Hey there, all thoughts viewers, you are about to watch a conversation recorded live in Chicago with TRW founder and CEO Don Wilson, sometimes called the smartest man in trading. We had an absolute blast hosting that show and we hope you enjoy it too. Don, well, thank you for being here. Really appreciate it. Great to be here. Truly the perfect guest. Truly the perfect guest about what's next in trading. But just to begin with, why GPUs? Well, obviously AI is becoming more and more useful. And, as it becomes more useful, people, use more of it, which means they need to use more GPUs to, to run inference or train new models. And I actually have this theory that within the next ten years, the world will spend more per year on on GPUs than it does on crude oil. And that would, of course, make GPUs compute the largest commodity in the world. So it seems like you would kind of need a market for a very modest call, just the largest market in the world.

Joe Weisenthal: Yeah. It's funny because, you know, I certainly an oil often coming out of, you know, sandy desert, but now they're literally turning the sand via, chips into the commodity itself, or, like, breathing life into the sand. Just to back up, I have a million questions about that. About this. For those who don't know, why don't you give us the sort of, you know, the 32nd or the 45 second description of, what you do or what AWS. Yeah. So, I started off standing in the trading pit in Chicago and the Eurodollar option pit, yelling and screaming and, and, then I would go home and write code on my Macintosh computer, and I built models and essentially, you know, I don't stand in the pit and yell and scream anymore, and most of the pits are gone, but but we kind of do the same thing now with computers.

Tracy Alloway: I heard a story that you were once on vacation with your family, and you were in Italy. I think in Florence. And instead of, I don't know, eating gelato or something like that, you decided to invent a new Greek letter for derivatives trading. So. So this is cool. Yeah. So? So here, I mean, the the you're confusing two story, so. Oh. Dairy. Okay. Actually, what happened was, I was, there was a new exchange that had launched an interest rate swap futures contract. It was called IBC. And, I looked at the contract and I figured out that actually they had not designed the contracts properly. And so, although they were telling everybody that it was economically equivalent to a regular interest rate swap, it wasn't because it had this additional convexity bias in it. Which is we could talk about convexity bias. It goes even more in the weeds than a lot of your podcast go into. But, but so when I was in Florence, I had this idea of how you could create an interest rate swap futures contract without this convexity bias problem. And that is that is what I focused my time on there. What was the letter. Well so so back to the letter that the letter that was about after a really unpleasant period in the Eurodollar option pit, where, you know, all the market makers lost tons of money because the shape of the skew shifted dramatically. As the fed started hiking in a very predictable manner. And, and nobody had really developed a measure for, linear skew, and so over, over the weekend, I said, you know, during the week, I said, well, you know, this isn't that much fun. We're losing a lot of money every day. But the good news is that that means we have something to learn. And so, so I spent the weekend working with the quants and we came up with, you know, kind of a measure of, of the, linear skew between the inputs and decided to use the Greek letter psi to describe it. And, you know, so by Monday morning, we had put it into the risk and under the sheets and, and before the open, I explained to the traders, you know, how to talk about it, how to use language around it. And, and before you know, it, we had made the money back because we were able to trade, manage this risk better than anybody else because we had a whole language around it. Amazing.

