AI Buildout Meets Capex Wall, The Browser Company Effect | Drew Houston, Jacob Andreou, Adam Fry, Ian Rogers, Molly Cantillon, Jonny Dyer, Mike Shebat

You're watching TBPN.

Today is Thursday, October 23rd, 2025. We are live from the TBPN UltraDome.

The Temple of Technology. The Fortress of Finance. The Capital of Capital.

Did you see this graphic in the Wall Street Journal, the business and finance section, talking about the circularity of AI deals? And I thought it was just a beautiful image that reflected nature. It looks like a conch shell.

Looks like a conch.

It looks like a conch, Max. Right? It looks fantastic. I was laughing to myself about Satya Nadella just being like, yeah, I was into circular deals with OpenAI like 6 years ago, actually. So Larry Ellison, you're not— yeah, you're not as cool. Don't take a victory lap. I was doing crazy complicated deals with OpenAI back in 2019. But, you know, we've talked enough about OpenAI on this show. It's quickly turning into the OpenAI show. We have someone from OpenAI coming on the show later today, actually, to talk about the browser. I am excited for that. But we gotta go a different direction. We gotta talk about some other stuff.

Okay.

And so I want to talk about how we name companies, how we build brands in Silicon Valley. And I want Jordy to take the lead on breaking down his thesis for the uninspired company of Silicon Valley. But first, I want to tell you about ramp.com. Time is money. Save both. Easy-to-use corporate cards, bill pay, accounting, whole lot more, all in one place. Jordy Hayes, you have the floor, sir.

Thank you.

This is an idea or an essay I've been thinking about for a while. Finally got around to writing it. And yeah, generally, I felt like the timing was right. The Browser Company's acquisition closed this week. 6 years ago, they started the company. So I'll just read through what I wrote, and then we can chat about it. So 6 years ago, The Browser Company of New York was born. This week, its acquisition by Atlassian closed. Regardless of how you feel about the product or the ultimate acquisition price, it's undeniable that Josh Hirsch and the team brought an incredibly fresh perspective to what a startup brand could look and feel like.

6 years is kind of crazy. I feel like I found out about The Browser Company like 2 years ago, maybe 3 years ago. What a grind. Congratulations to the team over there.

Overnight success.

Seriously.

The name, the brand design, et cetera, were all incredibly thoughtful, but they weren't new. Roughly 150 years ago, it was standard practice to name a company like they did. The Prudential Insurance Company of America, founded in 1880.

Standard practice sounds like a good name for a company. That sounds like a real name.

Standard Oil Company of New York, founded in 1911. That was a spinout. The Edison Electric Illuminating Company of New York, founded in 1880. And there was a bunch of others over history, again, this was just like the way that you named a company.

Yeah.

You weren't trying to think of— back then you weren't trying to think of like a cute dot-com. No, it was, why don't we just name it what it is? And so what I'd like to do—

name it what it is or name it who we are. Goldman Sachs. Morgan Stanley.

So what was, what was amazing about this naming convention, this isn't in the essay, is just like how explicitly clear it is what the company does. The Edison Electric Illuminating Company of New York.

Yep.

I can imagine what they do.

Yep.

It's good marketing.

Pretty, pretty simple.

So, The Browser Company of New York was a perfect name for a specific reason. Juxtaposing a 150-year-old naming convention with a modern tool such as the web browser was an incredible way to stand out and signal to the world exactly what their mission was and that they would be bringing inspired thinking to the category. The Browser Company's name, brand, and marketing materials were so effective that they catalyzed a wave of companies to adopt the same naming convention. Between The Browser Company's emergence and their eventual exit, I'd estimate that somewhere between 50 to 100 companies adopted this type of legacy naming convention. So naming a web browser company the Browser Company of New York signaled this sort of original thinking. And the problem is that the 2nd, 3rd, 4th, 5th, 6th, et cetera, company to use this sort of convention basically does the exact opposite, right? It signals like, OK, I saw somebody do something cool in my industry. I'm going to do the cool thing as well.

Yeah, the genericization, the making that that whole pattern generic. Like, I was not tapped in. Clearly, I did not know that The Browser Company in New York was 6 years old. I found out about it a couple of years ago. And there are people that probably found out about it after they found out about 5 copycats. And so, in their mind, they're just like, oh, The Browser Company in New York, that's one of those copycats that went all around.

Yeah, but it quickly became an anti-signal.

Totally, totally. I completely agree.

Personally, I'm not automatically bearish on the companies that are kind of like a little too inspired on this front. But I think a lot of the people are kind of missing what made it great, right? And so at least one of these companies, the Interaction Company of California, managed to break through the noise and deliver a truly novel consumer AI product experience. But ironically, they did this under the Poké.com brand name. This is a great lore. So Poké came out, they had an incredible launch, very cool, sort of novel consumer AI experience. And so my point here is they should probably just abandon that original name and just run with Poké, 'cause it's super memorable. And I think it's perfect, right?

This is what Instagram did. Instagram was Burbn originally.

Yep.

And then once Instagram hit, they were like, we have a new brand. We have a cool thing. And then of course everyone copied Instagram. Interestingly, Instacart copied Instagram's name and it worked.

Yep.

So yeah, not always a total anti-signal, even if it's—

yeah.

Yeah, so in defense of copycat branding, naming startups is really hard. Every founder's gone through this. There's only so many domain names. There's only so many English words that mean something, and there's just so many companies being created. It's quite challenging. Another defense and kind of like a misinterpretation, people misinterpret this quote that's attributed to Picasso, which is, good artists steal. Great artists— good artists copy, great artists steal.

Picasso said that originally? I know that there's a book—

There's a variety of people that have said it throughout history, stolen, remade.

It'd be so funny if he legitimately copied it word from word for his direct rival, like another painter.

So yes, taken literally, if somebody just says, okay, they're going to see like, okay, this is good, I'm going to do the good thing. that someone else is doing. But a better interpretation of it, and my interpretation of it, is that you're better off stealing from the past, but recombining inspiration from a variety of sources to create your own unique style and approach. Yeah.

Before you continue, great— good streamers stream, great streamers Restream with Restream.com. One live stream, 30+ destinations. Continue, Jordy. Sorry.

So yeah, so the browser company hit so hard because there was this crazy juxtaposition, and I don't think anyone had done it before, so it got this amazing response, right? It felt fresh and new. But after they executed against it so perfectly and so loudly, you can debate, was the product truly a hit? That's up for debate, but the brand was— The brand punched through, for sure. The brand went through. So if you just go after— if you just sort of clone, To me, it signals that you're in a rush, which I think is fair. A lot of companies are in a rush. You don't value naming, which is concerning. I think names matter a lot. Didn't want to spend the time to find a great domain, which is fine. Again, a lot of founders don't have a broker like our buddy Rob at Snagged.

Yeah.

But it also could signal you aren't seeking out inspiration from outside of the tech bubble, which I think is concerning. The world's a big place. You should go outside of of our little bubble. And then you didn't seek out influence or advice from people that understand the value of naming, know how to acquire great domains, because you can go to founders that have been in the game for 10 years, and they will have a domain guy. They'll give you the playbook. They might even have a domain name lying around that's great, or you could talk to a VC, and they're like, oh, this portfolio company was using this name 10 years ago. You should pick it up and use it.

They might even have a billboard for you. It might have been on adquick.com. Out-of-home advertising made easy and measurable. Say goodbye to the headaches of out-of-home advertising.

Couldn't have said it better myself. Yeah. So in this case, browsers were all modern, right? You had Netscape, Edge, Chrome.

Yep.

And so it was juxtaposed, but it was also counterpositioned, right? And of course, the browser company naming saga, it's not an isolated incident. I brought it back to Linear. Linear launched in 2019. The product design, the website was so good that people were just like, I'll take one, please. One Linear brand, please. One Linear dashboard, please. I even talked to a founder earlier this year, and he was like, yeah, our website and product is going to look exactly like Linear.

He said that, admittedly?

I was just like—

That's insane.

Dude, there's so many— I mean, it's great, but there's probably something better if you just really thought hard and spent the time to think through what could this be? And so anyways, I think copying great design signals that you don't value design, in my view.

Yep.

Yeah, that's a good point. And so what this comes down to is that I believe the tech industry needs to learn how to copy or steal from outside the industry. Even today, every AI company wants to be the Apple of AI. I think you see this in some of the campaigns that they're doing, which are good campaigns, but sometimes they feel like, okay, You guys went to an ad agency and said, give me one '90s Apple ad, please. Give me one '80s Apple ad, please. And so I said, it's counterintuitive, but I bet the Apple of AI will probably not build an iconic generational brand by trying to emulate Apple advertisements from the '80s and '90s. They'll do it by being themselves, or said differently, stealing from the past, a variety of sources and a variety of categories to create their own unique style and approach. It's perfectly respectable and even fair to take inspiration from obvious sources and industries. We at TVPN have been vocal about being inspired by ESPN, SportsCenter, Complex, and others. But the key is that we took that inspiration and applied it to an area, tech, that none of those groups had ever played in, right? So we weren't copying SportsCenter to build a sports podcast, right? So I said, if you're starting a company today, I urge you to take inspiration from the outside world and other industries and avoid the trap of becoming an uninspired company. So happy to get this out there. And again, I love— I truly love— I feel lucky to love the hunt for a great domain.

For sure.

Yeah.

I'm thinking about the different— when you say every AI company wants to be the Apple of AI, that feels like OpenAI and Anthropic's most recent campaigns to me.

Well, yeah.

And even OpenAI's Super Bowl campaign. Right.

That didn't feel Apple to me. I don't know. Like all the pixelated, like the dots and stuff, that felt a little Apple-y. I don't know. That didn't really spring out to me like that. But I was just thinking back on the DeepMind documentary on the Lee Sedol match, the AlphaGo series. culminates in Move 37. Like, that was part of the brand building. And it's odd because typically, I think they paid for it themselves. Like, typically paying for someone to make a documentary about you as opposed to just like waiting for someone to make the documentary about you, it looks like it gets a lot of criticism as like, oh, this is just marketing. But At least in that particular documentary, I think a lot of people have watched it, and I think it was very effective marketing. And I think it actually built DeepMind's brand in that moment. And I feel like if you could turn back the clock, OpenAI would've really benefited from having a documentary, paying for a documentary on the Dota match that they did, or some of the other initiatives that they were working on during the OpenAI Gym, the original reinforcement learning stuff that they were doing, because— The AlphaGo documentary, you've seen it, right, Tyler?

Yeah, yeah, many times.

Many times?

Well, probably like 2 or 3 times.

2 or 3 times, yeah. And so that's how—

He didn't just watch, he studied.

Yeah, you sat your ass down and listened. Isn't that how you—

Did you smoke a heater while you watched?

Yeah, he said he sits down and goes, can you smoke?

But I feel like that was the introduction to Demis. And it really takes you through, because it opens with him talking about video games, talking about artificial intelligence, and it felt like it built him up as a character in many ways, more than just him going on podcasts.

Yeah, that's probably true. It was also my introduction to the game of Go, and I've since become a big Go player.

You play Go a lot?

Yeah, a fair amount. I mean, none of my friends— no one knows how to play Go, so I always have to teach them. But I've gotten a lot of my friends into Go. Yeah, I mean, it was definitely the first time Demis was really like— I think a lot of people watched that who were outside of tech.

Yeah, no, that makes sense. Well, before we move on, let me tell you about getbezel.com. Shop over 26,500 luxury watches, fully authenticated in-house by Bezel's team of experts. Dwarkesh Patel has dropped an essay with a friend of his, and Tyler, you had some analysis on this. Did you get a chance to read through his thoughts on the AGI build-out with Romeo Dean?

Yeah. So, so basically, The main point of the article is they're trying to address this question that— or address this thing that Sam Altman said, which is that at some point they want to be doing basically a gigawatt in infrastructure per week.

Yeah.

New infra or whatever.

Yeah, 1 gigawatt per week.

So it's like, OK, how do you get there? What are the bottlenecks? How does this actually work out? So there's a couple things he points out. The first thing is there's this idea of the fab CapEx kind of overhang. So if you compare like Nvidia revenue to the CapEx of fabs, you'll see that like the revenue kind of like dwarfs this. So there's like actual specific numbers, but it's something like this past year of Nvidia revenue is more than the past like 3 years of TSMC's entire CapEx.

Yes.

And if you go up the stack, if you're looking at like ASML or Applied Materials, like all of those that are kind of just like a bit kind of lower level than TSMC, you see that Nvidia revenue basically accounts for like the past 25 years of R&D and CapEx of those companies.

Yeah.

So basically the point is that if Nvidia like really wanted to, they could basically fund like all these new fabs without like, you know, they just like increase the margins for these companies by a ton, but they could just like spend and just Have these fabs like grow massively?

Yeah. So this, I mean, this doesn't feel like an answer to the question that most people are asking, which is about like the circularity of deals, the buildout of the actual data centers. I haven't heard a lot of people like nervous about the nature of the fab buildout. No one's saying like, oh, TSMC is over their skis. Like it's Oracle. that people are worried about, right? And so the fact that TSMC is not overbuilding because the $6 billion in CapEx turns into $200 billion in revenue, that feels like, wow, congrats to TSMC. They're in a safe place. They can basically go to the market, go to OpenAI and all the hyperscalers and say, oh, you guys want to put up $1 trillion in spend or revenue? We got you, you're not gonna put us outta business if you don't deliver, which is great for them. It just feels like TSMC's in a really strong position.

Yes. Yeah, that's true. But I think Dwarkesh is saying like they're underbuilding if we wanna be on these like timelines, right?

Oh.

So like if we wanna be doing a gigawatt per week, like how do we actually get there? So the fabs need to be producing way more.

So yeah, this is good because it's basically like what is the shape of the overbuild? What is the shape of the buildout? And where you don't wanna wind up is just like some sort of like fat, overbuilding like data centers and you're building like these massive data centers, but then you haven't built any nuclear power plants and then you also haven't built any fabs. And so you have like a bunch of data centers with nothing to put in them and nothing to power.

Yeah. The fabs are really important because chips are something like 60, 70% of the actual like CapEx cost of a data center. Like they're just, there's so much like compared to power or—

Yeah, but everything else. I mean, I wonder how much they're rate limiting on the actual build out. Like, it seems like at least some of the narrative has been there's a much longer lead time for energy infrastructure than there is for TSMC line time scale-up fabs. But also, I haven't actually seen anyone break down like this what it takes to 10x production at TSMC. Because I don't know where their capacity is. Are they close to max capacity? Is TSMC running 24/7 already and they haven't— and they don't have any other facilities?

I think probably, but what Dwarkesh is saying is that like their CapEx generally is just very small compared to like what it could be compared to— obviously Nvidia is like way higher, but like if Nvidia, you know, makes some deals here, like there could be A massive increase in CapEx. And then he says, well, maybe that won't happen with TSMC, but maybe, you know, with the Intel deals, like this is kind of what is being done.

Yeah.

Yeah.

Wow. Interesting. So he says further up the supply chain, a single year of Nvidia's revenue almost matched the past 25 years of total R&D and CapEx from the 5 largest semiconductor equipment companies combined. So you take ASML, Applied Materials, Tokyo Electron, SK Hynix, Micron, TSMC. You add all of those together, it only takes like a year or 2 to actually pay for that just with what Nvidia is actually producing because of, because of the margins and the actual ROI on the CapEx. The reason we're emphasizing this point is that if you were to naively speculate about what would be the first upstream component to constrain long-term AI CapEx growth, you wouldn't talk about copper wires or transformers, you'd start with the most complicated things that humans have ever made, which are the fabs that make semiconductors. So every time you see someone saying, oh, you know, worry about the transformers and the energy infrastructure, it is potentially more bottlenecked by the fabs. We were stunned to learn that the cost of these, the cost to build these fabs pales in comparison to how much people are already willing to pay for AI hardware. Nvidia could literally subsidize entire new fab nodes if they wanted to. That's interesting that we're not seeing those types of deals. It says, we don't think they will actually directly do this, or will they? Wink wink, Intel deal. But shows how much of a fab CapEx overhang there is. For the last 2 decades, data center construction basically co-opted the power infrastructure left over from US deindustrialization. That's where you get You know, the Memphis site was like a former washing machine factory or something like that. And they had a lot of power 'cause they need a bunch of machines to move all the different washing machine parts in there to build the washing machines. But of course, that actual business has moved overseas. And so Elon was able to go in and say, gimme that power. Google's first operated data center was across a former aluminum plant. Hyperscalers are used to repurposing the power equipment from old steel mills and automotive factories. This is honestly, A compelling ode to capitalism. Let's hear it for compelling odes to capitalism. As soon as one sector became more relevant, America was quickly and efficiently able to co-opt the previous one's carcass. But we are now in a different regime. Not only are hyperscalers building new data centers at a much bigger scale than before, they are building them from scratch and competing for the same inputs with each other, not least of all, not least of which is skilled labor.

Ryan says they should rebrand as the Semiconductor Company of Taiwan. To their credit, it's still the Taiwan Semiconductor Manufacturing Company. But yeah, flip it for the modern era and just copy the browser company.

Yeah. Well, if you're looking to invest in one of these companies, go to public.com. Investing for those that take it seriously. Multi-asset investing trusted by millions. Maybe you're deploying $2 trillion in AI CapEx a year. Someone else needs to be producing be producing $2 trillion worth of all other data center components, not just the chips. The people upstream of data centers. The people upstream of data centers, the companies producing everything from copper wire to turbines to transformers to switchgear, we need to build more factories. The problem is that those factories have to be amortized over many— a many-decade lifespan. And at usual margins, those factories are only worth building if this AI demand lasts for 10 to 30 years. There's an interesting— so he goes into some of the the gas turbine manufacturing stuff and ARR projections. The really interesting takeaway is the final chart on AGI timelines, the distribution. Where is this electricity? Here we go. So not only does China generate more than twice the electricity than the US, but that generation has been growing more than 10 times faster than in the United States. The reason this is significant is that The power buildout can now be directed to new data center sites. China State Grid could collaborate with Alibaba, Tencent, and Baidu to build capacity where it is most helpful for their AI buildout and avoid the zero-sum race in the United States between different hyperscalers to take over capacity that already exists.

China will basically run the experiment of we have worse chips, but we have like an incredible amount of energy.

Yeah.

Efficiency doesn't really matter.

Yep. And so, basically, he sums all this up in this chart, which is the probability density, like the probability of the US hitting AGI over time and the probability of China hitting AGI over time. And the conclusion is basically like, if AGI is coming soon, America's good because we have a head start in the race, but we're running slower. China has a slow start, but they're running faster. And so, if AGI comes in 2035, China has a higher probability of it emerging there. I guess the question is like, how bad is it to be a year behind? How bad is it to be, you know, just 6 months behind if AGI emerges? If you believe in like the fast takeoff of like once it's there, you know, it's like the immediate win condition, you better be advocating for some American electricity.

But yes. There was a reference to Tyler Cowen's post from 2023, do not underestimate the elasticity of supply. This was in the context of energy. So Tyler's post said, he said, when I first read about the discovery of a vast new deposit of lithium in a volcanic crater along the Nevada-Oregon border, I can't say that I was surprised, not because I know anything about geology, but because as an economist, I'm a strong believer in the concept of elasticity of supply. Now, about elasticity of supply, in which economists tend to have more faith than do most people, time and again over the centuries, economists have observed that resource shortages are often remedied by discovery, innovation, and conservation. All induced by market prices. To put it simply, if a resource is scarce, there's an upward pressure on its price. New supplies will usually be found. So you can reference this with people's energy concerns around AI. The big question is like, where's the energy going to come from for all these new data centers and planned CapEx? But Tyler Cowen's willing to bet that we will discover new sources of energy. We'll figure out how to make more gas turbines. There was some reporting. I think there's already like an immense amount of pressure on the Colossus data center development in terms of— I saw Wired was posting a video. I sent it to you yesterday.

Yeah, I watched it.

Like somebody using this like thermal camera or some type of camera to view how much gas was being like let off. And it was just like, clearly they're just Have, have all these gas turbines firing.

That was the— there was some good stuff in, in, in that video, but that was the dumbest part because it was like Elon has been publicly saying like, this is the biggest data center with the most energy.

Yeah, it was, it was framed as a bombshell. Yeah, it was framed as a bombshell.

We pointed a camera at it and it's big. It's big.

It's the big—

it's bigger than everything else.

It's like, yeah, it was framed as a gotcha. But Elon was saying that.

