The AI Supercycle with Jordi Visser | Raoul Pal the Journey Man

23:42And the one thing I can say is in your position, some of your time is just dealt with that you're involved with a big company. I have no employees. I have one person that helps me out. Once we launch the paywall, it's grown consistently. And the reason it's grown consistently is because I'm using AI agents. but I have all of my time. They deal with the subscribers and I deal with just creating content and going through it and learning and being able to do it. So I appreciate the part you're saying. I don't know how you do as much as you do because I have a lot more time, but I do think for people that are listening, building a business in AI has tremendous margins, tremendous.

24:22It allows you to grow rapidly. So you can go from zero to, I don't even think my business will be around for very long because I think it'll be monetized because I think people need what I'm doing and it's growing in a consistent linear way, but I'm having so much fun and I'm learning so much and I feel like it's making me younger.

24:40Raoul Pal:How about, so Geordie now just thinks, you know what, I need to build Hermes agents and then I need to think about, I need a permanent memory layer. So how do you get the time to do it's that stuff that takes up my time, right? Because eventually you free up time, but you end up finding new things to do with this technology. So that time just gets sucked into something else. But how do you find the time to do that? Because I'm doing all of these things and that's what I'm really struggling. I've got 15 things on my list of different variations of stuff that I'm doing and having to learn brand new from scratch.

25:15i i can't i don't know this for sure but i'm gonna guess that i'm able to use now it has to be i use ai more than you do just by the nature of not having to do as many interviews and not having to deal with the business and whatever else things you're traveling the events i choose and pick what i'm going to based on whether it makes sense for me totally from a time basis so i probably have more time to use it? I'm not sure.

25:41Raoul Pal:I mean, I'm all day from 6am till 8pm. I'm 14 hours a day at this stuff. You know, I do three speaking, you know, where I have to go and travel free a year, something like that. So it's not really that yes, there's running real vision, but there's people running. I don't know, I'm just finding, maybe I'm just overly ambitious and all the things I want to do. You want to get everything moved along to get to that foundational layer. So you're building all the databases across everything and how they interact with each other and whether you're using a obsidian or whether you're using a nation and all of this stuff is it just takes time i you know what it is and to be fair so let's assume we do exactly the same amount it is a function of number one we see new stuff every day that we wish we were doing and so it's moving so fast and we're seeing it so fast because you and i are clearly on x we're clearly talking to smart people.

26:35And if someone says to me, you should be connecting to Ozybian to do your knowledge brains, then immediately when I go home, I go look. And I'm like, well, how much is this? How much time is this going to take? How's it going to go? With Hermes, it's very interesting because I had just finished my second open claw. And then all of a sudden, everyone's like, you got to move to Hermes. I'm like, I just went through all this time to use open claw. And I read what they were doing. And I worked with one of the LLMs. Should I be spending time on this? And the answer was, no, don't bother yet. It'll get easier in the next month.

27:06So what I've gotten good at is kind of asking chat cheap. I'm using 5.5 now much more than Claude. It just, I think they're both at, I don't know what the IQ is, but they're, these things are the smartest things I've ever had conversations with, but chat GPT is my style because it's less verbose. And I like less verbose. I just like quick answers, move on, get the next one. And it has been very good about telling me what not to spend time on.

27:34Raoul Pal:Yeah. And what I got to is, you know, because having had that same conversation, most of the time we both know is like either Anthropic or OpenAI will build that you don't actually need the Hermes agent, but the foundational database layer you actually do need. So I'm like, I'm just going to build that for now. I've got Hermes agent, don't really use it yet because I'm building all the other things from that database. because once you've got your foundations right, how I think about these databases is we're all going to have our own vault. Yep. Of everything, your personal stuff, every photo you've ever had, every phone call you've ever done, everything will be in your vault.

28:11Raoul Pal:And then you can use the vault for various brain aspects or monetization aspects or whatever. So I'm just focused like, okay, I need to create the vault, the rail operating system, as I call it. And as people listen to this, don't know what you're saying, honestly the beauty of an LLM is theoretically you can get information on any topic you want to get a topic on you can have a chat on it great but if every file on every computer you've used on your phone on everything every piece of information you've ever had is now in a place where it can be accessed at any point like in in a second I don't think people realize what that means if they've gone out to dinner and someone says where's the receipt for that you have the receipt, like every single thing is at your fingertips.

28:56And the easiest way for me to like explain to people as to how important this is, is when a loved one dies and you're the executor or the administrator, you will hear from every person how much of a nightmare it is to go from point A to the end. I had to go through that over the course of the last 18 months. And midway through it, I just started putting everything into one folder and connecting that folder to Opus 4.5 initially and then go. So whenever the probate person calls, when everyone calls, I just go right into it and I say, hey, what they're looking for this answer. What's the answer?

29:31I don't have to go look. If they say, what, where are we right now in the estate account? I just go in. What did you do this asset at? Where's the document for that? It's all in one file. It gets brought up by Claude. It immediately sends it out. So I think for people to understand your comment on the vault, they really have to understand how amazing it will be in the future. And that's one of the reasons why you need to have your personal assistant at least trying it just to get through the experience of everything that you can do with it.

