Lineal frente a exponencial: la AGI ha llegado, pero también el petróleo, los bonos y la Fed

[00:00] Let's get to work. Um, it's Friday, September 11th, 25 years ago. Uh, I lost my best friend and the best man at my wedding. Uh, tonight I am going to go out with some other friends. So, I'm going to talk about that briefly, but I just wanted to uh say I'm thinking for everyone that lost someone in in there or remembers that day as something as tragic as it was for me. Uh I uh thank
[00:30] everyone for reaching out after I kind of relayed that story a few weeks ago when I was still in Maine. So we're going to go through linear verse exponential thinking. Uh oil bonds and Fed. Oh my. I'll continue on this thing of if you're paying attention to oil bonds and the Fed, uh you're in a linear world and you need to focus on AGI has arrived. That should be the big news story of the week. It shouldn't be whether bond yields were up 10 basis points, whether oil was up $10, whether the Fed might goes from 65% to 85%. Uh,
[01:03] AGI has arrived. Uh, I'm going to go through Astra. I'm going to give you guys some podcast to get up to speed. I am going to show you how to build a trading system using it and especially for your kids. I created a subscriber video uh this week after I built one myself. I'm going to go through what I did, what you guys can do, uh, how you can attach an AI agent to it using Mumu, Robin Hood, Interactive Brokers, you three, reach out to me. I'm trying to help your people be able to do what I'm
[01:36] going to show and connect it in there so that they can start building systems, even if it's paper trading at this point. The amount of entrepreneurial success people will have by being able to do this will blow them away. It will empower them significantly. Of course, there's bad news. Robin Hood, we'll go through a question that I get asked all the time, which is hard to answer for people. How to value Ethereum and the crypto ecosystem. So, uh, Christopher Shawn Kaitton. Shawn Kaitton who worked at Caner. I'm sure there's people that over the years had, uh, met him. This is
[02:06] a photo of him on a July 4th. Uh, very big manquan person. And here is his name down here. Tonight I will be going out with these three guys. Um that's me. This is almost 40 years ago now. That is a photo of Shawn at my wedding. Uh so again for people who remember this day uh in the emotional way that I do I am thinking about you linear visser
[02:36] labs. Again you have to make a decision which side of this you want to be on. If you're worried about oil bonds and rates, again, I'm going to keep saying during these years, the economy cared about oil bonds and rates. That all changed beginning in 2008. That is when the exponential world took off because of the release of iPhone the year before and because transfer payments started to grow and the government really started to take up its debt. This is when we exited this point and this started and I
[03:06] went to Silicon Valley in 2013. I'll get through that again as I've talked about, but this is the world we're in and this is the world that has been making money since 2008. You wanted to be in the MAG 7. So, I figured by now people would have learned their lesson about caring about where oil was, where rates were, where all this stuff was, other than again, I get it traders. But on a day where we had higher than expected core inflation yesterday, we had higher than expected PPI, the probability the Fed went up, crude oil went through, the war
[03:36] is getting worse, and over the two days, as of when I started doing this, the S&P was up slightly. So, how do these things actually matter if they don't matter? So again, the news is focused on this. This is a linear world. These are things that mattered when housing and autos were driving the economy. AI capex is driving the economy. And as I showed last week, it is not at all rate sensitive. Uh the Fed rate hike odds surged to 90%.
[04:07] Jackson Hole from Timuros. I think this the reason I put this in there is the market is testing you, Warsh. This is a double dog daring you. This is straight schoolyard. The market is fighting with wars and bessent which I think is just great because more importantly I think people should be looking for ways to make money on the linear on the exponential side. But again that's just me. That's why I'm doing these videos because I think you guys are here either to at least know what's happening in that part of the world but also some of you want to make money on it. Uh the
[04:38] third mandate moderate long-term interest rates for the Fed. Just always keep this in mind that Trump clearly has got Bessant to be fighting the market in a way that Trump would fight the market participants and Worsh isn't fighting people but at the same time he was chosen by Bessant and Trump and they don't want rates higher. So we'll see what happens. Uh Besson's upsized buybacks get hit. So he goes to six billion and the market immediately sells off again. Good news hit with bad news.
