The Startup Powering Billions In Trades Every Day

1. 0:00 – 1:15

   Intro: A New Kind of Stock Exchange

   1. JFJared Friedman0:00

      OneChronos is a new kind of stock exchange, kind of like a Nasdaq. You probably never heard of it, but just three years after launching, more than 0.3% of all U.S. equities traded every day trade on OneChronos. That's billions of dollars every day. OneChronos is growing so fast because it uses advanced mathematical techniques to allow institutional traders to make sophisticated trades that were previously impossible. Here's an example. Say you're bullish on Tesla and think it'll dominate the auto market. You wanna simultaneously buy Tesla and short sell GM and Ford. OneChronos lets you bundle that into one atomic trade. Now, that might not be something that you as an individual investor care about, but if you're a hedge fund making a billion-dollar bet, doing that trade atomically could save you millions. To get the full picture, I sat down with founders Kelly Littlepage and Steven Johnson to understand how six years of quiet determination and a deep belief in a seemingly esoteric mathematical principle created a company that has suddenly changed the way institutional markets function. [upbeat music] I'm very excited to have here today the founders

2. 1:15 – 2:45

   The Origin Story

   1. JFJared Friedman1:15

      of OneChronos, Kelly and Steven. Thank you guys so much for being here today.

   2. KLKelly Littlepage1:19

      Thanks so much for having us.

   3. SJSteven Johnson1:20

      Thank you for having us.

   4. JFJared Friedman1:20

      Take us back to the beginning. Um, how did you guys meet, and what did you guys do in life that led you to start making OneChronos?

   5. KLKelly Littlepage1:29

      We actually met in middle school. We grew up in the same neighborhood at one point, then ended up adjacent, and became really good friends in high school. So it's a lot easier to start a company when there's just no unknowns about your co-founder, really.

   6. SJSteven Johnson1:45

      Yeah, we grew up, uh, south of Denver in Colorado in the, in an ocean of suburbs, where, uh, there wasn't much to do but tinker with computers and get in trouble with computers. So kind of bonded over that. Went to school on opposite coasts, but stayed in touch just as friends, and over our careers where we kind of ended up working in similar technologies but very different industries.

   7. JFJared Friedman2:08

      I learned today you actually got kicked out of-

   8. KLKelly Littlepage2:10

      [laughs]

   9. SJSteven Johnson2:11

      [laughs]

   10. KLKelly Littlepage2:11

       ... middle school for hacking the school's computer system or something. Uh, the middle school transgressions were more behavioral, shall we say. Uh, talking back to teachers. The subsequent high school suspensions were more directly related to, uh, to computer incidents.

   11. JFJared Friedman2:27

       [laughs]

   12. KLKelly Littlepage2:28

       All very harmless pranks, all very white hat hacking, but, uh-

   13. SJSteven Johnson2:31

       Very misrepresented

   14. KLKelly Littlepage2:32

       ... not a lot of tolerance for that. [laughs]

   15. SJSteven Johnson2:32

       Very taken out of context. [laughs]

   16. JFJared Friedman2:36

       [laughs] And then y- y- you ended up going to Caltech for college. How did what you guys did in college relate to what you guys are, are doing now? Do, did it implant some

3. 2:45 – 4:30

   From Auction Theory to Wall Street

   1. JFJared Friedman2:45

      of the seeds?

   2. KLKelly Littlepage2:46

      Probably the guy that had the most impact on me in undergrad, though not knowing at the time, was I took undergrad econ with Preston McAfee, who's one of the most famous auction theorists. And coming in, I'd never even heard of or thought about an auction in an academic context, and I was studying physics. But I got really interested in some of the computational richness and complexity, and really the notion that you could engineer good outcomes in an economy, and that this is a CS and a math problem, not just a classic econ problem.

   3. JFJared Friedman3:17

      And then how did you guys learn about trading? How did you even, like, know about this problem?

