Building AI in New York Before It Was Cool

Transcript generated automatically from the episode audio and lightly cleaned up (filler words removed, misheard names corrected where clear). Speaker labels were added afterward and may contain errors.

Rob May: Hey everybody, I'm Rob May. I'm the co-founder and CEO at Neurometric.

Anna Kirk: Hi, I'm Anna Kirk. I'm the chief of staff at Nadia Partners.

Ryan Eppley: And I'm Ryan Eppley, co-founder CEO of RootAccess. And this is the second episode of the AI in NYC show. And AI and NYC is covering anything in AI at the intersection of AI and New York. So, we're talking to different builders, investors, anyone doing interesting things in New York City and the impact it has in the city. Very excited for our guest today. He's been building in New York way before it was cool. We have some really amazing topics teed up for us, some current events on AI as well as Rob's going to introduce our guest.

Rob May: So, sitting here to my left, our guest today, Stephen Messer. I' I've known Stephen for about a decade now. And, Stephen has the very interesting background of, well, three things that I'll use to describe you. Number one, got a law degree from NYU, but never practiced law. Number two, built a tech startup in NYC before it was cool, which we'll talk about. And then number three, was a very, very early AI entrepreneur back when most people didn't think it was a viable place to build. So, Stephen, welcome to the show.

Stephen Messer: Thanks for having me. I'm excited to be here.

Rob May: Yeah. So, we should probably get started and talk about how you and I met because it it's kind of relevant to the New York thing. We were invited to a dinner maybe 2015 or 2016 by James Cham from Bloomberg Beta and it was in San Francisco and we ended up sitting next to each other talking about why we like the East Coast more than the West Coast.

Stephen Messer: Yep. That's very true. I think I called San Francisco the Boston of the West Coast for the ugliness of the people and food. And you were living in Boston at the time and that's how we became friends.

Rob May: I was. Yeah. Yeah. And then you started reading my newsletter and we started hanging out when I came to New York. And so you've been here a long time. You're a super proponent of New York. You're one of the guys that got me to move here. Yep. You really you and Charlie O'Donnell who we had on last week were always on me. What tell us a little bit about what it was like? When did you guys start link shares?

Stephen Messer: We started link share in 96.

Rob May: Yeah. What was the texting here like then?

Stephen Messer: You know it was actually great. Remember most of what started off you had things like Pathfinder which was one of the early players in the market. You had AOL which was in Virginia. So most of really what you had in the tech other than maybe Netscape early on in some of the search engines you really had it splintered. You had Leos in I think it was Waltham Massachusetts. You had so what you had was this very broad and diverse background of people but because it was media New York really became a center place for people trying to define what is this new medium going to be so it was a great place to be here for experimentation and all ideas were welcome which was pretty great yeah

Rob May: and explain to people what what you guys did

Stephen Messer: so link share provided something called affiliate marketing affiliate marketing is today the engine that drives almost the entire economics of the internet which is essentially think of it as a commission for some kind of action that you want to take place. It could be a click-through, it could be a lead, it could be a sale, it could be you name it, and you'd get a commission. And LinkShare provided the entire infrastructure to allow somebody to track accurately all the way through. We invented a non-cookiebased way of doing it, so it was trustworthy. And then we were the only network that tried to bring everyone together in one place so you could sell multiple people's things and that was pretty revolutionary at the time. The whole thing was

Anna Kirk: And what was it like to raise money for a company like that in that time period in New York?

Stephen Messer: It was [ __ ] hell.

Anna Kirk: I can imagine. What year are we talking here?

Stephen Messer: This is 96. 96. It was brutal. It was about as brutal as it got. We almost had to move to the West Coast. Tim Draper, who's an amazing venture capitalist, came out, loved what we did. And was like, I want to do this, but none of my partners will do it if you're in New York. And I was a New Yorker by trade. I also felt that we were both media transactions. And when you go out to the West Coast, they have a very hard time with anything that's not like pure and easy to understand. And more importantly, if it's not in the common vernacular of what everyone's doing, they have a really hard time. And we were not at that time. People were saying advertising is the way to do everything and that's the only way to do things.

And we thought this is insane. We're a very big first principles group of people. And we were like, so you have advertising that wants you to stay on the website, but they have to have enough clickthroughs to justify the like none of it made any sense to us. And but we were really it was just so we said no and we weren't gonna go and Tim's to this day fantastic supporter of us even though it didn't work out. He came and spoke at one of our events but that was really it. We ended up having to raise money from Internet Capital Group and Comcast at the time who were both Philly based. That tells you a lot about the number one financial capital sector and I had to go to Philly to raise money from amazing investors but still I had to go to Philly.

Rob May: Yeah. But then you guys had a sort of typical startup story almost died a time or two almost shut down. Oh yeah. Ultimately made it had a fantastic exit. You and almost death multiple deaths. Yeah. Yeah. Yeah. And then you and your sister who you did the company with, Heidi, sort of became like New York legend entrepreneurs because you were you were at that time and place and then when that was all over and done, you decided to do a new company.

Stephen Messer: Yeah. Stupidity doesn't ever leave you no matter how successful you end up getting and how much luck falls on your side. You do end up starting again and again and you know this Rob as well. You have been pulled back in despite saying I will never do this again. You know humans have stupid biases that make them forget you know cognitive dissonance is a [ __ ] and you end up doing it all again but yeah we ended up starting Collective[i] all over again but for the listeners yeah linker almost died multiple times brutal deaths and we somehow squeaked by and then it turned into a cash cow that exists today we sold it to Rakuten Rakuten media is still linkshare producing ungodly amounts of cash on the original PHP stuff that we wrote years ago. They haven't changed a line of code and God bless them for that.

