but Gemini is growing and he says Google already has nine products with over 1 billion users. Search, Android, Chrome, Drive Photos, Play Store, YouTube, Maps and Gmail. So says they know a thing or two about a thing or two.

Host 2 (20:27)

Yeah. At the same time bootstrapping a new product flow, like actually funneling you from Gmail into something else is really hard for almost 30 year old company. It's always been through acquisition they buy something.

Host 1 (20:44)

I mean Meta and Threads.

Host 2 (20:45)

Meta did do it with threads.

Host 1 (20:46)

So maybe when you own one of the best advertising networks in the world, you can aggressively funnel people. The question is how much do they try to integrate the value of Gemini into the core search product?

Host 2 (20:58)

Yeah. How would you do it? Close them out? I think what Connor did over at Threads is a really, really smart integration where you actually see threads content inserted into the Instagram feed. So while you're scrolling Instagram you see a thread and you're like oh, I want to see that. And then it just takes you over to that app and then your dau of that app. And it was very easy to create a Threads account if you had an instagram account. How. How should Google Surface a Gemini flow where you're in Gmail or you're in Google Search, and then all of a sudden you're in Gemini?

Host 1 (21:38)

I think there's a bunch of ways you can do it through search. You can do it through Chrome, maybe.

Host 2 (21:42)

Through the AI overviews. Like, you know how you go into AI mode. It should be like one more click and then you're just fully in Gemini. Maybe.

Host 1 (21:49)

Yeah. I mean, I think there's a reason that OpenAI wants its own browser. Perplexity wants its own browser. I don't think they've properly even leveraged Chrome yet in order to drive Gemini usage.

Host 2 (22:00)

Yeah, yeah, yeah. I was thinking about. I was thinking about the browser, like the agents in a browser.

Host 1 (22:05)

Browser wars. Of the browser wars. 20, 25, over, over. Before they started.

Host 2 (22:11)

Yeah, with a whimper. Yeah, I was thinking about the browser because I used to. I used to enjoy scraping web content. So I'd go to a website and I'd write some JavaScript to, like, pull it all out into a spreadsheet. And I was thinking like, o with an agent in my browser, I could do that really easily. I could have it write the JavaScript to scrape everything out. That'd be really easy. But then I realized I could just copy paste the whole thing into an LLM or just give the LLM the URL or even just. I was thinking about the Midas list. I was like, if I wanted to get the Midas list into a spreadsheet, I used to have to write JavaScript to go through the Midas list and save everything and put it in a spreadsheet. Now you can just go to ChatGPT and say, reproduce the full Midus list and it'll just do it. You don't even have to tell it the URL. So I'm still searching for use cases of AI in the browser.

Host 1 (22:59)

This was a good highlight from the interview. Kurt Path. He says, when I go on my Twitter timeline, I see all this stuff that makes no sense to me, honestly. Just fundraising. And then the next post from Chris Offner, the unbridled joy of listening to someone smart who's not trying to sell you anything. This is why I think this interview hit so hard, is he's not sitting there, like, trying to get to the next tender offer. Right. He's just being honest about what he's seeing. And some people aren't gonna like that.

Host 2 (23:30)

But, I mean, same with Sutton. Like, Sutton doesn't have bags, and he was able to come on and make A pretty. Pretty clear argument to put Tyler out of a job.

Tyler (23:42)

I mean, so I do think the most interesting quote out of this. I don't see a ton of people talking about it. It's that he basically doesn't think, like, even if we get AGI in 10 years, GDP growth will actually just stay at 2%, which I think is, like, very contrarian. It seems to me that, like, most people are like, we won't get AGI if they're saying that gdp, gdp, GDP growth is going to stay, that means we didn't get AGI and it's just like, better autocomplete or et cetera. But he basically loops in, like, all, you know, innovation just into the same 2% GDP growth that's been going on. Because if you look back, you can't really see.

Host 2 (24:20)

Yeah, that was a super profound idea. This idea that he sees AI as an extension of the Internet. And he's like, yeah, when I hear the story about, like, recursive AI, like, oh, wow, the Foundation Model Lab AI researchers are using AI to improve the AI that they build. He was like, well, yeah, like, the Google engineers would Google stuff. Like, yeah. And like, you know, the Ford. The Ford Motor manufacturers would probably be like, cars, let's drive cars.

Host 1 (24:47)

Parts to make more cars.

Host 2 (24:48)

Exactly. So, like, he was like. He was kind of like every technological innovation since we've invented technology and started.

Host 1 (24:55)

Yeah. And all you have to do is look at OpenAI, and fundamentally, it looks a lot like Google. Yeah. It has different vibes, different. A different leadership style. But, yeah, you know, they have their YouTube, they're building their AdSense.

Host 2 (25:09)

Yeah.

Host 1 (25:10)

So.

Host 2 (25:10)

Well, if you're writing code, you need to check out cognition. They're the makers of Devin, the AI software engineer. Devin, crush your backlog with your personal AI engineering team. Everyone's from Canada, says Yaxin. Basically every famous computer scientist is Canadian. Andre Karpathy. Hinton. Alex Net. Rob Pike. Yaxine Yassin, Elon Grimes. Elon, Canadian. Grimes. All Canadian. I love it.

Host 1 (25:40)

Everyone likes fake news. Will Brown, thoughts on the Carpathia interview. Agree with him on pretty much everything. Literal transformative AGI is not imminent. That's fine. Doesn't mean it's a bubble.

Ryan Cohen (25:51)

Doesn't mean it's a bubble.

Host 1 (25:52)

Let's go hit that gong. Normal technology that improves productivity and growth is very valuable. The market already reflects this.

Host 2 (26:01)

That's great. Nothing ever happens and everything is fine. And we're so back and it's already priced in the exponential RSI takeoff.

Host 1 (26:11)

Famous last words. Suspended Cap says the Karpathy interview is scaling. We aren't sure compute scathing. We aren't short compute. We are short breakthroughs that replicate things the human brain can do. No doubt we'll get there, but wouldn't be surprised to see market sputter on Monday. While this makes the rounds. Really honest stuff.

Host 2 (26:33)

No one in the market is talking about this. The private markets and the public markets are fully disconnected. The public markets, the Nasdaq is up 1.46%. The vibes on X are way different than what happens on Wall Street. I remember when Nat Friedman and Daniel Gross went on Stratecherry and they did a great interview series about AI back in 2022, 2023. There's a whole series of them and in there they talk about Nvidia specifically and how important Nvidia chips are going to be. And Nvidia was like a $300 billion company or something at the time. And it like almost like, I don't know, 3x or 10x maybe it was like, maybe it's 8.

Host 1 (27:14)

Okay. To be fair, JP Morgan did downgrade Oracle's credit rating today on the Carpath epo.

Host 2 (27:21)

You think that's what happened? They were like, oh, they're watching DoorDash. They're like, we got to downgrade this. This is wild. Do you know who Conor Leahy is?

Host 1 (27:30)

Yeah.

Tyler (27:31)

So he collapsed a famous. I think he's a big doomer.

Host 2 (27:34)

He's a doomer, right? Yeah, so he says. I want to pre register the following opinion. I think it's plausible but by no means guaranteed that we could see a massive financial crisis or bubble pop affecting AI in the next year. I expect if this happens it will mostly be for mundane economic reasons. Overleveraged markets, financial policy of major nations, mistiming of bets even by a small amount of and good old fraud. Not because the technology isn't making rapid progress. I suspect such a crisis to have at most modest effects on timelines to existential dangerous ASI being developed but will be used by partisans to try and dismiss the risk. Sadly, a bunch of people making poorly thought through leveraged bets on the market tells you little about the underlying object reality of how powerful AI is or soon will be. Do not be fooled by narrative. So he's saying yes, we got to go even harder. The bear market's bullish for doom potentially underrated AI doom scenario. The bubble pops so hard that everyone goes bankrupt. There's mass famine and starvation and humanity population collapses all because we went to leverage long AI what do you think about that? Plausible? It just might be. Matt Palmer says I've been saying this. Update your models and timelines. Foom cells. Mankind is back. Mank back and Figma is back. Think bigger, build faster. Figma helps design and development teams build great products together. How it feels listening to this podcast. Han Wang can we pull up the picture? It's balloon popping. We will see about that. Who knows how this will all play out.

Host 1 (29:13)

Francois Chalet says the reason it is so important for everyone to keep pretending that AGI is definitely right around the corner is that there is now over 1 trillion of investment riding on this belief either already expended or committed. Current and recent past capex cannot be justified by current use cases and technology. Currently spending 10 to $15 to make $1 to ever be in the black you need dramatically better tech applications and you need them fast before current data centers depreciate which is a three to five year time scale. Gnome Brown was calling out said, didn't you just say two months ago that.

Host 2 (29:50)

Gnome's at OpenAI.

Host 1 (29:53)

That you think AGI is about five years away? Of course, yes.

Host 2 (29:57)

And Chase Coleman, the founder of Tiger Global came out of hiding on the same weekend. Karpathy said, AI Labs are promoting slop as innovation to secure funding and we're decades away from AGI. Sell it all today.

Host 1 (30:11)

Bukha Karl it was a good panel. Listen, Brad Gerstner I should have thrown.

Host 2 (30:18)

That one in the mix. I listened to the full Palmer Lucky on Joe Rogan episode was a lot of fun. Palmer will be on the show tomorrow and I listened to the Carpathy episode which was great. Oh, another interesting Carpathy tidbit. He invented the term hallucination. He hallucinated calling what LLMs do hallucination, which is very funny. I just thought that was really cool. Anyway, we should move to the Napoleon jewels that were stolen from the Louvre in a major robbery. The thieves stole jewelry from display cases in Galerie de Paulion.

Host 1 (30:59)

Finally a brazen heist.

Host 2 (31:01)

Finally a brazen heist.

Host 1 (31:02)

There's been little or no brazen heists.

Host 2 (31:05)

If you're new here on this show. We are anti heist. We are anti crime. Do not do this. If you have information that could lead to the arrest of these thieves, let us know. We will send you a hat. That's the best we can do. But Luke Metro says Is this the cure for male loneliness? I hope not, because we don't want more brazen heists. We want less brazen heists. Leave the Crown Jewels alone.

Host 1 (31:27)

A truck Mounted furniture elevator to reach the gallery housing the artifacts and then sped away on motorcycles. Officials said tourists were streaming into the world's most visited museum on Sunday when a group of thieves burst in through window of a gilded gallery on the second floor, second floor and made off with a set of priceless royal jewels. It was broad daylight, roughly 9:30 local time, when four individuals driving two powerful motorcycles in a truck with a portable furniture elevator parked outside the Louvre museum.

Host 2 (31:55)

I didn't know what a furniture elevator was. Have you ever heard of this before? It seems like actually a really awesome invention because if you need to get a couch to the top of like your house and you can't take it up the stairs, you just put it on this. It looks like a fire truck. Basically the fire truck pulls the ladder up and then the elevator slides up. But they move the funny.

Host 1 (32:13)

When I saw the picture of the furniture elevator outside the museum, it looks like just kind of makes sense.

Host 2 (32:20)

Oh yeah.

Host 1 (32:20)

Normally when you see something like this, you're not thinking to yourself that's out of place.

Host 2 (32:23)

Yeah, no, no, no way.

Host 1 (32:24)

There's maybe one person that worked out at the museum that would think that doesn't seem scheduled.

Host 2 (32:29)

Yeah, totally. It happened during normal hours. The entire heist was less than seven minutes.

Host 1 (32:35)

So two of the perpetrators, at least one wearing a high visibility yellow vest.

Host 2 (32:39)

That looks blend in so much. It just looks like workers, like, oh, they're going to go work on the balcony equipment. Like really the value of the professional equipment. But there was a crown angle grinders.

Host 1 (32:52)

To cut through the window and get inside.

Host 2 (32:55)

They attempted but failed to set fire to their truck. So they were trying to burn the truck on the way out to get rid of all the fingerprints, I suppose. And they dropped the crown of Empress Eugene with nearly 1400 diamonds before they sped away. It was found damaged, no one was injured. Eight pieces in total were stolen, including an emerald earring and an emerald necklace that belonged to Empress Marie.

Host 1 (33:17)

Such a brutal fumble.

Host 2 (33:20)

That is a fumble of 1400 diamonds. Massive fumble, absolutely brutal. An emerald ring and an emerald necklace. Some of these things are massive. I didn't know this, but the Wall Street Journal takes us through some history of burglaries. There's been a long string of very history.

Host 1 (33:40)

This is a very French, extremely French.

Host 2 (33:43)

Leonardo da Vinci's Mona Lisa was stolen from the Louvre in 1911 by an Italian carpenter, only to be found two years later. He got away with it for two years. In 2010, a thief broke into Paris Modern Art Museum and made off with more than $120 million of artworks, including some from Picasso and Matisse. A thief. The thief, dubbed Spider man, was sent to prison for the crime and later appeared in a Netflix documentary. What do you think they're going to do with the stolen jewels? We were debating this at breakfast. It's going to be hard to fence this. I think that's the term. You can't just go on ebay and throw it up there.

Host 1 (34:18)

Facebook Marketplace.

Host 2 (34:20)

Facebook Marketplace was built.

Host 1 (34:21)

No? Yeah. It's an interesting challenge because obviously they're never going to be able to capture the same value necessarily as if you were able to do an open auction. If you're able to do an open auction and you could buy this legally and the person that bought it could then go flex it around and publicly be the owner of these incredible historic jewels. That would get you sort of the maximum price. That being said, given the historical significance, I think there's a small pool of buyers, there's probably 100 people in the world that would be willing to pay tens of millions, potentially more, just to know that they had it. And just to be a family secret.

Host 2 (35:09)

Do you think there are any buyers that would flex it? Because obviously if you're an ally of France, you're not going to do that, obviously, because they'll immediately come and be like, those are stolen. Give it to us. Doesn't matter what you paid, you got to return it. If you're America or Germany or Britain. Right. But what if you're Vladimir Putin? What if you're Kim Jong Un? Like if you're in North Korea, like France is going to invade North Korea for a couple rules. Kim Jong Un just rocking it himself. I wouldn't be surprised. I don't know. I feel like there are some countries that are actively not friendly with France that could just say, you know what, we're going to buy this and we're.

Host 1 (35:48)

Going to rock it. We're just going to rock you.

Host 2 (35:49)

It's going to be in our museum now. It's a stolen artifact. A lot of museums have stolen artifacts.

Tyler (35:54)

Yeah.

Host 1 (35:55)

The other play is to, you know, if you're a family that does, you know, is into long term planning, you could say, all right, we're going to store these for literally 150 years as a family. And once 150 years pass, we're going to be able to pop up and we're going to be able to talk about our jewels publicly because everybody that was around when they were stolen isn't here anymore. Nobody's really going to care. It's not going to. We're going to be. The entire world population will have turned over and we'll be able to suddenly come out and make up some story of how our family got them.

Host 2 (36:31)

Yep. We need to up our security because I feel like someone's going to steal that horse if we're not careful.

Host 1 (36:38)

They're going to use a furniture elevator to get out of here.

Host 2 (36:40)

But if you want to manage risk in your business, go to vanta.com, automate compliance, manage risk, prove trust continuously. Vanta's trust management platform takes the manual work out of security and compliance process and replaces it with continuous automation.

Host 1 (36:52)

Okay, this story from Friday. Saturday.

Host 2 (36:55)

Yes.

Host 1 (36:57)

Cracking up.

Host 2 (36:57)

Tell me.

Host 1 (36:58)

So, long story short, some OpenAI researchers were able to use ChatGPT to find the results to a handful of problems.

Host 2 (37:08)

Yes.

Host 1 (37:09)

That were previously marked as unsolved. What are they called? The error Erds.

Host 2 (37:15)

It's from Paul Erds, Paul Erdish. He's a mathematician.

Host 1 (37:19)

And so they were able to. They said, update. Matab and I pushed further on this. Using thousands of GPT5 queries, we found solutions to 10 Erdos problems that were listed as open. And so they put this out and it looked like ChatGPT had made this incredible novel discovery. And it turns out they were just using ChatGPT as this effectively web search, knowledge retrieval. Knowledge retrieval.

Host 2 (37:41)

So someone had put these up there. Yeah. Okay.

Host 1 (37:44)

And they were kind of positioning it. You could read it both ways. If you went back and looked at their original post, you could see how they were saying, we just found the solutions, we didn't generate them. But the way that it was written was definitely like a lot of people would have just read it as, oh, wow, ChatGPT5 is doing novel discovery.

