AI Needs Crypto

26:31I said, all asset markets will be 24-7, 365. All markets will be settled immediately on chain. Humans won't be trading them. Instead, it'll be 100 billion agents trading at a frequency that you can't even imagine. And all those fees will be captured by crypto protocols. And that tweet pretty much kept me in crypto. Like that tweet and the thinking behind that tweet kept me going right through to pretty much where we are today.

26:55Raoul Pal:I was the same at that point. I would just zoom out and say what's actually happening here. Every bank, every financial institution has said they're going to move on chain. Stablecoins have been shilled by the Treasury Secretary of the United States saying it's going to be a$3 trillion market. and here's us lot going oh crypto is going to be awful i'm like same as you is that the outcome has to be asymmetric from here wildly asymmetric sometimes that because you're so close to the wood and i was very close to wood every day you know the problem is that when you're doing a daily show when you're watching a market daily the move between 64 and 65 000 becomes a story and the move between 65 and 66 becomes a story right because we're so close to the wood most people look at 60 ,000 and then they log back in at 70 ,000.

27:45And then to them, it makes a whole lot more sense. But sometimes we're very, very close, which is an advantage and a disadvantage. Anyway, now the thesis is clearer than ever. And you asked me what my thesis is. I think it's maybe coincidental that today the BlackRock report came out. And in that BlackRock report, they basically confirmed, I think, my thesis and your thesis. So for me, the thesis is that up until now, the NASDAQ has had a run. And when I said the NASDAQ, the NASDAQ being a proxy for everything that's AI. And that run has been everyone building infrastructure for what they believe will be AI, right?

28:23So you've got the data centers, you've got the chip companies, you've got the energy companies, you've got whatever it is. And that NASDAQ run has pretty much been everybody building the infrastructure for AI. Great. So we've built the infrastructure, We've got the Frontier Labs. And what is this whole thing about AI? AI is a bunch of agents. That's what it is. When they talk about building out the infrastructure for AI, they're building the infrastructure for a bunch of agents. Fantastic. That's what drove the NASDAQ vertically. That's what drove the Cosby vertical. That's what drove world markets vertical.

28:58Great. They built the infrastructure for AI. They built the infrastructure for a bunch of agents so that the agents could do what? So that the agents could transact. Great. So they built the infrastructure so that you could have all these agents who eventually transact. And I kept asking myself, who is going to be the benefactor of the transactions? And I tried to, I mean, I convinced myself that it was blockchain. Why? Because agents are going to meet on the internet, wherever they meet, they meet in, I don't know what you call it, and they transact and their transactions have to be, they're trustless because my agent and your agent don't know who the owner of the agent is.

29:42It's got to be immediate settlement because if I want to make an API call, you've got to let me call that API, get the data and pay you absolutely immediately. So it's got to be trustless and it's got to be immediate. And I kept saying to myself that the only way that I can see this happening is blockchain rails. But I kept asking wherever I went, wherever I went, I kept asking people, how are these transactions going to happen? Because I needed someone to say to me, it's not going to happen on blockchain rails. And even like last week when I was in New York, I went to one of the big summits and I asked people, I said, I asked one of the big developers, I said, please tell me my thesis is wrong.

30:21Tell me that these AI agents aren't going to use blockchain rails to conduct all their transactions. And he says, they're not. And I said, well, what are they going to use? And he says, well, what do your kids use when they do transactions? and you've got to set limits for them and they're not allowed to go on certain websites. You give them a prepaid debit card. And he says, one thesis is that AI agents will use some kind of prepaid debit card with limitations where they're allowed to spend and not allowed to spend and stuff like that. And when he said that, he said prepaid debit card. And I just thought, no, just a bunch of smart contracts about where an agent can spend and can't spend and what the parameters are.

30:54Raoul Pal:And a wallet preloaded. And a wallet. And a wallet and some kind of proof of identity or credit score if required. And then I started to do research into the number of agents. I was like, okay, well, how big is the workforce in the world today? So the workforce in the world today is 3.3 billion, somewhere around there. Let's just call it for our numbers. They reckon there's going to be, I mean, I think I saw a number earlier today. I don't know if it's how accurate it is, but it says here, there's going to be 1 billion transacting deployed agents by 2029. and i would take the over on that number 10 times 10 times i would take the over but hold on get this they're going to be executing roughly 217 billion transactions a day okay so take the over take the under i would take the over on it 100 times but we've just seen that meta muse thing and they integrate it immediately with shopify's um with paypal with everybody it's like okay yeah and so what are these agents going to be doing agents today we today we know agents as i give you a question you give me an answer i give you a task you give me the solution but the new economy is actually every agent interacting and paying each paying each other a minuscule minuscule fee for the service what's a service maybe it's an api call maybe it's a data call maybe it's maybe it's something we don't we haven't even been able to work out yet but it's little microtransactions, and as this says, 217 billion of them per day, right?