Ryan Eppley: And Stephen, what were the main moments for you where you kind of had a light bulb moment? You realized, okay, I need to course correct. This is a moment where we need to pivot and to find that new direction for link shares.

Stephen Messer: We never pivoted. No, we just we bared knuckled through. So, what happened in our world was there are two things that are challenging. When we first started we were a network model. Nobody gets network models. So, everyone was like, "This makes no sense." And we had a competitor at that time, Bree, which is a horrible name, but another Waltham company. And they would go out there and say, "I'm going to build you a software and you pay me a million or two million bucks and we'll give you affiliate marketing." And we were out there saying, hey, you just pay us two to three% of transactions and anything that flows through us, you'll get money. And the challenge of affiliate marketing is if you're selling a network, the first player you sell is like, "That's awesome. How many websites do you have? And you're like, well, after you, you know, and that sort of ends the conversation. And then we started getting people to sign up because it was relatively inexpensive.

And network effects models tend to have this thing where you give away a lot in the beginning to get people to sign up. And they'll take a chance because it's really low risk. And we started getting better and better at getting that. And but remember, two to three cents on a dollar that is you have to have a lot of transactions flowing through you. And we started taking off and growing very quickly in around 2000. And this is boring. Stop me. In about 2000 and we started getting really to a nice place where transaction volume is picking up. And I don't know if you remember, but in 2000 there was this little problem of dot companies blowing up pretty quickly. But prior to that, we were actually having a hard time selling people because venture capitalists were throwing huge amounts of money at Webvan and E*Trade, ETY, excuse me, and all these other places.

And they were they were they wanted to buy the expensive version of affiliate marketing because our competitor would say things to them like oh you're going to own your affiliates which is if you think about logically is insane. They're a separate website who's essentially being paid by you. And by the way I'm telling you the story because when we get to collective I it makes it ties directly in. And so and we were like yeah no that's crazy. We ended up winning that battle, but we in 2000 we lost about twothirds of our customers in the span of three months with about 180 days worth of payables. It was pretty ugly.

Anna Kirk: And you mentioned two of your biggest investors were Philly based. Now you can't walk down any street in Soho without bumping into a venture capitalist. What do you think has caused the New York ecosystem to change so much over the last couple decades?

Stephen Messer: So, I'll I'll give you my honest take. I think it was hard to get venture capital in New York because for the most part, venture capitalists are the C-minus students of finance. Well, I mean, I don't mean that to be mean, like I'm not saying it in a negative way. I'm actually saying it in a factual way.

Rob May: They'd probably agree with you.

Stephen Messer: No, no. I'll say because look, if you are a venture capitalist and you look at the economic model of VC and you compare that to any other asset class, you'd have to be mathematically challenged to say this is where I want to spend my time. Very few VCs end up with at the top making a lot of money. Most of the funds are generally poor performers. Unlike a hedge fund, you can't mark tomarket every year. So, you can't take cash out. You can't long and short like you can in a hedge fund. I'm giving the alternative asset class. Forget like debt which is like a much bigger asset class and even equity is much bigger. So like if you've gotten to the place you're like this is where I want to spend my time and make my wealth, you have already failed the basic finance. So I think and the reason why I say that's hard for New York is because you have amazing opportunities in finance. If you were a junior, like if you just graduated, you could probably $150 million hedge fund where you can do 10x borrowing easy and now you're magnifying your returns. So why would you do venture?

So I think what's happened now is a lot of people are realizing that between secondaries and between the lack of innovation around venture that you can be in New York and start innovating. And if you think about the minor iterations in venture that have changed the market, you can look at things like first round capital came out of Philly from an entrepreneur modified the venture world a little bit by promising you at the end of a meeting that you get a 50 or 100k check immediately and they'd be your lead and they stole people were flying from Silicon Valley to Philly which isn't a great airport. But they would fly there because they were like this that that's a example of where I think the east coast has a real advantage which is we think about innovation and can you make this market much more efficient and I think the idea for example that a venture capitalist has to stay with you through the AB and the C and that it signals there's a problem. No secondaries are normal now. Yeah. And that's why you're starting to see a lot more real investors coming in.

Ryan Eppley: Yeah. That is you bring up an interesting point and you think about just the ownership math and fund economics of that like it is an outliers business and if you are a VC fund that owns 5% of a company think about the exit that you have to have to rightsize that investment and IRR the biggest factor is time and you're building a model that is not time dependent if you're a seed and you can get out some secondary in an A and a B your IR goes through the roof. Yeah. You don't have to have those crazy all returns.

Stephen Messer: And nobody bothers to do the basic math because of this whole idea that I'm adding value by staying in. I frankly don't want a series A or B investor on a board of a C and D company. That that's not their skill set. They shouldn't be there, right? But yet they somehow think that's a value ad. I don't get it. But good for them. They are C minus students. Should I pause? I mean, you tell me. I'll keep going.

Rob May: No, no, no. I mean, I agree with you, right? It's part of We were talking about this at coffee before the show. That I've stopped doing a lot of early stage investing for that reason. Yeah. I mean, it's I think it's fine to be a small percentage of your portfolio or whatever, but you know, you can't use leverage, you can't there's just all these things that you can't do and it makes it harder. And particularly when I started doing it 10 or 11 years ago, your average holding period for an early stage company might still be 7 8 n years. And now realistically it's 12 to 15.

Anna Kirk: Yeah. And so we at Nadia, we're we're building we're incubating. So I'm I'm not taking any offense to this whatsoever, but I've just noticed out of our portfolio companies, most of them prefer to raise from family offices now because it's really flexible capital. Diligence is much more expedited and you don't even need to worry about going through the hassle of going through an IC. And I think we're going to see that more and more, especially as some of these companies don't need to raise as much cash to do what they can do.

