Max Levchin: The Interview

0:00Have you ever thought about opening up a coffee shop? If I didn't do what I do for a ring, I would just open a coffee shop. Enjoying my coffee. How did you get into this? It's one of these self-fulfilling prophecies. So I grew up in Ukraine and there's a culture of coffee drinking. And my dad was really into drinking espresso. So when I was like five years old, he would buy a shot of espresso at a coffee shop. So I'd have this visual, like a bag and a shot of espresso. I'd be drinking it right alongside my dad. For drinking espressos at five years old? What? There's no prohibition on drinking espresso.

0:30at any end. I recently had Keith Raboyan last fall and he said that you were a first-rate technologist and also a first-rate business strategist. So how did that combo develop? I sort of walked in thinking I'm going to get a computer science PhD and just be a professor and walked out thinking I can't wait to start more companies. Majority of American banks derive a disproportionate percentage of their income from late fees. A firm was founded in in many ways to fight all of that. And we are vestidiously precise about the cost of credit. We'll do north of 47,$48 billion of loans this fiscal year alone, and yet that's a drop in the bucket.

1:06We've been public for five years, and every single letter has a Big Lebowski quote. Have people picked up on that? Until very recently, no one caught it. And then finally, one of the analysts wrote a note on us saying, cue to fiscal 25 quarter that really, really tied the room together. Ah, show us kiss. They finally got it.

1:36also dude having an amphibious rodent as a you know domestic that ain't legal either what are you a fucking punk rager now that's what i'm gonna say i'm gonna clip all of these into one you should i i can uh i can go on i'll do side by side with the actual movie every bit of stupefying some great one-liners just too good many learned men in but in 15th century rambam a lot of lebowski has these great sentence fragments where you have no idea where the characters coming or going to just like cuts into this like yeah and you just naturally drop it throughout the day yeah you said it man nobody fucks with the Jesus so I recently had Keith Roy actually it wasn't really recent at this point the year is just flying by but I had Keith Roy on last fall and he was stringing out this rare combination of talent that he's very like obviously rarely, never seen before and have come across.

2:47You were one of those people. He said that you were a first-rate technologist and also a first-rate business strategist. So how did that combo develop? Where does that come from? I mean, I grew up in Soviet Union, so obviously compensating for some lack of early business education right there. Keith is too kind. And I think the, I, I love computer science. Like I, I was sort of smitten by some very early programming experiences as a, either a teen or maybe even a preteen. So I was like, I was like 11 or 12 when I sort of first got access to the very basic programming tech. And it was like one of these things where I'm definitely going to do this for the rest of my life.

3:38and I was on track to do computer science as kind of the final destination. My plan was to get a doctorate and teach. And a couple of kids at school recruited me to go start a company with them, which failed. But I got bitten by the startup bug. And as I kind of went through life starting more companies and failing at them, So I go, at some point you have to learn business. Otherwise they're going to keep on failing because I was sure that they're obviously not failing for any software related reasons. So it could only be the non-software reasons. Internet businesses are basically some business strategy and some technology.

4:20And I was very confident in my ability to produce the latter. And so it must've been the former, but I'm not sure where Keith got his, uh, assessment. with we have known each other a long long time it was at paypal we got to got to become friends there i asked him if there was a paypal group chat and he said no that's that's what he knows there's not a paypal group chat not that i'm invited to either so if he's not invited i'm I'm not invited either. Okay. I'm really curious. You have been an engineer throughout many different cycles now. What is it different now to be a technologist, to be an engineer in this?

5:12I mean, you mentioned earlier, you're spending most of your time reading research papers. What does it take to be a leader in AI and also a CEO in this era? the cool thing about this moment in time there's a lot of very cool things about this moment in time we're definitely living through an extremely uncharted territory again like the i think it's possible that i am the luckiest boy in the world in a sense that i got to come of age when the internet went from being largely green screens to being the web, which happened to be on my campus. So Mosaic was launched at the University of Illinois the year I matriculated.

5:58So I sort of walked in thinking I'm going to get a computer science PhD and just be a professor and walked out thinking, can't wait to start more companies, basically because of the first graphical browser and then I got to watch mobile and cloud and and and now AI which is kind of I'm not sure what else comes after but it's already more than a lifetime worth of amazing experiences it is the moment in time to be a CEO with a computer science degree the the unfair advantage for a long time has been if you know how to build software and you have product ideas is you can build some amazing things.