Elon is extremely proud of how much energy this thing uses. He actually posts the exact energy figures constantly. But I did think that we have Saagar Enjeti from Breaking Points joining the show tomorrow, and I think that he lays out a really good case for like what public sentiment is like. And you can see in that video there's a politician from Memphis who is like, whoa, whoa, whoa. Like, if you guys are going to come here and build a massive data center in Memphis, like, you have to make sure that the water stays clean. that we don't wind up with really polluted skies. We don't wanna be Beijing. We don't wanna be, you know, some industrial capacity, like we, some industrial hub. Like we've lived through that. We've seen like, you know, Detroit was at one point dirty. The industry sort of left, but like we want like a more modern, futuristic, eco-friendly. It's not to be like a total decelerating, like tree hugger. It's just like, let's not give each other cancer, people.

Yeah.

Like there's probably a way around this. And so I think the, The actual discussion is gonna get super, super heated. And I think Sagar's correct, and we'll unpack this with him more tomorrow. The discussion will get extremely heated and it will turn into these extreme ends of like, you want us— you don't wanna cure cancer with AI and like you wanna kill everyone with, you know, this, you know, this insane pollution that's coming out of whatever natural gas turbine.

Surprised we haven't seen somebody Like protesting on top of a data center yet?

Yeah.

You'd think somebody would've scaled one and—

Well, it also—

People used to do this with trees, right? They would just climb to the top of a tree that was gonna get cut down.

This happened at Anthropic, right? But for doom reasons, not for environmental reasons. And so there's like a very clear alliance that should form, right?

Happened briefly. He was hunger striking. He got hungry.

He was there for a couple days. But I mean, no one debates that there are tons of AI doomers that are actively protesting the lab in one way or another. Like, I would put Eliezer Yudkowsky's new book, If Anyone Builds It, We All Die, that's what it's called?

Yeah.

I would put that in the AI protest canon. And it would be interesting to see if there's a team of rivals that emerges with Eliezer And then just the folks on the ground who are like, look, I actually think that AI is slop. I don't think it's going to be super intelligent or kill me. I just find that my electricity prices went up and I don't like that. So we're on the same team, like strange bedfellows, basically, where they believe different things, but they arrive at the same conclusion. So they team up.

Same cause.

Anyway, if you want to go visit Memphis, book a Wander. Book a Wander with inspiring views, hotel-grade amenities, dreamy beds, top-tier cleaning, 24/7 concierge service, You really should, if you're a true technologist, go on a trip to see a data center in person.

I don't know if they have anything in Memphis, but they have Pigeon Forge, Tennessee.

Oh, OK.

It's the Wander Smoky Woods, and they have Wander Smoky Peaks.

Truly, that's probably like an amazing wander.

And Wander Gap, the Great Forest.

That's probably an amazing wander.

And you can probably see smoke rising from colossus.

Colossus.

It would be incredibly— it would be incredibly Like wild if Colossus just started burning so much like dirty coal that it was just like this black like smoke rising. Yeah, I think, uh, they won't go that far.

Yeah, yeah. Just what about just a good old-fashioned wood-fired data center? Tyler, can we figure out what it would take to have it— have you over there? I want you to have a little, uh, We'll build you a chimney in the UltraDome, and we'll build you a hearth. And you can stoke the fire with wood and cedar logs. And then I want you to capture the heat, turn it into steam, turn the flywheel to generate the electricity to run a fine tune of GPT-OSS. Work through exactly how that would work. Is that possible? How many logs would you need? A wood-fired data center. Wood-fired data center, which is like the— you know how good a warm— it's almost fall here. We're cozy maxing. We're already in brown and green. As far as I'm concerned, the holidays are here. And nothing would get me more— it would really bring the Christmas spirit forward to have a wood-fired data center.

Yeah, generating tokens the old-fashioned way.

Exactly. Just hanging out at the hearth. Generating some tokens.

Can you imagine late nights with the fellas just tossing wood in the hearth?

Yeah.

Yeah.

Just, well, you know, that's like super intelligent.

I'm pretty sure that's like how a steam engine works. Like, I'm pretty, I mean, there's, I guess it was coal, not wood, but, uh, if you're on like the, the Titanic, you're, you're down in the boiler room and you need to boil water to turn the spring, the, the, to spin the turbine. Uh, you're taking coal and shoveling it into the fire to make heat, to boil water. And there's no reason why we can't be doing it here. There's no reason.

We got to go to the chat.

Except all of the environmental regulations.

Adam Hoopes says, wife is going into labor. I'm between Jordy and John for my son's name. If Jordy or John can't decide, I'm letting Sam Altman spin up a model and name him himself. How about Data Center?

No.

Data center hoops.

What about Adam Jr.?

Adam Jr.

I think Adam's a great name. Adam Jr., we'll call him Junior. Or have we talked about this?

In all seriousness, Adam, just a true congratulations. Congratulations.

This is incredible news.

And log off.

Yeah, log off.

Log off. We'll still be here. You can catch the recording. But we're excited for you and your family.

I did actually talk to someone who— Who had a baby and the new mom and the husband threw on TBPN when they were in recovery. Because oftentimes you're hanging out and you need something to watch. And they were like, yeah, let's watch it.

Why not?

It's going to blow your mind when I tell you who it is too. Because it's a very funny couple. Anyway, yes, naming the kid AGI would be funny. Aggie for short. That would be— That'd be great.

Tyler is also a strong name for a baby boy.

Yes.

How about 2.6 million? Isn't that somebody in the chat?

Yeah.

2.6M. Anyway, the timeline was in turmoil a little bit yesterday, near Cyan and Rune going back and forth. Before we dig into that, let me tell you about Addeo. Addeo is the AI native CRM that builds, scales, and grows your company to the next level. Stick to the CRM. Maybe stay away from the short-form video consumption. It would be this would be the argument from from near. Basically, people are going back and forth with Rune. Where does Rune stand on Sora? Where does Rune stand on slop? And so short-form video.

The Rune deleted his post.

He did. He did.

So I don't know what it said.

He committed seppuku effectively. It's a it's a harikari on the timeline that we experience. Deleting your post is— it is—

A sign of honor.

It's a technology— yes, it's a Technology Brother ritual, effectively. But we have some traces. We have some fossils that we can dust off.

We might have a screenshot. So what was the general idea of what he said? So we can get into this comment from Nir.

He was saying He was saying there's sort of a moral panic about short-form video. I was kind of saying that with him too. And I think there was like, there was a big like, this all started with like the Meta Vibes thing where everyone was like, this is brain rot. This is gonna one-shot everyone. Get ready to die. This is the worst thing ever. And my like, I was like, yeah, it is kind of sloppy. It is sort of Infinite Jest-y. But like, there is like the possibility that this is just not popular, that like it is possible that like the slop trough just doesn't taste good and people just actually want to do something else. Like chess, AGI—

It was so obvious. It was so obvious on day one that the product was kind of cool.

Yeah.

But it was— it may as well have been a playlist of like cool AI videos. Yeah.

Like people refuse to believe that there is at least a probability that in 50 years people are still watching humans create content. Like, I don't know, there is a probability of that. It is possible. I think most— I think many, many people in like traditional media, just in not AGI-pilled crazy world would be like, yeah, 100%.

The more people live online, the cooler it will be to see a dude like wingsuiting. 'Cause it's like—

Totally, totally.

Okay, we're, you know, 99% of people watching this are just inside on a screen.

Yeah, yeah.

And this dude is flying 20 feet away from a mountain. Yeah.

And so is there a possibility that the Meta Vibes app becomes so addictive you can't turn it off and it destroys your life?

Maybe.

I've read the sci-fi. I would put a probability on it. I would say it's a couple percent that that happens, that it is bad. But there's also just the possibility that people just are like, nah, I'm not into it, right? But people weren't even— No one was saying like, hey, this might just not work. Everyone was like, Meta Vibes will kill everyone. And that was kind of the mood on the timeline. So Rune was reacting to that saying that there's a moral panic about short-form video. It's not actually that bad. He's going back and forth. Some people were pressing him on like, don't you think this is bad for kids? He was like, I don't actually know that it's that bad because I watched a couple hours of TV a day as a kid. Is short form— is watching short form for 3 hours a day fundamentally different than watching TV for 3 hours a day? Because everyone will say, oh, the real trade-off is like reading James Joyce versus TikTok. And it's like, that was not what was going on in the '90s.

Yeah. And what does television do, right? They're like, they're using hooks, right?

Totally.

An episode of a television show starts and it starts off maybe dramatic and it pulls you in. And then right before an ad—

All of a sudden it'll slam you an ad for fall. The world's best generative image, audio, and video models all in one place. Develop and fine-tune models with serverless GPUs and on-demand clusters. Yes. So we don't have a way to quantify brain rot, right? Like, is Barney with ads on it on Nickelodeon— I'm getting a lot of things wrong.

Actually, actually, we do have a way.

Is that more brain rotty?

We do have a way.

Okay.

I have twins.

Yes. Oh, A/B testing.

We can A/B test them.

One gets—

I refuse, actually.

Yeah, I mean, this, this you could I mean, it would, it would, you would effectively be potentially sacrificing a son for humanity.

Yes.

But imagine one son gets 5 hours a day on the iPad from a young age in the Sora app, in the Meta Vibes app, just scrolling. The other is, is gets, you know, personalized tutoring from, you know, I mean, we also have Tyler.

Let me volunteer. Maybe we can Uh, how much, uh, Sora have you watched?

None.

I never go on Sora.

You're, you're, you're, you're purely like, if I, if I tell, if I order you, if I, if I, as your boss, I give you the task of generate a video for me, then you will do it. But you, you truly have not consumed like any? No, not in the last week. The first, uh, day or two. What about you, Nick? Have you consumed any AI video Sora or Meta Vibes? Not since we got access. Because you were really into it on day one, right? Day one, I was super into it.

It just takes too long to load, and then it generates, and then it says, you have violated policies.

Yeah, yeah, yeah. Can we hear him at all on the stream? Sorry. I can hear him. Anyway, what about production team? Hands up if you've watched more than 10 minutes of AI video in the past week.

Zero.

Zero. Philistines. Uh, well, um, yeah, so I mean, it will get better. I don't know. But anyway, this all concluded with, uh, Near Cyan and Rune going back and forth. Rune said, I like the way Sora looks. I like the content. That was his final— he wasn't saying, I'm not, I'm not doing some 3D chess coping on like, it's not that bad for you. He was just saying like, I like it. I think it's just good content every once in a while. It's fine. And so Nir says, do you actually believe this? Like, do you realize the product is the API, not the app as containment isn't possible? And that short-form video consumption is on the order of several hours per day for US adolescents. I didn't comment on the launch, but I'm sure I'm not—

Yeah, this is what I was saying, like, on launch day, which was like, if you really wanted to give Sora the chance to become a consumption platform, you wouldn't have made Sora 2 available via API for anybody to generate. You would've like at least tried to contain it as much as possible, right? People still could've like screen recorded and posted it elsewhere, but it was just very obvious immediately that it, there was like an incredibly innovative like AI creative tool.

Mm-hmm.

And like probably not gonna work as a like new standalone consumption app because There's a lot of people, the TAM for people that will be entertained by AI video is like the entire world in the fullness of time.

Yep.

Anybody that's logging onto the internet will probably at least laugh a few times a year from an AI video, right? Or they're certainly gonna see them. But the TAM of people that today only wanna watch AI video is like basically zero.

Yeah. I'm trying to imagine the bull case for the API. If I'm an OpenAI customer as a business and I'm using GPT-5 or some cache stuff and I'm using some reasoning tokens over here and I'm a big API buyer and then they come out with Sora and I'm like, I really would like to use that in my business, like I have an actual business use case. I'm trying to think of what that would be. I don't really know.

The business use cases, I mean, think about how many platforms are trying to help people generate ads for social ads, right? Generate content for various videos they're making.

Yeah, so if you're like Kennan at Icon and you're an AI generative ad company or now like an agency and And the API is not available for this new tool, you're going to be bummed out, right?

Yeah.

And so it just feels like— I mean, I agree with all the rationalization about it's better for the long term strategically to create— to actually bootstrap a real network, keep everything in the network. But at the same time, it just feels like the OpenAI ethos is just like, do everything all the time, all at once. And that's just like their— Their core thesis. Anyway—

Interrupting our timeline, Palmer Luckey has hit the timeline and says, it says something about society that the most controversial thing I have said in recent history is that I wish I would have married my wife sooner. Not that whales might have better oral history than humans, not that America should restart nuclear testing, marrying young. So, Obviously, his comments on TBPN earlier this week became somewhat of a current thing on the timeline. And of course, people were taking one part of his statement completely out of context and ignoring the fullness of his point.

Yeah.

But it's certainly been interesting to see the response. It's been very split.

Yeah, it was a true wedge issue, both inside and outside of the core tech community. I saw people who literally write for similar outlets on opposite sides of the issue, but I don't think of them as left or right or pro-tech or anti-tech. There were pro-tech people who liked the take and hated the take, and there were anti-tech people that liked the take and hated the take. It was all over the place. It was very, very hard to map. It was fun watching the conversation. get started. And it surprised me.

It started with him calling me a turbo normie.

Hilarious.

And saying that—

I was just laughing about it.

I know, we were laughing about that. And then he took it even further.

Yeah, I mean, it was framed in like, let's see how far we can push this envelope way. But you can clearly understand what the core message is, which is that—

He even said, easy for me to say. I met my wife When we were teenagers.

Yes, yes, yes.

Now, and the broader conversation of how having children in your 30s at any point is exhausting.

Yeah.

Now, basically, the younger you are, potentially, in some ways, it will be easier.

Yeah, yeah. I was trying to think back to like, okay, would I— like, let's say that I I actually did have kids at 20 or something like that. So I know some people that had kids at 20, like guys that I grew up with in high school. And their becoming a dad experience is framed kind of like, oh, it was like an accident or like a surprise. And they all turned out fine. It didn't detract at all. It's just like, kind of just turned into like a funny bit of lore about them. And it's like, oh yeah, they have like a kid that's like much older than like the rest of their cohort. So they are in some ways like, you know, like they don't have a huge support group because all of their other friends had kids later. But in terms of like the mechanics of like what their life is like, it's not that much different. And I feel like no matter what, when you analyze kids, Just the fact that you love them is so— it's such a wrench in the system of like, let's analyze this rationally, because you wind up just saying, well, like, yeah, they're exhausting. They're exhausting at 35. They'd probably be extremely exhausting at 20. Yeah, you'd have more energy, but you have less experience. And also, you love them. And so, it's amazing. And it would be amazing then. It would be amazing now.

Yeah. My take has always been the time always comes from somewhere.

Yeah.

So, if you have kids in your early 20s, then by the time—

Get ready to not club.

Yeah, that's true. But it's going to take up a lot of your time. Maybe it'll be massively distracting from your career, right? I think about this all the time. Early after my son was born, it was like, okay, there's now— I ended up just completely sacrificing my social life, right? There was work time and there was family time and there was like very little in between. Sometimes there was work time that was a little bit social.

But what is the social life for most people?

For most, it's, it's their work.

No, no, no, no, no. It's like a mating ritual.

That's true.

Like a lot of the reason why people are going out, people have a social life is because they're trying to meet someone. Exactly.

Peacock. Yes, exactly.

They're trying to meet that person.

Yeah.

And so, we've just kind of extended our lives, which is incredible, and then injected an extra 10 years of like, let's find the person and delay the process.

Yeah.

In my view, if somebody that has kids in their early 20s, well, by the time they're 28 and actually have hopefully like a skill set, their kids are in school, and maybe their kids are a little bit older, and they're able to potentially— they're not hitting 30 and then having to go through this crazy experience of having kids, and then it's still going to be a distraction from their career. But even then, they maybe have more responsibility, higher opportunity costs. It's all these different factors.

So I think also people were very triggered by the fact that he specifically said, like, if he had said like 19, it would've been different than 16 because 16 is like underage, right? I think that people were probably latching onto that, that it's like very young. Or do you think it was mostly just like, He said it in a provocative way. Yes, but I actually don't know about the history of when humans had kids. I feel like there was a time in the Castle Ages, like the Medieval Ages, when people were having kids at 16. That's true, right? He wasn't factually incorrect about that?

Yeah, I'm sure. I mean, you should look back through the Coogan family texts.

Oh, yeah.

Go back a few hundred years. Open it up and—

I actually did go back, and on my Swedish side, there's this, like, 400-year chain where the family just kept using, like, the same naming convention. So, they would say, like, Olofsson, son of Olof, and then it would be like Jacobsson Olofsson. Jacob Olofsson, Olof Jacobsson. Jacob Olofsson, Olof Jacobsson. Jacob Olofson. It would just go back and forth because they just kept naming themselves.

No, my dad was the 4th in a row of the exact same name.

Exact same name.

Yeah. I would have been the 5th. My mom was like, nah.

Well, if you have another son who you give the same name as the dad, you can use the nickname Trip for—

oh, no, no.

It would be Skip because you skip a generation.

Really?

Is that the origin of Skip?

Yes.

That's where Skip comes from. And so, if you have the same name as your father, you can be called Junior. Everyone knows this. If you have the same name as your father and your grandfather, people can call you Tripp, because you're the triple. And if you have the same name as your grandfather, but a different name than your father, you can use the name Skip, which is very fun.

Skip Hayes might be coming up.

Yeah, Skip.

It's a good name.

Might be coming up.

You might have to do it.

In other news, there was some news or fake news this morning that the Trump administration is in talks to take equity stakes in quantum computing firms.

Yeah.

Companies including Ionic, Rigetti Computing, and D-Wave Quantum are discussing the government becoming a shareholder as part of agreements to get funding earmarked for promising technology companies, according to people familiar with the matter.

And this is so important, because the free market just can't solve this. It's not like there's a whole bunch of multi-trillion-dollar companies that are—

yeah. Yeah, hyperscalers that are heavily invested here, venture capital firms that are happy to fund promising—

It's not like Google—

It's not like they have the software.

—had a breakthrough last week that people were pretty excited about. It's not like Microsoft's working on this.

No, I think the admin did this specifically to torture Scrawley, because Scrawley— this feels very personal, right? So Scrawley is very loudly and proudly in short supporting Rigetti and Ionic and all these companies because I think the general sense is that there's a spectrum from science experiments to scams, right?

Yeah.

And people tend to kind of sit somewhere there in terms of their opinion on quantum. Josh Wolfe had some choice words earlier. Let me pull it up. He said, this is beyond stupid, absolute waste of taxpayer money. Quantum is utterly irrelevant fraudulent BS. So strong words there. But do you know the Rigetti lore?

Do you know the Rigetti lore?

But before that, Shkreli had hit these, these companies had been selling off pretty aggressively. And of course, like you get an announcement like this, let's, let's look at Rigetti and see how they've been doing. So.

Oh, Ragov in the chat. This is a, this is a very interesting take. My conspiracy is this was a planted story by someone in the admin to help with exit liquidity in expiring call options.

I think that's potentially what—

There's something odd going on here because—

Because there would just be so much blowback from generally smart people about this that are taxpayers, that are Americans, that are going to say, I don't want to be a shareholder in Rigetti.

Yeah.

And so there was pushback and concern about the Intel deal, but at least there was broad consensus that we don't want We want into—

maybe this is a hot take, but like, I actually disagree with Josh Wolfe that quantum is utterly irrelevant, fraudulent BS. I don't think it's fraudulent. I mean, I'm kind of in line with the Shkreli take that it's just like it has a very narrow use case for very specific algorithms that— and we don't really know the business application. It's not like LLMs where We're just gonna be stuffing quantum computing all over the place. It's like, there might be something that we use it for, but we're not exactly sure. And so, it is very much a science project. But I personally don't believe that, like, I won't say all of the quantum computers, but I don't believe that they are all fraudulent. Like, I think a lot of them are saying, look, you, the shareholder, are investing in a science project. I'm going to test some things. I'm going to run experiments. It might work, it might not. And you're like, I'm 100% down for that risk. Yes, take the risk. That's what I'm signing up for.

It's worth it.

Everyone's happy. Everyone's happy, even if it goes to zero. And so, I don't see it as fraud at all.

Even in Google's efforts, which are probably the most advanced in the world, Sundar is not out there saying, hey, really, Alphabet should be a $10 trillion company because of our quantum efforts.

Yes.

project that they believe has real potential. So they're investing many billions of dollars into it, but it's still even internally structured as a science project.

And so, yeah, Equus Global says leave quantum to the Googles and the Microsofts, fund with free cash flow. I think that's one good option, but to be clear, it's also great if VCs want to take a crack at that and say, hey, there's this really weird idea in quantum, that Google and Microsoft are not going to even take a swing at because it's so bizarre. Everyone thinks it's a failure or not going to work. It's such a long shot. And so we are going to do it outside of the system. The analogy here would be like—

We had a quantum startup on that raised like $1 billion. I don't remember their name.