Stephen Messer: Now, look, I know that New York will create the venture firm that only does common stock. And the reason being is in every single part of the finance world, the goal is to make things highly liquid. Venture is the only part of the market where they create such complexity it takes six months or more to do diligence or they just go in blindly and hope it's regular diligence but you have to do this this liquidation stack and all but that kills the valuation because with secondary getting stronger and stronger you really would rather see funds start getting created that say look we're going to play a sector of every startup in this category it's such little cost to do that if I did a Vanguard-like fund and I had a let's say a $500 million fund that just created liquidity around seed and a that arbitrage alone because the valuations fluctuate very quickly.

So you that would create such great algorithmic training opportunities for people looking for alpha and yet no one thinks about that way because they're like oh no no we have to do an investigation of this stuff and I had a great board member who who was at News Corp and I you know and this is like going back in time but he he was the guy who was in charge of Columbia Pictures when they did Slumdog Millionaire. I think it was Columbia Pictures, but it was Slumdog Millionaire.

And I would go out with him at night and I'd be like, "Dude, how the hell like like a film based in India with a heavy Indian accent in America? Like you thought that would be a home run? How how the hell did you figure that out?" And he was like, "Steve, it's very easy." And I was like, "Tell me, tell me." And finally, I got him drunk one night and he was like, I was like, "You got to tell me how you knew." And he said, "12." And I was like, "What is what is 12?" He's like, "All you need to know is 12." And I was like, "What the hell does 12 have to do with anything?" Is that like a lucky number or anything? He's like, no, you always do a film slate of 12 films a year and one will hit. And I was like, there's got to be more to it. He's like, no. He's like, that is the secret of Hollywood. They're always crap. One turns out to be great and you just don't know which one is that rare moment. He's like, we've tried doing everything from picking and doing all these matching and trying to put together people. He's like, no, it's just 12. He's like, I know it sounds crazy. He's like, I just learned the lesson and you don't sweat over it. And that is very much tech. You sometimes don't know.

Rob May: Is that why your fund does one a month or so? Is that is that the 12 a year?

Stephen Messer: We don't do any external investing. I mean, that's the reason why like it's a builder market and you need to be treating venture funds more and more like an index fund. I think is

Ryan Eppley: but I do like the idea of like this startup ETF, right? Just kind of, you know, arbitrage together a bunch of seed and series A type companies, you know? I have you heard of something like this on the market before?

Rob May: Actually when I was so I was a general partner at PJC for a couple of years and we looked into doing it and it was it's difficult. You'd have to be a bigger fund than we were to pull it off. And then you have to figure out how do you get how do you make sure you don't have a negative signaling issue, right? How do you make sure you actually get a good swath of investments and you don't just get like all the people that couldn't raise money from Sequoia?

Stephen Messer: So I think look I think if you were the f like what what you learned from first round is if you're the fastest money in the first one to put it down the best go to you first because you're just looking for as capital raise you're looking for speed. I mean that is sort of the if you're looking in the sequoas and the others they're just trying to get to know everyone so they can be the first money in and move fast. They're not getting any signaling other than I like this group.

First round was in the middle of godforsaken nowhere. The fact that they would move fat quickly and lead, which is like the fear of everyone, which is, "Oh my god, if I if I pick this wrong, and I thought they did a great job on that." So, I expect the East Coast will be better at that. And even in the end of the day, if you if you have sort of a negative signaling, it tends to be such a small amount versus the dollar amounts, especially when you're talking about seed A and B, it's really now you're more talking about like, okay, if I get OpenAI wrong and I'm in the, you know, because that seems to be the greater fool theory right now, which is like, can I get in before the next fool who buys it at a higher price? If you can get in there, that's sort of a different game. But I don't think people need to play that game anymore. You can essentially go buy on the market, that kind of stuff. I think it's more the lower stuff that you're trying to get into.

And by the way, I also think hedge funds would love to be able to long short that versus the public stocks. Yeah. You know, and I so I think that's the thing that will be the biggest innovation around software companies is you'll see liquidity around stock in a much different way.

Rob May: So going back to you know it's hard to raise money here in New York. Was in the early days and now you're in this scenario where I mean this is the number two AI market when you tried to get me to move here but I moved to Boston in 2010 because Boston was a clear number two after Silicon Valley in 2010 and sort of 2014 2015 New York passed it in terms of tech and startups and now is one of the biggest I mean after Silicon Valley the biggest AI hub in North America and I would argue better AI hub because it's very applied and the stuff people are doing. We're not obsessed about AGI and all that stuff here. And so what do you think happened? What do you think drove that change?

Stephen Messer: Well, one, it's nice to hear you say I was right even though you denied it for a long time. So, thank you, Rob, for that. I appreciate that. I don't need that for my ego, but thank you. Look, I will say there are benefits in New York again, which which is the large amount of capital, family offices, hedge funds, etc. I still don't think we have the heritage. We're way too conservative. I think if you look at New York, New York continues to grow really quickly. We will have a when we finally have a major breakout moment of Facebook, a Google, etc. And the people who make that much money are now ready to invest quickly because they know about it. I think that's when New York will break away from everyone else and its natural strengths. I mean, remember before World War II, New York was Silicon Valley. So, I mean, this is the California anomaly is thanks to Silicon Chips. But it's an anomaly and the reality is it's a great place to live.