6:40The thing that's happened with the LLM powered models is you can now go from, I have an inkling in my head to I'm gonna build a prototype to I'm going to, not necessarily always on your own, but with a bunch of co-conspirators to build something that's production ready in record short time. And if you do know what you're doing as somebody with a sense for how to build software well, you don't need to vibe code yourself to a prototype and then hand it off to somebody who knows what they're doing. You can actually build something amazing and just ship it. And so the moment is quite unique.

7:26The reason to read the papers and the reason to sort of keep track of what's happening on the ground of the research is every morning something launches that's nothing like you've seen before. the pace of innovation is accelerating, at least has been for six, maybe 12 months now. And if you blink, you miss something. Just kind of stay close to it. What are you looking for in these research papers? Has it gotten redundant at this point? Like, what is new? A lot of it is I just look for inspiration. I think the thing that's happening right now across computer science writ large is these glimpses of what the future will look like.

8:09And for a long time in my professional career, academia was always either too abstract or too far into the future. You'd read a paper. So I spent a lot of my younger years looking at things like cryptography, which was sort of the first love that I've had in NCS. And you'd look at like homomorphic encryption and be like, wow, that's a fantastically interesting concept, but we don't really have nearly the computes. Like it's not going to be a thing in my immediate professional future. Or you'd look at quantum computing and like quantum attacks and post quantum encryption and be like, that's cool.

8:49But like, this is really going to happen in my lifetime. Will they have enough qubits to be dangerous? And with AI, it's different where you're literally reading a paper like, yeah, somebody's going to implement it tomorrow. tomorrow and it's going to be available and then everyone's going to take advantage of it and so you read the papers like what what might come next what what will be the important news of tomorrow so it's you have to stay current i heard you used to write out code by hand and so now i'm assuming you're writing code through llm so now i just don't claude code to write out my code for me by hand send it right to claude code do you think it's important for engineers to even learn the basic principles of coding now?

9:30Like learn to code was like the key kind of skill to go after for a while. Do you think that's still important even at a basic level? I think so. I think the software engineering is kind of an interesting combination. I'm obviously biased, but it's a science. Computer science is a science, although Peter likes to point out that no science that is a real science needs to have its name include the word science. Like we don't say math science, we just say math. And so, but as a computer science graduate, I, I, I have a lot of affinity for the science of computer science. It's also an art form. Like if you read code or maybe we used to read code for, for fun, some code is more elegant than other and you you would readily see it like that that's a beautifully written piece of software and like that's kind of garbage but you know it does the job and so there's a matter of taste and elegance in programming and it's also a craft when you actually look at well-made production grade code so not like a project you built on your desktop but something that you deployed on thousands of servers you can identify if you know what you're doing like that's a well-crafted piece of software.

10:45It's like a thing that, you know, someone could like stamp their brand or whatever, whatever the, uh, the term craftsman of, of, uh, yesteryear would use for this. And so because it's such a combination of these three things, I don't think the LMs are going to naturally always deliver beautifully crafted, elegant, and yet scientifically correct code. And so you'll still need some degree of taste, even so much as to converse with the LM to steer towards the right outcome. And for that, you have to understand what you're doing. Like we may not spend a lot of time, maybe soon enough, we'll spend zero time learning the exact minutiae of syntactic decisions of each programming language.

11:29But the conversation with the machine around what you as a software sculptor would consider to be elegant, I think will remain. And that's certainly important to me as a programmer. And I think without having a solid foundation in computer science, I wouldn't be able to have that conversation. Have you started on any side quests? I have lots of side quests. I've always had lots of side quests. So the thing that's actually what's so amazing about this moment in time for engineering, for those of us who just know how to write code and take pride in being able to write a lot of code. Unless you're current, and even if you are current, there's always a thing you've never done that, like, I've never built an iOS app.

12:14Like, I know how they work. I understand the frameworks that exist. I've seen the tools. I know people who are very good at building great iOS apps, but sort of the ramp up to like, oh, I got to set up Xcode and all the infrastructure to build a really good iOS app. It's a lot of work. Maybe I'll do it next weekend. With LLMs, the barrier to entry into an area of programming that you've never done before is nil. So I was annoyed at some remote control functionality that I have at my home, in my home AV system. So I decided I'm just going to build an iOS app to completely get rid of my stupid remote and have a remote on my phone.

