The Delphi Podcast
The Delphi Podcast
Logan Jastremski: Solana vs Hyperliquid - Who Wins The Global Exchange Race?
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Join Tommy Shaughnessy as he speaks with Logan, Managing Partner of Frictionless Capital, about the high-stakes race to build a global blockchain-based financial system. Logan shares his updated thesis on why monolithic, high-throughput architectures like Solana are winning the battle for real revenue and trading dominance over modular designs.They dive deep into the "Global Exchange" vision, the physics of time-to-inclusion, and how innovations like Proprietary AMMs (PropAMMs) are redefining market making. Logan also explores the broader implications of AI, from the compounding power of Grok to the societal impact of a robot-led workforce.🎯 Key Highlights▸Analyzing the Frictionless chart showing that crypto apps on Solana are the fastest to reach $100M in revenue globally.▸ Why Logan remains bearish on the "incorrect" modular design and Layer 2 scaling solutions. ▸ How the speed of light and server location prevent current centralized exchanges from being truly global. ▸ A look at the next engineering frontier. ▸ The "crypto-native" evolution of order books that uses massive Oracle updates to eliminate stale quotes. ▸ Why Logan believes the market will split into either fully regulated exchanges or fully decentralized global engines. ▸ Why Elon Musk's Grok is positioned to be the smartest model over time. ▸ The "freaky" transition to humanoid automation and the potential for a 20-day "atom-moving" data center revolution.💡 Want to stay updated with the latest in crypto & AI? Hit subscribe and the notification bell! 🔔🧠 Follow the Alpha▸ Logan's Twitter: @LoganJastremski▸ Frictionless's Twitter: @FrictionlessVC🔗 Connect with Delphi🌐 Portal: https://delphidigital.io/🐦 Twitter: https://twitter.com/delphi_digital💼 LinkedIn: https://www.linkedin.com/company/delphi-digital🎧 Listen onSpotify: https://open.spotify.com/show/62PR1RigLG2YN5Pelq6UY9?si=18ac7ccf36ab4753Apple Podcasts: https://podcasts.apple.com/us/podcast/the-delphi-podcast/id1438148082Youtube: https://www.youtube.com/channel/UC9Yy99ZlQIX9-PdG_xHj43QTimestamps00:00 — Logan’s return to Miami & Fund 1 vs. Fund 2 thesis04:00 — The $100M revenue chart and acting on alpha08:00 — Why throughput alone isn't enough: execution as the North Star 11:00 — Solana’s parallel execution and the 2021 NFT "gas wars" 13:30 — Being "bearish decentralization" for its own sake 17:00 — The Global Exchange vs. New York physics limitations 22:00 — Multi-Leader design: 20ms inclusion and regional information 30:00 — Blockchains as the leading indicator for price discovery 34:00 — Rebuilding modern finance: efficient 24/7 global markets 38:30 — The hollowing out of regional, non-KYC exchanges 42:00 — Estimating the $20 trillion daily global trading market 45:00 — From AMMs to PropAMMs: the crypto-native order book 52:00 — Hyperliquid vs. Solana: product-led vs. engineering-led 59:00 — Retail order flow and the "payment for order flow" win-win 01:05:00 — Timeline for the multi-leader world (2027/2028) 01:11:00 — Returning to fundamentals: cash flows and PE ratios 01:15:30 — Bullish Grok: the vector of compute and real-world atoms 01:21:00 — Automation anxiety: Uber, gig work, and surgical robots 01:26:00 — Dystopian gambling, inflation, and the Zillow reality 01:29:00 — Financing the $30k robotDisclaimerThis podcast is strictly informational and educational and is not investment advice or a solicitation to buy or sell any tokens or securities or to make any financial decisions. Do not trade or invest in any project, tokens, or securities based upon this podcast episode. The host and members at Delphi Ventures may personally own tokens or art that are mentioned on the podcast. Our current show features paid sponsorships which may be featured at the start, middle, and/or the end of the episode. These sponsorships are for informational purposes only and are not a solicitation to use any product, se
You're now plugged into the Delphi Podcast.
SPEAKER_03Hey everyone, I'm Tommy. Welcome back to another Delphi Podcast episode. Today I'm in person with Logan, who runs Frictionless, an incredible fund uh down here in Miami. Logan's been on the podcast before. Uh, was it two years ago?
SPEAKER_01I think so. About two years ago. We talk about ETH and decentralization and low throughput versus high throughput.
SPEAKER_03Was it the worst podcast appearance of your life, Ethan?
SPEAKER_01No, it was good. I think it was very factual. If you look back at it now, uh I feel like a lot of that would still be kind of like evergreen content, which is hopefully the the goal of uh how we approach investing. If our thesis changes a lot, then uh we're probably pretty wrong.
SPEAKER_03I don't remember too much of that episode, to be honest. I remember it being a lot of fun, but I wonder what we got wrong, what we got right.
SPEAKER_01I should have watched it.
SPEAKER_03Um we should just post that episode and see if people pick up on the differences.
SPEAKER_01It was I I remember distinctly though, we we talked about Ethereum uh L2s, why L2s, in my opinion, were not going to win, why um you needed high throughput, why why decentralization was objective and measurable, um, and that kind of the fears were overblown about high throughput blockchains not being decentralized was kind of the main things that I remember.
SPEAKER_03So when we spoke then, you were full in on fund one. Yep. Now you're moving on to fund two. Yep. And we talked a little bit about your thesis for fund one versus what you want to do with fund two. What's the difference in thesis between frictionless one and frictionless two?
SPEAKER_01Um, it's a good question. I think uh we got a lot of things cracked in terms of our understanding that uh blockchains really needed to scale. Um and so understanding that you had to transition from low throughput blockchains like Ethereum that were amazing um in terms of kind of uh turn completeness and kind of pushing the industry forward, but kind of maybe optimize for the incorrect things. Um, I think we got that fairly directionally correct. Um, I think we also kind of, if my memory serves me, or if I just knowing myself, we probably touched upon the modular versus kind of monolithic or definitely did. Uh and I I'm assuming my point uh there was that the monolithic chains were the correct architecture design, um, and that the modular design was going to be incorrect. Um, and so in large part, a lot of the things that we probably talked about a couple of years ago, I think has panned out. I think the thing that we've gotten much more of an appreciation for as kind of fun one continued was that throughput definitely mattered in terms of you need scale for application engineers to build more expressive applications. But the thing that we have really come to appreciate now is that throughput only mattered in a sense that it enabled scale, but the other part of that for at least the underlying infrastructure was execution and trading. Um, and so I would say our fund one portfolio definitely has some trading-related uh portfolio companies within there. But I would say for fund two, we've definitely have a full appreciation now for the things that make revenue and where's the value going to accrue, and just kind of doubling down on our current thesis. But in large part, it's more of a continuation than like a pivot by any means. Um yeah.
SPEAKER_03So just to recap, like with fund one, your thesis was high throughput, monolithic chains will win, uh, pressure on ETH, L2s, the modular thesis. And now with fund two, you want to stick with the high throughput thesis, but you want by extension, like the second order, you want to focus on apps and projects that lean into that and make real revenue based off trading and execution.
SPEAKER_01Yeah. So we actually plotted this. Um and we showed Raj and Anatoly this uh in Abu Dhabi, and they're like, why haven't we seen this? But it was a chart from uh essentially the time from an application to get to 100 million in revenue, um zero to a hundred million and how many days it took. Um, and a lot of those companies um are now currently being built on Solana, and this is not just like crypto companies, this is across AI as well. Um, the fastest zero to a hundred million dollar revenue companies in the world are crypto companies. Um and this has been stolen and remade multiple times. I'm kind of pissed off about it, but it is what it is.
SPEAKER_03I guess uh wait to that chart on Twitter where it's zero to a hundred million revenue and the amount of time you see the the line.
SPEAKER_01Uh you did? Yeah.
SPEAKER_03I've seen it copied a lot.
SPEAKER_01Yeah, I know. Um I've told I've been told that we need to be more public about our research because everybody takes it, which I guess is bullish. Um, if I don't know if you remember, but one of the first things that we published was essentially the throughput of different blockchains. Um, and at the time people were like, You guys are fucking idiots. I don't know if I can cuss on this, but they're like, you guys are idiots. Um like this is not true. And literally all we did was go to each individual's uh website, and it was just the block times and how big the blocks were. It was like super basic. Uh and we just said this is the amount of like kilobytes or megabytes of throughput per second. Um and looking back, like it it kind of made us second guess ourselves, but then like a year later, Paradigm published gas per second. I was like, this is literally what we've been saying for a long time. And because Paradigm wrote it, everybody thinks they're geniuses.
SPEAKER_03Um I mean they're a pretty smart crew.
SPEAKER_01Yeah, they're getting what you're annoyed on. I I mean I mean they are smart, but more broadly, it was just like I think one thing that I've come to appreciate is like we definitely have a lot of alpha. We're not always the most public about it, but I want that to change because I mean the good thing is you're acting on it though.
SPEAKER_03Like you're actually investing based off of it.
