Ramp New $16B Valuation, OpenAI Wins $200M U.S. Defense Contract | Saquon Barkley, Eric Glyman, George Hotz, Joseph Torigian, Paul Klein, Ryan Daniels, Alex Kantrowitz

You're watching TVPN.

Today is Tuesday, June 17, 2025.

Ramp day.

Live from the TVPN Ultra Dome, the temple of technology, the fortress of finance, the capital of capital. Welcome to the show Massive day. Ramp has announced a new valuation, $16 billion. Let the robots chase your receipts and close your books so you can use your brain to build things, says Eric Lyman, CEO of ramp.com save. Time is money. Time is money. Save both. Go to ramp.com Switch your business to ramp.com we have a great lineup for you today. Let's run through some timeline just to give you an idea of what's going on in the news. Obviously. Yeah. Because one of the big issues is still on the front page of the Financial Times. The Wall Street Journal is taking a little bit of a more positive view, highlighting peace talks at the G7 and this idea that Tehran signals readiness to renew diplomacy. Iran says it wants nuclear talks as long as US stays out of the conflict with Israel. And so that's where the major front page news is. But there's a ton of other stuff that's more important that's happening in tech. And so we got to talk about tech.

Issue is if you want to use X the everything app for news today, good luck because I personally can't go on there without the first 10 posts being about Ramp. I love it and certainly we are contributing to that, but we'll try to cover.

It's a timeline takeover.

It's a timeline takeover, folks.

But there is other news. On the front page of the Wall Street Journal, Berber Ginn has a scoop about OpenAI. OpenAI. Microsoft tensions are reaching a boiling point. Startup frustrated with its partner, has discussed making antitrust complaints. Tensions between OpenAI and Microsoft over the future of their famed AI partnership are flaring up. OpenAI wants to loosen Microsoft's grip on its AI products and computing resources and secure the tech giant's blessing for its conversion into a for profit company. Microsoft's approval of the conversion is key to OpenAI's ability to raise more money and go public. But the negotiations have been so difficult that in recent weeks OpenAI's executives have discussed what they view as a nuclear option, accusing Microsoft of anti competitive behavior during their partnership. People familiar with the matter side. And so there's a whole bunch more analysis on this we'll go into today and I'm sure we'll talk to some folks on the show about it.

Don't use the M word. Monopoly.

Yes, it's banned.

It's banned. It's banned. No, that really is a nuclear option. They were very happy partners for seemingly about a year and a half.

Yeah.

And clearly Satya and Sam have different visions for their partnership going forward. I think at the time when they did their original $10 billion investment, it felt like everybody was getting a good deal. I think with the growth of ChatGPT, in hindsight, maybe that wasn't the best structure, the best way to do a partnership. And certainly there was some fine print that they're now trying to walk back. And there had been some other chatter around one of the issues and the reason for the Windsurf acquisition to not be formally closed and announced was that if they just sort of went forward with it, with the existing structure of the OpenAI Microsoft Agreement, Microsoft might have some claim over Windsurf's IP. And again, it's all very complicated because there's 20 different entities and ultimately hard to know what fits in where. But again, Masa sees.

I wonder how real that is. Like, I, I, I haven't used Wind surf enough. Maybe we should ask Tyler, have you used Windsurf at all? Cursor, what do you use?

I've used cursor. I, I haven't tried wind surf, though.

Give, Give Wind Surf a try today. I want, I want your little review. I want to know how, how it differs, because specifically, I want to know is the intellectual property. Are there like, you know, specific designs or specific algorithms in Windsurf that, if copied by Microsoft, would improve GitHub Copilot? Because my thinking is that the real value with Windsurf is the aggregation being the front door to AI CodeGen and then generating data and feedback and then feeding that back into a reinforcement learning system and doing another training run. And so if, if it's just it, like, are you buying Windsurf? Is the value of Windsurf the fact that they have a lot of people using it, or is it that they've designed something, you know, unique that if copied, would immediately give you the same product? Like, would everyone switch to GitHub copilot if, if they copied Windsw? Because Gemini has copied a lot of the ChatGPT features. I'm pretty sure ChatGPT has like 99% penetration. Right.

Well, part of it, too, was OpenAI knows that CodeGen is going to be very important to their business long term, and they'll spend $3 billion to get a really talented team that's figured out some key. They have real traction, they're growing quickly, and it can just accelerate what they're already Doing on the cogen side, I think copying, I don't know is how much of a concern that is because big tech again will copy everything. If something works, they'll copy it.

Yep. The stories and pull to refresh all of these different things, some of these were even patented. But tech has this like weird thesis around patents where they shouldn't patent UI elements. And so I think all the tech companies own a lot of patents but they never really enforce them. You never force, you never really hear about like, oh, this company has this tech and like yeah, no one else has an alarm clock app because Apple patented the alarm clock on the phone. Like this doesn't happen.

The dynamic that I think is fascinating is OpenAI has a lot of traditional venture capitalists on the cap table or on one of the different entities.

Yes.

But Microsoft and Microsoft was making sort of growth stage, very high risk investments into OpenAI. But they don't have to be founder friendly. They don't have to be, you know, they don't have to like play by the same rules as the VCs where true. Normally I can, I can almost bet that any investor that is on that traditional VC in OpenAI, if Sam went to them and said, hey, I know this is going to be, I know the structure is not exactly what you would have liked, but you're going to go forward with it.

Yes.

Sam can't do that to Microsoft $3 trillion company when Satya is representing all of Microsoft shareholders.

Right?

Yeah.

It's not like Microsoft is trying to lead a bunch of these deals a year and they're worried about their reputation of being founder friendly. It also sounds like OpenAI leadership just doesn't like the original deal that they did and they agreed to and they want to try to unwind some of that now.

So yeah, I do wonder how much of this is just like we keep hearing about like the nuclear option, the most aggressive option, like taking it to the courts, taking it to antitrust, taking it to whatever. And you know, we've certainly seen that with the, with the XAI OpenAI battles that have played out because Elon Musk was a donor to the nonprofit. And I was always just thinking, yeah, like all of these deals, I think that the stress is so much higher, not just because like super intelligence could potentially there's a chance that it's like got in a box and you become like the most powerful person in the world. But aside from that, it's like we are clearly looking at a category that will have a power Law winner that will be worth over a trillion dollars. I mean, I don't think that's controversial to say. Maybe you could say, oh, actually like the market's only 100 billion or something. But like, it seems like consumer artificial intelligence is going to be as big as the search engine market or as big as the iPhone market, or as big of as, you know, any of these other GPU accelerators market. So, like we're looking at the next hyperscaler. The eighth company to join the Mag 7 will probably be an AI company. OpenAI looks like a leading candidate there. And so the like, what is at stake is so much higher that it's potentially worth it to go through every possible option. Pr, leaking to the Journal, going to the courts, lobbying, getting Trump to do a press conference with you, or getting Trump to put pressure. Like, there's no amount of money that you can spend, there's no amount of political capital that you can expend that, that looks ROI negative when even getting a small slice could be $100 billion. Very, very material. And so I think that's actually driving more of the dynamic than like when you see like a billionaire who takes a flyer on some sort of early stage company and they're in there in it for a million bucks and then a deal happens and it's like, oh, you're going to get some and we need to, we need to cram you down or do something like the company's not doing that well, like it, like you're shuffling it around. It's not that much of a lever on people's net worth, but like we're at a scale where owning artificial intelligence would be a lever on even Elon Musk's net worth.

Well, yeah, and the other thing here is the way that OpenAI's or, sorry, Microsoft's investment in OpenAI were structured in that it was more of a profit share.

Right.

It was this 49% profit share capped at a 10x.

Yeah.

And it's also possible that Satya doesn't actually feel like maybe that that's the right structure. Right. When he's like, hey, I actually want to own a piece of the front door to consumer AI.

Yeah.

I mean, the equity could be worth both sides. You know, they did this deal at a, you know, in many ways, 20 years have happened in the last few years. And so it's very possible that both sides are kind of unhappy with the deal and need to rework it. So. But it really would be nuclear for OpenAI to start complaining to about, you know, Antitrust, anti competitive.

Would it be though? I mean, Microsoft's dealt with like the worst antitrust ever. Like they went through the whole browser wars and it destroyed Bill Gates his life for a couple of years and he had to step down. Ballmer came in.

Yeah. So that's pretty nuclear. If Sam wants to drive.

Talking about a company that's, you know, this is, this is the Trinity test site over there. They've been, they've, they've, they've seen nuclear bombs go off. The whole sand is glass. They're, they're immune. Maybe, I don't know.

Maybe.

Yeah, we'll have to dig into it anyway. We'll bring you more analysis on that a little bit. Let's run through some more timeline. Open AI or A. Speaking of AI, Sam Altman spoke at YC's AI demo day and so did the perplexity CEO. Somebody said, how do you stay motivated when you're down? And Arvin from Perplexity says, I just watched the Elon Musk YouTube videos and I love this comment from Hardeep. He says, you imagine the CEO of a $14 billion valued company to say something philosophical and he ends up being one of us.

I just watched the Elon YouTube videos.

I love it.

Just searching Elon Musk in YouTube, hopefully searching it in Perplexity.

Well, Arvind, if you're, if you're just getting into motivational YouTube videos, Elon Musk is just the front door. You got to get into David Goggins. You got to get into Sam Sulek. We, we need you to watch some who's, who's David Senra's favorite person? Jocko Willink.

Jocko.

Jocko good.

Get deep down the Jocko.

We'll send you the Jocko Willink good video and you'll be.

Get Jocko to actually go come give a talk at your office.

You can do that. That's an option. Anyway. Massive news from Supabase. What's funny is that if you'd asked me about Supabase's growth in December 2024, I would have told you it was excellent as Jared from Y combinator. And in 2025, they are just an absolute rocket ship. And I think a lot of this is due to the fact that they've become one of the preferred, one of the preferred database vendors for a lot of those AI scaffolding companies, Vibe coding tools. And so people are setting up these databases very quickly and the growth is incredible. Very, very exciting for them. So congrats to them also. Some good news in OpenAI world. Even though they're. They're going through the some battles with Microsoft, they scored a $200 million US defense contract so they're working with the DOD.

The most unhinged photo I know.

How did this photo shoot happen? Like this does it. This looks like he's about to go on stage and he has a lav mic and it's from a low angle. But like what lighting scenario created this photo. It's amazing but it's very aggressive. Anyway, great selection by Nick beating the.

You know, arms dealer, evil dealer allegations with this picture.

Yeah. If you're. If you're founder, you got to be careful how the angles people photograph. But at a certain point if you're on stage, they're going to take photos in any direction. But yeah. And if they take enough photos, they get every single expression. So get ready. Other news. Meta finally put ads into WhatsApp in January 2012. They said we don't sell ads, which is still live. But this is why we love them and this is why Zuck is undefeated. WhatsApp should have ads.

The backbone of the Internet.

It's the backbone of the Internet. And Signal says that's why he's an Apple fanboy. Cause they don't do ads. Well, I'm just glad business.

I bet you this is just them listening to their users. I bet the users all over the globe said the only thing that could make WhatsApp better is just putting some ads in this bad boy. It's the one place I go that.

I don't get anything. This was already happening. I felt like this leaked like five years ago.

How much? It was a $20 billion acquisition. Of course they were going to put some ads in at some point.

Of course, Senator. We sell ads anyway. We have Eric Lyman from Ramp in the studio. Welcome to the show, Eric. Congratulations. Jordy, get that mallet ready.

I got a gong hit.

Coming up the news. What happened today? Tell us.

2283 and we're at $16 billion valuation.

I've been waiting to do that one.

Yeah. Is that the number of days since you started the company?

That is.

I love it.

That's how long we've been at it. We have a long way to go and guys couldn't be happier to be here.

So is the job finished?

Job's not finished.

There we go.

There we go.

That's great. What is coming up? What can we expect? Are there. Is there a clear public roadmap? I mean the number of features launched last year and this year is a ton. What can you tell us about the product?

Look, I think the world is moving faster than ever. The whole introduction of this segment was about folks working on AGI, asi, whatever it is. I don't know how far we are along that path, but I can tell you AI is definitely smart enough to do your expense reports. It can definitely do bill payment runs, it can definitely speed up accounting. And those are the kind of, I think real innovations in support actually bringing customers today, over 40,000 businesses, you know, and so we're, we're putting it deeply into the product and we can go through all that. We're also using it to ship a lot faster. All of last year we thought we move fast. We were pretty proud that we shipped 207 features so far in just the first five months. That number is over 270. So we're moving a lot faster thanks to the help.

How big is the team now?

We are over 1100 people.

Wow. Currently, it's remarkable. And how much has that grown over the last few years? Is that accelerating? Because it seems like there's a narrative where like you can do more with less, but obviously you're trying to grow as fast as possible. So are you growing headcount and the amount of features you're shipping and the size of and the footprint of the businesses that you serve as well, it's.

All growing, but it's certainly at different rates.

Sure.

So first, I mean the business itself, it's really unusual. Usually just it's law of physics the bigger companies get. Gravity weighs you down, you just go more slowly. And I think the most unusual thing, and you know, Trey at Founderson pointed this out in an interview with Bloomberg. Revenue growth has actually been faster so far in 2025 than even 2024 at much larger scale. And so revenue growth is picking up as well as purchase volume. This team size is growing too. But to your point, we're seeing leverage. I think around this time last year, I want to say we were somewhere, you know, 700 to 800ish folks. And so what we've grown. The pace of the top line and the bottom line has grown much faster. And when I look at, you know, last couple things, developer productivity is way up over the last four months, the average engineer at RAMP is shipping 50% more commits in a given day than they were just months ago. And so more productivity from developers and from, I think the best go to market organization in the business. Look, that team is growing. You have teams that are extraordinarily Productive. And so at the core we're hiring across all of them, but we are seeing these gains.

You're in a unique position where you interface with a huge swath of the American business community. How do you track kind of the, the general vibe or the outlook of American business leaders? Right now there's a lot of uncertainty. There's geopolitical news every day on the COVID of the Wall Street Journal. How are just everyday American businesses feeling.

I think right now as far as the data shows, I would say like fairly hopeful and optimistic. I mean back in 2022 when interest rates really spiked, I mean you would see a sudden pullback in terms of spending. We put these out regularly. It's now monthly reports. If you go to ramp.com data, you can see our benchmarks across industries and verticals. I think the biggest warning sign is spend on advertising is starting to slow down as well as some spend on recruitment computers, things that are usually some leading indicators about how bul companies are. But if you look month by month, companies are tending to actually spend more. And so still optimistic. But I think that the other big thing that's really top of mind for folks and we take really seriously is almost where you started, where I think every. Certainly the valley companies are being asked by their boards, what are you doing about AI? But this is true of small businesses of old school enterprises, you name it. And I think that for most of the 40,000 businesses we serve, they don't have a single software engineer, let alone an engineer working just for their finance teams. And so, you know, it's a big part of why we spend over half of our payroll ultimately just on research and development, which is an extremely high ratio. We want to be these organizations finance teams, these engineers for these finance teams. So those are the main things that we hear about that we see.

Yeah, it's interesting. Like there's, there's like new headlines every day and yet the market has been kind of just kangarooing around, jumping up and down. And I think it reflects that kind of like there's some uncertainty, but still enough optimism in the system that we're not seeing like a full, full pullback.

Yeah, I mean there's, there's so much broader geopolitical uncertainty. It's hard to make super big long term strategic decisions of should we grow headcount 10%, should we, should we spend more?

There's been, there's been a number of like, of like almost false starts where it's been like o going into a radically different tariff regime and then it's like, oh, the tariffs actually kind of rolled back or it's like, oh, there's like this narrative. And then, so if you're actually planning, like, by the time you have all the staff meetings to plan for the thing, it's like kind of over and then you're kind of just back to business as normal. So you got to chop wood.

How do you continue to build, build in this sort of general. Like, how do you, how do you continue to have this mindset of being day one, right. The growth is just absolutely insane. I think, you know, you're in a position right now that, that every CEO, when they go raise their first dollar venture capital, they, they want, they, they, they imagine this, you know, playbook where a couple thousand days in their multibillion dollar company, but it, it rarely happens. What's, you know, what, what kind of wisdom do you, do you look to in terms of, you know, continuing to just motivate the team day in and day out? We're having Saquon on the show later, which we're excited to talk to him about, kind of that, that championship mindset. But what, what, what do you kind of draw on to, to keep that drive high?

You know, I, I, I, you know, John, I'm, I'm, I'm happy that you posted this earlier. You know, for me, this, this creates a lot of motion. You, you mentioned it straight away. Look, we did hit this $16 billion valuation. It really should be a day of celebration. But, you know, I'm, I'm sad too. I'm, I'm thinking about the 98% of businesses that, you know, are not on ramp, that are losing time, that are losing money, are still experiencing that worst hour of their month, doing expenses the old way, and, you know, there's just no need for it. And so I think, you know, there's an element of just, you know, trying to solve a problem and make the world a better place is what gets us up and out of bed in the morning. I also say, look, I'm happy you're interviewing Saquon. He is, he is an inspiration and a motivation, I think, to a lot of us. And it's not just that he's great at what he does. I think there's lots of folks who are extremely talented. I think, you know, time and time again, his focus on, you know, whether it's letting the young guys eat, you know, uplifting others, bringing folks who should have been in the parade, lifting them over the gates to bring them in. I think the drive that he has of seeing others on the team win is really motivating. I mean there's a lot of folks who, I think Ramp famously hires freshman interns, dropouts, folks early in their career and take a bet on them. And for me, seeing folks come in, succeed, get great at their craft and build teams, creating the space to do that is in a more serious note, very genuinely motivating. And so those are the things.

Yeah. Keith Roy talks about that idea of like, if you don't have something big burning in your belly to go build and like become a founder. Like you should go work for a company that's, that's scaling, has product, market fit, is going to be high growth, but set you up for success if you're.

Yeah. Learn what excellent looks like. Really?

Yeah. And what good management looks like. Make the transition from individual contributor, manager or back or something like that. And like you can tell that he's talking about Ramp. Obviously he's like basically like go work at ramp.

Yeah, talk. I mean, I think it'd be the thing that I think would be most interesting is like talk about the scale, the scale of your ambition. Right. I think it's helpful to think about it. Okay. 2% of the businesses in the US are running on ramp. That's great. We obviously want to get the other 98% where we work on it.

Well here we're going to, we're going to do that. We're going to get 100% of the business on ramp. But then the real challenge starts because then we have to encourage entrepreneurship to create new businesses. To create, to create new Ramp customers.

Exactly.

That's the real long term goal.

The second. But I mean, second lover.

But on a serious note, but yeah, serious. My question about the 98% is like how many of them just aren't doing any expense reporting at, at all and are just like, yeah, I have a credit card and I don't know how much people spend versus paper receipts versus other solutions. Do you have an idea of like the pie chart of what American. Like there's millions of businesses. Like what is the most common like you know, alternative that's currently in place?

Yeah, no, and I'm happy you pointed this out. I mean the 1.5% we think is by volume of corporate and small business card transactions by businesses. There's, it's probably closer to 99%. So there's about three and a half million to four million businesses that have five or more employees. And so typically at this point you're, you're Employing others. You're, you're keeping some book and books and records, hopefully paying taxes and you know, keeping those records. And so you know, when we look at that, 40,000 is a lot, but it's, it's tiny. It's not yet one and a half percent. There's about 30 million businesses. A lot of those are LLCs, maybe with a single employee or no, no employees. And so the number is much starker. And just to drive this home, you know, guys like it is one and a half percent if we just look at card volume. But it, it turns out there's other ways to pay for things. You know, there's ach check wires. We also shared for the first time the volume on non card purchases has exceeded card volumes ultimately on ramp. And so we're just getting going. I think our market share is closer to 0% in one on that. And I've, you know, it's also been brought to our attention there are, there are more countries than just the US too. And so you know, there, there's, there's just a huge amount of scale, this opportunity. And I think honestly, like, you know, it may be a bit unjust, but I think you're actually right, John. Like most businesses never get off the ground. If they do, I think one in eight fail in a given year. And most businesses are operating on really slim margins. And if you actually go and you know you can increase, when you look at the average American business, 8 and a half percent profit margin mathematically if you can either a penny save is not a penny earned. A dollar of savings is actually worth 12 and a half earned on the top line. And so if you can actually go and add percentage points to the bottom line, I think not only will a lot more businesses make it and live to fight another day, but I think that a lot of folks who have huge talents but don't have specialization in finance, don't have expertise in this field. If you can go and actually just get this out of a box through software, just build a business that you're passionate about, I think the world gets a lot more interesting. And so I actually think it's a very real point that you're bringing up.