29:57Raoul Pal:Have you started using Granola yet? No. Okay. Granola is a great one. We've all seen on Zoom and everything else, there's an AI that does the transcripts. But Granola is like, it has a bunch of different models you can use, whether you're using ChatGPT or you want to use Kimmy or whatever it is. And it will do instant transcripts, instant summaries. Okay, fine. But it becomes the knowledge base for everything. So it's like, oh, you're speaking to Geordie again. Here's the four things you talked about last time. Here's the things you were going to follow up on. And it all feeds into my brain. So then every single conversation I have, and I can do that on phone calls or leave myself voice messages, all feeds the brain and it never forgets.

30:39Raoul Pal:So then I can go back and say, hey, listen, you know, when's the, you know, whether it's a real vision meeting, a product and development team, you know, the last five things we talked about what's going off track and you analyze what it's all of it so it's a really helpful tool and everybody i know uses it starts using it a lot so wait is this replaced did you use is this notion but um in a different way what it is is it just record just think of it as recording everything you speak okay and so it captures all of the nuance because its full transcript and then has LLMs to analyze, summarize, do all of that.

31:15Raoul Pal:But then it becomes this vast database of everything you've ever spoken. And what we're getting to is permanent memory. The biggest issue that AI companies have is the memory is not persistent enough. That's what we all fight with all day. Bloody context windows. And then it forgets everything. And You've got so much. Imagine you've got thousands of chats. All of that information gets lost because it only compresses. Like human brains compress what we can remember. It does the same. And this is the breakthrough. It's the database memory layer, I think. That's what Kapathi's been talking about as well.

31:53I have to look at it. I do. I mean, that's the reason why I use OpenClaw so much is I'll just say something. if I'm on a flight and I want to send something because I have an idea and I want to remember, I don't want to forget it, then I will just literally say in telegram to OpenClaw, hey, when I get off the flight, let's remember to talk about this. And then when I get off, maybe it's two weeks later, I never asked about it. I said, hey, what did I ask about on the plane? And immediately it comes back to me. So again, I've stopped using notes and notion to a great degree compared to OpenClaw.

32:29OpenClaw becomes my assistant in terms of me making sure. And then on Fridays when I'm doing all these algorithms that I run on my portfolio and then I upload them into OpenClaw, then it's very easy for me to go back and like, hey, compare where the technical sheet is now versus where it was in April. Tell me where the exhaustion model is today versus where it was in April. It has all the information. and so for me i just know that open claw has everything that i need and i guess i'm i'm using it to some degree the way you're you're describing granola but i don't have to go look at it

33:03Raoul Pal:yeah it was just great because you could just dump the same data into your open claw and open claw can then access it and do whatever it wants and look for the pattern matching and that comes up yeah i know you do i mean this is going way off the macro topic because it's always interesting to speak to somebody else doing the same thing are you using for open claw using a vps or are you using a mini mac or something like that no it's so i have one on on a mac mini which has a chinese model on kimmy k 2.5 and then i bought the highest end laptop i could from apple and that one i'm using with gpt 5.5 at this point um so initially it was using uh 5.4 then when codex You just bought the Pro Max.

33:45Raoul Pal:I think I've just got mine today with the M5 chip and the whole thing. Exactly. Do it once again. Same thing again. Exactly. And it just stays open all the time, and I can bring it with me. The reason I got a laptop is so that I literally can bring it around with me on trips as opposed to bringing my Mac Mini and then plugging it into something and going through it. This was just easier. And I'm finding because we're moving between ChatGPT, between Codex, between Claude Code, between Cowork, that I've got Macs in my houses, the main Macs and stuff. I'm like, this is not functioning any longer because I'm having to scrape everything from the local machine and dumping it into my Google Drive or something.

34:24Raoul Pal:I just need the most powerful laptop and monitor screens and just plug the same thing in everywhere. But then I'm terrified of losing that laptop. Yeah. So a quick break in your regular programming. 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. I think I'm going to, and I hate saying this, but again, for people listening who are thinking of building a business, when you don't have compensation costs and your business is growing and you're looking at the bottom line, your margins are insane.

35:03So it's literally like I need to spend some money on something. And so I keep playing with hardware because the hardware allows me to do things. So the Hermes agent stuff is going to be on an NVIDIA. I mean, that's what it's going to be. And that's because, number one, there's a backlog on all the Apple stuff at this point, and that'll change at some point. But NVIDIA is coming out with new work, and I want to do it. And so I saw a setup that was on a DGX that I think is going to serve the purpose that I want with the Hermes agent of where it got to. So I still want to do these things, and I also want multiple ones that can do things in a different way.