It's wonderful. But they are an echo chamber that is both good and bad for them. I think New York is open to lots of other ideas that would not be acceptable in the valley. I don't think Collective[i] would work in the valley as an example. I think our current business works really well because we have thinkers here and sometimes people thinkers are good. I think they're both good and bad. So I have a theory about sort of what makes California work so well that some people don't like. I'm happy to share it. But I'll I'll tell you what I think New York has advantages over that, but it's up to you guys. I it'll piss people off, so it's up to you.

Rob May: Yeah, you can share it. That's fine.

Stephen Messer: Are you sure?

Rob May: Go right ahead. Sure. Half of this podcast's purpose.

Stephen Messer: All right. Well, look, I have a lot of crazy opinions and you can disagree. Let me have you on. But look, I think the interesting thing about California is that they are extremely great engineers who really don't want to think. They are the smartest dumb people you'll ever meet. They they don't broaden their horizon beyond anything else and they're always looking to follow what everyone else thinks is smart. So why do I think that's an advantage? Look in New York when you have an idea that's new, people really want to debate the idea. They want to and debate is the form of way you understand the idea and it helps you to think it through in a way that you can add to it and create some value and in Silicon Valley nobody wants to add it to anything they that's why there's a cult of Larry or cult of Mark the mere fact of I can say Larry and Mark and everyone knows the name should tell you how small of a pond that that place is because if I say Mike here who which Mike could it be if I say Larry it could be I mean there's 500 of them media mogul real estate developers.

Oh my god, Larry Silverstein. I can go through like a thousand of them, but there you don't. So what happens is people are in such a rush to be successful that they will blindly follow you. And do you know the speed of an organization when they're like Mark says and then they don't think they just go and do the best that they can do with their engineering degree and they will create the best you pick it design web page whatever it is. They'll never think they'll never question. And that's why Silicon Valley moves in these like memes where you're like okay like like even the meme of that we're going to talk about today which is like is AI like a bubble? Okay, like are LLMs a bubble?

Well, they can't even deal with the like okay are we having like any gray in this? It just has to be this is good. This is what I do and I'm going to be the best at it and they'll move so quickly because they don't need to know anything other than my part of this is to do this small thing and I'm going to do it really really well. And I think that's an amazing thing that they have. It's also their weakness. It's their Achilles heel. And where I think in New York the benefit is yeah, it takes longer to get everyone around the table to agree to it, but then everyone brings something to it. And so we tend to create much more interesting companies that I and we'll we will end up having our big breakout. Everyone likes to forget that Meta or Facebook didn't start in the valley. Started in on the east coast. And then it went there to get all the talent and to get followers who just follow and so but I think we'll we'll end up keeping one of them and really rock it.

Yeah, we've had some successes here like MongoDB I think is a big example. Data dog these are these are doubles and triples. We haven't had our home run and I think that's what I think the entire weight of the company country shifts because again no one's from California out there like Silicon Valley is an anomaly. It is a small capital base but they know what they're doing. They'll put money out fast. I think New York has way more money, not enough going into venture, but it doesn't matter. You see the sheer gravity of New York that we've become the second biggest and we have great companies, but it will naturally take its place again at the helm because all those people will fly and move here and they'll have we'll have the combination of both of those.

Ryan Eppley: What what is a home run? Is that a financial return? Is that a cultural kind of like phenomenon that happens? Like what is a home run?

Stephen Messer: Look, that is a company that throws off so much cash and draws so much talent that it is the gravity, it is the sun that pulls everything into its orbit, right? Like I said, people didn't grow up on in California and go to work at Google. They flew from everywhere in the world to be paid more money than God, to have the ability to have an impact on something and to be a part of something bigger. New York by its nature is a son, right? We are by we bring in the best from around the world who are the smartest who come here for a lot of different reasons beyond tech. And so if you had that kind of a of a gravity on top of New York, the talent would flee.

As beautiful as as Palo Alto is, one, you can't survive as a single person in Palo Alto. Two, it is the ugliest people outside of Boston that you'll find with no style. And the food sucks. So you end up in a place where and by the way, if you work in Europe or Asia, it is the worst of all time zones because no time zone fits. Whereas in New York, you have a 12-h hour time zone for Asia, six for for Europe is a great place to work, and you have the best of everything. So, I think what'll happen is we'll have our call it meta or whatever it is, and they decide to stay here because they love the New York area, and then everyone who's left will move over here. And I think that's the natural order of things.

Anna Kirk: It also begs a question like what what is the pull to moving to New York? And I do kind of see this natural story arc where especially during this period of time where we've been heavily innovating on foundation models, of course they're going to flock to the Bay Area, of course they're going to look for the best possible engineering and technical talent that they can. We are entering a phase of that journey where, and this gets into the theme of kind of our our core discussion topic today, too, where we're graduating to the application stage. So, we're getting to the point where maybe we're starting to enter the trough of disillusionment a little bit and we're starting to think through what is the unsexy stuff that we have to do to actually get this done. Which, which brings us to the first article we wanted to talk through today. Ryan, do you want to introduce that one for us?

Ryan Eppley: I mean, you know, I had to say before we hop into topics, just because I'm so, you know, excited about this is last show I said we were hiring two engineers, right? We hired our first one and I'd say almost half our applicants now are from Silicon Valley applying and the number one reason they say as much as I know they're excited about our company and they say that in the interview is they're like we want to be in New York, you know, is that something that how many how many times are are you hiring now and you hear people from Silicon Valley saying hey I hear what's going on in New York. This is very exciting to me.

Stephen Messer: Look, no matter how well you've done in California, in Silicon Valley, you're in a small pond. Like that's the hard thing. Like, like you asked me for a New York story, I'm going to do it early. Is it okay? Don't get in trouble. Go ahead. I'll give you a New York story. So, we sell LinkShare and it's a $500 million exit. It's all cash.