Joe Weisenthal: So we've established your street cred when it comes to, solving problems in contracts for financial instruments. If I think about a GPU future or something like that, the first problem that comes to my mind is standardization. Because of course, you know, all different types of chips, different types of memory, different latency, I guess. How do you go about addressing that? So so that's a great question. And right now, so what we've done is we set up to two companies. One is called compute exchange, not very creatively named. We have a tendency to do that. At what is your initials. Right. That was my trading badge. And yes, also my initials. Yeah. I, you know, I mean, we did better later on with Cumberland. Crypto. Yeah. That's nice, that's nice. That was actually a reference to the Grateful Dead song about, Cumberland Mines. I didn't I didn't know that either. Yeah. One of my partners who does the more creative naming came up with that one. He's a he's a dead fan. Anyway, so, the other company is called Silicon data. And silicon data is job is to create indices that will be, that will become tradable, you know, will be viable to have futures contracts listed on them. And right now, they've created a number of different ones, but one is the H100 index, another one is the A100 index. And believe it or not, those indices are both available on Bloomberg. Oh, amazing. That's, love hearing that. If we were in the studio, I would already we would be looking at I would already be looking up the, chart as you were talking about it. Who are the natural participants? Because when I think about AI or training, you know, imagine someone goes to one of the big cloud vendors and they sign a long term contract or whatever. Who are the participants who would be better off in an environment where there was sort of where there was a liquid market for compute. So, what we found and who actually uses, compute, extend source compute. And we find that because there are something like 70 different cloud providers that, participate, you can, you can often get better pricing. And one of the things that you can do is you can specify if, let's say that you're an AI company. Yeah. And you kind of you know, roughly what kind of cluster you want, you can specify that. You can even say, you know what? I'm, I don't I'm indifferent between locations or, you know, I if it's in the Middle East, I'm still okay with it, but I want to pay $0.20 per GPU or less. Whatever it is, you can kind of express your preference curve. Compute exchange can conduct an auction and then, you know, find the, the kind of best price compute that matches your needs. So that's, that's kind of the idea of how it works. And, and, you know, it probably doesn't work if you want a 10,000 cluster. You know, monster for doing a huge training run. Okay. But for, for inference, it works great. Or for smaller training runs, it works really well.

Joe Weisenthal: Is the broader impact the idea that once you establish a liquid market where people can, you know, presumably hedge their exposure, that that would bring down the cost of capital. So, so that's right. So, so once you have a liquid market, then it's much easier. Then then you have much more confidence in the indices. And you can then list futures contracts. And and so what does that do. It enables the NIO clouds that are going out raising capital, buying a bunch of GPUs, putting them in data centers and kind of hoping that they can rent them out and not really knowing what, the they're going to be able to rent them out for six months from now, let alone two years from now. So a Nio cloud could, buy the GPUs, sell a strip of futures contracts, and I envision that these will be traded kind of like electricity futures, where there's one for every month. And. And if you want to hedge the next three years, you sell 36 of them. And now you've locked in your pricing, obviously their cost of capital is going to go down, which in turn should make GPUs more readily available. And then on the flip side, if your running an AI company and you raise a finite amount of dollars and you kind of know how much training you're going to do, but you don't know exactly what configuration, and you can go ahead by the compute in the derivatives market. And then once you have, clear view on exactly what configuration you want, then you can swap those derivatives for actual compute. Talk to us a little bit more about so I'm trying to think, oh yeah, the sell side. So like we have these like big, big clouds, right. The ones that everybody knows. And then you mentioned the neo clouds. Do you see that changing? Like what do you see is the future mix of cloud vendors in the future. So, that is a great question. I think that that the whole space is going to grow, but that the, you know, AWS gcp's of the world will make up a smaller percentage of the whole. Okay, that's my guess. But how come, because there are such proliferation of, other companies buying GPUs and deploying them. Okay. So good answer. You know, Joe asked you who would be the natural market participants for this. I'm going to ask you the opposite question. Who wouldn't want this? Because I think of some of the hyperscalers. They seem to like controlling the GPU supply and maybe squeezing some of their competitors. Would you expect resistance from them? Yeah, I mean, I think the hyperscalers benefit from opaque pricing and kind of bundled pricing. And of course, they would prefer to have all the GPUs could. But Nvidia I would also prefer to have all the GPUs. You know, that's always a good thing. But but I think Nvidia wants the GPUs to be widely distributed and, and they're really the ones that make the call.

Joe Weisenthal: This isn't the first time that there's been an attempt to create futures markets out of something, out of technology. I think there's been multiple efforts decades ago to like Dram futures. It doesn't seem that fundamentally different, although maybe it is. Why did those fail? Like when you think about like, what's going to be different this time? What was the failure that caused? Why didn't Dram futures take off? So so the thing about Dram was that the price just kept on going down. So yeah, in a very predictable way. And so why would you want to buy a futures contract if you know, the price in the future is going to be lower? Whereas GPUs, you know, we've certainly gone through periods where GPU demand was super high. And then we've gone through a period where, you know, there was kind of some excess. So there's not a consistent trajectory of pricing. I, I think that there will be a consistent trajectory lower in terms of, I don't know, however you want to measure it, dollars per flop. Sure. Dollars per token, I, I think that that's going to continue to decline. But the, you know, in H100 is going to be a useful GPU for a very long time. And, and so, you know, and, and over its life, I think there will be periods where there's more demand, less demand, and, you know, a little bit more cyclicality and, less predictability.