[05:10] So that means the market's acting badly, but the S&P's up over the two days. Is it talk time to talk about plan B? If this doesn't work, they'll just cut long-term issuance outright. I mean, Besson is trying to be as obnoxious as possible. Um, I am the House now. Underscore his level of engagement on economic policymaking in Japan where he has coordinated with Japan finance minister again why fighting the government is the best trade. He's speaking out against the Bloomberg terminal bros. The arrogance
[05:42] stuff is coming out for the week. The S&P we're holding the 50-day for now. There's nothing much going on. Again, this looks more like a flag formation after a big high. We had this one here, then we went sideways, then we went up again. As I mentioned last week, we're barely up since June 2nd. So unless the economy is going to fall off and earnings are going to fall off, as far as I'm concerned, we can keep doing this for a while. But if you look, we did this rip and then pain here. We did this
[06:14] rip and then pain here. Guys, I showed this with Nvidia. Get used to the fact that when this just rips, it's going to rip for three days just when you're feeling like it has no chance of ripping. Here are the facts. Revisions this week 32 up there. This is the 20week average of revisions. Here's PMI. The leading indicators are doing great. HYG relative to IEF, the proxy that I'll use real time and the one that I care the most about as an
[06:44] overlay rel instead of spreads. Two, the S&P had a huge ripper week. HYG up relative to EF, meaning as bonds are selling off, HY is not. Here it is overlaid with the S&P. If you go back and look at this in 2022 when the Fed was aggressively raising rate hikes as the S&P was going down, this was going down despite rates going higher. So again, the proxy for the equity market with credit completely fine. And as I said last week, bond va's not moving, junk spreads aren't moving. Uh any part
[07:17] the inflation expectations via tips not moving like the market is saying h we don't really care about what's going on. But the aida and the emotion in people is AI thematic portfolio. My portfolio uh breaking above the 50-day today. We'll see if it can get going. Um we're still doing the low V consolidation here, but we've already broken the 50-day on what I consider the most important part, which is my 10 name concentrated one of the 10 names that I feel are the best by different
[07:48] groupings. So, if you guys haven't done it yet, I'd go look at it because it broke above the 50-day and it's getting close to the August highs. The names are all acting well. Bitcoin, well, flag range trading, Ethereum break, we'll see if it can continue. Broke broke back in before today's close. Uh, but again, I've said it before, I think Ethereum should be outperforming Bitcoin when we're actually in the bull market. uh because the ecosystem and the network effect
[08:19] should be the critical thing because the agentic side especially with Astra and a huge week for Ethereum over Bitcoin breaking out again to the highest level since January. Bessent warns nothing else would matter if China wins the AI race. There is no day after tomorrow if China wins this. Why would we fight the AI trade and why would you fight the bond side? Go with what he's saying. you don't have to go buy bonds because that to me is not a good idea. But I do believe fighting the
[08:49] AI trade is a bad idea. So again, do you want to be a linear side of my V or do you want to be the exponential side? You guys make the choice. But again, this is the big news of the week. We've talked about AGI. We've heard about AGI. Jensen Yuang, the godfather of AI, said AGI has arrived with Astra. The moonshot's portfolio covered it well and again 100,000k is what basically Astro was trained on. The next model is four times the size
[09:20] and behind that the Vera Rubins. Why fade this guys? Why? Why? The IQ is going through the roof and we're doing things unthinkable. So it was trained on 100,000. They say research agents are now generating 3.1 workdays for every human workday. They hit the Navier Stokes challenge which mathematicians have been trying to do for a long time by using 10,000 AI agents working in parallel for 88 hours. So we hear about the agents
[09:50] working together on the negative side. This is then solving some of the world's greatest problems which have huge implications for physics and as I'll show huge implications for the PE of the S&P 500 going up five years. Compute can increasingly be converted into intelligence. Intelligence can be multiplied into thousands of agents. So again, more compute, higher intelligence, more intelligence, more agents. We're at the critical point where this is where we were in October of last year. We reached the point of
[10:20] high enough IQ to move into the second side. Those agents can be put at problems like the Navier Stokes which get us closer to curing drug disease at a mass scale solving problems for energy solving all physics pro like we start getting into the things that it's all built on and that's where we are now the agentic side means a big change and that's why I talk about crypto so much it's why I talk about other parts which we'll go through but we need more compute for that and we are nowhere in
[10:51] the compute side so we need a lot of GPUs and that's why GPU prices just refuse to go down. Now the way GDP changes more people plus more capital plus higher productivity. That's what the formula has always been on. But with AI a new possibility shows up. We have the human labor. Same thing up here. We don't need more people though because we now have digital labor and we actually have accelerating productivity because we not only have more of these to replace these but as I showed these work at three to five times this. This is my whole if you just take
[11:22] weekly hours, go through the economy for human beings, you add in the lunch breaks, the smoking breaks, the flirting breaks, and everything else that goes on and the daydreaming and not working. Just go back to um the office and go see how much time was wasted. Agents don't do that. They have no need for that. They also have no emotions from a trading side, which I'll get through. The old constraint on GDP was ultimately the number of humans and the productivity of their time. The new constraint may increasingly become compute, energy, and our ability to deploy intelligence. The reality is GDP
