Sam Lessin: Google is winning AI, incumbents beat startups, and OnlyFans is a screaming buy

behave in the interest of humanity. It was it was a cool like thought experiment almost. Uh it was very cool. Anyway, uh Patrick Oanis was shouting it out. He says one of his favorite genres of podcast is recurrent expert guests on on the same show.

If you're interested in the nitty-gritty of AI model progression, you'll enjoy this. So go check that show out this weekend. Uh, it's a great listen. And we have Sam Lesson from Slow Ventures in the studio. Welcome to the studio, Sam. Gentlemen, there he is. It's Jessica Lesson's husband. Put it on the flesh.

I saw her like literally minutes ago. That's how I am. Tell her to tell her to come by at least and say hi if if she's not uh busy. Yeah. Yeah. She's uh she's she's busy this moment. But, you know, we're 30 minutes. We got the crossover of the century. TBPN the information. It dropped today. It was great. I love that.

Yeah. I actually I I knew about the development of that story from dinner conversations. Oh, yeah. It was a lot of fun. Yeah. Oh, I bet. I bet. Uh Abe is Abe is the man. Yeah. Uh there you go. It was It was fun hanging out with him and um kind of the So, are you guys you guys are the cover of the weekend section?

Is that right? We are. We are. The big Are you guys going to do a print edition for us? That's coveted space. I always thought that they should actually do a print edition. I think it's time. Everything goes in a cycle. I got the I got the Wall Street Journal here.

I need I know I was once a once weekly uh uh Saturday dropped off on the doorsteps of every% hire the kids to like deliver it with the with the you know throwing it out the window.

Look, I was in an airport lounge in Germany two days ago and they had all the the print newspapers and I'm like look there's a missing opportunity here. The Germans want to read about the information. They want to read about technology brothers. Yeah, of course. Let's get it to them.

You guys are big in Germany I'm sure. Well, somebody started translating our streams into Dutch, which was interesting. They're doing other other things. That doesn't seem like the first choice. Like, don't you want like Mandarin or something? We're trying to get We're trying to get big on Waybo. Yeah. What's next?

Yeah, but we're expecting to have some push back. Some push back. Um anyways, uh it's great to have you on as always. Congrats on all your success. You guys are everywhere. This is this isn't the new studio yet, though. Well, you're a part of it. You're a part of you're a part of that studio.

to come down next time you're in LA. Come by in person. Be great. Tuesday will be the first show. Yeah. It's a lot closer than Anglewood Cliffs, New Jersey, you know. Yeah. Yeah. What's what what's there? Is that the new Is that the newb CNBC?

You can't you can't make this whole CNBC rival thing and not know where CNBC is. Everyone always says like uh Yeah. Uh Andrew Reed was like, "Oh, they're making Squawkbox for tech. " And I was like, "Got to start watching Squawkbox. " I guess Netflix is also Netflix has a new billion dollar New Jersey studio.

Maybe maybe some New Jersey and maybe we're taking the fight to New Jersey in the future. We'll see. Let's pick LA's fine. LA Where were you where were you traveling this week?

I I I went I I've had an insane travel schedule, but I was uh Monday I went I was New York uh Frankfurt, New York, San Francisco was my was my week. What's going on in in Germany?

Germany was a personal thing I had to take care of but but New York was the Salana acceleration uh what is it the Salon big Salana conference which they call ship or die very dramatic brand nice uh you're infamous there right the GM the GM guy do you still get comments on that you know I saw I saw Raj backstage you know I was I was one of Salana's first investors I like my my um Jason Calcanis I was the first investor in Uber was like I like do that for Salana right and so you knowve I've known those guys for a long time and I saw Raj backstage and we were reminiscing when I did the GM thing which became one of the first meme coins and blew up and got if you remember it got Raj banned no I know from Twitter it was amazing about that yeah this was like Twitter still right so it was like somebody at Twitter was like didn't get the joke obviously didn't have context that like you guys are boys and just yeah the thing the thing is I unfortunately have other people on on Twitter legitimately threatened to kill me, which that's not so good.

You got to get that taken down. But the the Raj was a pretty funny one to for them to respond so so quickly to. Yeah. Yeah. Quick reaction. Uh there's so much so much to cover in the news. It's AI week. It's AI week. Isn't it every week AI week, boys? This one was particularly AI week.