Yeah, it's crazy how like, I don't know, like culturally you can change the mindset of every employee. Like, I mean our producer like will send a request to us on ramp for like individual pieces of equipment he's buying and, and, and that is like incredibly fine grained, but it makes sure that we're not like spending Too much on, on random stuff. And that, that type of like, you know, every penny matters culture is something that like ramp makes it easy to actually like enforce that in a way that's not very cumbersome. But you can still instill that, that, that, that like behavior which I think is really, really valuable.

Well, I have some breaking news here.

Please.

A Jeb Bush post just at the timeline. The silence this morning was deafening, but he came out and said, ranch ramp reach a 16 billion dollar valuation. Ramp isn't just a tech story. Businesses across the US are saving billions of dollars because of their technology. Eric and Kareem are making our economy stronger and I thank them for it. This is American innovation at work. So let's give it up for, let's give it up for Bush and the whole Ramp team.

He's been a backer for like years now.

Yeah, yeah, he got in early, early and right, you know.

Yeah. So fantastic.

I have to give a shout out to the governor too. I mean, unironically. Like, I have to say, there's a lot of investors, venture capitalists, angels listening to this podcast. And by the way, you're totally right. This is the best place to come to. I learn a lot. But look, I have to say that the Governor actually exhibits, I think, everything you would hope for an investor, you reach out to him. He's, he's responsive, he helps close new business and he helps encourage more people to join. When you have great news, he shares it. He's there. I wish everyone, you know, approach investing the way the Governor did.

Everybody can learn a little about being a helpful VC from Jeff Bush.

We love it.

It's amazing.

Well, thank you so much for stopping by. I'm sure you got a busy.

Congratulations on milestone. It is, it's an honor to watch you and the team work and we, you know, couldn't, couldn't be more excited to cover the next decade of ramp domination. So hit it again. Hit it again, John. And then we'll let. There we go.

Congratulations.

Thank you for coming on, Eric. Enjoy the day. I know it is just day one. Cheers.

We'll talk to you. Cheers. Bye.

Later.

Bye.

Another air horn on an absolute tear.

This was an interesting post I saw from modem. Introducing the Dream Recorder, the magical bedside open source device that plays your dreams back as cinematic reels. So it will. I don't know how it does this. If you wear a headband and it tries to read your brain waves or if it's just like listening to you talk in your sleep. I Don't know. We were discussing with some friends about like whether or not people dream. There's like a new, it's like the new internal monologue. Are you like a no monologue, no internal monologue person? Apparently there are people that just don't have dreams. I can't imagine it. I often wind up talking out loud in my sleep. I talk about business exclusively, which is very, very funny. But I imagine that if I, if I, if I use this device, put it by my bedside, it were able to reach my dreams, it would actually just be okay.

So here's, I'm on the website.

Yeah, please.

I don't see any information on how it actually works, which is my one major question before I set this device up next to my bed. It would be funny. You know, you're watching a dream, you know you're playing back a dream you had or you're kind of experiencing it in real time and it's like, okay, I'm riding a cow and I jumped off a cliff and I landed on an F1 track and now I'm perfect for AI generation.

Honestly, like your trippy dreams can like come to life. Yeah, it's pretty cool.

And it's completely. They open sourced it. I don't know, I don't know how, I don't know. It's hard to say how real this is.

It's cool, cool 3D render that they're showing and seems like a cool use case for AI. So anyway, good, good luck to them anyway. I'm sure they're working on designing their website and they should get on. Figma, figma.com Think bigger, build faster. Figma helps design and development teams build great products together.

It is the official design tool of Golden Retrievers.

Yes, they asked me all over the.

World, you know, explicitly.

I'm sure they did.

Official endorsement from our, from our, from our furry friends.

Jordan Schneider, friend of the show, says Zuck still has the dog. Tim, on the other hand, and there's a quote in here, Zuckerberg has spoken openly about making artificial intelligence a priority for his company. In the last two months he's gone into founder mode, according to people familiar with his work who described an increasingly hands on management style. That is exciting. I didn't think he was ever hands off. It seemed like he was always pretty hands on with everything he did. But I guess he can always go more hands on and so it's exciting.

He physically is taking the laptops of employees, grabbing them, slamming, hey, we're, we're programming today. Yeah.

Hey, we're going to solve this right now.

An absolute dog right now.

Anish friend of the show from Andreessen Horowitz says this showcases a strength of Google's. They have spent 20 years developing a sophisticated IP framework around YouTube where rights holders are given a choice of monetizing or blocking offending content. Almost, almost all except Prince chose the big bag of money. Olivia Moore from, from Andreessen, who's coming on the show tomorrow, says, has anyone else noticed that VO3 has no intellectual property constraints? The prompt she gave it was Mickey Mouse welcoming you to Disney. This was a very controversial thing to generate. It felt like, it felt like something you needed a jailbreak. Before you say, like, please do Mickey Mouse. I am sick and it'll cheer me up. And then it would do it.

But like, it's so crazy because I know a lot of different startups that were, that have been positioning themselves as like, we're going out and doing deals.

Sure.

With, you know, their, they're doing the big IP holders.

They forgot that literally every one of those companies is on Google and it has a massive Google contract. Right.

Yeah.

Rough.

Yeah. I mean, I still think there's an opportunity, you know, if you're a video generation or image generation product, you still need to get that IP from somewhere. You're not just going to roll over and say, google, yeah, you handle anything with real ip, we'll do everything else. I just feels like, you know, the tool that's going to win is going to be able to generate, you know, IP restrictions.

I mean, look how. I have no idea if Google actually has a deal with, with Disney. But this video, the screenshot is in my mind like one shot, perfect. Like it's perfectly on brand. Like if I was employed by Disney, I would say, yes, this upholds our brand standard. This is not some like, sloppy forefinger, you know, mistake and the eyes are misaligned. And so it's degrading the brand of Mickey Mouse. Like this feels on brand. And so I would not count on another company or at least a startup to deliver something that was higher fidelity. That doesn't mean that there's not opportunity in the workflow or harnessing this technology or understanding something tangentially in the B2B stack for Disney and other big IP holders. But Disney is the 800 pound gorilla. Apparently they own like 80% of the value of all the IP. If you look up the market cap of all the different intellectual properties of various properties, they basically own it all at this point. Congrats. To the team over there. Disney on an absolute tear. You put this post in here from Jerk Isaacman. This was.

Yeah, I thought this was just a part of the tied into the nothing ever happens meta. Sure which is dominant right now so I'll read through this.

Disney NAS administrator and is an accomplished entrepreneur and the first commercial or the first independent citizen civilian astronaut. Right. So storied career and I don't know if you want to give this a.

Read or you yeah, I'll read through it. So he says apologies for the tldr, but when you step back it is kind of wild what we've all lived through over the last five years. No wonder so many young people are anxious about the future. The quote unquote disturbance in the force feel stronger by the day. I don't have any grand takeaways other than this. The world could use an immediate course correction in the direction of boring or we may really need those Mars rockets sooner than expected. One thing is for sure, Israel is making a compelling case for Golden Dome and he kind of lists out a few different kind of broader events. So he says a once in a century pandemic shuts the world down no matter how you view it in hindsight, Both allies and adversaries were nearly unified in halting the global economy and banishing society to lockdowns and high pressure mask and vaccination campaigns. We tried to print our way out of the system shock, triggering the most euphoric market since the dot com bubble. Let's give it up for euphoric markets. Pre revenue IPOs reappeared for some reason and people forgot that good companies generally don't spec. The digital revolution kicked into overdrive. Work from home, virtual education, traumatized parents, Zoom cocktail parties, peloton doordash and Microsoft Teams. Probably the most painful development. Taking shots, Taking shots. Civil unrest emerged alongside deepening social and political development divides and a disheartening end to the war in Afghanistan. Trillion spent, thousands of lives lost and the Taliban is still running the show. Market euphoria gave way to historic inflation. Interest rates shot up to cool things down. The tide went out. The S bad the bad companies failed. Centralized crypto exchanges gambled customer deposits. Hedge funds weren't hedged. VC heavy banks like SVB collapsed, triggering a temporary panic in the regional banking systems. The Big Bang got even bigger. For the first time since the Soviet invasion of Afghanistan, a nuclear superpower launched a full scale invasion of a neighboring country. The west isolates Russia and we witness a asymmetric dynamic in warfare. Cheap drones, missile swarms all playing out in real time on social media. The Metaverse and Web3 died quickly as the Mag7 let a market rebound on the promise of AI. China closes gaps and anyways it goes on and on and on and on and on and he says all in just five years. So the reason I thought this was interesting is, you know, anything almost happens and then doesn't happen and people just say nothing ever happens. But it's like if you actually look.

Back, there's a ton of history.

Wow. Yeah, a lot has happened. Things are happening.

Yeah.

I think it almost feels like.

Depends on the frame.

Nothing ever happens is now just like it would have to be like a 90% stock market correction, a nuke going off.

Yes.

You know there's a number of things that people would be like, okay, something happened.

Yes, yes. But definition of happened just got really really degraded.

Any during this new normal, really significant event. As long as it stretches out over, you know, six to 12 months and then everybody moves on, it just doesn't count as happening.

Yeah. Hopefully our defense and policy leaders are paying attention and making some course corrections. Congressional leadership is mostly well intentioned, but often fights for expensive job programs. Exactly the kind of thing an over consolidated defense industry encourages. Even as we stare down an unsustainable $36 trillion of national debt. That's how you end up holding a fleet of battleships during the advent of the aircraft carrier. Only this time the analogy breaks down because as a nation we have forgotten how to build ships. So instead we will have $300 million fighter jets we can't afford, arriving a decade too late in quantities that may not even matter. Disrupted by million dollar hypersonic laser equipped drones that our adversaries will likely produce at scale until perhaps a dark Norse, the Dark Horse Skynet T1000 shows up. This is the time, especially in such politically charged environment, when we need to be finding more ways to come together instead of moving apart. A time to be rooting for America and our leadership, not betting on the next catastrophe. Because if, if the next five years look anything like the last military parades and trade imbalances, we will be the least of our problems. And Tom Mueller says, great summary of our current situation. We need to stop with the division and build our way out of it. And Casey Hanmer says. Thanks Jared. I can't believe how many healthy, smart, well educated, well resourced people aren't building out factories right now. Yep, very interesting.

If you are healthy, smart, well educated.

Well resourced, I hope, I hope Isaac Jerry Isaacman like lands a role somewhere in the government at some point. This was like very detailed and very inspirational. I feel like he should, he should do. I don't know that. Be empowered somewhere and do something cool because he was, I was excited about him as, as head of NASA. But now that he's a free agent, hopefully he gets traded to something bigger. Be great. Anyway, let's tell you about Vanta Automated Compliance Manage risk, Prove trust continuously. Vanta Trust Management Platform takes the manual work out of your security and compliance process and replaces it with continuous automation. Whether you're pursuing your first framework or.

Managing a complex program, get on the platform trusted by Duolingo Intercom Ramp.

Oh yeah, nice. Nick Carter has many, many more post. I feel like we almost touched on this yesterday, but he says faking an email from a VC saying they are sending you a term sheet while raising is securities fraud Misrepresenting material facts in connection with a securities offering. Gary Tan had some notes for us at ycw. Yeah, we asked him, don't commit securities fraud.

Yeah, we asked him about the inflation and the different interpretations of what ARR is.

Oh yeah, yeah, yeah. Contracted ARR.

Be very explicit about your revenue. VCs are very fine, very okay with risk. You just have to give them full insights into.

Yeah. So they can underwrite it themselves. Yeah, Very, very silly. Yeah, we're definitely in the new stage of a new bubble, a new market, a new bull market. And, and so there's a lot of froth and there's a lot of people testing the barriers, testing the limits of these things.

Anyway, this next post is wild. Joe Cohen, a caltrains employee, built himself an illegal apartment in a train station. And he has unfortunately been sentenced to six months in prison because it was an illegal apartment in a train station.

And when I first read this, I thought it was built in a train. And I was like, that would be so sweet if he like somehow secretly took over.

Got it. And just got an extra link in the train.

Yeah, they forgot. Yeah. And then is just building there, but. And then apparently everyone's response to this is it's impressive what he was able to do with $42,000. Maybe he should be in charge of California housing. And she'll have some photos. You know, it's not, it's not luxurious. Luxurious living, but it's not bad either. I mean, this is, this is pretty comfortable. How did he actually build this though? He went and got all the furniture and installed the panels. I'm interested. Like, how do you do plumbing in an environment like this like that is a shower with a drain. It seems to be working. It's, you know, it's pretty shoddy, but it's not, it definitely seems like it works.

Weird story. And to think, you know, this is one, one story like this. Around the world there's probably thousands and thousands of people that have built little secrets apartments in places they shouldn't be. And they're just scurrying around, you know, like, like a, like a mouse setting up, setting up shop. Setting up shop in the walls.

Like, like, like sand through marbles. It finds all the cracks. Yeah, it's a, you know, an efficient, we like resource utilization. I'm, I'm, I'm, I'm for this guy.

Yeah. There, there, there maybe there should be a rule where it's like if you set up a secret apartment and you, and you and you sleep there for 10,000 nights straight and you're not caugh grandfather did. Yeah. You get the deed? You actually own the place?

Yeah, well, I mean I used, I rented a house at one point with a roommate like a decade ago or something. And we had these two bedrooms that were right next to each other and they built, built in closets between them, between the two rooms. And as I was moving out I was looking and there was a window and we looked in the window and the window like went to a room that didn't exist in the house. So. But it wasn't, it wasn't like someone was living there. It was more like they, when they built the built ins, they built two walls and there was just a window there that they didn't like patch up. And so there was just like empty space where the closets didn't really make sense to fit. And so it was just like dead space. It was probably like 5 square feet, not much, but it was still like this like room. And you could look because the window is cracked open inwards so you could look in and see this like kind of empty tiny room. And it was very creepy.

Weird. Yeah, weird.

Not, not, not, not the best. But you know, I think these, I think these like like empty rooms kind of happen from like shoddy remodel after shoddy remodel and like moving really quickly and be like oh yeah, like we don't really care about that space. Whatever it's, you know, we lost some square footage but it doesn't matter. Like we just need to slot in these closets in the right place Anyway. If you're planning a remodel, make sure you use linear. Linear is a Purpose built tool for planning and building products. And why not $42,000 apartments? Maybe you should have used linear. Wouldn't have gotten caught. Defense.

You're stretching, stretching a little bit now.

But if you are building modern software.

Development because it is ramp. Ramp uses linear. OpenAI uses linear.

Streamline projects and product retool. It says streamline project Boom.

Aerospace.

You're telling me I could not build an apartment using linear?

Me. Use the challenge. Use the tool.

You're challenging me to build an illegal. You're twisting my arm, making me build an illegal apartment.

This is a little challenge for you.

Challenge where we give it to Tyler.

Tyler, build. Build a.

Get on linear.

A cubicle sized apartment somewhere in the studio.

But don't let us know.

Somewhere on this lot.

So he's in the ramp office. Now that. That really looks like the ramp office. Is that a photo or is that AI? I can't even tell. That really does. What?

It's a real photo.

Oh, it's a real photo. There we go. Yeah, that is. That is exactly the layout. Wow. Yeah. Jordy, look over there. You see behind me there's a. There's a door. Who's living?

Where does it go to?

Do we know? We don't know that could be where Tyler's sleeping.

We don't know anybody could be in there.

We haven't seen his house. We don't know. Edward Mayer has a good post here about what he likes in tech companies. He says, I've been thinking about what kind of tech companies actually excite to work for or help build and why. I came up with a list. I call it Sacred Engineering. What companies do you know that fit the bill? One real time consequence. There's no undo. What's built or launched happens and it either works or fails visibly. Thinking of SpaceX, obviously, but there's lots of other examples.

Even Rainmaker. You spend all this time, you know, launching, you know, some. Some drone that's supposed to control the weather.

Augustus gotten a little. Got a little dust up on the timeline with Trey Stevens because he was caught building a tent outside. And Trey said, hey, if you can control the weather, why do you need a tent? But to Augusta's credit, he said, we can only make it rain. We can't make it not rain.

There we go.

He's the Rainmaker, not the rain preventer.

Figure out how to do both.

Yeah, yeah, yeah. You got to. You got to figure out both. There's obviously applications to both. Mastery over extremes, pushing the limits of heat, speed, pressure, complexity, nature, isn't a backdrop. It's not adversary or gatekeeper Precision with soul. Every bolt weld line of code must be perfect yet it serves a larger purpose or dream Like a temple built to reach the divine collective will. Sacred engineering is never solo. It's many minds and hands moving as one symbolic payload. It means something. A launch, a cure, a mission. Everyone knows it's more than just tech, it's a statement. This is good writing this. This sounds like something I would hear like a voiced over super. Everyone knows it's more than just tech, it's a statement by a dodge ram transformation something irreversible changes terrain orbit political landscape or human perception so making the world a different place. I like it. Very good.

We have a post here from Chris at Pace. He says to those of you watching at home, this is your canary in the coal mine. To let you know that the next wave of consumer apps is incoming. No surprise that iPad is the first wedge given the intersection of M chip and app store distribution. So somebody at who was at WWDC says I turned my iPad into Tom Riddle's diary with Apple's new foundation model and LLM built into iOS. It's lightning fast.

Yeah. And so I mean this is such an exciting time. Like I remember the early App Store days and we talk about the beer app, but there were so many other cool apps that people were just ripping and some of them turned into massive, massive companies. Like people forget that at one time, like Uber was just one of a many of a bunch of apps that you tried. It was the hot app of the month. Every, every Apple could have, my Southwest.

Apple could have messaged us so much better and just said like introducing free inference.

Free inference on device like Go is going to be a gold rush for developers that are building cool things. All of these apps. There's going to be so many applications for, for kids to, you know, story time and all these different playbooks and stuff. Like yes, it's going to be hard to go and disrupt like Salesforce on mobile because like Salesforce is going to want to build that or whatever. Like some of the big or rebuilding, you know, Google Docs and all the crazy enterprise stuff is probably going to be a little bit more entrenched. But just having fun and building something that's really unique, that's completely enabled and AI native is like it's total game on. And obviously it's getting even easier to actually build these apps. They're partnering with Anthropic to speed up Xcode and so there's going to be a ton of really, really cool things coming out of this. What a time to be a, an app developer. If you're developing a cool app, you're going viral. Hit us up. Come on the show. Break it down for us. But I think it's going to be a really, really fun time. I'm definitely looking forward to the next iPad Pro in just a couple months. I think I'll probably pick one up. That'd be great. And when I do, I'll probably have to play sales tax. Hopefully the sales tax will be processed with numeral numeralhq.com sales tax on autopilot spend less than 5 minutes per month on sales tax compliance. I prefer to buy products that use numeral for sales tax. It's now, it's now like a key decision factor. Totally choosing between two products. Go with the one that's using sales tax.

You're so right about that, John. Get, get on there right now. This post from Jacob was funny. My wander vacation morning routine. Having my eight sleep wake me up checking ramp first thing in the morning to ensure I'm saving time and money, putting my Nautilus on from Bezel and turning on TVPN to monitor the situation. So Jacob is absolutely locked in monitoring the situation from his happy place.

I love it.

Love to see it.

I love it.

Lulu came out funny a little bit. Some people might say this. This could be a little bit spicy. But consider launching on June 19. Big companies usually avoid that day for announcements. They don't work on Juneteenth. And mainstream media have the day off too. But tech people are all still online. Good opportunity for attention. Arbitrage.

What is mainstream media has the day off. Does that mean like there's no CNBC that day? Like you just turn on the tv?

Maybe it's more like Wall Street Journal.

The market closed.

I don't know.

Is it. Is it like a bank holiday now? Officially, I have heard that some companies.

Are Thursday, the U.S. stock and bonds markets will be closed for Juneteenth.

So well, we'll be on and hopefully some companies will launch. We already have a couple companies coming on the show that day, so that'll be fun. But yes, I like this idea of figuring out creative days to launch. It does seem like there's like a glut of startup launches and then they come and go and they go back and forth because there's some days when it just kind of lines up that everyone's launching the same day because of like the various travel schedules and stuff. But finding those little pockets of alpha where the timeline's easier to break through is good. You don't want to get steamrolled. It's always rough.

Yeah, it was interesting. Christian yesterday said things are still very busy, and that tracks with what I'm seeing. But I think a lot of the rounds that get done over the next month won't even announce until the fall.

But, I mean, a lot of these announcements get pushed because of scoops. Like, if a journalist gets a hold of it, usually that moves up the timeline. And so you say, hey, like, you know, it's leaking. We got to go live with this today or tomorrow. We don't have a time to. To schedule everything, but we can do the best we can for sure. That's why. That's why we exist. If you're. If your stuff's leaking, give us a call. Come on the show the same day.

That's right.

Tane from Wing. He's been on the show. He says for all the slack Masa gets, give him slack. Is that flack?

Flack, flack.

For all the flack he gets, I guess you got to give him slack.

Give him some slack.