32:29Now, again, I've challenged myself a million times to try and come up with who the custodian of that is going to be. And this morning, BlackRock published a report, and they basically confirmed what I've been suspecting, where they basically said, to actually quote them, they said something here, which I think was, this is the buzz line here. They said, at the core of this convergence, AI and digital assets both arise from common foundations. So they go to, they say, the rise of agentic AI and machine-to-machine payments will likely increase demand for blockchain rails. Existing rails, such as ACH and card networks, support substantial automation, although the onboarding requirements and settlement economics make them less suited to always-on low-value transaction requiring programmable execution.

33:13Basically, in that line, what they said, out with the old financial system and in with blockchain. And that was like, for me, the cherry on top where I actually made the show today. And I was like, I've been waiting for confirmation of my thesis. I've known the thesis is right. Everyone's been telling me that the thesis is right. But when you read a report like that, and BlackRock basically confirm exactly what you've said, I think we now have the biggest, the most exciting opportunity in the world. and I actually woke up this morning. I tweeted, I said, we're in the most exciting opportunity in the world.

33:45And again, I'll show it to you with a chart and this is a really basic chart, but it's just showing you. The purple line is AI returns. So AIQ returns. The other two lines are crypto returns, and they diverged on 1010. This was the infrastructure for all the agents that eventually are going to do payments using these rails. So up until now, you can say the infrastructure has been built for all these agents. The frontier models are building. If you've used GrokBot, I think GrokBot is probably the best representation because I've got like 10 bots inside my GrokBot. I've got like 10 agents doing stuff.

34:25And I know that they're going to transact. And so the infrastructure that was built with this chart was made for the transactions that are going to take place on these charts. And then from that point on, it just becomes, I know for you, it's a very, very broad selection. So for you, I know it's a, you just want to own the very broad rails, but for me, it actually goes into which altcoins are going to benefit the most.

34:51Raoul Pal:I want to get into that process. So you've laid out your thesis, which I completely agree with. I wrote my whole agent thesis about a year ago about all of this. And it's, it became so fucking obvious that this is almost the purpose of blockchains. It was like it was built for the machines. So you see that. We see the whole financial system coming on board. We see this rise of speculation. We see the rise of prediction markets. We haven't even started with identity. We haven't even done any of the Web3 stuff yet that still is going to come on chain in time. So it's like, yeah, I get that too. I don't think there was a, I don't think that this was a full bear market in terms of the full cycle, I feel like it was a mid-cycle.

35:35Raoul Pal:And that, like you, I think there's potentially a very, very, a longer and more dramatic move to come. I think you're talking about Bitcoin. And for Bitcoin, to be honest, I mean, I don't really care. A lot of my wealth is in Bitcoin. I'm very bullish on Bitcoin asset. But you're referring to Bitcoin bull markets and Bitcoin bear markets. I don't think crypto altcoins have had a market, a bull market yet. No, they haven't. Because even And ETH has been in that bloody triangle, that wedge pattern for five, six years now. So I agree. Let's define a bull market not as prices go up. Let's define a bull market as companies and protocols actually make revenue.

36:12So, I mean, you can define the 2000 bull market around an internet bull market. In my eyes, that wasn't a real internet bull market. It was the promise of the internet bull market. And I think with crypto, we've had the promise of the crypto bull market. But I think now we've actually got protocols that are making a million, two million, three million dollars a day in bear markets. And, you know, you take like PumpFun, which is, I mean, PumpFun is the most ridiculous protocol for people to launch meme coins. And they're making between one and five million dollars a day, depending on the day. OK, now you're starting to talk about crypto protocols that are actually making real money.

36:49Right. Hyperliquid, you know, making it's hard to say, but five million a day, four million a day, three million a day, depending on the day. So you finally got real use cases that are actually working in crypto that are generating revenue. And so I think that this is crypto's first real bull market. And again, you can say a bull market price is going up, but I'm saying price is going up, technology being used, revenue actually being generated. That is for me a bull market.

37:14Raoul Pal:What's hilarious is it's almost like a record low for VC investment because everyone gave up on the applications layer and the applications layers come through. We've had some really big use cases, whether it's Polymarket, whether it's FOMO, whether it's hyperliquid, whether it's, you know, we're seeing an applications layer being built. OK, I want to move on to the next bit. So we've got the thesis. Now you talked about the process around the research. So now you're trying to find the opportunity set within the thesis. So talk me through how you do that. So a quick break in your regular programming.

37:49Raoul Pal:If you're serious about your future, grab my free report called Prepare for 2030. I think you've got five years to make as much money as possible. And this guide will help you navigate what's coming. The link is in the description. Download it now. So first of all, yeah, I say this is, I think, where my investment style and your investment style diverge. Why? Because you're much more macro and you're playing much bigger macro trends. And I'm looking for the opportunities inside. So the first thing is, to me, you've got to ask yourself, when is a token valuable? And the answer of when a token is valuable is it's got to have network effects.