We got Microsoft build Google IO open AI IO anthropic uh a big uh are you are you taking a victory lap on the idea that this is a sustaining innovation that Google's going to win you know I you guys actually I'll tell you a funny a funny answer to that you guys know my associate Jack Reigns you know this guy Jack sitting he's right over there Jack he can come wave for a second come say hi Jack sends me this long missive this week about how amazing Google is and how Google is going to win the AI wars.

What's up, dude? And I'm like, what value are you adding? I've been saying this for two years. I'm like, thank you. I'm like, I wanted to check out all the little Google I was like I was like, Jack was like, Jack, you know, you're an associate at an early stage venture capital fund.

Telling us to buy Google stock is not exactly guys. I want to see slow as the next crossover. Yeah, crossover. Become an RAIA 30%. Never. We've actually looked into it.

It's definitely not a thing we will ever No, I mean I mean the the the my favorite comparison I forget who posted it, but they were like, would you do you want, you know, Google at, you know, 17 times earnings or Costco at 50 times earnings like at this moment, this AI inflection point? Look, here's the negative on IO.

The negative IO is that per usual, Google's actually terrible at marketing, right? And so they just launch everything. And so you're like, I have no idea even where to focus right now, right? Like it's it's not my value comes in figuring out. Thanks, John. Focus on generative AI videos, V3. That's the thing.

It's like so many things you're like, I don't know. But the net of it is like they're just crushing it, right? Like from a tech perspective. And I got to say, I'll tell you a funny story. I spent like 20 hours quote unquote vibe coding over the last week cuz I was on so many airplanes.

And it turns out airplane Wi-Fi, vibe coding, perfect match of activity. So, I just like was like messing with these things till my eyes bled. And you know what happened in hour 11? What?

I stopped using chat GBT and went all in on Gemini cuz it's actually better G 25 Pro is just like way better at a bunch of these coding tasks. It's better at structuring. There's like a bunch of stuff. So, it's it's I think it was absolutely a Google week. But were you using codeex? No, I wasn't using codeex.

There's always Look, this is one of my other insights from the week. spending a week, you know, literally flying around the country vibe coding in the air. Yeah. Is um you know, the the whole what model to use thing is a complete [ __ ] mess, right?

And I think the analogy I now have in my head is it's kind of like imagine walking into a crowded room where you don't know anyone and saying who should I trust, right? And then by the way, every two seconds a new person pops into the room.

You have no and like I people are like oh or or you can trust this person 95% of the time. you can't trust him with your life, you know, but that's a totally that's a totally separate chain of thought which I think is important. I was talking to some entrepreneurs today about if you think about what AI is today.

It's like having it a B minus employee or like an 80enter. And the question is how do you get value out of 80enters is actually more difficult than it looks in a lot of ways, right? Because the reality is so much of work is about trusting.

But isn't it more isn't it more their B minus because it's you're averaging their effectiveness on different things in in my experience.

Like for certain tasks it's like 80 90% of the time it's amazing and then some percentage of the time it's so bad that you would want to you would ask the employee is everything okay like you never kind of but I think the other the thing I really think about guys is you know if you think people are always like oh there all these tasks that AI is more efficient at and so clearly those jobs will just go away and it'll be it and I think that's exactly the wrong model because the reality what people don't understand is the real reason you hire people to do work is not to do the work.

It's to have a throat to choke when the work goes badly, right? It's like, why do you have an accountant? You don't have an accountant to do your taxes. You have an accountant so that when the IRS comes and says, "What's wrong with why did you get this wrong? " You're like, "I didn't do it. That guy did it. " Right?

And so the I think that's the interesting thing is like there's all these models like great, good news. You get rid of the employee, now you get a queue that's mostly right and you approve and you're like, "I don't want to approve that stuff. " Like the whole point is to have someone else. Yeah.

I think our our example is like we use a lot of the different deep research products and like we still would love to hire an amazing researcher. Yeah. So that you can fire them when they're wrong, right?

Like you need like just the point is like there's this whole I think there's a lot of like new well it's an agency thing too.

It's like if you could tell if you could tell deep research to like try itself and then try other products and then combine the outputs of all the different things and maybe that's a research agent that can you know there's all these things.

But I think the ultimate model in my mind is I don't even care what the average persons are using. I don't care about some review online that's four stars for coding for this type of I don't care.