12:56and normally a year ago, I have to learn how to build iOS apps. I have to learn how to interface with this audio video system that I have. That's a lot of work, maybe next weekend. This time around, like, well, I can just ask my favorite agent to go research this stuff and set it up for me. And then I can just focus on exactly the functionality that I want and the implementation quirks that I'm interested in. And now I have a remote that I haven't touched my physical remote in quite some time. It works really well. So that's a side quest that manifested itself as a new app on my phone. It's also the first app I've ever built.

13:27I've since built a bunch more apps because it seems fun and easy now. Like what? Did you build your own DoorDash? Like everyone at OpenClaw? No, I think that that's the silliest thing people have said. DoorDash is not important by way of having a great app. It's important because it integrates with all your favorite restaurants. So until OpenClaw can also do things like call every restaurant and negotiate with the owner and install the right kind of tablet and software and extract the menus and all the things that DoorDash did, I think DoorDash is actually quite safe in their business. Well, what companies do you think are most vulnerable because of this?

14:02I think companies that have built software poorly and just sell that software are very vulnerable. The bar for quality of software is going up rapidly. The, yeah, it kind of sucks and has bad interface, but it really serves an important function and I can't be bothered to hire engineers or build the same thing myself, like that excuse is gone. Like if you really hate some piece of software that you're using and it's just software, doesn't have some deep sort of proprietary data, proprietary source of value, it will get replaced. Like there's no reason why not. So I think that part is, but that's sort of the, it's long overdue to get rid of bad software.

14:48sorcery is brought to you by brex the financial stack trusted by more than 30 000 companies including one in three venture-backed startups in the u.s nearly 40 percent of startups fail because they run out of cash brex is literally built to help founders avoid that unlike traditional banks that let your money sit idle shipping away at it with fees brex is designed to help you spend smarter and move faster their all-in-one solution combines checking treasury and fdic protection into one powerful account. You can send and receive money globally at lightning speeds, get 20 times the standard FDIC coverage through their partner banks, and even high yield from day one.

15:28But same day and even same hour liquidity, access your funds anytime. Companies like Scale AI, DoorDash, Service Titan, HIMSS, Anthropic, Flexport, Robinhood, and Plaid trust and use Brex. Start today at brex.com slash sorcery. That's B-R-E-X dot com slash sorcery. Turing is training the next generation of AI with tasks that require real expertise and real world judgment. That's why companies like NVIDIA, Anthropic, Salesforce and Gemini partner with Turing. Turing builds realistic reinforcement learning environments and data systems based on real operational traces. The kind of infrastructure frontier labs need to train superintelligence.

16:14Visit turing.com slash S-O-U-R-C-E-R-Y. What are the largest changes that you've made in this new era of AI to a firm? How has this changed the way that you're either leading or engineering? I think most of what we've done has actually been fairly organic. We've definitely focused our engineering teams more and more on mastery of the tools that are available. So for a while, we'd sort of had this point of view of like, you're writing code, you're writing code, do whatever you like and have whatever tools you want. And as the tools improved, we had become more and more sure of what works for us and what we don't like.

16:56And so we've been switching the. From all comers, welcome as far as tooling is concerned to some fairly prescriptive ideas and what works well for us as a company. Yeah. I think more than anything, the expectation of how much can we ship and what quality we are willing to allow has increased. And that's good. These tools are fundamentally about productivity gains and that those have been amazing. How do you think it affects commerce? How do you think commerce gets impacted? I think commerce is a very broad topic. There are many different kinds of things people buy. And some of them are already agentic.

17:44Like DoorDash is agentic commerce. When you tell DoorDash, you're not speaking to a human. You're not discussing your basket with a person. You're telling an app, I need more bananas. Like you can tell your app or soon you'll be able to tell your app, I need bananas every week. Like my potassium got a bananas. And so that's agentic. Like it's just out there replenishing bananas for me. can't have any cramps while you're biking no very important um but i think some parts of shopping are inherently entertainment like when i'm picking a gift for my wife or i'm trying to find something that will amuse my children i don't want to outsource that like my taste is you know just the right amount of personal and quirky that outsourcing it to a robot no matter how good it is at being me won't feel authentic or at least i wouldn't believe that it will feel authentic to the recipient.