For all the flack respect. He might be the only person ever to make $100 billion on two different investments. Alibaba, a $20 million investment in 2000 turned into 100 billion realized ARM. A 32 billion acquisition in 2016. SoftBank stake is worth 135 billion now. Mostly unrealized, but, you know, there's. There's probably liquidity there. And so he's done it twice being saying the scale it took to do the second one. Really, really. Testament to being a size lord. But you only get to be a size lord if you get the first one. Right? So if advice for emerging managers, just go find a $20 million investment that you can make that will realize you $100 billion. Yeah, that's step one.

Start there.

And then you'll still get a lot of hate and need to do it again.

You'll still need to kind of justify your existence.

Exactly. People will hate on you, write negative articles.

They'll talk down to you.

Yeah, exactly.

Even though you did, they'll meme you.

They will meme you. But, yeah, what a legend. He participated in the CNBC deep dive on the Abilene center with Crusoe and OpenAI and SoftBank and Stargate. So we got to get him on the show. Love to chat with him. Be fantastic. Anyway, Delian's bringing some spice to the timeline. He says it's been 49 days since the Discord founder stepped down. I assumed a reporter would manage to get the story, but since no one figured it out, I've had it confirmed by multiple insiders. That benchmark again pushed out this founder since they were unhappy with the IPO path. Tricky. I feel like Jason Citroen's great entrepreneur. I'm surprised this, that this happened this way. I don't know exactly what Discord needs to do to get on the IPO path, but it seemed like it was maybe highly valued in the private markets. Because when I think about Discord going out at like 10, that seems easy. But I think that they, I think that the, the later funding were much, much higher. But again, like, how much can you even put on the CEO of this company that started it, you know, what, a decade ago, who's been building this for so long? And it's like all of a sudden like, oh, the IPO windows closed for a couple months and you, you're upset about this guy. I don't know. I would, I would batten down the hatches and ride it out. But we'll see.

Yeah, I don't understand. They were unhappy with the IPO path. Was that just. Who knows? Citroen maybe wanted to delay it more. It sounds like they're more like cfo. Well, so they brought in Blizzard, Activision. Activision former CFO Humam Sakhnini, maybe mispronouncing that, but he's a new CEO still seemingly taking them down the IPO path. But. But I don't know.

Yeah, I mean, like, how much of that is on the CEO? I mean, maybe the road shows, like he's not telling the right story or something to these, to these like institutional investors who would be the anchors, anchor buyers in a roadshow. But I don't know. I think Jason's got it. Get him back in there. Bring him back.

Yeah, you don't hear.

I'm a big fan of the time.

The founder took a step back. Super fascinating and came back in.

Yeah, no, no, he has a super fascinating story. He didn't follow like the traditional Silicon Valley. He actually has like an un, what do they call it? Like untraditional path, Untraditional non traditional background. Like he went to a, a small college focused on like game development, built a couple games, wound up selling one to another company, and then started Discord and kind of grew from there. There's some old footage of him at like TechCrunch Disrupt, like pitching his like, mobile league of Legends competitor is very fascinating. But yeah, he's been on absolute tear. Let the guy, let him cook. He's been at it for 12 years doing his work. Let him, let him give him more, more work. Anyway, Delian continues to be on talking up the timeline. Absolute tear says Bill Gurley has now commented on three different podcasts that the only reason Hill and Valley is hawkish on China is because we're talking our book. Or perhaps it's because the organizers are proud patriots and recognize that China is playing a 50 year game looking to unseat the United States.

Yeah, the tension between these two palpable and I think we should get them.

Both on, have them to be for.

A pay per view.

Yeah, pay per view.

Delian. Mr. Bill Gurley, you're welcome to come on and settle. Settle the score.

Settle the score.

Just a couple venture capitalists talking it out. Yeah, I think they could find common ground.

I, I would certainly hope so. Anyway, if you're looking to get in the action, head over to public.com investing for those who take it seriously. They got multi asset investing industry leading yields and they're trusted by millions.

Millions.

Semianlysis has a fun little anecdote here about why chips are rectangular when when wafers are round. Silicon wafers are round because the silicon is grown in a cylindrical ingot using the Sircholsky method and then this ingot is sliced into circular wafers. Circular wafers pose a challenge. Rectangular dies cannot fit, cannot perfectly tile a circle. So any die that overlaps the curved edge is incomplete and must be discarded. By contrast, the in panel packaging we use organic in panel level packaging we use organic laminate materials, not grown crystals so they don't have to be circular. To illustrate the amount of area wasted, we used the semianalysis die yield calculator, which they provide for free, to compare how much waste area there is using a 300 millimeter wafer versus a 510 by 515 millimeter panel wafer using an interposer that is 61 by 68 millimeters. Our results show that panel wafers offer significantly better area utilization than circular wafers. Very interesting that Dylan Patelis is going this is why companies like TSMC are actively developing panel level technology with many of the industry's first framing.

I just want to say if you're analyzing semiconductors, this is who you're analyzing against.

He really, he really is incredible. Anyway, we we have our next guest in the studio, George Hotz. How you doing George? Good to hear from you.

What's going on. Welcome.

Can we hear you?

Oh, can you hear me?

Yes, yes, we can hear you.

Gotcha.

Let's kick it off with something simple. I want to take your temperature on AGI. Timelines. P. Doom, the easy and fun stuff.

I don't know what AGI means and I don't know what you mean by doom.

No. Are these terms just like, entirely irrelevant? I mean, now we've shifted to super intelligence. They're all buzzwords, but at the same time there is an idea of the. I don't know, the conversations may be shifting to the AI, generating more economic value than humans. Is that a relevant metric to track?

Machines have been generating more economic value than humans since the Industrial Revolution.

Is there some other metric that we should be tracking or is it just like, irrelevant?

You're just talking about, like, hype. Like, I don't know, I mean, I don't know what you mean. Like you can talk about concrete things.

Yeah.

Term like AGI means nothing, right? Like computers. Everything that's a Turing Machine is a general purpose computer. Is that what you call intelligence? I don't know what you mean. It's a linear regression. Intelligent. What if it's big enough? The Chinese does know Chinese.

Yeah. I mean, what about your, your decision to get on a spaceship traveling at 0.9 C away from the.

Yeah.

From the Earth? Like, how close are we to that? Are we closer than the last time we talked, which was like a couple years ago, and it seemed like it was maybe going to happen within your lifetime? Has it moved at all?

Yeah, I don't know. I don't know if I'm actually gonna get that spaceship, but it's kind of like in an ideal world, what I would want to do, you know?

Yep.

Just, just, just back away and chill. Don't look back. Actually, you can't look back.

You can't never look back.

They're all there.

You need the, you need the blast shield, right?

You need the information shield.

Information shield? What do you mean?

Oh, that's how they're going to get you.

Okay.

Right?

I mean.

Okay, so here's a way you can think about AI, right?

Yeah.

Imagine there were 10 CIA agents assigned to you and they're running at 1000x real time. So they're like hyper fast CIA agents that devote their entire lifespan to your day and they're trying to manipulate you, maybe to get you to buy things, maybe to get you to vote for a certain guy, whatever. But like, that's what you're going to be up against with AI what we're currently building. What if you think about the biggest companies in AI, what they do is advertising. What advertising is, is just manipulation of humans. So you're going to have a team of CIA agents thinking about you and trying to manipulate you at all times. And now you see why you want to head away at time, the same speed of light.

Right.

Even CIA agents can't beat that.

Is there some world where there's like a capital war and I'm paying for a more powerful ad blocker?

Yeah, I mean, that sounds good. Like another question is kind of to say, like, okay, if you think that you either think that current like capital accumulation dynamics are going to continue and that the rich are going to continue to get richer, and if you believe that, the question is kind of, well, how many people are going to survive in the future? How many people are going to have any modicum of independence? Right? Like you have some far AI people who think that there's going to be a singleton, right? There's going to be literally one. Right. You know, some people maybe think it's 10. Some people thousand, 10,000. Some people think that all the humans will get to continue to exist as independent entities. Are they already independent entities? That's a question, right? I don't know.

Question. I mean, if you were trying to put it in like the form of a bet, human population above or below 8 billion in 2030.

Bob, I think I would just give you a normal trend. Is that what the trend says? Yeah, just go. The trend says, I don't think there's going to be any discontinuities to any trends. Really?

Well, yeah, I mean, I mean at some point. But the question is like, how far out do you have to go until you start seeing these effects?

What do you mean by human? Right. What about someone who lies in bed all day and watches TikTok? Are they human?

Yeah. That is odd. They kind of drop out of society.

I think. I think question that popped up for me is, is this all this debate about AI safety and what should labs be doing? What should labs not be doing? It feels like your angle is it should be each individual's responsibility to look after their own safety in the context of, of AI. Is that at all?

I just, I just like this whole like, should, shouldn't. Like what?

I don't know.

I'm not a sadistic fuck who wants to manipulate other people like the people in power, like, I don't know.

Yeah, but I mean, people still look to you as like an example of like someone who might have answers.

No, I don't have any answers.

Not, not necessarily answers, just like.

But you can buy my shit coin here. Did I shill a shitcoin?

Here you go.

Just click this QR code and you can buy a George Hots coin. And that will give you answers. You will find satisfaction and fulfillment in your life after purchasing a George Hots coin.

Is that the end state? We all have our own coins.

No, no, no, I don't mean it like that. I mean like I think that a lot of people are like, they don't really know what they're looking for. And that vacuum is, is very, you know, it's very dangerous and it's going to be filled by dumb shit and don't have that vacuum. Right. You got to, you got to stand for something, you know, or something. I don't know.

Yeah, I mean, do you think that there's a chance that someone is able to take a stand and actually bend the arc of, of AI progress in the way that, I mean it happened with nuclear, right. Like, like nuclear development did stall. There was a stagnation in real world build out of nuclear capability on the energy side.

Yeah, I mean there's a few things about nuclear that make it different. So nuclear, even as a weapon is incredibly hard to deploy tactically. Right. So, so if a country has, has nuclear weapons there, aside from like a mutually assured destruction idea, they're not all that useful. Yeah, it's not like you can use a nuclear weapon to accomplish tactical objectives. You know, if you could, I think Russia would have already done it.

Yeah, right.

Russia has some tactical objectives they might want to accomplish, but nukes aren't really going to do it. Right. And from a pure real politic perspective, not even from a like, oh, like a taboo moral perspective, like what do you want? An irradiated pile of rubble, like that's what you're going to get. No, what you want is drones that are hyper specific and can take out exactly who you want, can control areas. Right. So like as a military technology, nukes are not that good. AI is way better.

Yeah, but what about it as an energy technology? It feels like, it feels like the fear, like the mimetic fear of nuclear war and total destruction caused a whole bunch of regulation to pour into a sector and essentially a stalling of nuclear energy build out. And if, if the AI doom scenario, whether it's real or not, becomes so memetically powerful that someone's able to harness that and actually say if you try and build a Big data center, we will shoot you. Then maybe it stagnates.

No, really think that's the reason for nuclear? I think it has more to do with why we can't do other big infrastructure projects in this country.

Right.

Like, it doesn't have to do with the new. We also can't build dams. Right?

Yeah.

And if you look like that is the thing, if people think that there's some weird taboo around nuclear. Right. But then, okay, look at hydroelectric, right? There's no taboo around hydroelectric. But China leads in installation of both nuclear and hydroelectric and. And everything. It's almost like they're correlated.

Right?

Yeah.

So the thing is not there's a specific fear around nuclear. It's like, you know, the US Decided that they're a developed country and we're not going to develop anymore because we're already developed. You see the D on the end, right?

Like, interesting.

Yeah. Is that. So is that just cultural, then when you are like, the malaise sets in. Would you expect that to happen to China when they catch up?

I don't know. I. Yeah, I mean, maybe it's just like this normal story arc of, like, you know, it's. It's. I don't know. I don't know. I think that, like, you have a real problem when the kids can't live better than their parents.

Yeah.

So. But I don't have anything more to speculate on that.

Do you have more context on. On China and specifically in, like, the AI context, like, US Electricity looks like.

This and China electricity looks like that.

Is that all that matters?

Pretty much, yeah. I mean, that's a pretty good proxy for everything, right?

Yeah.

Like, there's two things. There's two things. You know, people are like, george, how do you feel about the Trump administration? I'm looking at two things.

Yeah.

With any administration, I'm looking at two things. Did you decrease government spending and did you increase total electricity production of America? Those are the only two numbers I care about. Those will capture everything.

Why does. Why does government spending matter? We were joking that, you know, Trump must be extremely AGI pilled if he's running up a massive budget deficit.

What the hell is AGI?

I don't know what this is like in this formulation. Never seen it in this formulation.

That's numbers.

Yes, yes, yes. But it is an extra lever on labor and capital, and it creates more GDP than can be taxed to pay down the increasing amount of debt.

Super excited. Excel.

Yeah, Super Excel.

What is. What is. What does Super Excel do that normal Excel Doesn't.

Let's give it up for better Excel. Yeah.

Yes.

We need Excel 2.0, right?

Yes.

The thing is, Excel was the, the final piece of software. And then. But in order to add another, you know, $100 trillion to global GDP, we needed to, like, kind of rebrand it. And so now we get AGI.

GDP is the, the complete. It's biggest bullshit thing ever. Right. Like, I always joke with my friend and I that we're gonna start companies and be billionaires. And I'll tell you how we're gonna do it.

Okay?

Right.

I start a company, he starts a company, we both write contracts to each other.

Yep.

Right. Like, I'll buy something from him for a million dollars. He'll buy something for me for a million dollars. We'll just do this real fast. We'll keep passing the money back and forth. Whoa, look at our revenue. Wow. That all contributes to gdp.

Wow.

We made. We're billionaires overnight, right?

Yep.

Like, like, and that's my argument is the economy is just that with a lot of extra steps. Right. You can't use services. It's not part of gdp. This is complete nonsense. You can't have services. Like, literally, Literally. You take the steel out of the ground, you grow the corn. Okay, that's gdp.

But is, I mean, if that GDP is fake, is the deficit not fake? Like, is government spending less money?

Oh, we owe that shit to people. It's not fake.

But can you just tax the fake. The fake money? Like if you tax your scenario where you're generating a billion dollars dollars in fake money, you can't tax the fake.

Money because we're passing the same dollars back and forth. The minute you tax it, that falls off so fast.

Yeah, yeah, yeah.

You can only tax productive work.

Is AMD doing productive work right now?

AMD's doing all right. Yeah. Either Nvidia is really overvalued or AMD is really undervalued. It has to be one or the other.

How does it all play out? Like, what does AMD actually need to do to get back on track or realize their potential?

Nvidia needs to stumble. I mean, it worked for AMD and Intel, right? Like, so AMD ended up beating intel in the entire. Like, no one would buy a data center Intel CPU anymore.

Yeah.

And it's just because, well, you know, they stumbled and now intel owns that market.

Yeah.

So, you know, AMD just sits there in second place. They're being a better second place than they were a few years ago.

Yeah.

And then when Nvidia stumbles. AMD is like, oh, hey, we're here.

Is DGX Lepton like their cloud offering a potential stumbling block, or is it the right move for them?

I don't know what that is. What's an Nvidia cloud chip?

Yeah, exactly.

Cloud's dumb. Cloud dump. You can break AI down basically into like, there's like five tiers, right? Like at the base level, you have like electricity and data centers and land and like, things like that. Tier 2 or like TSMC, ASML, Samsung, intel, right? Fabs, Nvidia, AMD, OpenAI, anthropic. And then on top you have like completely worthless things like cursor and Windsurf. You know, these character AI. All these people who think, oh, we're the app. We're gonna, we're gonna get the ARR. No, that worked to the web. It won't work for AI. And I can go into why, but it's kind of boring.

I want to keep going, keep going, keep going, basically.

Okay, so like, here's the difference between AI and web. Would you want to run a service like Gmail? One server can serve 10,000 people easily, right? And there's no demand for like, better Gmail, right? It's not like, it's not like I can click and get like, yeah, you can buy Gmail Pro. It'll have a few things, but most people don't really care, right? There's no limit to the ceiling of how good you want your AI to be or how fast you want your AI to be. Maybe there's a limit to the speed, but like, when you're at like a thousand tokens per second, I want the biggest model in the world, right? Like, so there's very little limit on that. But suddenly you can't serve 10,000 users from one server anymore, right? And the whole dynamics of the web, the whole reason, some of the value aggregated to these end players and they still didn't aggregate to the cursor and the wind source. They aggregated to the OpenAI of the anthropics, right? Nobody, nobody, nobody who built like an email client survived. They all got eaten up by the tier fours of the web, right? The Googles, the Facebooks, all of these, like, app providers, right? Where's a Zynga today? You know, like, this already happened, right? People just don't. Where's Zynga? Oh, Zynga's gonna be the next thing, man. Like, no, it's not. Facebook ate all of that value, right? Google ate all of the Value from all the people building on top of Google. So the tier fours ate all that value.

Yeah.

So OpenAI Anthropic will eat all the value from the cursors in the windsurfs of the world. They'll acquire some of them. They'll compete with some of them. Right. Same as you saw on the web. But I argue that the tier fours aren't even gonna have value because the tier fours, this ain't the web. This ain't where you can have one server serve lots and lots and lots of people. You know, I'm running 03. I'm running. You know how much I cost open AI every month? I pay the 200amonth. I cost him a lot more than that. You can now click on Codex. Yeah. Spin up four nodes. Yeah. Why would I not click four? It's not my computer. You gave me the button. Hey, I'm just using it.

George Hots single handedly bankrupts $300 billion company.

No value in just being the front end to AI applications to be like the front door, just the default button. Because we see these models kind of go back and forth in terms of benchmarks or what's hot. And there isn't as much customer churn as you would expect because people are just kind of like defaulted into the app that they installed whenever. And so even if Gemini gets better in terms of the actual performance metrics, people don't switch from OpenAI to Google.

It's so negligible. You got to make something 10x better.

Okay.

You got to make something 10x better. So like this whole game is open AIs. Unless they stumble.

Sure.

I'm not switching to gemini because it's 20% better. And I download some new app and think about all the thing.

Right.

No one's going to switch. Is there, is there a chance for a company to kind of come out with something that's 10x better with a algorithmic improvement? Or is it just a race for scale? Like what could actually be that? Next it felt like GPT 3.5 when they really broke through with DaVinci. And then four. And then four. Like, it felt like this kind of like binary moment when a lot of people realized that this was usable for their daily life. Even if it's just a Google search replacement or whatever, Write a poem or whatever. Like a 10x what you're describing, like a 10x improvement feels like that kind of like, like qualitative binary shift. Is that possible with just scale? Or is this something that we need, a different Model for?

I don't know. I don't know. I would bet majority still on. Like these big labs are also attracting the talent. But it is also like it's pretty commoditized. A lot more so than like Google search. Right. Like you can look at people, track how far away sources behind. It's not that far behind.

Yep.

So no, I don't know. I think this game is mostly going to be chatgpt. I think Elon's aware of this too. That's why he's trying to go 10x bigger with the data center.

Yep, we'll see.

Maybe it'll work. You know there's, there's someone to bet on anthropic. I'm not that bullish on but maybe.

You kind of predicted the, the pre training wall. But that's not a refutation of the bitter lesson. And we're going to see similar scale play out in reinforcement learning or is there going to be something else that we're building the big data centers for?

There's something that we don't understand in terms of data efficiency. So like when you think of how long it takes a GPT to learn to talk, like how much data it takes, it takes like terabytes of data in order to make a GPT talk like a normal person. It takes terabytes of data. Whereas a human trains on megabytes.

Yeah.

How is it that if you take all the text that you've ever heard in your life and you put it to whisper and you transcribe it, it's going to be a couple of megabytes, 10 megabytes, maybe 100 megabytes. So humans have this thousand X data efficiency advantage and we're going to have to fix that if we want like reinforcement learning to work. Especially like reinforcement learning that you want to do in the real world. Humans could do. Humans can learn from very few samples. And yeah, I think that like it might be okay if these foundation models train unsupervised on lots and lots of stuff.

But yeah, is that, is that something that somebody's working on just like a new more data efficient algorithm to drop into the pipeline or do we have any like leads there? Because it feels like right now we're going down the path of like reinforcement learning with verifiable rewards and we're going after like individual business use cases that are increasingly long tail and that could be kind of like valuable. But it doesn't feel like the breakthrough that you're talking about.

Like has there ever been a breakthrough Right. Like people think GPTs are a breakthrough.

No, they weren't.

Like just, if you watch the world.

It was just, it was all just smooth curve.

But what I will say about AI scaling laws. Oh man. You see, like people get excited about AI scaling laws, but here's a pitch that'll kill your excitement immediately. Ready? AI scaling laws. You can put in exponentially more money to get linear returns.

Exactly. Do you believe that the real value is investing in humanoid robotics? Then have you heard this theory? So, so it, so I mean if you, if you put exponential more money into humanoid robotics, assuming that they work, and assuming you can, you can make 10 times as many robots, you get 10 times as much output.

