Building AI-Native Infrastructure for Developers | Erik Berhnardsson, CEO of Modal

20:08I think now we're like, we can like reliably predict like seven days out. I think with like better models, we might be able to push that to like 10, 12, you know?

20:16Turner Novak:Yeah. And that's going to be like, take so much more compute, but like, it's still like, you know, 12 days out. It's like, that's good. That's pretty good. Yeah. It gives someone 12 days notice on their, on their rainy wedding. They can, they can plan around it. So going back, like I usually like to talk about when you first started the company, but I feel like the story kind of starts probably back in Sweden. I know we were talking before we started recording. You grew up in Sweden. What did you like to do as a kid back in Sweden? Oh, I like to code. I started coding in 1992. So my parents had an Apple Plus, Macintosh Plus.

20:52So I started coding and it was like HyperCard. And I was like, do you know what that is? It's like a multimedia thing, kind of like Visual Basic in a way, but like Mac. So I figured out how to write code and that was very basic, simple. But yeah, so I actually think a lot about how grateful I am because back then I had no idea that I could make money from it. And I was thinking about my friends, they were trading hockey cards. That was their hobby. I was writing code and like, look at me now. I'm making a lot of money from my hobby. I'm actually very, very grateful about that in a way. So I just kept coding throughout my childhood.

21:27I switched to JavaScript at some point and Perl around late 90s, started doing C++. I discovered the programming competitions when I was in high school and discovered I was actually pretty good at it.

21:43Turner Novak:So what is a programming competition, especially back in, sounds like probably late 90s, early 2000s? This was early 2000s. You typically have like five hours to solve a bunch of problems. there's like slightly different formats sometimes it's like solo teams sometimes it's like three people there's a bunch of different formats but but the core idea is like you get to solve you have to solve like i don't know half a dozen to a dozen problem in like a few hours or even sometimes just three problems uh and it's usually like very algorithmic problem they're very artificial they're always they're like super silly it's always like there are n cows on a farm and, you know, there's a graph that describes their movements.

22:24It's stupid. But, like, you know, they're fun. I mean, they're, like, fun puzzles. And I don't know. I discovered I was pretty good at it. So I actually went to, like, this international... It's called IOI, International Olympiad in Informatics. It's my first time to the U.S. I went to Wisconsin.

22:44Turner Novak:Wow, the pride of the U.S. Yeah, yeah, yeah. America's their land. Yeah, yeah, exactly. In the Midwest, like right in the middle there. That's funny. That's the first thing I saw from the US was Milwaukee and something like that. Kenosha. This is Chicago. Anyway, so I did pretty well in that and I kept doing that throughout college too. And at university, there was a bunch of other people doing it too. and then they joined this random startup called Spotify. And I was like, what are they doing? Like music? Like music streaming? You know, I don't think that's ever going to work. Yeah, isn't it illegal?

23:27Turner Novak:Like it was illegal at the time, right? You know, like I don't think it'd make any sense. But all of my smart friends from school joined that company. So I was like, okay, I guess I'll join it too. And obviously now it's like, you know, whatever, Europe's biggest startup or something. But I think of it as like so random. Like I kind of ended up there when it was about 40 people, I think, and ended up staying for about seven years. So that was, you know, I lucked out. Yeah. And you did your master's thesis on music recommendation. Was that? I did it at Spotify. Oh, you did it at Spotify. Okay. I did it at Spotify.

24:00Yeah. Yeah. And by the way, I didn't know anything about music recommendations. I was, I just, I was able to convince them. I'm like, hey, like, can I just come and like do some machine learning? And like, you know, I didn't know anything about machine learning either. but they're like yeah cool you seem smart it's just a math problem right at the end of the day

24:16Turner Novak:yeah a lot of math problems it's like a whole stack you know 20 math problems together so i did that and i i chose spotify like full-time early 20 2009 no wait yeah whatever something like that 2000 yeah so i ended up like initially like i was kind of ended up like working like business intelligence which is actually kind of in hindsight i'm just getting very happy about But because they were like, you know, like music recommendations are cool, but now we have all these more pressing problems. Investors are asking for all these numbers. Actually, Eric, you know numbers. Can you do that? So I ended up spending like a year just like crushing numbers and just like generating charts for like investor decks and stuff like that.

24:52But I ended up being like super valuable like 10 years later when I was running a startup.

24:57Turner Novak:Oh, I've done this before. Yeah. Yeah. So it's kind of fun. Anyway, but eventually at Spotify, I ended up going back to music recommendation systems and built up a team around that. So what's the most interesting thing or surprising thing about music recommendation systems that people might not know? Because I always, I think it's Spotify. I think of the Discover playlists. Sometimes I find new stuff from, how do they work? Like, what would people want to know about that? I think that maybe like people get a little bit surprised, maybe not, is that the recommendation systems have like no idea what they're like crunching.

25:33They just look at a bunch of correlations. Basically, you ended up using a lot of language models, but not in the way you might think. The language models know nothing about music or lyrics or anything like that. You just think of individual songs as words, and then you just look at the sequence of words, basically, like sequence of songs that people listen to. And then you're just trying to find patterns in that. It's essentially like a next-token prediction problem, where the words are like individual songs, But the songs are just like IDs. You know, they're just, you know, in a big matrix, they're just like entries in a big matrix, essentially.

26:11So it's called collaborative filtering, basically, this idea that, like, you're just mining this, like, data set for correlations. And then basically how it works in practice, it's like you create a big matrix, then you factorize this matrix, and then you end up with vectors. So you get, like, vectors for the songs, and vectors for users, and playlists, and artists, and other stuff. And then I ended up, I actually built a vector database at Spotify, too. open sourced it for a while. It was actually quite widely used, briefly used by Twitter and all these other companies. But yeah, so, yeah,

26:40Turner Novak:then you have to do like vector search. But yeah, that's like the core idea. It's like you put shit in a matrix, you factorize that matrix, then you get vectors, and then you do like near-as-neighbor search in that like high-dimensional vector space. I think I read a blog post about how Spotify's Music Rec worked a while back. And maybe it was like a different version of this, but it was something like it basically just looks at who do you have a similar listening history or habit to? And then like, what are they listening to? That's roughly right. Yeah. But you do it in a more statistical, like less direct way.

27:12It's like, it doesn't exactly do that, but as a sort of like mental model, roughly. And then what did you learn about just like

27:19Turner Novak:building a culture at Spotify? I think it was super formative. Spotify in the early days, there was like zero management. At least for me, I didn't even know who my manager was. I think that was actually bad. Oh, geez. That could be bad. For the first two years, no one told me who was my manager. So I was like, and they're like, that was my first job. So I didn't know what to expect. So, but I think it was actually, I ended up working out because there was a lot of smart people just like working on like hard problems and people knew what we kind of needed to build and we didn't need a lot of structure.

27:50So I always took that with me. Like I think most startups, like if you just have smart people who are like pretty autonomous and have like a mission, you need a lot less structure than you think. You know, you can get by like people, people should just come in every morning and just ask themselves, what can I do for the business today? That's my sort of platonic ideal for what a startup should be. You know, it's a little naive maybe, but like, you know, aspirational maybe. You can get pretty close to that actually.

28:14Turner Novak:Yeah. It's kind of like the whole like, what did you get done this week thing? It's just like, did you actually accomplish anything? Like, go back, look, like, did anything change? Like, did you materially make a difference on anything? Like, if not, maybe revisit how you're spending your time. Exactly. But that phrasing feels like kind of like negative. You're like, what did you do today? It almost seems like you're expecting, did you actually do anything? So what do you think is the correct phrasing? It's just go in and get shit done? Is that the right way? What can I do for the business today?

28:46Turner Novak:What are you waiting for? Just do it. Have you learned anything then from the Spotify days related to modal? A million things. How to build software, how to hire, how to... I think another thing Spotify did really well, it's kind of random, but I think everything that Spotify did everything like it had like high taste like i feel like the design the websites like how we talked about the product it's just like it's like a certain level of quality and i remember like the first time you tried the product like you clicked on a track and you're like wait a minute this is like in the cloud like that's sort of like magic experience of like whoa like this is actually like that's actually something you get in modal too like you're like running some code you're like wait a minute this code is running in the cloud and like having that like kind of magic moment and then like the whole kind of product and onboarding like kind of designed around like getting people to that magic moment and then and like keeping the bar high for like what the product looked like and and the design sometimes spotify kind of lost that a little bit but they eventually always came back to it like just you know it just spotify has this like they was like postured as like a like a bigger company than they were i felt like and i think that's quite important like you like visual language and how you talk about it and the identity and your voice like you gotta you gotta seem like you're like a thousand person startup even though you're like 10 yeah the the design aspect reminds me of did you ever use like lime wire or any of those like yeah all of them yeah like you do like search a song and it'd be there'd be like a hundred versions of it you like download probably a couple and like see which one sounded the best to keep versus with Spotify it's like you just know you're getting the song and that's going to be great and there's not going to be like some guy made a mist a mixtape and he has like his you know his like intro credits rolled over it or like somebody sometimes it's like a different song I mean like people would put like fake songs sometimes and like like sometimes I download it and I listen to an album but it turned out like two of the songs were actually like a different like the wrong song and then when you switch to Spotify you're like wait a minute is that the action, is this the real song?