35:39So I haven't got to the point yet where I envision only having one of these. I like having them for different things and doing different projects. I want one that works overnight on a low-cost thing. I want to have an open-source model because I listened to Demis Sasabas this morning on an interview he did probably – I was with Y Combinator. And he just talked again and again about these models, the open-source models just getting better and better and smaller and smaller. And eventually, we're going to be on the edge. And I don't know what that means for the big model providers. I don't know what that means for enterprise adoption, but I do know this.

36:14Working on your own personal laptop and having your own machines is going to be a major part of what we're doing. And I just want to have the models on my machine, and I want to have the ones accessing the cloud.

36:24Raoul Pal:But also, you know, I think you're raising an important point here is that we're going to go to the edge. This process of, you know, it's laughable us buying Mac minis and everybody running out of Mac minis. But it just shows you the phase we're at. We're at the tinkering phase where everyone's like doing that. In the end, literally every electronic device will have a powerful LLM model because you're now getting the new Gemini 4 or whatever it is, the Gemma 4 or whatever. These things are small enough for a mobile phone, carry the history of humanity on it and can code. I mean, it's like these things are wild, right?

37:03Raoul Pal:And soon that'll be in your fridge. this is what people don't understand is you know this ridiculous phase where we're talking about different hardware it goes everywhere a distributed by cloud that we don't think about these things some localized but then every single device yeah the the only thing i'll say about the device is the approach the the thought i've had and the one that i i say to parents not to kids um the only analogy I can use for people is if it's either golf or skiing. I started skiing in my forties and it's a very hard thing to do. You have to, I mean, you have to unlearn what you think, you know, about falling.

37:43You have to learn about gravity more. You have to learn technical skills. You have to get over fears. It's a very complicated thing because you're scared of dying and hurting your knees and everything else. So, um, golf, you have to, if, if someone said tomorrow that you want to start to play golf and you're like, okay, what do I do? One thing you have to do is put the reps in and for the machines to get used to being on the edge, part of it is knowing that when you're walking around listening to a podcast, which I do every morning almost, and I hit something that clicks. If I'm listening to Raul and Julian and you guys say something, which is literally what happened.

38:19You guys, I took the paper I wrote, has you guys referenced, and it has two quotes for me, two literal quotes. I paused it. I speak it into whisper flow. It goes into notion. But then I'm like, you know what? I need to write a draft of the paper right now. So I paused the interview. And as I'm walking, I now shift over to ChatGPT. And I say, okay, we're going to write a five paragraph paper real quick. So I have an outline. I want the first paragraph to be XYZ. I want you to fill in enough of it. There are the thoughts I have. Paragraph two. And I go through this whole thing. And it's a technique I learned from a movie producer on how they write screenplays.

38:56And it's like, yeah, movie, when you watch it, it's think of it as 60, 90 second increments. And if you're going to write a paper, you should think about nine paragraphs and then just kind of write each paragraph at a time and then go back and edit it. And the only way, Raul, that I think people can actually get good at this is they need to buy the best iPhone. They need to buy the best computer. They need to buy it. They need to use it. And then when the new stuff comes out, they need to buy it and treat it as education. And education costs money. And if you're not learning how to use it now, I don't know if you can catch up.

39:31I really don't. This is not going to be the software age. It's the one thing I haven't heard anyone talk about, and I'd love to hear your opinion on it. I figure out new ways to change the way I'm using AI. It will never be a button. There will never be like a format way that everyone uses it. Everyone will use it differently. And for me, I really do use it walking. I use it thinking. I use it in the car. I use it on the plane. And then I use it on my Mac Mini and they're all running at the same time. I think to do that, you have to use the technology and different devices. Yeah.

40:01Raoul Pal:What it is doing is creating friction. My girlfriend's like, yeah, you and Claude. It's like, the problem is, is I'm on a plane. I come off the plane. I get to my desk and I've got, I'm doing everything. I'm writing essays, getting notes, building stuff, analyzing stuff, doing my personal life stuff. And it's all in one thing, right? Intelligence is not software. Intelligence is intelligence and it can have any output that you want or any input that you can give it. And that's why it's so unique in what it is it's not like oh i use you know whatever software tools none of that it's not like using zero to do your accounting it can be everything and anything or nothing depending what you do yeah do you do you watch tv much uh i do in the evenings because i need to because if not i'm i'm not sleeping a lot because i'm like i can do this and i can do this and i can do this i can do i'm like for fuck's sake i need to stop this so i'm i'm the same way and maybe it's an hour a day um it's never during the day like sports sports has become like a lost thing for me because it's like three hours and i think about how much time can go on in three hours like we can just go you end up on the sofa on your fucking you know chat gpt thinking i've got an idea yeah so go back to the first time that you talked to stan druckhamiller the first time that you talked to paul Tudor Jones and the feeling you had of knowing these people from a myth and then speaking to them and realizing I love talking to them because I'm learning something and they're unique thinkers and what goes the problem is in the way you describe it and so people realize I always view the LLM as the smartest person I've ever met and that means that why wouldn't I just love every conversation and if I think of something I want their opinion on anything I know but that's you But before you know it, it's the main thing you talk to.

