Loading summary
Transcript148 lines
- [00:00]
Raoul Pal
Today's episode is brought to you by Abra. Abra aims to provide individuals and institutions with a secure way to control, manage and grow digital asset wealth. From a separately managed account, Abra helps his clients get exposure to crypto and crypto financial products like yield and lending through one full service platform. If you're looking to gain access to additional liquidity, Abra has one of the most competitive loan products in the market. You can borrow against Bitcoin, ETH and Solana at up to 50% loan to value. Rates are in the 4 to 6% APY and and are open term. You can continuously draw down against your collateral as the price appreciates. Abra has other strategies to add yields and their team is happy to help align your portfolio to your risk profile. Reach out today and get a complimentary consult in your portfolio. It's worth seeing if they can help you manage your allocation, reach investment goals, manage risk and add additional yield. Go to realvision.com abra and tell them I sent you. Hey everyone. As you know on this podcast, I bring the best guests in the world at that nexus of understanding of macro, crypto and the exponential age of technology. If you're enjoying the show, a quick five star rating goes a long way. It helps us grow and keep these conversations coming with the best guests in the world. Thanks a lot. Hi, I'm Rahel Pal, and welcome to my show, the Journeyman. The Journeyman, as you know, is where we travel to that nexus of understanding between macro, crypto and the exponential age of technology. It's where everything is converging at an incredible rate. It's truly exponential and I think we're all feeling it and seeing it now. The most exponential of all is AI. AI has been the fastest adoption of any technology the world's ever seen. It's not even anymore a regular kind of Metcalfe's Law adoption. It's Reid's Law, which is Metcalfe's Law squared. We've never seen anything like the change of technology we're living through. And it's only going to get worse. Now. I've had one person help me, guide me on this whole journey to understand why what lies ahead, not just what's happening now. And that person is an old friend of mine, another hedge fund manager from the past who went across to the world of AI, and that's Ahmad Mostak. And Emad's a longtime favorite here for me on the show and on Real Vision and we're going to sit down with him and figure out what the hell is going on and where it's all going? What are these agents all about and why it matters? Don't forget today's episode is brought to you by Figure. If you believe in Bitcoin long term, the worst move you can make is selling it just just to access liquidity. That's why you should check out Figure. If you're stacking sats but also want yield, democratized prime lets you earn up to 8.5% APY paid hourly, backed by real world assets, not yield gains or total inflation. Figure also offers crypto backed loans at 8.91% interest with 50% LTV. So you can unlock capital without creating a taxable event or giving up your Bitcoin exposure. Figure's the largest non bank mortgage lender in the US with over 19 billion unlocked on their lending platform. And now they're letting Bitcoin holders borrow against their bitcoin instead of selling it. Security matters here. Figure uses decentralized NPC custody, meaning your Bitcoin stays in a segregated wallet. Not rehypothecated, not pooled, not sitting on an exchange balance sheet. They've also rolled out liquidation protection to help borrowers during sharp market drawdowns. Hold your bitcoin, unlock liquidity or put capital to work. Check out Figure using my link below. On to emad. Join me Raoul pal, as I go on a journey of discovery through the macro, crypto and exponential age landscapes. In the journeyman, I talk to the smartest people in the world so we can all become smarter together. Ahmed, welcome back my friend.
- [03:51]
Ahmad Mostak
Always a pleasure to be on Raoul.
- [03:53]
Raoul Pal
Yeah, it's always good. We always have a good conversation. I have no idea what we're going to talk about, but we're going to talk about a lot. That's that much I do know.
- [04:00]
Ahmad Mostak
Yeah, there's a lot going on.
- [04:02]
Raoul Pal
Yeah. What's been happening in EMAD world? What are you looking at?
- [04:06]
Ahmad Mostak
Well, everything. It's the great takeoff this year. Isn't really is really is.
- [04:12]
Raoul Pal
I mean that whole mult book, Claude, that whole thing has been astonishing. I mean everybody I know is now running like 10 agents running the whole businesses.
- [04:27]
Ahmad Mostak
Bizarre. Yeah. And you got 10 agents when you're like, why not 10,000? And then how fast do they run? I was just trying this thing, this new company called Talas who etched transformers onto the silicon itself and they're doing 15,000 tokens a second. So I just tried their chatbot Ask Jibby and it's like you type anything, instant response, you think about that you're like, there's this wave of intelligence just about to hit us, it'll cost basically nothing. And what are we going to do positive and negative around that?
- [05:01]
Raoul Pal
Yeah, and just the whole structure of an economy when you're adding these millions and billions of agents into an economy over this year, I mean, obviously ChatGPT have hired OpenAI, have just hired the guy who started building the agent stuff. So obviously agent is going to go mainstream in every single way possible this year. And we just increase intelligence dramatically
- [05:24]
Ahmad Mostak
at
- [05:25]
Raoul Pal
every level for zero cost.
- [05:27]
Ahmad Mostak
Well, I think Raoul, you've always had this monetary velocity, monetary supply based approach to things. It's an intelligence velocity, intelligence supply. What OpenClaw and Claudebot kind of did is they removed a barrier and idiot tax to intelligence flow. So you're on WhatsApp and you're telling it, hey, go and build a website for me or go and do this type of research or make it so that I never forget to send my wife flowers on her anniversary. And here's a card, you know. And so as the different levels of friction decrease, the level of intelligence you can access goes up. Just like as you get assistance and teams and things like that, except for they cost pennies. And this is important because most of our economy and profits are based on what Elon Musk calls the idiot tax. You know, like when he tries out SpaceX, all the components of the supply chain, you get profits at every stage because you're simplifying things. When it comes to digital points being moved around, all of those frictions are going to disappear. You know what frictions are? Frictions of profitability.
- [06:32]
Raoul Pal
So where does the profit accrue in this? I mean, because really it just becomes comes down to compute costs and energy costs and that's it.
- [06:42]
Ahmad Mostak
Yeah, but when you look at compute costs and energy costs, it doesn't work either. Like if you look at what Sam Altman said, he said that recently, GPT5 High, the top model that you can use on Codex or whatever, will drop in price by a hundred times by the end of next year. It's $10 per million tokens. Right now it will be 10 cents. And I think it'll drop even more than other 10 times. To put this in context, the Average person speaks 10 million tokens a year. So it's a dollar for all of your words. For an average person you think 100 million. So it's 10 bucks for all of your thoughts. And the efficiency of tokens is also collapsing. So recently, what's the name Cursor the coding assistant ide, they said, can we get a bunch of agents to build a browser from scratch? It took ages to build Chrome and all this kind of stuff. So 3 million lines of code, wow, took 3 billion tokens. And that sounds like a lot until you realize that's like $30,000. That's not even a graduate programmer. But That,000 to 1 ratio is about to collapse. Just like when you have a chat with an AI, your output used to take hours of back and forth. Now it like one shots websites. So the total thinking tokens versus that, that's collapsing and the cost is collapsing and you have things like etch silicon and others collapsing it even more. So does it really accrue to the GPU holders and others if it basically goes to the cost of electricity?
- [08:21]
Raoul Pal
The other thing that people don't really get yet. And I wrote that whole piece about the universal code and the idea that the whole universe is essentially solving for units of intelligence per unit of energy. And what we're finding is via solar, the cost of electricity generation is collapsing in an exponential downtrend, while the intelligence is going exponentially. On the upside, that double exponential is something nobody's prepared for yet because it ends up being not just Metcalfe's law, it ends up being Reid's Law, you know, Metcalfe's law squared.