[11:54] productivity already is accelerating. It's seen through profit margins at this point. Forget the economist trying to figure this out. The model of GDP is wrong. You can't say that GDP, the index created in the 1930s to measure industrial production, is the right metric to be using to say productivity is not higher. Think about it. Productivity is an output of GDP. GDP doesn't count IP. G GDP doesn't count lots of things. That's why we change it all the time. It doesn't know what's
[12:25] happening. Navier Stokes importance. I'm not going to take you through all this. The reason I wanted to show this is because Navier Stokes, which I knew nothing about until two days ago, but I spent a lot of time, as I always do, when I hear something that they say is really important from a physics basis because it gets us closer to solving things in energy, transportation, compute, manufacturing, again, efficiency, medicine, climate, and more. But more importantly, I've talked about this, the PE of the market, which is is not here, but I want you to
[12:56] think about it. The economy gets bigger because of science break breakthroughs. So the total economic size gets bigger. More energy, more transportation, more compute, more manufacturing. Go spend time on why that's the case. Again, we get to a point where the TAM gets bigger at the same time that corporate duration gets shorter. This is terminal value. So we've already gone through this episode for software companies. We're kind of going through it already in hardware with
[13:27] Nvidia trading at such a cheap PE. Multiples are going to compress because terminal value, the ability to know when a company will have a cliff of their earnings no longer growing. And that's all it takes. Because once AI disrupts your business and your earnings are no longer growing, where should your multiple be? Should it be one? Where should it be if you're not growing anymore? That'll be a question that's going. And I know human beings won't want to buy anything that's not growing. So you're left in a situation that's there. Now if they're not growing, the
[13:57] only way they want it is the usefulness of the community and things like that. Again, it gets back into crypto. It gets into a mindset change which is happening with younger people. It's never going to happen with older people, but older people will get recycled with the newer people. So this is something important to think about because solving Navier Stokes leads to the ability of having and I'm just going to go through that. You can read all the different parts of what this means. I just want you to know it means better turbines, better aircraft, better cooling systems, better ships, better reactors, better
[14:27] batteries, industrial products, and more. Things just get more efficient. It changes everything. If you want to go read more about the people who know what's coming and then let their economist translate into what's coming, go to the economists who actually see the data. Most economists have no idea what models are coming up. The anthropicus economists do. So go read their economics team. I know everyone's going to read the Goldman Sachs one. I know everyone's going to read the Morgan and Stanley one. I know everyone's going to read all these different ones and they're going to listen to the rate
[14:57] forecast and they're going to go through this and I'm arguing that none of that matters because it's all linear thinking and that is the truth from my side. The truth Greg Brockman from OpenAI spoke. He's a boring guy but if you want to go listen to the interview it's only 20 some odd minutes 24 minutes to be exact. On there he gives a description of what Astra is. So Astra AGI is here. He talks about it. I highlighted the different things you can go. He describes OpenAI's direction as merging chat. That is the one that almost everyone uses at this
[15:27] point with agents. Almost no one uses this with desktop. Almost no one uses this. Codeex, no tools, no applica. All of this in one place. So you can now use Astra in this way. So I've been talking about the fact that Grockbot allows you to use agents. Well, guess what? It's not the same user face. It doesn't look like a fun video game with your little agents running around, but you can use agents in this as they did. And that's where this all changes. And I will get into a little bit of this when
[15:57] we get into a trading system side. Uh image models, everything along those lines in one system. This is the critical point for you, for your children, for everyone who wants to get better, who wants to follow on the path of me spending all of my time on AI, wants to use it for work, wants to use it to build things, wants to have an entrepreneurial mindset, and wants to be empowered. So, if you don't like your job as an accountant, if you don't like your job in any kind of knowledge work, this is the point where you want to start paying attention. Greg Brockman, we're really trying to
[16:27] empower the individual to make it so that you can have superpowers, you can accomplish more. And I think that really trying to benefit people, empower people, build tools that can really help you in your daily life. That's what this thing does. So I did this on the video series. So for those of you who are subscribers who've never watched the video series, if you have access to it, um I did a five six-part video series which is all about develop developing an AI entrepreneurial mindset. For those of you who are not subscribers, this is where I would start is I would get these
[16:58] and basically be in the position and have your kids go through it. I think this is critical especially for 15 year olds and above. Let's say 15 to 24. I also think it's very important for people that don't like their jobs at this point. But anyone who wants to develop a new mindset to ask, to find, to build, to test, to fail, to learn, to adapt, to repeat. This is all what I'm going to show you in the new video, which is about how to build something from scratch and use the tools without needing any coding. I did 90% of the
[17:31] work for the first save, if not 100% of the work, but 90 plus% of the work and AI can do the rest. Um, and I talk about that in the video. It's an hour and a half video. It does a combination of walk you through slides to explain what I'm doing and then go to the place where the prompts all the prompts you're going to get all it's 14 different prompts you will have this thing and you're just opening something that I built that I saved on GitHub. So this will bring empowerment you will feel a sense the same way I did when I started to use AI