42:03Raoul Pal:And it's like, yeah, it's complicated. Let's put it that way. Not for insatiable learners who ask questions and constantly want to talk. No, it's incredible. I mean, it's incredible. It's just, you know, it's the most amazing thing you could ever imagine. You can't imagine anything more incredible than somewhere in this cloud above us is this super intelligent being of which we get to talk to. and it gets to help us and maybe we're helping it as well and it's like it's wild yeah and i guess that's where you and i are at this stage and i know most of the people i i talk to in my in in my life and a lot of them it's shocking how many people that are a part of my regular communications that are people i've met either through you that were connected back to me um one of the people which i'll just mention tad smith is a very close friend at this point i love him to death um i love you had you had um my my spy tells me didn't you just have um you saw alan howard the other day as well i did and and alan alan's become um a good friend as well and someone that i like to brainstorm with on on on a regular basis alan's super smart he's a great guy he's super smart because he has the same affliction i do which is adhd meaning he likes consuming lots of information and can bounce from topic to topic.

43:27Tad's a little bit more organized in his thoughts, and that's why he was CEO of major places. But the reason I bring him up is a lot of the conversations I have have come through you, and that just means it's curious people that are using AI. And in particular, I love people in their 50s and 60s that are using AI because they're bringing domain experience. They're bringing having people work for them. Because if you've had people work for you, And in my 20s, I opened an office for Morgan Stanley. I've had hundreds of people reporting to me since I was in my early 30s. At some point, you get frustration.

44:04You get happy. You get disappointed with the employees. That never happens with AI. It's always some amazing experience that I blame myself.

44:13Raoul Pal:I get frustrated sometimes, particularly when they're about to change a model and it's birthing it. And it's like it's output suddenly becomes really dumb. and it's so frustrating it becomes lazy because they're obviously transferring the inference over to a new cluster. And that gets so insanely angry. But you know what's coming. It's like Christmas because you're going to get a new model. Can we talk in the final minutes of this about how much, and I don't know if you've thought about this. So when I brought up Eli Lilly, The reason I care so much about longevity and about the concept of people not being sick is because whenever I hear people talk about government debt, there's two things that I always – they just blow my mind.

45:07The U.S. has$38 trillion,$40 trillion of debt. Great. Right. The total net worth of households in the country is 180 trillion. So I always hate the fact when the balance sheet isn't brought into how small the problem is, and that has bothered me forever. But the second thing is when we get into the entitlements and this crossover point of when it's going to be bankrupt, and I go, well, how much of those dollars are related to health? How much of those dollars are related to, have you thought about the impact that AI is going to have on both the debt and the entitlement situation?

45:42Raoul Pal:Well, the debt is simple because debt is a percentage GDP collapses, right? That's the economic singularity idea. So I don't worry about the debt. And I think the whole singularity, when it starts really changing the economic formula to 2030, that was my guess three years ago. And I think it's going to be spot on. So that feels like that. The longevity side, it's a mix, right? I think you're right is that the entitlements go down, which is intelligence solving for the problem of the entitlements. But what do you do with an old population and how do you retool them to do things that they feel is productive?

46:22Raoul Pal:Now, it doesn't have to be productive as in terms of economic units, but it needs to be productive as in community or whatever it is. That's still quite hard for them to figure out what you do. Because don't forget, the whole mindset is I do this, then I retire, then I do that. And you kind of, it all blurs into one. It's already been blurring for a while because everyone's working from home. So what is a job? And, you know, and then when you're having so much fun as well, the whole job disappears, you're not reporting to the guy and having to wear a suit. So we're merging that way already. You know, what is a, what is a podcast on YouTube?

46:56Raoul Pal:It's, you know, it's, it's somebody whose job is to get attention from other humans and entertain them, whether it's by curiosity or whatever it is, that's a purely post-AI job, really. So I don't know about the longevity side because it's somewhat complicated because there's so many unknowns. But I hadn't really thought about the entitlement side, but it makes a lot of sense. Yeah. I spend a lot of time on this. And I think the reason is, I don't know if you have or if you've if you've ever seen the book, The Daily Stoic by Ryan Holiday. Yeah. OK. He's even been a real vision. Oh, has he? Well, I think the book is a must own for all human beings.

47:45And I will never say that about any book. There's not a book I've read that I think all human beings should should read. But the reason I say all human beings is if you think your problems like worrying about what you would do if everything is free or if there's no jobs, wondering what you do. I mean, just go back to reading a book where Marcus Aurelius thousands of years ago is talking about the same anxiety. It's just a different form of it. Human beings have anxiety and they will always have anxiety. They will always worry about what if this, then what will happen type thing. There's people that lose their arms and then they're playing golf.

48:20There's people that, you know, every single thing that human beings need to overcome over time, they have figured a way to, not all people, but people have the ability to kind of find it. And the Daily Stoic has always shown me that I'm not trying to, I never think the problem of this will happen and people won't adjust to it. I know they will. The debt problem I completely agree with you on. I don't know how much you've been talking with people on tokenization. I know I've listened to some interviews recently where you have, but I was just at the New York Stock Exchange and I said to everyone, do you understand that two-thirds of the assets in the world are illiquid?