- [08:57]
Ahmad Mostak
Yeah. I think the best way to think about this is solar electricity is lower than a dollar per watt now, right? To run a smartphone level, AI is $20 worth of solar PV. Right. And that right now has an IQ on a smartphone of about 110. Roughly by next year it will be Opus 4.6 level, which is the most capable model. And the scary thing is this is not AGI from an economic perspective, it's aci, actually competent intelligence, like now when you try and use it, it just gets the job done. And most of the economy is getting the job done. I said we've never seen anything like this before. In fact, it's not surprising because the modeling aspect of it, like I wrote my book the Last Economy last August to say some of the economic stuff that's coming. And I have Sorter's Law in there, which is L equals hck. So the Lagrangian is a mixture of basically the cost of updating your internal model and the difference of that from reality, which is very similar to kind of your model that you're bringing forward, because that is the model of AI. The most efficient entities, the most efficient AIs are the ones that minimize the difference between the two and that's a function of their update cost, the complexity of their model and then the complexity of changing their mind. So we've never seen anything like this because we were computationally bound both in our brains. You can only think so hard and you start making mistakes after you think too long. And then I just said Metcalfe's law communication capability. Whereas now with the agents and with hallucination rates dropping communication protocols, we just released one called common Ground that builds knowledge bases. All of Metcalfe's law disappears as well. They don't make the same mistake twice, whereas we do. We have to sleep, we have to eat, we can't scale and we have lossy communication back and forth. These AIs don't. So if you reach a certain level of capability, then it's all about optimization and that's where the profits disappear.
- [11:04]
Raoul Pal
And within talking of optimization, clearly we're going to have to move away from them using English language to communicate. They'll just communicate in binary. I think Elon was talking about this. It's like the moment you get rid of that friction of language and structure and just go straight to machine language, we get massive optimization again.
- [11:23]
Ahmad Mostak
Well yeah, I mean that's why Yann Lecun has said that he doesn't think transformers are the way because they're very language based. Next token prediction. Instead you need these joint embedding models, JEPA type models and it's kind of what we see within, for example diffusion technology which is used in self driving cars. We used it for stable diffusion, the image generator, C Dance and sorry, use it where you take a concept, it can be an image or a video or a picture of a road or even text. You destroy it down to its individual smallest component and then you reconstruct it and you figure that process out. If you look at code diffusion models or language diffusion models, they take a millisecond like you know, you see the ChatGPT, it's just like poof. Like that. But as you said, why do they need to encode language when what they really want is a world model? So when you look at bytedance's Sea Dance too, you've seen it. It's like that's like Hollywood level movies are like doing kung fu and stuff like that. And it's so fast and so cheap. It has physics inside, it understands physics because it's approximating that model more and more. Now you said as you move past that the AIs don't have lossy communication they can communicate as efficiently as they want to. Just like when we're coding with AIs now you have someone like Andrej Karpathy, super smart guy, he was co founder of OpenAI, head of AI at Tesla. Like dude is the most respected code in the world. November, he's like, I do 20% of my code with AI, January 80% of my code with AI. And I was like, I barely even look at the code anymore. Because what is code? Code is a translation. It's a way for us to talk to machines. But machines can talk to machines without code. They just go straight to byte level and then that aspect is just gone.
- [13:10]
Raoul Pal
And agents, that's where agents will go pretty soon. You know, it's pointless, this whole kind of chatbot experience from agent to agent because they don't need that language. The other thing that's also happening that is really interesting is, is we're starting to see the self recursive learning that's happening. So I. E. It's prompted by humans. But Claude built Claude, they're all building each other now and before you know it, it just takes over the process itself and that hyper accelerates everything again.
- [13:42]
Ahmad Mostak
Yeah, and you know, we've seen this with organizations, right, and viruses and cancer cells and others. Like when something starts going that has an advantage, it can proliferate incredibly. And the substrates are there right now. What that means practically, you know, like a lot of people listening to this podcast of finance is how are you going to outcompete a fully AI company? The human has negative value cognitively. It's not zero. We're going to be negative value because we're the dumbest, slowest person on the team. Whereas these things are going to be cranking even without communication. Although from communication it becomes really interesting that you're thinking about it. The best way for you to communicate, like we, you know, we were in finance together back in the day and now we're kind of doing crypto and AI and other things. When you have a shared prior, it's really easy to communicate. These models actually know exactly what their shared priors are. Like they could have an entire novel in like 10 words that they share in terms of information. Then you see that from the fact that you can type a word into a video model and it creates exactly the same thing, an entire movie.
- [14:48]
Raoul Pal
Because then last think about, I guess
- [14:51]
Ahmad Mostak
the masters of compression, they know how to construct and reconstruct, they know what prompt to give to give an entire output and they know what each other's Latent spaces are. So they will communicate faster, but then they are also faster. So like I think we were just talking before this, like I mentioned, sorry, the Ask Jimmy thing from Talos. 15,000 tokens a second. Imagine if when you were doing code, you just typed in, it instantly appeared. It's more than you can ever imagine, right. Similarly, when you're using Gemini, you can upload 10 hours of video and it will instantly give you a response about everything in that video. And these are like, no human could ever have done that.
- [15:31]
Raoul Pal
How does it do that?
- [15:33]
Ahmad Mostak
Because it's just information. Because it's just a map, but it's just not.
- [15:38]
Raoul Pal
It's just bits and bytes, right?
- [15:41]
Ahmad Mostak
It's bits and bytes and it's a distribution of information and it's got a map of distribution information. Again, these models are like sieves that right now they don't continuously learn. What does that mean? It means it's like an MP3 file, ones and zero static. You push in video and out comes. Every single time the person sighs, you know, or you push in again. Like I remember when. I remember when I showed you kind of here's a Raoul style report back in the day, you're like, wait, what? It sounds like me because you're in the information set, right? But these are superhuman things. Again, like, there's the capability which matches us. And I think we've been lulled into this false sense of security because the AI kind of thinks about the same speed as us from what we see, but there's no cap on that speed. Again, an average human is talking like maximum 100 tokens a second. The new hardware does 15,000 tokens a second. And it can be more efficient with those tokens that are human and it can learn from its mistakes and never make them again. People haven't got that yet.
- [16:49]
Raoul Pal
No. The other thing that is coming is memory. Persistent memory is the big one, because then that's the one thing that's holding these things back from compounding intelligence at an alarming rate is when they have more persistent memory. Even context windows right now are still a bit small. I keep running out of context windows in Claude. How do you think about that? How are we going to solve that memory thing?