38:30And for those people who don't know what network effects are, even though I'm sure that everybody at Real Vision must know what network effects are, because I think you're probably one of the first people that taught me about network effects. But every user has to add value to the network exponentially. Now, tokens become valuable when there is a network and there's scarcity. So the fact that there's a network is great, but if there's a network but no scarcity and tokens can be minted forever and the token supply is growing as fast as a network, it doesn't really matter. It only matters when there's scarcity.

39:05So the first thing is I look at, the first thing is, is there a network? And again, let's talk about hyperliquid. Is hyperliquid a network? Yes. Why? Because every user adds liquidity to the network. And therefore, the more liquidity you have, the more people are going to trade on that network, right? The quicker you can get in and the quicker you can get out with the lease slippage and the broadest range of positions, that is it, liquidity is a network. Trading is a network. So the first question becomes then, is it a network? Is the network actually growing? So yes, you could have network stocks, economics, but is the network actually growing?

39:43So once it's a network and the network is growing, you say, do they have real revenue? Yes. Great. Then you ask one last question. And the last question is, one second last question. do they have a mechanism to transfer the value back to the token holders? If the answer to that is yes, great, but be careful. If the answer to that is no, probably the best token opportunity that you've ever seen. So let me explain to you what I mean. There's a network, the network's making revenue, and they find a way to reward the users for the value. Everyone knows the value of the network, right? Because all they do is they forecast the rate of growth of users.

40:28They forecast the revenue and they forecast how much of the token everyone's going to buy back and burn. Example, hyperliquid. Everyone's now forecast that hyperliquid is going to buy back and burn three, four, five million dollars a day, depending on the market forecast. They divide that by the number of tokens in circulation and therefore they get a metric for the value of hyperliquid. And we'll talk about the metric in a second. If they don't yet have a mechanism to return value to the users. Currently, those are the most undervalued tokens. So in crypto, the terminology is, do they have a fee switch?

41:04Fee switch means a switch that returns the fees back to the users. So right now, those are the best opportunities in crypto. Protocols that are networks, growing in users, making money, but haven't yet determined how they're going to give money back to the users. So if you forecast and you say, I believe that they're going to give 50 % of their revenue back to the users, and you've got a smart way of forecasting how they're going to give revenue back to users, then you have an edge. So what are the best tokens that I've bought? I bought a token recently called Collector Crypt. It's called Cards.

41:41Essentially, in a nutshell, what they do is they take Pokemon cards, they put them into a vault, a real world vault, and they give you a token representation of the Pokemon card. Now, you may think that that's nothing, but a Pokemon card is a real-world asset, and the token is a representation of the real-world asset on chain which allows people to trade. I'm giving you a very simplistic representation of what they do. Right. Now, the velocity of the trading becomes 100x what it is. They're accruing 100x more fees than if people were actually trading physical Pokemon cards. Now, they haven't yet worked out how they're going to pass that back to the user.

42:20And the reason is because there's no SEC clarity yet of what they're allowed to do and what they're not allowed to do. Best opportunity in the world, because people haven't factored in that they're going to give the fees back to the users. That's an opportunity. So the first, so do they have, is it a network? Is the network growing? Is there real revenue? And then is there a way to give revenue back to the users? That's the first class of tokens. the problem with tokens is though that people value tokens and this is what i'm working on right now which we by the way we haven't cracked yet but i'll walk you through how we're thinking about it so the first the only way that you and i know how to value assets companies revenue generating assets is you take the revenue they make and you multiply it by multiple and divided by the number of shares that they have that is how you value a share right you say what are the total number of shares times that by the share price, that's the fully diluted value of a company.

43:18Right?

43:19Raoul Pal:Great. So everyone takes that valuation into crypto and they say, well, you know, if there's a billion tokens outstanding, right? And the value of the token is$3, that means that the fully diluted valuation of the protocol is$3 billion. Mistake. And that's one of the biggest mistakes that people make in crypto because they don't understand crypto. Right? You compare Coinbase to hyperliquid. Coinbase has, I don't want to mislead the user, so let me actually give you the number here. And this is what I'm spending probably 80 % of my days doing right now, because I know if I crack this formula, the price is huge.

43:53So Coinbase have a market cap of$52.4 billion, which means that if you take the number of shares times the share price, the value is$52 billion. dollars. Hyperliquid, so this is Hyperliquid, they have a$20 billion market cap, but an$89 billion fully diluted. So if you take all their tokens times the price, it'll give you$89 billion, but there's only a$20 billion market cap. Now, this is where people get things wrong. They take a valuation of a share and they try and import that to crypto. There's a fundamental difference. When you have a share, all the shares or most of the shares are tradable in the market and there's a price.

44:36With crypto, it's not like that. Firstly, a lot of the tokens may not be in the market because programmatically they're not in the market. Like if you think about Bitcoin, there's still a million Bitcoin that aren't in the market. The second thing is that so already you've got the market becoming, instead of being that big, it's becoming that big. because these tokens aren't in circulation yet. Then crypto has got this thing called staking. What is staking? Staking says that if you want to launch a token on Hyperliquid, you have to lock away some of those tokens. So that makes the pipe even smaller.