What I care about is what are the smartest people I know in the world using and like how do you make that transparent so I can just be like oh it's this type of task I'm going to trust this model for that and like that needs to be up to date it's much more of like an identity problem or like a human trust problem right than it is you know a technology problem in a lot of ways and it's just changing really fast so it's it's been a fascinating week but yeah net I mean pulling pulling all the way up yes I'm taking a victory lap on Google wins and yes I'm taking a victory lap on incumbents win um because that's what's happening Mhm.

So you're you're saying to Jack like, "Thanks for reading my memo from two years ago. " Yeah, exactly. Well, in fairness to Jack, he wasn't on the team two years ago. He was very excited to discover notebook l. But love you, Jack. I got we got to get him out of the pool house.

He uh Anyway, uh I want to talk about some other stuff. Uh what what do you think about the uh Johnny IV news and and hardware generally? I mean, where to begin? I know. I know. This is my favorite part of having you on the show. We don't have to prompt at all. We just like two, we could give you three words.

Johnny, I've hardware. Well, okay. A few things. First of all, I mean, my my my my tweet on this when it was happening was like, well, I guess all these hundred million dollar AI researchers are cheap, right? Like the these it's the designers that are going to get paid bank on aqua hires in this in this era, right?

Look, I think it was marketed in a really interesting way, right? Like it's marketed as like they tried to pick the biggest number possible. It's a 2% they paid 2% for this thing, right? For him, they market it as it's this big number, I think, because that's it's kind of the pissing match of AI, right?

Is you're supposed to look really big and powerful with big number. And I'm sure Johnny doesn't mind being told that he was worth $6 billion, but like it's kind of one of those interesting PR stories of like how you shape the narrative as much as anything else. The hardware thing is interesting.

Like on one hand, I actually think it's like if you're open AI, you do need a hardware play. You're going to get [ __ ] otherwise, right? Like how, you know, you know, Meta discovered this years ago, which is like being beholden other people's hardware is really tough, right? In this case, it's only more so, right?

Because of kind of the way these chips are going to fall with Android and iOS. So, I don't think they're wrong and like Met has got its glasses. Everyone's So, like it's not strategically wrong from first principles.

I think the problem is that I have is I just think that like the likelihood they're going to successfully introduce a new piece of hardware is like zero, right?

Um like and that's just you know it's you know you're gonna have a phone and they could try to compete on the phone but it's very hard for them to imagine building like a fully featured phone that you know whatever and then also having something that's so great well well the one thing that does seem clear haven't they said this is a a new type of device that will integrate with your phone and your computer and that screams to me okay well is is app yeah I said septum piercing where you My take was eane.

Did you see ecane? So you this was so when I left when in 2014 No, but sorry to finish my thought that the issue with that is if you're saying we're going to create a new device that integrates with your Apple ecosystem that seems like a bit of a like uphill battle. Look what happened to Pebble, right?

And the Apple Watch, right? In the end of the day, it's like look, you know, you have the iPhone is a hugely breakthrough device. This is, by the way, not my line. It's a friend's line, but like hugely breakthrough device. You already had a phone, right?

You know, it wasn't like a new thing you were putting on your body. The glasses work. That's a thing that was already on your body. The watch work. It's a thing, you know, same thing. It's It's really hard to imagine you're going to get through with some new device.

So, my my joke, but it's only half a joke is look, you know, 12 years ago, I got obsessed with the idea of making an ee. So, imagine like bringing back the cane, right? That's what you meant. And I have no idea where you like Novaane. Yes.

And if you think about if you think about the can it's actually you you put a shitload of batteries in it big antenna right and you know what the nice part is like the world is getting less safe so you can people with it the world is like the world is aging so like everyone needs canes you know like there's all these like macro trends in favor of bringing back the cane and then you know could you drop like a GPU and a microphone and a camera in the top and like talk to your cane and like you know yeah smart Smart cane.

Smart cane if I don't want to be pitched any more non AI canes. Okay, I've had enough. Like I'm just saying like if they actually if OpenAI came out like we've we're bringing back the cane.

I'd be like all right listen for all my Well, you you know that Johnny Ive got his start at a design firm that was working on bathrooms. And so maybe there's all like smart toilet, you know. Yeah. Well, there is someone just came out. There's a there is like a new Fitbit for to for your poop. Okay.