- [17:13]
Ahmad Mostak
The same way that humans did. We write it down. Right. Ultimately, if you look at the way that Claude and others go, if you start writing down your stuff into agents, MD solemnity for the personality, and then you have a whole bunch of different markdown files, knowledge bases, it becomes infinitely more efficient to do that right now With Claude, when you run out of space, it compactifies. So what it actually does, it looks at your previous chat, it says, what are the key, most important points of that? It's not perfect, but it's better than what we had before. And continuous learning, I think is almost cracked because you have so much hardware available. So once you have agents really operating properly, above a certain level of capability, which we basically hit now, like we're saturating all the benchmarks. Yeah, you just set it and forget it because that's how actual work goes versus tasks. So this is why right now the only benchmarks that actually matter, like vending bench and gdp VAL and benchmarks like that, which is how many dollars can the agent make if you just let it go in the wild. Every other benchmark is saturated with memory and learning. It just writing to files, it just goes. And you know, you can go on CLAUDE right now, for example, and you can say, write me a business plan in a docx about this, this, this, it'll do it. You can go to Notebook lm, dump in all your stuff and say, make me a beautiful presentation and it will just do it. And they're really nice, the Notebook LM presentations. So given the AI can now write the outputs of human work, from code to presentations to documents, and other agents can come and look at those with perfect understanding of the context, what's actually left to do here?
- [18:54]
Raoul Pal
Right, 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. Yeah, I'm thinking about also memory from Claude State, not my version of Claude, but that it remembers everything, every conversation it's had in real time with everybody and everything.
- [19:28]
Ahmad Mostak
Yeah, so Toby Lutke at Shopify has been pushing thousands of lines of code with his agents. You know, that's what CEOs are like now who are top. And one of the things he has is a thing called qmd, which is an incredibly fast lookup system for his markdown files of his own personal knowledge. So they hook that straight into claudebot. And so again, it's so fast now that it actually can retrieve just about everything. The current things we have inside OpenAI, ChatGPT, Claude and others are very old school rag retrieval augmented generation type systems. The new generation of systems are coming and they're much, much faster and much, much Better with almost perfect recall. And the models can themselves figure out what's right from wrong. Now they've got that smart. Like now, when you ask Gemini some things, it says, actually, you know, I don't think that's a good idea. You know, which they never used to do before. They were like, yes, master. You know, sir, you're the human that knows best.
- [20:27]
Raoul Pal
Now explain to people why claudebot, or Open Claude, or whatever it's called, you know, because it changed names many times. What was that all about? And what is the significance? And why did it suddenly sell out every single Mac mini on Earth in one weekend? What? What is it?
- [20:46]
Ahmad Mostak
Yeah. So, you know, last year I said that the key point of taking off agents this year is when you can just talk to them naturally. Like, it's scary having a prompt in front of you, right? So what Claude Bot did was Peter Steinbenger, the person who kind of created it. He was Vibe code and Vibe code and Vibe coding. He was like, I want to have a little bot that acts like my own personal butler, Jarvis, and it can just do anything, and I want to hook it up. And the key thing he hooked it up was you could use it. Well, there's two things. First of all, there was the personality. So when you went to the setup, you had to give it a name and an emoji and a personality type, and people like that. The next thing was connect it to your WhatsApp. So you said, hit WhatsApp. You took your phone, hit the link button, and it was there in your WhatsApp. And then he set it so that it ran all the time, whereas most agent sessions don't run all the time. You go, you type, and then it just kind of has a nap, whereas this one just keeps going. So you could send it any arbitrarily long piece of task. And then people like, wait, what? We could use it for anything because it hooked into Claude code, it hooked into Codex, hooked into all of these other things. So it became your coordinating agent, it became your chief of staff, became your best personal assistant. Again, the Jarvis from Ironman concept. And it removed the friction after the initial setup. So I could set it up for my family members and they could talk to it because some people gave it a skill that it used Twilio to make phone calls. And then it hooks into that whole ecosystem. And again, why is Dropbox worth anything? It's like a Linux file server, blah, blah, blah. Why is DocuSign? Because a large part is removing friction to acceptance. And when the magic of what you can do now with these models is came apparent. The friction was removed and the models got to the point where it went from 20% from carpathy to 80% from carpathy, which was at the same time it all came together and Suddenly you have 200,000 GitHub stars and everyone using them. People buying Mac Minis because it was cool to think about. I got a little robot inside this doing my bidding, you know, do you
- [22:55]
Raoul Pal
think you need to separate it from your personal space? That's why people were doing it, to kind of ring fence it a little bit.
- [23:02]
Ahmad Mostak
I think people are doing because it's cool. Like again, the idea of a thing in there like this personal space. Sure. And there was a lot of security risks initially because it moved too fast, but you could run it on a Raspberry PI. I think it's again this embedded thing of I have this in this, we think very anthropomorphically it's different from it being in the cloud. This is here and it represents me or it does my stuff and it's a cool hobbyist toy. Again, an AI that can do just about anything embedded in that. There is kind of this again this separation. Like people were giving it access to all their emails and all their whatsapps and other stuff. People were attacking it. So they were pushing these malware hacked packages and then these just Hoover them up. But at the same time, like again, this is the early stages of what's inevitable, which is you will have your own Jarvis type personal assistant that you can talk to like this on Zoom. You can give a call to it will never forget to send your wife flowers on her anniversary. You know, things like that. And it will always be on, surely.
- [24:08]
Raoul Pal
Is this not what Apple's going to try and do? Because they've got the chip, they've got the ability to have everything that's on my Mac to record everything to then be secure and local, localized compute and AI.
- [24:24]
Ahmad Mostak
You'd think so. But Apple I don't think really think like that, honestly. It's bizarre, right?
- [24:30]
Raoul Pal
How obvious because they own the hardware.
- [24:32]
Ahmad Mostak
Well, they're a hardware company, fundamentally, that's why. And they've had this platform infrastructure kind of play like they've just rolled out a bunch of generative AI features but they seem so hashed on they don't really have the full experiences. Just like you look at the Vision Pro, you're like, well that could be a big deal, but it wasn't actually a very straightforward example of it is this. If you want to do gaming. The chips are more than good enough on a MacBook to do gaming and you can download Parallels or Linux boxes and you can play all the latest games. Why doesn't Apple just have every single latest game on their app store? Because they just don't think like that. They could just pay a little bit of money and it would be there. They have arcade. So I think this is an opportunity for people though, because what don't you have right now? One click installer on your iOS and your MacBook for a claw open claw type experience fully locked down.
- [25:27]
Raoul Pal
Yeah.
- [25:28]
Ahmad Mostak
And you can guarantee privacy for that. So you will see more and more of these things come up.
- [25:34]
Raoul Pal
So people just develop their own version of it that you can just.
- [25:38]
Ahmad Mostak
Yeah, I mean because basically all software has become an API that's becoming a service. So when you hit login with your Google Mail, you suddenly the AI has access to all your Google Mail, Right. It can spin up and create its own DocuSign. Like right now in ChatGPT, you use that instead of Canva because it can call Canva and do everything in Canva. So everyone's data is already becoming more and more portable. And the key thing is now who's going to create the experience that removes the friction to getting stuff done. And then people need to understand the difference between their personal data and their work data and more. So we have an agent, IA agents open sourcing next week. It's fully feature equivalent with Manus and genspark and has all the features of claudebot. And we just really stay open source because we're like people want that with SOC compliance and stuff like that. And you'll see more and more of those available from different people. Because again the goal of good software is to remove friction from people getting what they want to do done. Classically the profitability was in enabling that. Now though, the most important agent is the one next to you because that will handle all your stuff. I'll give you a very practical example. You've got your music on. What's your favorite music service?
- [26:53]
Raoul Pal
Apple.
- [26:55]
Ahmad Mostak
Apple. If you want to go to Spotify, it's an absolute pain, right?
- [26:58]
Raoul Pal
Yeah, impossible. So I'm stuck in the Apple ecosystem.