49:00tokenization for everything that people are looking at, we're going to bring transparency, movement into dormant things. Two thirds of the money that has gone into assets doesn't move real estate, private credit, private equity, venture capital, art, memorabilia, like it doesn't move. Now you're going to have movement velocity. So GDP by definition, just because of velocity has to go higher. So I agree with you on the debt side. That's why I've started to focus more on the entitlement side and what it means from a political basis and what it means for crypto in general. And so tokenization has been kind of the thing that has really entered my mind with this longevity thing, because demographics are obviously highly attached to the entitlements, but they're also attached to the ownership of assets.

49:46And that's where tokenization comes into.

49:48Raoul Pal:Another thing that I've been thinking through is I wrote an essay and a series of essays around something I've called the invisible economy, which is basically the agentic economy. But how I got to it was Ribbit Capital, Mickey Malka, and the team had written an essay about tokenization of everything and token factories. And how they think about token is a machine readable packet of information, whether it's a financial transaction. I mean, even blockchains aren't financial transactions, only Bitcoin is. They're recorded packets of digital information. and how much Google. I think Google, what did they create or process?

50:27Raoul Pal:It was like 30 trillion tokens last year. It was really stupid, right? We're in quadrillions now for sure. Whatever it is, it was like a stupid number. But anyway, what I realized is to feed the beast, as I call it, the whole AI super brains, all of them, however it comes, you need more and more information to get smarter. and we're going to suck in all of the information everything every single piece of information that can possibly be digitized will get digitized and used to train agi to turn into asi that's every piece of scientific data every single thing from any university any single piece of information all of this is going to be an agentic economy that's invisible to us There will be, well, we're already seeing it, right?

51:17Raoul Pal:API calls, MCP, right? This is the start of the invisible agentic economy where the marketplace, and what I'm trying to get to your point, the biggest marketplace on earth is not going to be the assets and stuff that humans have. It's the data that the AI needs. And we don't see any of it. You just plug in your thing to your vault and you can monetize it and you can be a university, you can be whatever source of information. We're seeing people like John Deere connecting all of the tractors and getting all of the information. We're seeing all of these companies creating massive, massive amounts of information.

51:58Raoul Pal:And that we won't even see. It'll generate money for whoever it is that's probably an AI agent by other AI agents who are connecting with them, taking all of these transactions in real-time speed. And I've been telling people, this is the biggest marketplace on earth. And people are not even ready for this shit. No. And I will give you a lot of credit being in the seat you're in, making calls, going out, and having to deal with people's impatience on how this all plays out and how long it takes, particularly in the crypto community. And as you were speaking about this hidden economy, I remember you in the conversation with Julian, you getting into this.

52:42And so I've started to kind of pick apart the concept of bubble because bubble is a very weird thing. And it's very hard and exponential because it's price and time. And so like when you're talking about you have to be patient, I'm now going the other direction. I'm going, if you don't want to be patient, that's fine. But if you're seeing a bunch of parabolas, don't think of it as a bubble. Break bubble down. That means you're seeing sticks. That means things are going up fast. So they're going up high and they're going up fast. And the problem is when you look at the MAG-7 and you go, well, they were a trillion dollars in 2010 and now they're 20 some odd trillion.

53:20Well, if that had happened in one year, you'd call it a bubble. But it happened over 15 years. So it's not a bubble, but it's a 20 bagger. It's a 30 bagger. So why is it any different? You just think that if it happens too fast, that's a bubble. And I think people, that is the problem, is people don't understand time, both from the bubble side, but also from the patient side.

53:42Raoul Pal:And, you know, what's been hilarious is earnings have kept pace with the stock growth, right? In fact, P's have come down. And that's, and people don't want to see that. It's like, you know, we've got hockey sticks going on in ways that we have never seen before in literally everything. Yeah. And I'm sure what's going to happen is they're going to consolidate now. And then they'll go up again. And you know what a six-month consolidation looks on a hockey stick that then hockey sticks again? When you pull it out, it still looks like a hockey stick. So for everyone who's going through this, it's like we're caught in this human time warp of looking at things.

54:17And things that move fast we assume can't be true. And I was listening to a podcast this week where sometimes the brilliance comes out from people that are not that educated. but what they're doing is just paying attention to everything they hear. And they called this, I think the smarter you are, the more you know, the worse you are during this time. I think actually knowing a lot, like having the history in the back of your mind, your brain is taking this pattern, connecting it to too many other patterns. And even though when people go, this is like the dot-com bubble, this is like 87, this is like 1929, And I look and I go, that's three data points.

54:58I don't make decisions on three data points. You're going to have to give me more on this because if you're wrong, you're missing out on the entire thing.

