Transcription
Um, shortened week, but as has been the case all year, uh, a lot going on, particularly uh, with regards to AI and the software side. So, um, I just want to bring this up at the beginning. I am, uh, I want to make sure people hear this. My, my overall views for the year are that somehow or another, the market will finish the year up. Nowhere near as much as you would expect if you thought GDP would be good, profit margins would expand, and earnings would be good. Uh, this has a lot to do with my views, which I'll go through today, on multiple compression for anything related to software and then the difficulties that come with people just dealing with cyclical items, uh, on the hardware side.
So, I'm going to spend a lot of time on SAS again, um, because I want this to be kind of the last time. I, I don't like getting involved in this. I think this is, uh, hurting people by trying to buy into something that is clearly going through an AI disruption that I understand most people don't spend enough time on AI to fully grasp the concept of bespoke. Um, and that is what's happening. Entrepreneurs are rising rapidly, and enterprises are what people are focused on for revenues. But that is not the right way to go through it, in my opinion. Uh, I'll go through that. I'm going to go through the dispersion side again. Uh, I'll talk a little bit about the videos and the stuff that is on the paywall. Uh, I'm going to try to help all of you. Uh, but particularly people that have reached out on the trading side. This is going to be a different environment. Uh, Kovar, as you guys are learning, in terms of covariance matrix, I do not see any change happening. I think this is a new regime. Uh, not only do I think it is, I think as I go through this, hopefully some of you will will see why. Um, we'll go through Bitcoin, where things stand with that. When is it going to break away from the SAS side? The recap, the SAS debate, all that you see how much is related to SAS, you just can't get away from it. It's the gift that keeps on giving.
So, um, I did put this out, uh, on Friday, uh, to literally go through. But I just want to read you something because this, you know, I, I write things, especially now with the paywall. A lot of this is for the people, uh, the individuals who are out there who are trading the market and are are curious as to what's going on, but obviously for asset managers as well. And in most of the institutional conversations I've had this week, the question has been, "When does this stop? My VA has gone up. When are we going to get back to a more normal point?" And what I wanted to basically get through here is that the dispersion is massive. Uh, as of, uh, yesterday on Thursday, there were a total of 185 stocks in the S&P with a greater than 15% absolute performance year to date. Last year on this day, there were only 81. Um, basically in both directions, we're getting massive movements. I go through in that paper why, in my opinion, this is happening and why this is it, um, for the next 5 years, 10 years, whatever case you guys want, uh, this is it.
So if you're trying to pick a bottom in software and you think that all of a sudden V's going to come down with inside the market, just know my opinion is no chance. Uh, and I don't like saying no chance on anything, but basically AI will be a disruptive force, and behind AI is humanoids. So I, I don't know how people don't understand that you can't forecast things out three years. I've talked about this. The speed is there. So the software side, this is basically a chart of the S&P overlaid with the PE of the software index, just to show how much we're breaking away. The software side has only gone down when the S&P has gone down in terms of people's viewpoint of it. This is a structural shift. Um, and as I'm making the argument, and I wrote in my Substack this week, this is just the beginning of a bigger story.
So, if you're trying to pick the bottom of SAS, in my opinion, like I've said before, you are an AI bubbleista. You still believe that this is a bubble, you do not understand that this is the year that there will be billions of Einsteins doing work while you're sitting there not believing in AI. If that is who you are, you are going to have a hard time adapting. Um, and here's what you're going to be if you have a job in 5 years. This is what you're going to be in the elevator with is effectively humanoids in with you. This is not going to stop. And that's why I will just keep showing this over and over again. This is a supersonic tsunami. It is speeding up. It is going at light speed. And it is disrupting anything in its path that is not based on scarcity.
So, I'm going to go remind you that the physical world upgrade, the paper I wrote at the beginning of the year, the first phase of AI investment, 2022 to 2025, was narrowly concentrated in the digital brain builders, hyperscalers, GPUs, and core networking, producing enormous capital intensity. I could have included SAS in there, but limited economic diffusion. As a result, traditional manufacturing indicators like PMI remain subdued despite record AI capex. The timeline for this transition stretches not quarters or even years, but decades. The investment sides of this transition are not a passing trade. The road to artificial intelligence is expected to take at least four to five years. And what follows is an era of embodied intelligence driven by robots and blah, blah, blah. Um, not a speculative bubble. Positioning has been heavily skewed towards services and software for over a decade. But the AI narrative expands beyond code into physical infrastructure, commodities, industrials, and hardware. They are poised to become the structural winners of this new era. I'm just reminding you to go back and read the paper. For those of you just joined the paywall, even the old papers that I put up there, I would go read. They have shelf life. These are not things that I'm writing to say the S&P is here. These are all related to AI, and I'm telling you, out of the many, many people I talk to on a weekly basis, which seems to grow, I still put it at about 20% I can have a conversation with at all regarding AI where they know enough, in my opinion, to be making decisions on what's going to happen over the course of the next 3 years. It's not a slight against them. They don't have the time or maybe they don't have the access. This is going at light speed. So I, I just leave it there.