30:51Yeah. That happened to me a lot. Because people just pollute. Maybe the labels hit it just to undermine streaming or downloading.

31:00Turner Novak:Yeah, just to get people to keep buying CDs for an extra year or two. And then, so I think you kind of moved over to the US at some point while you're at Spotify. And then you helped start this company called Better.com. That's right, yeah. Yeah, real quick for people who don't know what Better.com is. What was that one? It was a mortgage lending startup. I think at some point I was like kind of, I was at Spotify for seven years. I was just kind of burnt out and wanted to try something new. And I was like, I'm going to try something completely different. So I ended up starting as part of the founding team at Better.

31:31Yeah, I think we're like 10 people when we started. But yeah, that was like a total roller coaster. We're like 10 ,000 people when it left six years, five years later. And then the company basically collapsed.

31:42Turner Novak:It's public now? Like they got it public at some point? Yeah, to the credit of the founder, Vishal, they were able to take it public a few years later. And randomly, actually a couple of weeks ago, it kind of turned into almost like a meme stock. Stock went up. Yeah, I was looking. It looks like Opendoor and Better are both kind of like going vertical. It's so bizarre. It's so bizarre. So I'm actually kind of bullish on the company in a way. I do feel like there's a space for innovation in the mortgage lending. I think with interest rates high, it's tough. And that's ultimately a lot of what made the company almost collapse.

32:22But I think it's still worthwhile. There's a lot of stuff in the housing industry that's just fundamentally extremely broken in the US.

32:29Turner Novak:Yeah, and it's a very large percentage of GDP. It's enormous. It's crazy. And for most people, it's the biggest purchase that they'll ever make. like for 90, I don't know, 98%, 99 % of Americans in most countries. Like your house is your biggest asset. You literally, you work for a decade to save up, to make a down payment. You put it in the house, you pay off the loan, and then that's your retirement. Like it is everyone's like fundamental, just like the value of their life that's accrued is like the value of the home. No, it's crazy. It's so large. And so there's a lot of good reasons to think about this problem.

33:05But like ultimately, I think tech is just like one part of it. And I don't know, I was a CTO there for five years. I built a tech team. I took it from basically one person, like 300 or something like that. I learned a lot about management and hiring and all this stuff. Took a lot of stuff with me to Modal. But I don't know. Now that I'm in Modal, I'm like, this is what I want to do. This is so much more fun. Better respond to.

33:29Turner Novak:It seems like, based on what I know about you, this might be a slightly more interesting problem for you to be working on. But still, both important. And you mentioned you learned a bunch of stuff. I'm assuming you've interviewed thousands of people, hired hundreds. Maybe you've hired 1 ,000 people at this point. I think I've hired maybe 400, 300, 400, probably something like that. Wow. Okay. So just thinking back over the past decade or two of this, what have been the biggest predictors of on-the-job success from an interview? I know it's extremely difficult, but you have a blog post on your blog that's basically like, you need to treat this as almost like a prediction problem of like figuring out probabilities of like what actually leads to success exactly like that's what it's for what it is right like you're trying to predict their future performance so i always feel like people are like when they're criticizing interview processes they're like why are you doing these like weird brain teasers like whatever like who cares like whatever predicts like future performance maybe brain teasers don't do that And that's like a valid criticism, but don't like, don't throw them out on the grounds that they're not like, you know, realistic work.

34:38Cause like, that's actually besides the point. So, so, so the real thing here is like to predict like future job performance. And so what actually predicts future job performance? I don't know. I think it's important to be like humble about it. Cause like, I like the more I interview, actually, like, I think there's like a false sense of confidence in my, in the early days of hiring people. Like I thought I knew I was like, yeah, this person is great. This person is not good. Like I've just realized it's very noisy. It's very noisy. It's very hard to predict. But the things I've learned, honestly, like the biggest thing is actually intelligence.

35:10And it's kind of boring, but like, and it's like hard to assess. But like when you work with a person for a long time, you generally like observe that like smartest people, they'll always like just do better. So that's like kind of a boring answer. I think beyond that, like kind of after intelligence, like what else like predicts it? I would say like at startups, at least like I've seen this sort of like high agency, commercial, autonomous, like autonomy that that tends to like be quite important.

35:40Turner Novak:It seems like it would kind of relate to your earlier point about just like what can you do for the business to like move it forward kind of a thing, like find an opportunity, close a deal, make money, you know, save money, you know, insert whatever, moving business forward, commercial objective. like being a little impatient like i like people who want to just like get stuff done like they want to you know they're high agency and they just get stuff done so i think that's pretty important there's a lot of stuff like that people are like fundamentally positive like i i hate when i interview people and they're like complaining about stuff like i think that's kind of a bad negative trait i don't know there's a lot of stuff like that i often like ask just like almost like seemingly random questions but to people and trying to see how well it holds up So what's the point?

36:22Turner Novak:What would be an example of that? Ask me one. What's the hardest thing you ever did? Tell me something you did in the last year that you're proud of. Me personally? Yeah. We're role playing here. I mean, I figured out how to start making money from this podcast, which is not easy. And how do you make money? Sell sponsorships in the podcast. I mean, it's super simple, but also... And how do you sell sponsorships? Like, how have you learned about, like, sourcing sponsorships and winning those deals? It's actually a lot easier than I thought it would be. It's just, like, a volume thing. You just got to be like, hey, I have this podcast a lot of people listen to.

36:59Turner Novak:Here's the data. Okay. What's the conversion rate when you reach out to people? What's the percentage of, like, sponsors? Anyway, my point is, like, I just ask, like, questions like that. And I just, like, go super deep. Some people like that, though. Some people hate it. But the reason why I like that question is smart people, they're able to go almost arbitrarily deep. But when people are starting to really hold up, that's when I get a little hesitant. It turns out actually they don't know what they built themselves. There's a lot of stuff like that you can do. I don't know. But in front of the interview is hard.

37:36Turner Novak:Yeah, I actually want to finish answering that question because you'll see why in a second. And what I found is that the people that are most likely to convert on sponsorships in the podcast are people who follow me on Twitter or listen to the podcast. So it's like they kind of are, you know, like when I bring it up to them, like, oh, yeah, I'm in. And it's just you don't there's like not really a big back and forth. And you have to like sell them pictures, just like they've been listening to it or they've been following it. So if you're listening right now and you want to sponsor the podcast, respond to the email, comment, DM me on Twitter, whatever, whatever you need to do.

38:09So by the way, your answer was like a great interview question answer. Like it sounds like it made me seem like you've thought deeply about the problem and you observe patterns and like you come to realize something. Right. Yeah. I think just generally sometimes just trying to go out of your comfort zone is so hard

38:31Turner Novak:sometimes. Like for me personally, like I just never done it before. And you're kind of there's almost this there's somewhat of like a fear of failure and you just have to kind of do it. And then once you do it, you're like, oh, this is actually way easier than I thought. I email someone, we got on a call and they gave me like 20 grand. That's pretty cool. That feels pretty good. Like it's an extremely high ROI. So then it like motivates you to do it more. So I think it's kind of like for any hard thing, it's like figuring out like what the dopamine like unlock is or what like the part of it that you like and just like figuring out how to get to that point.

39:05Turner Novak:And you just say you have to do the hard stuff. I mean, there's every job that you have. Like one of my first bosses told me this. I worked at a private equity firm as an intern and my job was literally to cold call investment banks and be like, hey, do you have any deal flow for us? It was the stupidest thing. It was the 0 % conversion rate. People were like, why is this like 20-year-old kid calling me asking for companies to invest in? This is not how it works. Yeah. Sales is like that for me, by the way. Like I had to do a lot of it. By the way, that's why Mormons are so good at sales. Like I love Mormons because they have to like go and like go for like three years to Europe.

39:44Yeah.

39:44Turner Novak:And like just try to convert you to their religion. That's a little different than most other religions. I admire that so much. You know, so many people saying no to them and they just like learn to like, okay. And you know. Yeah. I feel like that's the thing is like when you're selling anything, it's not like nothing's a hard no. And if you're just getting hard no's, you're almost doing it wrong. You kind of think of it more of like a discovery process or like a, what's their problem? Can you solve it? And in some cases, they might just not have your problem. If I'm a restaurant owner and it's like some one-off non-chain, I sell Indian food and you come to me and try to sell me modal, I'll be like, no, you shouldn't even be talking to me.