Don Wilson: So I know the the Trump administration has said that they want this market to happen. Right. So you seem to have some regulatory, I guess, tailwind behind you. Yeah. I mean, I don't think that I don't think that this is a controversial thing. I think that it's pretty clear that, you know, okay, once we figure out the right index construction and have kind of sufficient data that I don't think the CFTC would complain about the, the product. This is a little bit of a sideways question from your attempt to build a, this market. But speaking of the cloud, in the in your main business, is it, I assume your sort of major customers or users of, the CME? Are you excited about the CME migration of its back end to Google Cloud? Because they tout it. They talk about their partnership with Google, etc. as a client or customer. What are you enthusiastic about this move? We interviewed Terry earlier. Yeah. Said that we was excited for sure. Yeah. So so the problem with so it depends on what you put into the cloud and it's totally fine to put, a lot of things into the cloud, but the thing that you don't want to put into the cloud is a matching engine. Okay. And and the reason for that is you want the matching engine to be as, deterministic as possible. And so that means that, if you send two orders into the, the matching engine, one, let's say, you know, a couple of microseconds behind the other one, you want the one that gets there first to be filled all the every time. Yeah. And if you put stuff into the cloud it's very hard to make that happen. You wind up getting, you know, a wide distribution around, which order will be for filled first and, and even as you kind of stretch those times out, you could have an order that comes in, you know, maybe a couple milliseconds later, be filled first, that, is super disruptive for liquidity providers. And it means that, the liquidity in the market is going to suffer. But this is so, I mean, and you say it's not ideal for them to put the matching to have a matching engine in the cloud, but this is the direction it's going. It. Yeah. And it's unclear exactly which part of the matching engine within the cloud. Okay. Is it some kind of, you know, a dual structure? I don't know, but but that's what matters is. Yeah. Deterministic matching engine. I mean, if Google can figure out how to make a matching engine in the cloud deterministic, you know, go for it. Can you skeptical that that's even possible? Can you just describe the sort of theoretical problem? What is it about cloud computing that makes this particular problem the deterministic aspect difficult, as opposed to traditional infrastructure? Well, when you have on prem, computers, you can, kind of it's all right there. You can control where the wires go and. Oh, I see, and so when it's in the cloud, it's a little bit more. Well, yeah. Nebulous, I guess. And, it's just hard to do as a good pun, I admire it. So, you know, you mentioned that you have this long and storied career in the trading industry. Starting from old school trading. And now we're here talking about GPU trading and what's in the cloud and what works and what doesn't. Tell us what your company is actually doing when it comes to practical application of AI. This is a question we're asking everyone. We asked all companies to spill all their proprietary secrets, excluding the engine, excluding the engineers. We know that they're we know that they're generating. We know people. So don't say yes. We know that they're using cloud code or whatever. So besides the engineer. Yeah, you're right. That's that's kind of the boring. Yes. The boy, the. And then the other thing is then they when we ask this question, people say to watch a machine learning things that. Yeah, which has been around forever. So let's talk about actual. Yeah. Yeah. So, so, I think that the way that you make, we make trading decisions is going to change dramatically. And and it already is. You can, use AI to, interact with your proprietary data, your proprietary models, and suggest trades. That's pretty cool. Are you doing that right now? Yeah. So, okay, so we're we're I mean, we're starting to do that. But but we have some tools that kind of do that now and, and. Yeah, the other thing that's really interesting is to fiddle around with, with agents, and have different agents interact. And so you could, you could kind of think about, you know, maybe you have, you have a couple different analysts, AI analysts that, both, work on some stock. And then you have kind of a, risk taking agent or maybe a couple different risk taking agents that interact with those analysts and then come up with trades, based on that. So, I mean, these are, you know, that's a little bit of a theoretical concept, but I don't think we're that far away from things like that.