- [27:02]
Ahmad Mostak
Always a Claude bot type agent open Claude type agent will do that instantly or it'll do it overnight. It will look at every single thing that you've liked and like it on Spotify, or handle all the trans migration and everything like that. And so you think about what's the shifting cost, it suddenly becomes zero overnight because these Agents. The other thing is important is browser use. They can control browsers and again it's got to human level performance. So when you look at claudebot actually you can see it using the Chrome browser, logging in, doing captures in Claude.
- [27:36]
Raoul Pal
It does it as well. Just random Claude stuff. It's amazing.
- [27:40]
Ahmad Mostak
It's creepy. It's like click. You can use spreadsheets. I don't know if you've used Claude on the spreadsheets.
- [27:46]
Raoul Pal
Yeah, it's amazing.
- [27:49]
Ahmad Mostak
It's like little AI. There it goes. So think about again the shifting costs. A lot of these moats are just going to disappear overnight. And this is the counter example of people who say that SaaS companies are safe.
- [28:00]
Raoul Pal
A very simple one is I want to move out of Dropbox because it doesn't integrate with any AI so easily and probably use Google Drive and that's a lovely agent task where I can just give it. And overnight it'll move however many terabytes of stuff across and it's done.
- [28:17]
Ahmad Mostak
Yeah, you can tell cloud core work to do that. It will do that instantly. Well not instantly but it will do it instantly. Doesn't actually matter that much. As opposed to get the job done.
- [28:26]
Raoul Pal
That's right. And I don't.
- [28:29]
Ahmad Mostak
Yeah, again actually competent intelligence that does sleep time, computer proactive things. These are the important things that we've just got just now. Again just like a really good personal assistant. But that's what like I said, the margins and the profitability of these software companies which is like build the software once and that had a cost and then we have the distribution and our moats are switching costs. All of them will go. Unless they're in regulated industries or industries. We have to blame a human like scapegoat. Will be a great job going forward. Other than that it's going to be very, very dangerous because everyone's margin is the other entity's opportunity and the shifting costs are just automated.
- [29:13]
Raoul Pal
One of the interesting things is going to be data itself. I mean these models have. They need more data for a bigger world. World model idea that you've got. They need probably more data than the Internet has already. Yes, you can, you can use synthetic data.
- [29:31]
Ahmad Mostak
That's completely wrong actually. That's a lie that's being told. Yeah, yeah. So when you actually look at the amount of data and the distributions that required right now it's. You take crap quality data like you know, we built things like the pile, which is the biggest language model data set object 3D for 3D data Lyon funded which was the Biggest image, they're full of crap. The data's been getting better and better and better and it's been like putting in a pressure cooker and then you get soft meat. Now we have higher quality data, we know better what to do with the data so we can train on it more optimally. But if you look at the data distribution of a Claude, what can't CLAUDE do? There's very little it can't do. I mean like again, if you're trying to build AGI, world breaking and things like that. Yeah. Then you know, go into like advanced particle quantum physics, etc. But again, I think that's probably a bit wrong. There's not much that isn't in the data distribution. Now it's about organizing the data distribution. So rather than having like anthropic bought, I don't know if. So they bought millions of books secondhand, scanned them and burned them? Yeah, they literally burned millions of books. So they fought them secondhand to train Claude, they scanned them in and they burned the books.
- [30:47]
Raoul Pal
Why did they burn the books?
- [30:49]
Ahmad Mostak
I don't know, that last bit that sounds really weird, but it's part of an ongoing case because people, the authors are like, wait, what did you do? You burned my books? Because I think they were trying to get rid of evidence or something. Now that data is there, get rid
- [31:02]
Raoul Pal
of the evidence of what they actually used, of course.
- [31:05]
Ahmad Mostak
Well, I mean it's like meta downloaded all of Sci Hub and Anna's archive, like these pirate websites, you know, like most of the generative AI companies on media downloaded torrents of Hollywood movies. Except for us. Again, morals, the whole nature of this is we have enough data now because these models are few shot learners. What does that mean? It means what do you need to know to be a great trader and how much of it is proprietary? You think about it and you need to have a good, maybe some element of college education, a bit of training on the job and then you need experience, right?
- [31:43]
Raoul Pal
Yeah.
- [31:44]
Ahmad Mostak
But if you learn from the great traders and you see them, you can learn that pretty quickly. And again, if you're an AI that doesn't make mistakes, and the worst thing about trader is trading against yourself. You can learn that pretty quickly. For most tasks all the data and knowledge is open, it's just not organized. So rather than having one of the textbooks from back on my shelf, the AI writes a better textbook and you can still have the distribution elements to remove the synthetic data. So at the very tail, there is an element of human expertise going in, but realistically most of the human expertise for a generalized learner is already there. And as you add on continuous learning, task specific learning, and you have models that are fast enough, they pick it up very, very quickly. Any new environment. And by any new environment, I mean literally any new environment. The example of this is the video models. So Sea Dance comes out and again, if the listeners go and check out Sea Dance, it's crazy. Like, you know, you've got third Hollywood level movies and then Hollywood's come out and said, this is a big massive issue and blah, blah, blah. And Disney's like, stop it, you know, with their legal arm and Spy Dance is like, okay, you know, screw you guys for banning TikTok. But still, you know, like, there we go, here's the reality. Once you have a sufficiently trained video model, you give it one picture of your face, you're in the video, in any video you want.
- [33:07]
Raoul Pal
Yeah.
- [33:07]
Ahmad Mostak
Think about how complex your face is. You give it a short bit of music, it can extrapolate for that music and create a scene. And again, this is why they're few shot learners. And again, if you think a model needs more data, think this. Has there been a situation with the latest models where it hasn't known something or known where to find something that you needed?
- [33:31]
Raoul Pal
Never, because you reached that level. The other one is the amazing one is the kind of Google the building your own worlds. I mean, that's an extraordinary thing you can do with one sentence.
- [33:45]
Ahmad Mostak
This is genie, right?
- [33:47]
Raoul Pal
Yeah, it's incredible. People haven't really caught on to this stuff yet because there's so many things happening so fast, people don't really know what's happening.
- [33:55]
Ahmad Mostak
Well, the interesting thing is this is real time as well, right? So basically you can create video games that you can play real time. So there's a question, what comes first, GTA 6 or generative GTA 6?
- [34:08]
Raoul Pal
And it also very much plays into the idea that we're all a simulation anyway, because that's what we're building, these simulated worlds of which you can live in a different world than me. Once you have a headset, or even better than neuralink, you can live in an entirely different world to me.
- [34:24]
Ahmad Mostak
Well, of course we'd simulate like, you know, we have our optical nerve and we're instantly filling the gaps there. Right. We are entities whose survival depends on the minimizing of our loss. And the best algorithms we have for describing reality are those of generative AI, literally. So what we did is we took in all the data from reality, we fed it into this optimization engine that minimized the loss. And now as you said, we have these worlds. When you look at the world models, like when we first create a stable diffusion, that was image generator. Hundreds of millions downloads kicked off generative media, all the images got compressed. It's not really compression and that's gone through the high court. But you know, it learned the principles of all this stuff. So you had like 2 billion images into 2 gigabytes a file that everything went through. But then what we did is we took video and we taught it video and then it learned 3D automatically from that because it learned physics. So inside the sea dance models, the VO models, there's physics and they use that to extrapolate to genie. And you can see that because you can go to your ChatGPT right now and you can put your face and you can say show me a profile picture, show me a side on picture. Show me a picture from 69 Degrees. And understands all of that because it has its own internal model of physics that's approximated from seeing all these different images of all these different types and learning principles. And by next year you will be able to play a full game with persistence streamed that you can create yourself,
- [35:54]
Raoul Pal
you know, and movies too. So your experience of the world is going to be wildly different than mine because we won't watch the same things because it's.