   4. KLKelly Littlepage3:23

      Similarly, when I was an undergrad, I heard the story of David Shaw and his general approach to wanting to make money in finance and use that money to fund basic research. And that really resonated because, especially at this point in my career, I thought of myself as a very average scientist, but maybe I'd have a better chance at building something that creates value and can still fund basic research. So I decided to go to a hedge fund, is the short answer, and at that hedge fund, I realized that some of the biggest problems were really the impedance mismatch between the way that current capital markets are wired and the way that traders, or more accurately portfolio managers in this context, are thinking at a high level. And that really pulled me back to my experience in undergrad, thinking about these types of auctions, knowing that there were better auctions out there, but also knowing that there are very good reasons that they're not currently used at that point in capital markets. And, and it wasn't that we thought about or necessarily aspired to start a company or do this, it was really just talking about the problem and then just getting increasingly interested in and

4. 4:30 – 6:00

   The Idea Behind OneChronos

   1. KLKelly Littlepage4:30

      convicted in the fact that this is something that had to be done.

   2. JFJared Friedman4:33

      Is it fair to say that it was sort of a combination of two ideas? There was seeing the problem from working and trading at a very high level at a hedge fund that traded, like, large amounts of money, and then seeing the solution from your undergrad years and realizing that, like, no one had put these two pieces together.

   3. SJSteven Johnson4:50

      I was doing a lot of cybersecurity work, trying to find signals in large data sets that hackers were breaking into companies, and Kelly was looking for signals in markets to find that stocks were mispriced. Um, and we were using modern data tools to do all this, and sort of had the realization that the stock market, there are many different exchanges, many different electronic trading platforms, and it looks a lot like a database, like a distributed database system, where you have these anomalies, like a write happens in one place, a read happens in another place, and the read is on stale data, and that's exactly a arbitrage condition that exists in electronic trading. And, um, so we kind of pulled all these pieces together over the years to, um, initially try to address, like, the thought exercise of how would you build an exchange that was immune to these latency race conditions. So that was kind of what the, the early years were about. Um, and then a lot of Kelly's research in auction theory came in when we, um, realized that the, the tools and techniques to run optimizations-We're

5. 6:00 – 6:45

   How Trading Actually Works Today

   1. SJSteven Johnson6:01

      at a point where, you know, we could potentially use those and do much smarter things than exchanges do currently in the actual, you know, semi-real time matching process.

   2. JFJared Friedman6:13

      OneChronos is, is, is sort of like the perfect example of, like, a overnight success 10 years in the making. [laughs]

   3. SJSteven Johnson6:19

      [laughs] That's a great way to put it.

   4. JFJared Friedman6:20

      'Cause you guys ... Like, the exchange launched in the middle of 2022.

   5. KLKelly Littlepage6:24

      Yep.

   6. JFJared Friedman6:25

      But what year would you say you started thinking about the ideas that eventually became OneChronos?

   7. KLKelly Littlepage6:30

      Back in, I think the earliest timestamp we have conceptualizing anything was 2011-

   8. JFJared Friedman6:35

      Yeah

   9. KLKelly Littlepage6:35

      ... probably.

   10. JFJared Friedman6:36

       So, like, over 10 years before it actually launched.

   11. KLKelly Littlepage6:38

       Yeah.

   12. JFJared Friedman6:39

       And then there was a, there was a few-year period when sort of you were ... The ideas were sort of,

6. 6:45 – 8:45

   Going All In: Quitting Jobs, Joining YC

   1. JFJared Friedman6:45

      like, circling around, and then in 2016, you guys decided to actually go all in, quit your jobs, start this thing as a real company. You did YC, but then it took another six years [laughs] from then [laughs] until when the exchange actually launched.

   2. KLKelly Littlepage7:02

      Yeah.

   3. JFJared Friedman7:03

      [laughs]

   4. KLKelly Littlepage7:03

      Well [laughs]

   5. JFJared Friedman7:05

      [laughs]

   6. KLKelly Littlepage7:05

      You pretty much summed it up. Um, um, I- I think, like, one of the probably funniest and most important moments of the, the company on the origin story was really a conversation we had with you, which was right after we got rejected from YC for the first time. [laughs]

   7. JFJared Friedman7:21

      [laughs]