Yeah, it does not seem like it's a bear market.

It seems like capital is a constraint.

No, I completely agree.

Making progress there. And so, yeah, hopefully this story It's a sad situation where we have to say, hopefully this story was planted by someone in the admin whose calls were expiring.

It happens sometimes. Well, you know who's not planting stories? Graphite.dev, because they're reviewing your code in the age of AI. Graphite helps teams on GitHub ship higher-quality software faster. Upslope Capital. I am without speech. I like that as a funny line that you can quote tweet things with. I am without speech.

Brother Buco Capital says, can't believe my tax dollars are going to be invested in rigatoni computing.

Yeah. And the crazy thing is, Jaguar Analytics says, this is the kind of stuff that used to happen repeatedly during dot-com bust. And it shows the red, all the different Companies that are—

Joby, Archer, Rigatoni.

And it's not Rigatoni, it's Rigetti.

Iron.

Founded by Chad Rigetti. And this is the lore. Do you know the lore? Chad Rigetti is married to Susan Fowler. Susan Fowler is the Uber whistleblower.

Whoa.

Bombshell. Crazy, crazy dinner table talk at that point. It's like one's building and quantum computer. The other one is blowing the whistle, writing books about Uber, working with Mike Isaac on that, probably. I guess they knew each other. Although they did go book for book. That's the funny— that's the other funny lore is that she wrote the blog post that went viral, right? Mike Isaac at the New York Times was doing the— he's at the New York Times now. I don't know if he was there. I think he was there then. Mike Isaac was doing the beat reporting on the play-by-play of all the chaos inside Uber as Travis was getting ousted. And then they both came to market with books, and they went head to head. And I don't know who sold more copies, but it's just a funny thing of like, when you blow the whistle, you've got to get that book deal. You've got to turn it into a 50-episode Apple TV original series. You've got to turn it into— you know what you've got to turn it into? You've got to turn it into a One Piece anime, 1,000 episodes. None of this like, oh, Theranos was a crazy story. Let's turn it into an eight-episode series. No, I want a thousand animated episodes of the story of Elizabeth Holmes as she goes on adventures and and then just continues forever. That would be good.

Speaking of Elizabeth Holmes, yes, she's probably pretty excited about seeing CZ pardoned.

The pardons are flowing.

Chong Peng Zhao, yes. Canadian businessman and former CEO of Binance, pardoned by the Trump administration. So I'm sure she's thinking, I'm up next. I'm up next.

We'll see if it happens.

Yes.

Although there is a rumor she might be getting out earlier, not because of anything special necessarily, but just the way the— when you Google it, it's calculated one way, basically. The autocomplete is just like, when did she go to jail plus the maximum sentence, and it adds up. But like—

There's like a best behavior.

It's like best behavior plus time served while you were under investigation and stuff. And so you can count old times because it's like you were in prison while you were, you know, going through the court case, so we count that. And so I think that if things go well, she could be out in the next like 3 to 4 years maybe. But yeah, pardon certainly seems interesting. And I think that's probably why we're seeing some posting.

Seems interesting. You're not recommending the admin pardon?

I haven't seen anything that would lead me to recommend a pardon. At the same time, I haven't I haven't seen a ton of— I think I'm not exactly sure who got the most burned. There were definitely people that got bad test results, and that's obviously seriously harmful. But most of the VCs who invested, the investors were accredited. They took a flyer on a dropout from Stanford. I don't know. I think it's kind of like a buyer beware situation with a company like that.

Yeah, but they put devices that didn't work into use around the country.

Yeah.

Which— And so, like, the investors took the risk.

Yep.

They knew the risk that they were taking.

Yep.

But you put it out in the real world.

But the nature of fraud was that they lied about that.

Yeah. And could negatively impact—

It's just so crazy that you were— Bobby Kosmic, I am not a weave. I merely studied the Antichrist lectures. And so I've never seen an episode I have watched one piece of anime, which is Akira, just the movie.

Didi Doss was highlighting some sections from Coatue's AI report.

Before you dive into this, let me tell you about Profound. Get your brand mentioned in ChatGPT. Reach millions of consumers who are using AI to discover new products and brands. Yes. Fascinating set of charts. Shall we pull them up?

Didi says, why we are not in an AI bubble environment.

You got to summarize these, and I'll hit the gong for every positive bull signal out there. Break it down for me.

Here's why. Multiples are nowhere near dot-com level.

We got room to run. We got room to run.

And we got camera— we got face tracking on the camera. This is a new feature. Shout out to the production team.

Give me the second good news code too.

CapEx is growing but funded by cash flow.

Woo!

Give me something else from the largest tech co valuations lower than 1999. We're still in business, baby.

Concentration in the market isn't necessarily negative.

What about the U.S. government is going to bail out everything from quantum?

That's also quite exciting. Brian Halligan over at HubSpot, who's coming on the show soon, says, "In my honest opinion, private markets are in a bubble, are in bubble territory, while public markets still seem rational." I think that's a good take. I think a lot of what's holding us back from even more craziness is just the overall state of the economy. You have trade wars, tariffs, you have a weak labor market, you have an unpredictable admin, you have war in the Middle East, war in Europe, and you have a higher rate environment than we've been used to.

Yeah.

Imagine if we didn't have those factors, I think it could be a lot crazier, right? But there's a lot of real reasons for people to be kind of—

Yeah. I also wonder if—

Not entirely risk-on.

I wonder how much different the AI is. There are some new companies, right? OpenAI, Anthropic. There's a bunch of startups in private markets But AI is like a growth story for the Mag 7, but the Mag 7 are so huge that it's just hard to like, you can't like turn it into a meme stock, you know? And I'm wondering if you go back to dot-com, like the biggest companies at the time, do you know the biggest companies during the dot-com boom?

Was it the—

It was the oil companies. The oil companies were the biggest companies by market cap.

Oh, by market cap.

Like ExxonMobil.

Not by like—

No, no, no, not by growth at all. So what I'm saying is like, you have this like, you know, like you have this cruise ship that's like the anchor of the economy and it's like the big companies. And at that time it wasn't big tech, it was big oil.

Big oil.

And big oil was like something that would be really hard to push. And the dot-com narrative like didn't push it at all. But now we have the big tech folks who is also like a big cruise ship.

And—

but they're so big that, like, it's hard for them to just, like, oh, yeah, like, the Mag-7 tripled this month. Like, that would be a really, really crazy move. It would take trillions of dollars of investment to actually drive that in the market. So I don't know. Bunch of bull cases. Tyler, what are you thinking?

OK, so I calculated how many— how big of a hearth I would need to fine-tune.

Oh, yes.

Yes.

So, OK, so if I'm using GPT-OS 120 billion parameters, and then I have—

What do you need?

Like 8 H100s, you can probably get a decent fine-tune within like an hour. So that's going to be something around like 8, 8.5 kilowatts—

Kilowatts?

Kilowatt-hours. And then if you're looking at dry wooden logs, about 2 kilograms each, you'll probably need something like 4 to 5 logs for that energy.

Wait, 5 logs of— and that's all you need to do this fine-tune? Yeah, it's just 1 hour.

Yeah, I was actually surprised.

That's insane. I was expecting you to say, I'm going to need a massive thing. I'm going to need all this gold.

No, it's just 1 hour. I mean, it's not that crazy.

That's crazy. But you could really fine-tune GPT-OS in 1 hour?

I mean, the model's pretty small.

Okay. How would you— would you need like a server rack or could you actually like build a computer that could house 8 H100s? Is that possible? Or do you need like a proper server rack?

Yeah. I mean, 8 is pretty big. I don't think you can just fit that in like a normal PC setup. But even then, it's still not that big. A lot of people have that at their house.

There's definitely the trappings of a banger YouTube video here, like those primitive videos where, like, I built a thing from scratch. I built a toaster with no modern technology. Have you ever seen those videos or anything?

They're great.

And I would imagine, I fine-tuned an AI model using logs. It would be a ripper for sure.

In other news—

I have a guess.

Charles Gasparino says scoop, scoop alert. People inside Paramount Skydance say David Ellison, advised by his father Larry Ellison, are reluctant to pay more than $25 a share for Warner Brothers as the buyout drama continues. His last offer, the third rejection from David Zaslav, was $23.50 to be exact. Company still weighing a public, a.k.a. hostile offer. for Warner Brothers, which they will argue cannot be sold to other players who are interested, including Comcast, because of regulatory roadblocks. Full story coming. So Warner Brothers is currently sitting at, I believe, $21 a share. Yeah, $21.34. So David Zaslav is holding out.

Did you know Andrew Ross Sorkin got a genuine scoop about this deal while Being on TV for 3 hours a day and promoting the book 1929 and being on one of the most aggressive book tours I've seen recently. Is that the goat noise?

Yeah, that's my goat.

I didn't realize— I never put it together that that was— okay, I get it now. Wow, that took me way too long.

That's crazy.

I just thought it was like, you like farm animals.

No.

Okay, that's the goat.

Yeah, that's why I play— I played it for Jeff yesterday when he joined the show.

Yeah, yeah, yeah. Yeah, no, that makes sense. I like the goat noise. We're building out the full farm analogy. We need the farm map.

Yeah, and we're working on getting live animals. Of course. Of course. So we'll go to the goat dam. The goat will make—

Truthfully, I don't feel like I have a good understanding of what David Ellison is building. I'm newer to coverage of the media conglomerates because I focus more on the Mag 7, more on the tech companies, more on the AI stuff. It's a fascinating world.

I think he's running the experiment of like just, you know, basically going to the world and saying, one media industry, please.

Basically, it feels like he's trying to build some sort of new conglomerate. I don't know how the synergies pencil out. I don't know how everything works.

We got to get Bari Weiss on the show.

I would love to.

Talk about her first couple of weeks over at CBS.

We also have to tell you about Turbopuffer, serverless vector and full-text search built from first principles on object storage. Fast, 10x cheaper, and extremely scalable.

Philip Johnston is on a tear. He says, I'm super excited to announce StarCloud's partnership with Crusoe, builder of OpenAI's massive Stargate data center, to enable— enabling the first public cloud in space next year. This is the first step in a partnership that will see StarCloud providing energy for much larger Crusoe data centers. over the coming years. There was also a post from Dalien. He was firing shots at Nvidia because they announced AI shipbuilding partnership with Saronic. He said Nvidia has announced AI data centers in space and an AI shipbuilding partnership in the last 2 days. Words and physics just don't mean anything anymore. Dear God, this has to be the top. Please, when will it end?

He said that in the post. I kind of assumed he was like, you know, top-pilled, but That was a wild place to go. Yeah, we were kind of dunking on this yesterday. I don't know. It seems like there's a lot of pieces that have to come together for this to be real.

Yeah. Dalien, from my understanding, will always do a deal that he's extremely excited about. But it's notable that he was very active earlier this year, and he sort of hasn't announced a new investment. In the last little bit.

So yeah, I mean, I think some of this is fodder for the public markets, marketing campaigns, just general outlines of remaining futuristic. Like there is a world where if you're Nvidia and you're looking at Andrej Karpathy saying like, hey, the current thing is gonna be slow, you're like, well then let's, What are we gonna do in the next decade while we're waiting around for ASI? Maybe we put some data centers in space. Maybe we build some ships.

More SaaS.

Maybe we do a bunch of stuff. More SaaS. That's what I wanna see. Take some real shots.

MacRumors is reporting that there's virtually no demand for the iPhone Air. And I thought it was notable. I have a friend who used to work at Apple. He bought the iPhone Air, used it for, I think, like 24, 48 hours and returned it.

He did?

He just said it was— Not the, the, um, the thinness of the phone was not worth the, uh, the battery, uh, trade-offs. And he also said that, um, the phone is so actively trying to conserve energy that it renders things like poorly and slowly.

Okay, that's annoying because I feel like there's a little bit of skill issue with regard to the battery where it's just like, just get a car that has a battery charger in it.

Yeah, I think there's something about phones.

Get one at your desk, like figure out how to actually charge it regularly.

Yeah.

Everyone has some point in their day where they can plug in. And maybe if you just are a wild man and have no routine whatsoever, but you're also addicted to Soar and TikTok, you're going to have a bad time. But I feel like if you have any sort of like, yeah, around noon I'm at my desk, and so I just put it down on the charging pad regularly, you're going to be fine. But I don't know.

Yeah. There's also something about phones where I think the iPhone Air is a bit cheaper, right?

Sure.

But something about a phone is it is so valuable to your life that if you ask somebody, if you were selling an iPhone from today to somebody in 2000, how much would somebody pay for that? Would they have paid— would a high-powered executive pay $100,000 a year to have the iPhone from 2025 and 2000. I think it's that valuable in terms of what it does for your life. And so I don't think a lot of people are making— at least the average iPhone customer is saying, if I can get slightly more performance out of my phone for an extra couple hundred dollars a month, it's probably a good exchange.

I'm just surprised there aren't more people that are like Tyler, were a couple of iPhones before. What were you actually on? 13 or something?

I was on— no, I was on the 11. So I was like 6 generations behind.

Is that 6 years? You hadn't gotten a phone in 6 years?

Yeah, I think that sounds right.

Have you— was this your first phone ever?

No, I was like— I was probably, uh, like freshman in high school. Yeah.

Okay, okay, it's sort of adding up to me, but that's like— what was your first phone?

Um, Motorola, hand-me-down, but it was probably like iPhone 4.

iPhone 4.

Wow. Started on an iPhone.

Wow.

Well, no, silver spoon alert.

Yeah.

No, no, no.

Wait, so you don't remember where you were when Steve Jobs introduced the iPhone?

I would— I— was I even born? Yeah, yeah, I was like 2.

Brutal.

Uh, Tyler, are you excited that The Sopranos creator David Chase is developing a new limited series about Project MKUltra for HBO?

Do you know MKUltra? You know where you were when MKUltra happened?

What?

It's like from the '70s. Oh, you're giving him the tinfoil hat. This is the first— this is the first tinfoil hat. Oh, the tracking cam is phenomenal.

Put it on, Tyler.

There you go.

Why is he wearing the tinfoil hat?

You're safe now from MKUltra, right?

Isn't that—

isn't that— wouldn't somebody—

No.

I thought MKUltra was like the LSD.

It's LSD.

Oh, I thought it was remote. I thought this was the remote mind control. Control one.

No, no, no, no, no.

Exposed for not being wow.

Fakes get so. Fakes get so. Doesn't know his conspiracy theories. No, ridiculous. Anyway, Kotu has a bunch of other stuff. We should go deeper into their public markets update. We should just have someone from Kotu on again because we have a couple a couple folks on it.

But they said they studied.

They studied. They sat them. They sat down and. And read— they studied 10 books. They went back 400 years, read 10 books, analyzed 30 bubbles.

Let's give it up.

Why stop at 10?

Let's give it up for reading 10 books.

And their conclusion was that it's not a bubble, right? I think they think it's like 1996, which is awesome. I'm rooting for '96. Let's go. I also am rooting for the bubble already popped. It popped last Friday, and now we're reinflating. Anyway, we we have we have Jacob from Microsoft AI in the TVPN Ultra Dome. Welcome to the show.

Wow!

Finally, somebody that oh oh oh we got a short guy in the in the studio.

How tall are you?

Six seven.

Six seven.

I'm six eight. Oh, six seven. Are you getting the six seven thing? Do you have kids or anything? You know this whole thing? Six seven.

Come on.

Yeah.

You've got to be in youth culture.

I don't know it at all. I found out about it— I literally found out about it when the Wall Street Journal published an article.

No shit.

And it was like, here's the latest trend that's causing problems for teachers. And I read it.

What are you guys even talking about?

You don't know this?

OK.

No, I don't, dude. So there's this meme where in middle schools, if the number 6 comes next to 7, 6-7, Yeah, it started from this meme. All the kids will go crazy and go, 6, 7, and then they do this because— and if you trace it back, it goes like 7 layers deep into like this. There was a kid at a basketball game who was saying 6, 7, which was a reference to a basketball player's height.

And then one awful rap song that nobody had listened to before called Doot Doot.

Yep.

Doot Doot.

And that's a reference to 67th Street in Chicago. Which is like a popular area to sing about, basically.

I'm impressed. That's all from the article?

Yeah, loosely. Yeah, Wall Street Journal knows their stuff. Knows their stuff. Anyway, the Wall Street Journal didn't have your backstory, so you have to give it to us directly.

Yeah.

Introduce yourself.

Absolutely. Yeah, Jacob Andreou. I lead product and growth over at MAI, mostly focused on Copilot. Before that, I was at Snap for like 9 years. 2015, 2023, doing the same thing, basically. And then in the middle, venture for like a year and a half. Still doing a lot of investment. Greylock?

Was that Greylock?

Yeah.

So that was here in LA?

No. SF, in the city. But I've been based in LA from 2014-ish, and now just spend a lot of time in Redmond and Mountain View, my 2 favorite places on planet Earth.

Amazing.

Yeah. That's great.

Exactly. That's not nearly sharp enough. Like, you're gonna need to narrow that down. That's like a million products.

We're putting Copilot in everything.

It's a multi-trillion-dollar company. Like, what do you focus on vertically, horizontally? How is the organization structured? Like, what are you actually doing on a day-to-day basis?

Yeah, I mean, it's a super, super cool moment for Microsoft right now. We have a bunch of folks that have come together, a bunch of people that have been there. I run into people all the time at Microsoft Been there for 35 years. It's fucking crazy. I've never seen anyone who's been— am I allowed to swear on this show? We don't.

We don't.

But we don't have any censors.

But feel free to if you feel like that's the right—

you're fired up.

That's out of words.

Yeah, that's what you want the Microsoft comms team to really like take away from this, right, is that you went wild on a show and they maybe shouldn't have greenlit it. Then yeah, all right, we'll push the envelope.

Um, no, look, we got together an amazing group of people, uh, a ton of people that have already been there forever. And we're just trying to transform a bunch of the different things that say Copilot across the company. So we got a group of folks, they're super talented, working a lot on GitHub Copilot, VS Code stuff, you know, whether that's competing with Cursor or all the other folks kind of in that agentic IDE kind of world. And then me and my team, we spend a lot of time on like the consumer Copilot app. So the thing that's in your pocket, the thing you can pull out of your pocket, use voice mode, talk back and forth with web retrieval. And then we introduced a bunch of really cool stuff. just in the last week.

And is consumer Copilot, is that powered by OpenAI models? Are you model agnostic? Are you training your own models? Like, how are you describing? Because with some of the foundation labs, they have to put a ton of their marketing behind like their model.

Yeah.

And we were having this discussion about like, at some point consumers are gonna stop caring about the model. Yeah, yeah, yeah. So how are you thinking about it?

Yeah, 100%. Yeah, so Microsoft AI, when we're shipping Copilot, we're just building the best possible product.

Product.

That's it. And we're super product-focused, super customer-obsessed. And at the end of the day, like, whatever's gonna give people the best thing in their pocket, like, that's the job.

Yeah.

And so for a long time, that was OpenAI models.

Yeah.

And thanks to, you know, the amazing partnership that we have with them, we can put all their models directly into the product.

Yep.

Uh, 4o powered the mainline response for us for a really, really long time until GPT-5 came out more recently. GPT-5 now is honestly, it's an amazing set of models, uh, and it's been really, really popular inside the product. But There are parts of the product where we're already serving other models as well.

Sure.

So we introduced a couple of new modes this week. One of them is not powered by an OpenAI model. A few of our own first-party models just started to drop. So I think 6 weeks ago was the first time we privately started running LM Arenas on MAI fully end-to-end trained models.

Okay, cool.

So pre-trained before that. Yeah, before that we had no ability really to build our own foundation models from scratch. And in the last 6 weeks across both language and text, text and image, we've been in either the top 5 or the top 10 with our own first-party models.

Yeah.

And that's like a first for us.

Yeah.

And so it's been super cool.

So does that reflect a view that essentially, like, the foundation model layer is commoditising, it's high CapEx, but it is something that's maybe viable to get near the frontier?

Yeah, I think that a piece of it is a little bit that perceived intelligence, you know, when you try these different systems today, even when they're backed by different models, It is a little bit tough to tell, oh, this is way smarter than that. Like, things are kind of getting a little bit closer together.

We saw that with GPT-4.5.

Yeah, totally. And then we went to 5. A lot of people, even though it was a smarter model, were complaining about, like, the tone and the personality.

And so we're in the spiky intelligence era.

We're no longer in a race where just, like, smarter is just better every single time. And so to really start to build an amazing product, it's not good enough to just have a set of models that are marching up intelligence. You have to actually build what people want in the product.