SPEAKER_01Yeah, and that's that's kind of been my kind of North Star for the fun. Like, I think in large part, I don't know whether it's the crypto industry or just venture or um or even like kind of money managers more broadly. I think sometimes you can get sucked into playing the wrong games, which is like Twitter clout. Um and for better or worse, we've kind of just said like that's not the game we particularly care about winning. I do think it's helpful for fundraising and uh potentially like seeing deals, but our North Star has always been returning capital to our investors, and so we've definitely acted upon it, which has um been beneficial to us.
SPEAKER_03I feel kind of slighted. I'm pretty active on Twitter.
SPEAKER_01No, no, no, but it works. I I I'm just messed with it. There's different games.
SPEAKER_03The chart of of the Solon, the applications on Solana doing first to 100 million in revenue.
SPEAKER_01Yeah.
SPEAKER_03Like, so you're talking out of all the applications built around the world, regardless of web two or web three, a lot of those companies are built directly on Solana. Is it like one company? Is it the top like a percentage of the top 10? Like, how do we visualize this? Because we're just talking.
SPEAKER_01Um maybe our wonderful editors can throw up the chair. Uh I can send it to you. Yeah. Um yeah, it it some of them are on Solana. I mean, the fact and Pump Fund. Yep, Axiom Pump Fun. Um, one of our portfolio companies backpacked, they're not really built on Solana. Uh Ramp was on there, I think Lovable Cursor. Um, Hyperliquid was number one. Um, I think it was 89 days for them to get to 100 million in revenue. Um so I mean, so in broad part, like now what we're seeing is that these applications are making money, and every single one that is uh generating this cash flow is trading related um on the blockchain side. Interesting. Um and so for us, I think it's kind of given us more of a full sum appreciation for that trading aspect. Um and the blockchains, I think why kind of the high throughput blockchains have won in my mind, uh are continuing to win, is because that the additional throughput enabled um essentially more transactions per second, but ultimately made it better for trading assets. I think the thing that we've come to appreciate though is that throughput by itself, um maybe picking on Celestia just for a second. Uh the we've seen that throughput by itself does not necessarily equate to value capture. Um, and Celestia is killing it on the throughput slide. When we initially plotted it um in our like data propagation chart a while ago, they were relatively small, and I think by now they're the leaders by far. But the thing that they haven't really focused on was the execution piece. And so the thing that we've seen is that the value or revenue is ultimately determined by the execution. So I would say one thing that has been really interesting within the blockchain space, really for the last five years, has been the kind of design exploration of all the ways to scale. But if you say execution is the North Star and trading is the thing you need to optimize for, I think that very narrowly narrows down kind of the possible different uh design decisions that you uh need to make. So there was like L2s, which is a single server, there's uh app chains, which um could be decentralized or could be a single server. Uh, you had the modular roadmap, you had charting, um you had uh like application-specific chains with uh or like um what Avalanche was doing with like different chains. So I think in large part, if trading is the North Star, it's kind of helped clarify in our mind why kind of the high throughput uh chains have become more valuable.
SPEAKER_03Probably a basic question, but like why has Solana won every trading and execution enabled app that's on that first to 100 mil rev list, other than I think you mentioned hyperliquid.
SPEAKER_01Yeah.
SPEAKER_03Um happy to talk about hyperliquid too, but um like why have why didn't they choose like why didn't Backpack choose an L2? Why didn't um Axiom choose another chain? Like why have they all landed on Solana outside of Hyperliquid, which we'll talk about?
SPEAKER_01Backpack is different just because they're a fully centralized exchange uh and went the regulated route, but they're fans of Solana. I would say Solana took upon the hard engineering problems. Um I think one of the things that we touched upon in the last podcast was talking about single-threaded virtual machines versus parallel. Parallel execution was just enabling modern hardware with many cores to do many things simultaneously. Um, and this is something that Ethereum and the kind of adjacent EVM chains just got wrong from the day one. And so I think um because they took a different architecture path, it was very hard to bootstrap, but because they were also the first mover and kind of this parallel execution, it just enabled more of a white space for engineers to build more expressive applications. Um, I mean, if you recall, Solana kind of got its original footing from NFTs. Um, and that was because Ethereum NFTs were just way too expensive.
SPEAKER_03Uh the gas wars were nuts. Yeah, it was crazy.
SPEAKER_01Um I I personally paid like 500 to a thousand bucks on a Uniswap transfer. And so in 2021, when uh NFTs were going on, uh the joke was Solana was for the poor's. Um and it was because it cost a fraction of a penny to trade NFTs on Solana, but if you're rich uh and could spend hundreds of dollars, you could do it on Ethereum.
SPEAKER_03I mean, the what was the Ethereum the god? I can't remember the name. The one where you would compete in a transaction to create the generative art on-chain in the transaction?
SPEAKER_01Oh, there is a bunch.
SPEAKER_03Um the biggest one. It it was like thousands of dollars. They would mint out of the thousand, and yeah.
SPEAKER_01I I lost money off in FT, so I try to block it out of my sand.
SPEAKER_03Yeah, it's it's a tough road. Yeah. Um that is pretty interesting. So so because of technical reasons, they've all landed specifically on Solana.
SPEAKER_01Uh I think so. I mean, so I think because transactions were cheap, um, users started to congregate there, similar to what happened with Ethereum and like DeFi Summer. Like people just liked experimenting. Um, but it was way cheaper to run those experiments, and for a user to kind of onboard transactions were a lot faster, they're a lot cheaper. Um, it just felt more modern.
SPEAKER_03Um, like a super basic question for you would be like an Ethereum L2 offers such theoretically cheap transactions. Yep. So like if you're just deciding based on cost, like wouldn't you go there because there's nothing else built? Or would you or there's less built, you're competing with less people on a new L2 for block space?
SPEAKER_01I don't know. You kind of run into the same problems as what we probably talked in our last podcast. Most of the L2s historically were also single threaded. Um, and so you didn't have these localized fees markets that parallel processing chains enabled. Some of the more modern L2s tried to do parallel processing as well, which would uh isolate fees for specific applications or specific smart contract applic um apps, which was helpful. Um But I don't think I'm actually today I'm pretty bearish decentralization. Um, more so in the sense that I think we going back, we ran all these experiments and they were fine. Um, but I don't think people necessarily pay for decentralization. Um I think if people valued Ethereum uh like uh say Solana had a thousand uh nodes and Ethereum had 10,000, like if if people really value that, they would pay 10x the amount of costs in my mind. Um in large part, I don't think we've seen that. Um I I think people care about decentralization only if it makes a better product. Um but decentralization for decentralization, I don't think people really care about. And so um to answer your L2 question, I think L2s are fine. Um I think they have just generally failed to uh look uh scale. Um there is one really cool website, I forget what it's called, but it literally measures the throughput of all the L2s. Um and if you aggregate all of them, I think they're still less than what like a single high throughput blockchain like Slana is doing. So I still think there's some scale limitations, um, but that may be changing with Mega ETH and a couple others. Um but generally I just think the user base was there just from more historical times. Uh the assets that people wanted to trade were there, and so that kind of just stuck around and now it's about continuing to scale them.
SPEAKER_03I'm not a developer. Well, I'm probably a pro vibe coder at this point, but I do feel like there's just a very low cognitive load to picking Solana. No, there is. Yeah, like it's just like if I want to build a performance app, I go there, I kind of know things will be taken care of from an infrared perspective.
SPEAKER_04Yeah.
SPEAKER_03Um so I want to talk about the other routes competition-wise. We can either talk about hyperliquid in Solana, or we could talk about um like traditional markets going 24-7, ice trading on-chain, or their permission chain and going from there.
SPEAKER_01Um, I think it's all kind of the same. Um in my mind. Um they're all racing in my mind, or what they should be racing for, is to build the global exchange. Um and so whether it's the New York Stock Exchange or an L2 or Hyperliquid, I kind of view them all in a similar vein. Um where you have a single server um located in some jurisdiction in the world, say for example purposes, they're all in New York. Um, and if you push it to kind of the logical extremes, if these things actually work and you're doing call it trillions of daily trading volume, um people, what we have seen in TradFi want um really advantages as much as possible. And so in TradFi uh high frequency trading emerged because people wanted better and quicker access to information and also trades. And so you co-locate next to the actual exchange or the server, in this instance, would be New York, uh, so that you can um receive information faster and make better trades and have more information with your models. And so when I think about kind of whether it's kind of a more modern version of what exists today, or um a single server like Mega ETH or something like Hyperliquid that's relatively centralized in terms of its node set on like the same data center or geographic region. To me, I think you can build a great exchange product, but I think that's kind of missing the point. Um, if you actually kind of could reimagine this and build it from the ground up for a global exchange, I think really, in my mind, the things that you would want is build this credible neutrality. And this is where I think going back to my earlier point, like kind of bearish on decentralization for the decentralization, I think here decentralization actually matters because uh decentralization makes the product better. Um, where it it's not bound by a single jurisdiction, um and everybody in the world can have kind of equal and fair access to information. Um and so Can we walk through that a little bit?
SPEAKER_03So, like hundreds of countries around the world, and you you're saying that decentralization makes the product better. What does that mean? Like if I'm in the US versus Russia versus China using the exchange, like why does that make the product better?