   8. KLKelly Littlepage7:22

      Uh, so you were, you were nice enough at the point to say, "Let's, let's keep talking," and this was back when YC was doing a batch every ... well, twice a year. Uh, so there was, there was some time in between. And basically, we walked you through kind of at a higher level of the problem that we were going after and said, "We think it'll take, you know, five years and $10 million to, to build something like this," and you said something to the effect of, "We're gonna need to get that down to, like, six months and 500K," which was really just how do you, how do you demonstrate the MVP on some way to kind of get people interested, keep going from there? And we did find a path to that, but the cool thing is, is it was completely the right approach, what you were proposing. We did ultimately raise $10 million, and we did ultimately launch over the period of about five years, and every step of the way, we kind of had a path to, to launch and do an MVP if we had to. But as people and the investors that were very close to us appreciated the problem at a deeper and deeper level, the more they thought, "No, no, no, you get, you get one shot at launching this the right way, and you should build that. You should-"

   9. JFJared Friedman8:38

      Mm-hmm.

   10. KLKelly Littlepage8:38

       "... build that and maximize the chance of success." And turns out that

7. 8:45 – 10:00

   Solving Two Deep Tech Problems

   1. KLKelly Littlepage8:45

      we came in thinking that we had one deep tech problem. Turns out we had two deep tech problems to solve. We were initially very focused on the optimization side of just how do you actually scale these computations to the speed and complexity of capital markets, but the second part was how do you make dealing with the combinatorial auction, which is a very foreign concept to capital markets, especially before we launched, and it's still relatively foreign even with our success at this point, how do you make that accessible?

   2. JFJared Friedman9:15

      It ended up taking you guys six years from the time that you went full-time on the company until when the exchange actually launched, began trading, like, real, real money. Can you walk people through, like, what happened during those three years? How were you guys doing every day?

   3. SJSteven Johnson9:32

      [laughs]

   4. JFJared Friedman9:33

      What was that journey like to get there?

   5. SJSteven Johnson9:36

      Yeah, it was very much a journey of having to solve hundreds and hundreds of smaller problems that roll up into the bigger problem. I think that's, that's true for basically any startup. Um, the magnitude is much greater when it's a combination of the technology is really hard, the two deep tech problems Kelly was talking about. It's regulated, so there's a whole body of

8. 10:00 – 14:00

   The Regulatory Gauntlet

   1. SJSteven Johnson10:00

      operations and sort of, uh, seemingly irrelevant stuff that needs to be done to get approved, especially for something where we had to spend a lot of time educating the regulators on this concept. One quote from one of the regulators that really stuck with me was when we were in the middle of walking them through, you know, over many different meetings, walking them through how our matching process works. Well, stepping back for a second, the way that pretty much every exchange matches orders is called price time priority. So first you look at the price of the order to see if it's the best price. If there's a tie, which one came first? And there's books written about, like, the, the advantages and flaws of doing this, but it's ubiquitous in almost all of trading, and, um, we don't do that. We look only at price and volume, and we completely ignore, except for, like, admission into the auction, we completely ignore time as a way of doing priority. And so the regulator at one point said, "Yeah, well, maybe it's time that we, maybe it's time that we look at different alternatives to the limit order book." And I was kind of surprised by that, because everybody had sort of indicated ... You know, publicly, everyone views regulators as these stodgy, like, people who say no all the time, and that's not really been the experience, um, when we were, like, really in the weeds with these regulators. So that really stuck with me, and that's, like, that's one class of problems that has to be solved, is you have to, like, educate all these people on your novel mechanism, um, when, you know, you wouldn't have to do that in other cases. Also things like, uh, just silly things like having to go buy a printer because we had to prove to FINRA that we were able to print something in our own office-

   2. KLKelly Littlepage11:51

      [laughs]

   3. SJSteven Johnson11:51

      ... instead of, like, the WeWork that we were [laughs] working out of.

   4. KLKelly Littlepage11:54

      [laughs]

   5. SJSteven Johnson11:54

      So there's, there's all these little things, um, and much bigger ones too, where, you know, we could end up ... We could have ended up paying 200K a year for this regulatory service, but figured out a way to do it in an automated fashion in the cloud, and it was essentially free.So just having to solve a lot of little problems like that took a l- took a big chunk of time. But I will say one of the big tailwinds, and Kelly kind of alluded to this, one of the tailwinds across all those years was the more comfortable people got with the model, the more conviction they had that it was the right idea. So we kept hearing over and over, "This is the coolest thing I've heard of in capital markets, maybe in my career." Like, those comments were not rare, but you'll never print a trade. It's too hard to get this done. Like, it's too hard to cold boot a marketplace. So that sounds like a, like a motivator and a demotivator, but it was actually a motivator and an even more motivator to prove everybody [chuckles] wrong about printing a trade. So we had years and years' worth of tailwind to keep us going.