It's not a bad exit. And a buddy of mine runs a fund of funds, which for your audience is funds that invest in other funds. And he invites me to a dinner on the Upper East. And I never go to the upper east side because you know it's on the upper east side. And he he invites me to this beautiful mansion and it's a it's a dinner with the top 10 hedge fund people of the of that year and they're going through their ideas and it's crazy because they're talking about the monolines about are about to crash and I end up sitting next to Bill Ackman who funny enough went to my high school who plays tennis much better than I do. And we're sitting down and he turns to me and says I don't know you. I know everyone else here, but I don't know you. Who are you? And I say, I'm Stephen Messer. And he goes, he goes, well, what is it you do? I go, I'm in tech. And he goes, oh, what's your company? I say, LinkShare. And he goes, I know LinkShare. Now, this is one of those moments in New York where you're like, holy crap. This is awesome. Like, inside the belt. Okay, I understand. Inside the industry. I get it. But, but here's this like successful hedge fun guy.

Yeah, I'm feeling really good at myself. And I'm like, yeah. And he's like, he's like, yeah, you work with this one company that we are backers on, and they rave about you every board meeting. You guys are producing a lot of cash for us. Thank you. I was like, that's great. He's like, can I ask how big is the business? I was like, well, we just sold it. He's like, that's fantastic. What was the final sale? And I said, it was $500 million. He's like, my god, that's amazing. He's like, imagine what you could do if you put your mind to it. Now, you have to understand his salary for that year personally was a billion dollars that year. Wow.

He earned individually a billion dollars. And that's the thing about New York, which is okay, in the valley, I'll have a Larry, I'll have a Sergey, I'll have a Mark, but everyone else is in like this like cast system of like, where are you at in this world? And okay, if you're a huge VC, you're below the founder, but you're in a good place. If you're a top engineer, and they're in a cast in New York, I can sit at a dinner. And we took the subway back. He didn't have a car and bodyguards. Like we took the subway home. That's New York. Nobody matters here. And yet everybody does something. And I think if you want to be the best, you don't want to be in a cast system.

Ryan Eppley: Yeah. No, it's a good segue because the dichotomy for our first topic here from the Atlantic. How big is the AI bubble? You know, we kind of talked about it earlier, but I feel like we're we're in that duality where it's more real than ever. The impact we see from entrepreneurs making in the economy today is really incredible with all different LLM based systems. But at the same time, we're seeing a lot of things get funded that you're like, I think either they don't have revenue or they're inflating their revenue and we're unsure which it is, you know. So, what do you see on the ground right now in terms of bubble or real traction?

Stephen Messer: So let's define bubble in the private markets because that's very different than public valuations. To me, I'll define a bubble as when the technicals of the business have been taken overtaken by the financial gaming of the economics because private markets are much easier to gain. And so if we talk about AI as being in a bubble, that's not going to be that that's not true at all. If we're talking about LLMs, which again, there seems to be no more gray in this world. Everything's got to be like 100% one or the other. Somehow AI is like all LLM. And that is far from the case. I think if you had asked me and when Rob and I met 10 years ago, we were talking about how I don't want to do anything around language, image, audio, and video because it was by its nature always going to be a commodity.

And we were like, that reminds me of LOS, AltaVista. And you know, their job was to organize the web. The web was public. That's not a great business. Google won not because they had a better search engine. Google won because they stole the economic structure from a third party who they had to pay a lot of money for after, you know, before their IPO. But taught them how to make more money so they could always beat out. So, so the winner of LLMs will be if someone can figure out an economic model. No one's figured that out yet. But if I look and say if we're looking at AI generally and you get out of the LLM space and I'll talk to you about why I think LLMs are are in a bubble in a second but if you get out of it I look at China dark factories that is AI right they are now able to manufacture at a low cost lower than their neighbors because there's no labor anymore how what's the robotic situation in our manufacturing world zero.

So you've got a you have a very fast path up especially if god forbid there is a war in particular in ti taiwan you will see manufacturing with dark dark factories spring up but I think there is a complete missing of investment there because every factory needs to get there fast because we can onshore a lot more onshoring is important for us we've learned given the tariff situation self-imposed wound but wars generally have hurt us we saw what happened during covid there's a real reason for that. So is that done? No. If I look at war and robotics and war and AI the very beginning we 40% of our budget barely any of it's going to AI and has done a great marketing job but look at their there there's stuff there nothing there. So are we talking about this I could talk about Collective[i] we focus on on essentially a study of how the world does business barely scratching the surface of that. So there there are huge amounts of upside on AI. Then when we get to LLMs there's difference.

Here's where you see financial engineering at its best. And this is why I do think we're probably in the LLM world is in a is in for some pain. One is everyone's economically motivated because it's driving chip sales. It's driving which is important for the public markets because that's the only place they're getting growth. So, you've got to be able to show you can get growth there. I don't know if that's a bad thing. Everyone keeps talking about a collapse. I think if there's war with Taiwan, people would be glad that we have a lot of chips on storage. Two, I think you're seeing a lot of the private players who've raised a lot of money who now have to publicly talk about it being a bubble because they don't want another competitor. DeepSeek was scary enough. All the models coming out of China are scary enough and so they're trying to talk down competitors happens to be just when they're going public and they've also the day after they raise a lot of money. Thank god there's no one cynical here who would think that

Ryan Eppley: the reason I saw it was it was a public company you know Zuckerberg just the other day casually said even if we have to burn a couple hundred billion dollars it is worth it in this moment you know that's right teetering on hyperbole or is that is that a real financial return that is investors are excited about

Rob May: I so I was just at the AI infrastructure conference in San Jose and the VP of infrastructure for Meta spoke meta is a little bit of an anomaly in the AI world because there's so vertically integrated. They build like they're building chips that just drive recommendation engines, right? Which nobody else would like you couldn't sell those chips to the general public. There's not a big enough market, but they can design silicon at that level. They obviously have been building their own data centers for a long time. And then all the way they control the monetization. They control all the way up, right? They control the network. They control the monetization. They control every layer.