I know some friends that have funded that. That's thing. I I funded that. You funded that? I know some people funded that. I funded that. It's called Throne. It's called Throne. I I think it's interesting uh for a bunch of reasons. One is it's just like personally as like a quote unquote biohacker.

There's so much information on how your digestion, what's happening with your digestion, your hydration, all these different things. In the end of the day, you want to copy behaviors people already doing. And you know, in college, one of my roommates, Brian, would send me pictures of his poop all the time.

So, like, you know, like that at all. You just got to double down. Brutal. Let's move on. Uh, hyperscaler capbacks. Give us anything to get us off of this. Well, here's my hypers scale. Hyperscaler capbacks. Here's my line on it. I flew through JFK today. They're spending $20 billion renovating JFK.

I'm like, this capex is nothing. Spend more money on AI. Like if you to spend $20 billion on like renovating JFK, all of a sudden spending a hundred billion on AI doesn't seem so crazy. Mhm. Uh what about uh say say as much as you can about what Meta, you know, Meta's playbook over the next kind of six to 12 months.

I know they've been having talent kind of retention. Llama was the Yeah. Llama behemoth delayed. Delayed. Look, in the end of the day, I again I as you know have huge respect for the team there and the leadership there.

I think the beauty of meta as I've said many times is it's a heads you win tails you win right you know from from a from a investor perspective right which is if you think about it it's like the most obvious highest value use case in the world for everything that exists is making ads way more targeted and effective right and that's happening and it's an incredible trend to ride and they're going to crush it you know Mark and that team is like deeply competitive they want to win win right and so just from a pure competition perspective I think they're going to like go to the mat to win win-win and like there'll be many chapters to that but the good news is they have a massively profitable business and a immediate use case right that is deeply aligned and so I just you know I learned a long long long time ago to never bet against Mark especially when he cares about it yeah it feels like with Meta there's huge value to having Llama be a a free LLM for them that they can vend into every corner of the app even in places the consumer doesn't feel or or see so uh profanity filtering and uh understanding what content goes where, ad matching, uh generative AI to clean up Instagram photos, all these things that will just be under the surface as opposed to what Google's doing, which is like now we have 10 new apps, notebook LM, like you're we're not seeing that from Meta and maybe that's the right strategy.

I look at the end of the day, Meta has the most natural beautiful use cases for AI, which is what you're pointing out, and therefore like they don't need to do a bunch of crazy [ __ ] right? They just need to like win at like the core. And then basically, again, the open source thing is clear.

like they actually for like think about ads. They don't really care if they're doing the AI or someone else's because it's all going to go to them either way, right? In the end of the day, it's just better if more people have access to it and the innovation will happen.

So, look, one of the things I did in my vibe when I got really into like all the platforms, right? I spent a bunch of time remember Digital Ocean, like Digital Ocean's doing a great job with AI, right?

Like it turns out like their services are really good and like it was really easy to boot up some agents on top of like you know a knowledge base on top of you know that's always the worst part is like deploy you write some code and you want to actually get it on a server it's always still a hassle and it's like they've actually done a great I was like incredibly impressed but like you know and then you say like where is llama showing up it's like well it's like right there like you just click a button and you're using llama it's like no [ __ ] got to get a killer video generation model out of meta.

Like they they're the only one that has another data source market. How do you see that market evolving? There's a bunch of foundation model labs doing video generation.

And I can see some of those getting to the point where they could crush it for big Hollywood movie studio, you know, making very specific types of scenes or assets, but then it feels like the low end of the market for, you know, if you're a random SMB and you want to make an ad or something like that, yeah, maybe use some vertical SAS, but also I wonder how much it matters.

I mean like you know I had this conversation actually with my good friend Dave Morin on our little weekly you know [ __ ] session more or less more or less we're talking about Dave also somehow cracked like the top of the angel there was like the top angel investor list and he he was the top dog it's amazing so let's we know the validity of that list we'll talk about we could I mean I just like to see my boy Dave at the top I like to see look my boy Kevin my boy my partner Kevin was like number four on on the list.

It was a very interesting list. But the uh I wasn't on the list, but you know, that's okay.