   6. JFJared Friedman12:55

      You not only had to solve these two hard tech problems that no one had ever solved before with the hundreds of sub-problems, and convince all the regulators to allow you to invent a totally new kind of way of trading stocks, you also had to solve the cold start problem of a new trading venue. You have to have buyers and sellers. Like, you had to have initial liquidity. How did you solve that cold start problem so that when you launched, like, trades actually happened?

   7. KLKelly Littlepage13:19

      Yeah. And that is... There's technical problems, and then there's exactly this type of commercial problem, and I have to say that this is the mother of all problems-

   8. JFJared Friedman13:29

      [laughs]

   9. KLKelly Littlepage13:29

      ... even compared to the technical side of it. However, going the other direction, we got people who were kind enough to plug into us when there was no liquidity day one because they did understand the technical part, and they understood what this could mean for the market. And they also understood that if we were successful, there was value in being there earlier first so that they would have more reps using this type of market-

9. 14:00 – 15:15

   Launching: From 200 Shares to Billions

   1. JFJared Friedman14:00

      Mm

   2. KLKelly Littlepage14:00

      ... so that their clients would see that they were innovative and got there first. So we're obviously very grateful for every subscriber we have, but we're especially grateful for the ones that plugged into a market that traded 200 shares on its first day of existence.

   3. JFJared Friedman14:14

      After thinking about it for 10 years and working on it full time for six years, you finally hit the on switch.

   4. KLKelly Littlepage14:19

      Yep.

   5. JFJared Friedman14:19

      And what, what happened that first day?

   6. KLKelly Littlepage14:22

      Test trades. [laughs]

   7. SJSteven Johnson14:23

      Yeah.

   8. JFJared Friedman14:23

      Test trades?

   9. KLKelly Littlepage14:23

      Yeah.

   10. JFJared Friedman14:24

       Okay. [laughs]

   11. KLKelly Littlepage14:24

       So-

   12. JFJared Friedman14:25

       200 shares.

   13. KLKelly Littlepage14:26

       200 shares, people making sure that the, the pipes work in a literal sense. There's an entire process that happens for reporting trades. There's the end settlement process. There's all these things that happen after. So typically, when people are onboarding a new venue, they do this, and they wait a few days. [laughs] So-

   14. JFJared Friedman14:45

       How long did it take to go from test trades to real commercial trades?

   15. KLKelly Littlepage14:49

       We kind of hit our run state in Q3, so we... I, I wanna say our first trade ever was June 21st, June 22nd.

   16. SJSteven Johnson14:55

       Yeah, it was, it was the last week of June sometime.

   17. KLKelly Littlepage14:59

       Yeah.

   18. SJSteven Johnson14:59

       And yeah, that was the first production trade, and then a while went by where most people were still just doing test trades.

   19. KLKelly Littlepage15:06

       Kind of started actually getting that nice hockey stick curve towards, uh, towards end of August, September that year.

   20. JFJared Friedman15:14

       You know, I work with a lot

10. 15:15 – 17:00

    Sticking With It for 6 Years

    1. JFJared Friedman15:15

       of startup founders, and a really common thing that I've seen is that often when things aren't working, even, you know, just, like, a couple of months into an idea, they tend to get demotivated, and they start, like, doing office hours to ask whether they should pivot. [laughs] Um, and you guys had this, like, six-year period during which lots of smart people were basically telling you that you were wrong and that this thing wasn't ever gonna work. No trades had actually ever happened on the, on the exchange. Um, what was it that kept you guys going for this, like, six years kind of wandering in the wilderness to finally get to that point where you'd launched it and it started to work?