Stephen Messer: The revenue streams are very diversified.

Rob May: Yeah. And so they have a really interesting opportunity to benefit from AI in ways that a lot of other people don't. Yeah, it's a great point. And I think it's part of part of what's hard about like this article and hearing like are we in a bubble or not is like I kind of agree with Stephen on like sectors of the AI market definitely are like OpenAI is probably a bubble. I think I like it's I don't see how they're going to lean into that valuation. I've actually thought about, you know, since they've got this supposed $300 billion deal with Oracle, it's like, should you should you short Oracle under the idea that like OpenAI won't be able to raise the 300 billion to pay Oracle and, you know, stuff like that. Like I think that's a little bubbly. But there's definitely pockets that are doing f I mean, and when you get away from LMS, like people still use a lot of SVMs and CNN's and stuff like that, and that's going really well. Like machine vision in your normal applications is great. I humanoid robots is probably another place where there's a bubble. You know, these five 1015 billion dollar valuations. I don't know if they're going to live into that anytime soon,

Stephen Messer: but that's why it's why you heard me talk about dark factories, not humanoid robots cleaning your house. I mean, no, look, I think you're right. I mean, look, the thing about it is it's easy for Zuck and OpenAI and I'm going to say Dario soon to start talking about a bubble because they need to stop other people from raising money because it is a commodity. It is a horrible place to be. But also, you have to remember all of Wall Street, you're now getting past the point where it's an innovation. Now, it's a trade. And when once you're in trade land, what everyone's trying to figure out is how to move the market both up and down.

They need to create volatility. So, they're benefiting from this because you're going to have people wanting to talk it down before the IPO because they have to go IPO. And everyone gets excited by these things. So, what you're seeing is the traders are really playing this game. Like when when they did their raise was that OpenAI going to like a whole bunch of people and saying hey Larry will you will you invest 2 million? No they had Goldman everyone else selling it to their private wealth people. This is the greater full game and this is the parts of where the financeers are trying to figure out how do I pass off to the next person the tag so that I don't get stuck with it because it's a trade. It's not an investment. It's now a trade.

And so I don't know if bubble means the same thing. And we're thinking I think when people think about bubble they're like oh this thing won't really work and it was a fool's errand and they're trying to point to some problems of LMS or they're not growing as quickly as they should have been no I don't think that's true I think we can see people use it they're incorporating into everything it's becoming a big major part of society whether it be commodity is a different story the second thing is we're now in trade land there is no thesis other than hyperbole around will this be a good trade and I think that's the bigger problem that we have

Ryan Eppley: I'd love to hear and we'll we'll put you on the spot first. We'll let you have the last word of number one bubble within the sector of AI specifically and number one real traction across all the different verticals or horizontal opportunities. What's the number one thing that you're bullish about and number one thing that you're like, all right, this this definitely seems a little bubbly.

Anna Kirk: Oh, I think coding agents are a complete bubble right now. And I think part of that, and I'll talk about this in a bit when we get to our next article too, but part of that is this idea of do you as a company own the distribution or do you own the underlying rails that are creating value? If you own the distribution and the access probably that technology is going to get commoditized and that's exactly what we're seeing in this market. I think it's going to be a race to the bottom.

On the other side of that, I think we are entering and I wouldn't call it a bubble actually. I would just call it a market correction like you look at customers expectations not even diplomatic like I'm such a I'm so bullish about AI and the capability of it to change things but I think we are kind of reaching a point where our expectations are getting corrected and you saw that in the meter article to the Atlantic article that touched on the study results from this think tank meter I think we're reaching a period of time where the focus will become How is this actually adoptable within the industry? And there are a lot of unsexy processes that come to come along with like operationalizing that technology. And so I think that's where I'm really bullish. Like I think the people who really stick it out through this kind of trough of disillusionment to actually figure out how do we adopt how do we actually like operationalize this within a really big company where your data is a mess. You know, I think that's going to be the really interesting part and the companies in that next wave that can own the actual implementation phase will do quite well. What about you, Rob? What you got? Bullish on something, bubbly on the other.

Rob May: I would agree with the coding agents with Anna. I think that's a bubble. I think you're going to see there there's still a lot of upside in AI in areas that we haven't talked about much yet. So, like I think you're going to see edge AI coming and I think that's going to happen. I think you're starting to see the typical AI technologies around LLMs move to other parts of the stack in terms of what they're doing with the w with the ability to write code and I was on a call last night. I'm I'm part of a Google group that talks a lot about AI and where it's going and it's private. It's run by a guy named Adam Saw here in New York and everybody was talking about how they use these these things and people are starting to set up all these infrastructures that give these LLMs like put them in a box and give them lots of rules and guard rails and very complex prompts and all that kind of stuff.

And that's the kind of infrastructure that you have to build to get more out of these things. And this idea that like they were all going to be super gigantic AGIS that you could single shot prompt and just, you know, get everything that you wanted, I think is bubbly. And I think people are starting to move because it's not just the valuation. PE people are moving off the like AGI ASI bubble as well, which even six months ago it was like we're less than a year away. Yeah. And now it's like, well, this might be another decade or more before we build these super smart machines. It seems like they're still missing a lot. So, I think I think it's been bubbly. I think you'll see a little bit of a correction and but I'm I'm still I think like everybody else very bullish long term.