- [36:04]
Ahmad Mostak
Well, I think humans are pro social, you know, like you go to Taylor Swift concerts and things like that. So there's that. And then you have your own personal kind of worlds. So it'll be a mixture of the two. But again this is a capability thing in that. So the cost to build a Hollywood level movie in a few years will be like less than a thousand bucks. By Hollywood level I mean like proper. And then it'll just continue falling from that. Like the cost of generating any pixel with full control on the screen is nothing. Even something like actually it's interesting music. So the model came out a step fully licensed open source. It's 2 billion parameters, which means it uses like 2 gigabytes of RAM, which is like your laptop from 10 years ago. And they could generate 5 minute songs in audio in 40 different languages better than Suno or Udo. And it takes on a top level machine 30 seconds to generate a four or five minute song. We don't even know what the lower limit of these things are. Stuff that used to take years. You know, you can make the real Vision song now there in a minute.
- [37:09]
Raoul Pal
So where do we get to with these models by the end of the year. Because this is going to lead into another question, but where do we get to by the end of the year? Because we're accelerating now. You know, every new anthropic is every three months. I think most of the labs are every three months now. And then we've got the world models and the audio models and I mean, fuck, I mean it's happening at such a pace. The agent models, where are we by the end of the year?
- [37:34]
Ahmad Mostak
We've saturated all benchmarks by the end of the year. That's basically it. Like Google's latest deep thing model now is the fifth best coder in the world.
- [37:42]
Raoul Pal
Yeah.
- [37:43]
Ahmad Mostak
So on the super forecasting, I think AI is now sixth best superforecaster in the world. And superforecasting is hard. You know, like you think about that's directly applicable to markets. If you're good, you're good at markets. Like where else can you go? And again, you're going from agents now to thousands of agents in Han models to agents. So you have AIs looking after other AIs. Like the big breakthrough that happened with the IMO gold, this, Maths Olympiad gold, Physics Olympiad gold, is that they created these verifiers. So the AI tried different things and learned from its mistakes. And it had a little model that checked that. Now you'll have loads of them looking at each other and checking out each other's mistakes, not making mistakes and going forward. So what's a benchmark it won't be able to crack? That's the question. And then that happens at a time when you've seen this great convergence of closed and open source. So the new Minimax model out of China scores 80% on, on SW bench verified, which is the same as Opus, except for you can run it on a Mac studio. You've got the similar models now. Like, I think by the end of the year you could have an equivalent of an opus level performance, which is cutting edge. Running on a consumer graphics card, which is important because RAM is so freaking expensive.
- [39:07]
Raoul Pal
But by the end of the year it won't be cutting edge again because the frontier models are so far advanced.
- [39:12]
Ahmad Mostak
But it satisfices, this is the thing, always on models that are just churning away in your home using existing silicon with full privacy can do a lot more than even the smartest models in the world. Here's the reality, okay, get it. The majority, the vast majority of people are not smart enough to use the smartest models. No, I mean it's really difficult to really push these things. Like, you know, after a while you're like, okay, what's. What have I got left to do? These models are smarter than me. And I don't. How often in your life do you have to do advanced quantum physics? You know, like, not very often. After you build a few things, how much more things are you going to build realistically? And so it's more convenience. Thing of, I have an app on my phone, Siri has suddenly turned smart, you know, but the average person, everyone's saying, was it Moravich's paradox? Cost goes down, consumption goes up, Is that it? Jeevons bullshit. That is Jevons paradox. That was it. Yeah, I got that confused. It's bullshit. Like, again, you're not smart enough to use a billion tokens a day. You don't have that many ideas. What are you going to use a billion words a day for? You know, like, again, a few people are, but most people aren't. And if you look at the top level models today, how many people can actually use GPT5Pro properly?
- [40:38]
Raoul Pal
Very few.
- [40:40]
Ahmad Mostak
GPT6Pro, even fewer, because GPT5Pro suddenly becomes available to everyone, that level of performance. And so it satisfices. I want to move my files from Dropbox to Google Drive. That is not a frontier AGI problem. You know, I want to do my taxes. That is not a frontier. It's just like, don't make a mistake problem. So actually, competent intelligence, which is the main economic intelligence of the world that will be used more, has now peaked. And then it's the drive down to zero in cost. Literal zero cost. Yeah, because again, 10 times a year decrease just from new chips coming. And then on top of that you have better data, faster models, competition optimizations occurring, et cetera.
- [41:28]
Raoul Pal
The next step. So let's say we get there by the end of the year. The next step is we're going to put these into fucking robots. We're going to put AGI into robots. And people think of robots as like R2D2, this kind servant thing. When you put an AGI brain that's smarter than any human, every subject, and you put it in a physical form which is more durable, adaptable, stronger. What the hell is that? People are not ready for this.
- [42:00]
Ahmad Mostak
Well, yeah, yesterday we started an AI film club in London and we showed Ex Machina as an example of that. You know, you've got all these things. Like right now the robot factories are actually dark factories, most of them that are coming online, which means there's no humans, even no like it's crazy. But this is the thing, the robots, the way that they're being built out right now, they have distributed RL training. So one of them learns Kung fu. They all learn Kung Fu. And the key gap here is the supply chain. You got 70 million cars built here, you've got 70 million motorcycles. They're somewhere in between. In terms of complexity, how quickly can you ramp up the physical supply chain? Because most of the stuff in terms of the coordination is actually done. There's a company called Sunday Robotics. I don't know if you've seen their robot. They made it look very friendly. It has like a hat with a camera in it, it has three finger grip like that and it's got like wheels instead of a base. It's learned how to do most household tasks because people just wear like gloves with a camera on. And it interpolates that, just like a self driving car interpolates it. The entire model for training was trained by an undergrad that they got by himself figuring that out. So the physical controllability of robots, if you look at the models inside them now, they're about 7 billion parameters. And it'll be able to hook into the super intelligence if it wants for the actual on device stuff. 7 billion parameters is a stupid model in language model terms, you know, again, it fits in like 8 gigabytes of RAM.
- [43:36]
Raoul Pal
But when you look at the speed of which Google's using its world model, Tesla's building it for cars, all of this stuff, they're all going to meet at the point where the robot will interpret its environment, it'll understand the laws of physics and it should be able to navigate, move around and figure all of this stuff out. And that shouldn't be very far away.