SPEAKER_01So let's use the kind of like centralized exchange example. Uh if you're in Russia and the server for the exchange is located in New York, why would I, as a Russian person, want to trade on a server that's located many miles away from me if I'm running a quant shop and uh you'll always get front run versus the guys in New York. Okay, it it's it's purely a latency like physics perspective. Um, you're disadvantaged if it's not in your home jurisdiction. Um and so I think even if the technology was better, and this is the same for mega ether hyperliquid, like hyperliquids in Tokyo. Um like if I'm a United States person and everybody in Tokyo is getting better prices than me, like if these things actually work out, like this is not like today, like today it's mostly retail, it's fine. Um, if these things actually scale to where you're doing trillions of dollars a day, and this is the everything exchange, I just have a hard time seeing the United States, uh, China, uh, whoever, like big global powers that actually trade a lot, saying, Yes, I'm okay with that data center being in Japan, or I'm okay with that being in New York, because I'm disadvantaged purely from a physics standpoint.
SPEAKER_03Interesting. Okay. So I I thought you were going to take it from a regulatory and like access view.
SPEAKER_01So there there is that too, which I think like the fully decentralized worked. But this is just purely on like, hey, I want to make trades and I want a fair shot, so to speak, at making those trades.
SPEAKER_03Okay. I totally understand. So if you're around the world in any country, you want um inclusion in a market that's in an exchange that's close to you, uh, so you're not front-run by other people that are physically close to the exchange.
SPEAKER_01And so this is, I mean, why I don't think you have a global exchange today and why you have regional exchanges. There exists the New York Stock Exchange, the London Stock Exchange, India, Japan, etc., because uh these, I'd say, physics limitations. Um now you have what has emerged is you have high-frequency trading firms around the world that have these private fiber cables that when both of these exchanges are up and running, will do arbitrages in between them to bring the prices back in line, which is, I would argue, a benefit to each of those exchanges to have a global price. But they're doing that on the back end, and it's you're connected to these exchange venues through the microwave towers and the fiber cables or even now with Starlink. But um, these are separate exchange venues around the world, and I would argue that they spun up not only from the regulatory standpoint, but also because of the physics and the limitations um of kind of the speed of light.
SPEAKER_03There's still somebody in between those two exchanges, though, that that has to even out the price, right? ARP the prices, yeah.
SPEAKER_01If the exchanges are up at the same time, they do ARBs.
SPEAKER_03Okay. So let's continue with this line of thinking. So we're talking we want a global exchange. We want nodes in each market so that nobody's front run. So the question is who achieves that goal? And I think we could talk through three options that you brought up, which is New York Stock Exchange, Solana, Hyperliquid. Maybe we could talk about how they're each moving in to achieve that goal and who you think's gonna win.
SPEAKER_01So when I think about the global exchange, uh really want you want like Almost equal and fair access to that data. So just being able to read that data. And then you generally, from the physics standpoint, don't want to be at a disadvantage to really any other party around the world. And so a very clever design that has kind of been thrown about that is just now starting to be implemented. And is, I would say, I think it's live in Sui. It may be live in Aptos, and it's not live in Solana yet, is this idea of multiple leaders. In blockchains, historically, you have one leader. And what I mean by this is generally at any given point in time, you send all your transactions to a disk one leader. So going back, if we continue with the New York example, the blockchain leader is in New York. You send all the transactions to the leader in New York, or if you're in Singapore, you still send your transactions to New York. Those are aggregated, then sorted, and then disseminated. So it doesn't really matter where you are in the world. But the clever idea that has now been started to implement, and I think is going to be as necessary as going essentially from low throughput to high throughput is this idea of multiple leaders. And so, for example, instead of having someone always has kind of an information advantage, you can disadvantage everybody to call it 20 milliseconds. And so what happens is you run parallel auctions. I think 20 milliseconds would probably be good, but under 100 is probably fine enough too. But I think we'll get to like 20 milliseconds. You aggregate kind of all these transactions around the world. And so it's interesting because from a physics standpoint, when you get to like call it 20 milliseconds, you can only see the transactions from a certain part of the world. Um, because light uh cannot physically travel to other parts of the world from like a fiber optic cable standpoint. So, for example, if you had one leader in Australia, um, you would probably only see every 20 milliseconds only transactions from Australia, because uh if it's 30 milliseconds away, you're not going to be included in that block and you're gonna have too late of data. Um, but it advent it makes it advantageous, obviously, for um uh Australia because they're able to kind of like uh update their worldview in kind of 20 millisecond increments. But if you have these world leaders in call it Japan, United States, Hong Kong, Russia, uh, UK, wherever, um, the thought process is you have these 20 millisecond batches uh that are happening simultaneously and continuously. And about 100 to 150 milliseconds later, each of those transactions are uh essentially uh sorted and then finalize. So you don't, because of the speed of light uh around the world, the world, it's like 100 to 150 milliseconds. Um, it takes a while to actually uh sort those transactions, but at any given time in 20 millisecond increments, everybody you can almost think of it as they're voting on what they believe the price to be around the world. So all the jurisdictions around the world, every 20 milliseconds are saying I'm buying and selling at this different price, and then 150 milliseconds they're slotted. But this entire process is continuous, both on the so you don't wait for these auctions to end. Um you just get the new price every 50 milliseconds, but people are continuously building these blocks every 20 milliseconds. That's awesome. So everybody's kind of disadvantaged at the same, but they're aggregated and finalized every 150 milliseconds.
SPEAKER_03That's so interesting.
SPEAKER_01Does that make sense? No, it definitely is a little complicated.
SPEAKER_03No, it I'm I'm getting most of it. So I kind of want to understand the impact though. So right now, if we're up late trading on Jupiter or whatever on Solana, and we're making a trade, we're selling Seoul, we're buying some random meme coin, and the current leader is in New York, you and I are both in Miami, so we're not really competing. But if a third person in Tokyo wants to buy and sell the same token at the same exact time, we clearly get filled before they do. So the question is we the question, I think, or the thesis is we want a global exchange that's fair to everyone around the world because we're not just going to be trading meme coins, we're gonna be deriving the price and flows and trades for every asset around the globe.
SPEAKER_01For all assets. And okay, the idea is so right now, block times are not fast enough for this to actually be useful because uh Solana has block times of 400 milliseconds. Aptos is actually the fastest at, I believe, 50 or 60 milliseconds. So their block times actually matter where in the world you the leader is. But for 400 milliseconds, there's enough of a global lag, so to speak, that it doesn't matter.
SPEAKER_03Got it. So everybody's transactions around the world have time to make it to the leader.
SPEAKER_01Yes. And this is why historically Ethereum has like 12 second blocks because they've mandated that you need low hardware requirements at low throughput. And so they've just given X everybody extra time to aggregate their transactions to this leader.
SPEAKER_03It's such a it's such a basic concept the way you describe it, that we want time to include everybody. I never thought that that would be the limiting factor. I thought we would always just want the fastest possible block times.
SPEAKER_01And so now the idea is if you shrink it to less like 60 to 70 milliseconds is like half of the world's circumference. So once you shrink it to that, um you that's where you can start getting regional information. So the thought process is once you have multiple leaders, um, and Anatolia has always used this example. It's kind of dumb, but it's funny. Uh there's a um call it a boat in Singapore that has uh $10 billion worth of iPhones and it sinks. The people that will be able to kind of capture that ARB, so to speak, are the people in Singapore because they are able to monetize this information first. So in this kind of 20-millisecond world block view, uh the people and traders that are seeing this boat go down uh could just start market dumping. And the people on the other sides of the world don't know this is happening, but these trades are now just market dumping the price, and they're able to capitalize on that alpha first. Um and um That's a great example. And so uh yeah, so it's it's taking the aggregate worldview price um and then kind of like finalizing it on 150 to uh yeah, 150 millisecond uh um kind of like time timestamps.
SPEAKER_03Um can't you cover the world enough if you go 50 milliseconds and have three or four nodes?
SPEAKER_01Like yeah, I mean we'll I think like we'll definitely experiment with like do you need two? Like, because then you just cut down the world latency in half. Um like but I always use the example of like Amazon Web Services. Um they have like uh data centers around the world because they want to cut down on the latency for you to ping the server. And so the more servers that there exists in the world, the lower it takes for you to ping that server and get a response. Um so they have Amazon East for people on the East Coast, Amazon West. Uh you just ping the server that's closest to you.
SPEAKER_03It feels like, I mean, I'd love your view on this, but it feels like we're getting to a world where obviously we have high frequency trading shops, we have hedge funds, we have us retail traders and institutions and funds, but it seems like, I mean, I know this is like a loaded term, but like AI agents will transact so fast and such low cost and at such high speeds that they would really favor regional markets.