45:11And then you've got other mechanisms in crypto which make the pipe even smaller. Staking, et cetera, et cetera. So now what is the price? It's not the fully diluted valuation. it's a function of this pipe. The tighter and smaller the pipe and the more buying pressure...

45:32Raoul Pal:Yes, the actual available supply versus the total demand for the thing. And the available supply... It's the pressure in the pipe. So with the stock, the pipe is this big. There's no way to make shares scarce. You never lock up shares to get benefits or whatever else. So shares aren't scarce. The pipe's always this big and the water's always flowing. Buying pressure is always flowing through this pipe. Unless they're doing buybacks like the big tech companies did. buybacks is just the company buying back its own tokens and never burn. They don't buy back and burn. They just buy back and keep it on the balance sheet, right?

46:01Now let's talk about what happens in crypto. Hyperliquid has$90 billion worth of shares. $70 billion aren't being emitted. They aren't being emitted for the next 10 years programmatically. So all of a sudden now the pipe's only 20 big. Of 20, you have to lock up$50 million worth of Hyperliquid to launch a token on Hyperliquid. And you have to keep it locked up and staked for as long as the token's listed. Now the pipe gets really small. And then Hyperliquid have this thing called the buyback and burn mechanism, where they take a percentage of their fees and create buying pressure inside this pipe.

46:34So to me, token valuations are a function of the pressure inside the pipe. So what do you have to work out? You have to work out how thin the pipe is and how much buying pressure is going to be in that pipe. And that's how you work out token valuations at a certain point in time. It is a function of the pressure in the pipe, which stocks don't have because you never lock up stocks. You never lock up stocks. You never stake stocks, et cetera. You can't use your stocks to participate in the network.

47:07Raoul Pal:This is also one of the reasons why crypto can be very volatile. Yes. Because the liquidity is never enough for the demand or the supply. In a bear market, there's not enough liquidity, so price goes down more. because the free flow is so small. So in equities, it's a free flow. So the free flow in SpaceX is like 5 % or 10%. It's ridiculously small. But yes. But SpaceX doesn't have automated buying pressure. It does via the NASDAQ inclusion. Once you put it into an index, you kind of get all the 401k money every two weeks. The same will happen in crypto when we create real indices. But imagine with every SpaceX part that was sold, they had to buy back SpaceX shares.

47:59Then can you imagine with a 5 % free float what that equation would look like? And so you said to me, what am I looking for all day? I'm looking for pressure in pipes. What's the scarcest token with the highest buying pressure? Because everyone's valuing these protocols the same way they value companies at terminal point. And yes, hyperliquid is valued at 90 billion terminal point. But today, the pipe is only 20 billion thick and there's buying pressure. And so that's basically what we're looking for, or right now what we're looking for. We're looking to find networks which have a lot of pressure in the pipe.

48:34And if you can calculate the pressure in the pipe correctly, or better than anybody else, then that's free money in crypto at the moment.

48:41Raoul Pal:Okay, you've talked about your research process. So you mainly focus on these kind of opportunities. So these tend to be application layer, higher velocity, low free float, but still a reasonable size, not always. What size are you looking at within these? So I think the layer one war is largely one. So in crypto, I think the layer ones have largely been determined. I think Bitcoin is the layer one for store of value. I mean, no surprise. I think in terms of the other layer ones, I think it's ETH and Solana have basically won the war. When I say ETH, I don't think ETH is won, but I think the layer two is on ETH.

49:20The Robinhood chains, the base chains, I think they're going to be the winners because they've got the credibility of settling on Ethereum and they've got the speed and agility of everything else they need. I think there's two more use cases that need to find a home. One of them is a very big prize. The Agentic L1 hasn't been won yet. I think there's a couple of contenders for that. I think Sui is definitely one of them because I don't want to get too technical, but it's an object-based language and an object-based chain. And that is much more suitable for AI transactions. Number one, I think Nier protocol.

49:57Nier is building a lot of tools. It could be Solana, it could be Base. So I think the Agentic L1, that war has still to be won. I think the TradFi L1, when I say TradFi, I really mean traditional finance L1. Where are the DTCC going to migrate? Where are the existing clearinghouses going to migrate to? I don't think that's been won yet. There's a whole lot of contenders in that race. But ultimately, I think the L1, the block space debate in crypto is pretty much finished. Right? I also think that the L1s haven't found a way to really capture value. so if you're a eth holder guess how much value you captured from the robin hood launch and robin hood launch just to give you some numbers in my opinion was the most perfectly executed launch that crypto has ever seen robin hood launched the chain the chain made 50 million dollars in the first two weeks in revenues for robin hood guess how much of those fees eth managed to capture.

51:01$50 ,000.