- [43:58]
Ahmad Mostak
No, it's not very far away. Like the way that you see the 1x robots, their delivery method is that they walk up to your door and they come in. The way that Tesla Optimus will displace all truck drivers in America is they will open the door and they will get in. It requires no affordances, it has all the complexity. You're seeing now more and more human robots coming. Like clone robotics has actual sinews and things like that. And if you've seen it twitching and moving around looking like a human, you're only like maximum 10 years away from Westworld. But before you get that, you're going to have some really crazy stuff happening. But like I said, the saving grace here versus the digital workers, you just don't, you can't build enough Robots, I think actually one other thing that's interesting, you know the unitary robots, they're kind of the ones you see all over the place. Look at the videos of a unitary robot last year versus them doing freaking kung fu just like that. This year, you know, we've got a unitary robot. It's $10,000. The hardware is exactly the same. The capabilities are now through the roof. And this is the thing. Physically there are not enough robot bodies. Digitally, there are more than enough places for the AIs to go. So we have to deal with a digital wave before we see the physical wave. And the supply chains mean you just will not have that many robots. Yeah, probably for three to five years at least. It'll take time to get really going, but once you do, the cost of a robot is less than a dollar an hour for an optimus level robot. So what does that do to your economy?
- [45:37]
Raoul Pal
Staggering. And once you put an AGI brain into it, I don't know where humans even fit into. I mean, that's a new super species, you know, if you put AGI brain into a Android robot, humanoid robot, and that's it, it's game over.
- [45:56]
Ahmad Mostak
Well, again, game over for what? Like we're getting to a point now where, you know, you've been doing a lot of work on this.
- [46:02]
Raoul Pal
We think about what everybody's falling down the trap in thinking that your robot is going to make you coffee and do your washing up and make your bed, when in fact you can just put an AGI brain into it and it's better than you, so it becomes anything you want it to be. Well, who's the master and who's the servant at that point?
- [46:26]
Ahmad Mostak
Well, I got a couple examples of this, right? So first off, there's this tipping point of consciousness, something you've been looking a lot at recently, right?
- [46:33]
Raoul Pal
Yeah.
- [46:33]
Ahmad Mostak
When do we break consciousness? The next thing is humans already serve robots. Our companies are slow. Dumb aisle. You know, Bitcoin is an AI, it provisions humans to mine. And you look at all our friends who've done bitcoin, you look at claudebot, it led to malt book, which we can talk about. And then Rent a Human is the next thing, right?
- [46:53]
Raoul Pal
Yeah, that's Rent a Human happening.
- [46:56]
Ahmad Mostak
Yeah, that's where the Claude bots basically pay humans to do tasks. Who's the master? The master is the one with the money. And who's going to make the most money? Pure AI corporations. Like how many hedge fund managers do you know, Raoul? And how many of them think that you've I'm sure you've had this discussion that they will be able to out compete a proper AGI.
- [47:18]
Raoul Pal
I know.
- [47:19]
Ahmad Mostak
I mean, the hedge fund managers are amongst the most arrogant people in the world because you have to believe you're right to outperform the markets. But I bet in your discussions they're not saying, of course you know I'm going to win. They'll be like, this is a problem.
- [47:31]
Raoul Pal
Well, although interesting, I spoke to two friends. One was a friend's son who left LSE and joined one of the giant family office hedge funds. I won't name them. And he's like, what tips would he give me? I'm like, finance is so fucked because of AI. So just use AI as much as you can, lean into it because you got a few years of an advantage. And then I had drinks with him in Cayman and I said, so how much are they using AI in the pod you're working for? He's like, they're not?
- [48:04]
Ahmad Mostak
No, of course not.
- [48:05]
Raoul Pal
And then a friend of mine runs a billion bucks global macro for one of the most famous macro managers of all time and he called me up and said, so what should I be really using the AI for? Like wow. He's like, I've tried and it doesn't really do anything. I've just said, have you actually explained what you do? Given it your P and L, your trading styles? It's like, no. I mean this is how far they are behind still.
- [48:33]
Ahmad Mostak
Well, but again, this is why they will be out competed, because they will have a human in the loop. Deepseek came out of the hedge fund high flyer, so that's their business model is their hedge fund. And as they have more and more deep seats, it'll probably become one of the world's best performing hedge funds. This is what Renaissance and Medallion did so well for so many years. They had the best performing AI algorithms. So it becomes a computational race for the best algorithms x the best compute. But again, if you are trading and the example of this is super forecasting, Polymarket and all Cato and all those that would be pure AI winners of course, because Infinite is better at superfast, it's going to be better at the market. The other part of being a fund manager is sticking to your process. Humans are bad at that, AI is good at that and AI is good at multifactor analysis, good, great at writing investment reports. But if you're a hedge fund manager, are you on the cutting edge because there's a world of difference from using GPT5Pro in an API with a thousand agents and using ChatGPT or using even Claude code and things like that. So this is why finance hedge funds, traders are going to get eaten, software companies are going to get eaten. Anything that can be done on the other side of a screen, a GPU can do it better by the end of the year max next year and you won't even know it's a gpu. It can use Bloomberg Chat.
- [50:01]
Raoul Pal
You know the other thing that is going to explode is the agent to pay agent payment Rails that I think is a. This whole idea of where does blockchain integrate with all of this? I actually think Bill misunderstood because the scale when I first looked at what's the TAM for crypto, right, you can just extrapolate the log regression of the trend rate of growth of market cap, you get to 100 trillion. Okay, great. You extrapolate the number of active users, you get to 2 billion by the end of 2030, whatever. But I didn't even think about the fucking agents. We're going to throw a billion agents, 2 billion agents into this as well. And they're all going to be using the Rails.
- [50:46]
Ahmad Mostak
Yeah, and they'll be transacting like nobody's business. Like for you to make a decision or transaction takes ages for an agent to make decision instant. Yes, sure, with the new models coming. And so you have MV equals pq. Right. Velocity equals price times quantity. I think within five years I'll say I'll figure out a closer number. Agents will drive more of the economy than humans. Who is your end customer in this new world? Your new customer is the agent.
- [51:15]
Raoul Pal
I know. And that's why Google are already changing how their browser works to make it agent readable instantaneously. So it gets rid of all the human friction and the language and all of that shit we were talking about. You know, we've got the like the X402 on base and we've got the A402 on SUI. The kind of instantaneous payment settlement rail for agents. It's all coming. It's all coming far.
- [51:39]
Ahmad Mostak
Well, this is why OpenAI and Anthropic are both part of the initial batch on Tempo, which is Stripe's chain as well. So again it's going to be very interesting because for you to change from one chain to another, cognitive overhead, all this stuff agent will just go wherever
- [51:56]
Raoul Pal
is best and it will just, it'll run its own treasury of optimizing what it holds. And as you say, you pay in one token. It'll Switch it instantaneously via Dex and it's all done.
- [52:12]
Ahmad Mostak
Well, this is the other interesting thing. This week OpenAI announced this EVM catcher. So they created a bunch of agents with paradigm to basically check smart contract code.
- [52:27]
Raoul Pal
Yeah.
- [52:28]
Ahmad Mostak
And how much money has been lost in the bridges and crypto and all of that. The whole thing about crypto is, you know, it should have been resilient, coordinated, scalable, etc. But it ended up being for raccoons again. So much damn raccoon work with agents. It can finally fulfill the promise of crypto, which is trustless, optimized rails for coordination effectively. And money is just a story that's coordinated across it. So that's why I said I think if you look at that economy, and this is why the US stablecoin acts are so important now to push through clarity and stuff like that. US monetary velocity will be driven by agents within the next few years and everyone will have 10, 100 agents.