Stephen Messer: You know, I would say it's I don't not again I push on this bubble thing because really where the bubble is where you have traders playing the game, not as their intrinsic value. Look, you're not wrong on the Harveys and the I mean, look, you can look at Windsurf. They didn't want to take the company for the full price. They're like you keep it. I don't want it and went over to Devin and you know good luck with that and but like like that's if you're an investor you should be thinking to yourself well isn't that a warning sign?

But I think what it is just there's happy years which is you've had six seven years where no exits and if you see something go from zero to 100. Do you care if it's got barriers to entry or do you think to yourself all I need is OpenAI or Anthropic or Facebook to buy this because I'll get an exit. And I get that. That's why I said when you're in a trader mentality, you stop thinking about fundamentals. Like every time I ask someone on the west coast or east coast even, you know, where are the barriers to entry? They're like, look, barriers to entry we can figure out. That's a weird place to start saying, I'm going to put a multi-billion dollar valuation on something and we're going to figure out the only thing that matters. Yeah. You know, and I watched I mean, you saw Cursor try to raise its rates and everyone just say I'll leave. Let me ask you of the of the people here.

Ryan Eppley: I moved off cursor when they did that.

Stephen Messer: Okay, everyone moved off cursor. How about this? How many of you gone to the place where now you have Gemini, Anthropic, and OpenAI for free, and when you run out of credits, you just switch each day? Yeah. Yeah. Yeah. I mean, I'm there. To token margins are very, very challenging right now. I'm not paying the 20 bucks anymore. I get enough from like bouncing between them and I don't really care. There's nothing. In fact, they're all very similar. They're all very similar. Anthropics got an ugly ugly interface, but like generally they're all pretty much identical. I can't tell the difference and I don't really care to tell the difference.

And I think this is sort of where, you know, you're you're you're seeing the trading mentality. I think I don't know if there's going to be a trough of disillusionment because you've got three players and I actually think of the meta players in particular, they may look if there is an encirclement of Taiwan by China and she keeps saying he wants to do it desperately and now he wants to live forever. So eventually he will. Yeah. 2027 that's the that's what he keeps saying.

How smart will all the players like Meta who have a revenue stream look for stockpiling chips that they will probably resell three years later a month away from burnout but people will pay full price just to get any access to chips. So I don't know if it's actually a bad thing or it turns out to be fortuitous which is why you know bubble talk is about for trading not for the business that we live in.

Anna Kirk: There is the underlying question of like where will that value acrue even if the value does sustain and this is a I was really excited to talk about this article this week it's an article in Colossus magazine by Jerry Neumann called AI will not make you rich which to get a bit more macro compares AI to previous technological developments and makes a statement that like actually value is not going to acrue to most of the operators or investors in this space and if you weren't one of the early foundation model companies, you're probably better off kind of riding this wave, but it's not going to make you billions of dollars. And the idea is that, you know, value is going to acrew to the end customers. It's not going to acrue to predominantly any new operators in this space. So I guess my question for all of you, maybe we can start with Ryan because you're to my right. Do you think that AI is a disruptive technology or an accelerating one?

Ryan Eppley: You know, it's almost it's more important about how you think about the actual enduser implementation. It has nothing to do about the technology. At the end of the day, it's just is it easy to have someone say instead of having this water today, I'm going to pick up a different type of water and it not be a big change for me dayto day. I think I think everyone in going back to your point, Silicon Valley is full steam ahead. AGI is here. We're going to be the first one to get there. That feels very disrupting to anyone outside of the valley. But when we talk to customers and this is just a personal you know kind of understanding is we try to dim down how much we say this is AI and LLM driven. We just say hey this is a more modern developer tool for you. Yeah.

This is something that your team won't feel a huge change in their day-to-day but they'll just go much faster and be happier. So, I think it's an accelerator for the companies that are trying to be customer focused and just really really get to the end user's day-to-day lifestyle, not hey, we have the coolest tech ever and everything's going to be better for you. All you have to do is completely change your entire, you know, way of life in your company. So I think it's a big accelerant when you do it thoughtfully for implementation.

Anna Kirk: Stephen, do you have thoughts on this?

Stephen Messer: I mean, look, I'm not going to talk about LM because I think LM are commodities by their nature. So, trying to debate what's going to happen to a commodity. Commodity is a commodity. I mean, cars and airplanes became generally bad investments to make because they're high capex and low long-term value. We at Collective[i], we focused on an economic foundation model because we saw high barriers to entry, but also huge economic impact. So, an economic foundation model just studies how the world does business. So imagine if you're a big company, you may not know this. Every Friday they have to ask their salespeople, what will they close? And then they get back a very interesting answer of a manipulated how do I tell you a great story and nothing's real and so that has real impact on people. They either wait till the end of the quarter and then fire a ton of people or find out they didn't put in enough money in. So these are things that have real pain points.

If you can predict what's going to happen to their business with a high degree of certainty, well, now you start to change the way they operate. You can start shipping products like Apple, I'm sorry, like Amazon does, right? Amazon ships the product to you before you buy it. That's how certain they are. So, this can change an entire economic order. Well, that's a business where everyone has to contribute their data to get the value back. So, I think what you're going to start to see is I remember in 96, 97, 98, the only thing people want to talk about was search engines. And honestly, it's great that Google's done well and they've sold up, but does anyone really care about search engines? I mean, we were desperately looking for an alternative, which is why we like ChatGPT when it came out. There just hadn't been a better alternative because nobody liked search engines. They really sucked. So, I think this is where you're when when sort of the LLM realizations come in, you're going to see all the other businesses that are actually changing everything that are heavily AI. And that's when you'll really see what's been what's changing the market.