    2. SJSteven Johnson15:56

       I mean, a big part of it was, uh, people were telling us that... People were telling us it wouldn't work, but they weren't telling us we were wrong. They were actually telling us we were right. We were 100% right about, like, this is how market structure should be, but it's just so difficult to break in and to, to get the initial liquidity you need to get everybody connected and to get people over the complexity hurdle. Um, even if they don't have to change anything about how they trade, understanding the model, there's no, there's no denying it is more complex under the hood than, uh, what people are used to. For me, at least, part of it was that the motivation of everyone telling us that this was the correct idea.

    3. JFJared Friedman16:38

       Mm.

    4. SJSteven Johnson16:38

       Um, and I think also part of it was we didn't really set out to just start a company, and because that would be fun, even though, like, it is pretty fun. That's not... That wasn't the goal. The goal was to solve this problem, and we had been thinking about it for so long that, you know, it seemed like, uh, it would be a shame to just abandon it even though we thought it was right and everybody

11. 17:00 – 23:45

    What Makes the OneChronos Model Unique

    1. SJSteven Johnson17:00

       agreed that it was the right thing to do.

    2. KLKelly Littlepage17:02

       What we knew we needed was more time, so we kept putting more time into it.

    3. JFJared Friedman17:07

       I think it's really interesting, like, the way that you guys thought about it is this sort of, like, first principles thinking. It reminds me of the beginning of OpenAI, which also took almost exactly the same time, actually, essentially, like, six years to go from, like, starting the company to, like, actually launching a product that anybody cared about. In a lot of ways, OneChronos is actually more like a hard tech company than a software company in terms of the amount of time and capital that it took you guys to actually get, get live. I'm curious to drill into, um, sort of like the conceptual origins of this because it's, like, such a, like, from first principles thing. Do you guys think you can explain to the audience sort of how markets work now, and then what was the key insight that, like, underpinned all of this that, that... The idea that you guys got obsessed with that seemed like the, the, the inevitable future?

    4. KLKelly Littlepage17:56

       So auction theorists think a lot about, um, different pathologies that can result in someone not bidding their true price, and true price in a sense means, like, price and quantity, just kind of their true economic value for something. So-The thing that really kicked off combinatorial auctions in a commercial setting was the FCC taking a look at how to license wireless spectrum back in the '90s. So the thing about wireless spectrum is it's a market with a lot of substitutes and complements, meaning that if you're building a national wireless network like a 5G network by modern standards, there's a bunch of different frequencies that can work for you, and you don't necessarily care which one. But if you're a national player, you care a lot about winning a contiguous allocation of these frequencies, meaning that if you're bidding on West Coast, East Coast, and you miss the middle of America, you no longer have a, a national network. So these are markets with really strong substitutes and complements. And the way that this used to be kinda dealt with in auction is you can imagine that, all right, let's just say ... I'm making up numbers, but let's say you have 800 frequencies and three geographies, is you as the auctioneer could say, "I'm gonna sit here, and I'm gonna auction these off one by one." And the challenge when you do that is that bidders have to start thinking really strategically about, "Hmm, how am I going to think about this if I lose part of this? Am I gonna then have to pay to ... Am I gonna have to pay spread to sell that back to someone? Am I gonna try to build a different network?" So they start having these counterfactuals that make them want to pay less for that, and you can end up in these really inefficient outcomes where everyone was willing to pay more, but they won't, and that problem is the exposure problem, as it's called. Combinatorial auctions give you the ability to get away from that, and instead express the mathematical optimization that you want directly to the exchange. You can get much more efficient outcomes because investors no longer face this exposure problem of, hey, what if I get this portfolio and it's unhedged or it's not exactly what I want? So it solves both the exposure problem, but also the ships passing in the night problem of people having mutually offsetting economic interests that they would be willing to trade with each other, but no way to discover that they actually have that interest because they're not willing to put out a piece of the overall picture that could then leave them exposed.

    5. JFJared Friedman20:26

       What's an example of, like, a specific trade? How, how does that work in, like, the, the, the old school world, and how does it work in the OneChronos world?