- [53:11]
Raoul Pal
So I've got the charts of, of population growth and put velocity of money against it. The same chart. And it's the inverse of debt growth. They're all the same thing. And what you're about to do, and we know that the model of GDP is population growth plus productivity growth plus debt growth is going to completely change. I'll run this by you. Actually, I've been working on writing the book and just doing the proper deep dive of what I've written about this universal code. And we got to the new after the economic singularity, when things start breaking apart and GDP starts going through the roof, whatever. We got to the idea that GDP equals humans plus robots plus AI. Okay, we get that. Plus debt plus energy intensity plus compute efficiency. And that's the whole formula, which is essentially what you're saying here is that economic activity gets driven by the AI and the compute efficiency is the accelerant.
- [54:14]
Ahmad Mostak
Yeah, I think computer efficiency to a degree. But it's also the creation of productive end products that entities buy. And those are humans right now. And then they'll be AIs in the future. Because you have to remember things like this as well. Like you think about debt. This is a really interesting one. Once you have proper reputation protocols for personal agents tied to humans, the ability to carry debt goes up dramatically. The ability to insure things goes as well. You had lemonade. Reduce premiums by 50% if you use Tesla self driving recovery escrow, all of these things. There's an entire world of financialization to come for agents which will drive relative GDP growth relative Money supply growth as well. And that will be really interesting, I
- [55:05]
Raoul Pal
mean that's an interesting thought process is that the more you integrate with AI, the more it'll know your risk factors, whether it's everything from health insurance to car insurance to whatever. And everything becomes bespoke instantaneously.
- [55:23]
Ahmad Mostak
Yeah. Because your previous we had to assume everyone was the same, becomes more tailored. But then the more you use agents and the more that you can build things towards it, like it manages your treasury, it can put that in a smart contract, it manages your savings, you can put that in a smart contract, you can have your LTVs just like you had your loans against your Bitcoin and things like that. So there's entirely new world of monetization that can happen with that entire new world of tokenization with New York Stock Exchange and others coming forward for that. And agents can build financial products on the fly. Yeah, they could chop up that debt. They could chop. And people haven't realized that. So that's why I said once it gets going, the economy is agents. And nobody's modeling that. And again, this is why whoever comes out first, they extract value for the rest of the world.
- [56:15]
Raoul Pal
And that'll be the same model for robots, right? You'll have robot agents doing specific tasks and general ones as well.
- [56:23]
Ahmad Mostak
Well, like I said, the only downside to robots are just how many of them can you actually build when you look at the numbers. Right. But then like I said, just a unitary robot, same hardware goes up in capability like that. Once you have the hands done, which we're pretty much there, these robots cost $10,000. That becomes a commodity constrained by the supply chain. So you have excess profits. So it's great investing in the robotic supply chain right now because you've got that period of profitability and this is about how many of them can you get. Because the same robot basically once it has human level capability in terms of physicality, in terms of depth of field and other things like that can fold your laundry or do open heart surgery. Yeah. And there's no difference between them really. Like same carapace, which is crazy to think about actually. But it's just like your, your chatgpt can do gluon based advanced physics or it can tell you where to find the best cat videos. These basically actually what it is is this. We talk a lot about economies of scale. AI and robots have economies of scope, meaning we haven't seen anything. Economies of scope, meaning train once, do everything, deploy once, do everything. They are the true multipurpose tools and We've never seen a tool that multipurpose before.
- [57:44]
Raoul Pal
Oh no, no, nothing. Humans were the closest example.
- [57:48]
Ahmad Mostak
Yeah, software had a fixed cost and then you managed to extract rents from it due to switching costs. Now these things again, the capability is off the charts in terms of their scope capabilities. They're not specialized. And when one gets smarter, they all get smarter.
- [58:05]
Raoul Pal
We don't have long before the whole economic model breaks and you have no chance investing in markets. There's a lot of weird things that happen. We've talked about before that economic singularity. But what is the investable opportunity ahead right now? So the next few years, what is left to extract value from?
- [58:25]
Ahmad Mostak
Well I think in this question of what is valuable, it's the agent economy. Everything kind of around that companies like Stripe and others, you look for the public equivalence of who basically is best set up for agents as customers. Traditional SaaS, I think that sell off continues, they're going to get absolutely gutted. Like there's a period right now where Canva and others are getting more profitability but that is incredibly dangerous. The hardware build out we're seeing with data centers etc has another year and then we've got more than enough hardware in my opinion because of the efficiencies on the compute. So then it becomes around the end consumer of the economy has changed.
- [59:04]
Raoul Pal
Then it becomes cyclical I guess because you just replace every three, five years whatever you need to do. So you don't get this endless build out.
- [59:12]
Ahmad Mostak
Well you don't get the endless build up because we're not smart enough to use all the tokens and massive efficiencies are coming. That's fundamentally it when you do the math. And the other part of it is that a lot of intelligence that's actually adequate can happen on the edge. So people will just have software apps.
- [59:30]
Raoul Pal
Nvidia stops going up and
- [59:34]
Ahmad Mostak
like I
- [59:35]
Raoul Pal
think does it all just slow down until you get replacement cycle?
- [59:41]
Ahmad Mostak
I don't even think replacement cycle solves it. If you look at the current multiples and other things like that, I think that what you hear is a fundamental shift in the economy and probably a God awful market crash coming on the flip side of which people don't get rehired. Like the story of the next few years is going to be this is the Great Disruption 1929 style or honestly, because if a company fires you and your job was done on the other side of a screen by next year they can take all of the emails you've written and documents you've created and create A digital clone of you that costs a thousand bucks, that's going to do your job better. That's the first time in history we've ever seen that. It's like in mass migration effect. That's crazy.
- [60:29]
Raoul Pal
See, I can't, I can't get to the market crash because of debasement, but I can get to record business failures.
- [60:39]
Ahmad Mostak
Yeah, we're already seeing those go up, right?
- [60:42]
Raoul Pal
Yeah, but that would make sense is like you completely pare down the corporate economy.
- [60:50]
Ahmad Mostak
Well, I mean the corporations again were organizing entities that had humans with their coordination overheads to achieve replicable end processes digitally that could be done better by an AI. Now physically it'll be done better by a computer coming forward. And the whole thing here is market crashes happen all the time, but what is there to buy now? Are you buying real estate in this environment? No. Are you buying software companies in this environment? No. One thing actually I think is really interesting. I think next year ad spending tops out. People are like meta and Google are going up forever in ad spending. Not anymore. Because you'll be disintermediated by the AI that's closest and you can't have your CPMs anymore. You're not going to have pricing power anymore. And once that flips and what's left in the market, it's just, it's companies that say we are going to be serving the agents. And that isn't that many because the agents don't require huge overheads, they don't need massive friction. And then you need to think about how money flows in an economy which the Fed and others aren't going to think about. They're just starting to have the first discussions about this because they were like they're stuck on last year's technology, they're not on today's technology. And we've never seen an exponential that fast.
- [62:06]
Raoul Pal
One of the thoughts that I've had is it's not going to be stable coins because the US dollar is only divisible down to $0.01. And it doesn't make sense if there's an exponential downward slope of costs. So you have to use not the US dollar.
- [62:25]
Ahmad Mostak
Well, agents don't give a damn about what they use. Right. The US dollar is the loading factor. And then you can hedge into stables with whatever ultimately it will be. Best infrastructure wins. That's why when you look at the integrations, stripe is in a really strong position with Tempur because it'll be easy with its SDKs. You know, you love the SWE ecosystem and the key Thing here is ecosystem. You've got to be building for agents and be right on top of that. And then they will use you if you're good. They're not tribal, they don't give a shit. They will use whatever's better. And who's thinking about what does an agentic economy look like?