The regime shift from concentration to dispersion. So number one, this is a regime shift. This is not a cycle. Number two, asset light is giving way to asset heavy. Buying the losers, meaning the SAS names. That is not adaptation. You are basically trying to play the game of the prior decade, believing that software is going to win in a world where you can create software in seconds. I, I just don't understand how people want to have this argument. This is a debate that is feudal. It is ridiculous. It's like debating religion. I think it is wasting people's time. Just get the hell out of the way of AI. If SAS does have a bottom, it's not going to be all of them. If you can pick some, I wrote something positive on Palunteer relative to Microsoft because I don't want to be involved in it. I have no interest whatsoever. Expect persistent internal volatility. No systemic risk, but structural messiness. This is the whole point. I think there could be multiple 10 plus percent corrections in the S&P, um, driven by hackings that are going to happen, uh, because of the agent swarms, driven by deleveraging, as I've talked about, as asset managers have to deal with leverage coming down because Kovar stays at higher levels. I think it's going to be messy, and I think you have to be ready for it.
Um, I wrote this paper, uh, a while ago. I want to go into this a little bit for Bitcoin at the beginning here because I want to remind people that I, again, I didn't expect SAS to go down this fast, but Bitcoin is related to SAS, and I don't think it can break away until we get the Mag 7, the hyperscalers, and anything built on code to go down so that there's nothing that people can invest in with certainty. Bitcoin does not get disrupted by AI. AI agents help move the transition from the physical economy to the digital economy. So, I'm going to tell everyone out there that's looking for things to come in SAS, this is where you go. When SAS finds a bottom eventually, it will, in my opinion, be Bitcoin that leads the way higher. And when it's the only growth asset that's working in the forever period of AI disrupting our lives, you're going to jump on board. That's the thing I know that will happen. Uh, it's just not happening now, but I would be spending your time on Bitcoin. I'm happy to go through that.
Raul Pal and Andreas Steno Larson had a conversation. Um, obviously I know these guys. I, I was at an event with them at Raul's event recently. Uh, I have tremendous respect for for Raul and and don't know Andreas well, but I've read a lot of his stuff. Uh, and I think overall we're on the same page, except for something important. So if you get a chance, go listen to this. I am a deflationista, which I think Raul is too, but this headline, "Why the NASDAQ could surge again." I do not believe that. And this is really important in this side. I don't think when I hear what people are saying and what these guys said of where the thread is. Yes, there's no bubble. I don't believe there's a bubble. Yes, you're spending money at a free cash flow for capex in the hope and the belief that you'll get the revenues. I do not believe the revenues will be there at the level that both of think. Um, and I think they need to think that because they've made this point in their head that you need liquidity and to some degree risk assets. And I don't want to speak for them, but I have a very different view. I think the NASDAQ and technology has to underperform for Bitcoin to do well, and I do not believe that people are spending enough time on the deflationary situation that is going on with the AI models, and I will get through that in this. It's very important to me to go through. But this is the way I've broken down what happens for Bitcoin. Let me just move this so that you can see all of them.
Phase one, that's where we left. This is the acceleration and the liquidity offset. This is the Jeff Booth theme. This is the Lynn Alden and Sam Callahan theme. Uh, which is that Bitcoin trades like a risk asset. It is highly correlated to liquidity, which Raul believes in as well. Um, he also believes that there's a deflationary shock from exponential innovation. The difference is, I don't think the government can now spend enough money to get us out of this. I don't think it's possible. I think this is incredibly deflationary when you get to the intelligence level and you have hyper competition. I have always believed the way that Bitcoin works is for the bottom end of the K of 8 billion people versus the 10,000 people with all the money in the world, there needs to be disruption to public equities. I've done this on this video. I've done it in writings. I've done it in interviews. I am a believer that all public companies filled with bloat, filled with software stacks that everyone is focused on, is a game that doesn't exist 10 years from now, or at least if it does, it's in the last legs of it. And that during this time of the capital structure changing and everything else, this is a very dystopian period, which is why it's very dark. The scarcity repricing, which is gold, which is copper, which is all of these things that are scarce. That is where we're going to end up. But right now, we're in this one with people leaning towards the more obvious ones here and the ones that are driven by, you guys don't want to see that, um, driven by the money going in spend by the hyperscalers. So I believe that deflation is going to cost the hyperscalers.