Tracy Alloway: Let's go just on the, the cloud, trading a little bit more. I want to. Yeah, just on the, the cloud trading. I am really interested in this, topic. What is the current state today? Just so that we understand where you're at? Like, what is today's snapshot of usage of the platforms? I mean, as far as where the matching engines. Yeah. No, no no, no. Oh, sorry. The on the GPU trading like how active is it? Right now? Like, what do you like, where is the state of the business? Oh, I mean, I, you know, I think last month we conducted, I don't know, 5 or 6 auctions, but it's early. Yeah. But but it's happening. So when I think about how, like, futures contracts are born, it's usually bespoke options. And then you get the index like us. And then you get a forward and then a future. That's kind of how I think about it in my head. Is that the process that you imagine for this? I don't necessarily I think that, I mean, the, the simplest, I mean, I yeah, I suppose you could do some privately negotiated, compute swap or something, and maybe that will happen first, but, no, I think the first thing is a futures contract that settles to an index. If the spot market becomes really liquid and you have very standardized auctions, let's say. And, you know, one of the things that you asked about was, well, how do you deal with the lack of standardize? You know, and so one thing is, you go to a certain type of GPU, you know, H100, for instance. But even with Nat, you can, you can, configure them in different ways. You could use InfiniBand. Yeah, some other way of connecting them. And so what's important is you need to decide on some benchmark and one of the things that silicon data has done is they've actually built some measurement tools that measure, you know, how fast a GPU cluster is. And so, you can then say, okay, well, in order for this GPU to be kind of eligible to be in the, in the index, it needs to meet a certain standard. And and you can there are a couple different vectors you can measure by. So I think that that's kind of how you would do it. And then if you got very liquid auctions, you could actually have a futures contract that cash settles to the auction price. And then, you know, people could have the option of either essentially just, you know, cash settling their derivative and walking away, or cash selling their derivative and participating in the auction. And they would know that they would, you know, that price would transfer from one thing to another. That that might be a future state of the world. And the initial state is probably just a generic index. And the futures cash that ought to be indexed. What would a market failure look like in GPU trading? Because your analogy is the oil market. And, you know, weird stuff happens in the oil market. Could we get negative GPU prices? Or if everyone wakes up one day and decides they want to use ChatGPT as their psychotherapist or whatever, some people are doing, could you have a GPU shortage where maybe people can't deliver into the contract? So, there are lots of ways that markets can break and go wrong. And, and, you know, I, I remember to this day that, you know, when oil futures went negative, it was during Covid. I was sitting at home, I was trading oil futures and I bought oil futures for negative prices. You were one of the ones who actually got it. So amazing. But my, then, what was that, 2001? Or 2021? So yeah, my then 14 year old said to me, please, please, please, I want to buy negative you price futures contracts. And I said, well, you have no way of, of taking delivery of the oil. And he said, I will go to Cushing, Oklahoma and figure out how to do it. You really raised, son, daughter, son, son. You've really, he's he's been learning for his, we have an episode about taking physical possession of oil. I do not recommend it. Turns out if you keep it on your desk for long enough, it evaporates into the atmosphere and poisons your colleagues. Yeah. Anyway, anyway, a little bit of a tangent. But, so I think on the, you know, upward trajectory, if there's tons of demand, you know, that's something that commodity markets are really good at dealing with. And, you know, the price will go up and more supply will come in. And I think that's all good. On the downward side, you know, you can always just turn the GPUs off. So I don't think they trade negative. How how much of the volatility that do you when you anticipate market volatility in the price of GPUs? How much of that like embedded electricity costs. So when you buy compute, right, you're buying the chip but also the power. Like how much of that volatility will be the power. So the industry lingo that's used is total cost of ownership. And and you know, what percentage of the total cost of ownership is the price is the power price. Okay. For an H100, it's less than 15%. Less than 15. Yeah. Okay.