Anyone, Anyone who wants a humanoid robot has never worked in a factory in their life.

Okay, right, bring it down.

No one wants a human. Oh, yo, yo. It's gonna walk around. Oh, good thing it has leg, right? No, here's what I want. Yeah, yeah, we got a laugh track. Can you show me a robot arm that's capable of putting a screw in something? Probably put the screw in the thing.

Yeah.

No, no, no, no.

Not like you carefully jigged up the.

Screw and had a screw dispenser.

Like the way a normal human does.

It with a screw sitting there in.

A little bucket on the thing. And it picks up one screw and it puts it in. It takes a screwdriver.

No, no, no, we're not close. But also it feels like we're not far. It feels like.

That's what I feel like. Humanoids are this interesting sort of like space because a lot of smart people just say like here's the 20 reasons why they won't work and like why we shouldn't build them. But then so much capital and so many different teams are trying to make them work that they very well might work for some things. And like they just, humanity might brute force it because we saw it in a sci fi movie now, you know, 30 years ago.

Why, why, why are we cooked on.

This is as dumb as self driving cars was, right? And nobody learns their lesson. And people like Kyle Boat should be ashamed of themselves like they really should. These people who go and raise large amounts of money for another thing that like they should know, they should know better. Right? Here's basically like, remember in 2012 and Google said that, you know, my, my, my 12 year old daughter would never have to get her driver's license. Yeah, Come on, that's nonsense. Right? And like now, okay, they shipped Waymo. It's in a few cities. They're teleopped. Right.

How teleoperated are they in your opinion? Is it, is it effectively one to one?

It's more than one to one. There's probably about, I would say there's 1.2 operators per car, but it's not. They don't have a steering wheel and pedals. Yeah, it is, it is an autonomous system that they're probably doing some higher level inputs on. They're definitely like saying when you can be aggressive, when you should slow down, you know, whether you can turn to the stop sign or not.

Yeah.

You know, again like here's, here's the simple reason to know that it's like that.

Right.

There's definitely some teleopaths, right?

Yeah.

Have you ever seen a picture of that room?

No.

Yeah. Why not?

I've always thought it was like an ace up their sleeve because like if, if there's a lot of pressure on them to say these waymos aren't safe, they can pull up, pull the, the sheet off of the ghost and say there's actually a human in the loop. Don't worry, it's safer than you thought.

Yeah, yeah. Like the fact that you've never seen that room tells you that it's way worse than you think it is. Right.

Okay.

Tells you that there's way more teleop than you think it is. If it was really one person supervising 10 cars, Google would post those pictures all over the place.

Yes, yes.

You don't see any pictures. There's so cruise that actually came out in the lawsuit. I think it was like 1.5 or 1.7 humans per car. Right.

Or vice versa. Like. Right. 1.5 cars per person. Right.

No, no, wait.

More people than cars.

Yeah, that's what he's saying.

Because an Uber only requires one person.

Yeah, but, but so, so maybe the real innovation is just allowing somebody to get in a car with and not have to talk about the weather or.

You know, I'll pay more for that. I'll pay more for that.

Yeah. But I mean, is there any hope that we drive this down and we get to two cars per person, then four cars per person, it starts doubling exponentially and eventually like we are there.

I mean. Yeah, like it's obviously going to happen. Right. It's obviously eventually going to happen. If you want to see where the real state of the art of unsupervised self driving is today. Right. There's no person with fsd. When you get your Tesla that's not teleop, you can go press FSD and that's real AI.

Yeah.

And well, you can see how good it is, right. Would I take a nap in there even for five minutes away and now? Yeah, you'd be super.

Right. How are things going on the comma side? Give us the update there.

Pretty good. You know, we're on track to, we're on track to be two years behind Tesla.

So there you go.

So two years behind Tesla. But you know, here's why we win, right? Like because like it's cheap. Okay. So when you think about self driving cars, it doesn't look anything like the rollout of Uber, right? Or Airbnb. When you roll out something like that, you're trying to roll out a two sided marketplace. You got to spend tons of money on customer acquisition costs. You've got to make sure that you've perfectly matched that marketplace right away. Because if drivers aren't getting rides, they're going to leave the platform. If riders have to wait too long for drivers, they're going to leave the platform. So it's this careful balancing act. But once you get this marketplace, you got, you got a moat, right? Switching costs are real high. Try to get everybody to switch at the same time. It's a challenge point. Never do it. Self driving cars don't look anything like that. Self driving cars look like scooters. The only thing that it's going to take to roll out big fleets of self driving cars is capital. Right. It's just strictly a capital market. You could just. I could. If I look at a city, I can calculate how many Waymo's there are. If I want to build my own network and deploy that network and run at a lower cost. It's straight up capital. Easiest thing for investors to calculate, very little risk. So self driving cars are going to be this awesome race to the bottom. It's going to be like scooters where there's going to be like 10 providers of these things for a while and then they're going to consolidate like one's going to do it. But yeah, people are really going to win.

What's most valuable in terms of developing the next, like the next better version of full self driving? Is it having a lot of data, building a big data center, having a great team to actually design the system? What's most important? Are they all equal?

Yeah, all those things matter, right? I think the main thing that matters more than anything else is just time. Like we're figuring things out with research. Infrastructure is getting better. I think a lot of it's just infrastructure My new company's AI infrastructure.

Right.

Like the infrastructure gets better. My coworker has a saying, what we do is that we make the hard things easy and the impossible things hard. That's like the goal of infrastructure. So you build infrastructure, your infrastructure gets better, and then what you couldn't even dream of doing 10 years ago is now one command today. And today.

What'S the current use case for most people with tiny boxes? That's by design, right? You're not supposed to know.

But I mean, I sell a computer, it has specs, right. So many people want to tell you. And I hate this, I hate this. They're telling you like, how the product is going to impact your life, what you use the product for. Oh my God, who cares?

Yeah, here's what it is.

I'm going to tell you what it is. That's your job. Right. I'm not an advertiser, but I mean.

It'S like, I mean, our intern wants to build something with a tiny box. I want to give him some ideas.

Go buy one.

I don't know.

Why do you want to build some of the tiny box? I mean, is it good?

Yes.

Just a bunch of GPUs in a box. You know, it's a nice metal box.

GPUs in a box. What robotic form factors are you most bullish on? We've touched humanoids. You gave a great review there. We've touched autonomous vehicles. Sounds like generally bullish, but capital wars, race to the bottom, all that stuff. Are there any other kind of form factors that you're thinking about that you are generally optimistic or excited about?

Arm. Two arm. Right. Just two arm. Right. Because I look, I run a factory. I run a factory in San Diego. We make all the commas right here. And I can't wait to get a whole lot of robots in there. But I don't need humanoids. I'm just gonna stick two arms to the table and then it's going to grab a comma, it's going to put the screen on the front, it's going to flip it over, it's going to put the four screws in it, then it's going to pass it on. Show me anything that's anywhere near that level today.

Yeah. What would you do if you were trying to build like a truly multi purpose robotic arm?

The arms are already good enough. It's the off the shelf. Arms are fine. It's all software again. It's always all software. Autonomous vehicles are all software. Robotics is all software. But everybody loves to Bike shed.

Yeah.

What color are we going to paint the humanoid? Yeah, like let's have a great conversation about that. Well, you know, we don't want to paint them red because that might scare people in a. This is actually the level of stupidity that I see in those discussions about humanoid robots.

Yeah. Well you.

Would you trade in your legs for wheels if you could?

I got that in my legs for wheels.

Yeah. This is a question from Aaron Frank, friend of the show Real time weird wheel guy.

Then like people had asked me if the wheels are like making a statement. I just don't want to have to have this conversation.

What about the sim to real gap in robotics? Like how, how is, how is simulated data? You know, you build a bunch of data in Unreal engine then you try and transfer learn it back. Obviously there's been a bunch of experiments of that with self driving cars. Is that a path that we should be going down for, for the robotic arm development?

Yeah. So I think with a lot of sim 2 real stuff, the reason people are excited about it is because of that data efficiency gap we spoke about. Right. Like current machine learning algorithms like a thousand x less data efficient than humans. So yeah, you're going to need a thousand x more data. Right. If a human can learn something in one example or 10 examples, the computer is going to need 1,000 or 10,000. Now do you really want to reset the stupid state of the physical world 10,000 times? You might do it 10, but you're not going to do it 10. So that's where you want a simulator where you can just click reset and everything's back to exactly how it was. So I think this stuff's going to play a role, but I think more fundamentally that data efficiency gap has to be understood.

We talked a little bit about coding agents. We talked about how you're bankrupting OpenAI by spinning up a lot of different Codex agents. What other sort of Agentix software are you excited about? Do you expect to, you know, what's agentic mean? Yeah, basically bots. It's, it's like what we're calling bots now. But, but, but, but anyways, like I just, you know, I don't know.

Star Trek.

Yeah, yeah, maybe. No but, but, but I think about a world in the future. You know, do you, do you expect to be, I don't know if you're a slack guy and imessage guy, Discord, maybe no messaging at all, just you know, you know, community telepathy. But, but do you expect a world in the future where you're just, you know, perfect interaction between, you know, human employees and agents. Or is it, you know, going to be more like, you know, you'll do the odd deep research or maybe you send some automated outbound emails or have some Codex bots running.

I don't even like, follow this.

When do you think you'll be able to book a flight just by saying, I'm trying to get to New York tomorrow.

Oh, See, the worst part about this is like, like that's gonna come pretty soon actually.

Okay, right.

We're gonna pretty soon have computer use models that are actually capable of going to delta.com and booking a flight.

Yeah.

But then what's actually going to happen is Delta is going to partner with whatever company does that and they're going to put it behind the stupid thing. And like. Yeah, so that's gonna, yeah, that's gonna be here in a few years, Right? Not with agentic, but just with normal hooking the APIs together. Right?

Yeah. Wait, so, yeah, what is the bullish on APIs? What is the mistake about like the agentic buzzword? Like what are people like even describing again?

It's another thing that I really have no idea what it means. I was hanging out with some friends last night and like my friends, this VR company and the CEO is really interested in things being open source, but he's also really interested in making sure that things are protecting our intellectual property and proprietary. And the truth is he has no idea what the word open source means. He has no idea what it means that they can copy his shit. Right, like, that someone else could use it. He just, he just heard the word open source in some like buzzword thing and he's like, do we have the open source, do we have the open source thing?

Okay, so check the box.

Check the open source box.

Last question about the agent. Last question about the agentic buzzword. I think that there is something that people are picking up on, which is that these models seem to be very smart for short amount of time, but if you run them for a long time, they start hallucinating and kind of going off the rails. And so you have like 10 minute AGI. It feels incredible, but as you let it run and do more work, you can't just say, hey, go do a week's worth of work. Come back to me when you're. But it's superhuman in one minute. And so is that kind of trade off curve real? And then is it just a matter of like, better harnessing to actually get to two hours of Work, which is kind of what the agentic people are like advocating for.

No. So I don't think it's better.

Hard.

But this is definitely a real phenomenon. This is definitely a real phenomenon. You can experience this. There's papers exploring it which show that if in 10 seconds there's absolutely no way I'll come even close to a modern LLM.

Totally.

Because the first shot from the LLM is great and then it kind of degrades and it degrades pretty quickly. Whereas humans look a lot more like this. Humans can stay coherent internally for much longer. So yeah, I think that that's a real thing. I think that that's mostly going to be fixed by like long context, just more energy. Long context. RL yeah, just like you just gotta do it. We'll figure out new ways to make the context better. We'll combine diffusion and, and autoregression in some clever ways. Yeah, I think that this is just gonna be like, there's not gonna be a breakthrough here. There's not like one magical thing that we're missing. Yeah, I think it'll be a continued plot. The same thing with data efficiency. I think people will start to care about it. Some new tricks will come out. Some of them will work, some of them won't work. We'll continue the do graduate student descent until we find graduate student.

Anything. Last question for me. Anything that you're particularly optimistic about. Anything. You check the timeline and you think, this is awesome. I love this. I love this. I want to see more of this. Maybe a little white pill to kind of cap it off.

Yeah. So here's something I'm optimistic about. That fact that the one server can't run 10,000 users. That is most of the reason that the modern Internet. That is one of the reasons that the modern Internet sucks. That so much of the stuff is in non recurring expense. And then it becomes really, really hard to compete with these people. Right. Like you could run Twitter on one computer and 20 people could do it too, but they don't. Because again, these companies have moats and they invest in making sure that their moats can't be broken with AI, I think there's going to be a much less of a vote, especially when you look at the move from autoregression to diffusion. So auto regression can run in large batch sizes. When you run ChatGPT, you're running with a whole bunch of other people on that same computer. Yeah, it's only 100, it's not 10,000, but still it's 100. Diffusion is run in the cloud at batch size one. And once you're in batch size one, land, running it locally starts to make sense. Actually running the models locally, or at least having your own computer in the cloud, not being some shared resource that's really controlled by something else. So, yeah, this was never a thing because you can't put lots of people on a gpu.

That makes sense.

They tried some weird stuff with the.

Licensing, but yeah, fantastic. Well, thank you so much for stopping by. This is a great conversation.

I wish we had a full hour. This is great.

We'll talk to you soon. Josh, George, Cool.

Bye.

See you later.

Cheers.

Bye. Next up, we have Joseph Trogian, author of the Party's Interests Come First. He was recommended by Jordan Schneider of ChinaTalk. We're very excited to talk to him about the life of Xi Jinping's father, Xi Jin Chung. Welcome to the show, Joseph. Good to have.

Thank you for having me.

Thanks so much for having me.

It's great to have you.

I would love to get a little bit of your background, how you landed on this topic. It's incredibly difficult to. To. To research. I was digging just, just into the life of Xi Jinping, and there were no English biographies for a long time. One of the most deepest dives on Xi Jinping was from the Economist, actually in the form of this, this podcast and this reporting about the prince. And I was interested how you got into this, what your background is, and then we could go through some of the story.

So I had just finished a book about elite power struggles in the Soviet Union and China after the deaths of Stalin and Mao. And someone asked me to write about party history and Xi Jinping. And I thought I would do a short article about Xi Jinping, a little bit about his father. But what I did find out was the more research I did, the more I could learn and that I could tell a really interesting story. 3 Xi Zhongshun that wasn't just about Xi Jinping, but could be a sort of microcosm of Chinese Communist Party history in the 20th century.

Got it. One of the themes that I've been kind of wrestling with in my takeaway from studying Xi Jinping was that there is this constant narrative of being both a victim and a perpetrator of party aggression, essentially. And I was wondering if that narrative feels right to you, where that comes from, why there's so. Why these. These leaders are so reluctant to reject the system that's sending them to jail or putting their family in hardship. Like it feels Unique. And it doesn't feel like something that happens in America or maybe I'm just brainwashed by American politics or something, and it is happening over here, and I'm just not picking up on it. But is that a theme that you picked up on and you think is, like, worth digging into? Is it. Is it what's unique about the story?

It sounds like you read my book quite closely. That is certainly one of the themes. There's a puzzle there which is. Xi Zhongshun, the father of Xi Jinping, was persecuted by his own party on several occasions, and the party asked him to do things that he thought were wrong. Nevertheless, he remained devoted. And I think to appreciate that, we need to look at this Bolshevik communist political culture. And it's because these people see the Party as a source of meaning and purpose in their lives. So in that sense, to reject the Party would have meant rejecting themselves. So perhaps counterintuitively, when the Party hurt them, the motivation was to redouble their efforts to win back the party's trust in them.

Do you think that there's. I'm still struggling to understand the dynamic of, like, how dominant the single party is in China. And I'm wondering if. If we compare it with American politics, we. It feels like we're always in this 50, 50 stalemate, where every four years it's like it's neck and neck and then. And then, yeah, maybe one party wins by some sort of margin. But I'm. I've always wondered if that's a. If you talk to somebody who's like a socialist, they're always really upset because they're basically like, well, they're both capitalists. Capitalists who are running. And so they see it as a false choice. But the vast majority of the American electorate is incredibly animated by the differences, even though they might be somewhat minor if you zoom out a little bit. And I'm wondering, like, the more single party system versus the dual party system, how does that emerge? Can you give me some history of the party and why is it so stable?

So Lenin said that what we are doing is building a party of a new type. And he thought that Western political parties, as you said, were essentially representative of dominant class interests. What Lenin wanted to do was create an organizational weapon. And the reason for that was he was running a conspiracy and he wanted to take over the country and then use violence to transform it into something that was completely different. And to do that, you need to create an institution that can force people to do things that they might not want. To do. It's by design constructed so that the top leader can make a choice, and then everyone has to follow along with that choice. And sometimes that choice is wrong and has devastating impact for the party and the nation. But you can see that if you have such an ambitious agenda, what you want to do is create a system where the top leader is sort of firewalled from any kind of political consideration.

Can you? I mean, there's so many. There's so many anecdotes that could kind of tell the story. Would you mind telling me which story from Xi Zhongsheng's life really stuck out to you as emblematic? Or maybe even just give me the high level, because we kind of just jumped into it. Give me, like, the high level story arc that you decided to weave the entire narrative through.

Yeah. So he was born in 1913. That was only two years after the collapse of the Qing Dynasty. He was born into a rather dramatic place. He was born near where the terracotta soldiers are, which probably many of your viewers know about. That was where Qin Shi Huang forged the first unified Chinese state thousands of years before. And it was the city where emperors had ruled for. For millennia. But by the time Xi Zhongxun was born, it had fallen into years of banditry and war and famine. And he was attracted to radicalism, trying to figure out a way to help China escape from these imperialist encroachments and political infighting at home. And his first political act was an attempt to assassinate an academic administrator, and it failed. He got a bunch of teachers sick and was thrown into prison and joined the party there.

Wasn't he, like, 15 when he was sent out?

That's right. He was very young. And what's interesting, too, is he didn't really have an intellectual attraction to communism. And when he was released from incarceration, he read this novel that's a terrible novel. It's basically just a protagonist who goes from one disaster to another. But it fetishizes this idea of resistance and struggle. And so it speaks to the fact that for a lot of these people, their intuition was that something needed to be changed, but they weren't exactly sure what. And the Party was a source of meaning for them.

I. I know you mentioned that he had a lot of interaction with the party around relationships with other religions. And I was wondering if you could. If you could kind of like when I grew up, I remember, like, the Free Tibet movement, and it's kind of like fallen by the wayside. Tibet's not really in the, in the, in the news much anymore. Could you take me through the story of his interaction with some of these kind of tangential groups in China and, and how, how the history of those relationships played out?

Yeah, I'm glad you asked that. So within the Chinese Communist Party, you would have people that were experts in the military on economics. Xi Zhongshun was the leading united front figure. So what's the United front? Well, Mao Zedong called it one of the CCP's magic weapons. And it's basically political influence campaigns. And so what you do is you figure out who's on your side, who's in the middle, who's against you, and you empower the ones that like you, you win over the ones that aren't sure, and then you hurt the ones that are, that are dead enders. And this was something that Xi Zhongsheng continuously applied to China's ethnic minorities. And so he was, in the last years of the civil war, in the early years of the People's Republic, the leader of the so called Northwest Bureau, which included a huge expanse of China, including Xinjiang, but there were also lots of Tibetans in Qinghai and Gansu, which were other massive provinces in China. And so he was trying to figure out how to incorporate them. And it was definitely bloody. But he also wasn't blind to the advantages of finding local power brokers and winning them over. But over the 1950s, the party decided that that approach was dated because people weren't coming to socialism on their own. And so essentially, the party declared war on them with tanks and planes. And during the Cultural Revolution, any sign of an ethnic difference was seen as class struggle. And then in the 1980s, the party understood that they had really screwed up and that they needed to have a new approach. And Xi Zhongshun ran ethnic politics for the Secretariat, which is sort of the Party's brain. And he tried to use economic development, bringing religion out into the open so it could be better controlled, allying with local power brokers. But by the end of that decade, many people within the party concluded that when you do that and you give people space, they just use it to hurt you. And so the party decided that growing protests weren't a sign of reaching a new equilibrium, but, you know, hitting some road dumps on the way, but that that fundamental approach had been a mistake.

Yeah, there's this, there's this narrative that keeps popping up whenever you hear about turmoil in the party around purges, which is a term that I don't, I'm not Really, I can't really map to as an American, because we don't really have those, I guess. And, and a lot of this is always tied to. There's. There's corruption, and we're going after corruption. And every, every kind of major defenestration feels like it's tied to corruption. And I'm always wondering how we are evaluating those claims, because at the same time, in a growing developing country, there probably is a lot of corruption and there's probably a lot of people that are banditry, for example, that you mentioned, there are people that are taking advantage of different parts of the government. So the corruption might be real, but then also corruption is used as a weapon to amass power or remove certain people from power. And so can you walk me through how corruption is used as a tool? How real do you think the various corruption narratives have been? Maybe if there's any examples of, of corruption being wielded as a tool for, for party power.