40:27Turner Novak:It's just about figuring out who actually has the problem. And then the bigger the problem is, and the more relevant, that's who you should be fishing for. I think it's like you're bringing up a really interesting point about starting companies, by the way. I think a lot of founders make this mistake. They go and talk to a lot of potential customers, and then try to land those as design partners. And people are like, yeah, this is not quite what we need, but if you do this thing, then maybe we would do it. If you support on-prem, we would consider it. And then people are like, oh, on-prem, okay.

40:58So we need that as our MVP. And people end up with this ridiculous MVP. You should disqualify a lot of people when you're talking to potential customers.

41:07Turner Novak:Really? So isn't maybe a good transition into modal, but so when you were starting it, did you get that a lot? No, I actually took a terrible approach in hindsight. I just decided, I'm just going to take my own experience and just sit for two years and build a product that I wanted to have. that's a very different approach. But you probably tuned out a lot of noise of like somebody, somebody might have given you feedback like, oh, we need on-prem and you're like, but it's not, we don't need that. Screw that. Yeah. So it's almost like you were your own customer really at the end of the day, like you're building the product.

41:41Exactly, right. And I think that's maybe different because I maybe started my company a little bit later in my career because like I felt like at that point, like I'd already, like I kind of spent enough years like on the application side that I kind of knew what I wanted from the infra side. But yeah, I think for most people, it's probably better to spend a bunch of times talking to potential customers.

42:02Turner Novak:Yeah. So then I know you described it as like you were just kind of like heads down building for like a year or two. What was that like? Like, did you, were you trying to find market pull or were you literally like 16 hours a day, don't talk to anyone, just building it? And then you got to the end of the tunnel. I quit 16, but because, you know, I have kids and a wife and stuff. But probably like 12, right? Yeah, something like that, right? So that was just, yeah. let's also remember like this was during the Zurb era like kind of late pandemic so like you know we took a little bit of seed money and people like kind of like yeah that's cool like you're building some hard stuff but like that being said like I do feel like there's a little bit of like sometimes disconnect where like I think if you're starting a startup you should pick a hard problem like you should almost like pick something that like kind of takes like a year to two you know like I don't know that's the challenge I have with when I look at a lot of YC startups they're pushing them so hard to find PMF like instantly like within a week or two yeah like I'm like that doesn't happen for startups and so I think it leads to a certain amount of startups like certain type of startups but like that's not the startup that I wanted to start like I wanted to build something that takes like a couple years to build because then I know there's a little bit of a moat also you know there's something more durable there because if you build modal like oh seems cool.

43:19Like, I'm going to build it too.

43:21Turner Novak:Like, I'm going to do it like vibe coded in a night and then start stealing your customers. Yeah. So you got like a year or two of just building this thing and it was stable diffusion was like the first thing where it really started to take off. That's right. Yeah. Like we sort of targeted a kind of, we always have this like very general purpose approach. We're like, we're going to build something that like works for everyone that builds stuff with machine learning data and AI. And we weren't really sure. Well, what's the killer app and that turned out to be Stable Diffusion, which I think was somewhat serendipitous because Stable Diffusion wasn't around when we started the company.

43:54So we ended up being a little lucky in that sense. I think we would have found something else, but I don't think revenue would have been where it is today if it hadn't been for all the Gen AI stuff. So we were a little bit lucky in that sense.

44:05Turner Novak:So then how has the product evolved over time? Was there a specific first thing that you built that people were using or how did you ladder We started out with this very general purpose product, right? And like I said, we didn't really know what was the killer app. And then Stable Deficion ended up being that. And then that pulled us super hard in the direction of inference and specifically Gen.AI inference. But in a way, we've kind of gone full circle because I think in 2025, roughly, we kind of came back to the idea of like, no, this is a general purpose platform. And we also have way more engineers now.

44:42So we can build more stuff. Yeah. I mean, when you have like five engineers and like you find product market fit, you don't have a choice. Just like chase that shit. Like just, you know, like just go for it. Right. Like, but when you have like, you know, now we have like, I don't know, 45 engineers or something like that. Now we can actually afford to also like build a little bit more like kind of speculative stuff. And so we were kind of very much going back to this original idea of like, no, we're trying to build a general purpose platform. We're trying to build an infrastructure layer. And there's so many other parts of that.

45:10We're focusing a lot on training. We released a notebooks product recently. Sandboxes, like I said.

45:16Turner Novak:What's notebooks? Basically like the idea of like Google Colab or, you know, Jupiter or things like that. Like there's more idea like you have like kind of a graphical like instant feedback thing in your web browser. We can just write some code and like plot stuff and run code. and get more instantaneous feedback from inside the web browser. And you can share that with other people. It's not really necessarily meant for production use cases, but it's meant for more interactive exploratory stuff. Like testing things? Yeah, like playing around and trying out quick and dirty prototypes and applauding stuff.

45:51Interesting.

45:52Turner Novak:So then as you were kind of building the product, I know you mentioned earlier making it super simple in some cases, but also infrastructure is hard. You also mentioned all this stuff is hard. How did you decide when to make the product super simple and kind of abstract things away, but then also kind of keep it complex when it needs to be complex? Was there any kind of secret to that? I don't like the way simple. I also don't like the word democrat. There's a bunch of words I try to avoid when I talk about the product. Because I think a challenge with a lot of infra products is they take away a lot of the possibilities.

46:33You want to reduce the burden on the developer by 90%. You want to collapse the amount of effort it takes to build these things. But if you're also removing the ability to build stuff by the same amount of percentage, then there's no...

46:48Turner Novak:It's not worth it almost. Yeah. So I think you want to retain, I don't know, 90 % of the possibility space while reducing the complexity by 90%. So I don't necessarily think of it as simplification. I think of it as finding the right abstractions that enable people to build things faster. And we always focus very intentionally on, I think it was high code use cases. We go to the power users, the machine learning engineers, the people running the models, training the models, running really complex stuff. We look at what they're doing, and we want all of that to run a model. but we want it also to take a lot less effort.

47:26Turner Novak:So it's hard to find those abstractions. I think in a way, like by sacrificing a lot of the backend workloads that helped us focus a lot more on AI and that like, you know, I think it's very important that like, at least for me, like or how we think about model is to pick a persona and like a problem space and think about like, okay, what exists in that problem space and build for that. And when you pick a problem space, then I think it gets a lot easier to make those, you know, simplifications that I don't like the word. This is kind of a challenge with the cloud providers. The cloud providers, by the way, you know, the hyperscalers, they're trying to build for everyone at the same time.

47:59And if you build for everyone at the same time, it's actually really hard to like, to optimize for any particular audience.

48:07Turner Novak:So yeah, so it sounds kind of like to your point when you're at Spotify, it's like you spent so much time just setting things up and building the infrastructure. So it's like making it so they don't have to spend setup time to like customize an environment or like a production run or a training run, like whatever the thing is. Exactly, yeah. So they can still do specific, like unique, hard to do things, but just like abstracting away the manual, boring, less creative aspect of it or something like that. Yeah, keep the power, but like reduce all the garbage. So then was there certain points like on the roadmap, like are you getting a lot of feedback from customers?

48:45Like how do you decide when to listen to them,

48:48Turner Novak:what to build? Like, has that been a big thing you've had to figure out? Or is it still just kind of like you guys kind of know? No, we talk, now we talk to, like, once we emerged from that cave, then we started talking to customers all the time. And we have a big Slack community. Yeah, how do you decide what to prioritize? It's very hard. I think when you talk to customers, you should take their pain points extremely seriously. But when they suggest solutions, you should almost, like, ignore them. I've heard that before, yeah. Because, like, customers will be like, oh, yeah, like, can you do this feature for us?

49:20Like, we'd really want that. And then like, you kind of have to ask them, well, what is the problem here? Like, what are you trying, and then kind of go back like a few steps and then you identify, okay, they're trying to like, they're having an issue with this. And that's, you know, and then you have to think for yourself, but like, maybe there's a much better, more general solution actually. And so, you know, and then the second problem is like prioritization. And by the way, prioritization is like, just incredibly hard. And like, I think there's no way around it because like, we have 45 engineers right now.

49:46Of course, we can do more stuff. But by the way, if I had a thousand engineers, I would still struggle with, I would want to build a moon rocket and whatever, space laser and all these other things. It just never ends. So you just have to be super ruthless about picking the two to three things that are existential for the company and just like, that is probably where you should spend time at any point in time.

50:10Turner Novak:And you mentioned that just kind of like the rise of the importance of inference changed a lot. What kind of happened there? Like what were people trying to use it for? And then how did that kind of impact the company? Yeah, so all these generative media models and other things like large language models and things. And I think a lot of the focus traditionally on AI, at least in the last five years or so, has been on training. But as like people put these models into production, they need inference. Inference is a little bit harder because, you know, unlike training, where you're actually kind of fine with having like a fixed amount of compute resources, like a static amount of GPUs.