Now, to get your mindset in this, I'm going to again say, so I'm doing a video series on AI. I think one of the things you guys have to understand, the reason that I do the YouTubes is to hopefully give you some stuff that you can't get from the banks, that you can't get from X. Hopefully, I'm a stable-minded person, but I'm spending my time in AI all day long. I'm talking about what I'm doing. And these videos are meant to help not only you, your employees, but also your kids. I don't think unless you develop an AI mindset, you put in the reps like golf, you think in bets, and you build a relationship with AI, that you can possibly understand how dystopian this is going to be. That is the reason why the research is on there, the videos are on there. Thank you for all of you who've joined on. I continue to believe if you go on, I'm going to help you from the investment side in terms of giving you ideas, but I'm also going to help you in terms of the regimes that we're going to remain in as AI dominates, and then hopefully give some things to where you can find some places to make money when we come out of this in the scarcity side.
Bill Gurley had a great post that led to something important. Some shocked that Walmart and Costco have higher multiples than software companies, when so much value is in terminal value. The haunting question is, will this company be around in 30 years? So he's referencing Michael Mobius's paper. So I basically went to Mobius's paper that he showed in the links and basically brought up, "Give me a summary of this and connect it back to the uncertainty AI brings to long-duration assets and what should happen to valuations." And this is, uh, Michael Mobius back in 1997. "A stock price is determined by three dynamic expectations: cash flow, risk, and time. If a company's cap remains consistent as years pass, it effectively creates bonus excess returns for shareholders." How does AI disrupt this? Mobius notes that cap is determined by the rate of industry change and the barriers to entry. AI acts as a double-edged sword. When uncertainty rises, volatility follows because the market quickly prices in worst-case scenarios. What should happen to valuations? Based on the cap framework, we should expect a bifurcation in valuations, guys. That's what we're seeing with the dispersion. So, you can disagree with it. You can bet against it. But what this is saying is, if you connect the fact that AI is disrupting whether a business will be in place 3 years from now, forget 30 years ago or 30 years from now. For companies that are based on future cash flows, this is a really important part. I would go read the paper, go spend time on cap and go through it.
Now, I also talked about concentration. I just want to make sure people see this. This is the reason why I think people should be very worried, and a deleveraging thing should be a major issue. All of these periods in time, the dot-com bubble is this. The one thing about both of them, they saw deleveraging and they saw massive rotation. I think you have to be very careful when concentration is this big. I don't think this ends in a horrible scenario for the economy. I think it could, if the deleveraging happens too fast, but I still think this is not a systemic issue. But as I go through this for credit and some of the other things that I mentioned last week, there is a risk of it if, uh, regulators and governments, uh, don't get involved if it goes too sloppy. I think if software can stabilize, which has not happened yet, uh, it can, uh, this is, uh, another thing that went through, and I think you should look at this. This fits in again, and I want you to look at these periods. This is the new NS, uh, I'm sorry, a high-low logic index, which is something created in the 1970s, uh, which is basically taking the the ratio of these two to some degree, and when it gets to very high, uh, levels, you end up in a scenario. This is not just high-low, this is high-low relative to the amount of names making high-low. You end up with a scenario that it means instability, and the max drawdown from top to bottom on this can be fairly extensive. The main point is, I think dispersion creates instability. Instability eventually leads to a drawdown. I still think that's a likely scenario a couple times this year. At a minimum, I think it is going to be very hard for stocks to go higher if we're seeing multiple compression in tech names, which I do believe is going to be the case. Anything built on code.
Crazy market stat. The S&P 500 is flat year to date, but there are just 94 stocks in the index up or down less than 5%. So, you've just been all over the place in terms of the the movements. Here's, uh, Scott Rubner, put this out from Citadel. This is another way to go through it. Um, just a dispersion side. And what I want you to do is just look at the time periods when we've had this kind of dispersion. We didn't even get this in 2020. And I think it's because it wasn't a rotation. It was a synchronized just unwind, except for some of the software names. You really only have two times going back to in the last 30 years. One was during the Great Financial Crisis. The other one was during the dot-com bubble and during LTCM. What I want to highlight is, if you match up these periods, in every case, credit spreads went higher. That's the only thing we're missing at this point. You're going to see me focus a lot more on credit as part of the deleveraging, 'cause that's the thing I'm watching. You want convexity in your portfolio. Uh, if you agree that Kovar is going to stay high, which means you want VIX calls, you want, um, anything on the high yield in terms of CDX side, you want to find convex trades where you're making money. Not if the market falls, because I don't think the S&P can fall with strong earnings and GDP, with strong GDP and profit margins sitting up, but I think it can fall 15% in a week and then rally back over the course of the next month.