Host 2 (38:09)

I have a very funny story about this.

Host 1 (38:10)

And so of course, just to close it off, demis over at DeepMind just commented, this is embarrassed.

Host 2 (38:19)

Absolutely roasted. And the OpenAI researcher said, I deleted the post. I didn't mean to mislead anyone. Obviously, I thought the phrasing was clear. Sorry about that. Only solutions in the literature were found. That's it. And I find this very accelerating because I know how hard it is to search the literature. Don Lecun, Don Lacuna is not happy.

Host 1 (38:40)

We're not gonna read that one.

Host 2 (38:42)

He's making a play on. Hosted by their own petards. Which is where you are a victim of something that you. It's like stepping on a rake that you put down on the. He is not making an allusion to The R word or any sort of slur. So just to be clear there. Anyway, I had a funny story when I was in Silicon Valley in 2012, there were these, like, extremely hard programming challenges that I could do the first couple of, but I couldn't actually get that far because they were really, really difficult. And I didn't really have a reason to like go and get, like be like a competition programmer. I was just trying to build a company and. But I wanted, if you, if you could get. But if you could get the service to accept your result, you could get this badge that you could put on your website and it would link it into your profile and it would be very legit. It was like programmer cred. What I figured out was that there were a lot of answers out there that were on GitHub and open source. So you could go and find those, put those into the system, get authenticated as having solved these really hard problems. I got the gold medal and was able to put it on my website and made it look very official, but I could not do that level of programming. So I was doing the same thing basically. So, you know, maybe I shouldn't be casting stones here.

Host 1 (40:01)

Anyway, Elon.

Host 2 (40:03)

Oh, yeah, he's taking shots.

Host 1 (40:05)

Dunked on. Well, gabe over at OpenAI. Well, to start it off, Elon says my estimate on the probability of Grok 5 achieving AGI is now at 10% and rising. Gabe says 10% chance Elon declares he reached AGI of 4. Fourth time Elon.

Host 2 (40:21)

Has Elon ever actually declared he reached AGI? I feel like he has not declared that. Like, I feel like Elon says a lot of crazy stuff about like, I hope that we'll be able to do novel physics. I think the next version is going to be really powerful, this and that. But I don't know that he's ever actually called a version of Grok AGI. But anyway, clearly it clearly upset Elon because he replied and said, you call yourself a quote unquote researcher? Hitting somebody with the double quotes, we know with the pain that the, that inflicts, it's terrible. You never want to be hit with the double quotes. And he says, pathetic. And so rough, rough day on the timeline. Timeline in turmoil.

Host 1 (41:03)

Jensen Huang was at a Citadel event, future of global markets, and said, we are 100% out of China. We went from 95% market share to 0%. I can't imagine any policy maker thinking that's a good idea. And yeah, there was some pushback here. Matt Palmer says 100% out of China, but doing astonishingly healthy business in Singapore and Malaysia. Totally out of the blue and not for any intelligible reason. Swinging around accusations that Singapore and Malaysia are fronts for Chinese chip demand.

Host 2 (41:43)

Yeah, people have been going back and forth on Jensen's involvement in China. Shamsankar took to the pages of the Wall Street Journal. I'm sure we have a copy here actually talking about Nvidia in China. We actually have some folks coming on the show later today, hopefully to break down some of what's happening both in the rare earth world and also in the semiconductor world with regard to China. If you are joining us and expecting an interview with Ryan Cohen, we are working through some technical difficulties. We are hopefully going to be getting him on the phone in just a few minutes. But first let me tell you about graphite.dev, code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster, Palantir CEO Shanti said. He said why the China Doves are Wrong American business leaders cozying up to Beijing refused to see that the Communist Party wants wants us to fail. China's Commerce Ministry last week announced far reaching export controls on lithium batteries, products that use Chinese rare earth materials, and related technologies. The export controls, which President Trump characterized as a rather sinister and hostile move, are the latest reminder that the US Is funding its own destruction through economic dependence on a communist adversary. Many American business elites persist in denying this reality. Jensen Huang is just one of them. He said in a recent interview that while some Americans wear the label China Hawk as a badge of honor, it is really a badge of shame. The future, Mr. Wong says, doesn't have to be all us or them. It could be us and them. A nice sentiment, but the CCP leaders don't believe it. They often speak soothingly of their country's peaceful rise. But the party's history and actions tell a different story. Influenced by the Chinese civil war and much earlier Warring States period, the party believes that stability comes from control. This belief explains its ruthless effort to consolidate power. The Communist Party believes China and the US Are locked in a great struggle for mastery. In this worldview, it is not enough for China to rise. The US Must fall. It is a dark telling of the current state of affairs. It is a rough time in geopolitics. He closes by saying, the first step to ending our dependence on China is admitting we have a problem. We can continue as useful idiots decrying China hawks who point out that we're funding our own demise, or we can wake up to the reality that we're already in an economic war in which every purchase and investment will help determine which system survives. Very, very rough. I mean a lot of this is all teeing up a major trade deal and I mean we're starting to see the outlines of exactly how big the chip stacks are on either side of the poker table. If you want to use that analogy. It is a, the rare earth equation is certainly a big one. America has a bunch of levers to pull, but the chip control is one. It feels less, it feels like a smaller chip stack, honestly. Because even when Nvidia says hey, okay, we're going to sell H20s in China, China says, ah, we don't want them, we'll just stick to Huawei. We will wait, we will be a little bit later on the slop curve on the sloppification of our economy and we'll see you in 10 years when we're actually fighting it out for AGI. And at that point we will have our, our own full semiconductor stack on the frontier stack. Yeah, I mean that feels like what they're thinking of. I don't know what their interpretation of the Andre Karvathy interview was, but it seems like they are there. If they were truly like the next scaling run is what does it. They would be like, oh, let's take the Nvidia chips as well. Instead they're saying, okay, well let's actually gear up for independence over a long period of time. Wall Street Engine has a story here. JP Morgan says Nvidia CEO Jensen Huang's projection of AI capex growing from 600 billion today to 3 to 4 trillion by 2030 is financially feasible. Though ambitious, the bank expects the tech sector to fund it through operating cash flow, private equity and venture capital inflows and new debt issuance which we've been tracking. Even With a projected 1.6 trillion annual funding gap, private markets could contribute about 500 billion per year by 2030, leaving the rest to be covered by leverage expansion. JP Morgan estimates 40% of new debt, 430 billion would come from bank loans and 60% from bond issuance. So there's going to be a new 1 trillion of debt coming into the system. Now even with this increase, the tech industry net debt to cash flow ratio would rise only from 0.7 to 1.2x, still below the global average. So that's the bear case for like everyone's investing a ton, but we're not completely over our skis. It is a very aggressive build out. It might take a hit on depreciation or it might take time to really monetize all this.

Host 1 (47:02)

But yeah, and the key question is if you want AI CapEx to scale to 3 to 4 trillion, you're going to need revenue to support that. You're going to need a lot of revenue fast. We've seen tremendous growth on the agent side, developer agents, whether that's Claude Code, GitHub, Copilot, etc. Etc. But yeah, it's not that it's just financially feasible, it's just that it has to be backed up by something. And thank you for bearing with us everyone. We are getting Ryan on in any second. Yes, AWS being down is presenting some technical difficulties, but the interview is still happening.

Host 2 (47:49)

Yes, we're working on it. Thank you for sticking with us and thank you for checking out Julius AI, the AI data analyst for everyone. Chat with your data and get expert level insights in seconds. And that's Fin AI, the number one AI agent for customer service, which is also a sponsor of tdpn. So thank you to them. Dean Ball, who is coming on the show today. He says, I have been contacted by a person clearly undergoing LLM psychosis, reaching out because 4o told them to contact me specifically. I have heard other writers say the same thing. I don't know how widespread it is, but it's clearly a real thing. Jordi, have you received any, any inbound from random people who seem to have GPT psychosis?

Host 1 (48:34)

Hard to say.

Host 2 (48:35)

I feel like my entire like unread like DM requests tab is probably like LLM psychos, but those folks don't break through. And honestly, like the bot slop has been so bad for so long that.

Host 1 (48:49)

It'S hard to tell what, it's hard to tell what's actually Emmett Shear last week was being accused of having LLM psychosis. He said if you're trying to understand AGI corec concepts, you should familiarize yourself with thermal time. Quantum Darwinism, quantum Bayesianism, the free energy principle, the theta function integral theory, the Yolanda lab. I think he was, I think, I think he was joking here. Yes, but, but we've talked to him. Many people did not, did not take it that way.

Host 2 (49:24)

Yes. And it kind of spread around. Like lots of people have LLM psychosis and lots of people have LLM psychosis psychosis too, where anyone who's using an EM dash is seen as someone who has no brain left because they've been destroyed. And that's also not true. Yeah, Emmett went on a tear like this a couple Months ago. We had him on the show after that and he seemed fine. He was like completely normal on the chat and was clearly just having fun. And I think he's just vague posting. I mean, look at his favicon. He clearly likes to post about weird and esoteric things and just have a good time.

Host 1 (50:03)

Mike Isaac is highlighting. There's a company called Aerie is the first big company that this woman Rachel has seen that formally states they won't use AI generated imagery. It's one of their most engaged posts with 33,000 likes. Mike Isaac says, I do think none of what we do involves AI. Increasingly becomes a marketing thing over the next few years as we hit saturation. Smart posturing, in my opinion. So, yeah, aries saying, In 2014, we stopped retouching. Today we commit no AI generated bodies where people, real people only. We believe transparency isn't a trend. It's our promise to you. No retouching, no AI. 100% real.

Host 2 (50:46)

Interesting. Do you know this brand, Aerie? Are you familiar with this brand? Before this? I mean, good way to go viral and stuff. But yeah, I don't know. I wonder where this goes, because in Hollywood, there's been a big pitch for this movie. Doesn't have any cgi. We shot it all practically. And then what that really means is that, okay, they used miniatures, but they actually did a bunch of green screen stuff and they did some set extensions. And yeah, There were some 3D models over here and there, but it was mostly in the background. And it's just not. Not a crazy CGI explosion. Doesn't look like Transformers, but if you just look through the credits on one of those movies, Christopher Nolan's famous for this saying, we shot everything practically or we really steered away from cgi. If you look through the credits, you'll see a bunch of 3D artists because they're clearly doing some 3D reconstruction of certain things. But it might just be, oh, there was a boom pole in this shot, or, oh, there was a light stand in the back. Or, oh, you could actually see a taxi cab. And we were shooting in the 1800s, so we got to pull that out. So just go in there with the pen tool and replace it with AI. Anyway, we have our first guest of the show. We have ryan Cohen, the CEO of GameStop. Can you hear me, Ryan? How are you doing?

Ryan Cohen (52:10)

I can hear you.

Host 2 (52:11)

Okay. Sorry, we have serious, serious technical difficulties today, but I believe we have you here on the phone. Thank you so much for joining. Thank you so much for taking the time. How Are you today?

Ryan Cohen (52:23)

I'm doing well. How about yourself?

Host 2 (52:26)

We're doing great. We'd love to just get the story of GameStop from your perspective, kind of how you tell it. Now you're a couple years into this project. We'd love to kind of get the high level, just to set the stage.

Ryan Cohen (52:42)

Where do you want to start?

Host 2 (52:45)

Maybe the original idea, like, when did you think you'd become involved with this company?

Ryan Cohen (52:52)

I originally invested as a passive investor. I had a few conversations with the management team and the board of directors, and I realized that saw a lot of stuff that you read about, but you don't really believe until you actually engage with them. And as we went through conversations, I went from passive to active and filed the 13D, I believe, in August of 2020. And then I joined the board in January of 2021. And the rest is history. The long story short is there's been a lot of changes at the company. A lot of cost cutting, a lot of rationalization focusing on the basics of running a profitable business, and a lot of it was getting people out of the way. A board of directors that had perverse incentives, a management team that had perverse incentives. Ultimately what you see is that when people don't have their own money on the line, they don't give a shit and they're focused on all kinds of other stuff that ultimately doesn't matter to shareholders. And a lot of it was getting that out of the way so that we could actually focus on the business. And you see it in the results of the business today versus where it was not just when I joined the board, but you look at the results even prior to that. It's a tough business.

Host 2 (54:39)

Yeah, totally. What do you think the mainstream media gets wrong about the GameStop story today or has gotten wrong throughout the whole saga?

Ryan Cohen (54:58)

That's not an easy question to answer because we're talking in generalities and I don't want to stereotype. I can't stereotype in general. Everyone thinks they know. The mainstream media thinks they know I'm running the business, and they seem to have greater visibility into the business than I do. So everyone has an opinion, as opposed to actually just focusing on the results. And in general, the news should just focus on the results instead of comments from the peanut gap.

Host 1 (55:32)

Yeah. Let's talk about Q2 specifically. What made that one of the biggest quarters in years?

Ryan Cohen (55:42)

It's a focus on running a business like an owner and cutting costs and making money and leaning into the areas of the business where there's Margin potential, just trading cards and then getting rid of the bloat. As an example, I was just looking at this earlier today before we spoke 2021 there were 14 over 1400 people in corporate. Today there's like 400 people.

Host 2 (56:18)

Wow. Yeah.

Ryan Cohen (56:19)

And we're more productive today than we were in 2021. SG&A has come down by like 50%. So people build teams, they hire people. Ultimately what that means is they delegate their work to someone else. You end up taking on this crazy cost structure. And it works when the business is growing, but doesn't work well once the business stops growing. And that's where GameStop was. And physical retail is tough.

Host 2 (56:50)

Yep. How much of that corporate restructuring going from I think you said 1400 people down to 400. How much of that is just reorienting around a players aligned incentives, picking the best person to actually run a specific initiative versus using technology, using software, using AI. Is there any sort of narrative around that that you've found success with with?

Ryan Cohen (57:17)

It's both even finding the right people though.

Host 2 (57:22)

Yeah.

Ryan Cohen (57:22)

You know, you don't, you don't know. I mean I've been interviewing people for a long time and you meet people, I meet them and I get really excited because they're really good at the interview process and they know how to say the right thing. So what does that mean? Means they've got really good public speaking skills. So they over index there. And then you actually, when you look at their execution skills, they're not great. So I've been as equally excited as people that I've hired that have worked out as I have been of people that I've hired that I wasn't excited about. And then they proved to me that they can actually execute. So you know, the results speak for themselves. You really only know once you put people, someone in the position and you see what they're capable of. They can tell a great story, they can put together a great PowerPoint presentation, but you don't, you don't actually know until you see what they're, what they can actually do. So there's no question that artificial intelligence and just broadly speaking technology has increased productivity. You know, that's, that's been a big benefit. And then doing more with less.

Host 2 (58:39)

Yep.

Host 1 (58:40)

You take a $0 salary, you have billions of dollars of cash on hand. How do you plan to allocate it over the next two to three years? I don't want to, I don't want to go too long term.

Ryan Cohen (58:56)

There's, we don't have a gun to our head. So it needs to be a situation where our the downside is limited and the upside is really high.

Host 1 (59:11)

And.

Ryan Cohen (59:14)

That'S a different calculus than the world of private equity or venture capital or any money managers where they're incentivized to deploy capital because they get management fees. In this case, a fancy way of saying is it's risk adjusted. I don't want to lose money and I want a situation where there's a good chance of making money and a really low chance of losing money. So it needs to be a pitch that is pretty much down the middle.

Host 1 (59:49)

Does that mean you're waiting, you're waiting for a crash?

Ryan Cohen (59:59)

You never know what's going to happen in the financial markets. They can go from green to red and they don't flash yellow. So when that happens, we'll be in a position.

Host 1 (60:11)

Yeah. I just feel like it's notable that you have all these digital asset treasury companies that are just market buying obscene amounts of bitcoin and other tokens and you guys are taking a longer view.

Ryan Cohen (60:29)

Yeah, definitely.

Host 2 (60:32)

What are you excited about in the core business over the next few years between retail online? You mentioned trading cards. Can you unpack a little bit more about some of the key initiatives like the customer behaviors that you see as really big opportunities?