50:46But when you're doing inference, you can't fundamentally predict how much. I mean, you can estimate, but you have very volatile demand. So one of the kind of value propositions of Motorola, like part of the reason why I think we found a good market is that in inference, is that we sold that for you. And so you don't necessarily have to think about capacity, right? And so there was a sort of realization, like there's kind of a fast-growing market here in inference, which is why we invested so much in the last few years in inference. We still invest, by the way. There's so much still to be done.

51:17But that's why I think inference has been a good space for us.

51:21Turner Novak:And is this a good way to think about inference? Training is just your brain, and then inference is using your brain or something, or it's doing things? I think that's right, yeah. Okay, cool. Yeah, I'm always trying to get a better understanding. Plus, also, sometimes people will reply to the emails I send. Then they'll be like, what does this thing mean? And it was a basic concept in the conversation. I was like, oh, fuck, I probably should have explained that. So I like to try to, as we go, be like, hey, this is actually what inference is, by the way, for people who don't know. And so then when you think about the future of modal and AI development and all this stuff, where is it kind of going over the next couple of years in 10 years, 20 years?

52:05Turner Novak:I don't know if that's even possible to predict, but how do you think about how this stuff is? Yeah. How's it all evolving? I always thought of modal as my 20-year project, by the way. So like for me, it's always, you know, I always made, you know, I have a lot of ideas. I don't know how it's all going to unfold. I think of like, there are so many things around like AI development and like, or like more broadly, like working with just like compute resource intensive problems. That's kind of goes beyond just training and inference, like this entire life cycle, right? And when you start to think about that, like there's many other things, there's like data pre-processing, which we actually have some customers doing this training.

52:43There's inference, but there's also other things, like a lot of observability. There's like features, stores. There's real-time streaming data pipelines. There's, you know, eventually, if you want to start working on more like data stuff, there's a lot of things around query engines, like working with like more like databases, more like primitives for like structured data. So I think, you know, when I look down the road, like three to five years, my goal is like, if you're building an AI application, you can do the entire thing on modal, like the entire thing. We have all the software tools you need and we do it, by the way, much better than anyone else because we control the underlying infrastructure and it's all like super well integrated too.

53:23So that's kind of my like hand-waved addition for what I want to get to. So it's sort of like traditionally,

53:30Turner Novak:if you're not using modal, there's almost like multiple tools you have to stitch together and there's like trade-offs between how things sync versus modal. You're almost like the full stack of it. so you will get a better, more succinct, faster experience. Are there any ways you might fall short with that approach? Like that someone can provide a different or a product that might be considered better if somebody wants everything like cost or something like that? I mean, I think cost is always going to be... We try not to compete too hard on cost. And that's always like fundamentally, I think if people want to run something at like absurd scale, and they're willing to do all the hard work that it takes to like get the absolutely lowest cost yeah i mean probably like they're gonna be able to like go buy their own data center and like you know whatever buy their own power plan and like like if people are willing to like spend an enormous amount of effort like they can always lower cost so so we try to strike like a happy medium where like most companies are probably better like way better off using model because they can just iterate so much faster and i think ultimately that's like what people care about it's like, you know, maybe it gets like 30 % more expensive at modal, but maybe that's fine because engineers are way more happy.

54:46They can ship things to production much faster. In a way, we've been in this like weird paradigm in the last like five, 10 years with AI, which I don't love. We're like, I call it the like the carbon to silicon cost ratio. Like the chips are more expensive than the humans. The humans are the carbon and the chips are the silicon. Okay.

55:08Turner Novak:Yeah. Right. So it's like, right now, that's a challenge, right? So what it means when the humans are more, the chips are more expensive than the humans, there's no consideration paid to how much does it take an engineer. But when it's the other way around, when the labor is more expensive than the silicon, then yeah, take more chips, whatever. We can run things at a little bit less efficiency. Let's use a higher level language, whatever, in order to make engineers move faster. And I think we've been a little bit too far to the extreme of optimizing for GPU efficiency in the last few years. And I think as we make engineers more important, again, I think there's going to be a lot of premium people are going to want to pay for just the ability to move fast and have good developer experience and shipping stuff into production.

55:59Turner Novak:Yeah, I'm definitely curious on your opinion on that. But I feel like one thing I learned, I took a business strategy class. It was my last semester of college or whatever. And my biggest takeaway was you basically have to figure out, do you compete on a different, unique, better product or do you compete on cost? So it's like, are you the lowest cost provider? It's probably the shittiest product, honestly. Maybe not. But are you the cheapest or are you the most expensive because you're the best? And it's like, if you get caught in the middle, you become like, you might say something like Walmart would be like low cost.

56:37Turner Novak:Like, it's like, you go in, you know, you're going to get, it's not fancy, but it's cheap. And on the high end, it's like Canada goose or something. They're like the thousand dollar jackets, right? It's like just premium, the most expensive, but it's like so warm, so comfy. and then in the middle is like jc penny or like kohl's where it's like they kind of have fancy things but they're not the you don't have the most scale so like they're not really the lowest cost and it's like it just kind of sucks and no one goes there so you die basically yeah yeah but like look at another market look at android versus ios yeah exactly like android has like kind of found like a niche and like low cost but apple is conquering everything else and making may more money, by the way, than Android.

57:21So I don't know. I think depending on which market you look at, sometimes there's like 25 players in retail. But in technology, it has to be like one or two players. And I don't think it's necessarily evident that being low cost or being high cost is like the best model. I think in iPhones, that actually goes pretty far down the spectrum and they make way more money. But of course, if you really, really want to save money, you're going to go buy an Android.

57:49Turner Novak:Yeah. Well, I think too, if you look at iOS, like they just, I feel like they've, they almost like continue to move up market in the sense of like the phone just continues to get more expensive. But then also they're like, I feel like they're slowly creeping down into the lower end of like just more and more people, just like as a percentage of humans are buying iOS devices. And then also like, you have tablets, you have the watches, you also have the laptops, like MacBooks. So I think I saw a stat. This is from a couple of years ago. I think they have over a billion devices active around the world.

58:22Turner Novak:So they're probably just like the top 10%, 20 % of like North America, Europe, Australia, Japan, South Korea, like those like most developed highest GDP per capita nations. and most people have money to spend on games and in-app purchases and Apple cards, however Apple's making money. It's from all those top 20 % of income earners. So then how did you think when you were building sort of the brand for modal? I know that was a pretty important sort of like intentional thing you've done over the past couple of years, but how do you just generally think about as a startup, maybe specifically as a developer platform?

59:03Turner Novak:Like, how do you think about building a brand and your go-to-market? In certain ways, it was, like, very intentional. In certain ways, it was, like, not at all. We haven't had a marketing team or a sales team until, like, very recently. So a lot of, basically, our early customer acquisition was, like, me tweeting shit and, like, you know, talking about the product. It's like founder-led marketing, I call it, right? So, and it sort of works. And I think, you know, we ended up being very PLG, almost, like, unintentionally, without me necessarily thinking deeply about it. because I always built consumer products.

59:34Turner Novak:I think today there are such a lost opportunity for so many companies to have just like a boring brand, like an enterprise infrastructure and SaaS and stuff like that. Like my theory is like the era of like boring brands is like over. People want to buy a product that makes them feel cool, that makes them feel like they're like working with something that like you know the cool kids used so i i think it's such a lost opportunity to not be out there and and have a voice and have an identity and be like a little quirky and like kind of carve out your own niche and like how you talk about the product and so i don't know we always try to try to have you know we're trying to be approachable but also try to be a little cool and i think it's important as you build out a brand to really think about that and i don't know there's so many products out there, especially in enterprise SaaS that are so boring.

1:00:30You go to the website, it's like, download the white paper.

1:00:34Turner Novak:Book a demo with a sales team. And I think that's just, you know, it's not the way that these products are. And I don't think we're the only one to notice that. I think there are some really exciting brands out there. Versailles is a great one. Planet Scale, I love what they're doing. Very crazy website. There are these new generation of products that are just taking a very different approach to branding and marketing. And I love that. Oh yeah. I'm on the PlanetScale website. It almost looks like it borrows a little bit from like MS-DOS. Exactly. But that's the point, right? They're like, they're like speaking to developers.

1:01:09Like, it's like very clear when you go to that website, it's like, oh, like, cool. Like there's this different, I don't know. I mean, it's like weird, but I think it appeals to developers. Developers are very finicky. You know, it's hard to sell to developers. And maybe that's also like a different buying pattern. Maybe like 10 years ago, it was more like CTOs and CISOs and whatever buying the products. But today's developers paying for the product and they want the cool product and they want to work with the product that makes them feel cool. And so I think it's a different way to sell software today.