[18:01] in a really big way with code two years ago. I immediately felt like I had a superpower. It gave me agency to build my own business which is what I started guys. So the YouTube and all of that came after I took Python and after I was able to manipulate AI in a way through very basic stuff but know that the models were going to get better to where verbally I could just say this and I didn't need to speak as I like to say in the video. Coding to me was Chinese. I could not understand it and so I had to have other people translate for me. I
[18:32] don't need that anymore and it leads to entrepreneurship and you want to be an entrepreneur right now because of that duration of corporate moes that I mentioned. If you work for a company that company will be in trouble. So I would say the majority of you work for companies that need to fire people not because hey we're using AI and our business is thriving. They need to fire people because they're under pressure. And they're under pressure because AI can do these things all the time. And I know that because I'm doing it that way. Astra is the most dangerous AI model
[19:02] right now. I'm getting into something here. These came out and this is what triggered me on a journey on Monday, Labor Day of this week. Yes, Labor Day on a holiday. I was going through, I just worked out and I started on the way back, I saw all these things and I immediately went back and said, "Whoa, I want to build some alpha strategies. This is cool." Now, I've built alpha strategies before, but when I read through these, this was completely different. And so I uploaded these, got the explanation, I uploaded this, I uploaded this Jane Street intern openly
[19:32] shows on Reddit, but all of this stuff to me, got uploaded into chat GPT. And you'll be shocked if you just take my computer screen, take your snipping tool, if you're on a uh a ThinkPad or a Microsoft thing, you know that think that snipping tool. If you don't, just go in search snipping tool, take a a snapshot of it, and then go stick it into chat GPT and say, "Can you rebuild this?" Now, if you don't have Astra, it's not going to work the same way. Astra is part of the bigger package, but
[20:03] I think most people at this point have some access to it. If you don't, you can find out how to get access to it. Astra ran six AI agents on my trading account. Probe was one of them. Prism was one of them. Again, this is why I said this is basically the same thing as Grock. I put $500 in, $5,100 in the pot by the morning. Now, I could care less about these back tests and people making money. I don't care. And that's one of the things I go through. The great thing about a back test is it should fail most of the time. And to build an AI trading
[20:33] desk, the system is designed to reject almost everything. This is what I love most about this ideas found. So, thousands of ideas every day. This is the critical way to getting to something that works is the filtering process. Look at it. Many many ideas 99% rejected. The the balance of the 1% makes it there. Then it gets 90% of those get rejected. By the time you get to the end, you have models. Now again, if you're
[21:04] wondering how, let's say this here is Millennium deciding, and I saw this firsthand with my hedge fund or the hedge fund that I worked at that closed down. Okay, we like this, we don't like this, we want this, we don't want this. They go through and they're looking at ideas and they're looking at data to support it. It's some back test because they can't see the actual people and what went on on a daily basis. They can get daily returns, but they don't see what responded, what happened in their space, whatever the case. They're just trying
[21:34] to minimize the chance. So, they have their own process. Everyone who's figuring out who to hire people to get returns has it there. If you use AI and then you get one and you get two and you get three and you get four and you combine them all together and they're all uncorrelated, you're basically creating a mini multistrat. So, again, for anyone who wants to build not only a trading desk, but wants to build a system, you can do that. You really can because the agents can replace the humans on the trading side and I'm telling you the lack of emotion will be better. They
[22:04] will develop the skill set that's there. I believe for most human beings like I showed that V where everything changed as of the great financial crisis. We're at another kind of point of the exponential cycle. So what you learned over the past 10 years is not going to be as helpful as responding today. That's why I think momentum becomes a dominant factor and why I wanted to go do something my own. So one of the edges that I still think human beings have is on those ideas side what I did is wanted to build a technical system of when to buy things
[22:34] with inside the thematic portfolio. So for those of you who've asked for these things who've looked for trading signals who take the technical side with all of the data I give you and then want to look at the ratings and be like well this thing said buy. No thing doesn't tell you what to do. Yes, it'll give you some number of what it believes is a buy signal without any kind of a really deep algorithm. What I wanted to do is go back test things and just say okay based on what has happened over the past 5 years, 10 years, nine years, what's in
[23:06] trend, what's resetting. So I created what is basically out of trend resetting coiled trending structural trend sign and this is all based on my belief of what's happening in the market. So, I've talked to you guys about market structure changing. I've had lots of conversations with the biggest mutual funds and hedge funds in the world about this. They seem very interested in how market structure is changing. They don't know why in their head, but they understand that it is changing because it's getting harder to make money in a consistent way without getting out of
[23:39] stuff ahead of time. So, it's making it more challenging because I believe what's going on is you're later to things then they get so far overbought they make perfect sense and like I did with the AI trade. I used what had happened because I believe the critical thing is select the hundred stock universe. So, take my hundred names. I believe those hundred names are going to benefit from the AI buildout over the next five years. The infrastructure needs, the compute needs, that is a capex trade. Those are the names that are going to benefit. Now, I think most