AI disruptions in earnings. So, we're coming out of the earnings season. This is not going away. AI will be a disruptive force. Um, it's compressing timelines, changing the nature of work. Uh, Opus 4.5's debut that changed the world. This is from Jeff Seabird, and I'm just bringing this up again because he's literally talking about the fact that the world has changed forever. And so, if you take what I've written and you keep going through, um, this is why this is happening now. This is why this started in October for the most part. Uh, but then for SAS, I would say it really got bad post the November period. Here's the S&P. We are just going sideways. Uh, it looks a little roundy to me. I think it needs to break out, but even if it does break out, I, I think we're going to continue to see multiple compression, uh, in the better part.
Now, for the month, again, tremendous dispersion. Industrials, materials, utilities, energy. See this? These are all the sectors I would say that are the data center buildout and the things necessary for the $650 billion. What do you have down here? Tech, consumer discretionary with Amazon, Tesla, and this. So, you got the MAG 7 down here. This is the bad one. This is the one that worries me. Financials down. Financials should be up here if this is a PMI related boom. What is going on in financials should be worrisome to people, particularly when you have utilities here. There is a 9% spread. Utilities are outperforming financials in in a month where, or so far, where you would expect, uh, it to be the other way when you see those other sectors up. I showed you before the software side, we're seeing multiple compression in the S&P as we go sideways, earnings are good, you are starting to see this compression go. IGV, uh, here it is, quarter to date. Now, it was down again this week. Uh, we are now at the worst quarter. There's only two time, two, three quarters altogether, but two periods. It was worse during the Great Financial Crisis, and it was worse during O2 into O3. Now, in both of these cases, the market was down. Right now, the S&P is flat for the year. So this is not a relative thing. This is absolute. So if you did it on absolute, we're obviously in worst case than that. Software just cannot catch a bid. It did not catch a bid this week, and we had a new gutting on Friday. Cyber stocks slide as Anthropic. So now, now we're hurting something where the multiples are still very high because it seemed like AI and agent swarms would be the perfect place to have these names do well. Well, they were thrown under the bus. Um, IT services, this is the consultants. So this is Gartner. This is, uh, Accenture. Massive fall, uh, so far this month, down 17%. Only one month worse in the last 22 years. So, even the IT consultants are in there.
Here's a 200-day moving average, the yellow line of the S&P 1500 software index. I don't know why anyone wants to pick a bottom of this. This is not a friendly chart. But I will tell you this, uh, we have not closed below there except for one day here, one week, uh, sorry, one month or one week there, one week there, one week there, and then we bounced immediately back up. This goes all the way back almost until the iPhone launch, which is when this went up. 200-week moving average. Go back and look at it over time. It is a good judge for whether you're in a real true bare market or not. So software, uh, at best is going to be a value trap, but I think we are going to break. Um, another group within the financials got hurt again this week. Blue Owl blew up. Um, this is the, uh, this is my equal weight, uh, index of the private equity firms, including Apollo and Blackstone, Aries. You can just see, I mean, it's been down six weeks in a row right now as we are spreading into private credit and private equity. Uh, I mentioned utilities and financials. This is a chart of utilities outperforming financials for the quarter. And you can see, aside from these horrible periods, when almost every time we've seen a number this big, you have credit spreads, junk spreads. So this is junk spreads widening out. Uh, right now they're tight. Again, I think if you want convexity with what's happening, I think you want that stuff.
Um, a place again, since everyone is trying to pick the top, top of semis and trying to pick the bottom in IGV. I will reiterate this package thing that I created, which is an equal weight of the names that benefit from the move to edge devices that are more sensitive. A lot of analog names, um, if you want the names for the people on the paywall, you have access to them. Uh, for the clients, they've already gotten this, but, um, this looks like it's going to break out. And here is the report that I wrote on it at the end of the year. I still think these are the names that you want to be involved in.