Ryan Cohen (60:51)

I tried a lot of different things. So I originally went in with the the Chewy Playbook was we focused first on consumables. So we had a lot of success on pet food treats, litter, things that you could put on auto ship. And then it went from focus on consumables, getting customers on auto ship to we're going to be the everything pet store and we're going to expand our catalog and we're going to add all these hard goods. So I had all these preconceived biases where I was going to copy the Chewy Playbook at Gamestop and basically be like the everything store for gaming. And we, I hired a bunch of fancy people from both Amazon and Chewy and we expanded the catalog and we added a bunch of product and most of that product didn't sell. And we ended up marking it down and taking a big hit. And it cost shareholders a lot of money because once you have the product trapped in stores, you got to mark it down in order to move and get it out of the stores. Whereas within the consumable space, if we ever bought it, if we ever overbought inventory, we just waited and ultimately bought too much cat food or whatever it ended up selling. So physical retail is Very, very, very different than E Commerce. And I spent a lot of money to figure that out. And so what I learned is that we went into all these categories and a lot of we took some significant hits on losing money. We lost money. It's that simple. Bytron expanded all of these categories that were not core to the GameStop customer. And then along comes collectibles. And obviously after expanding into categories where there's very little success, your risk appetite at that point is pretty low. All of a sudden we see that they're like the GameStop customer. Really when it comes to trading cards. There's a strong appetite for trading cards. And that category has done very well. And we've gone from like 10% of our sales to over close to a third, probably for the full year is going to come from collectibles. And that's remarkable.

Host 2 (63:28)

Congratulations.

Ryan Cohen (63:29)

Yeah.

Host 2 (63:31)

I want to know more about the collectibles thing. Let's table that for a second. I'd love to know more about where you see the traditional video gaming market going. I'm experiencing kind of whiplash because I see mobile games and free to play growing and micropayments happening. EA is getting taken private. You have a lot of stuff going on on that end. But then you also have Palmer Luckey kind of bringing back the Game boy and the N64 with chromatic and I know you're partnering on that. What are you excited about? What is the shape of the traditional sort of like video game market look like over the next couple years?

Ryan Cohen (64:12)

The video game market is definitely going from physical to digital. So our ability to play in the digital world is limited. There's a lot of money that's being spent and we've taken a capex light approach, like a, a pretty risk light approach to the digital world. If there's not a clear path, us being able to make money and a payback period that's pretty attractive, then. This isn't a story of moonshots. There's no moonshots that are being done. We're focused on real returns.

Host 2 (64:57)

Yep. What about wearables, virtual reality? We heard a story about the meta quest or the meta Ray Ban displays and a friend of ours went and had to test them at Best Buy and didn't have a great experience. And it feels like if we do enter sort of like a wearable era, there's a renewed demand for in store experiences. Is that on your radar at all or, or is that more of like a futuristic thing that you'll deal with it when it comes down the pipe.

Ryan Cohen (65:31)

I mean, when I think about wearables, the Apple Watch is a good product. When I think about the Meta Quest, I mean, it's a joke. All of the virtual reality stuff doesn't feel like who's going to walk around with these retarded glasses on their head? Like, it's just. It does not seem like that the future. But if we could sell a product and we can make really high margins, then obviously we're going to sell the product. But we're not investing in virtual reality or in the metaverse unless there's a clear path to being able to deliver results for shareholders.

Host 2 (66:14)

Yeah, yeah, I meant more just as like Apple does have the Apple Store network, other companies that are trying to get into wearables and might need in store demos don't have one of those. But. But yeah, I mean your rationale makes a ton of sense.

Host 1 (66:27)

It's so interesting to think about the. When I hear you talk, Ryan, everything you say aligns with thoughtful capital allocation. And yet the broader world seems to believe that this is just about being kind of. This is. It doesn't.

Host 2 (66:47)

It's moonshots.

Host 1 (66:49)

There's CEOs out there that have. Have effectively meme stocks and they just act and talk a lot differently than you do.

Host 2 (66:57)

No. This is refreshing.

Host 1 (66:59)

What do you think about some of the conversations? There's been, I guess, rumors that the admin is interested in companies moving to biannual reporting instead of quarterly reporting. Do you think that's smart? What kind of moves from the admin this year have you been particularly interested in?

Ryan Cohen (67:23)

I like reducing costs. So if it costs us less money than to report biannually, then by all means. I think it's important for shareholders to have visibility into how the company is doing and then making a determination whether they want to stay invested or not. But that we have an opportunity to reduce our costs. I mean, so it costs a lot of money to be a public company. Yeah, we spend a lot of money on auto fees. So that hurt me at Chewy. That hurts me at GameStop. They. They charge us a lot of money. So if it me. If it ultimately means we spend less money on being a public company, then. And it makes us more efficient, then that's. That's fantastic.

Host 2 (68:13)

I didn't mention it, but thank you for Chewy. I've been a subscriber for probably a decade. It's been the backbone of my household. My dogs. Thank you as well. I'm interested to hear your take on kind of evolution of E commerce. There's a lot of chatter about agentic commerce and people buying through products through their chat apps. Have you looked into that? Do you have a current framework for thinking about the adoption rates of people buying stuff through an LLM?

Ryan Cohen (68:51)

Amazon started selling pet products in the late 90s and they were doing okay. Chewy comes along. We originally wanted to start off selling jewelry. We actually bought hundreds of thousands of dollars worth of jewelry. I went to a trade show and I bought hundreds of thousands of dollars worth of jewelry. We set up the website and everything. And then I was shopping in a neighborhood pet store and I had a at the time like a five or six pound tika poodle. And I kept on going to this neighborhood bed store every few weeks and I run into the office. By the way, I don't wear jewelry. I was not passionate about the cat. I didn't know anything that's important.

Host 1 (69:45)

You're just kind of. It was picking like this seems like a good.

Host 2 (69:48)

Yeah.

Host 1 (69:49)

Business and it's light, you can ship.

Host 2 (69:52)

Yeah. It was a little too mercenary. Like you got to get into the, you know, build something that you want so you can give yourself self feedback.

Ryan Cohen (69:58)

Yeah. But you think like your intuition is like there's margins in jewelry and so like you can make money. There was, there was a Blue Nile at the time that had really gross, high gross margins. And I was like, we can do, we can do well in online jewelry. Anyway, I went to this jewelry trade show. I bought hundreds of thousands of dollars worth of jewelry and then I was shopping kind of at the same time every few weeks at this neighborhood pet store for my poodle. And I was way more interested in the pet category than I was in jewelry. I understood the customer and so I ended up selling the jewelry. I ended up getting like 80 or 90 cents on the dollar and I went into the pet category and I liked that. Well, I understood it. I liked the predictability of the industry. I like the fact that once you're a pet owner, you're buying pet food pretty much for the rest of your life. And we shifted and so Chewy comes along. It's 2011 and I took the Playbook from Amazon. So the focus on fast shipping, having a great selection, being able to get onto the website, add an item to the cart and check out. Our average checkout time was like less than two minutes. So you know, you think about like godaddy and I don't know if you've ever bought a domain name but you go through the checkout process. They're trying to Upsell you on like a gazillion stuff. And then all of a sudden Chewy is like average gecko time, less than two minutes. We're not trying to upsell you. It's like we're going to get you your pet food as fast as possible at the best price. So Chewy comes along in 2011 and completely, completely disrupts the industry. And we've disrupted the independence, we disrupted Petco and Petsmart and we were delivering your consumables at a better price, faster, with an easier experience, backed up by great customer experience. And so you don't necessarily have to be first to the game in order to be successful. Amazon was first to the game, but we focused on the category and we were successful. And it's kind of what's interesting in technology in general is like you have these technology companies that are trying to do everything right. Amazon's trying to do streaming, they're trying to do E commerce, they're trying to do everything. And then all of a sudden you have Netflix that is really successful in streaming. You have Chewy, that's really successful in pet food. So if you focus on a category, you could be very successful.

Host 2 (73:03)

How did you process the D2C E commerce era? It was really hot in Silicon Valley. Every MBA was RA Venture Capital. Slap a $250,000 brand on a white labeled product and raise some money and then it kind of fizzled out. But like, how were you processing that at the time? What's your post mortem like? How should people think about building brands going forward? Is there still opportunity.

Ryan Cohen (73:32)

For us? Building brand was acquiring one customer at a time and making sure that they got their pet food or whatever the hell they ordered from us, best price really quickly. That's how you build a brand. It's not spending a bunch of money on, you know, pets.com, a Super bowl commercial. It is focusing on the best marketing is word of mouth. So you allow a customer through word of mouth, you have a great customer experience, you deliver the product really quickly at a great price and you have a happy customer. And that was the way to build. Chewy was getting big market leadership and making sure that customers were really happy.

Host 1 (74:20)

What are the three key lessons from Chewy that you feel like you and the Gamestop team are applying today?

Ryan Cohen (74:32)

Running efficiently. There's no question that like we ran very a focus on, on, you know, Chewy was. They're so different. Thing is, is you think like E commerce and, and, and physical retail are the same and that's Where I, I made a lot of mistakes because I showed up like a wise guy at GameStop. I thought I had a lot, all the answers. But, but E Commerce and physical retail are very, very different. So in general, within physical retail, what I've learned is like, you want to, you have to run lean and you're better off having less inventory than more inventory. So I cost shareholders a lot of money by taking the Chewy Playbook at GameStop. And then I learned physical retail, which was a muscle that had zero memory. And it was on making sure that when you get the product, you sell it very, very quickly. Because if you don't, the product depreciates very quickly. Whereas with Chewy, if I overbought, I sold the product very quickly, it didn't matter. So, yeah, they're not the same. They're, they're not the same. If you. The reason why in general, both concepts have worked out is because, you know, you keep on. I have my own money on the line, so I'm not gonna stop until I figure it out.

Host 1 (76:11)

And you're him.

Ryan Cohen (76:13)

But it costs a lot of money to figure it out.

Host 2 (76:16)

Skin in the game, the man's in the arena.

Host 1 (76:19)

People have said you're. Amen.

Host 2 (76:21)

Yeah.

Host 1 (76:22)

People have said you're too bold. As a CEO, do you think the traditional concept of a buttoned up CEO, CEO that maybe shies away from controversy is obsolete or should be forgotten?

Ryan Cohen (76:42)

I don't even know what a CEO means. I want, at the end of the day, I want whoever's in charge. If it ends up successful, they end up doing really well. And if it ends up not being successful, they lose a lot of money and call them, call me the janitor for all I care. At the end of the day, their incentives should be aligned with common shareholders. So that's the most important thing, is that the incentives are generally aligned.

Host 1 (77:17)

Yeah. You're the loudest shareholder.

Host 2 (77:19)

Yeah.

Host 1 (77:21)

How do you decide? How do you decide what battles to fight? Because you got a lot of stuff coming your way at any given point, a lot of opportunity. What's your framework?

Ryan Cohen (77:32)

My framework is what's going to move the needle. Like if we can make 80% margins but the upside is limited, it's not going to move the needle. Then I don't care if, if it's able to translate into billions of dollars of shareholder value and scalable, then we're talking. So there's a lot of things that are like small time and yeah, we can make a lot of margin, but there's, there's not a lot of upside in terms of shareholders, then who cares? But if it translates into, into something meaningful and scalable, then, then I'm interested.

Host 1 (78:17)

Do you think, do you think power packs could be that?

Ryan Cohen (78:22)

What do you think?

Host 1 (78:23)

Well, the chat is going crazy.

Host 2 (78:25)

It seems like they want to know.

Host 1 (78:27)

They want the update from your side.

Ryan Cohen (78:33)

Power packs is interesting, Power packs fiscal and digital is very interesting. We can't get enough inventory, so that's.

Host 1 (78:49)

Always a good sign.

Ryan Cohen (78:50)

I don't want to say anything because I mean users can.

Host 2 (78:55)

Do you see it as customers can.

Ryan Cohen (78:56)

Figure out for themselves?

Host 2 (78:57)

Yeah. Do you see it as the same sort of like it's more consumption focused so it fits within the chewy model? Is that the right way to think about this as opposed to, you know, there's less risk of getting stuck with a bunch of inventory.

Ryan Cohen (79:12)

Are we talking physical or digital?

Host 2 (79:14)

Physical I would imagine would be like the place where there's inventory risk in some regard. But if you view it as more of a consumption product then like the chewy story, there's presumably less risk in holding a bunch of inventory.

Ryan Cohen (79:32)

I like the trading card space. So we have GameStop has a lot of assets. We have the community and the brand that allows us to. If you look at digital power packs, we've kept, we've kept it pretty limited in terms of being able to get into the digital category and frankly it's not a marketing thing. We just, we can't get enough inventory. It's the same thing on the physical thing, but digital is more scalable. So if we had a choice between physical and digital because digital is more scalable, we're going to go towards digital. Can't get enough inventory, at least at a price where we can, I mean you can buy inventory at 110 or 120% market value but, but we're not going to start doing stupid stuff like that.

Host 1 (80:34)

How do you see the collectibles landscape evolving over the next five years? It feels like different players have picked a focus, whether that's live traditional auctions, something like you just discussed as well. But what is the shape of the market going going forward in your view?

Ryan Cohen (80:58)

They're all connected. Like if you look at the overall collectible space, it's looked at as a stair store of value and it's been that way for decades. If you look at trading cards, you, it's like nostalgic too. You know, I, I, I grew up, I didn't, I collected trading cards a little bit but, but there's definitely a comeback right? Now on trading cards. So it's looked at as a store of value. Whether that continues or not, who knows? I mean, everyone is like, well, it's going to continue. Everyone thinks crypto is going to continue. Nobody knows. But. But we're having a lot of success. When you look at our assets right now, like, the way it stands right now is we're selling the product very quickly. We can't get. The more inventory we have, we could sell the product. So we're going to run efficiently and if we can sell the product right, and if it happens where we can't sell the product, then we're going to adjust and, you know, we're going to lower our costs and going to focus on the things that make sense.

Host 2 (82:14)

Do you have a. Do you have a take on Labubu? I feel like if I find out about it, I'm ultra late and I might have top ticked it when I finally learned about that. But it seems like it's somewhat important to the collectible.

Zach (82:27)

Boom.

Host 2 (82:27)

The story there. Do you have any idea what's going on?

Ryan Cohen (82:34)

Do we sell Labuvu right now? I don't think we sell Labubu.

Host 2 (82:37)

Yeah. Yeah. I'm just wondering, like. Yeah, it's like kind of an odd strategy. Some of the unboxing characteristics, very popular. Seemed like it just kind of emerged out of China out of nowhere. I was wondering if you had tracked the market at all.

Ryan Cohen (82:53)

No. Should we sell it?

Host 2 (82:55)

I don't know. It might be too late. I have no idea.

Host 1 (82:57)

They're very demonic in our view.

Host 2 (82:59)

Yeah, we think they kind of just don't have the right vibe. And I feel like there's plenty of other collectibles that would be more on brand.

Host 1 (83:06)

Personally, how do you.

Host 2 (83:08)

What was that?

Zach (83:08)

That.

Ryan Cohen (83:09)

Is it females or males?

Host 2 (83:11)

I don't.

Host 1 (83:12)

I think it's all sorts of people buying them.

Host 2 (83:13)

I saw Tim Cook had one, so I'm.

Host 1 (83:17)

How are you thinking about. How do you think about M and A on the. In the collectible space? I'm sure people come to you all the time with sort of platforms that maybe have some scale, but not quite the scale that you have that. That would love to sell to you. But obviously companies are bought, not sold. So I'm curious if you've. If. If it's something that you would explore in the future.

Ryan Cohen (83:40)

They don't come to us as often as you think because they know. Well, they probably go to private equity or venture capital or these fancy hedge funds before they come to us. Because I want to make sure that I'm, I care about cash flow and the price that I pay. So we don't, we see some deals, but it's hard to compete against guys that are, or girls, whatever that are getting management fees.

Host 2 (84:18)

Do you think AI plays into the collectibles world at all? Just this idea that like if you have a piece of ip, you can instantiate it maybe much quicker across a whole host of images and videos and kind of build out an intellectual property world faster. Is that actually an accelerant to the collectible trade?

Ryan Cohen (84:37)

I, in general, I have been, I'm the person that's very cynical when it comes to emerging technologies.

Host 2 (84:47)

Sure.

Zach (84:47)

So.

Ryan Cohen (84:50)

Like autonomous driving is exactly. Everyone with like GM, for all of the big OEMs, they're finished because there's going to be autonomous driving. Oh, it's, it's. When it comes to AI, it's a big problem. At some point the computers are going after the humans and I don't think it's that far away. I think that the sci fi movies, when it comes to AI, I feel it, I feel like there's going to be a big problem when it comes to artificial intelligence. And at some point it's going to be the computers against the humans. China versus the us who's going to be the winner? Who the knows? But we got big problems with AI and it's interesting because you can't stop human innovation. And we've got this insatiable appetite to go into these technologies like artificial intelligence that are very disruptible. There's lots of money that's, that is being poured into it. But what the future looks like, we have to be very, very, very careful. So artificial intelligence, it scares me. I mean, I like the productivity benefits, but AI, once the robots come after us, scares me.