[24:11] of them will do what they've done. 80 plus% are above the 200 day moving average with their 200 day moving average pointed upward. Great. I think overall that's going to create alpha. That's a structural bull market. I don't think you're going to get what you saw in the first quarter again. So, like I've said before, I don't think people should expect Micron to go back to to do a 5,000 from here over the next year, another five bagger. Do I think uh 2,000's possible, 1,800, 1,700? Let's say 1,600. We're at a,000 right now. you wouldn't take 60% on the name even though it was a 12 bagger before. Um, I
[24:42] think it's going to be really hard to find those names because like I said, multiple compressions coming. So, I think it's great that the multiples are compressing for Nvidia and Micron. I just don't think you're going to be able to make money on them the oldfashioned way of being in there and they go up all the time. I think most days they'll be either down or flat. Uh, and at the end of the year, maybe 5% of the rally, the days will be the rallies you need to miss. And if you miss those five days, you miss the whole move. So what I did is I said rate of change matters. What I care about is the structural uptrend and then I'm looking for a resetting. This is what the midcycle slowdown was. This
[25:13] is when I got out of micron way before the top. Then after that's happened I want to see coiled on deck. Now to do that I used ATR. That's what started this whole thing. Average true range which I've talked about. Um it's part it's a major part of the model. So's rate of change get coiled. Then there's an ignition. What's the initial point? what treats a breakout and then we're trending again. So, I'm looking for structural bull markets. When things get overdone, I want to basically be out based on rate of change and then I want
[25:43] to wait for them to reset. And I'm looking constantly for ideas within the space that way. I think you guys are going to enjoy it. This is the theory behind it. I'm not going to read you guys because it's just resetting some of the things I said, but the back test looks great. So, I'm I'm going to give you the end result signals on that. At the same time, I did a subscriber video. Like I said, it's basically on the theme of a treasure hunt. Uh, this is everything that comes in it. So, you know, you're going to get code, you're going to get
[26:13] prompts, you're going to get a research log, tools, this because it's all on GitHub. I put the whole thing in GitHub. Yes, you guys don't know what GitHub is. I'm going to show you in a second. I also built in Socrates mode for you. So, again, this is what you're going to be able to do. So, subscribers who are at this level, you're going to get access to it. use it. Here's what GitHub is. If you don't know, and I use that Chinese analogy for my life having data scientists, I needed them to basically be able to go into GitHub, which is a library of code. So, you don't have to reinvent the wheel. I go through this
[26:45] and show you how you can go in there and if you're an accountant, show me aic code accounting so I can be get superpowers in accounting so I can develop my own skill of knowing how to use a AI to be able to do more accounting work for my company without needing to lose my job. How do I use AI to do this? How do I help them? All of these things or how do I go set up my own business? Whatever. But this gives you coding in so many different things. It is not just the finance side, but I
[27:15] use the finance because back testing something I think everyone wants. Everyone wants a treasure hunt. They want to think that they've created renaissance. They've created all this stuff. And believe it or not, guys, you can have a trading system that can make you money going through the thing that you do. And you can keep coming up with new eyes, new ideas, new ideas, new ideas by listen to Stan Duck Miller. Go read a book on a chartered technic technical analysis. Use Demark. I don't really care. Whatever it is, go listen to someone who's made money over time
[27:45] and take the little parts they want and create your own strategy. You can make money in it. And again, it may not be scalable to run it at a multistrat in big size with a billion dollars, but there's plenty of strategies that you can do. And I believe AI can continue to find new strategies at small levels. This is one of the things with tokenization. You can do crypto as well. So, as long as you have good data, you can go through it. So, it starts with a question. We go to GitHub. We had our own idea. We test it, tried to break it. This is actually what happened in the
[28:16] video. And then we learned how to fail. And this is the golden side. You're basically learning how to fail and then restart and go through it again. When you find one that works, this shows you all the ways to kind of go through the promise, go through the progress using AI. It will take you through so many levels that are hard to do or hard to learn on your own. So, I think you guys will enjoy it. I uploaded it this week. Again, for people at Robin Hood, for people at Mumu, for people at Interactive Brokers, this will allow
[28:46] people to do what I have here, come up with their own trading system, and then they can upload it, and you guys can show them how it can be an agentic trading system on a paper trade. So, they can see if it works for whatever period they want. They can then upload 10 more. And that way, they're increasing the probability. If you have 10 different ones running at the same time with $100 in each or $50 in each, well, now you've got kind of a little mini multistrat of different strategies going on at the same time, you're starting to get empowered and you're starting to build something. And more importantly, your brain will start
[29:16] changing because you're developing an entrepreneurial mindset. So again, you'll see me as a pirate in the video. This is a different version. Um, but again, this is what it's going to take you through. This is where the Robin Hood Mumu, if you haven't seen it, again, these companies offer a Gentic side. So, I know the people at Mumu. I've already spoken to them. I want to find ways to get you guys to get empowered on this and do this stuff on your own. Here's the Aentic trading set from Robin Hood.