Um, one data point that matters, this is the durable goods. I've shown this before. It ended up being right. I showed you guys this all starting from the summertime that the capital goods were already breaking higher. This is what measures the data center expenditures. So, we got our PMI lift up, catch playing catchup. We are headed towards 60, guys. It is going to happen. 30-year JGB yields, remember those? Panic move. Um, largest monthly decline now in 30-year JGB yields since 2016. We are headed for a deflationary spiral, guys. Uh, I do not know how this guy continues to put stuff in X. Um, I don't believe in software narrative, and I think there will be a bounce. High gross leverage in semis versus software will be reduced soon. That's all I need to know to be long semis and short software still. Um, you're still going back. Um, everyone looks for a bullish argument. Here's one that came out on software. Is it completely dead? At least it's logical. They go through the things that have killed it. There's 10 modes. These are the ones that are there. Blah, blah, blah. I really don't care. You guys can go read it on your own. I bring that to you so you can look.
Um, Nick Evans, who I've met, um, fund beating 99% of its peers, sees few software firms surviving AI. I'm with Nick on this. I see few surviving in the long term for the next 3 years. Yes. Uh, but that has more to do with the, um, the addiction that the enterprises have with this stuff. Gradually, as the agent swarms destroy everything or replace everything, you will be dealing with, uh, companies that will have access to a billion Einsteins to do their work, and they won't need to have this stuff that, uh, people call software.
Um, Sultan, who I know very well, wrote this piece saying software will be free. Again, I'm giving you both sides of the debate. Um, Dennis and the team put out a great thing here. If you're still buying software, um, I wouldn't. Um, if you want something quantitatively to show it, and you guys should be getting these guys, uh, are all over the AI side from a data perspective. Uh, software has seen the largest decline in positive EPS revisions. That's a bad sign. Uh, again, back to the Jeff Seabird in terms of what's going on. The world changed. Fiverr. Um, this is a note they put out in September. I remember it. If you guys haven't used Fiverr, it's an outsourced place. They have employees in Pakistan and India and various places around the world where you can basically pay them, uh, local money, uh, to do work for you tech-wise. And so revealed the restructure will allow the company to go AI first. So here's a company that I fully expected was being quick to this. They'd be fine. And this week, their stock price collapsed as the freelancer platform tries to put a positive spin on AI disruption. Nobody can hide, even if they're admitting it and trying to get ahead of it.
The uncertainty of long-duration assets. Again, this gets into the Mobius thing. And I just want to bring this up as we go through it because the valuations fit in with it as to what's happening. Private software companies release earnings early to calm AI fears. You didn't calm them. This was earlier in the week. But at least they're trying everything. Uh, Carly on BlackRock, the private, they tried to do stuff too. They're buying software loans to boost CLO profits. Not going to work. Uh, this was a story about Evercore saying that, uh, so far, there have been exactly three insider transactions in large-cap software. If they're not buying, why should anyone else buy? The SAS apocalypse is a credit event, not a tech story. Uh, this is where everyone should be paying more attention because it has destroyed a lot of equity, but it is also testing the debt world, and this is where the Blue Owl permanently halts redemptions. Uh, they came out and said they're not permanently halting anything, but whatever the case is, uh, there's a whole big story here that continues to happen. I think this has to do more with we're getting credit problems that are just spreading. They didn't just start here. Here's the private equity companies. This fall here was related to the auto subprime auto lending, which was Tricolor. Uh, this has been a problem.
So private equity, private credit, VC, as I did with the Mobius things, I am incredibly negative on all long-duration assets. If it's illiquid, I'm even more negative. These companies are trapped in things that are longer, uh, that take longer to get through, and they've got money trapped up. I think they should be trading at a discount. Not necessarily the companies, but at least the way people view them for endowments, for foundations, for pension funds, for, uh, sovereign wealth funds, anyone that has a lot of long-duration assets. The hyper-competitiveness, regardless of whether it's VC or anything else. I just don't know how you can value things when you're talking about humanoids in an elevator 5 years from now, and when you're talking about 3 years from now having certainly recursive learning in AGI. We're already at a point where it's moving too fast.
So, um, private credit should worry about a singularity in software debt. If losses spike too quickly, they could lead to a widespread crunch in a tail outcome. The knock-on effects for all issuers attempting to access credit, including hyperscalers, could be significant, potentially undercutting capital spending investment plans. This is where the negative thing that I think will be a story this year. I would be very surprised if at some point we don't have credit fears that rise to the level that the hyperscalers, quote unquote, would have to cancel capex, or this will be an issue. So if you think about what's happening with Blue Owl, with, uh, their fact, and again, it's just a fear side of AI, but the reality is it's because you have no certainty in the future. This is not a linear economy anymore. This is not one with cycles. This is something where the uncertainty will be here forever. It will never get better. Which is why I show it as a supersonic tsunami. The speed is going. The competition is coming every day from entrepreneurs that are eating away at revenues. They're eating away at growth rates. And it's bespoke. You customize things for what you want, which is really the way software should be.