Sure.

So an ability to go kind of end-to-end and be able to actually look at the way people are using the product, have that feedback into model training, especially post-training, and then have that then allow you to iterate the product, it creates this model-product-user loop that is kind of the main way that you'll probably push the envelope in terms of meeting the user needs. So that's what we've really been focused on.

And when you think about the user of the app, are you thinking more business user, prosumer, consumer? All of the above. Like, there's everyone from like the Xbox Call of Duty kid is a Microsoft customer to like the Fortune 100 company CEO. There's very different archetypes.

Yeah, long term, we have to serve that whole spectrum. To start, it's the consumer. And it's the person that is looking for this thing. You know, they're asking us for advice on health. They're trying to learn new things with Copilot. They're asking for a lot of help making big decisions. Those are the places where we're really playing the most. And now you can't ship a product from Microsoft without being awesome at productivity. And so in the last few months, huge push in productivity, making it way easier to upload like tons of files, reason over them, create documents based on the source material you've uploaded.

Yep.

And this has just like ripped. Like the, the, as soon as we put this in the hands of users, completely, you know, we, we kind of expected, okay, if we improve answer quality generally, this has always going to be good, improving file upload and improving the number of files that people could upload, users just were drawn to it so quickly.

So I mean, it feels like the real wheelhouse, like you're just going to knock it out of the park with the customers if they're running their life on 365. Like they have a Microsoft email, they have OneDrive going, they've set up their life in that way, maybe because they use it at work, maybe it's also hybrid work from home. That's going to be a really great experience for them?

Yeah, so the product today doesn't actually need to be rooted in M365 at all. You can just download it. You can just use it. And it's actually amazing for that use case. Long term, the system that has the highest context on you is going to be the best in both scenarios.

Of course.

It's going to be the one that knows when your meetings are so it can schedule that dinner that you've wanted to go to. And so we actually think that the long view here is that these things ultimately have to come together. But for the next 6 to 9 months, we're going to focus on nailing the enterprise use case in M365 inside of all the work apps that you already do all your work in. And then at the same time, on the other end of the spectrum for consumers, nail the Copilot app that you just download from the App Store and you just love and you use.

How should startups think about your guys' ambitions? Because I think that Sam had a quote— it was probably a few years ago at this point, maybe a couple of years ago— which was like, don't build a company that's assuming the model's not getting better. And I've seen so many companies this year— we've had some of them on the show, we've talked about them that are basically, they may as well have said like, we're building a different Microsoft Copilot for XYZ, or we're building Microsoft Excel and Microsoft Copilot for Excel.

Yeah, yeah, yeah.

As a new company.

Yeah.

And it's cool. Like these are big categories, right? They're extremely valuable products. It's worth new teams like trying to take a crack at some of this, timeless software. But I'm curious, I'm sure you've seen some of them, and you're like, wait, you know that we do that already.

Yeah, that's on our roadmap.

And we have hundreds of millions of people that use that.

As opposed to Bloomberg has an Excel plugin, and they've had one for a long time. That hasn't been on Excel's roadmap because it's a separate thing. But yeah, how are you talking to startups?

Yeah, look, I think if you get really specialized, like the Bloomberg example, we're never going to change all of Excel to go compete with like a specific vertical of a way people use Excel. You can't change like the core platform that much. But if you're trying to build something, and I won't use the Excel example 'cause it's too close to home, I think, for people that are working on this problem. But if you built like a plugin for Word that does something that's like pretty horizontal, like it uses Word for you.

Yep.

We know.

Like, this is certainly something we spend a lot of time thinking about. Word has had templates for years. And if you think about, you know, Prompt to template. Like that's the very obvious next step. Do you feel like there's a there's a responsibility for the Microsoft brand to send clear messaging to startups so that you don't steamroll them? I always wonder about that. Like it feels like Sam was kind of saying like you know I'm former YC you know head. I want startups to like me, build on my platform, etc. And I don't want someone to. accidentally get crushed by the steamroller.

Yeah, yeah. I think that when you're training foundation models at the scale that they are and have kind of been doing it for as long as they have, I think that the world is still adjusting to how quickly things are changing. And so I think the degree that the model builders can signpost and be like, hey, this is going to change a lot really quickly, I do think that this is probably just the right thing to do for the ecosystem. Now, that being said, for us, we're just focused on building an amazing product.

Yeah, and the other thing is some startups are overtly saying, I'm going to kill Excel.

They're coming for Excel.

If they're coming for you, it's not like you'd be like, hey, you know— The gloves come off.

It's a knockout drag-out.

Maybe don't do that because we're doing that. It's like, no, they're trying to eat your lunch. It's totally fair to just keep expanding your product in ways that are valuable for the majority of users.

But the agentic stuff happening in Excel, as a specific example, is really cool. And so it's one of those areas where it's like, look, focus. You can win a lot of things by being focused. So who knows, maybe the people doing that are still going to have an amazing product and build a great business. But Excel is going to get pretty agentic.

Do you have anything to show us today?

100%.

Let's take a look.

We've got a lot of cool stuff.

Yeah, yeah, show us some stuff. Well, so just quickly, the stuff that we— I'm going to get your camera here so we can see your phone screen if you want to pull it up.

So the stuff that we talked about, gosh, just a little bit this morning, We have a huge release of AI browser. And so I think we heard a lot about AI browsers this week.

Yeah.

And the biggest difference is Edge has, you know, hundreds of millions of people using it.

Sure.

And now every single one of them with a flip of a switch now has an agentic browser in the form of Copilot mode.

Sure.

I've used it to reorder stuff on Amazon.

Okay.

Kick off stuff for me. It's been awesome. We also—

And that's on top of Bard. Yeah, get into maybe a little bit more there.

Yeah.

Just while we're talking about it. But this is like you can chat with the pages that you're on and Yeah.

So sidebar is totally there.

Yeah.

And you can reason over multi-tab context and you can talk about the different stuff that you're doing. The thing beyond like actions, like we've seen a bunch of people talk about whether it was like Comet or the Atlas stuff from earlier this week.

Yeah.

You know, actions are just going to get better and better as the models improve. And, you know, Copilot mode in Edge is, you know, state of the art with respect to the actions that you can do. One of the really cool differences is once you enable it, it'll actually start to pick up like patterns and habits of your browsing.

Mm-hmm.

And it'll actually create these journeys and do things proactively for you. And you can jump back in where you left off. And so if you're like planning a trip and doing flights in a few tabs and doing hotels in other tabs and trying to figure out stuff to go do, Edge will actually see what's going on with that. And the next time you come into the browser, it'll actually have taken the next step in planning that trip for you. And with a single tap, you just jump straight back into flow where it's actually done the next thing. And so—

That's cool.

For me, when I'm like interviewing candidates and just doing a bunch of research in tabs—

It's perfect for this.

How do you— there was some conversation, concern about prompt injection. Is that like— how do you think about solving that problem? It felt like people were saying, like, hey, this is a big risk, slow down. But it felt like something that was somewhat solvable pretty quickly.

I think that's right. And I think that you have to look at this whole class of products right now as— it sounds a little maybe too over the top, but kind of like a security preview. We know there's a whole bunch of risks with allowing the browser and these models to take actions like this. You really have to supervise it. We have a ton of checks and balances. It'll ask you a ton of steps along the way to make sure everything's still on the rails. But we're learning alongside everybody else, and this area is going to evolve really, really fast. So we're ready to just keep playing whack-a-mole as these issues come up.

Yeah.

Are there any surprises? Do you have anything that might be teed up for us? We love hitting the gong around here. So we did—

we are now the first ever AI app to also drop groups.

Okay. Are you on the most aggressive version of Liquid Glass, by the way?

I am like completely liquided out.

Glass.

I am. It's all glass all the way around.

Very cool.

Okay. It looks good.

And H2O Max.

And with Pro Max, you actually get the battery life so it can power Liquid Glass.

Okay, cool.

There we go.

Yeah, exactly. So Groups is phenomenal. It's now the first place where if you're having a chat with Copilot, you can just add someone to the group. Use it for my wife when she's asking me questions I have no idea the answer is all the time. I can just put her in the chat and she can just take the rest for me.

That's cool. Like creating a group chat with an AI and other people. You're not bringing it to Teams. It's like, yeah, sharing this. And then, and then we're both on—

So when I'm in, when I'm in the thread, oh, I just hit share.

Yeah.

Invite with a link. I can drop it in iMessage or whatever. And then she's just in the chat and they don't even install the app.

Yep.

They can just use it right from mobile Safari, which is super cool. And so this is the first time across all these apps that you can finally do this, which is how, how it seems like pattern.

It's such a great and somewhat obvious feature. I'm somewhat, I'm somewhat surprised that It hasn't.

I was shocked when I joined MAI.

I was like, no one has built anything social in any of these products.

I worked on My AI at Snapchat before I left. And so no one had really done social AI since then.

You were like, why not us?

I was like, I'll do it again.

It's fine.

We'll run it back. OK, OK, sick.

And then one of the really cool things that we did—

Yes?

Copilot knows all about you guys, by the way.

Oh, really? Well, we do put out 15 hours of content every day.

OK. We introduced Miko, who's this new person that you can talk to, this new character. Really cute face. It's not going to sex with you. So it's completely different than the other AI appearance. Explicitly.

It's explicitly non-explicit.

That's not what it's going to do.

This is the counterpositioning.

We love counterpositioning.

Let's drink the gong for Jesus.

That's not what's going to happen.

Let's drink the gong. It's not traditional.

Full marketing.

Bull market and traditional values.

And so with traditional values at our core, yes, uh, as soon as we showed, uh, Miko to Satya, his favorite thing actually was like poking it, which I couldn't find a better way to describe that on air without it sounding a little bit weird.

Yeah, but again, okay, it's, it's, it's going to be strictly platonic for work. It's totally safe because this is from the company that brought you Microsoft Excel.

And, and we—

and of course, yes, of course, and I would trust that brand.

And so as, as you poke Miko.

Yes.

And okay, how many times do you have to—

you really got to hammer it.

Hammer it.

Clippy is back!

Let's go!

We, we have been at— we have been asking for this.

Is this breaking news? Is this a scoop for us? This is— oh, this is the first time this has been on air. Now Clippy is back! Clippy is reborn. Yes.

Incredible.

So Clippy walked so Miko could run. Yes. Yeah.

And so we got to pay homage to our roots. And so Satya, when he saw this, demanded—

Yes.

That Clippy—

I'm so happy about this.

Front and center.

Incredible.

I think people, people have— people forget.

We've been, we've been begging for this.

We've been begging for this. Obviously Clippy had like a bad, bad rap and people were like, it didn't do what it was promised. But I feel like All of the nuance of like where it wasn't perfect has completely melted away, and it's just lore. It's just great lore.

Your parents hated it, so the kids loved it.

Exactly.

And it was cool to love it, so everybody else got on board.

And so it's— we're so ready to bring it back.

Well, I just— yeah, I have so many good memories of being in the computer lab at school and just looking down and knowing that Clippy was there. He was riding with me.

How are we— what does it take to update the name of the app in the iOS App Store to just Clippy. You got it.

That's the next—

Right now it's an Easter egg, but we're working here together.

Baby steps.

Baby steps. Actually, there's like a team of 25 people that really worked hard on Miko, and they're not feeling that.

There's like 3 CMOs.

This is not, you know—

Bring back Clippy.

But that is very exciting. Thank you so much for coming by.

No, thanks for having me, guys.

And breaking it down.

Yeah, yeah.

What a fascinating time in the industry. What a fascinating place to be. It must be thrilling to be in your seat.

It's an absolute treat. There's 5 places in the world you can try to build frontier AI for a billion people, and this team and this setup, it's unbelievable.

So it's a real blessing. Now, as a true product guy, it's like going somewhere where you have that scale, resources, et cetera. And it's actually— it's good that we have just this flood of talent into early-stage companies and the labs and all that stuff. It's actually incredibly important that talented people go and work at these companies that already power use cases for billions of people, millions and millions of companies. So I'm glad you're there.

No, you got to do it, 100%.

Yeah. Anyway, so much fun, dude.

Good to see you guys.

You guys are fantastic.

Pleasure.

Congratulations.

Come back by again soon.

Our next guest is Drew Houston from Dropbox, hopping on at 1:00 PM. We have 30 minutes to take you through more of the news. But first, let me tell you about numeralhq.com. Let Numeral worry about sales tax and VAT compliance. I use it myself. It's the official tax partner of Lucy, the nicotine company, who Palmer also got in trouble with, uh, talking about nicotine. Uh, but that is a much lower— that's a much less, uh, you know, controversial topic. I think everyone has hashed out the nicotine story.

I can't believe it.

I hope everyone knows.

I can't believe—

The Economist published a— I can't believe Clippy's back.

Wait, what? I just can't believe Clippy's back.

You're still rocked by this?

I'm rocked. This is shaking my worldview.

I love it.

No, but this is super fun.

The nicotine thing is funny because everyone was getting mad about Palmer for basically, in his own way, teeing up the argument of like, Are there benefits to nicotine? And like, this is a topic that has been discussed by Gwern back in 2014, 2015. Then we started Lucy, we were talking about it a little bit. Andrew Huberman did a big deep dive on nicotine, kind of laid out some of the complexities, and he actually revisited that. It's in The Stack today. Do you have his post pulled up? What did he wind up actually saying?

Fully.

I'll pull it up here. Pull that up.

But the, uh, the really interesting one is from The Economist, who I regard as a pretty conservative organization. They're not really doing hot takes over there. Um, September 12th, 2025, published an article called, uh, What Nicotine Does to Your Brain. Their conclusion: the drug is hugely addictive, but it does boost mental performance. And so That's the formulation that I think Palmer was grappling with, which is there's mental performance, it's highly addictive. How do you balance those in a world where you remove the carcinogens and you don't wind up with actually smoking cigarettes?

But he was talking about that it's an appetite suppressant.

Exactly.

There's obviously benefits in not being overweight. There's trade-offs with everything.

Yeah. But anyway, what was Huberman's final— let's trust the experts on this one.

Go to the experts.

Huberman's final word.

He said, with Palmer Luckey's comments about nicotine on TVPN, your reminder that yes, nicotine increases focus, also is very habit-forming, duh, is not carcinogenic. I know, I know that it's habit-forming. I've quit nicotine, you know, 5 or 6 times.

Yes.

I typically quit it like every night before I go to bed.

Yeah, and then pick it back up. And then pick it back up the next morning. Is not carcinogenic unless smoked, vaped, Dipped or snuffed and is an unusual stimulant because it simultaneously focuses and relaxes you. Also raises blood pressure. Again, do your own research.

Also, 21+. Don't get in the game unless you're 21+, Tyler.

I've never— why are you looking at me?

Well, they went to the wrong camera.

The little jump scare.

Travis Kelce is teaming up with an activist group to invest in and revive Six Flags. There was another—

How much?

Oh, I don't know.

We have no idea. Left. Mike Isaac says, genuinely appreciate how Pop Crave left out the investor part of activist investor group because all the people in the replies think that there's going to be like a BLM takeover of the roller coaster park. Someone else was saying, like, did Travis Kelce, like, get 3 wishes when, when he was like 11? And he's like, I want to be a football player. I want to date the biggest pop star in the world. I want to take over.

It's, it's a $200 million deal. He bought 9% of the theme park operator. And we love to see big-name celebrities crossing over into the private equity world.

Wow. They have the ticker. Do you know their ticker?

Fun.

So good.

That's a great ticker. And a little fun fact, turns out according to ChatGPT, you cannot reserve and barter stock tickers. I think we need to do a PSA about this. We heard some rumor that people were trying to, in the way that you buy a domain, maybe you squat on it. Let's say that you think that, you know, after ai.com, after chat.com, after after quantumcomputing.com, there's going to be a wave in time travel. So you go out and buy timetravel.com and you wait, and then somebody comes and buys it from you and you make a nice return on your investment. That's totally cool in the dot-com world. That doesn't work with reserving New York Stock Exchange tickers. Like, you can't trade them that way, at least according to ChatGPT.

Oracle.

And so speaking of just funny stock tickers, we do not believe there's a market there.

I can see why. Travis Kelce was interested in this. Six Flags trades at less than 1x revenue. Whoa.

So hugely—

He's seeing potentially— —deep value here.

Are you bullish on— first, do you like theme parks?

No.

Never? Never been?

I used to go to Legoland a lot growing up. I enjoyed Legoland.

Yeah.

But I kind of grew out of theme parks at a young age, and I haven't been to one in—

I never had a fear of heights. I never—

More than a decade.

I never was like, I don't like them. But I also was never like, I gotta go a ton. Kids sort of changed that because kids really like them. It's fun to go with the kids. But Six Flags is— it's a different target audience. It's not as— there's not as much lore around every ride. But are you bullish on theme parks generally in a world where Bill Peebles is updating us on Sora and it seems like the march of generative AI is unrelenting? Should you go long slop and then also steel on the roller coaster track?

I think I'm somewhat takeless here. When Chesky was on a couple of days ago, his point of view was that as slop increases, people will— I mean, these are my words, not his, but digital life becomes more intense. People will want to touch grass more. They'll want to go on trips, things like that. But in some ways, I feel like the theme park is like real-world slop, right?

It is.

It feels like it. I imagine going to Six Flags for me would, for 4 hours, would feel like I would come away feeling like I was using Sora for 4 hours, right?

Yes.

Like, not exactly. It's not like really counterposition.

Yeah.

It's not exactly the Metropolitan Opera.

It feels like the physical You know, because I just I don't like to pay to be in crowds.

It's physical brain rot. Yeah, it's brain rot IRL.

But maybe Travis Kelce wants to turn it around, turn it around, tear down the tear down roller coasters, put in put in maybe he wants to increase the risk.

Right?

Football players go out there; they're going head to head, right? Yeah, yeah, high risk environment, a lot of injuries. Maybe he says, "Hey, Six Flags injury rate right now." This is this is the Nathan. fielder strategy. The sort of accident rate is so low on the roller coasters that people don't get the same rush as if they thought that there was maybe like half a percent chance that each— that a ride would end in disaster, right? And so then that would, you know, really bring you into the physical world and really make you feel human to be taking on that level of risk, right?

Yeah.

So who knows?

Well, good luck to them. The Sora update, we already talked about Sora a little bit, but new— you can register cameos of specific things. So your dog can become a cameo, and then you can create—

Finally. Thank you for listening to your users, Bill Peeble.

Finally.

Finally.

I mean, I— You're gonna play around. I'm for sure gonna do this. I like— I think it's gonna be hilarious to put my dog on a skateboard at the X Games. Like, if you don't think that's fun, like, you're just like humorless and like you're, you're, you're like a slop scold. You're a scold.

I think it's, I think it's—

Let me enjoy my dog at the X Games.

I've been joking a little bit. I like to horse around, but I think it's smart because they had so much pushback from using IP.

Yep.

And it's got to be personal. It's got to be like, it's got to be personal and people will be excited to generate images of their dogs at the X Games.

And I'm going to, and I'm going to take my dog And I'm going to—

he's going to do a kickflip and he's going to do a kickflip.

And I'm going to send that to the family group chat, which is my new social network. There is social media and then there is social networking. I think this is a great take by Chesky. And so my social network exists on iMessage. And so I will send the video.

I still very much feel like X is a social network.

I feel like it's a professional network. It's my LinkedIn.

Okay. Professional network. But my Social life has almost entirely overlapped with my professional life.

If I'm trying to figure out what I'm going to be doing in my social life on this weekend, X has nothing to do with it. That is orchestrated purely in iMessage.

Yeah, but it's the way that I catch up with the majority of my friends.

Yes, yes, I agree.

In terms of keeping a pulse on what they're doing in the world.

It is more social media. than pure algo media. So there is something there. But anyway—

Speaking—

Yeah, it's going to be fun to play around with.

Media.

What else?

Not really. But Ramp data is now available in the Bloomberg terminal. This is cool. You can track AI adoption rate, business size right in your terminal.

The best news I've heard all day.

Very differentiated and real-time data source.

It's cool.

Very cool move from Ara and the Ramp It's very good. Uh, this is crazy. The Unitree—

Before we read this, let me tell you about Vanta. Automate compliance, manage risk, prove trust continuously. Vanta helps you get compliant fast.

Uh, the Unitree G1 crawl policy has been deployed to hardware. Plenty of room for improvement, but it's a start. Let's pull up this video. This is a humanoid that can turn into—

Oh, wait, this is a humanoid? Oh, this is crazy. I mean, I feel like— I, I don't want to be too, too negative here, but I feel like the film— the filmmakers are really not stepping up. Like, I feel like the creative tools right now, both with like AI stuff, but also like you could buy one of those or rent it and you could shoot a horror film Right now, practically. Like, you wouldn't be, you wouldn't be like generating the AI.