SPEAKER_01They I mean, I I think whether it's AI agents or humans, uh, no one can beat physics, which is why I think this is a cool design, because um the goal, and it's my pin tweet, um, I don't think people understand it, but uh it says the goal for blockchains is be to become the leading indicator for price discovery. Um, and why I think this multi-leader worldview has the potential to become the leading indicator for price discovery is because it's aggregating in 20 millisecond increments the world's uh average price or the world's price for any given asset. Um and if this is the pricing indicator for the world, what will happen is the blockchain will be the true price and exchanges on the centralized venue will be the lagging price. So what happens now today is that blockchains, because they're not performant and not efficient enough, is that they're always lagging, like the New York Stock Exchange or, for example, Binance. And so people view the market making there in the true price and these centralized venues. But if you invert that and you can flip it to this global kind of trading engine, so to speak, this is now the world's most true price because it's representative of essentially the global worldview. And I think that's what everybody's trying to uh build, in my opinion. That's the grand prize. Um, so maybe to like wrap it back up, the reason why I think like L2s, like um hyperliquid uh and kind of like the old um New York stock exchange aren't going to win this global exchange is because the design and engineering to build this is much different than just building a great product. Um Hyperliquid has built a great product, they deserve all uh the praise in the world for doing that. They're making a lot of money. Um, but it is a different engineering uh design to build this global exchange. Um and I think purely from a uh fairness um so so to speak, I don't think the world and its limit will say uh hyperliquid is this global exchange that I feel like as a US person or anybody kind of outside of that market will trade on. Um I could be wrong, but this is kind of our broader thesis. If if we're wrong, uh it's because the world did not care about these properties, and that could be true. Um, but this is our assumption and how we're kind of investing.
SPEAKER_03I think that's the clearest explanation on why decentralization or distribution is a benefit, not a cost.
SPEAKER_01Yeah, it I mean it's literally like Amazon Web Service. They're globally distributed because they cut down on latency. The only reason why I think it's a bonus to have the decentralization is because it cuts down on latency for time to inclusion for the global market. Um it doesn't advantage any single specific person.
SPEAKER_03I want to ask you two quick questions before and then go back to your your main line of thought. Like in a world where we have localized markets and nodes around the world, we have all these venues, it's all feeding into one global exchange. I'm trying to think through like what the benefit of that is to the world. And I'm thinking through your example, the iPhones. Like, like the old days are like equity research analysts would give quarterly estimates, they would say how Apple is doing, we would you know talk about you know how they're doing, how we think the price should be, the price will go up, the price will go down. In your example, it seems like if people are able to effectively act on individual bits of information in a bottoms-up world, like the iPhones falling off of the ship, leads to less iPhones sold, leads to you know lower numbers, less revenue, yada yada. If people are able to effectuate those trades in micro markets around the world, maybe it's on a prediction market, which bubbles up to the actual price, things like that. It seems like our world becomes much more efficient from a pricing and liquidity standpoint. I I'm not sure if that's the right way to take it.
SPEAKER_01I I I think it's I view it as if you could rebuild finance modern finance system, the world is very clearly global, um, whether we like it or not. And I think having 24-7 markets um trading all the time and having kind of an aggregate worldview um is just a better product.
SPEAKER_03But but this isn't just 24 like the New York Stock Exchange could offer 24-7 markets. Well, your thesis is what you're describing is like efficient 24-7 markets.
SPEAKER_01Yeah, it it it it should be in the way that I believe it will work, is that it doesn't advantage any single jurisdiction and it allows everybody in the world to participate. And I think like the actually the funny enough, the ri original reason why I got into blockchain was Ethereum's uh uh banking non banked. Um Ethereum got me into crypto too. And and like I thought that was super noble because it was like, all right, you don't have a bank account, you can't invest in assets because you don't have access to those assets. Um, I think when I think about the world and like modernizing kind of our technology stack, what it would look like, I think it would be this 24-7 global market that gives everybody equal and fair access. Um, in addition to, I mean, whether you want to earn yield and stable coins or just get stable coins, like uh trade your currency from one currency to another, uh buy Bitcoin, um, invest, save payments, like all that. I think if we're all kind of I I describe it as like a single API. You opt into this API. Um, and this is the global API for finance um that just helps everybody coordinate payments, value transfer, and price discovery.
SPEAKER_03That's awesome. It is it is wild meeting people that have never had to use like Fidelity, Schwab, e-trade, whatever, and have only ever used crypto. Yeah, because when you talk to them about like the frictions and like how cool it is to be the frictions of the old world, they don't even get it. They're just yeah.
SPEAKER_01And I I mean, we were talking with some LPs and they're like, look, I I can touch and feel the AI, like I can take Tesla FSD, and like it drives me around. Like it's hard to touch and feel crypto. And I think that's fair. Um, most people, especially in the Western world, have access to these systems today. Like our currency is not necessarily hyperinflating yet. Uh but we have uh access to kind of like investment tools and generally we can move our money around. Um, and I think that's fair. And most people are like, if you say, oh, blockchains are trending towards high frequency trading, their eyes kind of roll over too, and they're like, I give two shits about high frequency trading, which I think is fair. Um but to me, like my co-founder Diego, he had this uh talk at our summit, and we were talking about it's not the next like version of the internet, it's not web two and web three, it's the next iteration of finance. Um, and so I think most people in the the United States and Western world just don't appreciate the how great we have it. And so the impacts of kind of this more modern technology are not as felt.
SPEAKER_03It's also just like it it's crazy how just such a basic education miss it is too, right? Like I've met a lot of people that have a lot of money in their checking account earning no interest. Yeah, like very basic thing that they can get with stable coins. I want to circle back to your your global exchange thesis, though. So I think you've accurately laid out an incredibly succinct way what you think wins. I'm now curious who you think the winning contender is, whether that is Solana, whether that's hyperliquid, or whether that's somebody else. And you don't have to give a specific one, but I'm just curious like which which is trending in that direction.
SPEAKER_01So I mean, on the exchange side, I my broad view is um you either go fully regulated, like KYC have the fiat on-ramps or off-ramps, and be able to like offer all the product suites that KYC provides, or you go kind of this entire global exchange. And I think the middle gets hollowed out. Um because you're if if you're not competing with the global exchange, then the global exchange um can do what I described, where it's like the uh the pricing, the the leading price for the global world. Um you're kind of screwed if you're a regional exchange that doesn't have KYC. Um if you're a fully centralized exchange that does KYC, you're essentially the connective tissue between the kind of past finance world and the new one. So you can have fiat on-ramps and off-ramps, you can do the credit cards, you can essentially do a lot of things that you could not do if you're not regulated. Um and so that's kind of how we positioned our portfolio. Um, I think Backpack is doing a lot of really interesting things on the fully regulated side. Um, they're the fastest growing crypto exchange in the world. They just surpass 100 million in revenue for uh the year of 2025.
SPEAKER_03He's a hell of a founder. Yeah. Really is our money's great. Yeah.
SPEAKER_01Um so they're gonna do a lot of cool stuff because and build unique product lines because they went the regulated route. That's just not possible from the decentralized route because it's illegal. Um and so um I think that part of the world is interesting. And then if you go the fully decentralized route, I think if you can build that, which I think will take maybe a couple years, um, is that the goal is to be the leading indicator for price discovery. So this is the place where people check to actually be the true price. Um and so it's why I have not been bullish um L2s, it's why I was not bullish modular, and now not bullish hyperliquid. And it's not because like I think um like hyperliquid came about and it's just like all of a sudden I'm bearish. Like my point of view has really stayed the same. It's just hyper liquid in my mind is the same as an L2. It's a regional exchange, um, which is a great regional exchange and very profitable, and people love the product. And I give them all the flowers for that. But I don't think this global exchange, they'll be able to build that because it's a different product. Um, and so for our standpoint, I think you either go fully regulated or fully decentralized and build the global exchange. Um I think Solana and Anatoly have been very steadfast and talking about, hey, we want to build this global exchange really since day one. Um recently, Aptos put out a piece uh called the Global Trading Engine, where they kind of acknowledge that blockchains are for trading and that they wanted to focus on trading. And so to me, kind of the first crux step is the ship pointed in the correct direction. Uh Elon always talked about when I was at Tesla, like we're all vectors. Um, you need everybody pointed in the right direction, and you need the magnitude all the faster the magnitude, the faster you go in that direction. And so I'm they've been slow, but they're at least pointed in the correct direction. Now it's up to them to see can they actually ship, can they get things in order? But I like that they're pointed in the correct direction. So it's those on like the global decentralized side, I think, are interesting. And then yeah, the fully KYC side.
SPEAKER_03So when you say the North Star is the place to discover the the leading uh be the place for uh the leading indicator for price discovery. And that's do you view that as spot or perps or both?
SPEAKER_01So I think uh so actually like Apple stock. Yeah, everything. Um like all tradable assets. Um I pulled some like basic numbers and I'll publish an article and um some other cool stuff shortly. Um probably in the next two weeks. Uh if I don't do it by two weeks? If I don't do it by end of February, uh yell at me. But um I'm putting a note to yell at you. Please do. Um yeah, so I pulled some basic numbers. Um I think a conservative estimate for like all trading volume across all assets, across all exchanges around the world is about 20 trillion a day. Um that's like FX, uh derivatives, uh spot, etc. Um and so I think what is interesting is how much of a chunk can you eat into that call it 20 trillion a day? Um and this is also compounding at about 8% annual growth rate per year of the trading volume increasing. Um and so I think that is actually the market that you want to go after. I wonder if that 20 trillion is higher under a global design or I so and it's funny because um Maybe it's 20 like as is right now. Yeah, but is it higher? I think it's the market size is gonna be way higher because you're gonna have 24-7 markets. It's like what happened with Robinhood. Like if you looked at retail and the amount of retail traders, like even 10 years ago, it was a sliver. And now like hedge funds have to look at what happens on Reddit because they don't want to get blown out because if they're shorting, you'll have a GameStop comment. But now multiply that not just from the United States, but globally in 24 7 markets. worth uh all assets, um, I think the trading volume's gonna balloon. Um and so I think that twenty trillion a day is relatively conservative.