Tracy Alloway: GPU trading obviously one of the things you're working on, but you're a busy guy and you've got other stuff up your sleeve. What are you doing in the realms of tokenized trading? So, so that is an area that that I'm, we're super excited about that. And, and we've been thinking about this for a very long time. So in 2012, when we started talking about Bitcoin at the RW and there were a number of of traders at TRW that were very excited about Bitcoin, you were very early into it. 2012 was still pretty early. Very early. Yeah. And so we were having these discussions of, of, you know, why is this interesting. Is it interesting. What about it is interesting. And, and we came away with the following thesis. You know, there's some small chance that Bitcoin could be digital gold. I don't know, you know, call it 1%. It's kind of an interesting product. So we should probably make markets in it. So we we set up Cumberland as the and you know we didn't call it W because at the time everybody knew that anybody trading crypto was obviously a crock. So you know, we wanted to kind of separate the band a little bit. And, but, you know, the other thing was this idea that you could move value instantaneously in a trustless ecosystem was super interesting to me. And I said, wow, if you could do that in traditional financial markets, that would make the markets so much better. So much more resilient. And so we should really figure out how to do that. So we started a company called, again, not very creatively named Digital Asset Holdings. Which, which created the Canton Blockchain. Initially the Canton Blockchain was a private permissioned chain. But, last summer it actually became a public chain. And, and that chain was designed specifically with tokenization of traditional financial instruments in mind. So it has a couple of characteristics. One is it has configurable privacy. And believe it or not, for people who are in the, you know, finance business, they don't want to broadcast to the entire world when they are buying or selling something. Right? I mean, obviously if it's above the reporting thresholds, you do. But, but and so, so that was kind of a fundamental characteristic of this chain that's different than Ethereum or Solana or any of these other things where if you tokenize something and put it on top, and you move it around, everybody sees it move around. So, so that's kind of, something we've been working on for quite a while. And how big could this get? Like, could it swallow everything? Could you imagine a world in which the expect, given any financial instrument, a stock bond, etc., that it all sort of ends up on chain? Yeah, I think that everything will be on chain. Wow. By when give us a give us a year. Oh, I'm always way too early on this stuff but but I, I think in the next five years it really all of these instruments will be on chain. Okay. That's a good time. Well we will we will have a live episode in. We'll come back to Chicago. We'll revisit that question. Yeah. It's the idea with tokenized assets also that you could use that for collateral management and use it as a way to move collateral. 100%. And so everybody's talking about moving to 24 or 5 or 24 seven markets. And if you want to do that, it's really important to, be able to move collateral 24 or 5 or 24 seven and move variation margin, 25 5 or 24 seven. And so, yes, that is that is a very important use case.

Tracy Alloway: So speaking of very exciting, sexy topics in trading, right after you, we're going to be, speaking with, Tariq, minister of Kelce. And so prediction markets are super hot. What is your what where are you at with them is making markets and any of these in any of the spaces right now. So a million years ago, we, actually made markets and prediction markets. I think it was, I don't know, Intrade or something. Yeah. And and it never went anywhere. Nobody cared. And I always thought, you know, prediction markets should be a thing. Everybody should care, but nobody did. And then, you know, then augur came out, I was like, oh, this is really cool. This is going to take off. And nobody cared. And, so it's it's taken a long time. So at this point, we, we use it as, you know, as a reference price. You know, obviously during the election, it was super helpful to, you know, kind of, use that as a gauge of. Oh, so you were actually using that because, you know, we hear stories about institutional investors maybe finding prediction markets useful. Perhaps. But you were you were looking at it. We were definitely looking at it. We were not using it as a hedge. And, you know, it was it was funny. Shane messaged me and said, hey, you know, it's it's, it's up on Bloomberg now. And I was like, no, no, no, it's awesome. Shane. The Shane Copeland from polymer. Yeah, yeah yeah yeah. But currently like do you, do you foresee like are you going to enter now either in making markets on some of these exchanges and would you get into the the sports, the sports contracts? I mean, so we're not so here I think it's highly likely that we'll, start trading some of the prediction markets. You know, some of our competitors already trade in the, in the sports markets pretty actively. We don't. So it's not necessarily a natural fit, but but, you know, I don't have, like, a religious opposition to it. Would there be different considerations for trading in a prediction market versus a traditional financial asset? Are there different things you have to think about, either in terms of like pricing the trade or maybe risk management? Well, I think it depends on what the prediction market is. I mean, if you're if you're trading a prediction market on, I don't know whether somebody will throw, you know, rubber object on to WNBA court. Then I mean, that's something