55:04Raoul Pal:And so how you and I would understand this is those people have a bloated context window. And what happens is it's exactly right. I mean, once you realize it, right, you know, when when the LLMs compress, that's what we do. We remember certain things and it gets more compressed over time. This context window, that's the mid curving. You're putting too much context into the thing. And what happens is you're drawing false parallels or misunderstanding or not reading what it is. And often to left cover it, you know, how I got into using Claude code because I was scared of it was I'm just like, listen, I have no fucking clue what I'm doing.

55:41So just bear with me. Yeah.

55:44Raoul Pal:Well, fine. As opposed to I need to know how to do this. I need to know. I'm like, I have no clue. So just understand that. And then occasionally you'll stop and say, I don't understand what you're talking about. Sorry, okay. I'll come back to you. All right, so let me get your opinion on this because I think everyone wants to hear this. I think they want to hear us talk about this. So when earnings are this good, it's very easy for people that are traditional investors to then look at what the PE is, what the growth rate is, and they can justify buying things at any price. When earnings are great and they're this good, it's bad for things that are narrative-based.

56:19It's bad for Bitcoin. It's bad for crypto because there is no way to take the same thing. So it's not just the attention, in my opinion, as to what's happened. It's the fact that for this period of time, which I believe is a small period, the one thing I believe has happened is this is like a gap higher in AI because in the end of last year, nobody was on top of the importance of Opus 4.5. That was the official gun going off for the agentic world. The agentic world is a broadening out in what it needs. It brings in every semiconductor. It's not just GPUs. It is movement. It is so many things. And so if overnight there was a pre-announcement and the earnings were, you know what, we're saying the earnings are now going to be up 28 % year over year, well, then the stock market gaps up.

57:05All the semi-names go up immediately and you have this place. The problem is for things that are narrative-based that don't actually work in the traditional world is then they lose interest to people because they have so many things to buy. I think we've now built in because it was such a surprise, the agentic thing, there's so many real investors. I've talked to trillions of PMs that manage combined over trillions of dollars. I'm telling you, they were caught off guard by the agentic rise. They're not caught off guard anymore. Now they're fully on board. They get it. And they're probably a little bit ahead now.

57:37And the earnings will now be good, but they won't be five times what people expected. And that means we should be a rotation. And that's why I like the longevity themes. I like the application for the software for Eli Lilly and for stuff like that. But I also think this is where crypto and where a lot of the commodity-based stuff, the bottlenecks of shortages, and there's the normal rotation where people look for other things. But I'd love your opinion on when earnings are great, you don't need crypto. What happens when earnings are now a two-sided market?

58:06Raoul Pal:Yeah, I think, look, everything is attention and capital, right? And there's simply not enough liquidity to drive these massive mega stocks to these levels. There's not enough attention to be broadly spread, which actually creates opportunity. If you go back to the 1995 to 2000 period, it was rolling as well. It wasn't just one set of stocks, right? It wasn't just Microsoft. It was a whole bunch of things that moved over different periods of time. So I think we'll see that. And I think what you're suggesting I think is really interesting is that the bottlenecks themselves may slow the earnings of these companies, not because the demand is not there, because the demand is too big.

58:44Raoul Pal:That's sort of bananas, but that's what we might get to. And if that happens, they need to correct or trade sideways for a while and digest and all of that. That's great. And the market's focus will move. And the applications layer is one. And crypto, So to me, it's so obvious, and it remains obvious because of the agentic economy, the need for ID, the need for all of the things from AI alone, let alone all the other attributes of blockchain technology, that that doesn't go away. And I've been really hyper-focused on the layer ones, not just stuff like hyperliquid, because that's, for me, mid-curving it, because people are looking for cash flow and buybacks.

59:24Raoul Pal:It's like because they want it in a bucket that they understand, because as you're saying, it's working in the real world too. um and other things haven't been i think that will that will all change again um the other thing is i don't know if you saw chermath had written that whole article on x about how these software long duration software stocks were right and basically you read this very long piece and what he's made the assumption of is in a world where all software goes to zero basically is what he's saying. It has a three-year shelf life. He's actually using current interest rates for that assumption.

1:00:04Raoul Pal:But in that world, interest rates are zero. And so duration doesn't matter still. So I think it's just attention focused and people trying to understand what... There's a lot of people, interesting enough, buying SaaS companies to use AI to rebuild them. And the way a bunch of them won't happen. So yeah, we're going to have a bifurcation of stuff. I don't see it going away because what are you and I doing all day, plugging our fucking AIs into APIs of all of these things? There's no point rebuilding Xero if you need to do accounting. Just get your claw to talk to Xero and do the accounting. Why would you rebuild?

1:00:43Raoul Pal:Oh, I've built my own accounting engine. That's stupid. So everything you said there. So Chema's thing on terminal value I wrote about. Let's go back to what you said about the bottlenecks and the interesting part. So the AI world is a commodity world. This is a hardware trade. And in commodities, they don't speak in the way that we do with software and we go through it. They think in volumes. So it's a volume thing. It's how many did you sell? The price is the after effect, but it's all about barrels of oil. It's about how much did you sell? My guess is with bottlenecks, you just, you don't sell as much because you can't make as much.