Host 1 (86:27)

What's your timeline there?

Ryan Cohen (86:29)

It's faster. It's faster than I would have thought. When I look at what's happening, I don't buy into emerging technologies. But when I look at the advancements in AI, this is no joke. And we have to be very, very, very careful about what's going to happen to artificial intelligence.

Host 1 (86:53)

Where do you think we are in the market cycle? Do you think it's 1999, February of 2000, is that even worth comping to or do you comp to something else?

Ryan Cohen (87:10)

In AI, it feels early, doesn't feel like we're at the end. It feels like we're like the second or third inning, but who the hell knows? But at a high level, I think that it's, I think we have to be very careful when you think about the future of humanity and whether AI is going to benefit the future of humanity. I don't know. I would if I was running a dictatorship and someone made me king and you told me, should we move forward with this technology? It's not clear to me whether moving forward with AI is going to benefit everyone. It's definitely going to benefit the few that are invested in the industry, but there's going to be a lot of people that are not going to benefit from artificial intelligence. So.

Host 1 (88:23)

Chat GPT is depending on pay paid users right now. They can't, you know, if they killed off humanity, that'd be kind of bad for business. Isn't there a way to solve that alignment issue?

Ryan Cohen (88:35)

You know, we've, if you look at the tractor trailer, the US in the 1800s and the tractor trail, I think it was like 80% of the population was working on the farm and then all of a sudden tractor trailer comes along and few centuries later it's like 2 to 3% of the population is working in farming. And you would have said how the hell can the US economy adapt to something that's so disruptive? And we did. But when it comes to artificial intelligence, I feel like it's different. Maybe I'm biased from some of these sci fi movies, but there is going to be a lot of wealth inequality that's created. I don't, I don't like it. I mean there's, there's, there's opportunities to be had. There's no question about it. But is it better for humanity in aggregate AI, what do you think?

Host 1 (89:40)

Well, like, yeah, I think for me, I think it's very straightforward to imagine the dark sci fi timeline, but it feels farther away. I mean at least in our corner of the Internet, people have been reacting to Karpathy's interview with Dhark Keshe that was, I think went live Thursday night. And it feels like the debate right now is AI Frontier Lab progress slowing down, is it just auto complete or. And if it's autocomplete and we don't have, basically if the rate of progress is slowing down, is there massive overinvestment right now? And what I'm hearing from you is simultaneously generally kind of the doomer point of view, which I think is fair, but at the same time it doesn't sound like if you were running a hyperscaler you'd be ramping up capex right now. Now.

Ryan Cohen (90:43)

Are we smart enough as a society to understand what the benefits are and what are the downsides and everyone has perverse incentives. So someone who's in the AI industry is going to tell us how AI is the best thing since sliced bread. But in general as a society, if you think across like human evolution over centuries, do I want to take this emerging technology and is this going to benefit the human population in aggregate over the long term? Cpd, what happens when AI becomes smarter than us?

Host 2 (91:31)

Yeah, it's pretty crazy. Do you think the solutions government intervention, just good stewardship by the leaders of the foundation model labs. Like who actually, who actually has the responsibility of stewarding the new technology most effectively?

Ryan Cohen (91:49)

It's governments and it's ones who have a long tenure. You look at America, the presidential cycle is four years. So, so I don't know if that's necessarily long term incentives, but who cares about where humanity is going to be not in four years from now, not when they check out, but in 100 years from now. Who's got a long term focus on this emerging technology and who cares about humanity over centuries? But I, we have to be very, very careful about this technology.

Host 1 (92:38)

What, what was your view on social media a decade ago? It was quite popular for a period to say that social media was destroying humanity. And maybe it is, maybe it isn't. We seem to have found a way through. But is your view that AI is as bad as people once maybe thought social media would be? Like what specifically when you think about one of the things we laugh about internally is just how easy it is to clock when somebody uses AI to generate a cover letter or a job application. It's a really good way to just get your cover letter application ignored is just to generate it with ChatGPT. It's just beyond obvious. And I'm sure your, your team has seen a lot of this too. But I'm curious, like, what do you think the before we get to the, you know, sci fi doomer scenario where the computers rise up and destroy us all. What, what's kind of like the immediate impact that, that you're worried about about?

Ryan Cohen (93:50)

Social media is one of the worst things to happen to humanity. If you look at Instagram, people are so easily manipulated. They filter videos, they filter pictures. You know, you look at these young people, their expectations, their lack of work ethic. China has censored all of this stuff for good reason because it's so easy to manipulate the layman. And so in general, when I look at social media I say, well, has it benefited humanity or has it been toxic? There's no question that social media is is toxic. I clearly will increase productivity. But at what cost? What cost? I think we have to be very, very careful.

Host 1 (94:58)

When, when public company CEOs talk about how much efficiency they're getting out of AI, do you think that they're actually getting efficiency out of AI, or do you think they're just, just pushing their teams harder to be more efficient and they want to blame the impact on, on AI?

Ryan Cohen (95:15)

It's both. It's both. But AI without question increases productivity. But you know, again, it's short term versus long term. So if you're running Johnson and Johnson are pro or proctor and gamble and you can use artificial intelligence and you could reduce your cost structure because you've got all these humans that are doing these mundane tasks and all of a sudden you realize the computers are going to do it better. Well, duh, you're going to reduce your cost structure. But all of a sudden when you, when you have all of these people that are unemployed and the demand for your product goes away, what's better? You want to keep people employed and have some kind of self worth and working hard and making money, or do you want to replace them and give them universal basic income? And then what does that mean? Does your consumption and demand for your product in aggregate go up or go down as a result of artificial intelligence? I don't know. But one thing I know for certain is that the CEOs, if they can reduce their cost structure in the short term, they're going to replace, they're going to take computers over humans. But is that better for humanity over the long term? You tell me everyone on UBI over the long term, how does that make them feel? People need a purpose.

Host 1 (96:50)

You don't think we'll create new jobs? I mean, we created email jobs. A lot of them could go away and, and the world wouldn't be too much different.

Ryan Cohen (97:02)

Artificial intelligence feels different.

Host 2 (97:06)

Yeah. I just wonder if we feed on.

Ryan Cohen (97:08)

Other things, it feels different.

Host 2 (97:10)

We could still have hierarchies and competitions on things that only humans can do. I mean, we already do this with sports and all sorts of things. There's probably still some sort of like, like reproductive battle to try and get to the top of the stack. Even if you don't need to go and work to make money, there are still other things that you do with your time to raise your status in society. But I don't know, it is a bizarre future to think about. I just Wonder if it's five years, 10 years, 50 years away.

Ryan Cohen (97:44)

We're in an era of instant Gratification, that's the American system. Is that now? Now, now, now, now. Innovation, making money today. But when you think about 10, 20, 30 years from now, and artificial intelligence, what does the world look like? It's not about job displacement, it's not about control. It's. But is artificial intelligence going to control us or will we control artificial intelligence?

Host 1 (98:25)

And you would argue that social media already controls how we feel day to day. We open our phone and decide our mood based on what's happening in faraway places that we have nothing to do with.

Ryan Cohen (98:43)

Social media is a big problem. What's, what's, it's. So if you look at the divisiveness in this country, social media, whether you're a conservative, you go on Instagram and you're, you're conservative and then all of a sudden you get, you know, these algorithms serve you, all kinds of, of things that are going to make you mad. And if you're a liberal, you all of a sudden go on social media and you see all kinds of media that's going to make you mad. Why is it that ultimately humanity, like it comes down to human. Why do we have to lose something in order to appreciate? Well, it feels like the only neutralizer is death. And war doesn't have to be that way, but it feels like the only way we can ultimately appreciate something is if we, we actually lose it. It's very, very, very sad, but that's what it comes down to. There is so much indivisiveness in America if we can just all come together and figure out comma, things that, that we both agree on, but instead we figure out the reasons why we're going to be divided and the politicians divide us and.

Host 1 (100:14)

Well, I think a lot of people have agreed, a lot of people on social media have agreed that they like you a lot. So there's, there's, there's, there's one white pill in there.

Host 2 (100:22)

Yeah, there's somebody.

Ryan Cohen (100:23)

Until they don't, until they don't.

Host 2 (100:25)

Well, hopefully it doesn't.

Ryan Cohen (100:26)

Do they like me over the long term? I don't know. We'll find out.

Host 2 (100:30)

Yeah.

Ryan Cohen (100:33)

True leadership is not about dividing people. It's figuring out how do we bring people together over the long term to benefit humanity.

Host 2 (100:44)

Yeah. How do you apply that sort of thinking to video games? Because there was a lot of fear mongering about violent video games causing kids to become violent. The government did step in and regulate video games with the esrb. Every game is given a rating and young kids, you know, they can figure out a way to get access to some violent video games. But it's. You know, parents are more in control now, and I feel like we more or less got the good outcome and people can enjoy video games responsibly. And of course, there's some negative scenarios. But in general, I feel like video games have been just like a cool medium for artists to tell stories. There's wide variety of experiences. It feels like we as humanity got through that test, whereas maybe we're still in the middle of the fight for positive social media and maybe just at the beginning of the fight for positive AI outcomes. What lesson should we take from how humanity dealt with video games?

Ryan Cohen (101:52)

I think the Chinese have restricted their children from playing video games. I let my kids play video games. Doesn't benefit unless they're playing like Mario Karts. Yeah, but you look at these like Call of Duty. They're killing each other. I want positive influences. I want things that are healthy. I want things where people are going to learn. Blowing someone's head off, exposing it to young people, governments. You know, you look at America, it's the land of the free. It's great. It works well for immigrants because they come from places where it's shit. And they come to America, they have all this freedom and their gratitude. They're grateful. But then you look at people where they don't necessarily have that level of gratitude, and they come to America, you give them all this freedom and they destroy themselves. So do I want my kids going and playing Call of Duty, blowing each other's heads off? And then you look at the Chinese and they're restricting the ability to play video games. When you say, well, they're censoring versus we're free. Well, what's better for humanity? Having boundaries and having rules or just letting people do whatever the hell they want, destroy their lives?

Host 1 (103:29)

Have you ever thought about getting. Have you ever thought about getting into politics?

Ryan Cohen (103:42)

I was born in Canada.

Host 2 (103:45)

There are some positions you could still run for.

Ryan Cohen (103:48)

Which ones? Senator.

Host 2 (103:50)

Yeah, I think you'd be. I don't know if you can be senator, but you could be maybe mayor, right? Isn't the mayor of New York not born in America or going.

Ryan Cohen (104:00)

What is it? There was a saying. My. My herd.

Host 2 (104:06)

You could be city councilman, maybe.

Ryan Cohen (104:08)

Yeah, because they'd kill me.

Host 2 (104:10)

Well, you could go to Canada. You could be prime minister of Canada.

Ryan Cohen (104:14)

Too honest. I'm too honest.

Host 2 (104:17)

You're too honest.

Ryan Cohen (104:17)

I have to be full of. I'd have to be sure they couldn't handle me. They couldn't handle me. Politicians, they're like diapers. They start to stink. They get. They start to stink very quickly. I couldn't play the. So you want to tell people what they want to hear? Yeah, and I'm not playing that game of what they want to hear. But when it comes to artificial intelligence and social media in general, is it beneficial to society and humanity as a whole? I don't like him. I don't like him.

Host 2 (104:53)

That's extremely awesome.

Host 1 (104:56)

That's fair.

Host 2 (104:56)

Well, thank you so much. This has been a really great interview. Thanks so much for calling.

Host 1 (105:01)

Yeah. Anything else? Anything else that we missed?

Host 2 (105:03)

You know, we'd love to. Anything else that you're working on that you want, that we didn't touch on, we'd love to talk about.

Ryan Cohen (105:10)

You guys got anything else?

Host 2 (105:12)

I think we're good.

Host 1 (105:13)

Last question. What's your relationship like with Roaring Kitty? You guys talk much?

Ryan Cohen (105:26)

Ask him.

Host 2 (105:27)

Okay. Yeah, we'd love to have him on the show. Maybe we'll get him on the show. That'd be a lot of fun. He's, you know, we're pretty new to the live streaming thing. I always enjoy talking to people who operate in the same medium and so conversation.

Ryan Cohen (105:41)

Yeah, we don't want day traders. We want. I'll have a conversation with him. If he's focused on decades and centuries, not on making.

Host 2 (105:51)

Yeah, that makes a ton of sense.

Host 1 (105:53)

I like it.

Host 2 (105:54)

Well, yeah. Well, we're rooting for you for the next decade, for the next century, for the next millennia. Thank you so much for coming on the show. This was fantastic.

Host 1 (106:02)

Yeah. Great chatting, Ryan. Cheers.

Host 2 (106:05)

Have a good one. Bye.

Host 1 (106:12)

Good fun.

Host 2 (106:13)

What a wild day on tvpn. Thank you. If you're tuning in for. For the first time, AWS has an outage, so there are pieces of the show that aren't working. We're cobbling stuff together. But we appreciate you checking out our show. Every day we go live at 11am Pacific. We talk about technology and business. We interview CEOs, just like Ryan Cohen, the CEO of GameStop. We interview folks in the private markets, the public markets. We've interviewed Mark Zuckerberg, Brian Armstrong. We have Brian Chesky on the show tomorrow from Airbnb. We have Palmer Luckey on the show tomorrow from Anduril. And we go all over the place. We talk to early stage companies, public companies, everyone in between. But we always love focusing on technology and business. Don't do a lot of politics here. There are plenty of other shows for that.

Host 1 (107:02)

We don't do a lot of AI Doomerism here though. But there is a place for it.

Host 2 (107:06)

There is a place for that too.

Host 1 (107:07)

It's important.

Host 2 (107:08)

So we'd love for you to subscribe to the channel. You can follow us on Spotify. This show is released as a podcast as well. We have a 20 minute version. We also have a newsletter, tbpn.com you can go and subscribe and get a daily update from us on what's happening in the world of technology and business news. We have another guest coming in to the studio, a live guest. But first I need to tell you about Fall, the generative media platform for developers. The world's best generative image, video and audio models all in one place. Develop and fine tune models with serverless GPUs and on demand clusters. And are we ready for our next guest? I think we have Zach.

Host 1 (107:47)

Let's do it, Zach.

Host 2 (107:48)

Live in the TBP ultradome.

Host 1 (107:50)

Coming in while the government is turned.

Host 2 (107:52)

Off about Turbo Puffer search. Every byte serverless vector and full text search built from first principles on object storage. Fast 10x cheaper and extremely scalable. Welcome to the show, Zach. Thanks so much.

Zach (108:06)

Hello.

Ryan Cohen (108:06)

Hello.

Host 2 (108:07)

We have you here. The government shut down, but you were still able to attend. Yeah, maybe like set the table for us. What does the government shutdown mean? This feels like this happens all the time. I remember a decade ago being in college at some pool party and somebody saying, oh, the government's shutting down. And nothing happened then. Is something gonna happen now? Is this even story that people care about, like what's going on?

Zach (108:31)

Nobody cares. This is the big front problem. Nobody cares. It's taking forever for them to do anything. And there's a couple of reasons why. One, government shuts down anytime. They can't pass a budget. We don't pass budgets all the time.

Host 2 (108:45)

Yeah. And that's why we do a lot of those continuing resolutions.

Zach (108:47)

That's exactly right. We gave up on budgets. We shifted to continuing resolutions which basically say copy paste whatever the last budget was plus minus x percent on this thing. Plus minus percent.

Host 2 (108:57)

So you can sneak stuff in into the cr.

Zach (108:59)

Oh, you can definitely sneak stuff in. And people do that all the time.

Host 2 (109:02)

And that's why lobbyists make money. And that's.

Zach (109:04)

I wouldn't know. That's why other. That's why other lobbyists do bad, shady things. We only do good, positive things.

Host 2 (109:10)

But then also, that's why I've seen a lot of founders visiting dc.

Zach (109:14)

That's right.

Host 2 (109:14)

Because if they're going to get money for their program, they need it to be Provisioned. At some point, it's going to be provisioned in a new budget, hopefully. But if not a cr.

Zach (109:23)

That's right. There's basically. Think about it this way. 20 years ago, Congress used to pass all of these small little bills throughout the year, 30, 50 bills throughout the year. Each one had money attached to them. Right. They were going to give money to this, money to that, whatever.

Host 2 (109:34)

They did everything piecemeal.