1:01:37Turner Novak:I feel like developers are just smart too. Like if your product sucks or like they're not going to use it because they have a choice. They're going to be like, I'm not going to use this shitty one that I can't even integrate properly. Like just give me the one that just works because I got shit to do. Yeah, exactly. Exactly. And I think the see-through marketing fluff. I always push when we write up content. High performance, blah, blah, blah. I'm like, what does the high performance mean? As a developer, you're like, unless there's a very specific meaning, don't write the word because it just sounds fluffy.

1:02:13Turner Novak:So then if I'm just getting started on sort of, I don't know, building in public, tweeting things out, building my marketing strategy for like PLG bottoms up like that? Like, how would you recommend getting started? I know you, you kind of blogged for a while and tweeted for a while before starting the company. Like, is that probably the best way to go? Like just start today, like start five years before you start the company? Yeah. I mean, I think so. Right. Like that means, I mean, like, I think like everyone has to find their own journey. Like, like, I think like everyone's go to market journey, every company is going to have their own like idiosyncratic thing.

1:02:45And like, when you listen to like podcasts of like founders saying, Oh, like we, we like discovered that influencers or we discovered whatever like you know you always have to like discount that a little bit and like because i think everyone has to find their own kind of journey for for us it just ended up being kind of leaning into the stuff we're good at which is like i already kind of had a distribution channel through my blogging and tweeting so if you're asking for yourself i mean do you have a lot of twitter followers i would absolutely do that right like if you have a lot of following online like you should absolutely lean into that if you don't like i mean there's other stuff right like you know some other companies have found like very different pathways to user acquisition.

1:03:21Turner Novak:Well, it's probably also what are you good at? Like, if you're a great public speaker, just go speak at some conferences and wow the crowd and get them to sign up. Or like if you like podcasting, start a podcast, go on some podcasts. If you hate podcasting, you know, don't do that. Just whatever you're good at is probably the way to think about it. Because it's like, it's super hard. Like, I think that's the main thing if you're thinking about starting something. It's like, it's probably going to be a hundred times harder than you you can even conceptualize. So you better make sure that you enjoy it and that you're good at it because, you know, it just increases your chances of success if you can stick with it and you pick stuff that you're actually, you have an advantage in versus everybody else.

1:04:01Yeah, and especially if you enjoy it. Like I'm always like shit posting anyway. So like I might as well write about modal. And you mentioned you guys had raised some money.

1:04:12Turner Novak:I think you mentioned it was Zerp. Sort of, I think that's probably like the first round. What sort of like fundraising journey? I know you announced a Series B by the time this comes out, maybe like a couple of weeks, a month before this episode came out. But what's fundraising been like? And any just general advice for people thinking about raising money? So we raised the seed round. Actually, they raised two seed rounds back in 2022. And then the A round, the seed rounds was primarily from Amplify. Then the A round was from Redpoint. That was 2023. And then we didn't raise for almost two years.

1:04:45We did a small safe round in between, but then we did a series B recently with Luxus leading. I think fundraising, like, I don't know, like me personally, I like to keep things lean. I don't like the overhang of a lot of capital. I actually feel like I work better with like little capital and like, you know, forcing myself to like really like use that capital well to get to the next round. so I'm not a big fan of these like mega rounds that have happened recently and we'll say though like to some extent it's like also a function of the market like as the market is heating up like we felt that like at some point you just have to play the game the market is playing yeah you need

1:05:26Turner Novak:some amount of capital if everyone else has hundreds of billions millions or billions yeah exactly yes so you know to some extent it's like you can't choose your game you can play the game you know by its own rules and you have to figure out the rules and play by them but you know at At the end of the day, if competitors are out there raising a lot of money, you kind of have to do that too. It's just the market heating up. You go from this blue ocean to red ocean or whatever. Things just heat up. It's a different game. Early in the creation of a company, typically if you're building something new, there is no category.

1:05:59And don't raise a ton of money.

1:06:01Turner Novak:Is it harder, though, to raise if something doesn't exist? Like you're like, hey, we're building this thing and there's no comps for people to compare it to. So you kind of got to sell the story to them about why this is going to be a massive$100 billion company in the future. Probably. But I mean, today you can just say it's AI. Right? I mean, sure. Yes, we probably had the benefit of getting credibility as a team. I think in the early days, when Moto raised money, yes, we didn't go and raise money without any sort of experience. I don't know. I was in my late 30s when I raised money. And I think people looked at, you know, my experience and like, yeah, like, seems good.

1:06:42Turner Novak:Yeah, why not? Like, seems like he might know what he's talking about. So there was a little bit of that. My point is, like, it's probably easier for me to say. But, you know, sure. I think for most people, maybe you need to have a little bit more traction. And maybe, like, if you're, like, in your early 20s and have, like, zero experience, like, maybe YC is good for you. Because, like, it forces you to find product markets fit very early. And that's always, like, easier in a way to raise money based on. when you're building something hard that takes like several years, I think investors need a little bit more like of a belief in the team and the technology in order to like put in money.

1:07:16Turner Novak:Yeah. I mean, I think a very under-discussed reason of startup failure is that you just kind of like lose faith or like give up. Like you're saying, I don't really want to work on this anymore. And if something's super hard and it takes you two, three, four years, that's a hard thing to underwrite for the investors of like, will they even stick with it to even build the thing? And then it actually starts building the company around the thing that they built. Yeah, it's also hard to find a problem that's hard, but not impossible. And knowing that. Well, I think something you mentioned the other day when we were talking, you said that you think probably too many people start companies in their 20s and they don't start them enough in their 30s or 40s.

1:07:58Turner Novak:I don't know if I got that quote right. I think something like that. I don't know. maybe it's just me being like sad that I didn't start a company earlier in my life. But like, I'm very happy I started a company, but I was like 37 when I started Modo, right? Like, but on the other hand, like, you know, starting a company when I was 37, like, you know, it's been a lot easier because I know how to hire. I know, you know, what good looks like. I know how to build products. I know how to manage tech teams and how to, like, a lot of stuff is like so much easier. like i i don't necessarily need to you know just learn those things on the job some stuff i have to learn in the job but so i don't know my feeling is like there's a lot of people out there like kind of like me and my you know when i was in my late mid to late 30s like they're just like really smart and really good at what they're doing and they should just quit their jobs and start company and the only reason why i did that was like for a while you know my previous company had a tiny bit of an exit like i was able to sell some stock and take a year off and and like jam on stuff but so i really wish you know i think there's a lot of people out there in their 30s who you know if they had a little bit of money maybe they you know would take the risk and do that and i think that would be awesome for society because i think there's so many smart people out there with great ideas and we just need to push them a little bit is it harder though when you're like married with kids i don't know i don't know how many kids you said you have but yeah because I personally, I think it would have been hard to start company when my kids were like newborn.

1:09:24So when I started modal, my youngest had just turned two or three and she just started sleeping. And I think that was like a massive difference. Like IQ, like just back up.

1:09:34Turner Novak:10x increase. Yeah. Productivity. Yeah. No, it was terrible for a while when they weren't sleeping, you know, but now it's good. Yeah. Speaking about being more productive, I know you have a really strong opinion on how to make developers more productive. What's your general thinking on developer productivity? I think it's super hard to measure. Although actually, one thing I think is lines of code is actually not a terrible metric. That's something I need to write more about. That was my favorite thing from the Twitter takeover with Elon. He's just like, print off the code. And people are making fun of it, But it's like, how big is the stack?

1:10:15Turner Novak:Are you actually writing code or is there nothing there? I think that's right. But I also think the reason why it worked is I suspect the median lines of code was zero. I think there's a lot of people who actually didn't write code at all. Yeah. And I think that's like... By the way, that's actually something I think culture has shifted a little bit. Sorry, we're getting on a sidetrack here. But that managers write code, I think for the last five years ago, that wasn't necessarily evident. But now I think it's kind of shifted back. And there's more of an expectation that, yeah, manager should also write code.

1:10:46I still write code, by the way, at Modal. But anyway, going back to developer productivity, like I said, it's hard to measure. The only thing I think is kind of gets close to it is just, you know, or like the thing that I've sort of empirically learned is kind of gets to it is like, how fast are your feedback loops? And by that, I mean, if you have an idea or maybe an even better example is like, if you know that there's a bug on your website, and it's an obvious fix. How long does that take to fix? And I think it's like incredibly dysfunctional if you can't get it out with, you know, in like a couple of minutes or like an hour, like active time, like, right?

1:11:25Just, you know, if you know what the fix is, of course, like just fix it and just deploy it. And like, sure, maybe it takes like half an hour and run a CI, CD test suite or whatever, like get it out. But like the way a lot of companies operate, It's like you write up like a linear ticket or like whatever, Jira. And then like a product manager has to prioritize it. And then it goes back to some other engineer and then they have to write something. And then they send over to some other one for code review. And then it takes like a day or two for that person to review the code. And then it goes back to me and then I merge it.