SPEAKER_03Um why do you think that I mean if we could take a very simplistic view, it seems like Solana has killed it on engineering. Like in my mind I just I kind of just feel like they won. Like I look at double zero, I look at Alp and Glow, I look at Fire Dancer, like I look at the direction. When I think of Hyperliquid, it just feels like they've won perps trading. So I'm wondering why it's harder for the product oriented company Hyperliquid to become the global exchange versus Solana. Seems like they're way ahead on that becoming the better product company. Is that is that a fair question?
SPEAKER_04Yeah.
SPEAKER_03Um I'm trying to like understand the velocity of each moving into each other's world.
SPEAKER_01Maybe I can talk about spot versus perps and then talk about they are converging um and what I think is the harder design to win. Yeah that's perfect. So on spot it's easier to do because there's no margin. You're just trading assets and then it's just how wide the spread is um if we kind of look back to the history of blockchains what you saw initially was AMMs really start um with Ethereum and Uniswap because you had these 12 second block times and throughput was so small um you would just update the curve. And then when you talk to TradFi and back in the early days they said I know order books. This is not an order book like screw this. And a lot of people actually passed on Uniswap because they said this is not order books. This is dumb. But the AMM design that they created was really unique because of the limitations that they had to build it from a throughput standpoint. And what it allowed you to do is essentially determine a fair market value for assets that were not that were not like largely uh traded from a volume perspective.
SPEAKER_03Resting liquidity for long tail assets automatically adjusts yeah it's it's a cooler innovation.
SPEAKER_01And so really and this is why I say looking back on fund one like we got the throughput piece right because throughput allowed applications to be more expressive because you had more transactions per second. But and we had a lot of experimentation especially within Solana in the early days SAM and FTX build serum um and that was really uh and it worked for a while until people actually tried to start trading on it in size because it started to get congested there's a lot of bugs but the idea was correct like if you had more throughput you could build more unique applications. So they tried order books. But the problem with order books was because uh the bids and ask uh you'd have resting liquidity on the exchange so all the uh rescuing uh uh bids uh would be an individual um update and so from a blockchain standpoint uh people could essentially pick you off and pay more to access uh uh these individual kind of like resting liquidity and you would the market makers would get adversely selected and because there was a lot of toxic flow. And so we kind of experimented with the order book design and found that because of the blockchain architecture um updating bids and ask on-chain was not super profitable for the market makers and so what they ultimately did is widen their spreads. So on-chain order books was not a super sufficient uh solution and now within the last call it six months to a year we have the proprietary AMs um which are very unique um and these are actually allow you to update the entire pricing curve with one Oracle update. And so what has happened is these proprietary AMMs are really I would say um the blockchain native design for order books. If you look back um uh Chris Dixon has this post about talking about uh we kind of just copy what we had from the existing system into the nude system and then eventually we find the native way to like build these applications. And I would say prop AMMs are actually the like crypto native way for order books and price discovery where it's instead of passive liquidity like an AMM it's active liquidity.
SPEAKER_03So so it was two so you have so we experiment with order books on-chain but because of how many updates were required it wasn't efficient and people could ARB against the existing centralized exchanges.
SPEAKER_01Okay. What proprietary AMMs do is allow you to update essentially your entire price feed with one Oracle update.
SPEAKER_03And so it Wait sorry could you explain the prop AMM side a little bit more? That's that's pretty interesting.
SPEAKER_01Yeah um so essentially what has happened is that proprietary AMMs uh now have are able to send uh I believe it's like 50 to 100 Oracle updates um per second um to uh the leader and every time they update their uh uh the price quote they update the entire kind of curve so to speak of liquidity of liquidity and that's a hundred times per update and and they're doing it faster and faster wow um and so because it's not resting because in the old model you'd have to update each individual like hey like you know like you have the order book and the depth of liquidity you have to update each one. With uh prop AMM you can update one Oracle that updates that entire breadth. So if I want to sell a thousand sole at 120 and an order comes in it'll automatically move my price up or down for me on my rest liquidity or am I thinking so the the Oracle update will give you a quote for that and uh yeah now they're scaling it from like one million like market buys and sells they want to do 10 billion up to like 100 million. So as a market maker I'm more comfortable with this because my liquidity is dynamically moving based on the you're you have a better understanding of uh like um not only like the order flow but also the the dynamics for um like what you actually quote instead of like your quotes being sale because that was the real problem. The quotes were sale on chain and you say hey oh shit like the price moved I need to update my price and they couldn't update the price fast enough and so someone would pick them off. And so because of that they said shit like I have to quote super wide like if the price is is $200 I'm gonna quote 202 on an order book on chain because I can't update my liquidity fast enough.
SPEAKER_03I would never want to be a market maker and that's there.
SPEAKER_01Yeah and so now this essentially new paradigm of liquidity provision has allowed market makers to update their quotes simultaneously and much more frequent where they feel comfortable type quoting tighter because they have a better idea of what the true price is.
SPEAKER_03So it the amount of like computation that's happening is like crazy to think about that like 100 Oracle price updates per update. Yeah. And then that obviously like magnifies into everybody's positions, trading leverage like and so now um I'm trying to wrap wrap it back to your original question of it's okay this is interesting I'll come back later.
SPEAKER_01Uh around like hyperliquid and uh Solana is so we essentially experimented with all this um now trading is pretty sufficient. I think people would be surprised I was looking at the numbers today Solana is averaging about four to five billion of daily trading volume uh daily um I think hyperliquid for the last 30 days their average was about six billion. How much of this was through the prop AMMs we're talking about uh it was uh I I haven't looked at those numbers my guess uh it's probably conservatively 80% probably 90% wow so um yeah this is a massive change um uh AMMs are just getting smoked so is the is your are we to wrap back is the view that the prop AMMs is the winning design for the global exchange? Um or no so um yeah so because of that uh quotes are now able to be much tighter um okay so I think that is one important distinction what hyperliquid did was they prioritize cancels. Um so the they still have an order book but they said if you're a market maker and you need to cancel your order because your worldview has changed on what the true price should be, we'll let you skip the line uh for all the taker liquidity and you'll prioritize uh interesting your cancels. So what right now what has happened is different experimentation in the microstructure of on-chain trading.
SPEAKER_03Um is the canceling a little delayed versus the prop AM structure like wouldn't you want accurate price discovery versus a bunch of cancellations?
SPEAKER_01I'd be probably out of my depth on this they I I mean right now it seems to be working um and like I think we're nowhere near probably the final design on like something like Solana like there's still these iterations but prop AMs have essentially enabled much tighter yeah much tighter execution.
SPEAKER_03But the prop so the prop AMMs have allowed market makers to re-enter and be comfortable with providing a lot of liquidity which creates tighter spreads more trading more volume. Okay.
SPEAKER_01And so when I so going back to kind of the differences um I think the goal in my opinion is like what percentage can you potentially take out of that 20 trillion plus of daily trading volume. And I think from the regional exchange standpoint, I think they're obviously great exchanges I think Binance does 50 to 100 billion a day. Maybe uh don't quote me on that.
SPEAKER_03But it's a very small portion of the 20 20 trillion.
SPEAKER_01Yeah but it's I mean still like it's still it's a m it's a massive number. And so I I think the goal is to eat into that big number um for all the reasons that we discuss I think the kind of global exchange is the better product that will win long term versus these regional exchanges because in my mind they should be the leading indicator for price discovery. But to your question on like the convergence, I think it's much harder to start as a centralized menu and then build a global exchange than to start as a global exchange and then build a hyperliquid. So all this was kind of preface for saying I think if you try to build a global exchange it's much much harder from an engineering point of view because not only are you need the throughput for um enough transactions per second, but you also need to propagate that trade data. So it's not only you just need big pipes which is why we're investors in double zero because we think throughput is going to be a prerequisite to enabling this. So like the amount of throughput that we have today is still trivial to compared to what we think we're going to need. And so this was all kind of like you had to go from low throughput to high throughput you needed to scale execution from AMMs to order books now to prop AMMs. And then now the next step in my mind is going from single leader to multi-leader to have the regional kind of order flow and faster time to inclusion. And then if you build that I think you can build a better price discovery engine than a centralized venue. And so in the short to medium term I honestly think hyperliquid probably kills it because they have built such a great product and they have the assets that people want to trade and they have no KYC. And so from that standpoint I think they're able to continue to make money and be a great product. I think their virtual machine kind of sucks um but that's a different conversation of just like building different applications within their ecosystem um but if you can build this global exchange I think this is actually going to have a lot more trading volume than any regional exchange. And because hyperliquid didn't go down the KYC route or regulated route they're not going to be able to offer a fiat on ramps the credit cards, etc.
SPEAKER_03That's pretty interesting. Is there any counterpoint or counterview where Hyperliquid is just such a straightforward product has all of the liquidity that trumps multiple exchanges on Solana even if it might be faster time to inclusion more accurate global price that you would just use hyperliquid instead?