Here is the chart. This is the tech sector OAS, the option-adjusted spread in the bond market, the high yield side. So this chart goes back 14 years, and it's overlaid with CDX for high yield. So CDX is still sitting down here, guys. Um, the tech side is up here. And the reason CDX is down here is because the other industries are not growing. I think tech matters, and I think contagion is an issue. Now, here is that same chart in terms of the tech side. Here is again, you're starting to get the OAS on the overall side working out. It was still at lows here, but this next leg in 26 has got it going. And again, this is junk spreads down here. You can already do them, the correlations in your head. This is what I did at Morgan Stanley. I traded emerging markets throughout the '90s. I would look at these, and I believe once credit starts to go, you end up in an issue very quickly because it's the fear of getting paid out. Now, these charts matter all of a sudden because we didn't have anything happening at all in high yield, but now we have something on a big sector. We've got credit card delinquencies up here. We've got auto loan delinquencies, the subprime side. Remember, this is what got us into trouble with Tricolor and everything else. Carvana and everything else. Uh, student loan delinquencies.
Walmart had their earnings. They cited a hiring recession. How are things going to get better if AI is now going to disrupt the job market more than it already has? Several indicators are showing warning signs, including a rise in US consumer debt and delinquency. Americans are growing more concerned about their employment prospects. Unprecedented jobless boom. And again, we had 130,000 job creation last month and 137 positive healthcare jobs. So again, that's not really helping disrupt or argue against the AI side. Uh, CNBS, remember, uh, Silicon Valley Bank when this all kicked in right here? Well, this isn't fixed yet. Um, all-time high. Is AI going to help this or hurt it? I see things being printed about how AI is going to create more jobs and people are going to go into offices. God, good luck with that one. Lenders are not paying, or lenders to the owners, they're not paying. Um, money managers are now worried that companies are overinvesting. This is where you start to get into the hyperscalers. And I just want to bring this up, and these next three should at least remind you of why this will be a story.
Dario Modi perfectly explains why Oracle and OpenAI are in a very risky position. Oracle and OpenAI. He says that you can commit to buying 5 trillion worth of compute if you're projecting 1 trillion in revenue rate. If you achieve 800 billion, it's still, it'll still be a staggering number, but the company will go bankrupt. This is the risk that they are facing. And Oracle's involved in. Now, I want to bring this up because Anthropic is winning, and on January 22nd, this was the news story. Anthropic's gross margin guidance as revenue is surging fell by a 23% spike in inference costs. These figures hold significant implications for investors. I just want to remind you, Anthropic gross margins are now being forecast down to 40%. This is the winner. I'm only bringing this up because the cost of the compute is going higher, which makes the cloud cost up there. We've already got delays on the cloud because the data centers are being delayed through bottlenecks and through the government, which I'll show more of. The spending is happening, and here's the issue. China is competing with models that are anywhere from 80 to 90% cheaper. We are in a deflationary spiral. So when you spend money on capex for out of free cash flow or using debt as OpenAI is, or borrowing, going out and taking tons of money, it's about the revenues, and there might be orders, but if you can't get the orders, or if they're delayed, that's a problem. But if the cost of your compute can't keep up with the inference cost or the buildout cost, then you're really in trouble. You got to do a hell of a lot more volume, and that's really hard to get when there isn't enough token usage. So, you got to charge more money, and if you charge too much money, there's people offering stuff already at a 90% discount. I don't know how you get out of this.
Oracle def, lawsuit defendants failed to disclose to investors that one Oracle's AI infrastructure strategy would result in massive increase in capex without equivalent near-term growth in revenue. There's a lawsuit going on specifically on this. Oracle stock was down big on Friday. And remember, this is their CDS. Here's the chart of the hyperscalers relative to the S&P overlaid with the private equity companies.