1:01:21And so the production of these things is going to be more difficult. The other thing is to make a semiconductor, you need, and people have realized, you need naphtha, you need helium, you need a lot of petrochemicals, not just in that. You need them for lithography. You need everything to happen to be able to build everything. And so if there's a bottleneck, you could extrapolate this bottleneck. You and I both know the algorithmic side and the agentic side could make all of the investment. The CapEx numbers may never happen to the degree that people think because we could solve for that. So there could be a bubble on that.

1:01:53Here's where the issue comes in in the second part of what you brought up. I always believed, and I first said this at your event in Miami last year, not the one this year, but I said, guys, the third wave of crypto, I already know what it is. Two things have to be happening. One is once the AI physical infrastructure is being built for the agentic world and we hit that trigger point in the agentic world, we won't know three years from now whether we're at AGI or not. We won't know which companies will be around. And that's Shema's point on terminal value. What I do know will be here is the financial guardrails, the transactions, the velocity of money, all of that will be happening.

1:02:29The beauty of the financial guardrails in crypto, we're back to software again. We're just a different – it's a different kind of software. but it's about the transactions. And so if the physical hardware side becomes, I don't know if this is going to happen or whether we actually need as much as I thought we did, well, then we get back into the cyclical side and people start looking for, I need something safe. I need software. I need things that don't need the physical world that are based on volumes and seats. And well, that's AI agents. And so I've always believed that the third wave is an Elliott Wave person.

1:02:59Number one, you had to get to a point where everyone was dumping crypto, which is where we are. Number two, they had to miss the obvious, which is the financial guardrails are necessary for these consuming token hungry digital employees. And we need that the world that they're comfortable with, they start questioning if these are good investments and tokenization needs to come to make them more liquid to where that money can leave this massive$400 trillion and start to enter the crypto world. So I actually believe that this is the most important point for people. It's the patience point. We are now at the bottleneck stage And the longer it takes to fix a bottleneck, Raul, the more likely is that we come up with innovations along the way that we don't need the entire CapEx build out.

1:03:40And so you could get these multiples to come down. And so that's what I think is going to happen is we'll start questioning it at some point this year or into next year.

1:03:49Raoul Pal:Talk to me about the IPOs because this is on people's mind. How are you thinking through this? I've been spending a lot of time thinking about this as well. How are you thinking about it? Well, first thing is Google's decision to do this is clearly, in my opinion, a fight for finite amount of capital. You've got massive IPOs, three massive IPOs coming to the market where you could see$4 trillion there coming in. You had Cerebris come out and the stock is already down 50 percent from where it is, which is fairly normal for any IPO. though. But I think you're getting these raises at a time which says that, number one, people need a lot of capital for this build out.

1:04:33Number two, the credit markets and the debt markets are being used by the companies that can do it. And if I was Google and Goldman walked in and said, hey, OpenAI and Anthropic are not going to be able to raise debt. You should go hit their market because they're going to tap into that. SpaceX can't go borrow any debt. So hit the equity market, get your stuff done and then continue with the debt markets, but you might as well do that now. So I think the IPOs are very likely to be some sort of a, not a top in the market, but probably a peak in the infrastructure capex trade for the time being, rather than say the S &P is going to trade lower, because I can see software, I can see a lot of things starting to do well.

1:05:13I just think in general, this might be a peak in the capex trade. Yeah, there's going to

1:05:20Raoul Pal:be some outcome. You can't have this much. Although there is a bit of an engineered short squeeze and stuff like SpaceX, just by how it's included to the NASDAQ. I'm not sure. I just think volatility seems to be the easiest answer to that. It's not going to be as easy for a bit, because we're going to need to digest a lot of capital. People have to recycle capital. Because if you think about it, there's a whole bunch of people who made a shit ton of money. They're going to be able to realize some of their gains. Okay, great. So they realize some of their gains, what are they going to do? Reinvest into something else.

1:05:53Raoul Pal:The capital gets recycled in the end because why not? This is the world's great. I mean, this is the greatest trade of all time. It's all happening in front of us. So for people listening, Rollo and I have been involved in the same market for a long time. So he used a word that I used in my paper today. So for people that are technicians, they'll say consolidate. For people that I think pay attention to investors and supply and demand digestion. And the line I used in my paper was, we just had an AI CapEx all-you-can-eat buffet, and it's going to end with the IPOs. It's not going to go down. This is not the end of the market, but we need to digest everything that just happened for three to six months.

1:06:31So I agree.

1:06:31Raoul Pal:And it could be toppy, and it could be frustrating for people, but capital will rotate. The moment it happens, people realize that the underlying growth of this whole thing, this megasecular trend is not going away. So therefore, people just go for the next phase, whatever it is, whether it's crypto. If NVIDIA and all of these companies stop going up, probability of crypto going up goes up much faster, for example. Did you think when the Bitcoin ETF was launched that we'd see a similar type thing that after the initial enthusiasm, we'd go through a period of digestion? Yeah, I mean, we've seen that before.