The um the uh but uh what I was going to say is the um the thing Dave and I were talking about is like look, you know, if you think about what's going on like with memes, right, in this whole world, you have there's this like high road low road thing.

Everyone's really excited about like super high-end video generation and Hollywood and D. You know what's happened in the media landscape? Those things don't make any money. No one cares about these movies anyway. like they're consuming like low-end crowdsourced memes on the internet. Like that's where attention is.

That's where the energy is. That's where entertainment is. And so I think obviously the pure technologists and like the high flutin people are really excited about like the narrative storytelling beautiful duh.

But when it comes to like what tools will actually be used, how things will be marketed, how humans will interact with each other and communicate in the store, I think actually it'll be much more budgety tools that actually win.

And the other stuff is kind of just, you know, intellectual, you know, masturbation a little bit. It doesn't mean you shouldn't do it. Like people Speaking of that, Speaking of that, why is Speaking of uh masturbation? Uh why is Only Fans getting priced at eight times earnings?

You know, I I so desperately want to buy Only Fans, guys. I've wanted to buy it for years. I had plans like there there's been a few iterations of this and No, it's so good. It's so good. I just What do you have against Only Fans? Uh uh I just so good.

This is But your reaction is part of the reason I want to buy it, right? The buyer pool is small, so the price is low. The Well, that's part of it. The other thing is everyone because the buyers, it's actually the best AI short you could possibly think of, right?

Because part of the reason it's getting sold, my understanding, and like the whole story, is like, well, we're going to have all these AI girlfriends, so like the creator economy is going to get crushed on these things.

And I think that is like a very 101 simplistic take on what's actually the other thing is a lot of the a lot of the cost that goes into basically generating revenue on only fans is from messaging which is done by people in the Philippines and AI.

So this is the thing everyone the open AI is like the greatest human AI platform in existence right from position because you have I think you still need real brands. I think that having human brands matters at the top end. Your ability to then like effectively automate.

It's kind of like the Facebook story which is your ability to slot an AI and bank it right on messaging is like off the charts with only fans and improve the quality. So I think it's interesting. I think it's like this classic thing where it's being sold. I've always loved the asset.

It's the number one creator platform in the world. It's so defensible because they actually take have a very low take rate because they're so big that no one can no one can assault them, right? Because of how they're kind of set up. What's the take rate? It's like it's really low.

I want to say it's like 10%, maybe 15, but it's like much lower than you think. And as a result, like you have this incredible cash engine. I mean, YouTube's like 45% and many other sites are 30% just for general content creation platform. So, just the upshot is like this incredible position.

It's it's I think there are huge product opportunities. There always have been with it. But then you have this double storm of like one people like you guys boo it because you're you patrician technology brothers, right? like and you won't like acknowledge human nature and people want to do this.

And then two, there's all these fancy investors who like think it's dead because they read the front page of the Wall Street Journal about AI and don't understand what's actually going on. I'm like, that's great. This is my zone. Yeah, I have no Well, you're going to have to become an RIA to do it probably.

So, good luck. Well, I got to figure out how to not do that. I also like for what it's worth, I'm a seed investor. I don't know how to raise the $2 billion I need. But call me if you got it. Well, speaking of seed investors, uh I don't know if you saw this chart from Pitchbook.

Emerging managers are on track to raise less money this year than they have in a decade. That sounds right. Uh it's now under $20 billion. During 2021 it was up at 60 billion was really really speed uh really spiked. 2022 was around 50 billion and now we're down in just one JFK renovation.

That is that is a lot of money for emerging managers. This is I assume is defined as we're going to put $20 billion as an emerging manager. Wait, how much did they get in 2021? Well, you put it you I mean you put it in like $20,50 million funds. That's the idea at least or that's what happened.

They're all I just I've been so skinn I don't see any of this working out for anyone, right? Like you know I I know it it's one of the the they'll always be the random winner. Yeah, but but isn't that the nature of the game? It's like startups follow a power law so do funds.

I just think we have to hit a size gong because the 64 billion that they raised in 2021 12 billion of fees and so let's just give it up. Congratulations, guys. But it's kind of guys, I think you guys are cheering too much. I do the math if you do the math on I'm joking to be clear.

I mean, yeah, being doing the math on being a small fund and being locked in for 10 years as like a solo GP with your emerging manager fund and your kind of shitty subscale, it's actually a pretty bad business, right? What's wrong with being locked in? No, no, no.