Rob May: I would say the with respect to that article like I definitely liked the containerization the shipping container you know analogy. I thought that was really good. I think it's I think in general there's always pockets to make money in any industry. Even if the general industry sucks, there's some piece you can peel off. It might be TAM limited, but there's some something you can do to peel off the most profitable customer segment and resegment that market or whatever. But I think in particular with AI, a lot of what you're going to what's going to happen is it's going to make it easier for everybody to do everything, which is going to lead to more competitive markets for a lot of markets, particularly a lot of markets that are digital.

Now, I think when you have networks and or when you have some kind of proprietary distribution, channel partnerships, relationships, like these kinds of things that are hard to get or that AI can't get on its own that require human relationships, that I think there's still a lot of opportunity there. I think if you play in places where humans are still going to want to show status over other humans, I think I think that I you know I always give people the example of like when cell phones came out, watches went away for a while because we used to wear watches to tell time and now was on your cell phone, but they sort of came back as status symbols and I think that'll that'll happen with some things because of AI. So, I do I think it'll be in general harder to find pockets to make money, particularly in tech, than it has been like SaaS investing was just so easy and crisp and 20 years of almost guaranteed return. Yeah. And the metrics just everybody knew the metrics and it was and so yeah, we're not we're not in that world anymore.

Stephen Messer: But SaaS also was a [ __ ] investment. I mean, you have never had a company that gets more than 20% market share. They almost have no market clout to be a get a premium. I mean, you can look at Salesforce as the best example of a bad business model, right? I mean, if if Larry and Sergey ended up with 20% market share, they would go to the top of the Chrysler building and they would jump off. They would kill themselves. And so it was it was a good it was again a really good model for a C minus finance student who could do a basic spreadsheet and figure out CAC and say, "Okay, and they're going to need a certain amount of money." But there's been no exits. I mean, your best option in SaaS was to sell the Thoma Bravo or Vista.

Rob May: Yeah. A deep payer and there weren't a lot of barriers to entry in a lot of SaaS businesses.

Stephen Messer: So yeah, look, I mean, Constellation Ventures was the greatest example of that of buying really small ones, but no VC wanted to go into those deals. Yeah. And those were the best deals to be in because there was no competitor really in those marketplaces. Yeah. But by the way, your point is 100% right about networks. You saw Sam is not a dumb guy. He was talking about for a long time trying to make ChatGPT some kind of social network on top of it because if you're thinking about competing against Meta who has WhatsApp, who has, you know, Messenger, who has Facebook, who has Instagram, you're probably thinking to themselves, if anyone's going to be able to create an economic model where there's lockin to their LLM, it's them. And you're trying to figure out how do I get out of it? And even their their crappy share version, you know, you could see was sort of a nod to Google and their productivity suites.

Rob May: Well, we're getting close to wrapping up and we end every episode with a best or worst of New York and or I today most interesting. So, our idea for today was what's the most interesting New York conversation you've ever heard? You gave yours earlier in the in the day, but one of you guys want to go or talk about it? Yeah. What what have you been easedropping on, Ann?

Anna Kirk: Yeah. First of all, everything. Second of all, this is a one-sided conversation. I saw a guy talking over the phone, and so I have no context for this whatsoever, but he says, and I quote, "I'll do what I have to do, but I draw the line at Machiavellian. I'm not willing to destroy someone." And I almost like that out of like with no context whatsoever. First of all, the man who said that, come find me because I want to know what the context was behind that. Second of all, which episode of Succession were you filming? Like, it just felt so classic New York Midtown. That's probably my favorite.

Rob May: Ryan, what about you?

Ryan Eppley: Yeah, you know, it was the other other Friday I saw, you know, a couple of the girls out and about ready to have a night out on the town and they were on the phone with it seemed like one of their boyfriends or something like that. And she's like, "You called a regular Uber, not an Uber Black. Do you know who we are?" And I was just like, "That's amazing." I was like, "Come on, guys." Like, you know, get in the Uber. It'll be fine.

Rob May: I so my so about a year ago I was walking down 7th A in Chelsea and I'm walking behind this mom and she's holding hands with this little kid and as I get closer I hear them talking and the girl's like no no no I want to stay up and they're they're you know late tonight and they're arguing or whatever and she's like can we watch a movie can we watch a movie and the mom's like sure and then the girl goes can we watch one of the movies daddy made and I was like oh so like I'm like who is this is this like a director's wife or an actor's wife I mean I wouldn't know anybody like that. Yeah, possibly like, you know, I don't know very many celebrities. Although I did like literally almost run into Alec Baldwin one time in West Village when he was pushing a stroller. But it was so like I walked at that pace for for two more blocks with them so I could hear the conversation which was fascinating because clearly this girl this little girl who was probably seven, eight, nine years old knew that her father made movies and I was trying to like get more information to see if I could tell who it was but it was really cool conversation over here.

Anna Kirk: Did you figure out who it was?

Rob May: I didn't. No.

Ryan Eppley: So, but he is now on a watch list for

Rob May: Yeah. Finally. Yeah. Finally, the mom was like, "Hey, what are you doing?" I didn't walk that close. Cool. Well, Stephen, thanks for being on today. This was this was a lot of fun. And for all of you that are watching, thank you for watching. Do all the things that you're supposed to do, please. So like us, subscribe, send us requests for things we should talk about, guests we should have on the show, and yeah, we hope to see you next episode.

Stephen Messer: Thanks for having me. Thank you.