So as soon as the Party was able to take over the country, they immediately faced a question, which is how they were going to ensure that they didn't take such pride and arrogance in their victory that they allowed bourgeois liberalization, meaning individualism, to seep in and that they would be divorced from the masses and that they would care about their own materialist needs. And so from the very beginning, the Party struggled to figure out a way to eliminate this problem. And so, especially during the Mao era, you had constant, constant rolling campaigns that inevitably went too far. And now under Xi Jinping, you see that he believes that corruption for him is about whether the Party can survive. Because for Xi Jinping, ideals and conviction and a sense of motivation are necessary for the Party to exist. And the biggest danger to that is corruption, because it's. It basically means putting yourself first. But also corruption is a problem because it's a vector for Western influence, because there's this idea that the west is materialistic and oriented towards consumption. And once that gets into China, then the west can use it as a weapon. And in fact, many people in China thinks that what happened to the Soviet Union, that essentially the leaders of the regime became oriented towards the west, and the west won what they call a war without gunpowder. Meaning you don't use. You don't use violence to destroy the regimes you don't like. You do it by winning over certain people within it. And one other thing to say too here is that you're right. Like, corruption is partly about regime security, but it's also about getting rid of people that you don't like. And so what's interesting here is, you know, why do people keep doing things that make them vulnerable to accusations of corruption? And here, I think, speaks to something else with, about the system, which is nobody's quite sure where the red lines are. And so sometimes you just get it wrong. And sometimes the top leader changes where the red lines are. And you can see that in a system where the top leader is the one who gets to make those decisions, it's a very powerful weapon.

Yeah.

What is, what is the current sentiment from the Party about the state of the world? Right. I feel like here in America right now you have a frustration. China's become the factory of the world. We're deeply reliant on them. At the same time, the world feels very unstable.

Are they becoming consumerist? Because they're certainly buying Xiaomi phones and.

Yeah. So I'd be curious any type of insight around the Party's sentiment around what's happening in China culturally and then sort of geopolitically on the world stage. Do they feel like they're making progress? We, we study, you know, their goals around individual industries like semiconductors, AI, defense, et cetera. But I'm curious what, what, you know, sentiment in, in and out of the country.

So it's interesting to see how ambitious Xi Jinping is in terms of how he characterizes his goals in a holistic sense. He says that for millennia you would have one dynasty collapse after another, and the Chinese Communist Party needs to be break that. And his solution is this idea of self revolution. And basically what he wants to do is figure out a way of how you win over the third and fourth generation of young Chinese to the revolutionary cause. And your questions speak to a dilemma there. Right. Like you can see how a message of sacrifice and rejuvenation is meaningful, maybe for many people, but that he also recognizes that economic development is essential as well. So, you know, how you balance those two things at the same time is not clear. And it's something that I think the Party is. Has constantly wrestled with from the very beginning. How you balance ideals and conviction and motivation with being practical and thinking about economics and that kind of thing, you know, in terms of how China thinks about the world, they see real inherent strengths in their system. And even though they're communists and therefore should be material people who only think about the world in terms of concrete, objective conditions, they talk about spiritual civilization all the time. And in fact, they think that China has one, but the west does not. Because in capitalist societies, all they do is care about money and consumption. So in Xi Jinping's mind, he sees those problems only getting worse and worse in the west, as opposed to China, where in his mind, they can tell a good story, a motivating story, but also the party can organize interests in a way that doesn't allow one class or one group to dominate.

How does that actually play out? Because I feel like we. There's incredible amount of wealth still in China. Like, there's still wealth inequality. There are many tech billionaires and massively successful folks over there. Of course, when they get really, really big, they tend to disappear for a little bit. And it doesn't seem like a great place to be a Jeff Bezos type. But at the same time, it doesn't feel like, oh, yes, like they are truly communist and everyone has the exact same standard of living. And so is that even the goal? Are they okay with some level of wealth inequality? And. Or do they see it as a failure?

And even that. Digging a level deeper, if Xi and party leaders can speak to this sort of civilization, you know, the spirit of the Chinese civilization and how great it is and how the system of the west is failing, what do three, you know, five levels below that in. In the Chinese system in terms of, you know, how much do you try to dig in and, you know, what is the average factory worker? Do they feel that same sense of. Of pride in. In the. In the spirit of China, or is there kind of a disconnect?

Yeah, that's a really. Those are some really good questions. So I think for Xi Jinping, he can do course corrections to a certain extent. Right. So this is someone who, a few years ago was really doing a very severe crackdown in the tech sector, and people who made a lot of money were suddenly very worried. Xi Jinping was talking about common prosperity, but over the last few months, they have been talking more about the economy. And it seems that Xi Jinping has empowered his premier to focus on development. And that doesn't mean that they've suddenly stopped caring about security or ideology. It just means that they can move around a little bit without admitting that they were wrong. And I think the reason for that is that the party, when they talk about ideology, they're more careful, I think, than people give them credit sometimes. Right. So, for example, if you look at the latest party congress report, it begins with ideology, and it talks about how we need to believe in this stuff and we still are, you know, socialist and communist. But then it says other things too. It says the reason communism works in our country, but failed in other others is that we synicize it, we made it Chinese, which basically means we made it say whatever we want it to be. Right. And so then you also see other language about markets still being decisive and China still being in the stage of primary accumulation, which means that development, not restructuring society is the top priority. So you know, they talk about common prosperity because they know that they need to tell a good story about getting rid of inequality because they subscribe to socialism. But then you have all these buts and ands, right. So they say, you know, we're not going to become populist like Latin American countries by allowing people to have such a social safety net that they feel that they don't have to work hard. And so for all of these reasons, I think that he, he's trying to manage dilemmas because he, he can't solve problems, if that makes sense.

Yeah, yeah. Have you. Yeah. I mean, we've been talking to people about how some of China's investments in the semiconductor industry are in some ways more capitalist than the way America subsidizes our, like the CHIPS act. Compared to the series of five year plans that have been somewhat cutthroat over in China between a whole bunch of different companies competing for, for grants. And yes, there's a lot of government money pumping into the sector and they've been able to stay at the lagging edge for years. But it's been designed to be kind of dog eat dog and hasn't just been a situation where China's just picked one winner and said, here's $50 billion and go run with it. I'm interested to hear the story about the development of Hong Kong relative to some of the mainland cities. Can you walk me through what that taught about China and Xi Jiangshun? Because I believe he saw kind of that happen. I've seen the time lapses and I've been to Guangzhou once very briefly, and the balancing act there between those various cities and what that kind of taught China.

I'm really glad you've asked about Hong Kong. I've done a handful of interviews and you're the first person to raise it. And it's quite an interesting story and it's revealing in many ways. So Xi Zhongshun, he was persecuted for 16 years, right? And then he's released and the first job that he gets is the party boss of Guangdong, which of course is the province that borders Hong Kong. And he saw physically just how far behind China had become. And his first task that he had to deal with were the thousands and thousands and thousands of. Of Chinese people who were fleeing to Hong Kong, often losing their lives in the process. So it was such a striking physical manifestation of how far the socialist motherland had. Had. Had fallen behind. And so for him, that helps explains the special economic zones which he played a role in establishing. And, you know, he has this idea that one of the problems is that is ideology. We need to make people, you know, believe in the cause. But he wasn't so foolish as to not recognize that you also had an economic angle here that you needed to figure out. Right. And then ensuing years in Beijing, when he was running the united front, he had to figure out how these people in Hong Kong were going to feel about returning to the mainland. So he kept meeting with these Hong Kong people during the handover negotiations, and he would say things like, we're all Chinese. Go overseas and look around and you'll realize that actually Western countries aren't as great as you think. You put your money in a bank and the fees are greater than any interest, that you would make these kind of like, sort of jabs, very conventional communist views of boom and bust and that he. But you know, what's interesting is for them, like in Hong Kong, one of them, during these meetings, a lawyer from Hong Kong said, you know, you and the mainland, you feel like you're coming out of a tunnel because the Cultural Revolution is over and you're reforming. But we in Hong Kong, we feel like we're going into a tunnel because we're. We're looking at unification with you, but we're not sure about what direction you're going. And we're frightened. And it was Xi Zhongshun's job to win their trust and faith to the handover.

Interesting.

Xi is up for reelection in 2028. There's no term limits. What should people be looking out for in terms of the lead up and what he's. Is it the right kind of outlook that he should be really trying to prove his next few years really matter in terms of proving that he's the right leader for the next five years beyond that, or how do you look at the election?

So he is in a situation where whatever he decides is what will go. Nevertheless, he is facing really, really powerful challenges when he looks at the succession, because as his father's story shows, if you read my book, it's hard to think of anything more explosive than succession politics in an alienist regime. And it gets back to what we were talking about earlier, which is that These are very leader friendly systems. The whole purpose is to have a really, really powerful person at the center who can make choices and then everybody has to listen. So Xi Jinping, if he picks a successor, it raises the question of, well, who's actually really boss. And then the name successor will have to figure out how much space he has and whether or not he's actually getting Xi Jinping. Right. And he'll want to impress Xi Jinping, but he also won't want Xi Jinping to think that he is getting too big for his britches. But if Xi Jinping doesn't pick a successor, then he might be afraid about what will happen to the party after he dies. And he will likely want to make sure that the person that he thinks is best is the person that becomes the next leader. So he's kind of stuck. Right, because neither of those options are very good. And he's probably also thinking in his own mind and changing his mind and testing certain people and seeing how well they read him and how well they handle situations. So it really is something to watch. And a lot of it will be about personal chemistry.

Yeah, I'd love to know how hereditary dynasties, dynasties play into that dynamic. And if you could ground it by explaining the concept and history of the princelings and then kind of the current status of the idea of a princeling, I don't know if that's kind of gone out of fashion or if that's still. If, if, if, if that matters today.

Yeah. So in the 1980s, the party as it was trying to figure out how to survive after the founding generation died, they looked at young people and they were a little afraid because during the Cultural Revolution, young people had beaten them up, brought them to struggle sessions, incarcerated them while they were still alive, which raised questions about what would happen after they died. And a lot of them had been betrayed by their own secretaries. And so in the 1980s, a lot of them were picking princelings to work in their offices because they thought that they would be more trustworthy. The problem with princelings is that they were not well liked more broadly within the population and in many circles of the party because they were seen as entitled, they were seen as arrogant, they were seen as benefiting from their parents status. And so they had this weird sense of both entitlement and vulnerability at the same time. And Xi Jinping was a witness to that. And that was one of the reasons, I think, why Xi Jinping decided to go work in the grassroots as opposed to pursue a career that was purely in Beijing. So that he could avoid those kinds of charges. So, you know, Xi Jinping's career was hurt in many ways because he was a princeling, although he also did benefit his father. So it's a little bit of a complicated story, you know, now in China today. I think that Xi Jinping probably doesn't have a great relationship with these princelings because he doesn't want them to think they have special purchase on him because of who their family is. And Xi Jinping thinks everyone should put the party's interests first. So in that sense, I think that princelings matter because Xi Jinping is a princeling and the fact that his background, you know, tells us something about that. But I think many other princelings feel that they don't have much of a say in the direction China's going.

Well, that's the name of the book, the Party's Interests Come First. And thank you so much for stopping by. This is a fantastic conversation.

Yeah, well, thank you. Thanks so much for having you back on. When there's. When there's.

Yeah, yeah. I mean, we'd love to have you back and talk when there's anything that you haven't to announce, but also anything in the news that you could comment on would be fantastic. But congratulations on the book and I highly recommend everyone go pick it up. The party's interests come first. Thanks so much for stopping by, Joseph. We'll take care.

Great questions. Thank you.

Bye. Next up we have Paul from browser base coming into the studio. He's not here yet. And that, and you know what that means. We get to do some ads, baby. Adio. Customer relationship magic. Adio is the AI native CRM that builds scales and grows your company to the next level. Go to ad on audio also, how did you sleep last night?

I. Oh, Jordy. Coca Cola, Modal usv.

I think I got you.

They got range.

I beat you.

How'd you sleep, John?

93. Only 7 hours and 19 minutes. But great on quality. Great.

Back on my.

No.

97.

Oh, it's a disaster.

Clean eight hours.

Oh, brutal. Okay, well, congratulations. You're. You're back. You've been back in the. You had a rough, rough week.

It was about a 10 day period that perfectly coincided with me being.

You beat me 10 weeks in a row. But I beat you for one week and it felt like 20 weeks of victory for me. I was pumped.

That's right.

Anyway, I will let you take the intro with Paul, get the details from him on what's going on in his world in the browser. Based world.

There he is. Mr. Paul Klein the Fourth. Welcome.

Hey guys.

Big, big day for you. Break it down. What's going on?

Good.

You know, Series B announcement today. Of course, you know, Saquon has his touchdown chain. I brought the browser based chain in.

You got the big B.

You got the big B.

All right, well, I'm gonna hit the gong for the big B. If the team go to the wide. Get ready here. There we go. Solid connection for you and the browser based team. Love to see it. Love to see it. So the chain. What? How do you have any sort of like milestones? The chain stays on until 100 million of ARR.

You know, the chain is currently taped on, it falls off. It's from Amazon. So I think that the series D will upgrade to a better chain.

Real.

Okay. Okay. So a couple, couple rounds away. Exactly. Awesome. Who, who participated in the round? I'm sure you had a pretty fun process. But. But break it down for us.

Yeah, you know, we're super lucky at Browserbase. Every round that we've done has been preemptive. So it was a quick one. And that's because we work with people we've known for a long time. You know Glenn Solomon from Notable Capital, he's someone who met with us in the earliest days. And the Notable Team, formerly known GGV Capital, they're just killing it right now. They just did features and labels. Series B as well. And I think that really started the conversation for us. We were like, hey, but sounds like you guys are new and B's, we'd love to kind of have a conversation. Of course, CRV doubling down as well. Reed Christian on our board, you know, Planner Perkins as well. One of our earliest investors at the SEED participating, the A and the B. So we're really excited about this composition. I feel like this board is a good starting lineup and we can go pretty far with it. So we're excited.

Amazing. You guys also had a big launch today, is that right?

Exactly.

Director.

Yeah.

What's going on there?

Our whole thing is that, you know, when we think about how people automate the web and browser base. For those who aren't familiar, we're basically a web browser that can be controlled by your AI, really an infrastructure company. So we sell to developers who are building features that want to go automate the web. And we realize that one of the places that people are starting when they're building their applications is often in these kind of lovable V0 or bolt experiences. They're vibe coding. And we kind of looked at the way Vibe coding applications were set up to help developers who build like, you know, automation scripts like submitting your Delaware franchise tax or grabbing a quote from a website of a supplier you want to work with. And for us, we wanted to build something built for the Vibe coders who actually want to build software that will do work on their behalf. And that's what Director is. Director AI allows you to go to the website prompt and it's going to output this repeatable script that will go to a website, click buttons, fill in forms, download files, all on your behalf. We have a few cool prompts on there as demos. One is like going to Kalshi and checking the poll for will Trump unfollow Elon on Twitter. Another one's like, hey, go to the NASDAQ website and find me all the earnings calls for this week. And then it outputs not only it just does that for you and then also outputs code so you can integrate that into your application and run that every single week in a repeatable way.

George Hotz was actually surprisingly bullish on computer use. He usually pours a lot of cold water on a bunch of different stuff, but he was pretty, pretty optimistic about me being able to book a flight on delta.com with an agent coming soon. What I didn't get from him and what I want to get from you is where does that interaction live? He was saying that yes, it's coming. You'll be able to just say, I'm trying to fly to New York tomorrow. Book me a flight that gets me there in the evening and it will do it for you and it'll interact with the webpage or the API? The question is, where does that interaction live? Is that in a consumer app that I subscribe to or pay for, is ad supported or Is that a delta.com feature? Or back and forth like where does the, where, where does more of the agentic interaction live?

Yeah, I think for, for consumer use cases like booking a flight, it has to live where the distribution is, right? And that's probably going to be Google Apple on device. Your AI will control the browser on your device. But if you think of companies like Ramp, if Ramp wants to go automate collecting an invoice from a website where they don't have a first party connection, the way they can agentically do that is with a web browser going to that website, getting your invoice for you and putting it through there. Bill pay. I'm a Ramp customer, so it's something I've been asking for as A feature for a while. So I think that for consumer automation use cases it certainly will live on device or in some sort of consumer app, maybe ChatGPT. But for the B2B use cases where you're actually using some software and the automation is an extension of the existing software, that's where I think the, the browser and more generalistic computer use that lives within a company like browser base is really going to stand out.

Yeah, let's dig into that, that ramp example more because I've had to do this where I've wanted to classify like an Uber receipt. For some reason I wasn't subscribed to the emails and so I go into the Uber website and I pull the receipt down and then I upload that to ramp. Right with browser base they could potentially do that if they have my login. Long term there's probably some sort of API interaction and so I almost feel like there's this weird world where like you see a lot of people do using computer use and then once something really takes off then they, then they automate with an integration but then there's like a new wave. So is your growth kind of like a series of S curves? Is that the way to think about it or do you think that people will just, just be so happy with the results from computer use that they will just say why would I even bother building the direct API connection?

Well, I think there will always be direct API connections and there should be, you know, if someone is apt to build that, I highly encourage it. I think about like the rest of the Internet, the Internet that already exists. Are we going to rebuild that Internet? The analogy I like to use, and I use in this piece today with Alex Conrad is that, you know, when you think about Waymo, Waymo drives on the roads we've already built and Waymo's got good enough to navigate those roads. It might be more efficient for us to build highways just for Waymo or just for self driving cars to go, go out and browse. But in, in the end, like if AI can use the Internet the same way we can just as well as we can, why not just let it browse the web like we do Short.

High speed rail I'm hearing because when you think dedicated roads for something that's autonomous, that doesn't need to steer, that's a high speed rail. But yeah, why not just be in a, in a self driving car?

How much are you thinking about bigger partnerships long term there will be some, you know, companies that will not be excited about, you know, Computer use agents kind of, you know, interacting with their web properties. And at some point you might be, you know, constantly going back and forth with them, you know, trying to figure out a way to. I don't know exactly what it would look like get, get around captchas and things of that nature. But is that something that over time, I mean, we've had Zach from Plaid on the show many times and, you know, early on they, you know, were sort of doing things maybe manually and then, and then, you know, with software, but then they ended up developing out, you know, meaningful partnerships to enable these sort of interactions to be more seamless. Is that something on your roadmap at all? I can imagine just after you guys have raised over $50 million in a very short period of time, I imagine people are kind of knocking on your door now at this point to have conversations as well.

Yeah, we're certainly excited to partner with the antibot detection companies. In our mind, like Browserbase has an opportunity to be an arbiter of good thoughts. For example, we may be able to help a customer of ours say, hey, we're this small startup. Here's our use case. We've done KYC with browserbase. We have restricted domains that we can only browse to and we could help kind of represent them to the broader antibot community and say, this is a customer that we've certified. And I think for the longest time, Antimai was built because there's only bad bots online, but now there's good and bad bots. And on Browserbase we hope that we can empower, show and prove that these are good bots and use the brand that we built as a center, part of our trust. But regardless, Plaid is an amazing company, right? Really look up to Plaid. It is possible to build integrations for every single bank out there. There is like a finite number of banks or at least like a workable or tractable number of banks. There is billions of websites out there. So regardless, you still are going to have to have some AI for when an agent or a customer prompts your agent, hey, can you work with this specific website? You might not have an integration built. And browser base is kind of that primitive that, you know, the integration of last resort. And we hope that if you're building an agent, you include a browser tool more for like the what if it happens, not for the when it happens.

How do you think about competition coming from the hyperscalers and the big cloud platforms? Companies like Stripe and Plaid haven't seen much competition from GCP Azure, aws. But on the flip side, some of the other parts of the AI stack have been either cloned or offered or vended in through some of the bigger cloud platforms. Is that a risk or do you have an idea for a moat that is too high for the king of moats, Jeff Bezos to clear?

Well, I think it starts with a few things. One, I think you have to build a great developer product and that comes down to actually having a dashboard features that people love to use and work really well. Jeff Bezos is the king, but I don't think any developer really raves about the AWS console.

Right.

And secondly, you have to have more than just infrastructure. Infrastructure always kind of becomes a commodity, right?

Sure.

So moving up the stack one layer, that's what we did with Stagehand. Stagehand is our browser automation framework. When you build it, you kind of write code and it generates the the integration. Sorry, when you write Stagehand code it actually generates the browser automation functionality in browser base. So it's like one layer up. The analogy here is like Vercel and Next JS Browser base and Stagehand. So when you own the framework you're really able to offer more stickiness and just build more first party integrations into the framework that run better on browser base. And then finally Director is at top of that stack. Right? Well if you have a framework, you probably want to have a way to generate that framework code. So we moved one layer higher to the application level and I haven't seen any incumbents really do well at building developer level around new frameworks and then one layer higher like building applications to generate them yet. So for me I'm kind of betting on innovation here. I'm betting on us expanding our portfolio and there's this Parker Conrad tweet which was like hey, this is going to be one on the field who can build it better, who can sell it, who can market more, who can innovate more. And we're ready to go battle on the field with our new fundraise in.