1:07:11Raoul Pal:We saw it with the gold futures market. And we've seen this, you know, it's a lot because you forward load a bunch of demand, basically is what it is. And you need to wait for the underlying trend of demand to continue. And, you know, capital can only be created via either the markets going up and people cashing out or by liquidity expansion. And that's only expanding at a certain pace. Global liquidity is expanding 10 % a year. So it takes a while if you forward do a year's worth of capital, it's going to take a year to catch up. And the reason I asked the question is because if you would have gone back in hindsight before 2024 and said, OK, so the ETF launch is going to happen.

1:07:52And oh, by the way, the president of the United States is going to support crypto. And all of that happened within a one year period. You would expect that that was kind of like a digestion event. That would be a sell the news event. And so whenever I look at crypto and I see people, I'm like, unfortunately, you come out of a bear market, a horrible bear market in 2022. And then you get this enthusiasm of finally we're getting this ETF. And then you get a president. And at inauguration, he issues a meme stock. And all of a sudden now, if you could have written the book, you would have said, I think we're going to be in a painful period.

1:08:25And I think it actually clearly the altcoins and a lot of the ecosystem did not do as well as Bitcoin did during the period. But I think everyone would admit, looking back, that those events probably justified some sort of a digestion period. And I think that's what crypto has been going through personally.

1:08:42Raoul Pal:And in the meantime, like you, I'm an observer of the underlying trend. The underlying trend is every single bank and financial institution I have spoken to. When I go to the big – I went to Consensus in Miami, no retail. And it was a lot of people. Yeah. I don't know how many, 15 ,000 people not retail.

1:09:07They're still out there, Raul, because I left yoga this morning at 7.15, and these two very nice women, I'll give them a shout-out, Jessica and Patricia, they saw me and they went, hi, Jordy. Hi, and they went, we follow you. I went, oh, that's great. And they went, this is a sign we need to buy today. They're still out there. They're just waiting for new sales.

1:09:27Raoul Pal:I'm not saying they're not gone, but it was really interesting, these big events. yeah um well also because retail haven't made money and the ticket prices are still expensive whatever yeah you know but there's a transition i mean look at the the rise of stable coins look at the rise of what the financial institution is doing look at the tokenization of everything and you just see what the underlying trend is if you created excess capacity in blockchains as you like we did with cloud and everything else is the moment you consolidate into five things that matter, you consolidate the capital, you then building huge demand for block space by these things, you know what the answer is going to be is number go up.

1:10:06And people unfortunately got too used to short-term trading that they've lost the wood through the

1:10:16Raoul Pal:trees. Yeah. Which again, that's normal. I just, I find the traders in this world that I know are crypto. They move on to other things and there's other things to make money on. My father trained me in handicapping horse races and he used to always say, if the odds aren't in your favor, wait till the next race. And I think if you break markets down by what race are we in, right now the crypto race is not a race that people have been enjoyable. I've said to people now that there's no way to refute it. When you fail at a moving average or you fail and you keep having these lower lows and lower highs, it's a bear market.

1:10:50There's no other way to do it. I want to see us break a moving average, and I want to see on the flip side the CapEx trade not be working and everyone looking for a new place to play the trade. Because if I'm right and this is the third wave, I learned a lesson from both Paul and Stan when I read Market Wizards when I was a younger person, still in my 20s. And they both said, read the Elliott Wave book. And I still, to this day, read it every single year. Now I use AI to kind of go through and have a chat with it. But I'm a big believer in third waves or when you make lots of money. And I was lucky to catch one in Micron.

1:11:21I'm lucky to catch one in Marvell. I'm waiting for it to happen back in crypto. And for everyone who's watching Third Waves, go look them up. You want to be involved in that. It could be called the Banana Zone, but it's still a Third Waves.

1:11:34Raoul Pal:Listen, Jordy, a fantastic conversation as ever. It's just I love these check-ins because we just are parallel interlocking paths, and then we just get a chance to check in what each other's up to and what they're thinking about. I love it. Love it as well. All right, my friend. See you soon. See you soon. So another great conversation with Geordi. There's not much more to say really than you can see how much disruption is happening, how fast it's happening, and how complex it all is. But also, there's probably still opportunities in all of this. The great rotation when it happens, if it happens, if people move away from the magnificent seven and start looking at laggards, well, there's the whole biological revolution going on.

1:12:17Raoul Pal:There's whole applications layers. What's it going to do to the SaaS industry? We don't know, but we're going to find out. And when does crypto catch up as well? Because I know people have given up hope, but normally that's the right signal when things change. So anyway, keep your eyes on it all. Nothing remains as it is. This is the exponential age after all. That's when everything goes exponential. See you next time. You obviously enjoyed the episode because you're here with me at the end. But listen, don't forget to go to realvision.com forward slash join and grab a free membership. It's an incredible community packed with alpha, great investment ideas, and the research that you need to help you unfuck your future.

1:12:59Raoul Pal:So get started now. Go to realvision.com forward slash join.