51:04Raoul Pal:$50 ,000. So I'm going to give you a different way of valuing layer ones that you should think about. If I were to turn the plug off, the switch off on ETH tomorrow, how much value gets destroyed and future value? All of Robinhood, all of base, all of DeFi, all of stable coins, all of digital art, the whole financial system's future rails, the whole lot. Now, that's a trillion, two trillion. I don't know what the future anticipated value of that is. And it's nothing to do with fees. Fees are irrelevant. They're a red herring. In fact, in technology, fees are a friction. And every technology that's ever scaled, we've seen fees go to zero.

51:50In blockchain, in order to keep the lights on, you need people validating transactions. right? I knew a lot of people validating transactions. In order to do that, they need to be rewarded. So they get rewarded in one of two ways. The first thing is token native inflation. So that's a bad inflation. So if there's token native inflation, it's a bad inflation. It's the same as government inflation because it means the value of the network is being deflated away. The second way is fees. And by increasing fees, you would penalize the chain a lot. So let's talk about Ethereum around your thesis. It's either going to emit enough ETH to keep the lights on, and we're going to hope that in perpetuity this can happen, or the fees have to be so high that the miners have to be rewarded because the more security in blockchain, the more validators you need.

52:39So security in blockchain is a function to a certain point of how many validators you've got in the network, right? That's the strength of a blockchain. That's why the Bitcoin network is so big. so there is also the bitcoin debate as you know around what happens when bitcoin stops emitting coins now it's only going to happen in 21 something 2150 2140 whatever it is but what happens then the fees have to basically fund the network right and then you know maybe they will maybe they won't we'll worry about that in in 2150 if we're still around um but so i think the layer one model in crypto right now it's working but it's only working because it's

53:20Raoul Pal:subsidized by token inflation at the moment yeah see i again i would disagree i disagree with this because if we're right that the agentic economy is coming the velocity of everything goes to the fucking moon agree i agree then then yeah and if you're charging 0.0001 cents for a trillion transactions, it doesn't matter. Agree. You can make a lot of money from a very low transaction fee. I mean, there was a great example yesterday. I don't know if you saw, you might not have seen it, but there was a friend of mine who's a digital artist, NFT artist, Jack Butcher. He sold using the X Money Network.

54:01Raoul Pal:First time this has been done. So people on X Money, so X, you said, pay me 10 bucks and I'll give you a piece of art. he made, I don't know, a million or$2 million yesterday from$8 art. And then he's now got this buyback and burn and complicated mechanism for it. Fucking clever. But low prices, high volumes, he sold 123 ,000 pieces of art. So - I agree with you. And I think the agente economy is very, very, very, very soon. Like I think people talk about, I think it's, I think the agente economy is 2027, 2028, 2029. I don't think it's 2040. I think as the article here said, it's, you know, you're talking about a billion agents by 2028 or 2029.

54:45It's around the corner.

54:47Raoul Pal:So all of your bets based around these basic theses. So most of them are looking for the low free flow where demand is. I've taken my bets on the layer ones. I've got my bets on the layer ones. I've got my bets on Bitcoin. I've got a little bit of money in Ethereum. I've got money in Solana. I've got money in Sui. I've got some small bets on other ones, Nier Protocol being one of them. And you just leave those? Never touch those. Those are savings accounts. The only time that I'll touch Solana, example, is if I may take some chips off the main chain to fund an application that I think is more exciting than the main chain.

55:25But I'm drastically increasing my risk profile then, right?

55:29Raoul Pal:Yeah. Because if I bet on Solana, I'm betting on a billion experiments that may take place on Solana. If I'm betting on one app, then I'm betting on... And what is the percentage ratio between your core holding and your active? Are they 50-50 or are they 20-80? Which way around? So it's a great question. The question is whether or not you include Bitcoin. So if you include Bitcoin. No, include Bitcoin. You see, because Bitcoin is my store of wealth and my store of value, then you're largely skewing it with all ones. but let's remove bitcoin yeah and i'd say probably right now i am see i've had a great run on zcash and i haven't rebalanced my portfolio so probably right now i'm about 56 no probably about 60 70 percent in l1s but only because because zcash outperformed i took a bet on zcash at 100 bucks and yeah it's not 60 it's 17 times that money yeah and i haven't rotated yet because because i look at zcash and it reminds me of bitcoin 2014 i've got some zcash as well i got in about 300 bucks or something and it's doing nicely yeah it's it's exactly bitcoin it's the same people it's the same culture it's the same price moves it's the same issues uh the devs left and then the devs came back and then and then you know like and then they thought there was a hack but actually they were it's i heard all these i've watched this video before and i'm watching it again now, and you know, if there's one mistake I made on Bitcoin, it's that I didn't accumulate 100 ,000 Bitcoin back in the day when I, you know, I mean, maybe 10 ,000 Bitcoin when I actually could have afforded to accumulate 10 ,000 Bitcoin, right?