18:36So forms of shopping that are pure entertainment are never going to go entirely robotic. And then there's everything in between. There's things like price discovery. There's things like deal hunting and all of that sort of fairly easy to imagine if you're, whenever you call yourself thinking, I wish I had an assistant to do this for me, you will now. That's where it's all going. You've said that credit has devolved. What do you mean by that? So look at the history of credit. It kind of began as a form of sustainability measure for farmers. So it goes all the way back to roughly 2000 plus or minus BC, where businesses, farmers that would go through these periods of abundance once they harvest and sell their grains and need capital to seed and process their crops, they needed to borrow money.

19:35And so that kind of emerged as a means of like, I know you're growing something. I know you're going to have access to revenue later. I'll give you money now and you can pay me back later. And sort of buy now, pay later is a fairly old concept. For a very long time until middle ages, you had a fairly well self-governing incident like there were no consumer protections although code of Hammurabi apparently has parts of it that describe things like don't screw your borrower don't overcharge them and obviously there's both um biblical and Quranic prohibition on excessive interest or in some cases pure interest and so it's the idea of like don't screw your customer is has been with us for a while and like in the last hundred years or so we've rapidly left the whole don't screw your customer idea as like a footnote of history where you now have you've gone from interest that's easy to estimate because it's simple doesn't compound to only really doing compounding interest and have it looked in a little while but majority of american banks derive a disproportionate percentage of their income from late fees which is sort of like so that's actually like a really good example of devolution of credit so a leafy was obviously conceived as a means of slapping your wrist and saying like if you're going to be late i'm going to remind you like it's not going to be a lot of money but it's going to be enough like i wish i were on late i should be better next time at some point someone said wait a second that's a hundred percent gross margin product when i charge you a late fee you just pay me more money there's no cost other than like sending you a nasty gram you're like of course of I'll pay you a late fee.

21:10So it's better if you're late a lot because then I'll make more money and it cost me nothing to create that revenue line. So over time, credit just started skewing more and more into this bucket of let's do things that are not at all pro borrower. They're just great for revenue. A firm was founded in many ways to fight all of that and destroy the ridiculous and the exploitive. What do you think the biggest misconceptions you're still receiving about Buy Now Pay Later? What were the biggest hurdles even starting? The starting actually wasn't all that hard. And initially, people didn't believe that for a long time, we would tell people, we don't charge late fees, never have, never will.

22:01And they're like, yeah, sure, you say that. And we'd say, all right, well, this interest rate is zero. So a lot of our loans have no interest and people are so used to this idea of like, it's a 0 % rate, but there's a tiny asterisk. And if you read the fine print, it's like, it's not really zero, zero for now. And then it changes. And so we've built a company to fight all of that. Like we are fastidiously precise about the cost of credit. We don't compound interest specifically to make sure that we can pre-price every loan and say you're borrowing$500 and your total interest charges will be$25 or$0 if somebody else or merchant is paying your interest for you.

22:41And for a long time, consumers were sort of like, yeah, you say that. But there's got to be something in a fine print. One of our core values is no fine print to remind ourselves and the rest of the world that we're not hiding our business model in there. So that was kind of the early stages, probably the hardest thing to persuade people of. These days, the perception is actually shifted. Our customers know exactly what we're offering and why we're here and they love us for it. I look like a logo carrier 24-7 because I love getting stopped in like flights or coffee shops. People are like, oh, you work for a firm.

23:21That's so cool. I love your company. and it happens to all of us not just me like a lot of people here love the experience of like having our logos on because people like the product so much people who don't use a firm sometimes will trot out objections along the lines of well somebody's borrowing money doesn't that mean they don't have their own shouldn't they just be like buying with the money they have like newsflash half the country revolves on their credit cards which means that they're paying interest over and above what they could pay back to their credit card issuer that month. If you're going to borrow money and pay interest, you should not be paying compounding interest.

23:59You should definitely not be paying deferred interest. You should, generally speaking, not be subject to late fees or snooze fees or any other kind of fees. And so building a better system takes time, but people who use us know exactly how good we are and appreciate us for it. what is it like being a public company and going through these cycles and and like i guess what is the biggest pushback been from institutions retail most people who understand who we are and what we're here to do and how long term we're thinking don't really have a lot of pushback. If you look at our performance, certainly over the last few years that we've been in public, we've consistently under-promised and over-delivered over and over and over again.