SPEAKER_01Yeah and that's kind of like and this is I think the nuance that's obviously hard to debate on Twitter but the bet that we're taking is this time to inclusion latency will matter if you want to build a global exchange. If you don't want to build a global exchange and this is irrelevant then a lot of the things that we talked about do not matter. You'll have these regional exchanges that dominate the world which I would argue is either Binance or hyperliquid or the New York stock exchange you're kind of the juggernauts. If that time to inclusion and it doesn't or it does not actually lead to better price discovery some people have said like this is a very clever design but it wouldn't actually lead to better price discovery.
SPEAKER_03If those things are false then yeah it won't matter um where do like the where does the massive supply of liquidity like what which design does that favor? Like if I'm a huge liquidity provider, um big market maker, do I want to provide liquidity on Solana with global exchange or do I want to provide basically I'm wondering like where the supply comes from and what that favors um market makers today favor centralized venues because that's what they're familiar with.
SPEAKER_01It's kind of like you ask someone what they want it's a faster horse. They don't want a car.
SPEAKER_03That's the crazy part is I totally agree with you and I like the global exchange view that Solana's going in, but I could see retail and market makers just picking hyperliquid right now.
SPEAKER_01Yeah for sure. I mean and so part of that is you want the retail order flow. What market makers actually make money on is the people that are uninformed which is historically retail. So Robinhood originally made all of its money from Citadel because they sold their retail user order flow to Citadel and Citadel would make a small amount of bits on each trade. And actually like I mean like I know originally they got a lot of shit for like payment for order flow but I I think it's a generally a win-win uh they get better execution those trades are not sent to the New York stock exchange and they actually get better fills. And Citadel makes a small amount of money. They actually did Citadel also invested in Kraken because they want Kraken's order flow. They led this round last round. So I think also when I think about this global exchange, the second part of this is who gets the most global retail order flow? The HFT is only one part but if it's just HFT against HFT then like they're also not going to trade there because they're smart people against smart people. You want the non-toxic order flow which is just global retail people um which is why I also think that uh global exchange will win. But yeah going back to like the market making side market makers are much much more comfortable today with a single server because that's what they're used to. They're used to co-location they're used to pinging the server canceling uh orders updating orders this new paradigm of prop AMMs and um not first come first serve because historically it's a latency game your first message in if you're there you get served first um the blockchains because it's global and not like as latency sensitive um will most likely be like batch options um so it's a different way to market make so a lot of market makers today are kind of kicking and screaming and saying like this is dumb and it could be dumb like and and this is kind of again the dice that we're rolling and saying like we know it's different, we know it's weird, but if it can enable kind of glob a better leading indicator for price discovery um it's globally fair then everybody will move towards this new system um and this will be the place where majority of kind of trading volume happens.
SPEAKER_03That's awesome I I mean I could see retail favoring the global exchange for to act on the information from your iPhone example. Like I would want as a retail user I'd want to use it. The crazy part though is I wouldn't want to trade Apple stock though I'd want to do like prediction market betting on specific bottoms up things which eventually drive the price. So I would really love like a prediction market on Solana that benefits from that.
SPEAKER_01Well wrapping back to our chart and 100 million to liquidity the reason why I think that's interesting is because if you can build this kind of liquidity hub, then all these applications almost act as uh order flow or front end to this liquidity engine. And I the reason why I think they're some of the fastest growing companies in the world in zero to a hundred million in revenue is because they're tapped into this giant liquidity engine.
SPEAKER_03How much does the ecosystem around the global trading price discovery matter? Like do you care if hyperliquid builds out an ecosystem of applications? Do you care if Solana are you only interested in the I Sorry the trading and execution set. I think um because it doesn't seem like hype is overly interested in building out a suite of products. Yeah.
SPEAKER_01Solana clearly has that I I think they've built a good product which is why I think like it's a great regional exchange and again they should get all the flowers um for as much money as well. It's an incredible product yeah um I think it's more to me like that credible neutrality ish where it's like why some people are like uh tempo chain um like uh I'm like super bearish just like stablecoin chains um it's maybe a separate conversation but like why not build on like uh strike chain um or tempo chain um and it's because I think like it's kind of like the same thing that happened with web too like you can get rug pulled with your APIs. Um like I I'm building Zenga on Facebook Marketplace and then Facebook's like holy shit they're making so much money we should capture that let's let's pull their API and so I think not that um potentially hyperliquid could do this if they improve their virtual machine um but my understanding it's relatively limited and worse than like Ethereum's virtual machine so it's that's pretty bad.
SPEAKER_03Um but with why aren't they interested in in solving their virtual machine? I'm just curious.
SPEAKER_01I think they probably will try to make it better but again it's it goes to different sets of engineering. I think they're very good exchange engineers. But building a globally distributed trading product where throughput matters a lot you have to send literally uh tens of gigabytes per second of data information um and that's probably on the low end like globally uh like synchronize that like do the trade like it's just a different set of engineering. And so I think like Anatoly uh at Qualcomm and Dropbox was doing um I think like kernel design if I remember correctly but he was like optimizing um like VMs um and so I think they just had like a special skill set that allowed them to build a highly performant virtual machine. And I think uh Jeff's background is more in HFT and trading which is why they built an exceptional trading product um but they may be lacking in the virtual machine. Now they have a fuck ton of money and they could maybe like build it over time.
SPEAKER_03So I wouldn't I think everything will get better. I'm definitely just interested in like rel which gets better on a relative basis.
SPEAKER_01I think so like my concrete like answer is like over one to two years I think hyperliquid probably people will start dancing upon the grave of the high throughput chains um because it's just making a lot of money um in terms of revenue and the assets that they can trade. It depends on how long um this multi-leader thing can play out I I think it's as important as kind of a product design or product feature as going from low throughput to high throughput. And the difference between low throughput and high throughput was Ethereum versus Solana. It was massive um I think it's as massive to have multiple leaders um as it is to go from low throughput to high throughput but that well how far is Solana from the world you're envisioning like I didn't ask that. So it depends on what engineers you ask uh the Anza team says that they're gonna do it this year, which I think is unlikely. I would say a more realistic timeline is end of 2027, maybe 2028.
SPEAKER_03So but there'll be milestones along the way updates along the way like so okay. But it's but basically the idea of that global market with leading price discovery in regional markets leading to exchange okay.
SPEAKER_01And it yeah and that's why my pin tweet is the goal of blockchains is to be the leading indicator for price discovery.
SPEAKER_03Okay.
SPEAKER_01Because I think it's a race to build that I think a lot of people don't know they're in the race but I think if you and this is all just in context of kind of the underlying infrastructure um and then some of the applications that tap into that liquidity but uh for the underlying applications to actually be or the infrastructure to actually be valuable I think you To focus on trading.
SPEAKER_03Um I just find it hard that you could win on price discovery without having the economy built around you. Like I feel like you need so many different types of applications feeding into the price discovery. Like if you only focus on perps and trading and have no other apps or ecosystems feeding into that, I feel like it's really difficult to win.
SPEAKER_01So it's it's the liquidity and the assets and then ability to access those assets. So like it what Solana has been doing recently is essentially going down from like top like traded assets on CoinMarketCap.
SPEAKER_03It's hilarious.
SPEAKER_01And then like now trading them. Like, but if you remember Monad's drop, uh the primary place for price discovery was on Solana. It wasn't like any centralized venue. Uh it was actually on Solana. And so it's a good strategy. Yeah, if if you think about that, like you you need all assets, um, not just crypto assets. You need equities, you need Forex, um derivatives. Um, I think the reason in the short to medium term, too, like Solana has just has been hard to actually be performant enough to do a derivatives platform. Like um the popular platforms uh today, like Drift on Solana, um, it to no fault of their own. It was just Solana, I don't think in my mind, was as performant as they needed to actually compete with something that was centralized, um, like a hyperliquid that was co-located uh from a data center standpoint. And that gave them enough of an edge to actually win the perps, where Slana, because of the decentralized aspect, was not performant enough yet to actually build the perps product. So they've done well on spot, but they have not done well on perps. And I think what you'll see as these block times come down as throughput increases, um they'll be not only able to do spot, but perps as well.
SPEAKER_03That's super interesting. So there, I was gonna ask you why hyperliquid won on perps, but I guess that if that makes sense.
SPEAKER_01Yeah. My questions or answers have been a little bit wandering, but no, no, it it totally it totally makes sense.
SPEAKER_03It's just such a crazy dilemma because I agree with you completely that the leading place for price discovery and that bubbling up from regional markets and having the best engineers to act it is the winning strategy. But it does seem like people just want the centralization and speed of the hybrid liquid writer.
SPEAKER_01People don't know what they want, and it's totally fine. Uh uh, I think the Steve Jobs quote is like people don't know what they want until I tell them. Um but and it's totally fine. Like these are like super niche technical topics that most people don't care about. Um, but I think the the reason why the leading indicator for price discovery in my mind is the North Star is because that's the true price. That's where people view as like this is the most liquid venue. This is the place where price discovery is actually happening. Um I think that design doesn't break physics. Uh it allows kind of equal and fair access, and I think it can work. Um now there's a lot of things that like again, market makers are uh unsure. Um TBD on like can this actually be the leading indicator for price? Does that actually work in practice? But from the physics standpoint, I think it's it all makes sense. It's not uh impossible.