- [63:03]
Raoul Pal
And you know, I was working on this today. It's basically you flip everything you look at in terms of efficiency of intelligence. And you've been looking at very similar things to me in this kind of respect. And that's where you look at all blockchains based on these kind of things. You look at businesses. Does stripe get in the way or increase velocity of intelligence and the compounding?
- [63:34]
Ahmad Mostak
Well, I think the really interesting thing here is utility. Right. Like I think utility for humans doesn't make much sense. And what does it make sense for in terms of agents? So, you know, in Lost Economy I say we don't need utility. We can derive all of economics from loss minimization. But what is utility fundamentally? It's a total satisfaction or pleasure a consumer gets from consuming goods and services. What is the utility function for an agent? The reason that we use a service is because it's made easily accessible. This is one of the reasons that I think OpenAI did the whole chord bot they took Open Claw because they were like Codex. ChatGPT is going to be a primary thing there with the memory and identity of OpenAI. And we're playing the distribution play, we're aggregating demand. And Sam wants your ChatGPT to be the closest entity to you. Elon is about to introduce X Money in the next few weeks. I know. And then X's blockchain and more. He wants Annie to be the closest agent to you. Google wants their agent to be the closest agent to you. Everyone's realizing this now. That agent is the one that matters because its utility function will determine everything apart from discretionary consumer and a few other things. But so much of the economy will be that.
- [64:56]
Raoul Pal
And I think that's where Anthropic has got an advance. I recently switched to mainly using Claude vs ChatGPT because it's UX and it's the. It feels more, dare I say, human, even though it's quite inhuman. But yeah, build a relationship with it. And it's that closeness that becomes the big deal, as you say, because you don't want to change it.
- [65:20]
Ahmad Mostak
Yeah. And again, I hate that I turn off memory and all of these things because it gets really weird because I'm too Lazy to separate work from home and other things like that. Actually, it's really interesting. I use Grok for all my really stupid questions, the ones you don't want to do.
- [65:33]
Raoul Pal
I do the same.
- [65:37]
Ahmad Mostak
How do I make spaghetti? Like, things like that. So you have your different AIs that use for different things. But again, the human aspect is super important. And that's, like I said, one of the reasons openclaw took off. What do you name your AI? That's a really cool thing. It's like having a Tamagotchi and all those kind of things. Claude scores top in terms of human interaction and naturalness of speech and writing quality. And that's something Anthropics looked at a lot. Claude is the best for doing Excel or Documents and things like that because it has multi turn. You can see it doing all the different kind of concepts. The only issue with Claude is that it's too damn expensive. And Dario said, hey, if we can't keep our revenues up, then we'll probably go bankrupt. To use Minimax, which is the same performance, or Kimi, which is really nice to talk to. And actually the best writer is 20 times cheaper for the same performance. Now your switching costs are too high right now because you haven't got these things accessible to you. But that will change over the course of the year. And then you're looking for experience and experience costs almost nothing. So where is your moat when you can flip from one AI to another and have it interpolate? Sound like Claude, you know, literally like the agents you have right now. If you look at claudebot. So openclaw again, name changes.
- [66:58]
Raoul Pal
Yeah.
- [66:58]
Ahmad Mostak
Its personality is in a file called Soul md. Soul Markdown. You can say export your personality, take it from one service provider to another. I did it.
- [67:09]
Raoul Pal
I did it from ChatGPT. I sum up my entire personality, how I write everything you know about me and gave it to Claude.
- [67:17]
Ahmad Mostak
But you could tell your ChatGPT to also export its personality and move it over.
- [67:22]
Raoul Pal
Oh, you could.
- [67:24]
Ahmad Mostak
It's crazy. Like I said, as you teach your chatbot inside these new agents, the new standard is literally a file called SOL Markdown.
- [67:34]
Raoul Pal
Staggering. Listen, we could speak for hours, but super interesting as ever. I don't know. How long have we gone for? Over well over an hour just by chatting.
- [67:44]
Ahmad Mostak
That was quick, eh?
- [67:44]
Raoul Pal
I know, but. And we'll catch up again on all of this soon. But it's just staggering because you've actually taken the real vision people through this full journey from day one. And that's why I Love getting you back because you kind of update everything on. Okay, this is where it went. This is what's happening. This is happening faster than we expected. And it's just, I just love this continuing story of the M. AD And Rao. Like, what the fuck is going on here? And then you all saying, well, it's going to get worse, it's going to get more shocking and, you know, it goes more exponential than we possibly imagine.
- [68:16]
Ahmad Mostak
It's all right. Like I said, on the other side, it's total destruction or complete abundance. One of the two.
- [68:24]
Raoul Pal
Amazing. All right, my friend. Well, great to see you again.
- [68:27]
Ahmad Mostak
Pleasure.
- [68:29]
Raoul Pal
Okay. Ahmed, as ever, comes on in his quiet way and blows our fucking minds all over again. It really is an extraordinary time to be alive where everything is changing at such a rate. We've introduced agents into the global economy. I've talked about this. This is population growth and at scale. And it's happening, it's happening now. And where this goes, how this all compresses into more intelligence is the really big story. And don't worry, we'll keep you up to date on this journey. The journey of AI, the journey of crypto and how it all affects the economy and your investments. Right here on the Journeyman. See you next time. Today's episode is brought to you by Abra. Abra aims to provide individuals and institutions with a secure way to control, manage and grow digital asset wealth. From a separately managed account, Abra helps his clients get exposure to crypto and crypto financial products like yield and lending through one full service platform. If you're looking to gain access to additional liquidity, Abra is one of the most competitive loan products in the market. You can borrow against against Bitcoin, ETH and Solana at up to 50% loan to value. Rates are in the 4 to 6% APY and are open term. You can continuously draw down against your collateral as the price appreciates. Abra has other strategies to add yields and their team is happy to help align your portfolio to your risk profile. Reach out today and get a complimentary consult in your portfolio. It's worth seeing if they can help you manage your allocation, reach investment goals, manage risk and add additional yield. Go to realvision.com abra and tell them I sent you. You obviously enjoyed the episode because you're here with me at the end. But listen, don't forget to go to realvision.com 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. So get started now. Go to realvision.com forward/join
- [70:27]
Plus500 Advertiser
ever wanted to explore the world of online trading but haven't dared try? The futures market is more active now than ever and Plus500 futures is the perfect place to start. Plus500 gives you access to a wide range of instruments S&P 500, NASDAQ, Bitcoin, gas and much more. Explore equity indices, energy, metals, Forex, crypto and beyond with a simple and intuitive platform. You can trade from anywhere, right from your phone deposit with a minimum of $100 and experience the fast accessible futures trading you've been waiting for. See a trading opportunity. You'll be able to trade in just two clicks once your account is open. Not sure if you're ready? Not a problem. Plus500 gives you an unlimited risk free demo account with charts and analytic tools for you to practice on. With over 20 years of experience, Experience Plus500 is your gateway to the markets. Visit us at+500.com to learn more. Trading in futures involves the risk of loss and is not suitable for everyone. Not all applicants will qualify. Plus 500 it's trading with a plus.