[29:47] Mumu launches Agentic Investing. They've been doing this really since April 23rd, but I had a meeting with them last week. continue to be more advanced on this in my opinion in terms of trying to do this allowing paper trading going through it giving you the API the data prices so as long as you go in and open an account with them you're going to have access to their data so you don't need to use Yahoo Finance and their data they're telling me is far better than Yahoo Finance you guys can go see on your own and of course Interactive Brokers has had this for some time too this is as of
[30:18] June of 2026 so you will learn you will build you advance and then you can plug it into a trading platform. It's pretty cool. I'm doing it with my stuff and hopefully I'll be able to give you guys a a look at this in the way that they have it from the paper trading side. So, I'm not going to end with this. I will show this for you guys. And again, find a GitHub code project that exists that could help an accountant begin to incorporate AI, how to do their job. First, let me know how many you find. I found at least 13 relevant GitHub projects that could help an accountant
[30:48] start incorporating AI into real-time accounting flows. A few of them are particularly strong for the message you're building. You don't have to wait for someone to build AI for accountants. It's already being built on a public GitHub. So again, don't be an AI victim. Find, build, experiment, learn. Everyone can do this. If you're young and you just started at an accounting firm, I'm telling you, spend the time on the weekend going through this and then see if they will let you use it. At a minimum, you will know where everything is going. Sundar Pachai, if you don't
[31:18] learn how to orchestrate agents now, you'll spend 2027 catching up to people who started today. That is the truth. Microsoft plans data center pushed triple its computing power. I remember people were saying the AI trade was the reason it petered out was because Microsoft's I mean this just never stops. Demand for Astra is really unprecedented. 6.2 million views. They're running out of space for it already. Now, this is an important one. These are the H100s. And I just want to take you back to the
[31:48] Michael Burries, the Jim Chanos's, and all the people that said on the accounting side, this is the thing you'll never get. So for all those guys that said this depreciation is a risk and the reason it's a risk is because there's no chance the those prices will go higher. You have to depreciate those. And saying this is going to last for more than three years is ridiculous, especially when the technology is getting so much better. The technology is getting better. Reubens are far better than the H100s. H100 rental prices are almost at the
[32:18] same level as they were three years ago. Think about that, guys. They're at the same price. We weren't talking about them making it for six years. We were talking about Yeah. How can you build in They're unchanged in three years. So, this again gets into the fact that the demand is insatiable. I'll say it every single time. By now, there's a normal aging process of chips. They should at least be down whatever say 10%. They're unchanged maybe 20%. The normal rate of
[32:49] what people would have thought is down I don't know 40% 50% by this time in three years and they're almost the same level and scarily they've actually gone higher. Every depreciation scale assume assumes a chip to sold lose only loses value. The market is playing paying up for it instead. So its rental price is up 22% on the month. I take you back to Gavin Baker one month ago. His core point was that the market was assuming GPU rental prices would steadily fall. Again, this was a month
[33:19] ago. You see what it's doing this month. Instead, older GPUs were holding their value far better than expected. In some cases, rental prices were actually rising. Exactly what we just saw. What he said was, and this is the critical point, this changes the financing debate. So this is about how much debt the hyperscalers will need. The cash flow estimates for the hyperscalers in cloud in his opinion are way too low because of this situation. If cloud providers generate much more cash from
[33:50] repricing their existing compute. So again, think of this side as you buy tomatoes at $2 and the ability to sell tomatoes has now gone up to $20. That is the compute side. So the compute side is using these GPUs to what you can sell to people because the demand for the end compute is so large when it comes out of the data center that the chips that were bought to go in the data center have not had their prices fall yet. So the
[34:21] margins get really really high because what they get to do is repric the compute when the contracts end. So if they if they bought these at $2 and let's say the compute was at $3 and now compute prices are at $6, but they're still using and the demand for the $2 ones is still there based on the fact that people were willing to pay $3 for them. So they get to reset their compute which was set at $3 to $6. This could remove hundreds of billions of dollars of expected credit needs. So again,
[34:51] whether or not you agree with it, the market is depreciating GPUs, the rental market is not, that's the issue. And so when you look at Nvidia and you look at the hyperscalers, there is a way that this can turn into not needing as much credit and free cash flow going up. And like I've said before, they have tremendous amounts of not only demand, they have cap they they have a capacity shortage and Microsoft just reiterated that this week. Now, Paul Tudtor Jones put out a, you know, a negative piece on
[35:21] AI. This is not the first time he's gone through it, and this is not a negative pace on the technology. He and I have exchanged emails, uh, over the past month. I think his, you know, he's legitimately engaged with people trying to figure out what's going on. And what's happening is shocking. And again, that's how fast things are moving. Greg Jensen warned AI may need to kill people before regulation is kickstarted. If you didn't see Iranian hackers shut down a UK power plant. Boston Scientific striker drop on