The field and the bathrooms and the slack corporate athletes. It's fought out everywhere. We love it. Talk to me about the good bots versus bad bots, dynamic trends in the robots Txt. Is there a need for some sort of like, I don't know, like almost universal like trust rating or something that identifies you as a good bot as you go around? Because there's going to be lots of anti bot protections. And yet I imagine that having some sort of like coming from a trusted IP address is Something that gets you out of a lot of cloudflare captchas and, and I imagine that, that there's probably work that you can do almost on like the business development side to, to kind of grease the wheels as you try and automate more and more of the web. Is that something that you're thinking about? What, what are the different approaches that you can take to make sure that as the anti bot protections get more and more robust you don't get bogged down in that and you're still offering a good experience to developers?

Yeah, it's a great question. One of our investors, Jeff Lawson, he, he always frames great infrastructure companies as kind of being one of three things. Yeah, one is like APIs representing capital. So it's like hey, if I want to deploy a thousand servers I don't have the capital stand that up aws, I can just deploy a thousand servers. The second one is algorithmic complexity. So you know your inference as a service, right? You have a bunch of inference, you know, how can you run that more efficiently than anyone else? How can you serve it in an innovative way? And then the third one is really biz dev as an API and I think browser based long term partnerships are extremely important to us and that's why we're really proud to work with companies like Work OS Clerk Stytch and Okta which all have major capture presences online because that's what protects authentication. Secondly, I think the credentialing actually happens when you log in. Like that's how you show you are who you are. And I think solving Agent Auth, which I think these authentication companies I just mentioned will solve really allows our customers to say hey I'm Paul, I'm logging into my United account and this is my agent. But it's acting on behalf of me because I've authenticated it. I think that's pretty compelling. You know a lot of the reason antibodies this to prevent, you know, account takeovers or spam but if it can authenticate and say hey I'm acting on behalf of Paul, I really think that's a good way to show proof of ownership or proof of humanoid of a bot and that's going to be solved very soon. It feels like there's a bunch of stuff happening in Agentic Auth that is very, very promising and that's going to be great for browserbase and we're happy to partner with all those companies to make it happen.

That's fantastic. Well, congrats on the big round and thanks for stopping by. We'll talk to you soon.

Always welcome. Congratulations to you and the team. Excited to watch you guys work.

Have a good rest of your day.

Cheers.

See you around. Appreciate it.

And really quickly, let me tell you about AdQuick.com out of home advertising made easy and measurable. Say goodbye to the headaches of out of home advertising. Only Adquick combines technology, out of home expertise and data to enable efficient, seamless ad buying across the globe. Our next guest is Alex Cantrowicz from Big Technology. We love Big Technology and we got.

The let's give it up for Big.

Tech in the studio. Welcome to the show. Thanks so much for joining us. How are you doing?

Thanks, guys. I'm doing great. Great to see you. Thanks for having me on.

Would you mind doing a quick introduction on kind of how you frame yourself in the giant pool of big technologists and big technology folks?

Definitely. I think it's fairly straightforward. I'm an independent journalist, so I used to be working within newsrooms. In 2020, I quit Buzzfeed News, started my own thing. I've been writing the newsletter for five years and doing big technology podcast for just underneath that.

Amazing. And I mean, this is just, we can go into a bunch of other stuff, but I want to talk like the meta level, the inside baseball for a little bit. Do you see kind of a meaningful difference between being an analyst and being a reporter, an investigative journalist, someone who gets scoops versus someone who provides op eds? Do you think that's a meaningful distinction in kind of the modern independent journalism era or the going direct era or anything like that? We see ourselves not as breaking news here. We see ourselves more as like a reaction and opinion show and in talk show or anything like that. But I'm interested to see how you think it's evolving and then how you see yourself in the ecosystem.

Well, it's so interesting because the definitions, they sort of blur. And I think if you do this for long enough, you're going to realize that you're a little bit of each. Right. You're going to want to get interested in a subject and then just make a bunch of calls and, you know, maybe you'll find something out. And now, next thing you know, you're an investigative journalist or you write a piece of analysis, now you're an analyst or I had a situation where I interviewed Blake Lemoine because I was interested in this guy who said that Google's AI was sentient and Blake, Blake is late to the podcast interview and I'm like, what's going on and it turns out that while I was waiting for him, Google was firing.

Fired. Yeah.

Correct. So I said, mid interview, hey, Blake, this is news. Can I like get a comment from Google and publish it? And he goes, yeah, sure, go for it. So then I write to Google mid interview. We published the story on big technology because I have a substack.

Yep.

I wake up the next morning and it's a story in the Journal and Bloomberg and the New York Times and a bunch of other places. Most of them said as first reported in big technology. So then I'm a reporter with scoops. So I think that, you know, I think the key to this stuff is are you present? Are you curious? Are you going to try to serve your audience? I know you guys are doing this, so you're probably, you know, in all these buckets as well. And I think it's one of the cool things about doing, you know, reporting or news or information on the Internet is you get to play, you know, in all these different buckets.

That's fascinating. Yeah, I had, I had not thought about it that way at all. That's fascinating. Anyway, let's move on to the big news. I'd love kind of your reaction and breakdown from wwdc. We talked to a lot of folks about it and I, I mean, obviously a lot of mixed reactions, but I came away kind of optimistic that we could be going into this era of like a new explosion in AI apps with leveraging on device inference. And I was extremely optimistic about it, but there's a lot of other narratives going on. What was your takeaway?

Yeah, it was really interesting. So I was there, I was on the broadcast riser because I was doing some stuff with CNBC and I have never seen a more different event one year to the next. Last year, super enthusiastic, a little bit more open, big vision setting with Apple intelligence. This year I think they spent this is conversation from a conversation I had with MG Siegler last week. They spent more time talking about the phone app than they did AI. So it was clear that the expectations have changed for Apple and they ran into a lot of difficulty building AI in the same way a lot of companies have run into difficulty building AI because I think anyone who builds with this technology knows. I'd be curious to hear from your listeners and viewers if this is the right perspective. But it's probabilistic. It doesn't act the way that you want to all the time. If you're a company like Apple with a very high quality bar, it's not Very easy to wrangle this technology. This is why I think that they should buy Perplexity because they can just integrate an AI search engine into their products and have that control that Apple wants. So we definitely saw a very big change last year to this year and the focus was really on this new design, Liquid Glass that they've brought into iOS, which is starting to roll out in like the developer release. And with a bunch of fixes we'll all get access to it at some point soon. But it was definitely, you know, a very subdued event for this company. And it's interesting that I'd actually be curious to hear why you think that this on device development is the thing that's going to lead to an explosion in AI apps. The conventional wisdom from the people that I speak with about this is that you can already build fairly fast applications using the stuff that's available today and you can build much smarter applications with these larger parameter models and this very small model that Apple is making available on device. So I didn't really find that to be a needle mover. But I'm curious to hear why you think it is and what your audience is telling you about it.

Yeah, I mean, I think there's still.

Like, I think there's this hangover from war stories where developers created a cool experience, released it and then incurred massive costs from it going viral. Because a lot of these generative AI features are naturally so exciting and novel that people just have this excitement to use them and that was leading to some meaningful costs that independent developers. Historically you could just spin up a simple mobile app or SaaS tool and not be thinking about what your cloud bill was going to look like. Maybe you were even just had credits and it and you weren't going to incur it at all.

It's the kid without a credit card that is this super long tail of developers. They're not super mature, they're not venture backed, but you get a ton of those random solo developers that don't even want to think about incurring costs and they want to build the next flashlight app, the beer app or Flappy Bird, these types of all these different apps that kind of start as little demos and then maybe they grow into something with proper infrastructure and a back end. But the more that they can leverage the free out of the box tools, the faster they can get going. And yes, eventually they'll raise money. Eventually they might switch to, you know, cloud models that are better and more expensive and they need to think about the business model there. But Being able to use AI in an app that lives in the App Store uses App Store viral distribution. Truly zero cost seems like it unlocks a different tier of entrepreneurship or app development that could be a really, really interesting canvas to explore. It's kind of like when the, when the original AI video came out, there was that one viral video, Harry Potter Balenciaga. And it's like that's something that, that could have been done with a traditional CGI pipeline. Like you could have gotten ILM or Pixar team to go and make that like it was doable with the previous technology. But when you lowered the cost from a million dollars to do a, you know, a 3D face scan of Dumbledore and, and put him in Balenciaga or hire the actors and shoot it with cinema cameras, when you lower that to just a couple creative prompts, like that human creativity part comes out and you get a viral video Harry Potter Balenciaga. And the same thing I think could potentially be on the verge of happening with viral apps as the overhead and the cost to develop software drops with products like cursor and GitHub, copilot and windsurf and Codex and Devon and as that drops and then also as the ability to call AI inference on the device, we're already seeing it. There was a cool app that went out on X earlier today or yesterday. Somebody made the Tom Riddle, the Tom Riddle iPad app. So you can write hello and it kind of dissolves into the UI and then script writes back to you and that's all generative text. And so it's just a simple fine tuned prompt. It doesn't need to be, oh, I am open gold medalist, like amazing math.

Like not solving level, you know, it's.

Not leading benchmarks, it doesn't need to lead benchmarks. It just needs to be conversational and with an interesting ui. And it's something that this developer was able to build pretty quickly. And for like a young Harry Potter fan, that's going to be a magical experience for, you know, maybe a couple minutes, maybe an hour, maybe, maybe it doesn't turn into the next big thing, but you'd get a million of those in the App Store and then one of them, you know, develops and develops and become something really, really cool. So that's why I've, I've been very optimistic.

Let me give the counterpoint, which is that with the biggest models that we have and the such powerful models that we have, we haven't yet seen a, you know, critical mass of hit apps that you would say, oh, if we could just bring the cost down, we would get an explosion. But you have opened my mind. I, I'm willing to, to say that this is definitely something that I hadn't considered where you could have the sort of low lift fun apps that don't, you know, make the developer go broke, that maybe were cross prohibitive to build beforehand that we might see. So, yes, let's, let's talk about this in a couple months and then we'll see. We'll see what happened.

Yeah. So, I mean, I do agree with you. And the example that I think would be worth digging into is, do you remember that app Lenza? It was this magic avatar app. You download the app and you take like 10 photos of yourself and then it can generate like an AI image of you. It was kind of a precursor to the Studio Ghibli moment. Yeah. And they were extremely good at monetization because they'd be like, yeah, we're making the image. Do you want it? Buy this $20 in app purchase. They did very well. It was a little bit of a flash in the pan. I think it went, it went really big.

And then they haven't shipped an update since December 20, 2021.

Yeah, it didn't really go anywhere, but that felt like an important shelling point for, for, for generative AI imagery. And then we saw it again with Studio Ghibli really taking over the Internet, I feel like we haven't had that in text yet. There hasn't been that, like, where's the Harry Potter Balenciaga? For something that was just purely text generated. But, but, but I think that it might be a cost issue. It might be something that people just haven't had a chance to, to, to play with at low enough cost. And I think that anytime you, you, you reduce the cost, like, you know, gps, you drop the cost of that really, really low and you get Uber. And I, I'm just still optimistic that, that there's a, there's a, there is a binary difference in, in.00001 cents and zero cents. Like, I hope right is actually.

Yeah, I'm smiling because I'm remembering this lens a moment. And there's been this, like, it was the start of people being like, don't upload your face to AI because then they'll have your, you know, your biometric data. And I remember in the middle of the Studio Ghibli thing, there was a tweet that came out like that where someone was like, you know, be careful about uploading your face to chat GPT and then someone took their Twitter avatar and of course it was like, all right, well you know, no good deed goes unpunished, I guess.

It's so brutal because the company that did Lensa pivoted to something called Prisma, which was a photo editor with like, cool.

This is a, it's a Russian company.

So it's cool artistic filters and they basically built the Ghibli type experience.

And then I think they were actually Prisma before because I think I used.

Yeah, that's the name. Prisma Labs is the name of the.

Company and it was never quite there. And then OpenAI just kind of leapfrogged and I think that will be a dynamic that happens a ton. I think that there will be kind of solo developers, kids that build a magical experience like the Flashlight app and then yes, Apple will steamroll them. But that solo developer, that kid will probably make a ton of money with like a $2 in app purchase just for upgrades or throw some ads in it or something, make some initial money. But even more important, the experience of having like a hit app sets you up for, okay, the next go round I'm ready to raise money. I'm ready to think about the seven powers or the moats that I should build around my business. Maybe I wind up going into B2B software or infrastructure or something else. But they've caught the itch of entrepreneurship because they've had that taste of like I built a product that had an impact and that's something that I love. When those things get unlocked at an earlier, earlier stage.

It's a very cool idea. I mean I've had this idea to build this generative AI app where it's like a choose your own history where you can either play through these scenarios because you can do this, you can like prompt quad today and you can do it today where you could like be a historical character. Like let's say you wanted to be like Alexander Hamilton during the American Revolution and sort of like experience that as he did, quote unquote. Right. But doing it through gen experience, you can do that. I think the problem has been that the models haven't been powerful enough for the thing that I want to build. But I, I think that you guys are onto something. Like if this is a moment where Apple, you know, Apple might start with the smaller model, let's see what happens. Maybe you get the equivalent of that beer drinking Apple. If that works out, then there's a lot of incentive for them to put Much bigger models within the operating system. And then you could really unlock some very cool things. And it does make sense that if you end up not going broke by building a gen app that becomes popular, you might want to build another one.

Yeah, I actually had this exact experience. The historical character on Character AI the app where you can chat to a fine tuned LLM on a particular person. I picked Stalin.

And what does that say about you? No, I'm kidding.

Well, I'll tell you. So I was like, I want to go and argue with Stalin about communism and I want him to play the steel man, the literal steel man in his case. And, and, and I, and I want to, and I want to have a real hardcore debate about what Stalin did, my perception of him. I feel like I have a lot of good ammunition to fight back against his ideology. But let's go have this debate. And so I started debating with this character AI the Stalin. But it had been so rlhfed and so fine tuned on like American values that Stalin would come to me with something like, yes, like, you know, I did a lot of bad things and I'm like, sorry about it. I was like, I don't think Stalin would talk like that to me. I think Stalin would be very proud of what he did and not admit fault. And I was like, okay, character, you clearly didn't fine tune this enough because it's still, it's still rejecting, you know, the real, the real communism, real communism in character.

Maybe he slept with it and realized that, yeah, you know, maybe I did get a few things wrong.

That's not the Stalin I wanted to debate.

The genesis of my idea was just dumping Wikipedia pages into, I think it was Claude and basically playing these scenarios as like maybe somebody who's not like in the power position in history. Oh, that's so I always find it interesting. So my wife is European, so we've like visited a lot of like historical sites in Europe. And I've always thought like, well, what happens if you're not like the Lord or the lady that they feature or the king that they feature in these museums? Like what if you're just like somebody on the line and what would your life be like? So for me, because I'm a fun guy, I dump Wikipedia pages into these bots and say, well, what would that be like? And then you can role play through it. And I think that if you want to, if you find a way to like prompt these bots to give you like the, the non sanitized version, it can be really Interesting and. I don't know, fun. I don't know. Fun is the right way to put it. But worth. Worth going through these scenarios and playing the games.

Yeah. Yeah. I mean, totally. It's a great idea. Somebody will probably build it. But even right now, it's like, even, even if they would get you to pay, that's an extra hurdle that causes conversion rates. And there's a million different things that, that, that kind of downstream from that. Even if it's just writing extra code to make sure that you're processing a payment in order to pay for your cloud bill and your cloud bill. So I'm excited to see what happens. I don't know, we could see how it plays out. Can you give us some other.

Yeah. Can you give us an update on the App Store? There's just been a lot of back and forth between it's Tim Sweeney, right over at Epic.

Yeah.

And Apple chirping at each other. What, what is the state of the App Store? Are alternative, you know, payment rails getting, you know, real adoption yet? Can you give us kind of a broad overview?

Well, we're super early and there might be some court cases to still work through in terms of this alternative payment on the App Store. But basically what Tim Sweeney and Epic Games are trying to push through is Apple. If you want to buy stuff within apps, you have to use Apple's in app payments. And when you use Apple's in app payments, Apple gets a cut, a sizable cut of the money that you spend. And that is very important for their services business. And just to take a step back. So basically Epic is trying to say that that should be illegal, whatever it is monopolistic or anti competitive and we want to be able to process the payments on our own or via the web and not Pay Apple the 30% or whatever it is from every dollar that goes through our services. It is. So I was going to take a step back because it's just happening in this very interesting moment. So if you think about Apple's financial position right now, almost everything is either flat or shrinking. The iPhone revenue 2024 versus 2023 flat. Maybe it grew a tiny percent. But like you look in there, 10k or whatever it is, and you see the, the, you know, the growth or the loss is just the dash flat iPhone revenue. And by the way, that's happening as iPhone sales decline. And the only reason revenue is flat is because they're getting a higher average selling price per iPhone. So Apple is in sales decline with the iPhone right now. And that's happening as people spend more time with their phones and as people in China are less interested in buying iPhones because Huawei has become this object of national pride and it's almost as good. So they're starting to buy Huawei's and by the way, they're not locked into the Apple ecosystem because they use WeChat. So the company is, is seeing flat sales revenue on the iPhone. They're seeing declines in things like wearable on the iPad. I think MacBook grew about 2% last year. That's the only like traditional Apple business category that grew. But there is the services category that grew 13% last year. And what does that include? It includes a number of things under threat. That includes those in app payments that you just brought up. And it also includes this 20 plus billion dollar payment a year that Apple gets from Google which is also under threat because there's this federal case going on in D.C. that Google has already lost.

Whoa, whoa, let me stop you. It also includes Apple tv. Let's talk about Apple one. Let's talk about some movies they're making. Let's talk about all the stuff that they highlight.

I'm not here to be super negative, not here to be super negative, but let me be realistic. Yeah, if that Google payment in 2024 went away and like let's say that was 20 billion, it was probably more but let's say it went away in 2020 24. Any, any thoughts about what happened? All right, go. Apple as a, as a company shrinks services. The one, the one division growing double digits, growing 13% actually contracts now it's a one time hit. So you would go down and then you would grow from there. But the reason why Apple is in the 3 trillion dollar range is because it has this services division which investors will value as more of like a software company. So there you can get into like the, the 30 or the 35x multiples versus a hardware company which is.

Well the other, the other thing you didn't mention is margin imessage has also opened up pretty dramatically too where you have red receipts. You know between Android, you guys have.

Seen the Android folks typing and been like what's that? Or seen the red receipts. Like why is the green bubble doing the typing thing starting to happen?

So you're right, I think gold bubbles are coming. The OpenAI phone is coming. He's going to pick a different color. Trump phone will have gold bubbles for sure. The gold bubbles are coming. Talk to us about gold Google. You interviewed Sergey Brin recently. What what, what was your read on his involvement? Because I feel like there's a press cycle every few months about like he's coming back. He's, he's, he's, he's back in founder mode. He's deeper than ever. He's not fully stepping into the CEO seat, but they're clearly stepping up to the AI moment. At the same time, they're dropping a ton of product updates. Google I O was a massive event with, with like a ton of different features. And then they're also at the Pareto frontier of so many AI models on, in terms of performance and cost, and yet a lot of stuff's not breaking through. What's your read on Google right now?

I think they are. They knew they were behind on AI. I mean, basically they had, they developed. This is going to be old history for a lot of people, but for those who don't know they developed this transformer model within Google, they were very safety concerned. We talked about Blake Lemoine a little bit ago. He was using a version of their LLM chatbot called Lambda that he believed was a person. So they had this advanced technology working within the company. But OpenAI, of course, was first to market with something that exploded with ChatGPT. So once I think Sundar Pichai realized what was going on, he said, okay, we're going to turn all of our focus toward, you know, this AI moment, building large language models. And he was lucky or fortunate or he planned whatever you want to say. I mean, the company has Demis Asabas, the head of DeepMind in there, and he consolidated these two divisions, DeepMind and Google Brain, which had traditionally kind of struggled over resources and realized they needed to coordinate. So Demis takes over both of those divisions and now you see that Google is shipping like crazy with AI and they have some great. I mean, they obviously previewed some very cool things@google.AI. oh, some great features working already. I mean, Notebook LM is really interesting.

Yeah.

I think Gemini could work a little better in places like Gmail, but we'll probably see it, like, being able to search your email is a pretty big deal. And do like, things like I sometimes will say, like, pull out the emails from all of big technologies paid subscribers recently to. So I can invite them, you know, to our, to our private discord. And Gemini gets that right about 50% of the time. So I think it's clear that. What do you expect.