SPEAKER_03Nice. Do you do you think there's any like any value of my idea or idea just talking right now that like having the economy built on the chain leads to better price discovery? Like I'll give you a rough example, like a major liquidation on Solana on a bar-len platform influences the price of a token, influences price discovery. Like that would be easier reflected if everything was built on one chain perceptually? Yeah, I agree. Okay, yeah.
SPEAKER_01I mean, it's it's like you you want the depth of liquidity, you want the assets, and you want it to be kind of globally available. Um like payments are super easy. Um uh you can like originally when like the high throughput race started, everybody was quoting like the TPS and uh essentially like uh simple uh payment transfers. Um but it's very easily uh paralyzable because uh if I'm not sending if I'm sending to me to you and someone my mom's sending uh dollars from her to my brother, like those transactions are not conflicting, and you can with GPUs or multi-core performance easily paralyze those. And so you can like scale that super easy. That makes sense. The conflicting transactions are the hard part.
SPEAKER_03Um I guess to round out our conversation, like we went pretty deep on trading specific names. Yeah, no, it's good. To zoom back out though, are you I'd be curious if you're like just generally bullish crypto here or not?
SPEAKER_01Um I think my because I know what horses you like.
SPEAKER_03I'm just wondering if you like the farm.
SPEAKER_01Like I've never been good at like the short-term time frames. Um I would say that has never been the game that we try to play. Actually, like, yeah, I mean I have met, I would say, a handful of people that are much better at kind of asset selection. Um and I've probably met maybe one person that's good on like the short-term time horizon. And so like, I want to learn um and figure it out, but I just I don't think I have alpha. I would say my broader um conclusion is that my biggest fear was always that the market would be based off fundamentals. Um, but I was also very bullish that market would be based off fundamentals. It it was just that transition period where because like valuations would get destroyed, yeah, value or valuations reset, but when that happens, then the industry will become more legitimized because it's all based off revenues. Um, and I think now crypto is no longer getting the pass of saying, um, hey, we trust you and trust your story. Um, we need to actually see you producing revenue. Um, and so from that standpoint, it's painful. And I don't know whether this is kind of the opportunistic time to buy. Uh I don't know. Um, but that being said, the market now is starting to focus on fundamentals and how much revenue you can generate. How we've always thought about the market was what are the products or services that are actually net better products that people will pay for? So you'll generate revenue. And by a byproduct, in my mind, generating revenue is essentially um showcasing that people appreciate your product or service. And so now it's really up to the products and the engineers to start making revenue to validate to allocators in uh kind of the market that people appreciate their product. Um, and so from that standpoint, I think there are certain applications like trading and trading derivatives that are doing fairly well. Um, there's others that I think are promising that will just maybe take a little bit longer time to have steady cash flows and steady revenues. But overall, I think it's very healthy for the market to kind of go through this. It's painful, but if we can essentially get to tokens valued on uh some cash flows or PE ratios, then I think that's a much better place for the market, and you'll likely have less massive swings, which people could argue good or bad, but I think nah, it's good.
SPEAKER_03Yeah, it's kind of crazy because like we're we're we're kind of feeling like we want fundamentals in crypto from your answer. Yeah. But on but on the flip side, like gold is down 35% today, which is an entire Bitcoin market cap. Like and like, you know, these AI projects are walking into venture funds in Silicon Valley, and you know, they leave a job, there's four of them, and they demand five billion dollar valuations with no products. Like it it like the dichotomy in the markets are just ridiculous to keep your hands on because like what requires fundamentals, what doesn't, what's a meme? And it feels like they're they're all all of them just at different stages.
SPEAKER_01Yeah. I mean, it's all at least should be based on the expectation of producing some cash flows. I think people within the AIR markets are just expecting the cash flows to be so massive that even if you have zero revenue from day one, I think X and I raised five billion or six billion in the first round at a twenty-five or twenty-six billion dollar evaluation.
SPEAKER_03Um SSI was one bill at 32 billion, making machines was at what, 10 or 2 at 12, I think. Yeah.
SPEAKER_01And so I mean, people's expectations are just like they'll figure it out and it'll be such massive outcomes. But in large part, I mean, eventually people run out of money to fund these things, kind of like in crypto, where they're like, hey, I've given you X amount of years and lots of dollars. Like, show me where the real revenues are, otherwise I'm not giving you another dollar.
SPEAKER_03I feel like it's slowing down a lot.
SPEAKER_01Yeah, I mean people people want to see revenues. Like, um, and I think this goes back to one of our like LP conversations. It's like, and it's harder because it's less tangible. Um, like it's harder to quantify, like, people don't really care about high frequency trading. Like, if prices are marginally better, maybe that's not sufficient. Um, and so, but with AI, like you can use cloud and build an application, or your Tesla can drive you across town without touching the steering wheel. Like, those are things that you can observe. And so I understand the hype and uh AI and I'm super bullish robots, and AI is super cool. Uh I think for the record, uh Grok is underappreciated.
SPEAKER_03But I I asked you yesterday we were walking around and I asked you your favorite AR project.
SPEAKER_01You said grok, and I was like, I don't know if it's I I'm pretty sure it's contrarian versus yeah, it's mostly just like what we've seen with uh GPUs is that you throw more compute at it and the algorithms are generally getting better, that the combination of better algorithms and more compute makes better models. Um I think Elon is the best at world in moving the atoms around uh and building things in the real world the quickest. I think he built the first uh Memphis data center uh in like 20 days. Uh it was 100,000 interconnected GPUs, the world's largest data cluster. And now he's going to build, I think, a million.
SPEAKER_03Um I mean Meta was building data centers under tents at incredible speeds. But they lost a lot of cloud. I mean, they just they paid $100 million a person and they attracted all the mercenary talent and they fell off. Like that's what happens.
SPEAKER_01Um Elon cracks the whip.
SPEAKER_03Elon is good. I mean, Grok feels like just the global source of truth.
SPEAKER_01I think it's I mean, today I think it's a worse version of Gemini and Claude. Um now over the long term, the my bet is the the vector, the compounding uh of being pointed in the correct direction with faster, like more amount of compute. Um, and generally Elon being at the forefront on the algorithm side that the model will over time be the smartest model. Um today that is not true. Uh Gemini is great, uh, they're very good. Uh Claude, I've been blown away by Claude.
SPEAKER_03Um Claude is I mean, my favorites are the last seven, eight months have been Gemini, like underallocated to that position, but they're just absolutely incredible. Um it's just such an incredible team. Like Demis, like if you read the Maniac book or like read all them, like his history, like they are some of the smartest guys in the room. They have all the data, like TPUs are incredible, like they're slowly releasing products. You check your email or Gemini, like you'll see the products released. Claude is miles ahead on coding, and you could rationalize investing tons in anthropic because they could create $100 billion companies coding. Yeah, Chad GPT's, I think, isn't is in a tough spot. Like, people don't care about 5.2. They're getting sued, like a kill sentiment, and they might lose that lawsuit. Like they might, it's in everybody's interest for them to go public, even Elon's. He would own a bunch, I think. Um, but I am bullish grok because it seems like Elon will kill it, and it seems like him merging it with multiple of his companies is a major benefit.
SPEAKER_01Yeah.
SPEAKER_03Yeah. Like imagine the data from space that he could put in grok.
SPEAKER_01I mean, hey, mine, mine is more simpler. I I just think that they're gonna I mean the the Twitter data and real-time data is like it's a huge differential. It is, yeah. Um and then but more so just how quickly they move on uh assembling all the uh GPUs.
SPEAKER_03I think I told you this yesterday, but your thesis on Grok is pretty similar to your Solana thesis.
SPEAKER_01Yeah, it's kind of dumb.
SPEAKER_03No, it's not dumb because like Twitter and like Grok is like localized real-time data, news, and information. Like it's very similar to your Solana thesis on trading.
SPEAKER_01Yeah, I mean the Solana, the funny thing about Solana, we we under we felt like throughput was gonna matter a lot, and Solana was just unapologetic about scaling the hardware and the throughput requirements. Um I think the AI kind of like thought process is similar. It's like uh algorithms are getting better, and at least for the moment, the more compute that you get, the better the model gets. And so uh which one's growing on the fastest factor.
SPEAKER_03That is pretty cool. Do you think that uh maybe two like random grab bag questions? What do you think open AI and anthropic trade on for Solana or hyperliquid?
SPEAKER_01I have no idea. Uh I haven't been following the AI market that much. I mean, I appreciate the products, but the the rounds and valuations I haven't been following up with. I do think it's interesting. Like now playing around with Claudmore um just like the past month, uh, it seems now that you almost need to push kind of like an execution. Uh, like you have to focus on execution for the underlying infrastructure to be valuable for the blockchain standpoint. You may have to focus on coding. Um for like I don't think it will be the only thing by any means, but the fact that the applications can like build and like almost like uh I don't know, not become alive, but like build the tools that it needs like on the fly. Like if you understand coding, you can just make things. Uh and if the AI can just like build its applications or find the data, um, like it's amazing.