    6. SJSteven Johnson20:32

       Yeah, it's a excellent question. Um, so in the old school world, everything trades through algorithms basically. So you have an algorithm that's saying, "All right. I'm gonna buy a little bit of stock A. Hopefully the stock, the price of stock B doesn't move. All right. I'll do a little bit of stock B. Falling behind on B, need to get back to A." It's sort of just juggling the risk and hoping for the best at the, the end of it. Um, what OneChronos or what a combinatorial auction lets you do is embed that constraint in your trade. So you could say all or nothing. The entire position that I want in both A and B, sell and buy, that has to happen all or nothing as in a single atomic transaction.

    7. JFJared Friedman21:14

       That's like a new API for stock trading. Like, for like, like, uh, like, like the core, like, API, like, call signature for stock trading has always taken, like, one stock symbol, and, like, now with OneChronos it takes, like, multiple stock symbols, right?

    8. KLKelly Littlepage21:28

       Yep. Yep, and more, more generally, multiple stock symbols and constraints that you want enforced on the way that they interact with each other.

    9. JFJared Friedman21:37

       What does OneChronos the company look like now? Like, how many employees? Who's on the team?

    10. SJSteven Johnson21:42

        About 40 people at this point. Um, so we, we launched in-

    11. JFJared Friedman21:46

        Only 40 people. So you trade 0.3% of all US equities every day with a team of 40 people. Who, who works at OneChronos? Is it folks who come out of the quant finance world or folks from other-

    12. KLKelly Littlepage21:57

        Certainly ... It's certainly a mix. So on the technical side, we have people that have zero career experience, zero finance experience, up to people that were some of the founding engineers that were now very well-known market makers, for example. So it's that entire spectrum. The, the overall culture is very much engineering led as far as the way that we think about products, as far as the way that we think about building these APIs and primitives. A lot of it is a reflection of our previous lives. So our third co-founder was a global head at Goldman for most of his career. He very quickly latched onto what we were doing. He got connected to us via an investor of ours coming out of YC Demo Day. So as you'll see, the, the continuing theme here is that YC has been a pretty important part in the entire company journey. But I would say, like, the unifying thing for everyone at the company is first principles thinking and an emphasis on just keeping things very small and very tight. And, like, we think a lot about WhatsApp type stories where you build a giant company, but the thing that's really cool is the per capita outcome for that company as opposed to just you built a bunch of revenue, you hired a bunch of employees. The really cool thing to us is what that ratio looks like.

    13. JFJared Friedman23:17

        What does the future look like?

    14. SJSteven Johnson23:19

        Stay as small as possible throughout all of this. You know, we think that small teams produce the best outcomes, and we have a, we have a little bit of an advantage, I think, in that we can treat, uh, to an extent we can treat each of these different asset classes or types of markets that we roll out as, um, their own team.

    15. JFJared Friedman23:40

        And which markets is this applicable to? You're live with US equities now. You're gonna launch foreign currencies and

12. 23:45 – 27:00

    The Future of Markets & Infrastructure

    1. JFJared Friedman23:45

       European equities. But is this applicable to every market? Like, is basically every ... Is basically every asset in the future gonna be traded on a OneChronos like exchange?

    2. KLKelly Littlepage23:53

       That's, that's the way we think about it, and we also think about the fact that there's a lot of, there's a lot of things in the real economy that are not capital markets assets, and there's a lot of cases where capital markets are approximating some activity in the real economy.And the reason they're doing that approximation is that existing exchange mechanisms don't allow buyers and sellers to express those complements and substitutes.

    3. JFJared Friedman24:18

       Basically, your hypothesis is that this, this concept is gonna like take over the world. It's just a better way to trade stuff, and it's gonna go all the way to electricity markets and commodities markets, and basically everywhere in the economy, and we're gonna rewrite the way that we actually trade assets.

    4. KLKelly Littlepage24:32

       Yep.

    5. SJSteven Johnson24:32

       Yeah. Where any place where there are, um, complements and substitutes, or even more broadly, uh, what they call non-price factors. So constraints, things you care about, about the transaction that are not the price or, you know, the number of units. So any- anywhere where those crop up, these types of markets, smart markets, have performed very well, and it turns out that is most types of transactions. So I think, I think the, the important part about the emergence of, you know, compute as this resource is-

    6. JFJared Friedman25:05

       Yes. Like com- like compute is the new oil.