Zach (109:35)

Piecemeal, exactly. Right. And there was an overall budget that governed things. The Congress all the time would pass laws, bills, whatever.

Host 2 (109:42)

Yeah.

Zach (109:42)

Now we pass two, maybe three bills every year. These are called omnibus bills. Right. Crs are one of them. Ndaa, which authorizes the military, is another one, the Defense Authorization Act. Right. The big problem we have right now is there is a big tension between the Democrats who do not want to pass a clean CR for a variety of reasons. It's not pejorative, by the way. It's just descriptive. And Republicans who feel it's in their political best interest to pass a clean CR. Today. If you're Chuck Schumer and you're sitting there, remember in March, Chuck Schumer worked with the Senate GOP to get a budget passed. If you're Chuck and you're sitting there in March, you have a problem now with your left flank who are looking at you and saying, why are you giving any? Why are you rolling over for the Trump admin? Meanwhile, if you're in the gop, you don't want to have a Christmas tree bill where you have this ornament attached for this amount of money and this ornament attached for that amount of money. And everybody gets their favorite thing. All you want is to say, keep it going exactly how it is. Let's put off any of the bigger discussions for a later time.

Host 2 (110:38)

Yeah. I feel like it just as like a fan of American democracy. I like the big fan. Yeah. Let's give it up for American democracy.

Zach (110:45)

Big ups for democracy.

Host 1 (110:46)

But.

Host 2 (110:46)

But I feel like. I feel like I like the piecemeal.

Zach (110:49)

Yeah, of course.

Host 2 (110:50)

Piecewise bills. Like, I like the idea of like, we're going to go to the moon. Here's the moon bill.

Zach (110:54)

Yes.

Host 2 (110:55)

And it's like, that's. And we all agreed to that. We all got fired up and everyone kind of got excited and we. And we voted for that. And it kind of happened like the classic of like, we want to build a bridge, so we pass a bill to build that bridge. Like, that's right. That's kind of what I was taught in, like, Grade school.

Zach (111:07)

That's what you imagine when you go to civics class in, like, eighth grade. You're thinking to yourself, oh, I want. Do you ever watch Schoolhouse Raw? Yes.

Host 2 (111:14)

Yes. Okay.

Ryan Cohen (111:14)

How?

Host 2 (111:14)

Billy?

Zach (111:15)

Yeah, I'm just a bill.

Host 2 (111:16)

You're on Capitol Hill yourself.

Host 1 (111:19)

I want it to be when there's, like, a systemic failure in the financial ecosystem. I think it should be a one. Everybody gets a vote.

Host 2 (111:26)

Oh, you want direct democracy? Direct democracy. California. Yeah, it's working.

Host 1 (111:30)

Should we bail out the banks?

Host 2 (111:31)

No, no. Or should we just let it out? Weird thing, because I feel like on one level, as a fan of American democracy, I do want the direct democracy. I do want the piecemeal thing. But then at the same time, I live in California and I've had the direct democracy thing, and I still don't have a train that goes to LA to California.

Zach (111:47)

And there's weird, like, feudal landowners now who pass down their homes in California with no property taxes.

Host 2 (111:51)

Yeah, there's like, all these odd. So it feels like maybe both sides have their own warts and edges.

Zach (111:57)

You want, like, I don't know, far be it for me to say what your optimal setup is. The reason I like the older setup is every one of those small bills gets airtime, gets debate, gets discussion, and then ultimately, at some point, you step back and you say, look, I don't want to spend. And most people don't want to spend all morning waking up thinking about politics. I mean, your audience doesn't. I come on here enough as it is. They don't want to wake up and think about that a bunch. So they elect somebody to think about it for them, but they still discuss at least what all the things are that happen. The omnibus bills instead, they're so slammed every time, you're just jamming a hundred things through. So if you're a demo. And look, Schumer's stated concern is that there were a bunch of Obamacare subsidies, ACA subsidies that were built in during COVID that were all set to expire this year. They were meant to be temporary, but as we know, oftentimes you pass something in government, it stays around forever, and it goes on for the rest of time. Dems want to codify this in law. They want it to keep going. Republicans don't want to take up any major topic. Clean cr, nothing attached to it. That's the crux of this. And Schumer is winning points with his left flank right. The AOCs of the world, who's not a senator but is considered primary marrying Schooner he's winning points at his left flank because he doesn't now have to give in or be perceived as giving in to a hostile admin. And the admin thinks they're winning because they look at this and they go, great, now I'm gonna rift people, I'm gonna lay off, I'm gonna shrink the government. All of the things I want to do. Already this, I mean a judge blocked it, but in theory allows me to do.

Host 1 (113:21)

Since this is a technology and business show, what are the current, what kind of groups are currently impacted? I talked to a defense tech founder on Friday and he was saying like it's obviously massively disruptive because you're in, you know, you're trying to get contracts done and nothing can kind of happen for you. Basically like add 60 days almost to whatever timeline you thought. And so that prevents new hiring because they don't have, they can't hire against specific contracts, et cetera. But what are the kind of immediate impacts that you're seeing or hearing about in the private market? Markets?

Zach (113:58)

I mean the biggest thing is this. We work with a lot of founders and I have a lot of founders who are clients who are here on O1 visas and we think the O1 visa is phenomenal. If you want to get an O1 visa today, your processing time is so much longer than it ever was before. Even though, by the way, consulars aren't shut down. Consular offices aren't shut down because they're fee based. They have their own revenue stream.

Host 2 (114:19)

Oh, that makes sense. Yeah, you pay the O1 fee. That's exactly right for the funding. So they don't need funding for the government.

Zach (114:24)

Anything that either has multi year funding or has its own independent set of revenue is able to stay open. It's only the things that require year over year funding from the government that close. And by the way, there's some nuance on that because the admin has kept stuff open. For example, normally the army doesn't get paid during government shutdown. The admin has said, look, we're going to take the $6 billion left over from R and D spend in defense. So this is impacting the defense folks, right? Any R and D spend.

Host 1 (114:49)

Oh, that would have have potentially gone to startups.

Zach (114:54)

It's not sbir so much but it's more science Y but yeah, exactly.

Host 2 (114:57)

R and D spending.

Zach (114:58)

It could go to startups. Yes, that could go to startups that do R and D stuff. And instead we're going to use that money that was already allocated but hasn't yet been dispersed to pay the military. So it's a little bit of robbing.

Host 2 (115:06)

Peter searching through the couch cushions.

Zach (115:09)

Yeah, that's exactly right. And the problem you run into is, look, they can do that now, but eventually you run out of that money too. And so at some point, somebody's going to have to not get paid. So if you're an O1 visa holder, you are in a tough spot because you now have a much longer lag time, much longer processing time to get into this country. Even if you were already here and just happened to be gone. And coming back for your renewal, that's tough.

Host 2 (115:30)

Yep. Yeah, that makes sense. So, like, how long do we expect this to last? Like, where should we be watching for, like, updates? Is everyone kind of pricing in like, oh, six weeks is standard, but then it's the question of, like, is it 6 or 12 or is it really wide?

Zach (115:48)

Who knows? The real honest answer is the longest partial shutdown in history was under Trump 1 and it was 35 days. We're at 20 days today. Okay, so that's the tail end risk. The problem is the difference between then and now is a couple things. One, it's not a big press issue like we talked about. Like, I can imagine. I'm watching people tune out of this interview when they're hearing about the government shutdown. People don't care.

Host 2 (116:09)

No, no, they don't.

Zach (116:10)

And by the way, both sides think they're winning. The admin side says, look, there's great stuff happening. We just have peace in the Middle east plus or minus a little bit. Right. Dems are saying we've had huge protests. We galvanized the base.

Host 2 (116:19)

Oh, yeah.

Zach (116:20)

No one's motivated to get this close. Couple leading indicators you want to look for. One, military payday. Like I said, if that happens and they don't have another way of getting dollars attached to it, that's going to be a problem. Two, Majority Leader John Thune, who is the senator from North Dakota in the Senate, Republican guy. All of the farm state senators are looking at the expiration of loan programs that have help keep farmers solvent every year. If you're a farm state senator, this matters to you a great, great deal. So the admin is looking today to figure out where can I get my dollars to go and continue to fund these things. But in the absence of funding this, it's really, really hard for a shutdown to continue without political pressure. Third thing you want to look at, and this is the big one, is political consequences. Because the truth is, as Much as Congress today has said, fine, I'm happy to abdicate my role and let the White House run the shutdown response. In theory, Congress controls the power of the purse. They are able to at any time, right. Come together and do something. If the Dems lose the governor race in Virginia, which is disproportionately affected by the shutdown because there's so many federal employees who live in Northern Virginia, or if a year from now, which I don't expect will still be shut down, but if in some number of months you start to see negative impact, as we lead up to the midterms, senators are gonna wake up and say, gosh, my political future right now doesn't have to be tied to the outcome of the shutdown. I can be an independent actor. I don't need to rely on the White House or anyone else. The Dems can say this too. I don't need to rely on Chuck Schumer to guide where my vote goes. That's the real when my rubber meets the road.

Host 2 (117:47)

Is the government shutdown going to increase the risk of thieves stealing the Constitution? Robbing the. We saw that. These. Finding the national security in a brazen Louvre robbery. Do you saw this?

Zach (118:02)

I saw this. And they, like, left it in the gutter. I heard.

Host 2 (118:04)

Well, they dropped a crown. They did get away with a lot of jewels. And I hope that the folks at the Smithsonian who are protecting our moon rocks are not getting furloughed because we gotta protect our crown.

Zach (118:16)

It'd be the worst national treasure sequel.

Host 1 (118:18)

By the way, what's the current dialogue around AI regulation and policy in Washington? And then what are you seeing across the.

Host 2 (118:28)

How long does it take to go from Carpathia on Dwarf Cash to Capitol Hill? Are they watching or are they listening to shows that are talking about that or are they talking the next narrative? How many links in the chat?

Host 1 (118:39)

Like, wait, it's just autocomplete. Always has been.

Zach (118:44)

There are a swath of think tanks that do a really good job of translating what happens in dc. Sorry, happens in sf. Yeah, exactly. To dc.

Host 1 (118:52)

Big computer.

Zach (118:54)

That's right. Everything is computer. You can build a really good think tank. You're the guest who's supposed to be before me. Dean Ball, who's an amazing guy, was the AI advisor at OSTP at the White House Office of Science and Tech Policy. He's done a really good job being that communicator. Exactly. And his think tank fai, which I'm involved with too, does a really, really good job bridging the gap There. But, and by the way, things like the progress conference just happened at sf, there's a lot of things now institutionally that are built to translate this on like a relatively quick timescale. The truth is, look, if you're watching right now and you care about AI policy, the White House has an RFI or quest for information out today where they're asking people, founders in particular, to write in and say, gosh, what are the things that are impacting, what are the regulatory burdens that are impacting my ability to do AI business in America? So they want the answer, but it's hard to get ground truth because when you're in dc, it's a little bit of a bubble, you hear, a little bit of an echo chamber.

Host 2 (119:48)

Is the core D.C. aI narrative just state by state regulation versus not? Or are there actually higher level discussions around, okay, if $1 trillion of CapEx is going to happen, like there might be, there might need to be some fundamental changes in the way we regulate data centers power. Even, just, even, even if you're just the ultimate AI bull, you might need to step in and say, we're going to help speed this up. Because at a certain point, if you're building 10 nuclear reactors, you just can't do that without the help of the federal government. Right.

Zach (120:15)

All right, so part of the problem is we have a federalist system. And so power by necessity is distributed across, across state, federal, local. Right. So if you're talking about, you mean government power, not, sorry, literal power, Although. Yeah, literal power is also distributed electricity. Yeah, electricity is also distributed across these states.

Host 2 (120:29)

Yeah, yeah, yeah. As you can see with Elon, he built his Colossus 2 data center right at the intersection of three different states. Because he needs to be able to go over here for a little bit. Yeah.

Zach (120:37)

Part of the problem is this, if you want to build a new reactor or you want to build an energy transmission line, you need so many different people to buy in. In the last 30 years of American governance, basically at every level, has been built around keeping things the same status quo for people who already have done fairly well. The other thing I would think about.

Host 1 (120:55)

By the way, it's so funny because I don't read a lot about politics, but I have such tangible experience in politics through trying to get things done with my hoa.

Zach (121:06)

That's exactly right.

Host 1 (121:07)

It's literally like the boomers control the HOA board. And I tried to suggest changes when I bought my house and I basically got death threats.

Zach (121:16)

That sounds about right.

Host 1 (121:17)

That's my framework.

Zach (121:18)

For California governance is the largest HOA in the country. Think of it that way. It is just completely biased against. All right, here's the other problem. You have a thousand people in congress who all have their own political aspirations. Look like a Marsha Blackburn. Marsha Blackburn wants to be governor of Tennessee. What's one of the biggest industries in Tennessee? Nashville. Nashville Music. Right.

Host 2 (121:37)

Yeah.

Zach (121:37)

So Nashville, the home of John Fios the drink. Yes, yes, yes.

Host 2 (121:42)

That's really swinging the.

Zach (121:43)

That's huge. Marsha Blackburn, what was the other thing.

Host 1 (121:46)

We saw about Tennessee? Like something about isn't Nashville. Nashville specifically is pushing heavy on AI regulation.

Zach (121:53)

That's right, exactly. They have the Elvis music.

Host 2 (121:55)

No, no music. Okay, that's right. All right.

Zach (121:57)

You guys know Suno, the AI audio app? They hate Suno. They hate Suno. Oh, they hate it. It's like public enemy number one when you're in Nashville. Right. So Marsha Blackburn, it gotta be.

Host 1 (122:07)

Isn't it an open carry out there? Do not. If you're.

Zach (122:12)

Do not go to Nashville.

Host 2 (122:14)

If you're a lightspeed you funded soon. Oh. Do not go to Tennessee.

Zach (122:16)

Do not walk around Tennessee. I wouldn't personally.

Host 2 (122:19)

Nashville.

Zach (122:20)

Marsha Blackburn wants a political future in Tennessee. She wants to get out of the Senate, which is a miserable place to be and get into the executive chair in Tennessee.

Host 2 (122:28)

Why is it so miserable?

Zach (122:29)

Because nothing gets done. You know, the House works. No, this is like real. I mean, look, it's frustrating if you actually are a motivated person.

Host 2 (122:34)

So you say that the House works, but the Senate doesn't.

Zach (122:36)

No, the opposite. The House works even less. The House worked in the last. Since like July 4th. I think the house has worked something like 20 days in total. Okay, that's insane. Like think about if you guys.

Host 2 (122:44)

Did they not get the memo about the great lock in?

Zach (122:46)

No, they are. They are doomed. The permanent underclass.

Host 2 (122:49)

It's terrible. It's awful.

Zach (122:51)

This is the problem. All right, so Marsha Blackburn wants a political future in Tennessee. She says, okay, if I pass the Elvis act today or if I allow the Elvis act to pass.

Host 1 (122:58)

They should have called it the Trough Act. Don't put slop in my trough.

Host 2 (123:05)

Keeping keep organic.

Host 1 (123:07)

I want organic food for all my piggies.

Zach (123:12)

Organic only organic AI for me, thank you very much. Okay, so if she lets the Elvis act pass with no federal preemption, she wins herself a lot of friends.

Host 2 (123:21)

Sure.

Host 1 (123:22)

Tennessee.

Host 2 (123:22)

She can be governor.

Zach (123:23)

Exactly.

Host 2 (123:24)

Yeah.

Host 1 (123:24)

If you want to. I imagine having all the country music fans and the. And big country on your side for governorship in Tennessee. That's got to be pretty helpful.

Zach (123:34)

It's financial capital and it's social capital, too. It's a big outcome. Now think about that same dynamic writ large over and over and over a thousand people, all of whom want their own thing. Look, the truth is, at some point there will have to be a federal standard. That happens. It's going to have to happen. And some of this stuff. Ted Cruz put out a bill around AI Sandboxes, which is in the ostp, the White House AI Action Plan, basically saying, look, we're going to set up special economic experimentations. Zones is one way to think about them. So people are tackling it from different respects. But unless you're thinking completely nationally and you want to be governor, or, sorry, you want to be president, if you're thinking about anything at the state level, you are right now not going to act federally when it comes to AI policy.

Host 2 (124:14)

Really quickly, you said a thousand people. Help me break that down. It's 100 senators, 538 House members, but where's the other?

Zach (124:22)

Like, there's a gap. I mean, I was being a little hyperbolic. It was a gang. It was a gang. There's a phrase, gang of 500. Okay, yeah, 500 refers to everyone in the House, everyone in the Senate, White House, staffers, agency staffers.