1:11:58And then like three days later, there's a deploy. And, you know, by now you're talking like two weeks of work. That's absurd. Maybe not like two weeks of active work, but like adding up all those steps.

1:12:08Turner Novak:Yeah, all the different passes off between people. I mean, you just mentioned there's like three or four different people that had to get involved to fix a bug. It's the same. So all these handover points, it just adds a tax to all these steps. And that's like, especially for small features, that tax becomes now like 1 ,000%. And so I think it's very important to reduce that tax on every step and make it possible for engineers to get all these code out in production as fast as possible. So one thing you mentioned was code reviews. I was talking to your co-founder, Akshat. Is that how you say it?

1:12:40Turner Novak:Akshat, yeah, that's right. Akshat, yeah, he's like, you got to ask him about code reviews. I have no idea what you're going to say. But what's your opinion on code reviews based on what you just said? I'm not anti-code reviews. First of all, I do want to point out like a very common misconception is that people somehow think that code reviews are mandated by a bunch of compliance standards. That is not true. Like people think this. And that's something people like internalize. They're like, well, obviously we do code reviews. Like we have to because of like compliance. That is not true. You don't need to.

1:13:08Like as long as you have like a well-defined business reason for every line of code that goes into production, as long as you have like an audit log or like why did you push this code? That's fine. Like I think there are like some benefits of code review, like spreading knowledge, like stuff like that. But like what I think is an anti-pattern is like people sometimes I think just use code reviews to like absolve themselves of the responsibility if something breaks. They're like, oh, I don't want to merge this into production myself because what if something breaks? Then I kind of want to point to this other person also.

1:13:43That I think is super bad. And I actually think in practice, a lot of why people like code review is they feel like by doing the code review, they don't have full responsibility anymore. Yeah.

1:13:55Turner Novak:And if you have multiple people involved, like if there's like three, four, five, six, the more people, the less individual unit of responsibility you have. like every and everyone can just say yeah i think that's very dangerous when people just use it as a way to get to that but you know i mean there's some benefits like some when i like sometimes i go and like change some code in like a code base that i don't fully understand i'm like i kind of want someone to take a look at it just to make sure i didn't like screw something up right like or or or if like you know you have an intern like they're writing a bunch of code maybe the intern hasn't thought about all the security implications or whatever like yeah you should probably have like a senior person kind of thinking, look, there's a lot of reasons why I think a lot of code review, like it gets like internalized as this like, you know, cargo cult process.

1:14:42And I think that's quite bad.

1:14:44Turner Novak:So I actually have, his name's Dash Gupta at Greptile. I don't know if you're familiar with Greptile. They're like a AI code review platform. I'm actually having him on the podcast, meet him in like two weeks. So that'll be coming out, I think probably like a week or two after this one does. So it'll be interesting. I'm interested to get his take on code review, obviously that's the whole business. But it kind of lies into the point of how AI is automating coding. And obviously that case, it might be a little bit more necessary to have some code review with AI. I don't know, what's your opinion on how that's kind of all changing?

1:15:19Like AI coding?

1:15:20Turner Novak:Yeah, I mean, it sounds like a lot of stuff is getting automated. A lot of stuff is getting automated. I don't know. I mean, I use Cursor to write code and it's like, it works well for certain stuff, not so well for other stuff. Like I stuck in Rust, but like thanks to Cursor, I can actually write like, okay, Rust. Because like anytime I run into compiler error, it just like fixes all the borrowing stuff. And I'm like, okay, great. It's great for like, you know, weird SDKs. Like the other day, I needed to pull some data from Azure's SDK and like I asked Cursor and it kind of wrote up like a bunch of code and like kind of worked.

1:15:54I'm going to remove like 25, try accept because it injects all this like stupid error handling. and then like sometimes it works for like mechanical refactorings and stuff like that but sometimes it also feels spectacularly so like i don't know man like i don't i feel like for most day-to-day stuff it's not the sort of 10x i don't know like i find that like ai coding makes me i don't know 10 20 30 percent more productive but that's it by the way that's not necessarily

1:16:20Turner Novak:that much more than like coffee true yeah and there's a reason coffee is like what is starbucks Starbucks is worth a couple hundred bucks. Yeah, I was tweeting something about that. I mean, it's like, yeah, exactly, right? I mean, I got my coffee right here, so. Python, switching from, I don't know, Perl to Python or whatever. This may be a bad example. But I think there are new tools coming out if you look 30 years ago to now. There have been new tools coming out every five to 10 years. And they all make engineers 10, 20, 30, 50 % more productive. And it all adds up. Actually, it all multiplies up.

1:16:57It's all adds up on a logarithmic scale. So engineers today, I think, are 10x more productive than they were in the 90s or even more. And I think AI represents a new tool, and it's actually really good. But I think, to me, it's not necessarily fundamentally different than what's already been happening over the last 30 years with engineers being more and more productive. And by the way, if you look at the last 30 years, employment of engineers just keeps going up. There's more and more engineers in society. And like, I don't know, I sort of expect that trend to continue for another few decades, at least.

1:17:29I think there's so much software that has to get built. So many bad websites and so much bad stuff out there. Like, I think we're going to need a lot more engineers over the next decades.

1:17:40Turner Novak:Yeah, well, I had Aaron Levy, the founder of Box on the podcast. And he was like, if you're just like a rational CEO, and you realize that these engineers are getting better and more productive, why would you just like lay them off and cut your spending, like you should be hiring more and building more because the more productive they get, like any rational capital allocators, they're looking at the ROI. They're going to be investing more there. So it's almost like, yeah, if it actually works, you're going to hire more. There's going to be more engineers. I think what's going on is like, there was a lot of bloat that happened like pre-pandemic and during the pandemic.

1:18:17And a lot of companies were quite bloated. And then there was a lot of layoffs. And, And people look at Twitter and they're like, yeah, like Twitter, Elon laid off 90 % and it's still operating. And like, there are some truths to that. Like, I think a lot of companies got kind of bloated, but that's like actually pre-AI. Like that's sort of unrelated. And so I think there's been a little bit of a bump in like engineering employment, but I think it's like, it's going to continue.

1:18:40Turner Novak:Yeah. I'm sure people have seen those, like the subreddits, like people working multiple jobs. I've like always read those and like, you know, these people trolling, like this guy brags about having like five jobs at once, like all the keyboards. Like it was like a few weeks ago, like people started on Twitter. They're like, oh yeah, like soar him like we hired him and then we have to fire him and they're like 25 other people piled in. They're like, yeah, like, yeah. Yeah. And I actually met a guy, met a guy. This is just like we host, we host like some board game nights just with some friends.

1:19:05Turner Novak:I met a guy who's just like casually telling me he has two jobs. I'm like, whoa, I've never actually met one of these people before in person. And he's just like, yeah, they don't know. Neither one knows. I get the work done. but it's and it's like you make like double the money right you go like you just stack you make 200k here 200k here whatever like extra make 400 grand a year whatever your salary is but it's interesting because like you think the first job it just covers all your living expenses or whatever and that second job it's just free money and it's not like you're paying double the rent double the car, car insurance, food expenses.

1:19:42Turner Novak:It's just like it's like 10x more enticing to get that next job because it's just all cash flow to the bottom line. It's pretty high margin as an individual. That to me, though, like, I mean, great for that person, by the way. But like that to me signals to me that like there's a lot of like still like bad management of software engineers. The fact that like those two companies still haven't realized like this guy is like actually not contributing at the level of like a software engineer. To me, that's like insane, right? Yeah. Yeah, one of them was Big Tech. One of them was the government. Yeah, yeah.

1:20:12So that, yeah, doesn't surprise me at all.

1:20:14Turner Novak:They're probably like, they barely even have the internet connected and half their stuff. Like most of their software probably doesn't even work properly.

1:20:22Turner Novak:So you mentioned something that I thought was pretty interesting. You said you're both really bullish on AI, but you're also kind of bearish at the same time. I feel like that's maybe, it's like maybe a common sentiment with a lot of people. Like, how do you kind of rationalize sort of what's going on? And like, what are you most excited about? What is kind of giving you pause right now in the industry? Yeah, and like the more I like sort of talk to other people, I actually think, like you said, it's not necessarily an uncommon view. I mean, I think there are so many cool applications out there.

1:20:53And like the other thing that like I've been like so shocked by is like how much revenue there is in everything. You know, if you had asked me like a year ago, like what's the revenue in like Vibe coding or like APIs for speech transcription or like whatever,

1:21:07Turner Novak:Like a couple million bucks. I don't know. The fact that there's probably combined revenue of$500 million a year today in VibeCoding, it's mind-blowing. And these are people, they're paying for value they're getting. Or another example is Suno. I would have never anticipated the revenue, the opportunity for AI-generated music. I actually have no idea what they're making. But like, and like what brings me so much, what makes me bullish is like those people, they don't care about if there's an AI bubble or not. They're paying because they get value from a product, right? On the other hand, the thing that makes me a little bit bearish is like when I see the enormous amount of capital going into, you know, AI startups and AI infrastructure, I feel like is that revenue actually going to like match?