    7. SJSteven Johnson25:08

       Yeah. Compute is the, is the new oil, and it's, um-

    8. JFJared Friedman25:10

       But it... There's, there's not like a exchange infrastructure for compute the way there is for, for oil. It's super underdeveloped.

    9. SJSteven Johnson25:17

       There's not. Yeah. I mean, the, the closest thing to... There, there is kind of a marketplace for compute, but it's this distributed marketplace where you have the major CSP, the major cloud providers, and they all offer pricing per hour or day or month. You have the neo cloud offerings kind of coming up behind that, and then there are brokers that will find you a block of 10,000 GPUs for the next week or whatever it is, for a month. Um, there's this kind of reservation system. There's a... So there's like a distributed marketplace for all this. But, like here's a fun unknowable question, like how much actual utilization are we at right now of GPUs across all the data centers? And like what could we do in the way of medical research and like, uh, academic research or things that make money? Like what could we do with all that idle capacity? And I think that question just becomes-

    10. JFJared Friedman26:13

        Yeah

    11. SJSteven Johnson26:13

        ... more and more important as the, as both the asset base gets bigger, and it's, you know, hundreds of billions of dollars of servers, and as the value gets higher.

    12. JFJared Friedman26:24

        Yeah. The GPU market seems super inefficient and fragmented right now. I know even like the companies in YC who need to do large training runs, they're constantly scrambling, trying to find a block of GPUs that are available for the time window that they want the, like the right kind of GPU, the right block size. It's actually like very hard. It's, it's like a, it's like a, a, like a, like a block size problem that seems like no one has solved. And-

    13. KLKelly Littlepage26:54

        That's al- almost as if you're describing complements and substitutes [laughs] .

    14. JFJared Friedman26:57

        Yeah. [laughs]

    15. SJSteven Johnson26:57

        [laughs]

    16. JFJared Friedman26:57

        It, it, it seems like actually a perfect fit for the

13. 27:00 – 30:45

    Should You Leave a High-Paying Job to Build?

    1. JFJared Friedman27:00

       combinatorial auction. Okay, so I was at MIT and some other places recently, and I was talking to a bunch of undergrads, and they were saying basically, if they go work for Jane Street or Citadel, they can make a lot of money right out of school, and it feels like a very high opportunity cost for them to leave that on the table and go start their own company where, you know, the future's uncertain. It might work, it might not, versus like I'm being paid like guaranteed cash upfront at these jobs. But you guys must have confronted this as well. Like y- you were working at a hedge fund, I assume in a, a s- a, a similar position. How did you guys think about not just the idea, but sort of like the life choice of deciding to like go all in on this, you know, sort of uncertain journey?

    2. KLKelly Littlepage27:46

       Yeah. Well, I, I think you hit on one really important part for our calculus is that there's the eternal debate of do you start something super early on in life or do you work for a period? And for us anyway, the reason that we had the staying power to build something really big was that we did work for a period, and we had that to, to set us up really. The opportunity cost is pretty enormous for if you have a job at one of these great quant trading firms. You have to really know if the startup is something that you feel so passionately should exist that you're okay with that type of an outcome regardless of what happens. I'm personally of the mindset that starting a, a startup for a company because you think it's a good business idea, but not something that you really care about, very, very hard to grind over any extended period of time.

    3. JFJared Friedman28:39

       Like we've talked about sort of like the intellectual decision-making where you guys like had this like from first principles thinking where you're like, no, like this, this, this idea like must exist. I can basically prove it like mathematically how this is going to work. But how was the like emotional side? Like were you conflicted for a while? Was this a, a difficult d- decision? Were you guys married at the time? And did you have to like persuade like your family and like other people in your lives that this was a good idea, or is it just like a no-brainer, just like of course I'm gonna do this?

    4. SJSteven Johnson29:13

       Yeah. So I was, I was married at the time. I'd actually just gotten married maybe, I don't know, a year or so before we went full-time on this, and um, but had been with my, my now wife for a very long time before that, and yeah, there was, there was definitely the, the personal, you know, team decision of like, am I crazy for doing this or not? And, um, I think one of the, one of the elements there is like how do you minimize regret? So that's another framing of opportunity cost, I guess. Um, the indirect opportunity cost.