that, that, you know, people in the audience can control. And so it seems like providing liquidity in that you would be at a disadvantage. That was a very particular example, by the way. I was going to go with Taylor Swift getting married, but you went with that one. Well, these are markets that people can directly intervene on an incestuous as opposed to inside of their own. Yeah. Anti-social behavior. That's right. And as opposed to, you know, will the fed cut 25 or 50 or stay on hold? I mean, that is you can you can trade that in, you know, Sofr you can trade that in in the fed funds futures. There are some binaries you can trade. And so the prediction market version of that is, is, you know, totally fits in with the risk that we already trade. So we mentioned in the intro, there's going to be this big, meeting in DC next week. And we just happened to sort of catch a bunch of the participants. When you look at the landscape for these new futures platforms, because that's what they are, right. The CME like does has regulation been part of their dominant, has regulation made it harder for other entrants to cut into sumi margins or volumes? So, I'm trying to say I'm trying to ask questions that are going to create some tension around the table next week. Yeah. Yeah. So so here, I mean, yeah, you should hear what Terry said about how we're neck. I it'll be on the podcast. I, I'm, I'm sure I can, I can probably repeat it without having, so it, here, once you have a liquid market in something, it becomes a natural monopoly. Yeah. It's very hard to move that to a different venue. It's happened before I, I was living in London in the mid 90s, and, the bond futures were, you know, on the floor of the life. It was this huge trading pit with a bunch of guys pushing and shoving and, and over the course of 12 months, the, you know, DTV, now called the euro was able to move the entire bond futures complex on to the computer. On a different exchange now. I mean, they gave hefty incentives to people. I think they went to the all the German banks and they said, don't you dare trade on life anymore. So it's possible, but I think that these things are generally, I don't think that it's really a regulatory issue that causes them to be sticky. I think it's just kind of that unnatural state of affairs, network effect, I guess.

Joe Weisenthal: So our theme for this evening is obviously the future of trading. And one of the things that seems to be happening is the sort of intermingling of professional and retail trading. And we again talked about that with Terry. I'm sure we're about to talk about it with, the CEO. But from your perspective, and again, you started this career back when I don't think there were any retail traders doing day trading. Really. How is that changed the way you think about trading? And can you envision a future where, I don't know, I fires all of us and we're all going to be just day trading from home as an insurance policy? I it's really or I should say Robinhood is really a full employment program, maybe for U.S. workers, maybe. Yeah. So so I mean, that is that is a thesis that, that I have heard is that what's happening is, a bunch of, you know, relatively successful people are losing their jobs. And they're retiring. But in their retirement, they decide to just, manage their portfolios on Robinhood. And so there's this surge in trading activity that, you know, wouldn't have happened ten years ago, and it's only going to grow from here. And, and I don't know, maybe that's right. Do you like it feels to me like, culturally because you're talking about like, why have prediction markets taken off when they've been around for over 20 years? I think I first heard about them in like 2002 or 2003. They've suddenly taken off. There's all this sort of, you know, the there was never a bright line between what's gambling and what's sort of hedging or what's trading, but there's clearly whatever line that is just feels like it's completely collapsing. Is this good? Do you have an opinion like should should, is, there is, is, and I don't know if any of our opinions matter on the question because it feels like culturally we're entering this world where everything will be tradable on any app and there's, you know, you're going to see a price for gold futures right next to one day, the line on a football match, etc.. Is it do we want this world? So, I don't think there's anything particularly wrong with it. But I am a little bit confused about, whether, prediction markets and sports are actually consistent with what the Commodity Exchange Act says is permissible. And so, you know, I know that, you know, you're next. Yeah. You're setting us up perfectly. Here's, you know, benefits from, his ability to list these contracts. And and I don't know if, you know, the CFTC is just kind of asleep, and I know they're kind of understaffed now. Or do you want to stay on stage? Yeah, you can, because you can just anyway. Or or maybe they've decided that actually, these are economically important. Transactions that that are consistent with the. Yeah. It's unclear to me. Amazing. All right. Well, we're going to have to leave it there, but Don Wilson, founder and CEO of W. Q thank you so much for being here. I really appreciate it was really fun. Thank you so much. That was our conversation with Don Wilson, the founder and CEO of Dr. W, recorded live on stage in Chicago. Thanks for watching. Shall we leave it there? Let's leave it there. All right. This has been another episode of the All Thoughts podcast. I'm Tracy Alloway. You can follow me at Tracy Alloway and I'm Joe Weisenthal. You can follow me at The stalwart. And if you enjoyed this conversation then please like and subscribe or leave a comment down below. Thanks for watching.