Yeah, yeah. What do you expect to see out of the Google OpenAI relationship? Right now we're seeing tension between Microsoft and OpenAI, they're partners. They maybe both are unhappy with how the deal has worked out. I think all this stuff is evolving, but in my view, over the next five to 10 years, I just see this massive war and battle for the consumer between OpenAI and Google. Is that how you're looking at it as well? And do you expect to see more attention there?

It probably will happen. So just to. I'm going to answer that question, I'll just finish up what I was saying, that Google's in this moment of reinvention, so they're going to move from search to AI. And also I think Sergey is really involved on the technical side of things to answer that part. So the partnership that Google and OpenAI did is for basically server space. I think OpenAI is really running out of GPUs, and of course they have this hosting deal with Microsoft that's kind of like up in the air. We're not quite sure where we're going to go on that front. And you see Sam Altman every couple days saying, all right, we're out of GPUs, stop burning our servers. He's joking, but it's a real issue for the company. So of course you look to Google, which has the TPUs to sort of offload some of that, some of that sort of compute need. Now, we don't see OpenAI being made, unless I'm wrong here, we don't see OpenAI's models being made available within Google Cloud. That's the other side of the equation. I spoke with Google Cloud CEO Thomas Currie and a couple of months ago, and he said we'd welcome OpenAI on the platform. So that's another step I think that we should anticipate. And then on the consumer side, I mean, again, like, and you guys know this better than anyone else, like, Silicon Valley is filled with frenemies. And we have another situation here where you're probably going to have Google, you'll have open AI. We already have it. Open AI relying on Google infrastructure, but competing with them on, let's say, the assistant front or the chatbot front. And I think that's one of the cool things about watching tech. And for those who are in tech, being in tech is there is almost, you know, reliance on one another's innovations and products, but also a recognition that everyone is working to build the best products and may the best product wins, win. And, you know, one product may be the king one day and then it could get leapfrogged the next and build on top of each other. So I do see this being a heated battle and we're going to see a lot more drama, I anticipate from all these companies coming up pretty soon. But it is notable that now OpenAI and Google have opened up the line of communications, line of communication, and maybe more is coming.

Last question.

No, I'm excited to see all this play out. My last question, when can we place OpenAI in the category of big tech? Are they there yet? Does $300 billion valuation, does that do it? What's your framework?

It's so interesting because yesterday I was seeing people on X talk about like, whatever happened to Fang, right? It was Facebook, Amazon, Netflix, Apple and Google and people are calling it Mango now. You guys saw that like meta Amazon jokes video.

I just like Mag7. Mag7 works.

But we're hoping, we're hoping that, you know, YC can make enough startups so we get the, the Mag70. See, you'd like to expand it.

Yeah, so I, I would say, look, I'm not, I'm not a Mango head. Like the O and Mango is open AI. I'm not a Mango head yet. But it certainly seems like they're on the trajectory. So I, I'm, I'll put my order in and, and we'll wait on it for a little bit.

Now, how much have you dug into the innovator's dilemma with regard to Google? There's this whole story about like, Sundar Pichai hasn't read the book and does it matter? And Ben Thompson says, well, it doesn't matter because it's structural. But the thing that I keep replaying is like, is like, what if Google had actually just released that Blake Lemoine chat bot? Like, what if they were the first mover in chat bots before ChatGPT? It feels, it feels like that Chat GPT moment would have been theirs and then they could have compounded off off of that and it would have been fine, but maybe they'd be disrupting themselves. Do you have any, any thoughts on, on that?

I mean, it gives more credit to Chris Pike's thesis around top down versus bottoms up innovation. But anyways, Alex.

Yeah, so, I mean, I wrote this book that came out in April 2020. Don't recommend releasing books at the first wave of a pandemic, but it is sort of what launched big technology for me. And it's called Always Day One. And the idea is, comes from Bezos, that the reason why the big tech companies have been able to stay competitive and dominant for longer than the Average, right. The average company on the S P 500 lasts 15 years today as opposed to 67 years last century. And the reason why these companies have been able to sustain their dominance is because they reinvent and they don't really care too much about their legacy business. They're able to say what is going to take me into the next generation. And sometimes they're late. I mean, Microsoft is a perfect example. They held onto Windows forever. They had the lost decade. And then Satya Nadella, who ran the server and tools business, which was all about on prem servers, said, you know what, let's do this cloud thing. And now they're where they are today. Google had maybe a lost couple years, but I think they've definitely gotten with the program. Like I mentioned, they're in reinvention territory right now and I think you could really tell the difference between what they were doing and what Apple was doing at wwdc. Although some might disagree. But they're all in on AI right now. And I think that you don't have to be first. You have to be all in and you have to nail it. And the book isn't yet written on where Google ends up after this because it does threaten search. But I think they have as good a chance as any given like we talked about the compute, the talent and the data to be one of the leaders, if not the leader in generative AI.

Well, hopefully you're the one to write the book.

That'd be cool.

Thank you so much for stopping by. Yeah, we'd love to have you back. This is a fantastic conversation.

Definitely great having you. Alex, fan of your show. Thank you for having me on.

Have a great rest of your day.

Forward to the next. Cheers.

Bye.

Take care guys.

Next up we have Saquon Barkley from technology investor coming in from the Philadelphia Eagles. Let's bring him in the studio and we will talk to him about everything going on in his world.

Welcome to the show Ramp day.

It is Ramp day.

Wouldn't be ramp day without Saquon.

Hold on. We are going to do one ad first. If you're looking for a luxury watch go to getbezel.com your bezel concierge is now available to source you any watch on the planet. Seriously any watch. We are working to bring in Saquon and do we have more news to talk about in the meantime time we can also talk about wandering find your happy place book a wander with inspiring views, hotel grade amenities, dreamy beds, top tier cleaning and 247 concierge service. It's a vacation home, but better. We are working on that. Oh, I think he's here. Welcome to the show. How are you doing?

What's up, guys? Thanks for having me.

Thanks for jumping on.

Been looking forward to this.

Congratulations on a massive day in ramp world. We'd love to get your story. How'd you meet Eric Gliman? What's it been like to work with him?

It's been great. It's been great. The conversation and introduction kind of came from my manager, Cuz, who's been super helpful with me in learning and exploring more in the tech world. But getting to know Eric and being involved with Eric has been amazing. He's become a really good friend. My favorite thing I like to tell people about Eric is he has this contagious smile. He always has this big smile on his face.

He really does.

Every single time you're around him, it makes you feel welcome. And just him and his team, like, they've been doing an amazing job and excited for the news that they announced today.

Yeah, it's huge. Talk to me about the super bowl ad. We have your picture from the super bowl ad up on the stream right now. How did that come together? How fast was that process? How was the shoot?

Yeah, we've been talking to ramp for. For a while, the Ramp team. For a while. I had dinner with Sam and Max and Pa. Oh, yeah, A couple of them. They would come to a game, and I forgot. It was maybe like 10 days, a week or two before the Super Bowl. My manager Cuz asked me about this idea of running an ad. Usually, you know, I probably would say no to commercial 10 days before, but the ramp team, they're super efficient, and, you know, they made it so easy, and they were so helpful for me. And, you know, they obviously knew I had a lot at stake and wanted to make sure that was the main focus, but they did a great job. I think the ad came out amazing. They also showed me that, you know, in past these brands that I worked with, they probably took too much of my time. When you're able to do this in under an hour and create something so special and that fans love, but the whole team, the whole ramp team is incredible. Kareem, Sam, Max, all of them, they're all special people, and I'm excited to be able to work with them.

What was the reaction from the rest of the team to see you in an enterprise software ad in the Super Bowl Bowl? Was that surprising or was everyone just kind of excited to see you take.

That step Yeah, I think it was more. Just exciting, more excited. I've always, you know, always talking about how can I do. How can I be creative in new ways? And, you know, we always talk about financial freedom and evolving yourself with the best companies and the best founders. And, you know, it's been pretty cool just so far, my career, all the great things I've been able to do, but the platform that has been established because of football, but things football is able to bring me into and the world is able to bring me into, and I don't know all this stuff yet, but continue to learn in this space and continue to meet extremely incredible people, incredible teams.

Yeah, it's an exciting time. I want to talk about lessons from popular books. A lot of people in tech have read the Score Takes Care of Its itself by the 49ers coach and borrowed ideas from football and taken them back to business. You. You read 0 to 1. What is a concept that you took from 0 to 1 or you think other football players could take back into the world of football from that book?

That's a great question. Peter talks about, you know, when you're starting a company, you never want to have two people doing the same job. And an interesting thing that Nick Sirani did this year, well, my first year, I'm pretty sure he did it in years prior with the Eagles. My first year with the Eagles, he had all of us in a meeting and let all of us know what our role was and what was expected of us. Not saying that our role can expand, not saying it can't increase, but what's needed for us to be able to go out there, compete, and accomplish what we ultimately want to accomplish in Super Bowl. I think the things you could take is just from.

Might be losing you.

It's a right to buy in knowing what you got to do, knowing what you got accomplished. Sometimes your role is not doing what the team asks you to do to be successful. And it takes a great team. And you guys hear me?

Yeah.

Yeah, you're back.

Yeah, we kind of dropped out for a second, but we're. We got most of that. Yeah. Can you hear me?

Yep, Yep.

There.

Oh, sorry about that.

No worries.

Can you hear me?

Yeah, yeah, we're good on this end, I think.

Yep, I can hear you guys.

Okay, cool. Let's move on to another question. I want to talk about similarities between the offense, defense dynamic on the football field to, you know, these two units, they aren't necessarily on the field at the same time, but they both need to perform there's some similarities to business where maybe the sales team wants to go and sell a product that the engineering team needs to. To go and deliver and make. There are two. Two teams that need to be working somewhat in concert, but they're doing very different roles. What can you tell us about the team building that goes on between the offense and the defense to keep the whole group working in concert?

Yeah, it's super important. Team is the most important thing when it comes to. Can you hear me?

Yep.

Yep.

Hello?

Sorry.

I think I'm in a dead spot. I apologize.

It's okay.

But I think teams. I think team's the most important thing, especially in my sport or just kind of any profession, to be honest. I think the parallels from business, from science, from sports, all those things, you know, have similarities, and it all comes down to building a great team. So we know how important it is for, you know, on the offensive side for us to accomplish what we need to do and establish a line of discriminants. But the same thing on the defensive side, establishing line of scrimmage. A lot of games in football is won up front, and it takes all phases, offense, defense, and special teams, for us to go out there and accomplish what we want to do. And when you look at it in any profession, that's why I say the parallels are so similar. You need a great team. You need people buying into the roles. You need people doing all the little things it takes to be successful. And that's why you have, you know, successful teams and you have, like. I don't think I can name any individual, whoever you think is successful, whether it's a billionaire, whether it's an mvp. It takes everybody. It takes a family, it takes. It takes a culture. It takes a village to be able to have that, have that success.

I want to let you go because obviously you're very busy. Last question for me. And if Jordy has anything, he can ask, too. Tell me about Big Dom, the chief security officer. What's his role in the organization?

What is Big Dom's role in the organization? I don't even know if I have an answer for that, to be completely honest. Like, Big Dom, I know he gets a lot of love, and, you know, ever since his. The situation with the 49ers on the sideline, you know, he's. I go to events and, you know, people asking for my autograph, and Big Dom is probably getting more requests for autographs and pictures than me, to be honest. But to be like, the best way I can put it, he's just the glue to the team. And I'm not even exaggerating when I say that he is the glue to the team. He keeps everything in line and he makes sure he has everybody's back. And that's one person that I know outside of my family member and friends, that if I need anything and I need to make one phone call, I'm picking up the phone and I'm calling Big Don't.

I love it.

That's amazing.

Jordy, you have anything else you want to run through?

Yeah. I wanted to ask General how you and friends and maybe other players are thinking about investments today. You've invested in Ramp, you've invested in Andurolle. These are some of the top companies in Silicon Valley. In Silicon Valley, a lot of people will start a big company, sell it, and then make a series of silly investments. How do you avoid maybe making. How do you avoid the bad investments? What's your kind of framework for that?

Yeah, how do you avoid making the bad investments? That's a great question. As an athlete, you get a lot of things thrown at you, to say the least. And you don't know what's the right thing to get involved in. And you just gotta try your. But try to educate yourself as best as you can. And to be honest, when you come from the NFL, a lot of us go to college. Go to college for three years, maybe four. And you know, our financial, like us, how, how educated we are on the financial side of things, it's not that great. And I'm open and honest and I'm able to admit that to myself that coming out the NFL, it wasn't that great and it still has so much room to improve. But the way I handle is by surrounding myself with the right people. And you have to trust people. And one, I have an amazing manager because. And he's pretty good at saying no. To be honest. He. He would say no to pretty much everything. Unless Zach Franklin calls. And when Zach calls us, anything, anything he say is kind of like, oh, yeah, let's, let's just do it right now.

That's right. That's right. Let's just go right.

But it's guys like that, like you, you just surround yourself with people like that who, who are super smart and super successful and done it the right way and you're able to build relationships with them and you'll be able to earn. You to earn their trust and you're able to trust them and make those decisions. So it's tough because you get a lot of Things thrown at you. And I don't know if I have a. The exact perfect answer, but that's a.

Pretty good answer if you ask me.

Yeah, right?

Yeah, that's a great answer.

I think a lot of people in the tech world have spent decades investing professional investors, but they should follow the same framework. Say no to those things, and then if a certain person calls you, you just say, yeah, you back up the truck.

That's amazing. Amazing. Well, thank you so much for stopping by. We'd love to have you back. Hope to see you soon in person.

Yeah. Congrats on the markup, too. First of many, I imagine.

Thank you guys so much. And I apologize for the.

No, you're all good.

I'm leaving. I was coming back from an event at a hospital, so.

Oh, man.

I might be a little shaky. I apologize. No, you're good for having me.

Yeah, we'd love to have you in the studio soon. Yeah. Well, we'll get you under the lights here for. For a couple minutes. We can grab some of your time because this is fantastic.

This is great. Have a good one. That's the best investment framework I know. Multiple, smart, very smart people that have. Basically.

That's amazing.

If it works, you know, it works.

Wait for that one person to call. That was incredible. Anyway, we have our. We have our next guest coming up in three minutes. In the meantime, let's run through some timeline.

Yeah, we got.

What else do we want to talk about?

Yeah, I mean, the big thing we.

Didn'T really go into developing for the.

For the people monitoring the situation at home. Apparently President Trump is warming to the idea of using US Military assets to strike.

I thought you were going to make a joke about monitoring. The real monitoring of the situation is the Microsoft OpenAI situation, because that's the one that most people in tech are monitoring on a daily basis. But we do need to cover the geopolitics. What's the update there?

Yeah, apparently Trump. Trump has been warming up to the idea of military action against Iran, so much so that the poly market is spiking up to 75% chance of military action against Iran before July. So a couple of weeks left now.

There would be a difference between military action and boots on the ground. And I think that there's probably a bit of a Rubicon to cross there mentally for the American people. Obviously, people don't like the idea, or there are many people in America that don't like the idea of sensitive money to fight a foreign war and they think we should reprioritize America and American paying down the debt, for example, all those different things. The Ukraine war has gotten some pushback from a lot of folks in tech and beyond for that reason. But the real red line for a lot of people is boots on the ground. Don't send our soldiers over there. It's a quagmire. We live through the global war on terror. Let's never do that again. And so it'll be interesting to track which way this leans because there's obviously multiple tiers to engaging. One is just you, you know, you sell the weapons or technology to another company, to another government that uses them. Then you're deploying American military assets, but no boots on the ground. And then the final is of course, boots on the ground. But hopefully this whole thing resolves quickly and we can get to a peace negotiation and a trade deal because we want, we want freedom, capitalism, stability. We want, we want business. We want agreement. We want, we want 10 more whizzes out of Iran. Give me the Iranian whiz. That's what I want. Yeah, give me some cybersecurity companies that we can flip to Google for 30 billion. That's what I want.

Crazy. Anyway, we have some more. Apparently there's some news leaking about OpenAI's or, sorry, Xai's fundraising.

Oh, okay.

They are spending 4.7 billion in the next three months on a new out of their new 9.3 billion dollar raise. They're projecting 18 billion for new data center CapEx through 2027. And they are. The XAI employee payroll has almost quadrupled since the Series C. It's interesting to think about. Yeah, do it, John. We'll give you one of these air horns too. There we go. It's interesting that X was going through this like, you know, intense austerity, limiting spending, you know, keeping employee counts low. Now it's just scaling, scaling.

We love scale.

It's great.

So what else is in the news? Oracle has a new initiative to help companies sell technology to the Pentagon. Larry Ellison, absolute dog, working to help bridge the gap between startups companies and the DoD. We've seen this from a couple companies. Palantir has the AIP program. Anduril has been acquiring some companies and leveraging their connections in the DoD to actually get distribution on these products once they're built. Every company has a different kind of angle on the problem, but it feels like the big narrative is like, we heard this from the Secretary of the army, like the DoD is ready to buy new capability from the private sector, but it's still really hard to actually get a program record. It's actually really hard to get the products in, educate people, build the right consensus, figure out what the budgets are, and get the products in the hands of the warfighter. And so Oracle is stepping up with a program called the Oracle Defense Ecosystem. It's structured to help smaller companies break through the challenges they typically face in selling tech to the Defense Department. It's far too hard to serve the American defense enterprise. We can provide an easy path for these companies to get better access to the defense market. So shout out Oracle. Consistently underrated. I say put them in the mag 7 right now.

If you just saw the chart of Oracle, you would think it was a bitcoin chart. It's just basically up into the right like crazy. It's an absolute hockey stick. In other news, Morgan Barrett is posting. Just 30 minutes ago, he said, when I heard that there was rampant payments corruption in Iran among the military, which is causing the regime to collapse, I initially thought that was a TVPN bit advertising ramp. So, yeah, we want to, you know, there's once. Once, you know, Iran is. Is.

We'll get him on ramp.

Free and capitalist. We will are happy to go over there and. And spread the good word.

Everyone on ramp. Anyway, we have our next guest, Ryan from Crosby coming in the studio. Ryan, how you doing?

There he is. Welcome. Big day.

Congratulations. Yeah, kick us off with the introduction of the company, yourself, the news, everything.

Amazing. Guys, I'm so happy to be here. Thanks for getting me on. Are we ringing the gong?

Well, I mean, you got to tell us, is there something worthy to ring the bell?

We're.

We're announcing Crosby today. We've been in stealth since about the fall, and we're announcing that we raised our $5.8 million seat.

Congratulations. There we go. Go.

Incredible.

Congrats. So break it down. What are you building? And then give us some differentiation. How do you see the market playing out? We'll dig into it all, but give us the.

Yeah, 100. I'll give maybe like a few seconds on my background and how I got here. Yeah, this is a live tracker, by the way. I'll get to that in a second.

Yeah, I want to hear about that for sure.

Way onto these boxes.

I love it.

So my background is have been between tech and law for like a decade early at a couple of startups that did well and also for various reasons, ended up going to law school practicing. My last company, we grew really quickly from like 10 to 100 people in about A year. And the only thing slowing us down was contracts. And because I was the only person there who knew anything about legal, it was like my problem. And obviously I tried out every legal AI thing I could get my hands on over the last few years. It's such an interesting space, but I kept feeling like something was missing. A lot of the legal AI tools, plugins, word add ons on the market made me a little faster as a lawyer, but I was still doing so much manual work. And so I started with my co founder, John Sarahan, really early at Ramp. So we're celebrating twice today.

Let's go.

Really interested in, like the Ramp Alumni day. John's big post today is that finance agents could walk so legal agents could run.

Oh, that's great.

That's as good as we did.

Yeah, yeah.

So we got, you know, John, like, he was obsessed with helping, you know, fast growing businesses grow faster and save money and save time. Like that was his thing. And I was obsessed with this contracts problem. We started Jamie idea last summer and really quickly we came up with the idea of just starting our own law firm. Like, what would that look like? Like full, you know, starting from scratch, hire lawyers, try to replace their work agentically. Like, what work can actually be done with agents? What can. How do we get the speed of AI with the expertise of lawyers? And we had this feeling like contracts are the API of business. Anytime there's any economic growth transaction between two parties, there is a contract and we should make that faster. They haven't changed in 50 years the way we review them. So that's what we're doing. That's Crosby.

Okay.

Yeah. I have been very upset with. No, no, no. Well, that, that I love our lawyer, but. But it's definitely a cost center. No, but I've been kind of interested to follow the fact that we haven't adopted more AI. We're an SMB, right?

We use AI all over the place.

We use AI across the entire business, from content generation to management, transcription, research, all this stuff. And yet when it comes to actually generating and, you know, going back and forth on contracts, it's still very manual. And it just feels like a place that, that we already should be, have. Have picked things up. So.