24:52And as much as the whole quarterly system is a little bit silly, like why is it a three-month boundary that you need to get a letter grade for your performance? It's a performance reporting system like any other. We've been very, very good with the actual metrics that we're graded on, that we want to be graded on. And as far as sort of shareholder feedback is concerned, there are two groups for any company, not just us, people who are inherently interested in the short-term performance, people who are inherently interested in the long-term performance. I have absolutely zero time for the former group.

25:29I only care about people who are here to ride along with us for the mission, which we said will take a few more years. We're just starting to get close to a couple percent of e-commerce and maybe in some foreseeable future, we'll get to 1 % of total commerce in the US alone. So we're very, very small on a relative basis. We'll do north of 47,$48 billion of loans this fiscal year alone. And yet that's a drop in a bucket. So before the final grade is issued, We got probably a few decades to go, but those who are willing to come with us on the journey, I think are seeing that we are excellent at executing on this mission.

26:15How do you measure performance through all of this? One of our, not one of our, our key sponsor, our favorite sponsor, Brex, who I think you're an investor in. I am. We have some merch for you, by the way. It is a limited edition hat. It's beautiful. And we'll give it to you. But so with Brex, you're all about performance, spending smarter, moving faster. How do you measure performance for a firm? And like, what are the different benchmarks you're trying to hit each year? Is it qualitative, quantitative? How do you determine that? Very quantitatively driven company. We measure a bunch of different metrics about the company.

26:55Like any fairly sizable enterprise at this point, we care about revenue, we've committed a few years ago now to get to profitability and did. So just measuring profit and growth of profit is important. And then the sort of quantitative qualitative thing that we care about is what we call building a high performance culture. We really care about making sure that we're not just hitting some numbers, but are doing so efficiently. We're doing that with minimum necessary team growth. In other words, acquiring more operating leverage and so on. So those are kind of the three major areas that we measure for ourselves.

27:51Keep on trying to find a way to embed a way to have fun in our metrics. We have a lot of fun here. That's for sure. probably sometimes more than more and more. Too much fun. There's such a thing as too much fun, but maybe more than some of our commentators expect us to have. But I'm can't take yourself too seriously. In fact, the more long and a tooth you become as a publicly traded company, the more you have to find ways to goof off. We do that. But trying to embed And it's sort of a fun metric is an unfinished project. Interesting. Number of Big Lebowski quotes per shareholder letter is a good fun metric.

28:39And have people picked up on that? Yes. So we've been probably for five years and every single letter but one has a Big Lebowski quote or five. Like it's plentiful. It really doesn't take a lot to find them. and um so there's like 19 letters or something like that that actually it's not true the first year we didn't write letters we just recorded um speeches which was terrible so we've switched so maybe 16 18 letters each one has a Lebowski reference and until very recently no one caught it and it was like a way for us to amuse ourselves entirely kind of in our own echo chamber and then Finally, one of the analysts, I think like three quarters ago, one of the analysts wrote a note on us saying a firm, and there was like Q2 fiscal 25, a quarter that really tied the room together.

29:35Flash up skis. They finally got it. So now there's definitely analysts that, not all of them write it, but definitely many of them will reference some Lebowski as like a secret language. I feel like I've interviewed a handful of public company CEOs at this point, and I can tell that they're bored by quarterly reporting and they're all trying to do different things, like keep it fun. so I guess with yours it's big Lebowski quotes some of them are doing live streaming some of them are doing like different kinds of like theatrical shows like that kind of thing like what like how is it what is it like Bored is not exactly right it's an important thing like you don't want to take to the moment where you're disclosing everything that happened in the last 90 days you know You can't make mistakes.

30:33You might be shareholder fraud if you do that and so on. So you have to be pretty careful. And it is also a moment where all eyes, or at least the eyes of those who care, are on you. And so it's a momentous occasion. It's not a joke. In fact, it's so momentous that a lot of people become so wrapped around the axle of like, oh, the quarter's coming up. I know CEOs, used to know CEOs that would lock themselves in a conference room for a week at a time before the quarterly reporting and surround themselves with numbers and pore over the metrics and really take it very seriously. At which point it becomes more about the quarter than about the actual, more about the report than the quarter.