Jason Karmann, you could make a humanoid robot horror, low-budget horror film right now.

Or, or action movie or anything. Like, that just feels like a character that you could immediately put into a movie and tell a whole story around. What do you think?

Or a character on this show.

Yeah, that's true. I think we might have to do it. We might have to.

I mean, there's something, uh, Beautifully dark about a humanoid that's walking, and then it suddenly turns into this insect-like thing.

That is very spooky.

Are we really gonna get one of these?

Okay, if we get one of these, we also have to get a couple Desert Eagles, because if it starts acting up, we have to be able to take it out, right? Like, we have to have, we have to have, you know, it's like, If we were going to actually do the hearth, I would also say, hey guys, let's get a fire extinguisher. And so because what happens if you're training and the log rolls out?

Yeah, and then the AI becomes sentient and then tries to kill us, right? That's why we need the fire extinguisher.

Well, that's what— no, no, we just need the fire extinguisher because it's good safety policy if you're using a warm hearth to fine-tune GPT-OSS on A rack of 8 H100s. Because if you're going to be burning a log inside, you should have proper equipment. And I think if we get this humanoid, you gotta be ready to take this thing out. You have to be ready to take this thing out. Jordy, what do you think? How are you taking this thing out if it rises up? If we get one of these, I want to be prepared on day one.

I'm thinking— I think you strap charges to it.

Okay, and so and it has just it has a vest that I don't tell anytime that that a vest that that and then we have it yeah it has explosives inside we release it at any moment it explodes yeah basically Kevlar it has a Kevlar suit over it explosives on the inside oh so it explodes inwards explodes inward yes I feel like strapping a suicide vest onto the robot is the is the last it's the last thing I would do I think that might be worse it's just charging it's charging at you and you're running Oh no, I know. If it, if it, if in one wrong move, we're all going down. We're all going down with it. Oh, this is wild. Anyway, let me tell you about figma.com. Think bigger, build faster. Figma helps design and development teams build great products together. What is this? Is this literally, this is literally ChatGPT? Should we play this Family Guy clip or is this a Sora video? I'm not familiar with this. Did you put this in here?

No.

Okay, let's move on. Fareed has a chart that looks so insane. It's ChatGPT's customer retention. What do you think about this? How do you square this with the earlier data that we saw that showed that, hey, there might be some slowing adoption? This looks like a—

well, this— do you think this looks like an incredible— This is unrelated to growth.

Yes.

This is just when somebody tries ChatGPT or how— How many months later are they still using it?

I mean, it's directly related to growth because if you have really great retention, you can just add 5%—

Sure, sure, sure.

—of new people that you're trying.

But I'm just saying you can like—

They're going to stick around forever.

Yeah, yeah, yeah.

As opposed to when you have a leaky bucket, you're like, we need to be doing Super Bowl ads every day because people are only going to come in and they're going to leave after a week.

Sure, sure, sure. But generally just looking at these new cohorts, what percent— and I think that this chart is not nearly as impactful if you haven't worked on like a consumer mobile app, right? Because this is like beyond best in class. But I think this is what you would expect, right? People try ChatGPT, they love it, they stick around, it becomes a part of their life. It becomes like, you know, a kind of muscle memory quickly. It's become a, you know, a let's ask chat type of thing. So—

Look at the difference So the 10-month line for the most recent 10 months ago they joined versus 2 years ago they joined, from a 45% 10-month retention to 70% retention. That is just incredible. They're pushing up those lines. It's not just that the lines plateau at 50%. So for every person that gives it a try, there's a 50% chance that they just never leave, right?

Yeah.

And then also it's smiling, so you get people to come back. That's a great sign. But also they're pushing up the retention curve so much, so quickly. And yeah, I mean, this is just product getting better. But congratulations to the team over there. That is fantastic news.

Shane Copeland was—

Wait, hold on. If you're trying to run one of these analyses on your company, you probably need to use Julius AI. The AI data analyst that works for you. Join millions who use Julius to connect their data, ask questions, and get insights in seconds.

Billions very soon.

Trillions.

Shane Copeland was locked in at— is this— I don't know what sport this is, but I can see on the jersey it's Cleveland.

Cleveland. Some sort of crosstown rivalry going on.

Some type of stadium. Very cool. Courtside. Love to see it.

Yeah. He's locked in on the phone. That's very fun. Dialing in. And everyone's saying, oh, he's trading. He's front-running.

Okay. So I'm worried about getting one-shot by AI again, but Melissa Chen has a post here. She says, actual shot, not AI, of a French detective working the case of the French crown jewels that were stolen from the Louvre in a brazen daylight robbery. Somehow he looks like he's smoking even without a cigarette in his hand. But surely everything you know about life is screaming at you: this case is officially screwed. To solve it, we need an unshaven, overweight, washed-out detective who's in the middle of a divorce, a functioning alcoholic who the rest of the department hates. Never gonna crack it with a detective who wears an actual fedora unironically. I mean, This guy really—

There are so many weird things about this photo.

Okay, okay, okay. Yep. I knew this was too good to be— too good to be true. The community note says this is— this photo is real, taken outside the Louvre after the jewel theft, but the man in the fedora is a passerby, not a detective on the case. Didn't stop it from getting 42,000 likes.

But look at the— look at the woman in the back. All the way on the right, her face is like perfectly lit. Like, this is a remarkable photograph just from a dynamic range perspective.

This is—

Yeah, so my putting on the tinfoil hat, I think this guy with the fedora is probably one of the thieves that dressed up in an old costume and is coming back to the crime scene.

I like this.

I like this.

You know, because obviously the thief wouldn't come back to the crime scene. So he's kind of trying to take himself out of the pool of suspects.

Even if you take off the tinfoil hat and just go with the narrative that's here on the timeline, it is hilarious to be like, yeah, I'm going to dress up like this and then go be a passerby at a jeweler's.

He's like, I'm going to go get a fit-off.

Yeah, he's like, I got to go. I got to go see what's up at the Louvre.

I have to be photographed.

Yeah, for sure. For sure. Drip— he's a dripped-out Frenchman. For sure.

Dripped out.

Let me tell you about Privy. Privy makes it easy to build on crypto rails, securely spin up white label wallets, sign transactions, and integrate on-chain infrastructure all through one simple API. Do you want to do this uncapped clip? Look at the lighting. Jack Altman, your team is on a tear. This is beautiful. Good job. I love this. I haven't seen the clip, but let's react. What is it?

Learned at CAA is stark Most people don't have it. Some do.

What is it?

You just know it when you see it in a room. Tom Cruise is, you know, maybe my favorite actor growing up. I've seen Top Gun a zillion times. I loved Maverick. I had the opportunity to meet him one day in a very kind of random setting. I was delivering a package to him, and, you know, he shook my hand and looked me in the eye, and we talked for about 20 minutes. for 3 minutes. And, you know, for 3 minutes I thought, wow, this— no one cares more about me in the world right now than Tom Cruise does, right? The way he just commands a room in his presence. But also when Colin Farrell—

You heard it here first, command the room, no excuses.

Never been in a movie before, and you spent 30 minutes in a room with him, and he just had so much magnetism about him, the way he composed himself and talked to and looked at you and just his general persona. I do think that to me, I do look for that in founders. It's the combination of a mind at work and an opportunity that they're addressing. I remember when I first met Evan from Snap, he basically was making the argument that But look at the generation of young people coming up with the Iraq War essentially having been, you might say, a hoax, right? The WMDs were never there. Rolling into the financial crisis, oh, you told us we had the best economic system and then it almost collapsed, right? And then COVID, right? He built SNAP as a platform as a reflection of those trends, right? So what was it? Well, everything disappears, right? No one kind of stores your data. You can't trust institutions to look out for you. I think it ended up being incredibly prescient of where we are today as a world, right? Where the institutional breakdown that we've seen. So I think that—

Hmm.

That magnetism about a person in a room and an idea and a market that they're going after.

Be goated. No excuses.

That's my takeaway.

That's the right takeaway.

Be goated.

John, by the way, we can lightly hear you snacking on those pistachios. Just a heads up.

I am snacking.

Please, no snacking ASMR on this show. People can find that elsewhere. We have some breaking news.

What's that?

You've heard about funds that directly buy equity in companies. You've heard about secondaries funds that buy stakes in other funds. And we've got something new. We got some breaking news. We've now got tertiary funds.

Quaternary funds next.

Layla says, I can't believe it's come to this, but here's a visual representation of this ridiculous structure. Bottom is actual portfolio companies purchased by PE funds 1 and 2 with respective LPs. A secondary fund comes along, buys LP interest in funds 1 and 2 from some LPs. This secondary fund raises money from other LPs. But now, we have a Tertiaries Fund. This fund will buy secondary interest in Secondaries Funds by raising money from other LPs. Does anyone even remember how many layers back the actual value is created? P.S., not to be confused with continuation vehicles. Stay tuned for that visual. So—

Yeah, this is starting to look like the collateralized debt obligations, credit default swaps, From the mortgage crisis. It's very similar. Like, remember in The Big Short, there's like the— you have your stack of champagne glasses. You pour the champagne in one glass, and then it flows down to you.

You didn't see that one, did you? No, I have seen that one. One of the 5.

Yeah, yeah, yeah. It's very, very similar, where at every tranche, you recollateralize and reaggregate, and ideally distribute the risk. But if there's correlation, then there can be problems. I don't really know how big this stuff is, and certainly not publicly traded yet. But wow, to be a tertiary GP—

This is asset managers doing asset management stuff.

To be a tertiary GP, that's a life.

Yeah, a lot of Big Short memes going around.

Yeah, we should have one on. There's probably a bull case for it that's actually pretty sound. It actually makes sense.

Yeah, I can see it making sense in some circumstances. I think the concern is no one can get liquidity, so you're just passing the buck, passing the buck. GPs are harvesting fees at each level, but there's no actual returns flowing from the underlying value, right?

Yeah, I believe it.

What else? YASA, owned by Mercedes, recently smashed their own record for motor weight, size, and horsepower. This production-intent motor weighs 28 pounds and produces 1,000 horsepower. Kyle Chords in the comments says, I do not understand the point of this. There are already 5 EV brands with amply good enough motors. The motor power density is nowhere near the limiting factor on EV profitability or adoption. If they can make an adequate motor cheaper, that would be much more valuable innovation, though still not the limiting factor. Yeah, I have no idea what their intention is for this motor. It could be they want to just use it in Formula 1, right?

I say, why not? It looks sick. Yeah, this is— I mean, they'll put in something.

People are—

why not 4,000 horsepower? Like, there seems to be no limit. Like, the appetite for horsepower is sort of endless.

Yeah, and somebody here says, wouldn't this also be useful for drones, copters, planes? And the original poster says, absolutely, yes.

And also the G63 Landelay, the G63 6x6. These things need—

There's an EV version.

Imagine if you put one of these at each— the G-Wagon has gone electric. What will they do with the 6x6, the 6-wheeled G-Wagon?

Rick Hoffman says, what are your intentions with my motor?

But imagine one of those at each wheel of the 6x6. You get 6,000 horsepower in a thing the size of a tank. It'd be fantastic.

More interesting news. Self-driving taxi company Waymo is now doing 876,000 rides per month in California, 6x increase over the past year. And a 69x increase since August of 2023.

Do you think they wind up working with Uber long-term or not? Ben Thompson was talking about, like, he hasn't been writing a lot about the Uber, Waymo, the self-driving taxi stuff, mostly because he thinks, like, he needs more data. Because even at Uber and Waymo, like, they don't necessarily know what the consumer behavior will be. Like, will the consumers stay with Uber? And they'll order a Waymo when one's available, but then when they'll want one app that can get them anywhere. And so if you say, well, I need you to drive me into the snow to go to Big Bear today, you want to do all that in one app. Or will people say, I use the Waymo app for most things, and then when I want to go into the snow and use a human driver for some specific route, then I'm happy to open up the Uber app?

Yeah, we should get Dara on.

Because they're actually sort of A/B testing this because some markets are just Waymo app, some of them are Waymo in Uber, and it's very clear that they're running an experiment.

Yeah.

We'll see. I mean, right now, Waymo is quite a bit more expensive on a per-trip basis than Uber. And so it's possible the rideshare market used to be bifurcated a little bit between Uber and Lyft. Lyft was always I remember when I was in college, I was using Lyft more because it was slightly cheaper. And it's possible that Waymo ends up becoming more of a luxury good and people just have the Waymo app and more price-sensitive consumers are just sticking with Uber. So we'll see.

Yeah.

Before our next guest joins, let me tell you about fin.ai, the number one AI agent. If you want your AI to handle Customer support, go to fin.ai. Our next guest is Drew Houston from Dropbox. Welcome to the show, Drew. How are you doing?

What's happening?

How's it going, guys?

Great to see you.

We've been excited for this one.

Yeah, very, very excited. Where should we start? Are you a Waymo fan? Are you an Uber fan? We were debating, like, where does the consumer experience go? I mean, I know you're probably in the position where you might not have to choose. But it's an interesting question in my mind of like, will people want the, you know, the super app that can get them the robot taxi or the human long-term? Or if folks just wind up moving over to Waymo entirely?

Yeah, I mean, I think it's gonna be interesting to see how fast it diffuses. I mean, I think there's still like that, or that quality problem and then getting to that third 9, fourth 9, fifth 9. I mean, that's really what's— and Karpathy talked about that on why agents are gonna be further away than people think. And so, but I mean, I think it's, I mean, SF, it's like pretty well established. So I think that's probably a sign of what's to come.

Yeah.

Did the Karpathy, podcast update you on Dropbox's strategy at all? Or was it something that you were already hearing rumbles of and you were like, okay, now I have permission to fully write that to long-term memory? Or how were you processing it? Because I think there was nothing there that was completely, oh, no one has ever given this take before. It was more like, no one has ever said it in this context with this authority.

Yeah, it was It was somebody whose opinion holds a lot of weight that isn't waiting, didn't feel like he's waiting out for the next big tender offer.

Totally.

So, it felt less conflicted than some of the other opinions that get shared.

Totally.

Yeah.

No, I'm very aligned with it. I don't think there was, to your point, I don't think it was a lot of new— or I agree with a lot of these themes that the industry has been thinking about for a while. But yeah, sort of like a splash of sobriety on the space. And I think self-driving is a good comp in some ways for why AI or AGI or this fast takeoff might not be as quick or on the kind of timelines that people are thinking about. Because yeah, Karpathy talks about this march of nines. It's the same amount of work to go from 90% reliability. That's like one unit of work. Then to get to 99% is like the same amount of work.

Then Yeah.

99.9%. I mean, we've certainly seen that in our— as we've worked with AI and as we just work on reliability in general, I think that's very consistent. And I think, yeah, the self-driving car took the world or captured the world's imagination like 15 years ago. And yet it's these like more intermediate forms of automation that come way before autonomy that create a lot of the commercial value and get a lot more adoption. So for example, like There's like billions of users of Google Maps today, but maybe thousands of people are going to get in a Waymo today. And then same thing for highway assist or what Tesla's done or some of these intermediate forms of self-driving that have maybe millions of users. And so we think about that a lot as we deploy AI. And you don't want to be either too early or too late, both with tech in general, but also with automation for sure.

Yeah, it feels like if you watched the Carpathia episode on Friday and you walked into the office on Monday and like ripped up your whole plan, like something's maybe going wrong with your leadership style. But it is an interesting frame to think about it in the decade of agents framing. And I'm wondering, like, you've been at this for 2 decades now, correct? Like roughly?

Yeah.

And so you're like fully equipped to think in decades because you have literally thought in decades multiple times, right? And so I'm wondering how you're thinking about the next decade. Like, in 10 years, what if things play out kind of to the mid-case and we get a lot of really powerful AI systems, but they don't come next quarter? How do you set your company up to actually capitalize fully over a nice long decade of just solid execution?

Yeah. Well, I think when AI first came along, people's imaginations immediately went to sci-fi movies and sort of the one omniscient, omnipotent AI assistant. But for a while, yeah, I think looking at— I was surprised when I was 30 when self-driving first came out. I was like, oh, yeah, maybe millions of people will be off the road. Maybe we'll never see another human behind the wheel when we go to work. And it just took a lot longer. And then I've been working with Even when you're just designing systems with engineering, and certainly as you start to apply machine intelligence, you see these dynamics where it's like the demos look super good and sort of convince you that it's right around the corner. But then when you actually scale these things up, it's a lot harder than it looks. And so we saw that even before AI with Dropbox to begin with. We could make a very good demo or a video in the first couple of months, but then to roll it out at mission-critical and massive— to roll out such a mission-critical product at massive scale, you can't have a bad day, and that just takes a lot more engineering. We see that with our AI products today. Along the way, we want to build the omniscient, omnipotent AI agent too, but when we actually look at customers and look on their screens, the more basic problem that people have that AI can help with today is just that Your screen's a mess, like work is a mess. You have like 100 tabs open. You're working across Google and Slack and your email and Dropbox and a dozen other things. The question I started with is like, how much is AI helping with that? The answer is like zero or negative.

Yeah.

The other thing is, I'm sure reliability was like a core focus for Dropbox during the early days, because If you're like an online file storage and then 1% of the time, a file just disappears into the ether, it's not saved, and then the user might have lost it forever, and that's going to stick in their brain forever. And so reliability at Dropbox means 99.9999999999 forever.

Yeah.

And it feels like the industry in spending a decade building building SaaS and building SaaS that was pretty reliable and having a bunch of infrastructure that makes it easier to deliver reliability, maybe forgot how important it is. And if you have an AI agent that does an amazing job 90% of the time and just totally botches it 10% of the time, it's not ready for production. And so you can have these magical— and I think in the context of customer support is a big one, is like, is your agent ready for production if like 95% of the time it solves the consumer's problem, but 5% of the time it makes them pissed off and they're just like, talk to a human, talk to a human, and they're just running down that conversation.

Yeah.

Well, I mean, even if once you get up to a million users, even if you have a 95% or 99% success rate, that's still like thousands or tens of thousands of people having a really terrible experience like every day. And this was actually part of the motivation why I started Dropbox. Like, step one was not like create Dropbox. Step one was like use all these other services.

Yeah.

And like have all these near misses or things where like, wait, um, what happened to my like tax return? It was just here a second ago. Where did it go? And going into the support forums of these other competitive— like, well, these other companies, these other products that claim to do this on paper, um, but then you go in the support forums and it's like battlefield medical tent. Like, people are like, oh my God, what happened to my wedding photos? And where's this? And where's— and I was like, I like can't believe that, you know, these things are so brittle. Um, I'm gonna be putting my most important stuff in here, even just as a customer of one like me. Like, I need something that I can trust. Um, and, and then that took a lot more like engineering rigor and discipline, um, than we'd seen in the space.

We've been getting into barnyard analogies around here. I want to take you through one, and, uh, I'd love you your feedback. So we think about a lot of the dynamics between the various players in enterprise software and foundation models as potentially mimicking the experience of a fox in a henhouse.

Sure.

And it might be the job of the CEO of the henhouse not to let the fox in the henhouse. And I'm wondering if that resonates with you at all in terms of if I'm I'm a foundation model company. I say, hey, I'd love to provide a bunch of great AI tooling to my customers, your customers, both. I will need all of your data forever, and I will need to significantly erode your moat or your customer retention. Is the job of the CEO right now to protect the henhouse or go on the attack and be more fox-like? Help me walk through this from your perspective.

Well, I mean, I think when it comes to using AI, it depends. I mean, first is, um, and it depends what your business is doing and how you're using it and so on. I think, um, I mean, certainly the CEO's responsibility to like strategically position their company and mitigate risk and things like that. Um, but certainly every CEO is thinking about how do I like usefully deploy AI?

Yeah.

Um, and so I think it helps to have an understanding of the technology. I think it's important to know who you're working with and what their incentives are. To your point, henhouses, I think a lot of people have apprehension about using some of these or about working with a foundation model company whose whole business revolves around taking data and sort of grinding it up into pellets for training the next one. And you watch Sora videos or whatever, and you're like, that comes from somewhere. Probably not their artists. So I think, and, you know, certainly in our business, that trust and privacy and not having incentive conflicts of like not advertising, not training foundation models is super important as, you know, all of our customers think about, you know, who should we partner with?

So is there a little bit of a tension there where you need to, Keep customer trust high and let them know that we're going to get you these AI features, but in the Dropbox way that is actually fully integrated. And we're not just going to take your customer record or your data and sell it off the back of the truck just to pull something forward. And that actually layers into the customer experience.