57:13And I kind of said to myself, when I saw the Zcash thing happen, I said, I'm not going to make that mistake again. I'm going to get in early. I'm going to put the money down and I'm going to forget about the money. But my thesis was sound as hell. and it was very very very tough sitting through that thesis two things happened i don't know if you remember but towards the end of last year two things happened the first thing is the devs left there was a big announcement that devs left they didn't they just left one dev company and started a new dev company but that was very scary and that tested the thesis and then the most important thing that tested the thesis was around the potential hack that could have happened right that's right and that was that was really testing the thesis because yeah because it was is that

57:55Raoul Pal:Is there now infinite supply? And we don't know because it's all hidden. And that's the one flaw that there was in Zcash. And I say it was, that flaw doesn't exist anymore because with the new technology upgrade, they can actually verify that what's in the shielded pool plus the unshielded pool equals 21 million. Before they couldn't. All they could tell you was in the unshielded pool, the non-shielded pool, they could tell you how many there were and how many there should be in the shielded pool. but they could never verify what's in the shielded pool. Now they can. So now they can. So it's made the thesis a lot more bulletproof.

58:35And I'll tell you that sitting through that period where there could have been a vulnerability in Zcash with a large part of your portfolio in Zcash, like let's say that over 10 % of my portfolio was Zcash, and sitting there knowing that this protocol could have been compromised. I mean, those were quite tough days because again, if they were compromised, Zcash could have gone too close to zero. So it was you had to hold with conviction. A lot of people ran away. A lot of people left. They said, you know, I even remember with all due respect to Arthur Hayes, I remember interviewing him at that period and he said, you know, the fact that this did happen made me sell all my Zcash.

59:15Raoul Pal:And I remember. And that's Arthur. And that's Arthur. He said he sold all his Zcash. And he's not, I mean, He's quite astute. You know, he knows the market. And he said, look, that dent. So a lot of people left. And I mean, we held. And that's become a very big part of the portfolio. So I'd say right now, 60, 70 % is in L1s. And then, OK, so I don't consider hyperliquid in L1. So hyperliquid is another big position. The competitor to hyperliquid called lighter, which is the more regulatory. They went the more regulatory friendly approach. And the reason why I took a big bet on Leiter, just to give you some insight, I knew that they were trying to get hyperliquid into the US and that there's a negotiation going on with hyperliquid to bring hyperliquid to the US.

1:00:03Raoul Pal:Yeah. I also know how Trump... It was Trump or Besant who mentioned hyperliquid. It's insane. Trump. Trump. That was when I doubled down on Leiter, ironically. And it's been one of the best bets that I've made. Because how does Trump negotiate? he tells you how amazing your life's going to be if you join him right he shows you how miserable your life is going to be if you don't do what he says and then somehow he and then because you realize how miserable the tariffs are and because you realize so when he said we're going to be bringing in hyperliquid into the US I was like oh and what's going to happen in the negotiation He's going to threaten at some point.

1:00:45Trump is going to threaten hyperliquid with the fact that they're not coming into the US unless. And then the market's going to panic and they're going to go to the next competitor, which is lighter because lighter has gone the more regulatory route in the United States. And that switch I made when lighter was at about$1.80 or something. I don't remember the exact numbers. Lighter is now$5.50. Hyperliquid was at about$80 or$70. dollars. It's at a hundred bucks. So that was a very good switch to make. Yeah. So the majority, so where else are my bets? So again, it goes down to the foundations of what is crypto.

1:01:22What is crypto? Crypto is the ability for anyone to create digital value and trade digital value. That's basically what crypto is. Anyone can create or migrate digital. You can create digital value from a real world asset. You can create it out of thin air. You can create digital value. And if you can create digital value, then you're going to trade it because value is there to be traded. So the majority of my bets are actually in the trading platforms. Very, very, very few of my bets are in non-trading platforms because I don't really care what application you build, whether it's super compute or whether you're selling, I don't know, chips for racehorses.

1:02:00I don't really care. I know you're going to trade it. And I want to be in the places where you trade it. And the

1:02:05Raoul Pal:more leverage the better okay so let's wind ahead to two years time and rand's thesis has paid off but something else came along that he didn't see because there's always something new in a cycle that you're like fuck i saw that but i didn't see it what's it going to be where is something going to break out in this market that none of us are expecting because it won't be what we know always. It'll be something we don't know. What do you think? I mean, like for me, the AI agent thing is so big and it's just, just, just, just, just started. And I think learning their behaviors and their trading and their hedging and they're like everything else.

1:02:50It's just so big that like it's enough to occupy, it's enough to occupy whatever we're doing times 10. So I must say, I haven't really thought

1:03:00Raoul Pal:like way beyond that like i don't know because don't forget in the past there were there were what were social networks going to do with blockchain what we were going to do with id what were we going to do with like data storage you know there's i don't know there's just to me is i always keep my eyes and ears open to something that is entirely not on the agenda today that some weirdo in a somewhere who's figured out um then let's see i like to go back to to core principles and for me core principles is what what is this technology this technology is about the creation of digital value when you have value you're going to let's well you're going to store it i'm not sure that there's much money in in the storage right you're going to send it okay sending is probably the money is probably in things like stable coins algorithmic stable coins partially backed stable coins fully backed stable coins i'm not sure what that application is um you're going to trade it you're going to leverage it yeah but there's also other elements there's the it's a database that is immutable that has authentication built into it there is a whole bunch of parts of blockchain that aren't currently used a lot that's what that's in my head so i'll tell you one application which i think is underused in crypto um what's the time where you are 208 why isn't it block 10642 why is it 208 because humans have this thing humans have this thing of, you know, we need 24-hour days because we need routine and the sun rises and the sunsets.