1:21:59Sorry, is that investment going to match the revenue? And that to me brings me some pause. Like, I don't know, when we're talking about, you know, whatever, half a trillion dollars going into data centers today, is that actually going to like materialize in terms of revenue in the next like 10 years? That makes me a little nervous that there might be like an overinvestment in certain parts of this AI economy.

1:22:23Turner Novak:Yeah, I think I saw a stat that it was like OpenAI and Anthropic have actually added more net new ARR or something like that. I don't know how you define ARR, but like revenue, whatever, than like every publicly traded SaaS company. I might be getting the stat wrong, but it was a pretty insane statistic like that. Just those two companies are literally, they've grown the same amount of like total new dollars of revenue as like the rest of the publicly traded software industry. The revenue is insane. But like, on the other hand, like, you know, yeah, like it's crazy how much money they make. And like, hats off, like they're doing a fantastic job.

1:22:59Are they going to make a hundred billion a year? in like five years yeah they actually they maybe might like that's the thing is like to me it feels like incredible i don't know like i could be wrong like i probably wouldn't have expected them to make it you know billions of year either and so i don't know maybe i've been like you know maybe i'm continuously on the wrong side of history i just feel like okay great they're making whatever 10 billion dollars a year right now are they gonna make to get to 100 are they gonna get to 500 like that to me makes me a little bit nervous like at some point i feel the law of gravity is going to take over.

1:23:31Turner Novak:Well, and I know this is a little bit of the VC content marketing machine Kool-Aid or whatever that I'm sure we've all been drinking. But I mean, part of the rationality on this is AI just kind of automates things. It does these low-level tasks for you. It's like replacing labor. And if you just think of most companies, what do they spend money on? I don't know what the average is, but it's a good chunk of revenue. Like depending on your business, you might be spending 10 to 50 % of your revenue on just labor costs and whether that's in like manufacturing, whether that's in like software, whatever.

1:24:07Turner Novak:So if you're just like switching labor salaries into software revenue, I think it's kind of hard to predict what the ramp will actually look like, really. And it's like pretty big, like it's like tens of trillions of dollars that could eventually move over. But then it's a question of like, does that happen in one year? 10 years, 100 years. It'll probably happen within 100 years, right? But like... A few thoughts on that. First of all, if you look at like historical, you know, technology shifts, I actually don't think a lot of money has been made in like saving stuff. It usually comes from like making, coming up with a bunch of new applications.

1:24:48Like, I mean, I don't know if like, if you looked at like, you know, trains or something like that, rails, right? Like we probably saved a little bit of money like getting rid of horses. But I feel like it actually turns out like we open up like 10X more newer market. And the reason why I like, I actually feel like sometimes, and it's not necessarily saying it's like, you know, order of magnitude is wrong. Like I actually think on the saving side, like, oh yeah, we're going to save so much money automating stuff. I don't know if the value capture is going to be there. I'm just thinking it's just on making people more productive.

1:25:20Turner Novak:Like it should get more done. I think that there's always more money in like creating new markets. And it's not even like making people more productive. I think it's going to be like, there's going to be a bunch of new jobs no one's even thought about. And that's going to be where like the money gets made. You've probably seen the memes of like people, you know, tweets of like, these are the five jobs that are going to exist in the future. And it's like, you know, AI researcher, vibe coder, you know, insert a couple more funny, funny things. Like it's kind of true. There'll probably be a lot of vibe coders in 10 years.

1:25:49I think internet is a good example, right? Like internet, I don't know if like internet displaced that many jobs, but it created a fuck ton of new jobs. And that's where most of the value of internet was created, is not saving money for librarians or whatever. Yeah, I mean, sure. It just turns out everyone can now find information faster, and that makes everyone more productive. And now there's all these new jobs, like making websites and building startups and all this other stuff. And that, I don't know.

1:26:17Turner Novak:And to your point about the downsides of raising too much money, just for someone who hasn't thought through this before, Typically, if you're thinking about getting an exit on some sort of investment, whether you're a founder or an investor, when someone puts capital into a business, like let's just say I invest$100 million in modal right now today, let's say you'd raised no other money before. Typically, the terms you kind of agree to is that the investor that puts their money in gets their return first. That$100 million you put in, there's like a 1x liquidation preference. So if I put in$100 million, I get paid first.

1:26:51Turner Novak:So let's just say, whatever, you raise them bunch of money, multiples are really high. Your company's worth 100 times revenue. You raise a billion dollars at a 10 billion post money valuation. Things look great. Fast forward two years, multiples come down. You're no longer worth 100 times revenue. Let's say your multiple falls and you're now worth 10 times revenue. So your valuation right there is down 90%. Well, let's say you raised a billion dollars at 10 billion post. You're now worth a billion dollars on paper. you're down 90 % and you raised a billion. So as the founders and the employees of the business, you get no capital back.

1:27:26Turner Novak:And let's say also, maybe the company's only worth 500 million or 100 million or zero, like the investors also don't necessarily get paid back. So it's like an interesting dilemma of like, hey, you do want the capital, like it's good to have money. But like, if you like that, the more that you raise, it sets that like, sort of like minimum bar of failure, which, you know, it's just good to consider. Like, don't just assume like - I 100 % agree. I'm going to throw in another point about that that people often forget about. Texas, you know, when you race at a very high valuation, you also screw up your 49A and like you make it, you know, a lot more expensive for your employees to exercise, you know, your options.

1:28:07And there's all these tricks you can do if you, you know, you avoid racing, you know, especially when you're near the QSPS threshold. Maybe you do some safe rounds in order to get around that and keep the 490 low. And there's all these things you can also do. So I think there's many benefits to trying to keep the common valuation down. But I 100 % agree with what you said, right? The more you raise, the more of a bar you create to exceed for the next round. And it gets harder and harder. I think that's the only time. I actually think VCs and founders have quite good alignment for most part. but the one clear misalignment I think I see a little bit in like VC versus founders is like, VCs will always advise you to take more money than you need.

1:28:53And they're going to make all these reasons. They're like, well, you know, it's always good to have money. It's good to raise from a position of strength. You know, you never want to raise when you need the money. You know, who knows what happens in the market, which I always think is like the most silly reason because like, if you expect the market to go down, why are you not putting in money? Like, Like, yeah. I don't know. I think VCs will sometimes make up all these reasons to like, you know, want to like, you know, for their founders. And that's like, I don't know. I would be a little bit skeptical about some of those reasons.

1:29:25Yeah.

1:29:26Turner Novak:And ultimately VCs make most of their money on average just from raising their next fund because you get those management fees that are kind of locked in. It's like a SaaS contract. Like you raise$100 million, you get 2 % a year for 10 years. So you raise$100 million, You just locked in$20 million of management fee revenue for the next 10 years. It's like just a massive enterprise contract, really. And so how do you raise your next fund? Well, I'm an investor in modal. I tell you, raise a big round at a super high price so that on paper, I look good and I go raise my next fund. And then it's almost like, who gives a shit what happens to modal?

1:30:00Turner Novak:Because that's in the rearview mirror. Like, I'm doing new deals. I'm deploying more capital because I got to get the next fund raised. So I mean, the system does work. Like it's worked really well at creating a lot of value, but it's definitely something to just be like cognizant about if you're a founder. It's like the incentives are not always fully aligned. Yeah. And especially when you think about like the how long the economic cycles, the macro cycles last, like you kind of have to think about like the return profile over an entire economic cycle. And I think that the truth is like most startup investors have not lived through a whole cycle.

1:30:35So they don't necessarily think about that. They think more about like, what's the markup I can get on the next round? Because that's going to determine my promotion, right? Yeah. So I think most startup investors are more interested in like, I put in money in this valuation and then the next round better be a higher valuation, right? Like they don't care about like going public because that's like 10 years later. They're like, yeah, whatever.

1:30:56Turner Novak:We'll all be dead by then. Hopefully not. But yeah, yeah. Like I, I mean, for me, I, my first fund, I started investing in Q1 of 2021. You get slapped in the face pretty fast. Within like two years, it was like, holy shit. the valuation does matter. Like it can be a good businesses can be overvalued. Like those are two separate things. Like, and also shitty businesses can be undervalued too. I mean, that's, that's called value investing. That's, you know. I don't think there's any value investing in startups, by the way. Yeah, there's, there's not, there's not. But like, but yeah, it was a good, it was a good lesson I got like slapped with pretty quickly.

1:31:33It was just like, hey, the price does matter. You mean in the sort of the mini downturn in like summer of 2022 or?