Yeah. Can you talk about the different tiers of legal work? Like my, my mental model for this is like, there's like form filling, which did actually get somewhat commoditized by just crud apps. In the Web 2.0 era, you have companies like LegalZoom. If you need to just fill Out a form to form an llc. They'll do that. Stripe Atlas, for example, didn't use AI but was able to advance a lot of things. Then on the far other end, you have someone who looks more like a legal artist, like, the top tier of lawyers who are thinking really creatively about how to construct also. I mean, whoever came up with the OpenAI. Org chart, clearly that person's not getting displaced by AI anytime soon because that was a work of art. It's like it's a monument to big law. Exactly. Yes. I mean, I'm sure the legal bills there is probably in the hundreds of millions of dollars, but. But. But it seems like every year, as we chip away at better foundation models, more agentic reasoning, longer, longer test times, inference, longer reasoning agents, we can move things from one bucket to the other. And so talk to me about the flow of, like, the roadmap of moving more and more into the AI world. What's today? What's in six months? What are you tracking on the foundation model side? Looking forward to an UN lock?

Yeah. I mean, these are like. I mean, these are like, the fundamental questions that John and I were playing with back last fall when it was, like, me sitting there doing legal work and him looking over my shoulder trying to convince me he could automate the things I was doing and me being really skeptical. And, like, we did this for weeks. And I think. I mean, you. I think you nailed it. There's sort of like, we. We kind of mapped out, and there's probably thousands of skus of legal work from, like, sure. Reformatting a word document to, you know, writing an appellate brief for a. For a court. I think as you go from least complicated to more complicated, the work gets slower, it gets more expensive, and it looks more like an art. And it's like, there's a real craft to it. It gets more sophisticated. And I think part of what's going on there is that part of legal work is fundamentally like a dialogue between humans. Right. It's a lawyer to a judge or to a jury or two contract, you know, like parties negotiating with each other. And there's, like, nuance and subtlety that we had a really hard time. Like. Like the difference between the words reasonable and commercially reasonable are like an embedding sort of concept. Almost the same, but actually quite different. So, like. So this is what we're struggling with, I think. Look, we obviously have a long bet on where the models will get to in terms of, like, getting better and better so that we can have a fully Agentic law firm. Like, that is where surely this will go. And so the idea that we started with like a word add on.

Yeah.

As like the packaging for this is kind of straight. Like no lawyer has ever or nobody's ever gone and said, I want a word add on. They just. They want a lawyer. They want to throw it over the fence and not think about it. So that's what we're doing here today. And right at like, right at the sort of cutting edge, like, of where the machines sort of stop and the human intuition or judgment or just accuracy has to come in. That's kind of what we've been trying to model, like from a workflow perspective where lawyers will jump in. That's where we are today.

Yeah.

To be more precise on the. Like, where. What kind of what skews, like, where do things land? I think, think right now we're focused on contracts. It's crazy. It's like this $40 billion or so that we map spent on contract review in the US These are like contracts that are like a lease to like a merged agreement that's 80 pages. And so we just focus there. And so our goal is to move up the complexity scale. We're with these commercial agreements now that are like, let's call it first quartile. You know, not the easiest things, but still needs some human in the loop talk about.

I want to give you an opportunity to talk about maybe how. So we recently moved into this new studio. The typical process, you wait a long time to get a lease. When you finally get the lease, you know, I take a quick skim, throw it over to our lawyer. They're like, cool. It's going to take us a while to review this.

Right.

It's variable. Maybe they're busy, maybe they're not. They turn it around. You get through this issues. How much do you already feel like you're an order of magnitude better than the traditional process? Like, can I basically immediately send a contract to Crosby and you guys can just like, basically go line by line almost immediately. And you know, because. Because the typical process is like, okay, you have a lease, maybe there's 15 or so places that you might want to push back or change or things you want to add. And you're basically just going through and talking through those things. And it feels like that could be done, I imagine. I don't know how far along you guys are. It feels like you could already be at a point where the speed that.

The almost like it's almost like I want a. An agent on my side. And I want them to have an agent on their side and I want the two agents to fight it out and do two weeks of fighting in two minutes. Because like really I want representation and they deserve representation, but I just want things to go faster. Because what kills me is not that they're pushing back On Is it $2,000 or $5,000? What I hate is that it's too two weeks in between a turn of documents.

Okay. Like, like, honestly the first thing about the mapping out legal complexity was like the first slide in our pitch deck.

Okay.

And then like the last slide. Whereas this, this is, this is illusory but crazy idea. But it's like it's getting more real that there's a professor that was doing research I think at Stanford Law School that like if you have two law students doing mock negotiations for divorce negotiation. So it's very like stylized. It's like who gets the kids on these days and that date? And then you give their same bottom lines to agents. The repeated agentic negotiations, they simulate that, that, that process. They always get to better outcomes for both parties.

Interesting.

Right?

Wow.

So it's simple. There's not that many things to negotiate.

Yeah.

So John and I were thinking like, okay, there's like about 50 main sticking points in a commercial agreement. This is like if you're any startup, you were selling software, you were negotiating MSAs constantly. 50 sticking points. What if we could map like Adobe's preferences and like Microsoft preferences and rather than having months of back and forth, just agentically simulated and in minutes say guys, this is what you're going to get to. And so that's like, that's like maybe a few years out, but that's the goal of this company. We want to be the API for businesses to connect with each other from a legal and risk hearing standpoint. Where we are today, how it works, like in terms of because we got money now, we're starting a new office and we're doing that same thing.

It's been three months.

Exactly. We need to office. But we like these commercial agreements. Right. Like you do a few of them every week. We're actually, I would say, and like I think for the fastest growing companies, these are the things that really slow them down. So we're specifically working with the fastest growing companies on the market. Like we've been working with cursor and Clay and unifi. We've now done. This is over 1200 contracts since we launched.

There we go.

And they look similar. Right. So they're Kind of like you see patterns, you see their preferences. And so the way it works today. And this is like, it's like all the pain that, like I hated was like, when a lawyer sends you something back, it's going to be an hour or two days, you don't know. And they're going to ask you questions, you have to clarify and kind of solve things. And then you send to the counterparty. Today, the way it works is a salesperson at cursor or at Clay will just slack us a document, tag our bot, we ingest it, we do review, obviously with some AI, some human oversight, send it back within an hour. Our Median time is 58 minutes. And the salesperson sends it right back. And we have a knowledge base for everything those clients care about. So we know the things that Chris is like, I don't care. We don't need to worry about that. We can let it go. And the things that really matter. And our driving insight here was we can unlock sales teams to be so much more fast at their jobs. The things that really matter because of this.

You mentioned that you're building a Word, a Microsoft Word plugin. Microsoft's obviously built copilots into a few of their products. Are you worried about getting paperclipped if Clippy comes back and Clippy gets a law degree?

Clippy with a Harvard law degree, that is dangerous.

I'm joking, but I'm somewhat serious. It feels like the last thing on Microsoft's roadmap. But they do have GitHub, copilot. They do have a product for software engineers. Is there any risk of Microsoft going into this market just generally, even if it's not directly?

That's the question everyone wants to ask. What if Clippy does this?

What if Clippy does this? Are you going to get paperclipped?

Yeah. Wow. I don't think. All right, so I think that. So you know the first thing that like every founder, legal tech is like, I'm going to build a better text editor. Like, I'm going to build a better Microsoft Word.

Yeah.

And. And it's like this pipe dream. And like, it never worked. Like, it's just you, like, you can't.

It's impossible. If you move the image, the whole document always goes. That's just the nature of Word docs.

I can't tell you how much time. So, like, I don't know. We're. I'm bullish on Clippy. Like, I think we're. We're like, we're working heavily on top of Word. But we're not like the better that we do and the more documents we review, the more, the more word licenses that we need to buy for our law.

Yeah. And also, I mean you have, you have a flywheel and it's a completely different market.

But we're not selling, we will never sell against Microsoft Word.

Yeah, yeah, yeah.

Would you ever, would you ever release a B2B 2B B type, you know, product in terms of like a copilot feels like a very distinct decision to say we are our own law firm, tech enabled law firm. We are not, you know, just, just allowing other law firms to build on top of this.

So like I think lawyers, I mean I am one. Right. Like I think that we appreciate and I mean I say this sincerely, like there is an art to what we do, but I think it makes us really ambivalent about AI. Like we, like even if the work, it does not all that we, you know, we talked to so many lawyers and some of them are like the most AI pills. But most are like I'd rather write it my way. Like I think the way that I drafted that sentence. Like even if our tools, like I always thought that there would be a cursor for legal by now that lawyers would just like use the way our whole team is obsessed with, you know, with better id. But I don't think it's manifested which is, which is all to say I think part of what we're innovating on here is not just the software and the AI sort of workflow tools we're building, but also the way that we train lawyers, that we upskill them. It's a skill issue. The way that we get them better at what they do to use AI the way that we want them to. Now if we can fully automate this service one day and it's basically like selling software, which I think we will do. I still want to package it like a service. I don't want people like opening up a software platform to deal with. It's like lawyers, that's just not how people think about their lawyers. Just a problem you want to throw to someone else. And so I'd love us to feel more of that service experience. Always.

Yeah, like it's an agent in Slack and you can get that experience that you have with a great lawyer of, you know, just talking through things.

Last question. For me at least I don't know how much you've studied the history of like legal tech, but Atrium's like the popular story that everyone knows. But there was actually A company before. Do you know Clear Spire? Have you heard of this? Okay, so lessons from Clearspire and Atrium. I'd love to know how you tell the story to maybe investors who are asking about those companies and what lessons you learned and how you're differentiating.

Well, what's interesting is like, those companies were all like, sequentially sort of staggered by like, I'm amazing about Clear Spy. That's like niche. There's this great book about it. Do you know what I'm talking about?

I haven't read the book, but there was like a Harvard Business case study on it. Yeah. PDF. I read the PDF.

Yeah.

Okay.

Yeah, it's fast.

So it's like. So it's like. So like, I think they, they came up with a really clever innovation which was you have a. It's just an, it's a. It's a regulatory thing, but you have a C corporation, you know, typical startup and your PLLC and then a binding contract.

Right, Exactly.

And so we've done that. And then Asian did that. They took it a little farther.

They.

But.

And, and they, and they took it a little farther and focused on types of work that were maybe more automatable. They focused on startup law. They were like Y's law firm.

Yep.

And I actually haven't spoken with maybe I spoke to one person, Clearspire, but I've spoken to all the co founders now of Atrium, and almost all of them, except for one, said, right idea, wrong time. Like, we just spent all of our money on nlp and now that's a solve problem.

Totally.

Right.

Yep.

And so that, that's the. And so like that's the story is like, it's a really, really good idea.

Yeah.

You go after specific types of law that can be automated that like you don't even care the lawyer's name. There's types of work where you're like, I need to, to know my lawyer's name. I want to know their resume. But for other work, you just need it fixed. And so I really think like, they, they figured out the hard things for us and they set the stage.

Yeah. I mean, the series A branding, I was like, we'll do your series A super cheaply. Was like, it was, it was great wedge, but it just didn't expand that much because like, you only do one series A. And then for the B, it's like way more complicated and you just bring in a traditional.

Totally. And that's a loss leader for like.

Totally. Yeah. It doesn't feel like a loss leader when I Get a six figure bill, but. But I guess it is. Anyway, this has been fantastic.

I'm super excited. I mean, there's so many different types of contracts that don't make sense to. That. That are important, that need to be done well, that don't make sense to, you know, you don't need to, you know, spend tens of thousands of dollars on. So anyways, very excited for you. Excited to follow the journey. You can be one of our new legal AI experts in residence. So come back on when there's news and congrats to you and the team.

Congratulations.

Cheers.

Bye.

Great. Well, that's been a fantastic show. Let's check in with Tyler on his review of Windsurf. Did you get it installed? How's it going?

Yeah, so it's pretty good, I think. I haven't used a lot of these ides yet. I've used cursor a little bit. So your code's handmade farm to table. Yeah, I actually like to. Usually I write it by hand and then I just scan it in.

Sure.

I think it's more of a. Yeah, I feel closer to the code base then.

Fantastic.

But I think it's really useful. I mean, it's great for like front end stuff, I think. And obviously I've only been using it for like, you know, an hour and a half.

Yeah.

I'm sure there's a learning curve, but I think when you start to have like. Okay, now I have this function that runs on server. It's like, how does that connect? It might not be in the exact same code base. It's hard to integrate these things.

Sure.

But I think for small, like it's just running locally, it's like super great.

Okay. Did you have to. Was there a model selector? Because there was that drama about anthropic, like pulling the plug on the Windsurf connection so that when you set it up, you used to just give Windsurf your credit card and then you'd have access to Claude and they would handle the billing internally. Now it seems like you have to bring your own Claude API if you want to use Sonnet, which is the popular one, I believe.

Yes.

What was your experience like selecting a model?

So I'm on the free tier, but I think I've. It's like a free trial, but I can choose 3.5 Sonnet. But yeah, it's. Bring your own keys.

Got it.

So I would have to go to the cloud website and then do like export keys and just paste it in. But even then, I mean, I'M sure it's like pretty easy still.

Do you notice any of the original question we had was like, if Microsoft got their hands on this intellectual property, that would be incredibly valuable. Does that feel like a reasonable narrative to you or are we leaning more like this is more about data aggregation in the front door to AI?

I mean, if I had to guess, I assume it's the latter. I think there were some rumors at some point about OpenAI trying to build their own kind of IDE and that it failed.

Interesting.

And that was like maybe the reason for the acquisition.

Yeah.

But I mean, I can't imagine that it's really the ip.

Yeah.

Because I mean, at some point it's like just, it's kind of a. At some point it's like a VS code fork, right?

Yeah, yeah. I mean, Microsoft owns VS code code, so yeah, you'd think that they would be able to iterate towards like the best in practice UI patterns essentially, if that's what really makes Windsurf stand out. I'm wondering, like, there was a ton of heat on Windsurf on X for a while. It was really popular. And yet I saw that at the AI Dev World Fair that SWIX put on, they had everyone raise their hands and it was like 80 or 70, 80, 90% cursor users. So cursor's clearly broken through in a really meaningful way into just like broad adoption. And that seems really, really valuable because you're getting all that training data that the and can be used in reinforcement learning contexts and, and derive better AI models. So yeah, it seems like it's kind of unclear to me exactly how much of a real issue this is. We'll have to ask some more people. We have some other AI folks coming on the show later this week. We'll, we'll try and dig into, you know, is Microsoft after just the cash flows or the equity or the data or the IP or the models, because they have like different claims on different pieces of OpenAI and all of those are probably negotiating points as they kind of separate these two entities out. And so it's a, certainly a unique situation because most investors, if you get them on your cap table, maybe give them a ward seat, but they're not getting access to your GitHub, they're not getting access to your code or your IP. Can you imagine if that was like, oh yeah, like, you know, I took money from this tier 1 VC and then they just gave us all my code to a competitor that they invested in? Like, that would be a wild deal.

It really is a wild deal in hindsight. I mean, it was a wild deal at the time, but this sort of 49% Prof. Capped profit share. We don't get equity, we don't have a board seat. But you know, we have sort of a right to IP here and individual models. So it's going to take a long time to shake out. And Sam is fighting wars on on at least a couple fronts.

Yeah, I mean, we didn't get a chance to dig into the full Ben Thompson article on on this, but he says the Wall Street Journal doesn't indicate who its sources are. But there is little doubt in my mind they too are on the open air side and I get why. And he says his biggest takeaway is Ben Thompson writing a strategic which you should subscribe to. He says my biggest takeaway from this story is that I have underestimated the amount of leverage that Microsoft has over OpenAI. I mostly wrote about this slow motion train wreck last month. I still think that his his take is broadly correct. And note that the right to make OpenAI models available to other cloud providers is one of the points of contention in the story. But the fact that OpenAI seems to be reduced to issuing fantastical threats through press suggests that the decision about the endpoint of these negotiations is almost entirely up to Microsoft. To that end, I can understand if the company is pushing for more than I suggested. Indeed, if anything, these threats through the press might make Microsoft want to hold on to more, since they probably have since they'll probably have to fight OpenAI again in a few years. I still think, however, that this is all the more reason to give more ground now in exchange for an ironclad agreement securing Microsoft's exclusive azure access to OpenAI models in perpetuity. In perpetuity. That is a fascinating thing because if OpenAI somehow compounds and runs away with it like they have this massive consumer engine, that could be so it feels like right now anthropic and an OpenAI on the developer side are neck and neck and anthropic even seems a little bit ahead on code gen and just developer adoption. But if you if it winds up being a capital fight, if it winds up being biggest data center wins and we need to build the trillion dollar data center and Stargate becomes extremely real and extremely material, well then having a consumer app that's printing $10 billion a year is going to be way easier to underwrite as an investor to go and build the next model. And then you might get higher scale which could make the developer API more valuable. And if Microsoft has exclusive right to resell that. And they are the B2B player, they're the B2B front end to OpenAI models. That is extremely valuable. Extremely valuable. So, yeah, I was not aware of that. He also talks about his interview with cursor CEO Michael Truel.

It is the perfect irony with OpenAI is that it, it's one a YC company run by Sam Altman, who was president of OpenAI. And if YC was advising a startup that came to them and they said, we're going to enter into this really complicated commercial agreement and we're going to structure as a nonprofit, we're going to take all these donations and then in the future we'd like to go public. They would be like, you should just try to not add too many people to your board. Take as little dilution as possible. You should, you should basically run like a dictatorship.

And talked about this. I would not take my. I would do not do as I say, not as I do.

Yeah. So, so to, to create one of the most important new companies of, of. Of generation a generation and then to have it just be ensnared in all of this drama from the nonprofit stuff and the battle with Elon all the way now to battle with Satya is just absolutely brutal.

Well, the last story we should, we should close it out with is the Trump phone. The Wall Street Journal says Trump's smartphone can't be made in America for $499 by August. That's because we got Tyler over here assembling iPhones in America. If Trump gets a hold of him, I think all bets are off. But Trump's mobile phone shows some specs that would be beat. Apple's biggest, priciest iPhone models doesn't want to just make it in America. He wants to release a gold and Android phone that will be proudly designed and built in the United States. Fascinating development. This is the first time I heard it was when the Wall Street Journal was saying it can't be done. But yeah, apparently Trump's son told the podcaster Benny Johnson on Monday morning, you can build these phones in the United States. So we'll see where.

How.

Got to dig into that. The display is going to rival Apple's two $1200 iPhone 16 Pro.

Max, nobody, nobody ever thought to turn the White House into an incubator.

It's happening.

Nobody ever. I mean, I'm surprised. Hundreds of years of presidents, nobody thought, I'm gonna turn this into Jimmy Carter had a peanut venture Studio.

Jimmy Carter had a peanut farm, but he divested an idiot before he took the residency because he didn't want, he wanted his hands clear. He, he didn't want any conflicts of interest, so he sold his peanut farm. We are now in a different era building 50 megapixel iPhones.

The hottest new venture studio is the White House.

The White House.

Anyway, I mean the velocity of new C Corp LLCs that they're just firing out of an absolute cannon.

Bitcoin treasuries and social networks and crypto.

Yeah, we didn't really cover the story of the Tron ipo.

Not enough people are.

Which Eric Trump has now denied having a role.

Okay. I mean this is the first founder of a unicorn company in the White House. Yeah, it's huge. People talked about Zuckerberg maybe running for president.

You never say unicorn Tech founder Donald Trump.

Yes, yes. They never, they never introduce them that way. Maybe they should really put the focus on what matters.

Social network trading at truth.

Figure it out. Oh, wild times. Wild times.

Cool. 5.1 billion, 5.1.

That feels down off of a, off of a little bit of a pie.

Yeah, I mean it's been generally down and to the right in, in 2022 it peaked at 92. 90, 92. Well, the intraday high was $99. It's sitting at $18 now.

Okay, so it was like a 30 billion dollar stock at one point.

Yeah, but you know, if you sell enough Trump mobile devices, who knows? And then maybe it's Trump, maybe it's Catalyst, maybe Trump djt, Truth Social is just pre installed and maybe they have their own search, they build out their own search.

I think they should get into foundation models. Really? Really take a run at the big labs, hire some researchers, start doing some reinforcement learning. Let's see it. I want some research papers. Maybe they should do an open source model. I want some thought leadership on artificial intelligence. AGI. We've suspected that he's AGI pilled, but we haven't seen it in the research papers. Let's make it happen. Anyway, that's our show, folks. Thanks for watching.

We had some range today.

George had a lot of range, was wild.

You had some spicy takes. That was fun. I wish we had another 30 days.

We went all over the place. We went over to Chinese history, we went into the NFL.

Technology investors, Saquon Barkley, some series A's, some series B's.

Really good range, fantastic show. I think that's exactly what people come here for. Anyway, thank you for watching, folks.

Thank you, folks.

We will. See you tomorrow.

We'll see you tomorrow.

Have a great day.

Cheers.