Host 2 (124:32)

Got it.

Zach (124:33)

And then people like me, the lobbyists.

Host 2 (124:34)

The reporters, you add all those folks, and those people have. They might be. Want. Want to be the governor.

Host 1 (124:40)

Sure.

Zach (124:40)

They might want to have an aspiration or even they want to do business with somebody.

Host 2 (124:42)

Sure, right.

Zach (124:43)

They have some aspiration. Whatever.

Host 2 (124:44)

There's a thousand live players.

Zach (124:46)

That's right, 1,000 live players. Really? 500 live players. And that's who determines, quote, unquote, conventional wisdom in Jersey.

Host 2 (124:52)

Okay, Right.

Zach (124:52)

So if you're, you know, Blake, you guys had on from Boomera a while ago, whatever it is, the number one thing you wanted to do to convince people that you want to have a speed, sorry, a sound law and not a speed law is convince those 500 people over and over again to shift their perspective. Once you do that, the conventional system works so difficult.

Host 2 (125:09)

I mean, I love Blake, obviously, and I'm rooting for him, but, yeah, that seems like such a hard challenge in the face of, like, you can't just go and do, do the Supersonic act of 2025 by itself, and everyone's like, yeah, this one makes sense because you have to puzzle piece it with 25 other things on the omnibus bell.

Zach (125:25)

Well, it's like when we were all in enterprise software, it's account based marketing.

Host 1 (125:28)

You know what I mean?

Zach (125:28)

Account based marketing. You have one person, you have all the influencers who sit around that person and your job is to win over each influencer over and over again. It's the same thing here you have one person, maybe five people, right. Who can actually do it. Majority leader, the president, speaker, so on and so forth, forth. You want to flip over all the people the think. Thanks. The reporters, the junior members, their delegation, all the people who have influence on them. If you do that, you win.

Host 2 (125:50)

Yep.

Host 1 (125:51)

Sorry, Jordy, I cut you off any. Do you cover nuclear at all? Yeah, because every time we see these, you know, one gigawatt, you know, all these gigawatts data centers being announced. Yeah. It just, it seems like the, the, like nuclear is just going to have to play a huge part in that. And yet we need to relearn how to make reactors here in America.

Host 2 (126:15)

Yeah. It seems like a lot of people are underwriting like. Oh yeah, like, you know, we're going to, it's going to be a lot of money. We'll sign, we'll sign on this line and then the money will come. One wire the money and then, yeah.

Host 1 (126:24)

We'Ll just turn on a new active reactor.

Host 2 (126:27)

Those are two very different things.

Zach (126:28)

So easy to do.

Host 2 (126:28)

Sorry.

Host 1 (126:29)

D. Yeah, one, one new reactor, please.

Host 2 (126:31)

Please. Yeah, yeah.

Zach (126:32)

Thank you very much.

Host 2 (126:32)

Add that to the term.

Zach (126:33)

I'll take one reaction reactor and 50 transmission lines. Thank you very much. I'm all done.

Host 2 (126:37)

You really can like wire $100 billion in a day. Yes, but you can just wire 100 gigawatts in a day.

Zach (126:44)

This is part of the problem. So we do full disclosure. We do lobby for a nuclear energy. It's a great company, I won't name them, but a phenomenal business we have. Look, if you look at the federal.

Host 1 (126:52)

By 30, if you're not heavily conflicted.

Zach (126:54)

You'Re doing something like no conflict, no interest is the name of the case. This is the thing. If you are looking at the federal government and you're looking at this federalism system and the challenges that are inherent to it, you're thinking, who's cutting across this credit to the admin. They are the first ones who I've heard of who at least thought about, hey, we ought to have somebody whose job it is to expedite these long term. Michael Grimes. Right. Who runs the US investment accelerator. Former tech banker for many, many years, Calbert Go Bears. Yeah, there you go.

Host 2 (127:23)

Polytechnical.

Zach (127:24)

And anyway, he is running basically what Turns out out to be the federal government's investment bank. Sure, he's running that across. Anytime there's a system they want to accelerate and deployment of an investment, that's who they can turn to. But even they're limited. That's the problem. So nuclear is disproportionately affected by this?

Host 2 (127:39)

Yeah, no, it makes a lot of sense. Do you have a take on like Sagar and Jetty has been saying, like, oh, the technology doesn't know what's about to come. The narrative. I mean, we see these things debunked on X and in tech, in the tech part of X every day where there will be a mainstream news headline About Sora uses 25 gallons of water every time and it gets sort of debunked in some research paper from Google. But that doesn't really make it back. No, of course not. And so I think Sager is identifying a potential wave of anti tactical sentiment across both sides of the aisle. What's the mood around that generally?

Zach (128:24)

I mean, look, unfortunately it's great business and great politics to be anti tech.

Host 2 (128:29)

That's the problem because it's a narrow community.

Zach (128:31)

It's a narrow community, it's disproportionately wealthy, disproportionately influential. Very easy to be a punching bag.

Host 2 (128:36)

Very sloppy. Very sloppy. No, it really is like very visual to just show a picture and be like, this is bad. Yeah.

Host 1 (128:45)

That's why I was shocked that OpenAI came out and announced that they'd be supporting erotica. Because that just feels like as these debates come around with power and infrastructure, it's very easy to be like, you said you were trying to cure cancer, give free education, but clearly a lot of your users are using this for adult entertainment. It takes you down a path that it just puts a target on your back.

Zach (129:12)

This is the big divide on the right. Right. If you're on the right way, there's two camps. One camp says let the slop flow free right. Tech companies, let them do their work, whatever, it's great, let them do it all. And says the goal is to be pro tech generally. Right. One camp says, we're social conservatives. I don't want my kid getting porn from OpenAI. I don't want my 6 year old being exposed to this shit. Pardon the language. And so they want to reel it back in. That's the big problem. Look, if you're OpenAI, culture is often downstream of politics. Sure, Trump won this election, huge wave of things that feel pro. Freedom of speech. And you can Quibble with the definition and the boundaries that all you like. But in theory, the messaging is around being pro free speech. And so a lot of what they're doing aligns to that. But I think there's a backlash brewing. And it cuts to. John, to your point, it cuts across left and right. Right. Josh Hawley and Elizabeth Warren have common cause on very few things, but one of them is tech.

Host 2 (130:02)

I made a YouTube video about that, like years ago about how it was. They just came out from completely different realms, but they were saying the exact same thing about tech.

Zach (130:09)

That's right.

Host 2 (130:10)

And that was in the context of Google and monopolism and large corporations. But it just keeps ringing true again and again.

Zach (130:19)

Because the truth is, the Neo Brandeisians who didn't like tech, who are opposed to these sort of monopolistic or what their view is monopolistic practices, a lot of those folks have re. Cloaked themselves now. They talk about tech's power in other ways. Right. It's very sympathetic. If you watch these videos.

Host 1 (130:33)

What about Swiss watches?

Host 2 (130:36)

I heard that too. Yeah.

Zach (130:38)

Can we do a little wrist view? A little wrist zoom in, please? Yeah, actually, I don't want to do that because you guys mock me every single time. So I'm just a glutton for punishment. The bigger thing is. All right, so look, you get these videos on X from this group, you know, this group, More Perfect Union. You guys see these videos? They're the big water sort of power pusher people, right?

Host 2 (130:55)

Sure, sure.

Zach (130:55)

They make these very slick videos. Basically saying, you know, if you allow a data center to open up in your neighborhood, you're gonna be, you know, in a drought and your crops are gonna dry up and God's gonna smite you with locusts and so on and so forth, like a horrible outcome. Okay. And then, you know, people fighting the good fight. You know, ifp Alex. Stop. You guys know them.

Host 2 (131:12)

I know.

Zach (131:13)

Yeah, yeah.

Host 2 (131:14)

Ifp. Institute for Progress. Exactly.

Zach (131:15)

Institute for Progress.

Host 1 (131:16)

He's been on, right?

Host 2 (131:17)

I don't know if Alex has been on, but yeah.

Zach (131:18)

All right, so you have Alec or Caleb from IFP Institute for Progress, who do great gods work on Twitter. Every time that comes out, they're there posting, saying it's not true. And so one.

Host 2 (131:27)

And so someone should make a data center that runs on saltwater.

Zach (131:29)

Yeah, there you go.

Host 2 (131:30)

That would be really easy.

Zach (131:31)

That's.

Host 2 (131:31)

That's an easy win. You can just.

Zach (131:33)

Yeah, exactly. Give me like sewer water in there. Yeah, I don't know. Just disgusting.

Host 2 (131:37)

Gross.

Host 1 (131:38)

Your slop is actually.

Host 2 (131:40)

It Would be such a good, it would be such a good retort. If somebody's like, you're using all this water. You're like, yeah, we use seawater.

Zach (131:44)

Yeah, Correct.

Host 2 (131:45)

There's infinite.

Zach (131:46)

Infinite. In fact, we're purifying it at the same time because it's so expensive.

Host 1 (131:48)

No, it wouldn't work. It wouldn't work. A dolphin was going to drink that. And right off the shore there's an endangered species that is going to go extinct because you took the water from them.

Host 2 (132:01)

The sea turtle is going to get stuck in the Nvidia rack and.

Host 1 (132:04)

Okay, I have a pitch for you. So we talk about like presidential libraries, right? Could we be moving into a future where each president gets one spac, each.

Zach (132:18)

President gets one free shot of insider trading and then it's all good from there?

Host 1 (132:21)

Okay. No, but what's going on with Obama's presidential library? No, no, no, no.

Host 2 (132:27)

Clearly not built by humans.

Zach (132:29)

Oh my God. The Arctic.

Host 2 (132:30)

Obviously an energy source. But three, because of the AI narrative. This could be what sets up Michelle for her run. Because energy is going to be so expensive. The Obama presidential library, if that's what it is, built by aliens power source, lowers the rate of energy for Chicago, for everyday Chicago. So true. Huge. Huge. That's a launch.

Host 1 (132:52)

We gotta get the tinfoil hat. No, I just saw something in the, we didn't even get to it in the show today. I just saw that there's a Cayman Islands entity that's gonna be participating in the spac. Do you know what the story is there?

Zach (133:09)

I don't have the full story.

Host 1 (133:11)

And what do you think? What's your sort of non political view on, you know, we will undoubtedly have another Democratic Democrat, you know, president. Do they, does this set a new norm?

Host 2 (133:28)

Where wasn't Gavin Newsom saying, like, oh, maybe we're gonna do a coin. We'll do our own coin. Like we'll fire back with our version.

Host 1 (133:34)

By the way, I think COIN on coin violence.

Zach (133:36)

He's peaked way too early. Oh yeah, like you don't want to be peaking this early.

Host 2 (133:40)

Sure, sure.

Zach (133:41)

Like a delayed peak a little bit.

Host 2 (133:42)

Yeah.

Zach (133:42)

That is, have some time. You're way ahead.

Host 2 (133:45)

I do wonder, you know, people are obviously latching onto all the, all the Trump projects, whether it's the COIN or Truth Social or this new SPAC or a variety of projects. But I wonder if we looked at his, the number of companies he builds per decade, if he's actually at a low period in his career because he was originally doing like Trump stakes. Trump University Trump, he had airlines, he had a vodka. And so he used to be in 10 different industries. Maybe he's actually more focused than ever.

Zach (134:12)

That's right. If you really think about it, in some ways, this is the double down, lock in period for Trump as a businessman. As a businessman, he's avoiding the underclass.

Host 2 (134:21)

Yes. But, you know, we got to give credit to Jimmy Carter, who was the most locked in president. He had a peanut farm, he sold it.

Host 1 (134:27)

Now he put it in a blind truck.

Host 2 (134:28)

Oh, he did. You looked it up. He put in a truck.

Host 1 (134:30)

Oh, wow.

Host 2 (134:31)

Okay.

Zach (134:32)

No longer. Obviously. I was a Carter fan before.

Host 2 (134:34)

No longer. Yeah, yeah, yeah. Put it in a blind trust because he didn't want people to think he's in the pocket of little peanut, big peanut, Mr. Peanut. Ruling over Jimmy Crow. It was a very small farm, but he was like, I'm out of the game. I'm out of the game. I'm locked.

Zach (134:46)

This is my thing with the Dems. Right. If you're looking at the next four years and you're thinking to yourself, okay, public opinion, often thermostatic, it's going to swing back at some point.

Host 2 (134:54)

Oh, that's good for praise for that.

Zach (134:55)

I like that.

Host 1 (134:56)

Yeah.

Zach (134:56)

And this is the problem, which is like, like, look, both parties, when they're in power, they love to overreach. Right. And so if you're doing, if you think to yourself, I have four to eight years to execute every goal I've ever had in my life before, public opinion is thermostatic again.

Host 2 (135:08)

Sure.

Zach (135:08)

It's no wonder there's. People call it the magafication of the Dem party. You look at them and you're like, gosh, AOC Zoron being stars in this party is so unusual relative to even where Clinton was in 2008. Hillary Clinton was 2008, let alone 2016.

Host 2 (135:23)

Yep.

Zach (135:23)

So if you're trying to put your finger out, particularly where the wind is blowing, think to yourself, if a Dem benefits from massive tailwinds coming out of this admin and they elect or they nominate somebody who is a magified version of the Dems.

Host 1 (135:35)

Right.

Zach (135:35)

An AOC or a Zoranish part of the wing. That makes me concerned. Right. Because then you just see the pendulum swinging back and forth from extreme to extreme for the next foreseeable future. And there are many, many times you're.

Host 1 (135:46)

Saying that America could be more divided than ever.

Zach (135:48)

Yeah. America more divided. First I'm hearing of this. More divided than ever. Yeah. This is true. Historically. Many times.

Host 2 (135:54)

Yeah. What else are you monitoring in D.C. right now? What did we cover? What's kind of like under the radar story that maybe people aren't focusing on yet but should be? Is there anything that's at the bottom of your list that maybe is bubbling up?

Zach (136:09)

I mean the big thing that I'm thinking about a lot is a lot of these questions around education. You guys see the higher ed Compact that the Trump admin put out?

Host 2 (136:15)

I did.

Zach (136:16)

All right, this is a big thing. Trump Admin is basically saying, look, we want a set of standards that universities will sign in order to be eligible to receive future funding.

Host 2 (136:24)

Sure, sure.

Zach (136:25)

A lot of the things when you read them, by the way, they don't sound crazy. You read them and you're like, gosh, I thought universities always operated this way. Some of them, by the way, are impossible for universities to do. Some of them are super reasonable and there's a huge spectrum.

Host 2 (136:35)

Sure, sure, sure.

Zach (136:36)

Generally, I think a lot of people are sympathetic. I understand the sympathy for it. The fear is, look, universities used to be be. I'm not saying they ever were this in practice. Obviously they've had a political bias all their own. I mean academy's had that for a long time. But as an institution, universities used to be beyond politics. Right. You didn't have somebody who ran.

Host 2 (136:53)

Yeah. You got tenure and then you can say whatever you wanted politically.

Zach (136:55)

That's right, that's right.

Host 2 (136:56)

You were good.

Zach (136:57)

Everybody knew there were some whack job professors. It was like your nutty uncle. It wasn't like the threatening evil communists.

Host 1 (137:02)

Totally, totally.

Host 2 (137:03)

It wasn't like don't platform Chomsky.

Zach (137:05)

Yeah, that's exactly, yeah. Nobody was looking at Chomsky saying this guy's deport him, send him back to Poland or whatever.

Host 2 (137:10)

He's got his theories.

Zach (137:11)

So here's the problem. Now you look at this compact, you go, okay, maybe you say to yourself, 60, 70%, whatever is reasonable. What's going to happen the next time there's a Dem in office who now says, great, our compact for universities to get federal funding is going to say there has to be dei.

Host 2 (137:25)

Sure, sure.

Zach (137:26)

Or there has to be this. Yeah, exactly. That's the thing I'm tracking for the future is look, universities are pushing back. I understand why. I think it's very reasonable for them to do so. And I get both sides of it. I see the value for the admin too. I'm concerned about the precedent of it swinging back and forth over and over again.

Host 2 (137:40)

Yeah, it's so interesting because maybe it's just like the teal influence. But I've sort of stopped paying attention to universities entirely because every day on this show we talk to some 16 year old that isn't planning to go to college and is already building a company. And then when we do talk to people about the education space, it's usually because they're building an alternative to homeschooling.

Host 1 (138:01)

Yes.