All right. Data center delays. 125% surge in data center opposition. Moratorium nation getting cancellations. AI even in XAI's Colossus. You can go through and see a notice of intent to sue for violations, folding, uh, there on costs, on pollution, on everything. And what happens if we already have the Democrats at 83% for the House numbers falling on for the Senate, and what happens if the AI, uh, you can't delay anything again? You have orders, you're not allowed to delay things. Alman calls China's AI progress remarkable this week. How China caught up on AI and may now win the future. This should be scaring the hell out of everyone. And again, I realize enterprises are not going to use Chinese open-source, and Dario Modi said that. But if the models are as good and they're 90% cheaper, then every entrepreneur is going to use them. And this is where the K-shaped economy and the end of public companies gets to. And again, I'm going to keep saying it. AI is the most disruptive force that has ever hit. It is deflationary. And the Chinese are offering models, and they are racing to keep up. DeepSeek's King Quen just captured 15% of the global AI market in a single year. They went from 1% to 15%. You can go read an MIT article. In fact, I highly recommend if you want to go read and go through it. But I'll read you some of it. Uh, found that Chinese open-source models have surpassed US models in total downloads. But these models differ in a crucial way from most US models. You actually get the model weights, numerical values that get set when a model is trained. So anyone can download, run, and study and modify them. And they keep getting better. They will not just offer cheaper models who want access to them. They will change where innovation happens and who sets the standards. They will become infrastructure for global AI builders. The adoption of Chinese models is picking up in Silicon Valley too. A general partner and Dre and Horowitz has put a number on it. Among startups pitching with open-source stacks, there's about an 80% chance they're running on Chinese open-source models. These are the enterprises theoretically in the future. They're not going to use Salesforce.com. They're not going to use this stuff. They will be AI-native, building their own stuff. They're not even looking to use US frontier models. The demand is also rising globally. Limited new subscriptions to its coding plan for, uh, a new company, or new ZAI reports that the system's user base is primarily concentrated in the US and China, followed by India, Japan, Brazil, and the UK. OpenClaw, and this is an important thing. If you've heard OpenClaw but you don't know what it is, this is the beginning of the agents. When the agents are running, as someone who has a Mac Mini, your costs go up if you're using US models. If all of a sudden now it's costing you $500 a day to run a bunch of AI agents to do stuff overnight for you, and you can do it for $50 with a Chinese model, and it's protected where you don't have to deal with the security. This is why OpenClaw accelerates AI usage for China, and it will hurt the inference ability of people in the US offering it. It's going to happen. It already has happened. I'm using an open-source Chinese model, and it's this one. Kimy's K2.5 had surpassed Claude Opus and became the most used AI model by token count, meaning meaning it was handling more total text processed across user prompts and model responses. China is catching up. You can now deploy OpenClaw in seconds on the cloud. You don't even need a Mac Mini. They're making this easy. Chinese, they just, uh, released a desktop automation that agent that runs 100% locally. Record low hallucination for the GLMI 5. Genie 3 released a world model released by Google, and then immediately there was a Chinese open-source competitor. DeepSeek 4. The rumors have surfaced, and the numbers are terrifying. The best model scored 83.7 in programming test, blah, blah, blah, making it high in this, assuming it's true. We don't know, but it'll be released soon. Why are Chinese AI models dominating open source? This gets through it, but they've captured the space. Research published by Sentinel 1 and Senus mapping 175 exposed AAI hosts across 130 countries over almost a year. Alibaba Quen consistently ranking second only to Meta's Llama in global deployment. More importantly, it's running multiple AI models, becoming the de facto alternative to Llama. Uh, token share by region. Again, you can see how much the red side here, the Chinese continue to move higher again by region. Uh, dominating on open router. You can get another chart here. But more importantly, from Shanghai macro strategist, uh, top AI models now have a combined 47% market share. Isn't this the biggest nightmare for American AI firms that have spent 10 times more than their Chinese counterparts? That is the other thing about this is we're assuming that the capex being spent in the US makes sense. It costs, they're doing spending less there, which means they're being more efficient. Again, it's just, don't even fight the fact that this is happening. It's very hard to measure, but this is where I differ from people that believe that, A, the US models are going to win, uh, in such a big way like they did, the Mag 7 did with everything else. And secondly, the Chinese have always been deflationary for things. And I think that is a very scary thing when you're thinking that these models are going to be used in a way that you can monetize when deflation is going down.