Yeah.

You might be a couple of months behind on a certain thing, but it's gonna wind up with a better experience over time. Is that accurate?

Yeah, and I think it's all of the above. I think it has to— it starts with just like your business and your incentives. Like you have to design some of this in from the beginning. Like, you know, we could probably make more money or be more profitable if we did advertising, for example, right? And so, and a lot of scaled, you know, the predominant scaled business model on the internet, certainly for consumers, is advertising. We're just going to draw a hard line. We're just not going to do that because of this trust conflict, this incentive conflict we would have. And even if our intentions are good, it's like customers just have to wonder, right? And then that becomes a big advantage because then we've been living up to that for the last 18 years and we have a track record of—

What is the specific concern if you were running effectively display ads? Is it privacy of like, I don't want you to use my data to target me with more ads?

Yeah. And really, this is up to our customers, not us. It doesn't really matter what we think. It really matters what they think. And I think, yeah, everybody feels a little weird when it's like the equivalent of like, even just like the TV in your living room, if it starts showing you— if it's like a smart TV that's showing— it starts showing you ads you didn't ask for, you know, it's like, whoa, there's a an invasion of privacy here. And so you have to be super sensitive to the psychology of your users. But I think at a more fundamental level—

Yeah, the other thing is like non-targeted ads.

I'm imagining the worst possible scenario that you are 100% not doing, where the file storage company edits ads into my files. So I'm like, wait a minute, I saved my high school transcript and now on page 6 of this PDF, they edited in an ad.

Well, the other thing is like non-targeted ads. were really frustrating because I remember in college I bought a Kindle and there was an option. It was like, do you want the ad-supported Kindle?

I went with the ad-supported one.

This is a very Amazon move is to be like, okay, for $100 you can get the ad-supported Kindle.

It's like $10.

Or for $120 you can buy it. And I'm a college student.

I got the bougie one. I got the no ads.

Yeah. So, but then I was getting like the ads that were— that when I had this that were popular were like the most— it was like romantic novels.

Yeah.

And stuff like that, like books that I wasn't reading. And so my Kindle would be sitting there and it'd be like, just absorb my entire life so you can target me with something cool for once.

Yeah.

Yeah.

Uh, interesting.

And the other thing is I think you have to focus. Like, you have to really be selective about the problems you solve and like figure about, okay, how do we take what we do today and build on that in a way, in a way that makes sense for like today's context? So for us, I mean, we always took care of your files. These are already like your most important information. But as we all know, like what used to be 100 files on our desktops, now 100 tabs in our browser.

Yep.

And then the big— a lot of the pain points we have are like, you know, instead of one search box or at home, I can Google anything from one place, search all of human knowledge. I go to work, I've got like 10 different search boxes. My experience is totally fragmented. And that's what I was living personally. That's what my company's living. That's what all of us are living. I mean, look under my Zoom window. It's all And we're all dealing with the same thing. And so we're like, all right, let's extend AI into that new world in a valuable way. Let's give people that one search box. And then another problem with AI is that our industry is training these— is spending trillions of dollars to train these models that can do quantum physics but don't know about— can't find your Q2 board deck.

Yeah.

And the missing link there is AI, these models can only be as smart as the context that you provide them. And that's really the bottleneck. And like, no one's focusing on that. And we— so that's why— that's how like we have this new product, Dropbox Dash, which really aims to close that context gap and finally connect your AI to everything at work. And so on the surface, it's an AI assistant, it's a search engine, it connects, it knows about you, knows about your work, knows about your company, it connects to all of your work apps. And unlike with a consumer AI tool or You know, where there's a lot of questions it can answer, a lot of questions, but a lot of questions it can't answer. So if you ask like, when does our office lease expire or where's that slide from last year's product launch, ChatGPT can't tell you because it's not connected to your stuff. And, you know, and then Copilot doesn't know about your Slack and Gemini doesn't know about your Salesforce. And so that's where Dash comes in to build this like universal context layer. And the front end is really around search and an AI assistant because that's what people understand. But here's— but we're like, this is a natural evolution for us. Dropbox has always been about organizing and sharing your information, or like organizing, sharing your most important information, connecting to every ecosystem, being trusted, having this track record of reliability and security. Okay, how can we— what is the manifestation of AI in our world that actually makes sense? And where is there a new problem that hasn't been solved?

I think it's just like the perfect strategy for you right now because We had this like search was great and then we wound up generating so much data that search, I personally, not a judgment on you particularly, but across all of my services, whether it's iMessage, Gmail, we talk about it all the time. It's not to hate on you. This is a technology-wide thing.

Right.

Search got really slow and not as accurate and it would, there were so many cookies that it was picking up. it would just get confused. And then we got like GPT-5, deep research, like the thing that can go work for 20 minutes to get you exactly what you need. And it's like, okay, now we need to actually bring these 2 things together and get really performant, something in the middle, and then size them. And you see this with the model router in the consumer product, but to actually bring something like this in the business context makes so much sense.

Yeah.

So I love that the pitch is not ASI for your documents. It's like, hey.

We're excited about that too, eventually. But in self-driving, they have level 5 full autonomy, totally self-driving, human out of the loop. But a lot of the commercial value gets created at the level 1, level 2, like, all right, let's give people a working search box. Let's actually give them AI that actually knows about their work. These are going to be the big businesses that get built along the way to that superintelligence.

Zooming out and going away from what's happening at the product level, what is your updated strategy around running the business and capital allocation? I know you're the single largest shareholder at somewhere around 20%. You've been buying back a bunch of shares, but what is the strategy and pitch on that front?

Well, mainly investing for growth. Dropbox has always been a pretty profitable business. And we took a lot of the consumer internet playbook in the beginning and built these like self-serve, viral, product-led motion that was able to scale to millions of, you know, hundreds of millions of people, millions of businesses. We have half a million paying businesses on Dropbox, $2.5 billion in revenue. But, you know, the new— yeah, there we go.

Hit that gong.

Yeah, we got to hit the actual gong.

Yeah, for $2.5 billion of revenue. Creating product-led growth.

We get excited about big numbers.

The godfather— are you the godfather of product-led growth? Is that fair to say?

I don't think I'm allowed to call myself anything.

OK, well, I'm allowed to call— I'm calling you the godfather of product-led growth.

No, no. I mean, if anyone listening is not familiar with the Dropbox playbook, please go study it. It is a clinic for value creation. It's a fantastic story.

Yeah, and, you know, we're doing that again. We're running a lot of the same playbook with Dash. And I talked a little bit about search and—

Yeah.

And AI, the sort of AI chat piece. But we're also just thinking very fundamentally about, like, all the things we kind of lost as we went from kind of the file and PC world into the browser and the cloud. And then it's not just search or AI. There's even just the concept of organizing things, as you were talking about. People are like, somehow we all got convinced that searching for things is like a substitute for organizing things. And it's just not really true in that all along the way, we had files have folders, songs have playlists, but you think links have— right? There's no container, like a collection of— there's no way to have a collection of mixed format stuff.

Yeah.

If you have a Google Doc and a 10-gig 4K video and an Airtable, there's no way to still organize. So, We have something called Stacks, uh, probably about to show up on this video.

Nice.

Um, that does that. But yeah, I mean, just looking at these very basic problems hidden in plain sight and, you know, you give people a search box that works, let people talk to all their stuff in natural language, organize their stuff for them, you know, go beyond documents to organize your images and videos. Um, and so a lot of ways where I'm putting our capital is just building the version of Dropbox that I would want to have today. Which is— and Dash is step one in that direction.

Yeah, and there's— I mean, we've covered a number of companies that have been going after enterprise search, and it feels like you were— and many of them have been able to achieve pretty wild valuations in a short period of time. But of course, you're not asleep at the wheel thinking, I don't want to dominate that market too. So it's awesome to see.

Yeah, well, and then I think the big thing our customers really want, or we're Dash fits in there versus Glean or some other companies is like, you don't have to spend 6 months and $50 grand to get started.

No.

And so one of the things we're launching today is like a truly self-serve version of Dash, where you can just download it, get your team up and running, link your apps, and just go. And that's something I think that's really been missing, because there's a lot of— there's a bit of a knife fight in the enterprise around a lot of different things. But then we look at any Most of our customers in the SMB segment or lower mid-market, there's a huge swath of the market has no options. And so, we think it took a lot longer because it's hard to make these things totally self-serve, but that's what we've done with Dash. And again, it's the version of Dropbox I would build for today's context.

Fantastic.

We won't keep you any longer. Obviously, you're very busy. We really appreciate you coming on.

Yeah. whole team on the launch. Yeah, appreciate it.

And we're super proud of it.

And next time, come on in person, come hang out in the UltraDome and put on a growth clinic.

I'm sure a lot of people would benefit from learning some of the story because from the godfather himself. Yeah, it's a fantastic one. The godfather. Thanks so much for coming on the show.

Great hanging, Drew. Have a great rest of your year.

Cheers. We'll talk to you soon. Up next, we have Adam Fry from OpenAI. That rhymes. I like that. The product lead for ChatGPT Atlas. We're diving into the browser wars. The new web browser from OpenAI had a phenomenal response. I don't know if you saw how many folks saw the YouTube video. I think it has over a million views now, but we will talk to Adam Fry about it. Welcome to the stream, Adam. How are you doing?

Welcome.

Doing very well.

Nice to meet you.

Busy, busy couple days for you.

Yeah, it's been a bit of a whirlwind. I think just like you said, there's been such a big outpour of a reaction. We're so excited to see the reaction and so excited that people get to use ChatGPT Atlas now.

This is not GPU constrained at all, right? Like, I mean, obviously it's like a, it interfaces with your core pool of GPUs, but it's not like a new, like an entirely new provision of GPUs, right?

Yeah, that's right. So, you know, it's, it's available to everyone.

Yeah.

So ChatGPT Atlas, you can go to the download link, you can download it. It's available free, plus, pro, everyone.

Sure.

You know, the agent part of it, which I know you guys are savvy about, and we've talked to our colleagues before, you know, that is more compute intensive.

Sure.

And so if you use some of the more advanced features of ChatGPT Atlas, where you ask ChatGPT to go do things, that can be more GPU intensive for our paid users for sure.

Is there an incentive to set people— I saw a screenshot, but you never know what's real anymore— that if you set Atlas as your default, you can increase your ChatGPT subscription limits? Is there some sort of incentive? We were just talking to Drew from Dropbox about, you know, product-led growth, how you actually, you know, increase customer adoption. Is there something going on there that you can tell us about?

For sure.

So, you know, ChatGPT Atlas is, you know, it's a browser and that's sort of at the core of what you do every day. It's sort of your operating system for your life, whether it's personal or at work. And so, you know, part of that is we wanted to really— we felt like the more people use it, you sort of discover these AI features. I think you guys have talked about this a lot in your show, but, you know, this new generation of browsers, there's a lot to discover. You have to learn how to use ChatGPT in your normal workflow.

Mm-hmm.

And so we've really wanted to incentivize people to really give it a shot. And the best part of these new browsers is using ChatGPT everywhere you go across the web. And so one of the things that we did to start this off was if you give it a shot for a week as a default browser, you can sort of extend your limits and really allow you to test out all those AI features.

That's cool.

What was—

was it always obvious that ChatGPT would build a standalone browser, or was there somewhat of a debate internally? I always felt like ChatGPT was a web browser, Right? It felt like it was browsing the web on my behalf and serving it up in an interface. And so, it feels like— it almost feels like this is the second browser that you guys are doing, but you're going and building something that looks more like a traditional browser.

Yeah.

It's such a good point. You know, there's that great stat of, you know, if you ask people off the street, you know, what browser do you use? Most people actually don't even know what browser that they use because the concept doesn't always make sense.

Yep.

So, you know, it wasn't obvious, but what it really started from was a very specific user insight. You know, we were seeing, you know, you guys probably felt this when you use ChatGPT, but you're on chatgpt.com, you're in a million tabs, you're doing a whole bunch of other stuff. You're in your docs, you're planning for your next show, and you're like copying and pasting back and forth between ChatGPT and all the things you're doing.

Yeah.

And so we just sort of sat back and said, you know, if ChatGPT is gonna be more and more helpful for you over time, and people are going to rely on it more and more for the work that they do, for their personal life, it has to be able to coexist where you're doing that work. And that's what the browser is, is where you're doing your work. And so we said, how can we actually bring ChatGPT into that when you invite it in to actually have the context of what you're working on? And I think that spark of an insight was really, once you believe that, we were like, Oh yeah, building ChatGPT Atlas is, is of course we've gotta do that.

What, uh, how, how are you guys thinking about what success looks like with the product over the next few months? I think you have an immense amount of pressure, obviously, that the ChatGPT retention data was flowing around the internet over the last, over the last, uh, and it's kind of hard to, hard to fast follow that cuz if you set, set the bar.

Yeah.

You know, here. Uh, but how, how, how are you thinking about it and how, And then I kind of have a follow-up question around, like, you have hundreds of millions of happy active users that you can be porting over to Atlas over time. But maybe first, like, yeah, what does success look like in the first weeks and months?

Yeah.

Great question. So the start is we think of it as a multi-year journey, right? People have been using browsers forever. And probably your existing one you've been using for Yeah. 10, 20 years. And so it's, you know, building a browser is sort of like, we joke, it's like you're moving around the furniture in someone's home. You know, you know exactly where this is, you know exactly where that is. And so change is going to take time. So we measure success over the long run. We want to be as helpful to our users as possible. But in the short term, you know, we're focused on retention. You know, we're focused on the— for the people who are giving it a shot, which is a lot of people because people are really excited about this. you know, are they sticking with it? And, you know, are they loving it? Are they sort of diving in deeper to using chat throughout, you know, as they browse the web to get advice on whatever they're working on as they go? And if we see those signals, which we're starting to see, that's when we sort of, you know, pour on the fire and sort of grow it and grow it. But so short-term retention. But we also understand this is, this is a long-term game for browsers. And, you know, we're investing heavily for a long time in it.

Are you thinking about mobile version?

We are bringing Experiences Atlas to mobile and to Windows. And so the team is, you know, furiously working on it. We want to bring it to as many people as we possibly can. That's why we brought it to all of our free users, all of our paid, all of our pro right away globally.

Yeah.

Um, we think we've got a great browser. We're really excited for people to keep using it.

Very fun. Yeah, it's going to be, uh, big shoes to fill since Sora rocketed straight to the top of the App Store when you go Well, when you go live on there, I'm sure it'll be—

I think Adam can take it to the top again.

I love it.

I'm a believer.

Yeah. No credit here. I love the framing of it being like a longer journey. Like, it just does seem like there's not going to be like the Sam Altman stealing GPUs moment for the browser where it just goes viral and everyone's like, I got to check this out, right? It's very hard to like accelerate something that's a productivity tool. And so Having that mindset makes a ton of sense. I'm wondering about new patterns, new language, new archetypes of features that we might be talking about in a few years after this all rolls out, if you see any glimpses of this. The thing that I'm thinking about is I was talking to someone on our team about Chrome extensions.

Mm-hmm.

And I was thinking about what the extension looks like in the future. And if there's some sort of like, I've prompted an agent to do a certain workflow so many times that then it kind of collapses it into some UI and it places it somewhere in the browser. And it's like, it's almost writing like the custom software. Like there's, you see, I'm kind of like painting a bunch of splatter on the canvas, but there's something there. How do you think about like these new patterns? How much will your—

Yeah.

thumb be on the scale, or do you want the community to customise things? Like, where is the line drawn? How are you thinking about that?

Yeah, so there's probably 3, 3 big things when I look forward a couple years that I'm really excited about. So the first is, you know, part of this Atlas launch was the agent in your browser where it can take action. It brings up a cursor, it does things for you. I'm really excited for that as it gets better. So it's still early research preview, but it's starting to do tasks where we've seen it actually be helpful in a amount of time. And I'm really excited for proactivity around that. Eventually, you can actually suggest, hey, I've already drafted 5 emails for you. What do you think?

Mm-hmm.

You don't even have to task it. And so there's this sort of proactive world of the agents that I think is gonna be really, really interesting and valuable for folks. The second is working with developers. I think, you know, as it's more computer-to-computer interaction, if you have an agent that you're delegating to, to go, you know, write a document, or go to a website, those websites will evolve to work better with a software that is actually going and clicking around. And we just are sort of scratching the surface. We're teaching these computers to click around the way that we have, but I think eventually the internet and website will evolve to co-work with agents to make things more productive for everyone. And then the third is, I think eventually models would be really good to create their own sort of applications within the browser. And so I think there's so many amazing places that this is gonna go, and this is just sort of the beginning of this journey. And I think to Jordy's point earlier, ChatGPT is already so retentive. If you can bring ChatGPT into the browser like we have with Atlas today, it's already gonna be so helpful to your life, and then it's just gonna get better from there.

Yeah, you guys need to figure out—

Can you share any numbers with us?

I wanna hear about this.

Well, you gotta, we gotta, I was gonna say like— Maybe just the numbers I think the most recent comp that we have to a scaled consumer tech company launching an ambitious product in a super competitive market was Meta launching Threads.

Yeah.

And when there's this early excitement around a product, you can see massive amount of downloads, and then people were ready to call Threads, basically say, okay, it's over. Good nice, good effort, basically. And then Meta figured out kind of the right flow to get people to cultivate a unique community there, get people moving over there consistently. And so I would expect to see something similar out of Atlas in that I think a lot of downloads and excitement and usage early, some people will churn, but you have that massive base of users that you can continue to show them. Like if they do the right prompt in ChatGPT. It's like, you know, I can do this on your behalf over here if you like. So excited to see how this plays out.

Yeah. And I think the key thing there is focus too and investment. And, you know, I get to speak on behalf of the team that's working on this, but they're over there working on it right now, just making it better and better. So if you stick with it and you look at the small signals, the small paper cuts that people are telling you about, we just keep filling them. keep getting better, it's— that's when your retention goes up. That's when people really enjoy your product. So, you know, the launch is one thing. We couldn't be more excited. The reception, to your question, has been immense. Lots of downloads, but not sharing any numbers today. But, you know, it's really about do people stick around? And we have to keep working on that to earn that from people.

Yeah, totally.

Makes a ton of sense.

Well, we'll hit the gong once you're charting.

Okay. I was ready to go. He was ready to go. Thank you so much.

Thanks for breaking it down, and congratulations on the launch.

Yeah, this is great.

Thank you guys for having me on. It's such a blast. Appreciate it.

Always a great time. We'll talk to you soon. Cheers. Our next guest is Ian Rogers from Ledger. He believes he's already in the Restream waiting room.

Ryan, in the chat, I have to address it. Well, Taylor says he laughs and says it's a good chromium wrapper, sir.

I think big things have small beginnings.

Chromium wrapper company of California. No, I think that they have the incredible advantage of being able to have a massive user base and get people over there, continue to iterate, and it just—

You know what else is crazy? I mean, the VS Code story. The browser wars look a lot less crazy when you think about what happened to VS Code, open-source code editor. You get Windsurf, you get Cursor. People really did build serious businesses off of those. Anyway, we can talk about that more later. For now, we have Ian Rogers from Ledger in the Restream waiting room. Let's bring him into the TVP and UltraDome.

How are you doing? What's happening?

Welcome.

I'm good. How are you guys?

We're fantastic.

It's 10:30 at night here in Paris.

Oh, thank you so much for setting it up.

You're looking, uh, looking quite sharp in that jacket too.

You look great.

You know, I was thinking, I was like, wow, I— well, I just came from— we've had a big day today when there's a party happening at the Ledger offices, and I just came from that. So this is what I was wearing, and I'm like, wait, I'm in my living room. Should I put my pajamas on? Like, what's, what's appropriate here? But you know, you guys are always dressed sharp. So I thought, you know what, I'll just keep it. I'll just keep the jacket on. Well, you know, even though it feels a little out of place in my living room, I got to be honest.

I did not think that this week we'd be talking to someone in Paris at the seat of the crown about security. Tell us what Ledger does. How are you going to say— are we going to store my valuables reliably?

Not focused on royal jewels, unfortunately. Otherwise, it may not have happened.

By the way, by the way, I had, I had coffee at Ledger. with a number 2 from the Louvre last Tuesday morning. I was asking him, like, you know, what's it like? Tell me about the job. And he's, you know, pretty, pretty, pretty mellow, long-term. We think long-term horizons, these sort of things. Wow. That job got a lot more interesting over the weekend.

It did.

It did.

But anyway, yeah. How do you introduce the company now? Give us the general update.