SPEAKER_03Uh I mean the we're we're like the 900th podcast to mention CloudBot probably this week, but like the idea that it just in plain text it will automatically build the products and services it needs to achieve its goals and like do that and open source it and leverage other people's is nuts.
SPEAKER_01It's cool. Like I kind of think the Cloudbot's more like uh Tunks and Cubes for Twitter.
SPEAKER_03I mean the Mac mini stuff is is so stupid. Like people buying Mac minis to just use the Anthropic API, like it just annoys the hell out of me. Like just it's cool. I think it could drive like real open source evangelism, though. Like it it could if people run local models at home, and they should. Um but I'm hoping nobody gets gets hacked messing with CloudBot.
SPEAKER_01But the big thing is the robots. The robots. The robots actually worry me. Uh because you have in a lot of sense. I mean, you you essentially get automation um and like when you think about the jobs that a lot of people do today, like Uber Eats are just Uber drivers, like that has like the gig economy, so to speak, has become very popular. Uh and if you remove like the Uber or like Lyft or whatever, and they're just all completely autonomous cars, and then you think about how many like things that humans do, from like bristas to like restocking shelves to construction. Uh Elon's talked about robots doing surgery, like with doctors, um the amount of things that humans can do, uh, and that robots can probably do better. That actually like kind of freaks me out a little bit. I've been like uh there's kind of these like sovereign cities that are like being developed, and and not in the biology sense. I think like it's hardcore, um, but just like almost these like high net worth individual like enclaves, or they're almost they're like self-sustaining cities. Yep. Um, I think those I know they're I think popular in like South America or South Africa, or um, but I think they'll probably become more popular in the United States if you have like social unrest, which I don't know what you do with the robots if you just have unlimited robots and like AI is like actually smarter than humans, and then you have put that in a humanoid form factor.
SPEAKER_03Yeah, uh Nick Carter was on the podcast before two before you here, and he gave me a book, My Birthday, uh, by Nick Bostrom on I think it was Nick Bostrom, on uh like AI and philosophy and things like that. And I'm like 25 pages into it, so I'm not far. But the first part is on just if we create unlimited robots that are doing all of our jobs all the time, where does like meaning come from? Where do we like derive our income? Yep. And it I mean, there's so many examples. Like obviously, most human jobs go away, but there might be a couple human jobs that are important, like managing the robots or managing taste. So those people earn insane amounts of money, or we're all even and the money just comes from the land, and like you might be short-term oriented and you might want more money, so you'll trade your valuable you know, land, which might be resources in some context, to me, who's like longer form, so it's it's confusing, I guess.
SPEAKER_01Yeah, but it I mean it I think the it the pace of iteration is just the crazy part with like AI and just the robots will take a little bit to kind of experiment to get the design right.
SPEAKER_03Um I don't think we get AGI until we get robots. That's my view.
SPEAKER_01But you can have probably AGI before you get it in a humanoid form factor.
SPEAKER_03I just think that you you want like the human understanding of AGI, you want to look at something, you want to experience the world, you want to argue with you.
SPEAKER_01Just want more data from the human robot.
SPEAKER_03Yeah, I don't know, maybe not, but I just feel like that's a limiting well, it's not really a limiting factor anymore. We have robots running around.
SPEAKER_01But the the ones that piss me off are the ones in brickle, the little You don't like the livery box? I actually I shouldn't have said that. The the the only reason why there's like 40 outside of my apartment right now.
SPEAKER_03I love when they line up, they're just ganged up. I love it.
SPEAKER_01They're definitely squatted up, but I want them to go in the road as a truck formation and just I I I saw one uh I think in West Palm. Uh it was going over the train tracks and the bright line just smashed it. I saw that on Twitter. And that that that was crazy. Um but yeah, I I don't know. The robots more broadly, I think, is I I I think it's like new grads now are like 30 to 40 percent of them also don't have like are having a hard time like getting a first job. And so like I don't know. There's just a lot of things um that kind of worry me.
SPEAKER_03Uh but I think it's the other thing, like it's just really hard to figure out how you redistribute or if you have to redistribute wealth and UBI in that scenario.
SPEAKER_01Yeah, I don't think the like socialism New York slash California style is good.
SPEAKER_03Like no, I'm I definitely think it's stupid.
SPEAKER_01I think uh very pro Texas and Florida. Um but I don't know.
SPEAKER_03Um I think meritocracy always wins to Texas and Florida's benefit. Like the people who are smartest put in the most work, develop the best products, the Elons of the world should have crazy wealth like they're doing. But but the thing that I'm worried about is there will be a time when we are limited by the robots we could afford, the intelligence, and then it becomes like status symbols. Like if you could afford an extra, you know, Chat GPT 10 deep think for the week, and you have one more use than I do, like you're miles ahead than I might be, right? Like you might ask that to you know change your genes to make you fly. Like, I don't know what you'd use it for, but like it gets crazy when you have more resources.
SPEAKER_01There's this movie, I I don't remember the name, but the premise is uh you live for a certain period, uh depending on how much time you have.
SPEAKER_03Oh, the one with uh I out of time or in time, yeah.
SPEAKER_01I remember that. Yeah.
SPEAKER_03And Justin Timberlake.
SPEAKER_01Yeah, and it's a crazy movie because uh like if you think about longevity, like uh there's like a bunch of different things that are happening with the robots AI and like potentially longevity so early, but if it starts actually working and people like start living 150, 200 years old Ultra carbon style. Like, and but then like you have like your time clock and it like it just gets super sci-fi. Um so I don't know.
SPEAKER_03It's it's but that's I mean, I am like I watched uh Blade Runner last night, the new one.
SPEAKER_01I don't know if you saw the original one, but I haven't seen the original, I've watched the new one a couple times.
SPEAKER_03So it's like uh it's like you know, future sci-fi world, you gotta gamble crazy to get ahead, and you gotta gamble with your life pretty much. And we're kind of already entering that. That's what I'm talking about. It feels pretty similar. Yeah.
SPEAKER_01It's uh yeah, Bology had this blog post about uh essentially everybody comes in investor, and I think like maybe that's been coming true. But I I think the shitty part is like you're almost forced into it with inflation. Um and like especially after COVID and 2020 and 2021, where they printed so much money, is that if you look on Zillow and house prices, like they've like essentially 2x. Um, and like just broadly, at least from places that I've been looking at, where uh from then. And so, like, unless you've made 2x your net worth or 2x your salary like in like five years, like the most likely asset that you want to buy has doubled in value.
SPEAKER_03When I meet non-crypto people and I tell them we've been our lives or America's been here since 1776, we printed 80% of the dollars the last five years. They're like, that's just not true. Yeah, and I'm like, no, no, that's true. That's crazy. Yeah, that is the wild part.
SPEAKER_01I don't know. A lot of I think uh, I don't know, broadly it's bullish in the sense that people will have more access to hopefully financial services to better information via AI. Uh it's cool what Starlink's doing. Hopefully, the robots actually do give kind of universal basic health like with uh like almost being. Like a doctor for anybody or doing surgery. And so broadly, I think it's going to be good.
SPEAKER_03It's kind of hard to swear it though, because like it's really easy to think of we're all even. The only way to get ahead is to gamble. And whoever gambles and wins has more resources for more AI, more robots, more compute. Like that's an easy dystopian thing to think about. The other side is like cancer is solved, energy is solved, space travel solved. So like I don't know. Like, is it really a dystopian world you're gambling in if you can't get cancer and you can't die? Like, I don't know.
SPEAKER_01I I yeah. And it's yeah. I think that will those kind of breakthroughs in medicine and yeah, there'll be a lot of interesting stuff. Um maybe housing price come down because we can build it with just general robot armies. I don't know. But I don't know if people are gonna I don't know.
SPEAKER_03Would you rather rent and buy a car, rent and buy a robot, buy a house and not buy a robot? Like your decision trees become crazy in that world.
SPEAKER_01Yeah. I mean, I've been thinking about like younger kids, like they're you're not actually gonna like probably own a car. You'll just like essentially call a car okay. Robot taxi, yeah, way most. Yeah. So I'd be okay.
SPEAKER_03I'm obviously okay with not owning a car and doing Ubers. I I don't know if I would ever be okay sharing a robot with other people, though.
SPEAKER_01Yeah, that'd be weird.
SPEAKER_03That'd be right. Like I thought about that. Like, would we pay for a robot to come in our house and clean all of our neighbors instead of us having to buy one?
SPEAKER_01Elon talked about he wanted to get his robots to like 25, 30k a pop.
SPEAKER_03That'd be amazing. Yeah, and then just finance them, call it a day. Yeah.
SPEAKER_01Well, you kind of like cars, like five, seven hundred bucks a month.
SPEAKER_03Well, I guess you made it cheaper. Finance, but you'd also have like some weird monthly fee. Yeah. Don't forget your subscription fee. The way he went with self-driving, he's definitely charging something. Yeah. He could charge me whatever he wants for a robot. Let's do another podcast on AI. That'd be fun. Let's run it back. That'd be cool. Logan, thanks for coming on, man. Thank you. Awesome. Always fun having you on.
SPEAKER_00You're now plugged into the Delphi podcast.