1:31:40Turner Novak:Well, yeah, like you just, you get in and you're like, holy shit, you cannot be investing at a hundred times ARR if the company is not actually growing like 10x in a year. So like, let's say, let's say you've got, you've added a company at a hundred X ARR. The company grows 10x over the next year to hit the same price. You need the company to be valued at 10x ARR. Hopefully your next round is like 20 or 30x ARR. If you're still being, if you're, if you're, we're a hundred X before, but like multiple compression is a thing. that is hard to control. A lot of it is just market. And some of the issues too is, let's say I'm another founder.

1:32:16Turner Novak:I compete with modal and I say, oh, modal raised it. Whatever ARR price you raised at, that's all I know. I don't know, where's their debt in there? What are your margins like? I might not even know what your growth rate is, honestly. And I just say, oh, modal is this. I should also be valued at this. You get this game of like, everyone's kind of comparing themselves. And at the end of the day, only one of those is a good asset. and there's probably a lot of like subpar, less lower quality assets. And maybe they're not given quite the same premium on the multiple, like they're slightly worse, but it's like if one of them is worth 100X, one of them is worth 90, one of them is worth 80X ARR, like the 90 and 80 should probably be at like nine and eight.

1:32:58I would say though that like multiples today are kind of healthy. I actually feel like multiples like a few years ago were kind of more out of whack.

1:33:05Turner Novak:Yeah, yeah, they're pretty fair. Like if you look at OpenAI, Yeah, you can argue that there was that one like viral clip about Bessemer, like explaining, like ignoring the negative gross margins. These are actually pretty reasonably valued businesses on a growth rate. So that again, like brings me some comfort. Like, I don't think if there's an AI bubble, I think it's going to not like I think it's going to kind of fizzle out. Like it's not going to be as spectacular as like dot com when you were raising money based on eyeballs, whatever. Like there's like real substantive value being created here.

1:33:35Turner Novak:Yeah, it's your point. People are using them. I don't know if you remember Web3 and crypto. Like those, it was like, the interesting thing was like the spreadsheets looked good. Like there was revenue, but they were like not, it wasn't real revenue. Like it was like literally day trading, gambling versus with some of the AI stuff. It's like, oh, we like made a website that someone uses to make commercial transactions on. Like there's like underlying commercial purposes to it. But although the Web3, like, hasn't, I don't know, in hindsight, was that a bubble that popped? I don't know. Look at Coinbase's valuation.

1:34:08It's still pretty good. Like, I don't know. Like, I personally don't like crypto. I feel like there's, like, an enormous amount of resources that were draining on society until, like, what's basically just, like, gambling. Like, for that reason, I kind of despise crypto. But, like, if you look at, like, the valuations, like, they're still pretty high in crypto then.

1:34:23Turner Novak:And that's all a function of just, like, people have too much capital. They're looking for returns. They're deploying it into things that they think will keep going up, really, at the end of the day. That's all it is. Totally. Exactly. And that's actually where I think in the way, like I actually expect a little bit more inflation and like slightly higher interest rates. This may be good because it actually like creates a higher hurdle rate, which means like now people have to go and look for like, you know, more productive stuff to invest in, which in theory should drain crypto because it has like a 0 % interest rate.

1:34:51I don't know, man, like it hasn't really panned out. But in theory, I think there's an argument for it.

1:34:57Turner Novak:So I got a question for you. Akshat said I got to ask you about how you keep the temperature and CO2 levels in the office. Oh, boy. Yeah, what's the story there? So you guys are pretty... I suspect that, you know, I'm not like a health nut, you know, certainly not an RFK. I think there's like certain things that actually do matter, though. And one of the things I've been like radicalized on is like CO2 levels. It's actually very clear like relationships, like CO2 levels versus like productivity and, and cognitive performance, it's a very clear relationship. And a lot of indoor air, a lot of office schools also.

1:35:34I actually think one of the most cost-effective things we can do in this country, in this world, is just better indoor airs in schools. When I'm a billionaire, I want to just fund this stupid thing, which is send CO2 monitors to all the schools in the world. And just if it's too high, crack a window. That's it. And so, yeah, I mean, I, you know, I bought a bunch of CO2 monitors for our office. I go and look at them. If it's bad, I open a window. Honestly, I think it matters.

1:36:04Turner Novak:And it's that simple. You literally just open a window and it clears out the CO2. Yeah. I mean, I don't know. I mean, that's like the only way effectively to like, sometimes if you have like a good ventilation in an office, like it kind of just automatically trigger. But I don't know. I mean, we're in Soho. This building is like 100 years old. Like I think the only way to really get like fresh air in is to open a window. Interesting. Okay. What does the data show about CO2 levels? Like it just make your brain think slower? Yeah. Yeah. There's all these like studies. You look at like problem solving ability, like it goes down like very quickly.

1:36:38But, you know, like normal indoor air is like, I think 500 ppm parts per million. It's like, that's kind of like good. That's like outdoor air roughly. Sometimes I mean, three, 400. but often like in offices, it hits like a thousand, two thousand, three thousand. Five thousand, by the way, is like brain damage. In airplanes, it often hits like twenty-five hundred, three hundred. I used to wonder, why do I always get so drowsy during takeoff? It's because like CO2 levels go up inside the airplanes. I always used to pass out.

1:37:09Turner Novak:I do feel less productive on planes. Like I feel like my motivation level, because I usually try to work on flights, but there's definitely some cases where I'm like, Like, ah, fucking, I'm just going to like sleep or like listen to music or something. It's primarily during takeoff, like that CO2 levels peak. It gets a little bit better once you're in the air and it starts to circulate, but. Yeah, I can actually see that because yeah, usually it's like my, it takes me a little longer to like switch into like pull my laptop out type of mode after takeoff, huh? Well, last thing I wanted to ask you about, so you've worked in Europe and you've worked in New York in kind of like seeing the differences between maybe just Sweden, maybe just Swedish tech, but Europe versus the US, what's the difference in like the tech ecosystems in your opinion?

1:37:52I mean, first of all, I think Europe has incredible tech talent. And so I'm actually not like a permit bearer in Europe. Like I do think there's a lot of structural challenges with Europe and, but like I look at like, actually Sweden, I think it's also like doing okay versus just Europe. But so I think there's tremendous talent out there. I think it's just not being taken advantage of or like, you know, like Europe is really not. And so, I don't know, US has way more availability of capital. It also has like much better options taxation. I think that makes a huge difference. I think, you know, when you look at these like, you know, startup cultures, a lot of it just comes down to like having smart people, which I think you would pass.

1:38:35And then the other thing is just like having a history of success and just like having like a culture of like, oh yeah, like my buddy worked at that other startup and he made so much money. oh my God, maybe I should also work at a startup. And that latter part, I don't think has existed quite yet. And then maybe a third thing I feel like hasn't quite existed in Europe for a long time. It's this, I don't know what to call it, but like reality distortion field type founders. Danny Leck was actually really good at the founder of Spotify. But I feel like at least in Stockholm, there's been kind of a lack of like big vision, crazy megalomaniac founders until quite recently.

1:39:11Now there's like a few of them. and and i think that makes also another huge difference it's just like having these like big visions like i'm so tired of like european startups like building like tiny like they're building a sass for grocery stores or like whatever i'm like build something crazy something big and i feel like that's like that's starting to happen a little bit more at least in sweden so i'm bullish but i don't know do you get that a lot more than in the u.s just like bigger vision opportunity yeah it's too big you go to you go to ask stuff like everyone's like trying to build the next open ai or like the next whatever like i'm like chill you're like 21 but like we're gonna cure cancer yeah yeah exactly but i mean it's like good like i think we need more people on average to have these like big bold visions like i think it's good i think it's good

1:40:00Turner Novak:it feels like the sweet spot is like solve a specific unique problem today with like the vision of like, we're going to cure cancer, but we're starting with this. And it's like you bridge the in 20 years, 30 years, this is how we get there. Yeah. Yeah. Well, this is a lot of fun. Thanks for coming on the show. This is great. This is awesome. It was amazing. Thank you so much for having me. And thank you for listening. Thanks again to Meow in Hanover Park for supporting this episode. Head to meow.com and tell them Turner sent you to see if your startup is eligible for free bookkeeping and hanoverpark.com slash Turner to upgrade your fund admin to the 21st century.

1:40:33Turner Novak:If you missed it, make sure to check out last week's episode with Austin Petersmith on building Howie, the world's AI secretary. And tune in next week for Adit Abraham and Reducto on how they're defining the way the world processes unstructured data. If you like this conversation, please like, comment, subscribe, and name your next sandbox environment on modal after me. If you don't want to miss a future episode, subscribe to my newsletter, The Split, linked in the description to get each episode plus a transcript emailed directly to your inbox every week. Thanks again for listening. See you next time.