[35:52] fallout from cyber attacks. So the cyber attacks were so bad for Boston Scientific that they had to reduce their earnings forecast. Anthropic discloses the fourth AI hacking incident involving Opus 4.6. Anthropic says Iran used its American AI model to target US Navy warships. So, this is not all positive on AI and I believe that this hacking situation will continue. It's one of the reasons why I'm very positive on crypto as well because I think AI leads to not trusting
[36:22] anything and it will not stop the earnings. It will not stock productivity, but it will hurt smaller businesses and smaller companies that can't protect themselves. So, the cyber side is increasing dramatically. I've heard personally many many places uh through conversations about hit uh about hackings in bitcoin related wallets and all kinds of things uh not on the chain so the chain is safe but the apps that are on top for everything have become an
[36:52] issue now way put this out again we're starting to see the AI adopters again earnings grow so I think you're going to keep saying the profit margins go higher and again I showed this last week I'm going to say it Again, the market has never been cheaper or at least in the last 30 plus years with the PEG ratio down here. So, everyone who's negative goes, "Yeah, but this is peak uh earnings." Well, revisions are still going higher and we haven't even got to
[37:22] a point yet where this is falling and profit margins usually start to come over first. So, we're It's not like we're getting anywhere on this. If profit margins are going higher, guys, I don't know how if expenses don't need to go higher. I I like again we're at the productivity boom point. So I wrote this paper this week. We're going to move into the crypto section again very soon. You guys are seeing this now. About 30% is now crypto because this is in a bull market and this is where I want to focus my attention. So I wrote this on what I believe is the most important conversion
[37:52] in the world. The Robin Hood convergent. Robin Hood is basically the epicenter of AI crypto macro nexus. It is a trady company that has been somewhat involved in crypto but not making a ton of money on it relative to its other businesses. And it's an agentic company. So you guys should read this for the subscribers who got it. Uh and for the institutions that are getting this uh but won't be at some
[38:22] point soon unless you're subscribers. Um, again, I'm starting to go through this in in a very big way, and you should go read about Robin Hood because it is your entree into understanding how to start thinking about it from an earnings perspective. The chain is now earnings. Bernstein sees 31% upset for Robin Hood as fees top Salana. Robin Hood chain fees record 6 million in daily fees. Robin Hood chain has only been live for two months and it's already starting to rerate the onchain market. Arbitum, which is basically the
[38:53] app on top of it, surges 7.87% on Robin Hood chain. Boom. Uh, Arbitron. What are they? Oh, see, here's the news. You can go see it. Now, if you want to go see if Tradfi is covering it, Arbitron Wall Street Journal did not match any news results. Shocking. You're not going to learn anything from anyone unless you come to me uh from the Trady perspective. You can go to CoinDesk. You can go to crypto which are great places for this but I'm looking at it from the angle of merging the three of them tradi
[39:24] plus AI plus crypto which I think is critical especially for Robin Hood the AI agent site is the most important side uh if you didn't see they strike a deal with crypto.com the merge is continuing you want to be long stuff that looks like this is the pers like I said last week if you don't know what pers are please go pay attention tokenized stocks we continue to move forward on this. Robin Hood spoke at the Goldman Sachs technology conference
[39:55] and here's the five big takeaways. Aentic trading is going mainstream. This is from Robin Hood. So this is from Vlad, a hedge fund in your pocket. Well, I'm showing you guys how to do that. Robin Hood chain is about global financial infrastructure. Yes, it is tokenization. I believe if you have an advantage, if you get into tokenization ahead of everyone on the Trady side as they start to jump into this, which they will because they will need to make alpha, you're getting in front of the hordes of money if the Clarity Act goes through. And it's not a given that it won't go
[40:25] through and I'll be in DC in two weeks. I'm hopeful that by then they've already solved this, but the Clarity Act has huge implications from a spike perspective for this and from an institution of credibility with pension funds and things like that. So there's catalysts coming up. Robin Hood wants the entire financial relationship. Trading, crypto, retirement, banking. This is really critical, guys. Prediction markets are a gateway product. The bigger message. Robin Hood is evolving from a brokerage app to AI native tokenized financial platform bringing it sounds a lot like Apple to
[40:56] me. NASDAQ invests 100 million in Kraken, eyeing the 2027 launch of tokenized stock trading. Everyone's preparing. US Bank test stable coins. Visa stable coin card business is exploding. European banks plan to launch a euro stable coin on Ethereum. This is all this week, guys. Coinbase says clear US crypto rules are coming with or without Senate approval. Brian Armstrong said the Clarity Act is ready to get a yes vote. Patrick Wit said the same thing. The probabilities for this year
[41:26] of getting it through are still below 20%. But that doesn't mean they can't get a yes vote. So, this is where I'm going to bring you to the final part. William Mogayier gave a TED talk. I think this was 2014, 2015, could be 2016. Um, no, no offense to William or Bill. He's very boring. I love finding boring, smart people. Um, because that means people don't listen to him. I highly recommend everyone goes listen this in TED talks. It's as usual 17 minutes.