Netflix threatens ByteDance. If you haven't seen this week in this sea dance, so I, ByteDance. These are AI generated movies, videos. If you haven't seen them yet, it is staggering. If you haven't seen this one, go watch it. Uh, it's insane how good everything is. Hollywood AI nightmare. The Door Brothers. This one I watched. Unbelievable. Um, artificial intelligence has stormed into Hollywood with breathtaking speed and alarming consequences. Um, it just is. Here's Netflix stock. If you don't think it's getting impacted, um, it is. Everything is being disrupted by artificial intelligence except for Apple, 'cause they haven't spent any money on it. Here's Apple's spin. So maybe Apple's going to be fine. I think it'll be interesting to watch. Uh, they've avoided this thing. Um, we'll see. I can see the argument, even though Siri hasn't worked. Uh, AI is compressing timelines, changing the nature of work. Another article that came out, this is the Jeff Seabird one. If you guys want to go read up on it since I mentioned it in here, if you didn't see this, OpenAI hires the OpenClaw founder, Peter Steinberger. Um, so Claude, this was created out of kind of Claude code. Now it's, uh, been he's been hired by OpenAI. It's latest hire changes everything about AI's next move. We are in the agentic world, guys. Uh, so get ready for billions of Einsteins. And of course, if you haven't used Grok, now he's got the names poking up on the agent. So we actually have the agents working. They're working on working on my chemical, uh, thing for me, which will come out next week. I got too busy. I put two papers out this week. One of them, uh, you definitely should look into, uh, in terms of the, uh, the nan side. Uh, I'll get into that in a second. Google releases Gemini 3.1. Grok 4.2 came out. So the model leapfrogging continues. Jensen Huang teases upcoming surprise chip. A chip that will surprise the world will be unveiled next month. I still believe Nvidia is the key to the market. As long as their stock holds up, and you'll get earnings from them next week. As long as their, their stock holds up, I think the S&P can survive until it turns, until credit starts to make things ugly. If Nvidia for any reason, uh, it doesn't matter what the earnings are, they sell off, then I think you'll probably get a get a correction at that point. Um, Nvidia and Meta did a deal again for a lot of chips. Rampant AI demand for memory is fueling a growing chip crisis. Everyone wants to short memory. They look at the charts and say it's got to be in there, which is why I wrote this. Uh, for those of you who haven't signed up, this is just again, if you want ideas. There's a bunch of stocks in here, some that I love, uh, in particular, that have gone sideways for really the last couple years. Uh, we're about to hit a flash wall. So this goes through, and this was after an interview with the CEO of Feison Electronics out of Taiwan. Great interview. Had it translated, went through it. I think there's a tremendous amount of ideas in it, um, in particular the demand drivers as to what's going on. This is just going to continue. So you've got the US cloud, the buildout, the China cloud buildout, but then you're getting into the edge device that is starting in the second half of this year, the consumer rollout, and then you get into the institutional on-premise education 2026. There are so many levels to this memory side, and there just isn't enough.
Um, JP Morgan launched its HALO acronym for heavy assets, low obsolescence. Um, again, I think this fits in with my abundance scarcity thing, but everyone's got to come up with their own ideas. Stan has jumped on board the EWZ trade, uh, one of his largest new positions.
All right, this is where I want to bring it up. So this is Bobby Maguire, for those of you who don't know, uh, one of the traders at 22V. And this is him with his Mac Mini OpenClaw agent setup. I asked him to send me a photo because you guys should be doing this. Bobby has embraced everything with AI. He's run with it. We started talking about it, I guess a year ago, he mentioned, and I noticed that every week that he came in, he was more amazed by it. So when I'm doing these videos and I'm saying you have to become AI-native, you have to think and have an AI mindset. Here's Bob Maguire. Here's me. We're not two young guys. We did it. We have no coding skills. If you guys know Bobby, call him up. Talk about his experience. For you guys in Boston, I'll be in Boston this week meeting, uh, some of you guys on the asset manager side. Uh, Mike Walsh, who's hosting or at least, uh, putting the thing together. He reached out to me after a call and said, "Hey, is it possible that I could build an app?" And I said, "You can build an app. You just need to go onto Claude Code, and I think you can do it with Jesse's simple instructions." So, uh, over the weekend, he ended up building his own app on the website. He did this lifestyle change one. Give kudos to Mike Walsh. I bring this up because the leaders of companies, if you're watching, and I've met some of you over the course of the last year and you've had me come in from a consulting basis, the leadership of the companies has to do this. If they do it, then everyone else will follow. I think these guys have done an awesome job. Um, I'm going to keep promoting this site. I'm trying to help people with it. The videos are going to be there to help you develop an AI mindset. The investment side is going to help you hopefully avoid some things, think about things, proactively give you risk management side. Please subscribe. Please send it to your family and friends, and let me help as many people as possible. Uh, and I'll see you guys next.