1:04:46Agents don't give a shit. For agents, it's one long number of blocks.

1:04:51Raoul Pal:Yeah, Melton was talking about this actually using the blockchain clock. Well, yeah. So I think the blockchain clock has got to be the most accurate clock. It doesn't make sense that we talk about dates and times, EST, non-EST. Ultimately, we need to talk about about blocks and i think the one underused mechanism around blockchains which i don't think anyone's cracked yet and maybe we're very very very early in this is like if if i give you a loan um when when did i if i give you a loan and i give you a loan on one chain and it needs to be settled somewhere else by another chain through an ai agent and whatever else when did i give you the loan and we're specifically talking micro payments and micro micro micro interest when did i give you the loan so i think the one aspect of blockchains that isn't yet respected enough is the time aspect of it yeah so so so like it it can't be a minute and a second and an est and a pst in a daily in a dealing room when i was in a dealing room i'd write a ticket because i'm a trade and i'd have to timestamp it yes now we could change that by you timestamp in advance you do other stuff right but blockchain does something you're right with time and price or time and information or time and whatever that is it's yeah you're it's completely underutilized still so so i think i think as a log of events put it that way yeah i think we i'm extrapolating a couple of years forward but right now the fastest blockchains have uh i don't know 100 millisecond times to finality, right?

1:06:30And so the blocks are being mined probably at that speed or maybe slightly faster. I think that there's going to be a lot of protocols that are going to be measured in block time. And I think we're going to come to a universal block time at some point, which is, for example, you think about Bitcoin having to transact with ETH and Solana and Sui, because that's just the way that the transaction was settled, right? When do they settle? And so ultimately, I think we're going to come up with a blockchain that's just going to be the time. I think there's one or two blockchains that we're looking at this, but I think one or two blockchains, we're going to pay them a minuscule fee just to use them as a settlement layer for multiple.

1:07:11So the oracle for time. For time, the oracle for time. I think that use case is probably underappreciated, specifically with thousands and thousands of thousands of chains. where, you know, if microtransactions really matter and you've got to settle them all at a specific time in microseconds, I think the time element of blockchain is not well enough respected. Super interesting.

1:07:35Raoul Pal:Right, final question about all of this because at the beginning you said there was three component parts of what you're doing and how you're doing it. One was your thesis, we've gone through that. Two was how are you valuing this stuff? Where are you looking for opportunity? And three is how are you using AI? So how are you using AI and how's the team using AI? Is that for building trading models or is it for analysis or is it a combination or is it for scraping for information? Because people are fascinated by this stuff. So the first thing is we automated everything. So I realized I was running a very, very, we running a very, very fat staff business in an age where most things could have been automated.

1:08:16And so what I did was I took the biggest innovator of my team. She's amazing. She's the biggest innovator in the team. And I said, if you want to innovate, then the first thing you're going to do is you're going to automate the entire business so that you can free up all the cash that you need to innovate. And so she literally went through our business process by process. And she said, OK, what's your role? Your role is to upload the videos to YouTube and you keyword search them. Great. She built an AI agent that does the keyword search. the clips so you know the clips that i spoke to you about where i told you that we clip all our content that's all it's an ai we used to hire 10 clippers or five clippers we now have one and he runs 40 twitter accounts and he has one ai tool he'll run this interview what tool is that we built it we built it um it learns how to listen for clips and it clips all the content and it can actually upload them directly to twitter so he runs 10 twitter accounts one guy runs 10 twitter accounts um so she automated the entire business literally uh the thumbnail how we make our thumbnails now we you can the it's the system auto generates based on on everything else or you can just draw it on a piece of paper with a pen just draw a little thumbnail and just show it to the ai and it'll generate that thumbnail for you in 20 seconds so that took about 20 staff and and made them two two or three so there's two and there's an overflow guy.

1:09:43That was the first thing that we did. So we literally took the entire process of our business. We looked at the things we were doing repeatedly and we built AI agents and that's now running. We have a thing called the newsroom. Now, what is the newsroom? We have 50 people that are interested in crypto, including all my researchers. And every tweet, every article that we read, for years, we've actually been copying and pasting into the newsroom. So if I'll like a tweet from Raul and I'll put it in the newsroom and you'll like a tweet from the Fed and you'll put it in the newsroom and Bob will like a tweet from NVIDIA and you'll put it in the newsroom.