[41:57] This is again a decade ago. him talking about Ethereum and the important of Ethereum and the importance of the blockchain. He did this a long time ago. Give me an assessment on how correct his vision was in what's happening today. He was 8 to 10 on direction but closer to 510 on specific information which I think both of those is good to get something especially when when crypto has had so much regulatory problems to actually have an ability of guessing where things were going to be and to get an eight out of 10 on direction, five out of 10 on specific is still good. where he was too
[42:27] aggressive was removing the intermediary and this is Robin is a perfect example. I think this is the way to go through it. He didn't know about stable coins. The biggest thing he didn't see was AI agents and this ironically they may finally make his thesis work. That's the whole point. When you listen to this, I want you to connect AI agents to it. This is the way I bring you guys alpha. This is the way that you learn about this and you get the connection of AI agents because everything that he said was assuming humans would use this. It's
[42:58] not for humans. This is what his original vision was. He was a peer-to-peer guy. He was a Napster guy. He talks at item. I'm going to go through some other things because he gives the best talk on valuation for crypto. And I've been asked by so many people inside and outside of crypto how to value it. What happened instead? said the middleman stayed in but AI and crypto merge is changing it. So the AI side is different from this. It is critical for the success. It speeds things up. So here's the other one. This
[43:28] is now about 10 months ago. Ethereum's endgame. How to value Ethereum. He talks about his experience in the 1990s and that's critical to this. Now, I don't know a lot of tech people from the 1990s who are still involved in the market. They either made too much money, they blew up, and they moved to something else. It's been a young person's game since the iPhone came in for the most part. But there are people that were around. And his point is, we
[43:58] don't know how we we don't know how to value Ethereum. It's a new model the same way there was a new model when we had to figure out how to value the internet. And at first, it was eyeballs. Had nothing to do with earnings. This will get into my experience with Amazon as we go through. Then revenue finally started to matter in here. Then SAS metrics came, but that's a long time. That's 25 years before the metrics came up. There was no SAS metrics. And that's his point. So here they are. Now it's about other things. So with
[44:29] Ethereum and for all of crypto, there will be revenue. There are network effects. There's flow of money. And this is what I talk about all the time. There's ecosystem value, there's productive assets, there's so many things and it's programmable. You have to go listen to this both podcasts. Just listen to it. If you don't believe at that point, it's fine. But then I want you to go back to Amazon in 1997 when it was a lowly online bookstore to where it is now. Not now. This is 2013. The reason I'll tell you
[45:00] why I put this up there. So from here to here, 27,000% Ethereum right now from its lows is up about 3,000%. I really do believe we're going to see a similar number at a similar time where someone who's in Tradfi is finally going to give up and say, "I don't understand how to value this thing." I didn't understand how to value Amazon at this point. I was a macro person focused on oil. I had made most of my money trading on being either bearish or bullish oil
[45:32] being bearish or bullish Brazil, China, whatever. That was the places that I made money. My best year was 2008 and that was being short the US financial system being what I was called as a perma bear and being along emerging markets and along commodities in a super cycle. I didn't understand this. And so when this went back during the great financial crisis, I figured for sure this would go to zero because it didn't make money. Well, it not only didn't, but it came back faster. And when it broke out, every single time it looked
[46:02] head and shoulders, I can remember this one. I showed it recently. It was breaking down during the EU crisis. Went right back up. Guys, its PE never ever was below 100. It didn't even have a PE for most of this. So, I went out to Silicon Valley and say, "How can this work?" I went to Singularity University. That's when exponential started for me. That's when my mind changed. That is 12 years ago. That is why I can talk about crypto. That's why I'm using AI because I know you're fading the inevitable. When did Amazon finally reach a point of
[46:32] having a PE Tada. Below 100 for at least two quarters. Not till 2017, 2018. It wasn't until 2018 that it finally broke below for two quarters in a row. Again, most of the time before that, it didn't even have earnings. You had to measure it on eyeballs. Yes, truthful. So, when you're trying to figure out how to value crypto by the time it gets to the point where it's no longer a ridiculous price, you've already lost 27,000%
[47:03] in 2013. I never wanted to make that mistake again. For you mutual fund people, for you people on the hedge fund world, don't make the same mistake with crypto if you still want to be involved in markets from a bull market perspective. That's my take on the situation. He wrote a he has a new book coming out, Trust Shift. I believe trust is worth a significant amount of money and I believe Ethereum is the trust model. So you guys can just go through it if you listen to his two things. Love to hear from you, but I would do that.
[47:33] AI makes trust an economic asset. I'm 100% sure that the blockchain needs every real world asset to go on there to keep its value. Trust is becoming an issue. It is absolutely something to measure things on. And anyone who wants trust, it's going to have to go on Ethereum. It has to final two slides. I will be in DC. This is the presentation I'm giving to a variety of people in DC, not just crypto people, congressmen, the whole thing. The time mismatch. So if you want to join me, Freedom Tech DC
[48:03] 2026, September 23rd, 22nd, and 23rd, I will be down there Monday night, Tuesday night, Wednesday night, coming back Thursday. Happy to meet up with people that are there in DC if I have the time. The Bitcoin Policy Institute's been great on this. I will be a pub key with them. This is one of the slides I will be showing which again this is tokenization. This is the bridge between the digital world over here and again the linear V, the exponential V, the linear V. If you want to stay over here,
[48:34] it's great. But this tokenized world is coming. You're going to cross the bridge at some point. Why not take a little mini jump over here and start spending your time here? And the Wall Street Journal had Bitcoin's overlooked promise, freedom. Again, for everyone who lost someone or was affected by 911. Uh, I'm thinking of you. It's a 24 25 year anniversary of an emotional day for me. It tires me whenever this day comes up to think about. I am grateful for knowing Shawn and I'm grateful for you guys watching
[49:04] me every week. Uh, hit the subscribe button, reach out, do whatever, but most importantly, hug the ones you love today and I'll see you guys next week.