Host 2 (138:02)

Or what Joe Lamont's doing at Alpha School or Andrej Karpathy's doing with Eureka. Like there's so many different initiatives that just live entirely outside of the original, the traditional education system. But it does feel like there's still pockets of the economy that are really tied to education. We talked to Huberman about this in Basic Science Research.

Zach (138:24)

Yes, that's exactly right.

Host 2 (138:25)

And I'm super blind to it because all of the AI research that we was previously done at these AI labs on university campuses just completely paid for by big tech.

Zach (138:35)

That's right.

Host 2 (138:35)

Right. And so I'm not worried like, oh, we're not going to get the next AI innovation because there's not enough funding or whatever. It's like, no, there's more funding than ever. These PhDs are making $100 million now.

Zach (138:47)

Well, my problem isn't AI.

Host 1 (138:48)

AI.

Zach (138:48)

I'm with you all.

Host 2 (138:49)

Yeah, yeah. You can't apply that to everything.

Zach (138:51)

Correct. You look at something like oceanographic research total and you have like one place, Woods Hole in Massachusetts, which is like the best oceanographic research institute in the country.

Host 2 (139:00)

Yeah.

Zach (139:00)

There is no industry that is funding hundreds of millions of dollars for oceanographic PhDs to go do weather research.

Host 2 (139:06)

Yeah. We need like Saronic and oral. That's right. That's right. Amadon and a few others that are like in that ocean. Ocean tech space, we need them to be worth trillions.

Host 1 (139:15)

That's right.

Zach (139:16)

And the NSF and basic research and universities are hugely important for that stuff. Still. The thing that I get concerned about is when they start to be the football moved back and forth. And frankly, if you look at. Look, this is a very Nixonian admin in a lot of ways. If you ask Young Republicans in D.C. i love Nixon, so I'll just put that out there. If you ask young Republicans in D.C. who they look up to, they will tell you it's Nixon staffers. They'll tell you to go read the Pat Buchanan books. And Pat Buchanan's got his own fair share of mishegoss. But if you read the books which are well written, you'll see the big theme the Republicans have about Nixon, is he apologizing too much. Pat Buchanan will tell you Nixon should have brazened out Watergate, never should have apologized, should have stuck it out, should have broken the power of the Ford foundation and the large institutional foundations. That's the Nixonian view.

Host 2 (140:02)

What's the view on Kissinger then?

Zach (140:04)

I think these people are more domestic policy focused than Kissinger, but they're more America first than the Kissinger in a non Kissinger.

Host 2 (140:10)

Because that seems like very different in terms of open up relationship with China. Let's go over there and do deals.

Zach (140:15)

This is the domestic policy.

Host 2 (140:18)

We like Nixon, but just, just the stuff we like. Just the stuff we like. That's how people always remember.

Zach (140:22)

Of course. And the truth is people like Buchanan have become.

Host 2 (140:24)

Steve Jobs was just a designer. He wasn't a ruthless manager. I've never heard of that in my life.

Zach (140:28)

He was a passionate fan of design.

Host 2 (140:29)

Exactly.

Zach (140:30)

And it's always known for it.

Host 2 (140:31)

That's it. That's it.

Zach (140:32)

If you're like a Pat Buchanan fan, you're looking at this moment and you're saying, gosh, we should break the backs of foundations or institutional life, you know, higher ed. Right. And so that's where some of the desire is coming from is this reading of history where you said, look, Nick Snadman didn't go far enough enough in defeating its enemies and they allowed the academy, they allowed capture by X number of people. And that's why they're motivated now to go and try to do this, even though I think it is not great for universities.

Host 2 (140:56)

Yeah, that makes a lot of sense, Jordy. Anything else?

Host 1 (141:00)

No, always. We can chat all day if you want to keep hanging out. I can tell you about replacement AI. They're running a billboard in SF right now that says our AI does your daughter's homework, reads her bedtime stories, romances her deep fakes her. Don't worry, totally legal.

Tyler (141:17)

Who needs parenting?

Host 1 (141:18)

And then it. And then on.

Host 2 (141:20)

If you want a positive Billboard, go to adquick.com out of home advertising Media. Easy, measurable. Say goodbye to the headaches of out of home advertising Only Ad Quick combines technology out of home expertise and data to enable efficiency mis ad buying across the globe. He's like, I'm out of here. Thanks so much for coming on. Great to see you, Zach, as always. But yes, this billboard was bizarre.

Host 1 (141:38)

The whole site is wild. If you go to their website, it says the only honest AI company and they say human flourishing.

Host 2 (141:45)

Have you seen this, Tyler? This, this, this the replacement AI. Are they, are they doomers or are.

Host 1 (141:51)

They, they're massive, so they're massive doomers.

Host 2 (141:54)

But are they pro AI?

Host 1 (141:55)

They're. They're positioning the. They're positioning their. Their movement as a startup. So it's. Replacement AI is a website. The title says humans no longer necessary, so we're getting rid of them. Replacement AI can do anything a human can do, but better, faster, and much, much cheaper.

Tyler (142:11)

I don't think they're doomers in the sense that they think AI is going to like kill everyone now. Just that, like, they're. It's just going to make everyone's life. Lives worse.

Host 1 (142:18)

Yeah. Okay, so they highlight some quotes here. This quote from Sam Altman, AI will probably most likely lead to the end of the world, but in the meantime, there'll be great companies.

Host 2 (142:26)

That's a wild quote.

Host 1 (142:28)

They have one.

Host 2 (142:29)

When did he say this? This is very taken out of context.

Host 1 (142:32)

Yeah, that was in like 20.

Host 2 (142:34)

They're probably asking him, like, what's the worst possible thing that could happen? And he's like, this is what the worst thing could ever happen.

Host 1 (142:39)

We have a quote here from Dario. I think there's a 25% chance that things go really, really badly.

Host 2 (142:45)

Very odd. Well, on the other side of the spectrum, you got the Starbucks CEO Brian Niccol says the coffee giant is all in on AI reveals real time artificial intelligence systems designed to assist baristas and transform star store operations. I got a tip for you, Brian. Get on Profound, get your brand mentioned on ChatGPT. Honestly, reach millions of consumers who are using AI to discover new products and brands. Seriously, people are going to be saying, I want coffee. Where should I go? Near me. You got to be on Profound. Get your brand mentioned.

Host 1 (143:16)

Something. We got to put on the schizo hat.

Host 2 (143:19)

Okay. Please.

Host 1 (143:19)

Starbucks Stock is up 6.66%. Coincidence.

Host 2 (143:27)

Very weird.

Host 1 (143:27)

Are they using AI to summon the demon?

Host 2 (143:29)

I don't know. Well, if they build something, they should do it on Google AI Studio. The fastest way from prompt production with Gemini, chat with models, vibe code and monitor usage. Also, we have to try that frame to frame VO 3.1 thing. We'll talk about this later, Tyler, but did you see these demos? So basically you've always been able to upload a single image and then say, like, animate this and turn this into something. But now you can upload two images, a starting image and an ending image and have VO3 interpolate between them. And so I saw this really cool video that someone made where it was basically a tour of ancient Rome. And so it's flying around through the Colosseum. And the Colosseum gets built up and then it fills with water and then there's boats and you go under the water and there's sharks and then you go into some tunnel, you come out, you're at the across. It was really awesome and I feel like we could do something really creative with this. It really got the creative juices flowing. So excited to build something around that maybe I'll plan out the whole project with linear. Because LINEAR is a purpose built tool for planning and building products. Meet the system for modern software development, streamline issues, projects and product roadmaps.

Host 1 (144:36)

People really were not excited about Starbucks getting into AI. They had other suggestions. Like I saw Ryan Peterson just saying, make the WI fi, make the WI Fi work. Let's go back to the previous era.

Host 2 (144:50)

Also our president Dylan Ebriscato went viral on X quote posting barely AI who is breaking some news about Uber. Uber is going to give its drivers in the US an option to make money by doing digital tasks. These short, minute long tasks can be done anytime, including while idling for passengers. So your Uber pulls up, up, starts waiting, starts doing some tasks to help train the next generation of AI. You can do the data labeling, uploading restaurant menus, recording audio samples of themselves, narrating scenarios in different languages.

Host 1 (145:23)

People also absolutely hated this.

Host 2 (145:26)

Yes, they did not like this.

Host 1 (145:28)

I hope they put some type of restriction in where if they detect that you're in a mo, like actually moving that they say you cannot do data labeling because somebody in traffic just daily.

Host 2 (145:40)

You'Re delegating the driving task to the AI while training the AI on the next thing and you get a task and it's like is this car about to crash? Yes or no. And you look up and it's a picture of exactly what you're looking at. You're like oh no, I should have been labeling this with the pedals instead of the buttons. I don't know. There's a lot of black mirror scenarios, but it does seem like a big market scale AI, obviously surge AI. We've talked to Merkor, we've talked to a few other of these data labeling, label box. Some of these are more focused on the real complex reinforcement learning environments with verifiable rewards. It's more expert driven, but there's still clearly a need for general AI data labeling. Tyler, do you have a take on this?

Tyler (146:25)

Yeah, I mean it was just interesting because I feel like generally the playbook of data labeling has definitely moved up the skill ladder. Totally scale. I don't know what their revenues are now from the same kind of the Filipino.

Host 2 (146:40)

Yeah, it's felt like the job was finished with the general base level RLHF. But I think there's still niche areas where uploading every restaurant menu scale didn't necessarily do that. They're probably still competing with that. When we had the president or the new CEO of scale on, he was saying they just got a new huge contract for $100 million. The DOD. What are they? Data labeling. Probably some stuff that hasn't been labeled before. You know, how much, how much, how many. How many rations do they have in. In stock over time or something?

Host 1 (147:11)

Tyler, you're pretty AGI pilled, right?

Tyler (147:14)

Yeah.

Host 1 (147:14)

Okay, well, why don't you earn $100 doing data labeling?

Host 2 (147:19)

Yes.

Host 1 (147:19)

That's your challenge for. That's your challenge for the next 24 hours.

Host 2 (147:22)

If you believe in Rocco's basilisk. Rocco's basilisk demands that you do that. You do data labeling.

Host 1 (147:30)

Sign up. Sign up right now.

Host 2 (147:32)

Help summon the shoggoth with data labeling tasks. One dollar at a time. You know, I don't know.

Host 1 (147:40)

Matthew Prince said was weird being in Vegas recently. So quiet, so many fewer people. Gambling, drinking, partying. My pet theory, Ozempic killing Vegas just like it's killing snack food brands, liquor producers and Napa.

Host 2 (147:53)

Yes.

Host 1 (147:53)

I didn't know it was impacting Napa.

Host 2 (147:55)

Yeah, I didn't know that either.

Host 1 (147:57)

Obviously. This is just kind of him riffing.

Host 2 (148:00)

Yes, but do you agree with him or do you agree with the rebuttal from near, which you can read.

Host 1 (148:04)

Near says I would go the opposite angle. With the rise of Robinhood prediction markets and sports betting, there is little reason to go to Vegas. It is more expensive, the house edge is higher and it's hotter outside. Less drinking contributes to the.

Host 2 (148:17)

Okay, so do you think it's the rise of online outlets for.

Host 1 (148:22)

Isn't Vegas just like. Isn't it like a recession indicator? Right. Like the real economy is not just not doing well right now? There's no.

Host 2 (148:34)

I would want to know.

Host 1 (148:36)

It's not like, you know, the average person that's going to Vegas is saying, I need. Or the average person in America is, I gotta be in Vegas this weekend. They're scaling AI Capex like crazy. I gotta be there.

Host 2 (148:46)

Yeah. Tyler, can you look up the rough percentage of Americans who are on Ozempic? I wanna know, is it like 1%, 10%, 50%? Because if Vegas is. If it's like 2% of Americans are on Peptides, I would expect that wouldn't show up in the Vegas data, but if it's like 50%, I might see an effect there, right? I don't know. What do you have?

Host 1 (149:14)

The kind of people that go to Vegas are like, you know, the most.

Host 2 (149:17)

Likely to be on Ozempic. Maybe, but let's just get the number. What do you think I'm saying like.

Tyler (149:20)

Around 12% have used it once.

Host 2 (149:23)

Okay. We're not exactly sure how many are insignificant. That could take a couple points off the off the house edge. But the question is.

Host 1 (149:32)

Yeah, the other is clearly under pressure. Yeah, people, I saw another quote that just said like the nightlife scene there has been hyper financialized to a degree that it's just not even fun anymore.

Host 2 (149:43)

Oh, sure, sure, sure. Yeah. Well, if you're running a casino in Vegas, you have sales tax, you gotta get a numeralhq.com, sales tax on autopilot, spend less than five minutes per month on sales tax compliance. But here's my theory for how you bring back Vegas. People aren't gambling for one reason or another. Maybe it's the Ozempic. You need to make Vegas high end. So Vegas needs to bring in an opera house, a symphony, an art gallery. The Louvre should relocate. Maybe they're behind it. Maybe we'll see, you know, fine art come to Las Vegas. This is the only way that they can sustain.

Host 1 (150:15)

Yeah, what if they set up a bunker in there that that has a bunch of stolen goods from all over the world.

Host 2 (150:21)

They already have an F1 race. They need to turn it into Monaco. They need to.

Host 1 (150:25)

I think somebody's gonna look this up and be like, they already have an opera, they already have a symphony. They already have art museums, but they.

Host 2 (150:31)

Should close all the nightclubs and only have opera and only have symphony. I think that would be the true solution. We talked a little bit about the water issue. Where else should we go?

Host 1 (150:46)

FT made a pretty egregious error when trying to calculate the unit economics of neoclouds and hyperscalers.

Host 2 (150:56)

Yeah. What happened here?

Host 1 (150:57)

I don't know if they've issued an official correction yet, but it came out out it was super bearish. And then a bunch of people pointed out that they had gotten the math wrong by.

Host 2 (151:08)

Well, if you want to get the correct data, I would defer to clustermax, the project from semianalysis. They are the most reliable analysts in the space.

Host 1 (151:19)

You said last week the new media, traditional media divide is far too simplistic. If you want to understand things things today, you have to know the difference between legacy Media, Traditional media, new media, Legacy new media, Neo Media, Post Neo Media, Alt media, and Neo Alt media.

Host 2 (151:35)

Yeah, yeah. You really have to study this stuff. It's more complex than punk bands. Like, you can't just say, oh, it's a rock band. There's so many different layers. This is actually a riff on what you said because you used the phrase legacy new media around me. I thought it was so funny. But obviously this is a comment on the colossal debate over Colossus magazine.

Host 1 (151:59)

And I think the reason we were talking off air this morning about this, I think the reason that it sparked a debate is that it's new media that looks and feels like legacy media.

Host 2 (152:10)

Yeah. Unpack this. This was a good take. But first, before you do, let me tell you about Addie Oduck. Custom Relationship Magic. Addyo is the AI native CRM that builds scales and grows your company.

Host 1 (152:20)

My point was that Colossus looks like Fortune magazine. Yeah. It doesn't.

Host 2 (152:28)

Like, it's the Joker Corps. No one cared until I put on the makeup. Right.

Host 1 (152:32)

Yeah.

Host 2 (152:32)

Isn't that from the Joker? What's the line?

Tyler (152:35)

Or the mask?

Host 2 (152:35)

Yeah. Oh, the mask. Right. No one cared until I put on the mask.

Tyler (152:39)

It's Batman.

Host 2 (152:39)

Oh, it's Batman.

Tyler (152:40)

Yeah.

Host 2 (152:40)

Wait, what? Batman says that?

Host 1 (152:42)

Yeah.

Host 2 (152:42)

He puts on the mask and he says, no one cared until I put on the mask.

Tyler (152:46)

Let me fact check.

Host 2 (152:47)

I think this might be the Joker or something. I don't know. Anyway, the idea is, like, people have been writing profiles about tech companies in a positive way. Like, Paki McCormick has had a really successful. Yeah, okay.

Tyler (152:57)

It's Bain.

Host 2 (152:57)

It's Bain. Yeah. Yeah. This makes me. That's right. Bain says no one cared until I put on the mask. Right. And so. And so it's like no one cared about. Oh, it's just a blog. Oh, it's just a substack. Oh, it doesn't matter if it gets a lot of views or a lot of attention and tag. What matters now is that Colossus put the. Put the profile in terms that the other magazines can understand because it looks like theirs. It looks. It's the same glossy magazine. And we're seeing this with Arena Mag.

Host 1 (153:25)

Yeah, it's like legacy media didn't feel threatened until the entrance took on the aesthetics of.