So Ledger's been around for 11 years, which is a very long time. in the world of crypto, but really always doing the same thing. It started with this simple idea that we have these amazing devices, phones and computers, but they were meant to share information, not protect value fundamentally. And if we're going to have digital value in our lives, and especially if the main invention of this new digital value, Bitcoin, et cetera, is permissionless money, well, then you have self-custody. And if self-custody, then security is paramount. So, let's Let's use secure element chips to protect private keys. And that started with really one purpose, Bitcoin, but then we've had much more cryptocurrency. But actually, the fact is you can use that same device to secure anything. So, passkeys, identity. We live increasingly digital lives and proof is increasingly important, and therefore security. The other interesting thing about the way it's— if you think about it, the technology and where it comes from, it's this smart card technology. So, it's the same chip that's in your credit card or your passport, but where your credit card protects the secrets of the bank and your passport protects the secrets of the government from you because you have it in your possession, you're not supposed to hack it, this is using that same technology to protect the secrets of the user or of the consumer. which is pretty powerful, especially, I mean, I was listening to your interview about OpenAI going into a world of agentic AI.

Yeah.

We will need to prove our humanness. It would be nice if we could own our own preferences and then, you know, federate them to the apps that we use. And I'm not sure I want to shove my credit card into that agent. I'd rather it asked me, you know, before it booked the flight, you know, like, here's the info. Would you like to verify this transaction on a secure screen? Yes, I would. Thank you. So, it's that. And today, we announced a new product, but that's what we've been doing for a few years now.

Yeah.

What's different about the new product? I feel like Ledger's— it feels like something where it's like, great idea, solved. And so, I'm interested to know what's happening with the iteration cycle. Yeah.

So, we have this product that people are very familiar with, the Ledger Nano, which is— it's just like the swivel, it folds out. You've seen it, which— and it's kind of the Air Jordan of crypto. If somebody needs a physical representation of crypto, they include a picture of the Ledger in it.

Oh, yeah.

But with Tony Fadell, who is the founder of— the Podfather, he created the iPod and co-invented the iPhone and founded Nest. I mean, you guys will appreciate this. When I started this job, I told Tony, I said, yeah, I'm going to run the consumer business for Ledger. And he said, there is no consumer business. That business is business to geek. Your job is to make a consumer business. And so, we started making these secure touchscreen devices. So, he designed one that's beautiful. It's called Ledger Stax. It's the world's first curved e-ink display. So, where a Kindle screen is silicon on glass, this is an organic substrate on plastic, and you get this like 5 millimeter curve. And, you know, it's Tony. He can envision the circuit board and the billboard at the same time. But that screen, first ever, any first product to have an organic TFT, took longer than we thought, and the screen ended up costing more than we thought. And then the product is in market for a higher price than we had hoped for. So we built a That retails at $399. And then we built another one. It's a beautiful device called Ledger Flex, steel case, really nice E Ink screen, but that clocked in at $249. And what we really wanted to do was to do this secure touchscreen, but at the price of a Nano X, which is, you know, one of our most popular models of all time. So, this is the one that we introduced today.

This is the Ledger—

Nice.

This is the Nano Gen 5, and this is on sale today for $179. So here you can see what the, you know, what the box looks like.

Very cool.

And that's— so that's what we did today. The other thing we did today is announced a bunch of features in the software, because really, I mean, you know, the point— kind of to the point you were making earlier, it's not about hardware, it's about what can you do with it. And these devices, they started as one thing, this long-term investment asset, Bitcoin. That was what it was for. But now, it's all assets because it's tokenized RWAs, tokenized stocks. It's also a yield-bearing savings account with stablecoins these days. And it's a spending account as well because We have a credit card that's a— well, it's actually a debit card, a crypto debit card that pays Bitcoin rewards. So, it's so much more. We support 10,000 tokens, 100% of the top 100 tokens. So, it's really a financial tool in your pocket and a secure one, which your phone is not.

Talk a little bit more. I love how many different Like really thinking about the full suite and just building these like purpose-built devices for different use cases. Talk about, I guess, more, you mentioned earlier, if you have agents operating in the world on your behalf, specifically online, like how you guys are thinking and kind of even timeline there, because I think this is top of mind for a lot of people that are realizing like, okay, I'm gonna have a browser that's not just like a window into the web. and a way for me to enter into forms, but it'll be acting on my behalf. And if it's going to be moving money or doing anything else or submitting my information somewhere, I want to have more real control over that. So I guess wanted to hear more about how you guys are thinking about that opportunity in the context of Ledger's business.

Yeah. I mean, I guess the way I think about this, I think about it all in context, right? What was the internet really? 2 computers could talk to each other. We get this revolution of information. What is blockchain? Well, blockchain is the ability to issue and trade, really, right? So, I can— we talk about scarcity, but it's really about issuance and trading, the kind of what's the core building block. And then, with AI, we have superintelligence and ultimately things that we can delegate tasks to. that will act on our behalf. But if they're going to act on our behalf, then they have to represent us out there, right?

Yeah.

And also, when we're interacting with other people, we need to know who it is we're interacting with. And so, also, I think those things need to be machine-readable. So, interestingly, in both cases, you're talking about tokens, right? We talk about eating tokens in the case of AI, and we talk about issuing tokens. In the case of crypto. But what we mean when we say those things is we're talking about things that are machine-readable and machine-translatable. So, I think proof is actually a really important word to consider here. How do you know this is me talking? How do you know these are the words that are coming out of this microphone? I think proof is going to become increasingly important. We already have it in a lot of companies. are like trying to take this challenge head-on. You know, Reddit is trying to know that someone's not a bot on the other side. You know, dating sites— Match.com has a big effort to try to know that someone's— so how exactly are you doing that? And, you know, who are you trusting? So I think trust brokers will become pretty important in the future. You know, like if I had— I could— if I said like, ah, yeah, here's a Warhol print, And you'd be like, okay, well, maybe it is. But if Christie's said it's a Warhol, you'd be like, yeah, it probably is. They probably did the work. And that's the same thing that we have in our lives with things like driver's licenses. It's difficult to get a driver's license. It's easy to read one. And I think we'll have a lot of those kinds of things where there's a trust broker. That trust broker will issue a token that says, not only am I a human, but I'm a unique human.

Yeah.

This token matches you know, matches a passport, one passport, not multiple passports. And then, you know, and then a machine-readable way, in a machine-readable way, you know, we'll be able to kind of do a handshake where it's like, oh yeah, that's a real person. And you could use that in, you know, any context, Twitter, YouTube, or X, YouTube, etc., right? And so I think that these kinds of, you know, this is, this is like the super basic version.

Yeah.

You know, the more, the only slightly more complex version is that, agent that they were talking about earlier in my browser, it's going to go book me a flight to Miami. Well, how? Have I given it my Delta login? Did I give it my credit card information? Right now, we live in this world where kind of our identity is locked behind all of these login password combinations across thousands of websites.

Yeah.

There's no need for that. All of that data could be owned by the individual, just like we own our passport and our driver's license and other things. And then it's federated to those applications as needed. It's really a much— it's more scalable, it's more secure, it's a much better— and it's less trust. I think we're already— I mean, I haven't installed the OpenAI browser yet and I'm like, wow, do I really? They've already got so much of my data.

I have an idea. I have an idea.

Bring it.

A giant gong that you can store Bitcoin.

Yeah.

What about that? What about that? A massive gong that you keep in your vault in your basement and you can store Bitcoin on it.

You can smash it every time you want it.

You could actually probably just sneak a little Ledger Nano up in the gong and that would work fine. But I want the full gong.

I was holding back. I was ready to tell you guys. I was ready to tell you guys.

Yeah, yeah, give us some numbers. What you got?

I was gonna say approximately 20% of all crypto is protected by Ledger devices.

Yes!

I was like, what can I give you that's worthy of a gong? And, uh, that's it.

Okay, that's insane.

Fantastic.

You guys probably— yeah, you got 2, 2 hits.

Uh, we're, we're running over time.

Last thing I wanted to ask, how— what's the Parisian crypto scene like? It sounds cool and you seem really cool, but But what's it like on the ground?

Thanks. I have to say, the Ledger party tonight was crazy. I barely escaped from there because there were just people in the street going, can I get in? No, it's interesting here. I moved here 10 years ago. I moved here 10 years ago to work for LVMH. And a lot— I mean, I really moved to LA in '95, and it felt like it was a million miles away from Silicon Valley at that time. You know what I mean? But then, Over time, it kind of grew into something that we had a MySpace, we had a Snapchat, we had a— LA kind of came into its own. And you feel the same thing here over 10 years. It's still France, and France is not America when it comes to startups, that's for sure. But they've really— I mean, there's so much going on here. It's really fun. Get over here. Let's do this. We'll give you the proper tools.

I refuse. No, I refuse. I'm not leaving America. I'm sorry.

Jon won't leave our borders, but I will. I'll take you up on that. Okay, cool.

Come on over anytime. Does Jon have a passport?

He does.

Unfortunately, yes.

Yeah, he does.

But otherwise, I could say no to every international invitation.

You go, Jon, come over here. We got seas and lakes and mountains. And he's like, we got seas and lakes and mountains over here.

Yeah. Do they have the Indianapolis 500 over there? Do they got the Kentucky Derby over there?

No, I am, by the way, I am from Indiana.

No way.

And I'll give you a fun fact. Kid Rock is on record saying, the only reason I would go to France is to visit Ian. So maybe, John, you could come.

Wow. Oh, that's lore.

That's deep lore.

That's deep lore.

Amazing.

Ian, great to meet you. Congrats to the whole team on the launch. And thank you.

Thank you guys for having us on.

For securing 20% of all digital assets.

Yes. Thank you for your service. Uh, we'll talk to you soon.

Have a good day.

Cheers.

I will see you again.

Our next guest, Molly from Knox, is already in the Restream waiting room. We're going to bring her in and ask her how she's going to let you interface with both iMessage and WhatsApp at the same time. Do I have that correct?

Yes, you do.

So this is like a total moonshot company. This is harder than colonizing Mars. This is trying to get Tim Cook and Mark Zuckerberg to play nice. How are you possibly going to pull this off?

Yeah, message interoperability has been one of the problems that just people have talked about for a while. I think the way that we do it is we have this insight where if you layer onto people's computers and you don't distribute within the App Store, you have more interesting things that you could do. And so it's actually possible to do exactly what we want to do here, which is connect iMessage and WhatsApp together.

Yeah. So I imagine that like If I was super, super hardcore, I could have like a Mac Mini that's running with literally iMessage and WhatsApp open and an agent that's clicking and screen scraping and uploading to a database and then surfacing that in an app. And it's gonna be very hard for WhatsApp and iMessage to like shut me down if I'm doing that. But that's a very prosumer behavior.

You need your own private computer.

How far are you from there?

Yeah.

To the actual product.

Well, what's interesting is the first version of the product was effectively a cloud server that we would ask people to log in. So, give us their iCloud details to log into, which, you know, Trey has a funny story about this. But at a certain point, I think you realize that it's just not the most effective and the right way to do it. It's not the perfect answer. And so, because it feels inelegant, There's a better way, which is what you described. Essentially, if you can insert mouse clicks and be indistinguishable from a human from the OS level, so Apple can't tell that you're a human versus you're automating your computer, it's over. It's like, okay, great. That's where you want to be. So we've thought a lot about that question. And yeah, certainly some of the features sort of play with that aspect.

I mean, do you have to like message the prosumer nature? Do you think that there's a world where you can get big enough that then you exert enough leverage over the platforms that they have to play nice with you? Like, is that the long-term goal to be able to walk into Cupertino and Menlo Park and say like, look, like there are gonna be millions of people that are upset if you shut off my API access. So let's make this easier for everyone.

Yeah. I think we already see it moving in one direction with RCS. It's probably going to continue moving that way.

Yeah.

So that would be amazing. On the prosumer thing, I think it just comes down to regulation and rights and what you can lobby for. So yeah, there's—

How solid, how important do you think iMessage is to Apple's position? Because recently we've just come to this like weird realization that it's like my, my, it's my number one social network. It's like how I plan my social life, my family life, my friends. I'm sure you're the same way. I have group chats with business friends and all these social platforms are like where I consume news and content and like people that are like friendly acquaintances. But mostly like my social network is iMessage and that feels like they're gonna be able to charge whatever they want for me to keep that every year when they release a new iPhone. So do you, do you buy that thesis as well, that it's very important to Apple's position in the market?

I think it's hands down the most important factor as to why people are buying iPhones, because no one wants to be the icky green bubble in the group chat where all of a sudden nothing works in the group chat and things are delayed and just disgustingly, you know, designed. So at a certain point, you don't want to be that guy. So you just herd mentality and sheep into buying an iPhone. But I think a larger point here is like, I think a larger point here is It's just like when you receive a blue bubble from someone, there's already this preemptive—

Okay.

Another farm. This is—

wait, sorry.

We're trying to build out the ultimate farm analogy for attack because you have, you have the trough, the slop trough. You have people who are goated, but then you have people who are sheep. And so you just helped us create a new link in the chain in the diagram. But anyway, sorry, continue.

Goats need sheep, you know?

Yes. What is a goat without a sheep to contrast with? Sorry. How's the actual product launch going? The rollout? Are you in beta? Is it handholding customers on? Are you general availability?

It's really good. Yeah. So it's general availability. Anyone can download it right now, at least on Mac. And then if they want iOS, they can request access through me. So, we have been getting a lot of people today just going on. The onboarding experience is mind-blowing. It just tells you your life recounted through you, through, you know, all the conversations and top people that you've met. And what's funny is I was texting Ben before this, trying to coordinate like logistics.

Yeah.

And I found out that my messages suddenly turned green and it was because I hit a quota limit Because this morning I sent out, I think, 3,000 messages to people.

Oh, wow.

All in my one-on-one group chats. Like, hey guys, can you help me, you know, post about this and, and help me support here? And yeah, so I guess I re— I realized what the ceiling is on, on, um, on the rate limit.

Oh, that's always a good sign.

I'm wondering if I'm gonna get rate, rate limited on my, my inbox. Um, I have 5,004 unread texts right now.

Uh, one, one of our friends. One of our friends, Will Manidas, has been talking about the desire for Bloomberg for private markets. And he kind of outlined it as iMessage plus Signal and X and Twitter as the main ones. So have you thought about that? And what are the nuances of how hostile each API is, basically, is what I'm looking for, like a state of affairs?

Yeah, I saw— I think it was Will's tweet. I saw it, and I thought it was— is interesting. On our roadmap, we have Slack and Discord, maybe email or a few other messaging platforms. But I think the big thing is actually 80% of your messages come through. It's not an equivalence, right? Like 20, 20, 20. It's like mostly people are on one thing and then they'll text this secondary platform for a specific group of people in a niche and then another maybe 5% of the time. And so, Um, I get the sense that it's, it's mostly okay to just focus on iMessage and WhatsApp right now, and then you can add integrations later or even have the community help you build these integrations. Um, but there are private APIs and I mean, Telegram API is easy and some other ways that you can. Yeah. Yeah.

I, I, I, I really do wonder about that because it feels like—

Yeah. To me, to me, to me, it feels like the more platforms you add, obviously the more valuable it is.

Yeah.

Because it 2, I probably get 80% of my texts on iMessage, 10% or maybe 5% on WhatsApp, 5% Signal.

No, it's definitely a power law. Like, iMessage is the most important. I wouldn't even put WhatsApp— I would put WhatsApp like 4th or 5th for me personally. It would go iMessage, then probably X DMs, then maybe Instagram DMs, then WhatsApp, then I would even put Signal above WhatsApp. But the point is that like a one app to rule them all, like if it's going to be a standard in my life, that's like the one unifying interface, like I want it to be across everything, I think, the control plane. But I don't know. There's a lot of— with all these like unified personal assistant stuff, there's a lot of expressed preferences versus revealed preferences. Like people might actually use it a different way than they say they want to use it.

Yeah.

But I don't know, it's going to be a fun, fun building process for you. So I'm very excited.

Yeah, it's, it'll be cool to see the longitudinal access come in too, right? Because like our thread on X will be different from our thread on email will be different from our thread on iMessage. And if you can just solve that problem of context coming in from all these different places, um, that would be huge for both people and companies. So it's going to be really exciting.

Have you learned anything from the previous attempts at this, like the beepers of the world?

Yeah, you definitely don't want to poke the bear and do the thing that feels not right and inelegant for too long.

Mm-hmm.

I think the first launch we had, which was sort of a semi-private beta launch in February, we had a lot of people come on and tell us, hey, Apple's watching and, you know, maybe would have issues with iMessage. I think the main thing is just be nimble and agile and when APIs change that We know that have changed, we just automatically update, right? Day of, 4 hours later, a new build is out. And so nothing really breaks as long as you are on the system all the time.

Keep shipping.

I mean, 4 hours in text message land is a long time for some people. If they're like, I didn't get the message from my wife that I need to pick up the kids or something like that, does seem like a tall order. But that's why they invented 996 or 0.002, as the Wall Street Journal put it more recently, which is basically no time.

I think, uh, but yeah, everyone, everyone has this problem. Yeah, it's very hard to create an elegant solution, but if you do, it will be incredibly valuable.

Yeah, well, uh, good luck to you. I don't know if we lost the video, but thank you so much for stopping by. Uh, we will talk to you soon, and have a great rest of your day.

Congratulations.

Sorry, that, that was kind of a ending. But we have Johnny Dwyer from Muon Space, his second time on the show. And we had a great— we actually played a clip from our previous interview yesterday, and we're very excited to talk to him about both about what's going on in his world. But I'll welcome him to the show now. How are you doing, Johnny? Good to see you.

Hey guys, good to see you guys again.

Welcome back.

Thanks so much. We enjoyed the— we enjoyed the previous guest segment immensely. We played a clip from that segment yesterday. On space data centers. Yeah, we were trying to contextualize space data centers. We're actually having the founder of the space data center company come back on the show to defend what's going on since a lot of people are skeptical. Where do you sit on it now? This all feels like I'm not ready to call it even a binary. I think we should be discussing timelines more than will it ever happen. But how are you thinking about space data centers?

Yeah, I mean, I think it's, it's, it's like many things, it's a when, not if. You know, is it a year from now? Is it 10 years from now? Is it 20 years from now? I don't know. But I think as you think about generally the future of the planet and where we're going on it, more and more stuff is going to move to space. And so, I think it's inevitable at some level that we will have large-scale, you know, IT infrastructure in space. I mean, you know, we're probably going to talk about or deal with Starlink in a minute, and that's an example of a place where I think we're starting to really see what I would call truly modern IT infrastructure starting to show up in space. And I think it's inevitable that data centers will go there as well. A lot depends, I think, obviously on the fundamentals of what is going to drive the need for future data center capacity, right? And how some of those needs are met, you know, on Earth and then how that would compare to deploying things in space. I think if you ask anybody even, you know, what a terrestrial data center will look like in 5 years, they're probably not going to give you a real confident answer. So I think, you know, trying to predict exactly what that's going to look like in space is hard to do right now.

Absolutely. What— yeah, what more can you say about this new partnership?

Well, yeah, I mean, I think, you know, one of the things we like to think we're doing at Neutron is really pulling a lot of the aerospace industry kicking and screaming out of the Stone Age. And, you know, you think about a lot of the stuff that's circling the Earth right now, and it's still connected with things that look like dial-up modems, frankly. I mean, I think we're, we're kind of in an era where a lot of the space technology is a few generations behind what we take for granted on Earth. And this partnership with SpaceX is now going to give our satellites things that look a lot more like an always-on fiber broadband connection. Rather than a dial-up connection. And so, that's going to completely unlock, you know, a tremendous set of things that have been very difficult or impossible to do in space historically, much as, you know, Starlink has demonstrated with terrestrial use cases. You know, an example that I'm, you know, we're all very excited about at Nuon is we've been working on this global wildfire constellation for the last several years. This will eventually be a 50+ satellite constellation that will get 20-minute global revisit latency to detect and track wildfires from the earliest stages of their lifecycle all the way through to help decision makers on the ground make better decisions. And a really critical piece of making that system work well is getting that data back to people that are making decisions at very low latency. These satellites produce enormous amounts of data. There's going to be 50 of them. So having this kind of fiber connection in the sky is absolutely transformative in terms of being able to deliver a capability that's crucially needed around the world, and it has not been possible to date. So I think it's a really good example of what this technology can unlock.

Is fiber a metaphor, or are you physically putting fiber in the sky somehow? I mean, people are putting data centers—

We have big glass spools and we're kind of unwinding them behind the tower.

I mean, we're going to put data, massive 1-gigawatt data centers in space. Like, I wouldn't be that— it doesn't sound that crazy.

It's a humanoid up there with a spool and he's just letting it out.