1:10:22And for years, we've been doing that. Like since we started Banter, we've been all depositing. So we took an AI agent and we basically said, this is where we scraped all the data in the past. Very simplistically, we kind of said, look, the more recent ones get a higher score because they mean that they're more relevant to now. The old ones get a slightly lower score. But basically what we told the AI agent is we said, tally up how many tweets from each place we recorded. Based on that, that's how likely we are to want to see the tweets again. And now that whole process is completely automated, right?

1:10:57So there's a bunch of agents that basically populate us with ideas, trading ideas, alpha, et cetera, et cetera. And that basically is is completely automated. So we basically automated all those processes. I'll give you like stupid things. Like we do interviews like this. So when I interview a guest, sometimes I don't know them as well as I know you. So, you know, your content I watch and I know you and I follow you, but sometimes you get a guest and you don't really know much about him. That process of getting a document takes me about 30 seconds. I've got an AI agent. It gives me a background.

1:11:33Who's this person where they come from in fact it's so funny because you know it's a we run the process for everyone and uh i just want to see if i've got the document if i've got yesterday's document for because you were on my show yesterday and they sent me you need to hide the bit where he says he's a bit of a dick that's the hold on yeah he's so arrogant and i've got to actually see i've actually got to see if i've got it uh effectively i got i get like a two minute in two minutes i I basically know everything I need to know about you. I get your latest tweets, your latest views, latest things that you've said on latest YouTube interviews.

1:12:09And so I know that within four minutes of reading a document. Right? So that's basically that. And then when it comes to valuing tokens, so we've programmed the pipe. In other words, look for tokens, look for tokens that have high buyback pressure, high buyback pressure, low circulating supply, increasing users i mean that's really basic stuff because everything's available on blockchain and you can actually start finding it so to be honest the thing that i'm working on right now with with a couple of the guys is to try and find to build the perfect formula for pressure so the ai agent can tell us the ai agent as i said one of the the ai agents actually did catch the arbitrum trade very early before people really caught on how big arbitrum is going to be when robinhood launched it basically said hold on you guys are missing like there's this part we we we found it at a multiple of five and it was like weird and then we like double checked the thesis three or four times and was like hold on a second this is like a real thesis like and then we we traded on it and then that trade yielded us like a five extra turn because we're just so early to trade because people didn't pick it up so again without ai yeah maybe i would have caught it but I probably would have caught it after the run.

1:13:25So, yeah, things like that. I mean, other things which are like...

1:13:32Raoul Pal:You haven't tried the trading bot yet where it just takes all of you guys out of the process because that's the obvious thing. We've tried it. We've tried it. We've tried it. To be honest, we haven't really had success, to be honest. You know what is really good? What is really good in blockchain is tracking wallets. So, for example, tracking team wallets and the minute that the team wallet, having agents that track team wallets, the minute a team wallet starts to move tokens, that usually has a correlation with price. Or so you look for specific token wallet movements. Like I'm not going to mention the name of the protocol, but I'm actually watching on my screen here that a team is moving tokens that they probably shouldn't be moving, right?

1:14:15And we've put an AI agent watching every transaction, watching, because sometimes what they do is they bounce it from one wallet to a second wallet and then from a second wallet to a third wallet to try and trick people. But we've built a bunch of, a swarm of agents that basically watch this kind of stuff. So I'm actually watching one now. And I mean, the price has gone down 10 % since we started the show because the team's actually moving tokens around that they shouldn't be moving tokens around. So I haven't actually dug into it, but I can see the price going down and I can see the tokens actually moving.

1:14:40So, and again, I wouldn't have known that if there wasn't a GrokBot running here and telling me exactly what's going on.

1:14:48Raoul Pal:Fascinating. so Ram where should people find you I mean most people know who you are but you know I'm on Twitter at Crypto Man Ram I do a show on Banter Crypto Banter and I've got another channel called Crypto Insider which is where I do long form interviews I think probably Twitter is the best place because that's the hub the hub for everything it's the easiest place to find you otherwise you can find me in Cape Town you can come and visit me in Cape Town yeah or I'll see you in Singapore and I'll see you in Johannesburg and I'll see you in Cape Town I'll see you somewhere alright my friend Good to see you.

1:15:19Raoul Pal:And I'll see you soon. Chat soon, man. This episode is brought to you by Pith. Pith Network is the fastest growing financial data company today. Trusted by Fidelity Investments, the US Department of Commerce, Revolut, Calci, Jane Street, Coinbase, and many more. They provide real-time financial market data across over 3 ,000 equities, commodities, crypto, FX, rates, and again, a lot more. They're the first and largest provider of 24-7 financial indices and offer the widest array of market data for the lowest cost. So go to pith.network to take advantage of their free trial. Token 2049 Singapore, the world's largest crypto event, returns to Marina Bay Sands on the 7th of 8th of October.

1:16:03Raoul Pal:On stage will be myself, Jeff Yan from Hyperliquid, Shane Copland from Polymarket, and the Real Vision community gets 10 % off tickets. Claim yours using the link below. See you there.