    5. JFJared Friedman29:45

       How do you think about that?

    6. SJSteven Johnson29:46

       So I, I think, uh, you know, when you're-- We were both about five or six years into our career at that point, and, um, I think that's enough time that, at least for me, I was pretty comfortable that there would always be a job. The-- So I was working at Accenture at the time. It's a huge company. I was doing R&D stuff there. They're always gonna be doing that. So it was kind of like I could go back and do the same thing I'm doing probably-And maybe miss a year, miss two years, whatever. Turned out to be 6 years before we launched. But, [laughs] uh, you know, miss some time. Um, but I think, you know, just continually checking in and being like, "Am I still net enjoying what I'm doing here? Yep. Okay, let's keep, keep going."

    7. KLKelly Littlepage30:28

       Very similar experience. Not something that I thought about too much once we'd committed to doing it. And similarly, I was okay with a very wide range of outcomes in life. Money has never been important to me, as long as there's enough to provide for

14. 30:45 – 33:55

    What It Takes to Build for Wall Street

    1. KLKelly Littlepage30:45

       your family. And kind of to Steve's point, I got comfortable with the fact that you don't need a hedge fund life or anything like that to do that.

    2. JFJared Friedman30:53

       OneChronos is, like, kind of a rare example, I think, of a Silicon Valley tech-style company actually innovating in Wall Street, in the financial markets. Um, do you guys think that there's room for more companies like OneChronos in the world? W- Are, are there, are there other big unsolved problems here that other, like, smart people from, like, a quant finance background should be going and tackling?

    3. SJSteven Johnson31:19

       Yeah, absolutely. Yeah, I think, um, the... One of the barriers to entry is it's hard to just Google, like, how does trading work at a bank? And, like, software engineers, we're used to everything being so open, and you can just kind of get answers to whatever you need publicly on the internet. Um, and it's very much not the case, and finance is one of many industries. Like a lot of the, the industries that are more... that have been, you know, established for a long time, um, there's kind of an information wall. So figuring out how to break that is the, is one of the big challenges, and that could involve, like, going to work there for a while or finding the right person, which is tricky. Finding the right person to, to work with. But, um, I think people who have solved that problem, it's, uh, it is, uh... The other thing that comes to mind is just, like, it's, it's a different type of engineering process. Um, you can't do the move fast and break things thing-

    4. JFJared Friedman32:21

       [laughs]

    5. SJSteven Johnson32:21

       ... when you're dealing with, like, $5 billion a day of stocks. And, and, like, if there's any sign that you're doing that, then people will sniff that out, and they, they won't work with you. And so it is... Like, you, you said it perfectly before, it is kind of a hard tech problem even though we're not building, you know, rocket ships or anything. It's the same kind of, uh, mission critical or, like, critical systems engineering mindset that has to happen for a lot of Wall Street. Not all of Wall Street, but for a lot of it.

    6. KLKelly Littlepage32:52

       And kind of to Steve's earlier point about we signed up for one problem and ended up with 200 subproblems-

    7. SJSteven Johnson32:57

       [laughs]

    8. KLKelly Littlepage32:57

       ... it's the, the more of these subproblems we understand, the more we see and think of problems in there. And our, our mission is really to drive the marginal cost of matching a trade to zero by generating infrastructure that is so strong and so good that it's just completely aligned with users on if we create economic value for you, we get paid. But if there's no value creation in these auctions, then we need a different model of competition than currently exists in capital markets. And a big part of that is figuring out all of these subproblems, and we would love for someone to come in and figure out a lot of these subproblems. And if not, we're just gonna keep chipping away at them on our own.

    9. JFJared Friedman33:41

       I feel like that's a great note for us to end on.

    10. SJSteven Johnson33:43

        Yeah.

    11. JFJared Friedman33:44

        Thank you guys so much.

    12. SJSteven Johnson33:45

        Thank you.

    13. KLKelly Littlepage33:45

        Thanks for having us, Jared.

    14. SJSteven Johnson33:46

        This was great.

    15. KLKelly Littlepage33:46

        This was awesome. [outro music]

Episode duration: 33:55