So you're not literally putting fiber in the sky. It's more about connectivity between Starlink satellites, or are you putting up your own satellites that are acting as kind of backhaul between the Starlink satellites?

We're basically using Starlink as our backhaul. So you can think about them as sort of our network in the sky. And then we'll be connecting with SpaceX lasers or laser terminals, they call them lasers, into their network and using that network as kind of the backhaul system. Not literally unwinding glass fiber, but actually the technology's very similar to fiber optic technology on Earth. And a lot of the same componentry is used. We're just sending the light over these kind of free space optical beams, not over glass fibers.

And then who's your customer? Is it someone who has a satellite or a constellation in space that needs to move internet data around while in space, or is it more for speed up terrestrial uses in some way or just improve the Starlink network?

So we're building satellites for customers. So our satellites, you know, serve a broad range of customers. You know, one of the customers I mentioned is this— there's a global nonprofit that's funding the work on the FIRE mission. So ultimately, we're building satellites for them. They're muon satellites that we deploy and operate. And this will now provide sort of a much faster, lower latency network layer for us to operate those satellites on. Other examples of customers we're working with, you know, we have a customer called Hubble that's doing Bluetooth Low Energy tracking from space. So you can kind of think about like IoT device tracking from space. We have—

Wait, the company is called Hubble?

Yeah, it's called Hubble.

What a confusing name. I know that there was a contact lens company named Hubble, and that kind of made sense because it's like a wild Totally different market, but if you're out there in space—

So I can imagine the commercial use cases for that, right? If you have a lot of hardware out around the world and you want to track it in real time, does that have defense applications as well? What's the actual focus there?

Yeah, so I don't want to speak for Hubble. They would speak better for themselves than I will. Customer. But I mean, I think their primary focus is on sort of enterprise and commercial use cases. So, you know, there's millions of asset tracking devices in the world today that are out tracking, you know, shipping containers and remote facility equipment, things like that. And a lot of those are running on kind of these bespoke and frankly very old-school satellite networks. That's how they get the data back. And I think Hubble's vision is that, you know, we should be doing the same thing we're doing with consumer devices, which is is strapping a 9-cent Bluetooth transponder on anything that's out in the world and then being able to track that thing from space. And so, we think there's a lot, you know, I think they believe there are many, many commercial and enterprise applications of that. There probably are also, you know, government applications of that. I don't want to speak specifically to them, but I think that's, you know, it's the type— all of these types of technologies, especially with spacecraft, in some way are inherently dual-use, so there's always going to be use cases in both Give us the fundraising update.

I feel like we missed you around Series B potentially. Do you have news? What's up?

We don't have news. I mean, we raised kind of a 2-part Series B last year and this year. It's a big round for us. I think it totaled about $140 million. Oh, we got the gong coming. So yeah, we're doing well on the fundraising side. is we're well-funded. We've got a lot of business coming in. It's been very exciting days around the company. So it's a pretty, pretty exciting path forward right now. We're really just trying to scale the company up a lot. We have a big facility. We're in the Bay Area. We signed a large facility earlier this year. So we'll be kind of 10x-ing our production capability, our hardware production capability in the South Bay before the end of the year. And then building out a lot of other new technology pieces that we're really excited talk about as we start to hit space with them.

Fantastic. Well, thank you so much for coming on the show.

Yeah, great to get together.

Congratulations on the progress. And yeah, thanks for breaking all that down for us. Now we know there's not literally fiber cables up in the sky. That's why we asked the dumb question.

Figured it out.

Awesome. Thank you so much. We'll talk to you soon.

Cheers, John.

And up next, we have Mike Shebet from Traba. We have a video update from him. Let's play that.

Come on over.

What's up? You're welcome. Yeah, good to see you.

Yeah. Um, can we play the little bit of the video that he shared? Let's pull that up, uh, and get the update.

What's going on? Welcome to the Ultra Day.

Request. Trava qualifies thousands of workers for the best match.

This is, uh, is this your key hero marketing video right now?

Describe proper form for lifting heavy packages.

A quick reel just to show Sure, sure, sure. Because in staffing there's so many different manual workflows. This video just basically breaks it down with how it all comes together.

Okay.

Very cool.

How old is the company now?

We're four years old.

Four years old. Are you still in New York?

We're still in New York.

Manhattan.

Yep. All 150 of us.

150 now. Wow. Okay. So I I I toured the when did I come by and do an interview with you? That was maybe like. 2 years ago or so?

That was across the street. We moved across the street to a bigger office.

You moved across the street. OK. So yeah, take us through the basics of the problem, the market. It's a highly fragmented market. But what's the high-level, like, 140 characters that you describe Traba?

Yeah, so basically millions of workers across the country work in industrial staffing. And then hundreds of billions of dollars get spent on staffing. But if you break it apart, it's really just like A highly fragmented industry with tons of manual OpEx and things like that. So we basically replace the whole suite with tech, and we provide double the fill rate, better quality workers, and workers within a couple hours versus weeks.

Yeah. And the staffing industry in general, I think of it as highly lucrative but highly fragmented. Is that just generally your experience? I guess to just jump straight forward is like, why build tech startup instead of private equity roll-up? Like, you're in a dress shirt, you could go do a PE roll-up, right? Like, I don't know. You're ready for it. Like, it's not like you have to do it the Silicon Valley way. Why are you doing it the Silicon Valley way?

Well, basically because of the advancements in AI, this never really was able to happen before. But on the staffing end, there's just so much manual work and tons of OpEx. So we basically just like from a top to bottom, end to end, just have completely Sure. And that's why building it as a technology company has been so much better.

And so, when you say AI or tech, you mean lead scoring, basically, understanding who's right for what job, placing them, screening?

Well, it's like a ton of things. So, we do the vetting. We actually guide the worker to the shift. There's fraud detection and essentially follow-ups with the workers. And then, as the workers perform shifts, that feeds into our But it's not the AI doing the job.

It's always—

No, no, yeah.

You're sticking it out for the next decade.

Yeah, yeah, yeah. On the backend, the customer doesn't really see the AI. On the backend, we're doing all that work. So what the customer sees is just like an incredible group of people showing up for work every day and helping them out.

Did you see the Uber news that Uber drivers are going to be able to do one-off AI tasks?

I saw that, yeah.

What is your take on that? Have any AI companies approached you where you're like— just, I don't know if I'm describing the company correctly, but I remember a couple case studies is like there's a big Taylor Swift concert in town, and the organizer might go on Traba to pull in a bunch of folks really quickly, or if it's the rush holiday season on an e-commerce warehouse. I don't want to put you in a box, but those are 2 examples, right?

Yeah, yeah, yeah. Basically, manufacturing, fulfillment, food processing, all those industries just have Crazy volatility in their workforce needs. So we help smooth that out for them.

Yeah, exactly. And so my question is, have any AI companies reached out to you and said, hey, you have a big pool of labor, we have a unique use case that we could potentially partner on?

Yeah, we actually have gotten a few inbounds from companies who are just like, hey, we need these workers to do all these different things. But right now we're laser-focused on just industrial supply. So focusing on that, we may expand out Yeah.

Have any humanoid companies come to you and said, like, we'll give you robots, like, put them out into the field? Or are they not quite at the level where they actually would be willing to let them do unsupervised activity?

Yeah. So I actually, like, see this— I talk to our customers about this all the time. Right now, when it comes to humanoid robots, they're not quite there yet. But we're going to be well positioned to essentially lead that charge when the tech gets a lot better in about 5 years.

Yeah, because you're still going to have the variability in workforce needs, right? Where you might need— even if you have 100 robots in a facility, and then suddenly it's rush season, and you need to deploy 100 more on top of that, it's like, where do you get that?

Yeah, and then all these workers are going to essentially have to get connected to new earning opportunities. So we'll be able to help them with that.

Yeah, it's a little bit of, I guess, the narrative violation behind Like AI taking every job or AI being useless is like, I'm pretty sure Amazon is the largest robot operator, but also one of, if not the largest employer as well. And so they're really doing the both robots and humans at scale thing.

Within the facility, it's a lot more dynamic. And a lot of these companies, they have to change things really quickly. So to just commit to a certain robotics workflow doesn't work today.

Yeah.

Which is actually why, if you actually go back to Elon Musk's tweets in like 5, 6 years ago, he actually said, he's like, we tried to make the Gigafactory all robotics, and then we went back to humans because we couldn't necessarily get there yet.

Yeah, and they had to do a hybrid.

Yeah.

Talk a little bit about where the structure of the market is going in terms of staffing agencies broadly. I imagine that AI and technology generally are like an accelerant —growth, right? You can just onboard people faster, just using all the tools in the mobile toolkit allow you to understand where people are going, placing them, GPS, all this stuff is useful. How does that play out? I mean, I imagine that competitors are going to try and catch up. Are you planning to just outcompete them? Are you looking to eventually get into doing acquisitions? Like, how do you think the long-term plan plays out?

So I was in a staffing conference in Dallas, and it was pretty crazy.

Electric. Yeah, it was electric.

That's why we're here.

We do it.

Literally this year, everyone is like we are getting disrupted. AI is coming for us, and they were just talking about how like they can't organizationally roll it out.

Yeah.

There's just like thousands of people at these organizations that have all these different tasks.

Yeah.

Yeah.

So we they actually approached myself and my chief of staff, and they're like we want really want you to help us move into this new age.

Sure.

So when it comes to roll-ups, we're potentially looking at a few acquisitions, and in that case, it would basically It'd basically be like companies that have really good retention on the business side, so they have customers that continue to spend, but then they're just like, we have terrible OpEx and just so many manual expenses, and we want you to essentially improve these margins and carry it forward. That makes a ton of sense.

How do you operate the business? You're obviously incredibly aggressive, I'm sure, around growth, but at the same time, you also feel like somebody who you're planning to run this business forever, I imagine. And part of putting yourself in a position to do that is maintaining good unit economics and just running a great business. How do you balance growth and burn and those factors?

We're very thoughtful about capital efficiency with growth. So fortunately, in staffing, the margins are actually pretty good to play with. So yeah, we're just like, we're operating in the physical world, so there's so much that's going on. Which is why our team just works so hard to kind of get in there. There's not one specific tool that can essentially get cloned. It's like hundreds of thousands of workers that use our app and thousands of businesses. So, you get those real network effects as you scale up. But yeah, it's basically growth, unit economics, 10x customer experience. That's what we always talk about internally.

Yeah. Take a victory lap on the 996 thing. It is hilarious. I mean, you didn't obviously invent the 996.

Yeah, Sana invented the 996.

Yeah. Yeah, yeah, yeah. You stole it from them or something. But no, I mean, I think you were the first to really go out there and—

You were willing to say it when it was going to essentially get you a little canceled.

Yeah, exactly. Yeah, you get a little bit canceled. And I think it was an interesting framing for the brand of Traba to say, look, we're not trying to create superintelligence. We're not trying to cure cancer. We've just found a great business that delivers value to customers, delivers value to our clients.

Everyone—

all the parties are happy with this, what's going on here. And so we're going to work really, really hard. What was the actual journey of the get a little canceled? Did that ever really manifest? Were people ever actually mad at you? And then now, what are you seeing out there? Has everyone else adopted this? Are they taking it further? Can they take it further? it too far, just walk me through your thinking.

Yeah, I think it really has to go into what the goal of the company is. So we are a very ambitious company. We want to be a massive publicly traded company. And if you work backwards from any very successful company, the early team was just all in. It was their number one priority.

Always.

So we basically were like, we did this thing called anti-selling where we were like, look, we're going to get people working in person with us. We're going to go build some awesome stuff. And if you're not interested in making this your number one thing, then— And that's all good, but it created something really great internally because every single person is just like— it's like a team sport. Everyone's in it together, and we're all trying to create leverage on the time that we've done. So make a lot of money.

Yeah, I think there's something underrated about this where people have been mapping the missionary versus mercenary thing onto just how sci-fi the project is, and they assume Like, if you're working on something sci-fi, everyone's a missionary. And if you're working on something that's maybe just a little bit more tractable of a business problem, everyone's a mercenary. And I think that that's just not true at all. It's not true. Like, we see it here. Like, we're building, like, a media company. But I feel like everyone— we have a really great culture. Everyone's really bought in. Everyone is thinking creatively and working really hard and getting a ton of reward. And at the end of the time, At the end of the tunnel is not time travel machine that takes us to Mars or cures cancer. It's just a show, but we're all happy and aligned on that. And so, what has the road been? Where have the difficult moments been in the journey where you've had to kind of reignite the flame of missionary?

Yeah. So, yeah, shout out to your team. They're awesome. They greeted me here, and everyone's like, So locked in and love what you guys do. Yeah, I think a big part of that is just you guys being authentic to who you are. And totally look like we're TVPN. We're gonna make this.

It's a show company.

It's a show, and everyone gets gets involved in that. With us, you know, there's like challenges every single week. It's just part of building a company, and I think some people really they opt into that. They're like, "I'm gonna go learn a lot. I'm gonna work with a great group of people around me." The likelihood of success goes up if everyone. Everyone's equally committed and really bought in on making something awesome. And then people just have a great time. Like, they look back and they're like, you remember the war stories in the days where you're like in the Airbnb trying to get the customer? Like, this crazy request comes in, you're like, how are we going to do it? And you're really only going to create that if everyone that opts in and joins the company is like, that sounds exactly like what I want to do with my life.

Yeah.

So we get a lot of ex-athletes because they remember that in sports and college and things like that. So it's great.

Tell me the story of Uber, your experience there. And then I want the update on how you think about opening markets now, because this is something that not every software company has to do. Like, OpenAI launches a browser, they don't need to send a team to Dallas.

Right.

But at Uber, you did. And at Traba, you have been doing this. So walk me through the best practice for when you're going hand-to-hand combat In a software business, a tech company, but you still have to deploy a team.

Yeah, so operating in the physical world, it's quite different. At Uber, I was a launcher where you essentially drop into a city, you get a bunch of restaurants on the platform, on the Uber Eats product, and then you have to essentially get the couriers, and then you run—

You would skydive in, right?

You skydive in, sell the restaurants, say we can make you some more money. What's different with Traba and Uber Eats is that selection. Like, people aren't like, OK, I want to work at like 5 different warehouses this week. They actually are fine with working in this like middle ground where like, I'm gonna work for 3 to 6 months at this one warehouse.

Okay.

So, let's stay.

And then maybe I'll switch again.

Sure.

So, our sales cycle is a little bit different where we do have market-specific like leaders where people do own a market.

Yeah.

But we go for these big accounts that they want hundreds of workers versus, you know, like 3 to 4 people.

Yeah.

Yeah. And then the workers kind of like, it goes viral because they get their friends, and it's like a very social job and things like that.

Yeah, that makes sense. Has the playbook changed at all? Are you in every market now? Are you still on the warpath?

Yeah, we're in most markets in the country. We're not in California yet. So I was in Vegas before being here to meet some customers, then came over.

Yeah.

And that's not a capital constraint, right?

No, it's actually more just the employment laws in California right now.

The nature of your setup.

Yeah, we offer 2 different ways to work on Travo. One is 1099, in which workers can get paid within 30 minutes, and the other is W-2. But we're very intentional about every market we go into.

Yeah, you want to make sure you check all the boxes so you don't have some— oh, we forgot to file this form, and then it comes back. You don't want any of that. That makes sense.

So we're coming. We're coming soon.

Let's go.

There we go.

Let's go. Yeah, I'm thinking about all the different uses for a flash mob of human labor. that we could put to work.

How do you— any learnings around the causes of burnout that you see within your organization? If you're pushing people super hard, that means you want to avoid burnout. I don't get burned out when I'm working on something that I love and I feel like I'm making real progress on it. I feel like burnout can come from not Maybe finding meaning in the work, or not making progress, and running into a dead end over and over. But what's your framework? And I'm sure CEOs have asked you this, of like, OK, if you're pushing your team super hard, and you're pushing yourself hard, how do you avoid that?

Yeah, so the number one thing is people don't get burned out if they feel like they're winning. So we try to get people in environments where they're just winning. So you've got to get the right role, get them some wins under their belt. school, if you're like studying nonstop 7 days a week and you're getting straight A's, you're not really burning out from that.

Yeah.

So, you're putting them in an environment to actually be successful. And then everyone's got something that they do to kind of like take care of themselves. Like a lot of people, they work out. Like usually at Traba, people will work until like 7-ish, hit the gym, come back. Like it's a very healthy culture, but it's very much like happy, healthy, wealthy is kind of like what people are aiming for. So, they're all there together and like Yeah, that's great.

How do you think about the— how rigid is the 996 thing? Obviously, some people shift that forward, backwards. Now, there's this article in the Wall Street Journal about 002.

What is that?

So you work from midnight to midnight. You're working 24 hours. You take 2 hours off a day. I don't know.

Is this real?

I mean, I can pull up the article. because it would be good to get your reaction, but it says AI workers are putting in 100-hour work weeks to win the new tech's arms race. And there's like a 50% chance that this is like, you know, someone trying to get a PR piece to be like, well, we work even harder than this guy.

Yeah.

Josh Batson no longer has time for social media. The AI researcher's only comparable dopamine hit this year is on Anthropic Slack workplace messaging channels, where he explores chatter about colleagues' theories and experiments on large language models and architecture.

So, he's on a social work platform socializing, somewhat of a network.

Yeah, yeah, yeah, yeah. If you're trying to 996, don't count your Slack time. That's fun. Yeah, yeah. I think so.

So, the way it plays out actually is like when we first started the company, we were actually like, let's make this a little bit more structural. Now with 150 people, we have all these different types of roles. It's more just like you got to be locked in and make work your number Totally, totally. But what I will say is you go— we are an in-person company. If you swing by Traba at like 8:30 in the morning or 9:30 PM, literally any day of the week, there will be like people there building really cool stuff. And that's pretty awesome because then you can just like show up and build with your friends and people feed off of each other's energy.

Yeah, you can't be too dogmatic about it. I feel like, I mean, we both have kids. We're also on the West Coast. And so we run more of like a 6:00 AM start time. time. And so we just stack some extra hours there because then you have—

Yeah, that's cute that it really picks up around 8:30 at Traba.

Oh, yeah.

That's why I'm like, wait, oh my gosh, I thought we were intense. And then you guys talk about this 002, and I'm like calculating the hours. I'm like, man, we have the burnout question, and then we're like—

The 002, it's like—

Is that in here?

I think if you sleep for 2 hours—

Okay. So So working 002, the most intense periods for many come while working on models or new products when time working extends beyond the 996 schedule. That stands for 9 a.m. to 9 p.m., 6 days a week. One startup executive jokingly referred to the schedule as 002, meaning midnight to midnight with a 2-hour break on weekends.

Yeah, yeah.

We don't—

I don't think anyone can actually do that. You're just going to burn out, and you need sleep. You actually do need sleep.

Yeah, yeah. It's more about just like, look, are you all in on this? Especially in New York, there's infinite things you could be doing. You just want people to think twice before filling out their social calendars during the week. It's like, look, we're trying to build something epic here, and it's awesome to just know that you're going to be around your colleagues building something.

Also, I just feel like there's something about New York City, Manhattan, where if I was at the office late until 9:00, And I get off, like, there's still all the restaurants around. Yeah, like, like, stuff's gonna happen. Whereas in LA or in Pasadena, like, older with kids, like, I want to be home with the kids, 5, 6, like, you know, doing dinner.

A little bit different.

There's some things super comforting.

Yeah, that have kind of shifted, but they still probably give it their all.

Yeah, parents look like— if you got kids at home, like, we definitely have a group of parents at Travo and they make it work. Uh, shout out to them for sure. But there's definitely a huge cohort of, you know, people that don't have the kids, and they're out the office a little bit more than the parents.

Yeah.

There's something super comforting building a business if you're working as hard as you possibly can, knowing that lack of hard work is not going to be the reason that you don't achieve the level of success that you want. Because if you're— people out there, a bunch of people are not pushing themselves as hard as they know they could, and then it's always going to be in the back of their mind of the whole time building the company that maybe I worked a little bit harder, I would've achieved like, you know, whatever.

And then success in a startup, it isn't just this binary thing where you either get the gold medal or not. There's like different revenue targets you can hit. There's different valuations, different like ways to please customers and things like that. So yeah, it's great to like live a life where you look back and you're like, I gave it absolutely everything I got. There's nothing else I could have done better and like keep pushing forward.

So, well, that's a good—

That's a great place to end it.

Place to end the show. Yeah, gave it all we got.

We gave it all we got.

Thanks for hanging out with us.

Thank you so much for tuning in today. We will be back tomorrow at 11 a.m. sharp Pacific.

Can't wait.

Have a great evening.

Goodbye.

Love you. Bye.