Like what you're saying, what you're saying is so real, which is imagine you have a startup and it's like, okay, you have to commit to this startup for a decade, even if by two and a half years in, like, is that real though? Is that real though?

I feel like most of my emerging managers, they they write the checks in the first year and a half, two years, and then yeah, they get investor updates, they pass them along to LPs, but like they could go and get a job at big tech and Sure. But it's still even if it's couple hours a week for 10 years.

Yeah, I guess that is kind of Look, it's also it keeps you other it's just messy and complicated. And the reality is again like if you're earning two and 20 and you're never going to see the 20, right? like the two ends up actually, you know, on a $50 million fund after you pay for actual things. You're doing fine.

Like no one's crying, but like it's not like some great business, right? Like I think the only reason to raise a $50 million emerging manager fund is like to have enough card flips to prove that you can size up a bit.

Now, I think the flip side is true, too, which is I don't know how anyone makes money on a seed fund that's over $200 million. I the math doesn't work, right? I think there's like this sweet spot of fund sizes that work. Too small. You don't have the economics to make it work.

even if you have a banger, it's like not that important. And like too big is just like the law of big numbers. You're never going to produce an important fund, right, from a returns perspective. So look, what we do, seed investing, we're I think in the sweet spot, obviously, cuz why would I not think that?

You know, we've made money. I'm very proud of that. I care about DPI, whatever. But here's the thing. Sorry, sorry, sorry. We're laughing. We have to explain because you can't see it. But the Chiron right now says seed investor. Don't become a seed investor. stay the [ __ ] out.

The um the um but then the the reality is like you know it's not even from my perspective it's not the best business. You do it cuz you love it, right?

Like but if you just did the math being an asset manager and earning 2% and just getting enough DPI to justify more 2% fees is for sure a more scalable business model which is why most people do it. Do you think do you think uh part of why every emerging manager jumps into seed is just like they can't do growth?

Because I I feel like the skill set of being a growth investor is in some ways more repeatable. You know, you get people that come out of investment banks and they have all the skills to underwrite a company that's at a billion dollars. There's less risk.

But at the same time, when you think about someone who's, you know, starting a new fund, if they're going out and raising a growth fund on fund one, like that's just feels harder to mar question. It's like who's going to give you LP level dollars, right?

And like by the way, if you're if you're going to later stage fund, you're trying to like model to like a guaranteed whatever it is, two and a halfx or whatever. Why would you give it to the crazy kid, right? The crazy kid you give money for because they might be right. Yeah. Yeah. Yeah. Yeah.

And so you want that seed bet. And then also I feel like a big part of the emerging manager dynamic is just like essentially scout funds and and there's a lot of like horse trading around like, yeah, we will do a we'll do the deal from our fund to fund.

We started that way like you know 10 plus years ago and like that when there was the market was much looser and and it was a different market and it worked for us and we were able to scale up into being like a legit seed fund you know with legit dollars and legit institutional investors.

I just think like like most things that path all paths get super crowded, right? And so once the path is super crowded, it's just really hard to differentiate. It's really hard to make a good business out of it. Um and I I do think like mo almost everyone's going to go away, right? Become a seed investor.

You hear you heard it here first, folks. Uh last last uh want your quick take. XAI's Portland data center fire. You think that's just a a little accident or you think Portland is doing Portland stuff? That would be really funny. Um, that'd be really funny.

It's It's um God, have you guys ever read the science fiction parable of the sewer? No. Great book. Great book. I'm just finishing it now, but there's a whole slub plot where there's all these people who are addicted to this drug that makes them like think that fire is the greatest thing on earth.

So, they're just like setting fire to everything. And like, it's funny. It's It's I I just looked it up. It's a post-apocalyptic story set in 2025. Wow. I don't know if you caught that when you read. Yeah.

And like they're basically about people like trying to walk north from LA to like to to Seattle to escape craziness. But the um Yeah. So I don't know if you told me that that drug was invented in Portland. Wouldn't blow my mind. Yeah. I think we got to get a This is fantastic having you on. Thanks so much.

Always a pleasure. Happy Friday. Happy Friday. Talk to you quickly. Cheers. Let me tell you about adquick. com. Out of home advertising made easy and measurable. Say goodbye to the headaches of out