Host 2 (153:33)

And we really need a new term for this because there's traditional media, there's new media, and then there's new media.

Host 1 (153:41)

Neo Legacy media.

Host 2 (153:42)

It's Neo Trad media. We're the Neo Trads. We're the Neo Trads. So this show is new media, but it looks like trad media. You can see the overlays and the stock tickers.

Host 1 (153:53)

Neo Trad.

Host 2 (153:53)

This looks like tv, but it's not. And Colossus similarly looks like a magazine, but it's really not. I mean, it is a magazine, but it's also. But most people consume the Thrive profile on the web, but it's not just a blog, it's not just a substack, so. Fascinating. Well, Tyler threw out an alternative term, adding to the fray of media terms. What do you want to call it?

Tyler (154:15)

Dark media.

Host 2 (154:16)

Dark media? Are we dark media? You're dark media. Look at how dark your background is. You're dark media.

Tyler (154:22)

I think the Doomers, if the Doomers released some new publication, I think that would be dark media.

Host 2 (154:29)

Yeah, speaking of dark, it's dark when you go to bed, sleep on an eight sleep. Eightsleep.com, pod 5, 5 year warranty, 30 hour risk free trial free returns, free shipping.

Host 1 (154:38)

It is so funny that eight sleep runs on aws and so.

Host 2 (154:43)

Oh no.

Host 1 (154:46)

My reporting last night is all off. I was seeing some other. But it's good.

Host 2 (154:50)

I took a fantastic nap on my eight sleep yesterday, so I'm feeling very refreshed.

Host 1 (154:55)

In other news, stitch fix is up 7% today after bill Gurley posted pictures of himself using Stitch Fix's new AI product, Stitch Fix Vision. No, I think Stitch Fix is just popping because they announced a version of you, upload some images and then it'll generate you in different outfits. This is Thrive did a company.

Host 2 (155:23)

Well, that's super interesting.

Host 1 (155:24)

Doji. Doji.

Host 2 (155:25)

I don't know how, I don't know.

Host 1 (155:27)

They basically created like a doji.

Host 2 (155:30)

I don't know the recent history of Stitch Fix, but I remember at one point they were, I think they were employing a lot of human beings to do the collection assembly. So you would send in some photos, you would say some of your opinions, and then they would put together a box of clothes for you and kind of act as like a virtual stylist. And these stylists obviously had real costs to them. And so if that's something that AI can do, like that is a material change to their business and their fundamental economics. So who knows if people will like it? Maybe they say, hey, this particular outfit is slop. But if they have a lot of training data and also they have the whole pipeline of logistics and fulfillment and E commerce set up, I mean, that could be, that could be significant for the business. It's exciting. I'd love to have the founder on and talk more.

Host 1 (156:17)

Cremu says over the last 15 years, Reddit's relationship advice has shifted towards recommending breakups, boundaries and therapies and against compromised communication and letting people have their space. So he did well. There's actually, I guess, a Reddit user that generated all this. This is concerning for a number of reasons. Reasons one being that your favorite AI models are all on Reddit trainers.

Host 2 (156:45)

Yeah, and this is what Pat Gelsinger has been sounding the alarm about when he evaluates the LLMs. He put it in different terms, but he kind of noticed a similar trend. What's your best advice for a relationship? I say start a podcast together. That works great. Andrew osorkin's new book, 1929, has wild details of Winston Churchill YOLO stocks.

Host 1 (157:13)

We gotta expand on that because if you're having relationships problems, whether friend or romantic, and you start a podcast together, that sounds like terrible advice.

Host 2 (157:23)

Here are the options from Reddit. End relationship. Communicate. Give space, time. Set respect, boundaries. Seek therapy, counseling, compromise, Other.

Host 1 (157:33)

So you're saying other is podcast.

Host 2 (157:35)

How about getting the ring? How about getting the octagon?

Host 1 (157:38)

Duke it out.

Host 2 (157:38)

Duke it out. Break a sweat. Break some knuckles. You know, throw down. Throw down. That's the correct answer. Throw down. If you're having relationship problems, throw down. Why is that not on Reddit at all? Redditors haven't even thought of that as advice for relationships.

Host 1 (158:00)

Throw it out. Got to get on there. Trung fan was sharing some highlights from Andrew Ross Sorkin's.

Host 2 (158:06)

Wait, I have one more piece of advice before we go to 1929. If you're having a relationship problem with someone, buy them a luxury watch on getbezel.com because your bezel concierge is available now to source you any watch on the planet. Seriously, any watch. You're like, oh, you know, Tyler seems frustrated with me lately. What should I do? It's like, get him a gmt. Get him. Get him a Patek.

Host 1 (158:28)

Get him a hitter.

Host 2 (158:28)

He's gonna be like, yeah, it's water on the bridge, John.

Host 1 (158:31)

Metal solved all.

Host 2 (158:32)

So let's go to Tyler Cosgrove with his Official review of 1929. Did you really read the whole book this weekend?

Tyler (158:39)

No, I was playing on it. No. Yeah, I'm not done yet.

Host 2 (158:43)

Did you watch any tiktoks about the book?

Tyler (158:45)

No.

Host 2 (158:46)

Did you generate a Sora summary of the book?

Tyler (158:50)

That would have been much more efficient.

Host 2 (158:51)

Okay. Did you actually read any of the book?

Tyler (158:53)

Yeah, yeah. I'm like, close to, like, halfway through Cool. Yeah. I told you guys I was going to be done, but a little white light. But I mean, it's really interesting. It's like, kind of crazy how many parallels there are between, like, the current day and bring it down, 1929. So I think we asked Sorkin a couple of these.

Host 2 (159:09)

Yeah.

Tyler (159:10)

But okay, so one example is this guy Durant. He's kind of this. He's famously for.

Host 1 (159:16)

For just being like this Kevin Durant, to be clear.

Tyler (159:21)

And he's like, thinking about starting this company that is essentially the same thing as a company we've had on basic capital.

Host 2 (159:27)

Okay.

Tyler (159:28)

It's basically a mortgage for buying ETF leverage.

Host 1 (159:32)

Yeah.

Tyler (159:33)

And it's like, there's so many parallels where it's like, oh, that's just like that thing that I saw the news about the other week. So a little bit bearish reading this.

Host 2 (159:41)

Sure.

Tyler (159:42)

But I think also. So, I mean, the main thing in the book is, you see the real reason for all this speculation is basically just. There's just so much new money coming in.

Host 2 (159:49)

Sure.

Tyler (159:49)

Like, so many of the big banks, they're just focusing on trying to get essentially like retail to take leverage or to just get people to buy stocks in general.

Host 2 (159:59)

Sorgen was mentioning the AI equivalent in 1929 was RCA like the radio. Radio waves. Like, does that come through or was the bubble broader?

Tyler (160:13)

Yeah, I mean, so he talks about RCA a fair amount. I think that's mostly as just a specific example of a stock that is being pumped.

Host 2 (160:19)

Sure.

Tyler (160:19)

But it's certainly not like the general fervor is not like, oh, everyone has to get in on this specific company.

Host 2 (160:25)

Yeah.

Tyler (160:25)

That everyone needs to get in on the market.

Host 2 (160:28)

The market.

Tyler (160:28)

The market is just going to keep.

Host 2 (160:30)

I mean, there's a little bit of that happening.

Host 1 (160:31)

Here's some extra context. So Churchill was day trading.

Host 2 (160:37)

Love it.

Host 1 (160:38)

He was trading 400,000 pounds a week on margin, but adjusted for inflation, that's $36 million that he was just slanging around while. Was he actually in power? In power yet? Let me see.

Tyler (160:56)

So this is like he's kind of in some kind of like, ambassador role at this point. But, yeah, basically he. He has this very, like, luxurious lifestyle that he has to keep up.

Host 2 (161:07)

Yep.

Tyler (161:07)

So he goes on this road show throughout the country where he just finds these.

Host 1 (161:10)

One sec. Churchill didn't actually come to power until 1940.

Host 2 (161:15)

Yeah.

Host 1 (161:15)

So, yeah.

Tyler (161:17)

So he's going on this roadshow, basically to. To find rich guys who will just give him money to keep traveling, but also just so that they can speculate and another thing is.

Host 1 (161:25)

Is.

Host 2 (161:26)

So is he kind of like the Leopold Aschenbrenner of 1929?

Tyler (161:30)

I think Leopold has much more of a differentiated view, much more alpha. Yeah.

Host 2 (161:35)

Winston was pure beta.

Tyler (161:36)

Yeah, Well, I mean, it's still alpha in the sense that the amount of, like, insider trading going on is kind of absurd. Like, basically, there's a deal that's about to happen and then the banks just deal out, like, oh, here's a couple friends of mine, you know, here's like, Calvin Coolidge. I'll just give him like, $400,000, basically.

Zach (161:51)

Wow.

Tyler (161:52)

Like, current presidents are, like, getting deals, you know, so. Yeah, so maybe it's not that different.

Host 2 (161:59)

Yeah. Who knows? That's fun. That's fun. Anything else stick out from the book?

Tyler (162:06)

I think once I fully finish it, I'll write something up maybe, and then I'll give a final review.

Host 2 (162:11)

That'd be cool, but very good so far overall. If you haven't had a chance to pick up the book yet, you can get on Audible or wherever books are sold. We're very. Thank you. We're thankful for Andrew Rossworkin to come on the show, share his insight and hang out with us for about half an hour last week. Interviews up.

Host 1 (162:28)

And it's worth noting that Churchill made it all back with a book deal he did on his World War II memoirs.

Host 2 (162:35)

Okay, here's a potential recession indicator from Jacob Silverman. Says the number of people who registered to take the LSAT last year in 2024 was 18,000. But in September of 2025, a year later, the number of people who registered to take the lsat is now 32,000. So a massive, massive uptick.

Host 1 (162:56)

Yeah, these are. The idea is that these are people that were trying to enter the job market struggling and are trying to buy. Basically buy time, maybe, or.

Host 2 (163:08)

Oh, yeah, new grads.

Host 1 (163:09)

Not finding. Yeah, just not finding opportunities that are.

Host 2 (163:11)

College graduation, who say, hey, maybe I'll. Or alternatively, these are people who watched Andre Karpathy on Door Cash and said, well, AGI is 10 years away. I can go be a lawyer. I'll take the LSAT now, get through law school, become a partner, and then I'll be able to cash out before AGI hits. So, you know, LLMs aren't going to be able to do law anytime soon. It's lop. Heard it there first. And so they all apply to the lsat. Could happen. Could happen. Anyway, let me tell you about Wander. Find your happy place. Book a wander with inspiring views, hotel grade amenities, dreamy beds, top tier cleaning and 247 concierge service. It's a vacation home, but better folks.

Host 1 (163:50)

Did you see Senate Republicans posted a video of Chuck Schumer? Yes, that was fully. It was a real quote, but it was fully AI generated.

Host 2 (164:01)

It made AI video and audio.

Host 1 (164:03)

Can we try to pull this video out, see how sloppy it is?

Host 2 (164:06)

Yeah, they made an AI video and audio of the actual quote to kind of illustrate it to the audience. Andrew Curran says, first time I've seen this. This is a real quote. But it wasn't set on camera. So they generated a video of Chuck Schumer saying it. Now, I would love to know what model they used because a lot of models have guardrails here. Let's play the video right now. Chuck Schumer thinks playing with America.

Ryan Cohen (164:33)

Every day gets better for us. Women, infants. Every day gets better for us.

Host 2 (164:38)

Okay, okay.

Host 1 (164:39)

The mouth movement is a little.

Ryan Cohen (164:40)

Chuck Schumer thinks every day gets better. Every day gets better.

Host 2 (164:44)

I mean, like, it's clearly based on some real image that they took. And so the lighting looks very real. The skin texture looks really real because the, the AI model doesn't really have to do that much to regenerate it from scratch. But of course, we've seen that, you know, all the models are basically, you know, indistinguishable if you don't have full context at this point. Awesome says we're going to see some AI laws come to fruition as fast as they push the COVID bill through. Back In March of 2020, the Democrats introduced the bill negotiations. Trump signs it on condition of ending government shutdown. That's kind of a, kind of a bold take. I don't know if that will for sure happen, but you could definitely see before the next election, both sides wanting some clarity on where the line is in terms of what you can use. Now. The nrsc, the Republican group, the Senate Republican group that created this video, did put an AI generated tag in the bottom right hand corner. And so they did disclose that they used AI. Will that be required? Will there be certain definitions of how aggressive the watermark needs to be? That's all up for debate, but we will continue to check it out.

Host 1 (166:04)

More insight here from Barbarian Capital. Thought we were out of noun plus capital names on X, but we always find more. They pulled up a chart here from, from Schroder's. It's looking at average year to date performance of companies in each category. They're looking at nasdaq, no revenues, MAG7 unprofitable NASDAQ unprofitable US small mid caps, NASDAQ with revenues, the S and P and profitable NASDAQ and then profitable small mid caps. What group do you think performed the best year to date, John?

Host 2 (166:44)

I have no idea.

Host 1 (166:45)

Nasdaq. No revenues.

Host 2 (166:47)

No revenues.

Host 1 (166:48)

Yeah.

Host 2 (166:49)

Wow. I was.

Host 1 (166:50)

Nasdaq obviously has much looser listing requirements than nyse. And anyway, so.

Host 2 (167:01)

In the HBO show, like, you don't want revenue because then people will comp you. But now we're doing it. The public markets. That is crazy. And I mean, that goes back to the 1929 thing that Tyler was talking about. Like, once you get to the level of fervor in the market where it's like you just got to be in the market and you see froth, like in stuff that's not even tied to the core innovation. It's like, wait, like AI is clearly like, you know, there is a transformative technology, there is growing revenue. Now there's debate about the level of. The level of Capex. That's correct. But that's wildly different than looking at some company that's saying like, oh, we're going to colonize the moon or build a quantum computer or do something else that's like completely off the current tech tree. And they're mooning that. Not, not to be too punny about it, but that's when you get into like real, real bubble territory. Well, I have one final post. Do you have anything else? One final post. Congratulations to Everett Randall, former partner at Kleiner Perkins. Now he's a general partner at Benchmark. He has announced his trade deal, potentially the trade deal of the year. EV Randall is out of KPOP this week. That would be amazing.

Host 1 (168:16)

What's your plan? What's the plan?

Host 2 (168:17)

And of course, Delian is already taking shots, putting him in an AI generated image without a watermark. So I think everyone can tell that this is an AI generated image.

Host 1 (168:28)

There's one person out there that doesn't. I don't know who it is.

Host 2 (168:33)

Maybe. Well, Everett's playing along in Life laughs at Delian's post because they're both having fun on the timeline. But congratulations to Ev, of course, and good luck with the next era of his investing career.

Host 1 (168:46)

One thing I missed about the MAGA SPAC is that Chamath will be in this one too.

Host 2 (168:51)

No way.

Host 1 (168:52)

With Donald Trump Jr. And Laura Ingram.

Host 2 (168:55)

I wonder. So this is just a spac. They haven't identified a target. Have they even defined. We need to dig into and understand, like, have they even defined an area that they would want to take a company public through.

Host 1 (169:09)

Well, they found an area that they would want to incorporate the Columbia Acquisition Corp. And that is the Cayman Islands.

Host 2 (169:17)

Okay. Yeah. So they're in the Cayman Islands. I think that's standard practice, but I don't know.

Host 1 (169:21)

Yeah.

Host 2 (169:21)

Anyway, thank you for tuning in. Thank you for watching. If this was the first show you've seen because we had the CEO of GameStop on today, please subscribe. Follow us on X. Follow us on YouTube, Instagram, LinkedIn, wherever you get your live streams.

Host 1 (169:36)

Shout out to our whole production team. They were fighting for their lives.

Host 2 (169:41)

We were under attack, probably by a state actor. We don't know yet. We will get to the bottom of it and we will bring you the news tomorrow.

Host 1 (169:49)

They could take down our show by taking down aws.

Host 2 (169:52)

Nice try, nice try, nice try again. Try again. Don't actually try again, please. And you can also subscribe to our newsletter at tbpn. Com. Thank you so much. Leave us 5 stars on Apple Podcasts and Spotify tomorrow. See you tomorrow.

Host 1 (170:06)

Massive day.

Host 2 (170:06)

Massive day.

Host 1 (170:07)

Palmer, Lucky, Palmer, Lucky and Brian Chesky and probably a lot more. See you then.

Host 2 (170:11)

Goodbye.

Host 1 (170:12)

Cheers.