It's Official The AI Bubble Just Popped Here's Why

The big tech hyperscalers have just gone into an AI capex death spiral, which could lead to an even bigger bust than what we saw when the dot-com bubble burst. The AI build-out relative to the TMT build-out of 99-2000 is multiples, even as a percent of the economy GDP. If you look at what's happening on the AI space just now, it pretty much checks all the classic bubble signs. I'm going to reveal exactly what's going on, why they've crossed this very important threshold, where there is absolutely no turning back. This is the Titanic moment, if you will, and why this will most likely impact the economy and every single one of you watching this video in three simple, fast steps. Step number one, let's go over why this is just completely and totally unsustainable. Let's look at this chart, and I don't think most people realize what is actually going on. And when I'm talking about these hyperscalers, I'm talking about companies like Google, like Meta, and Microsoft. Well, there's a lot of them out there, but companies like that, let's say. And if you ask the normal retail investor, or if I ask you guys watching this video, you would probably say they are wildly profitable. I mean, they are absolutely printing money. And to a certain degree, you would be right. But lately, that's not the case. Let me show you what I'm referring to. Here's a chart from today's date going all the way back 2007. On the left, well, we start at negative $50 billion and go up to $450 billion. Now, the editor is going to throw up this chart. We've got two blue lines here. I want to focus on the light blue line. That's what this chart represents. So this is the free cash flow for these quote-unquote hyperscalers. And I think the ones they specifically include in this chart are, got to look down at my notes here, Amazon, Google, Facebook, Meta, Microsoft, and Oracle. So what we have seen since 2007, as you would expect, is their free cash flow, which you basically think of as profit, has gone up, up, up, up, up, up. We took this big dip, likely. That's probably the survey sickness. But then you go way up to a point where their profit, if you will, was right around $250 billion a year. But now look at it. Would any of you have guessed, if you didn't watch this video, that their free cash flow now, those mega companies, those huge titans of industry and tech, that now their free cash flow is actually negative and steeply negative? Now, I know a lot of you watching this video right now say, wow, wow, wow, wow, wow, timeout, George. This is very misleading. Because yes, it is true that they have negative free cash flow. In other words, they're losing money. But it's all because of AI CapEx, because they are investing for the future. Well, maybe, only if there's an economic model. And we have to compartmentalize an economic model with actual demand. See, demand can go higher and higher and higher and higher. But that doesn't necessarily mean you can make money. At the end of the day, there's infinite demand for selling $20 bills for $10. And right now, that is their business model. And also, we're going to get into this more throughout this video. If this CapEx turns into overhead, then all of a sudden, you got a big problem. And here is the big problem. A lot of people say, oh, George, I love your videos because you take complex topics and turn them into something that's very digestible and simple to understand. I'm going to make it real simple for you to understand right now. Here's the problem. There's no more money. But they need a lot more money. That's as simple as I can make it. And that's what's happening right now with these hyperscalers and the AI CapEx, oh, by the way, is propping up the entire U.S. economy. And for those of you who want me to simplify it even further, I got you covered. This is where we are right now. And this is no bueno. It's in the no bueno zone, as I say on this channel. That's all you need to know. You can click off the video right now. No, no, no, no. I got a lot more for you in steps number two and three, which we're going to get into right now. Step number two. So what do you do if you've run out of money, but you need a lot more money? Well, you have to borrow. And that's exactly what the hyperscalers are doing. And in my view, this is the first step to the death spiral. And I'll go over that specifically in step number three. But let's head right over to the Wall Street Journal title. The quarter trillion dollar onslaught of AI bonds is testing investor limits. Tech giants are borrowing even more than expected, weighing on bond prices. So we see here over the past couple of weeks, investment grade corporate bond market has struggled to absorb a combined 75 billion of bond issuance from wait for it. The most profitable company in human history. If you listen to CNBC, that would be NVIDIA. I believe they're valued at four, $5 trillion. And they're having to go out to the bond market and borrow money. And this group would also include SpaceX. No surprise there, because even the most optimistic analysts project that they're not going to even sniff profit until 2035. I'm not making that up. And even Amazon themselves are having to go out to the market and borrow mo' money, mo' money, mo' money, like that, in living color skit, to keep up with all of this AI CapEx spend, which they hope, they hope will actually yield a return on investment. But this is a lot different than these CapEx spends that we've seen in the past. Again, more on that in step number three. So when you combine what those guys are doing with the hyperscalers, with everyone else, this is leading to hundreds of billions of dollars of new bonds coming into the market, new corporate bonds. So it's not like treasuries. Those are way different, because those can be absorbed by Eurodollar banks. And I don't want to go down that rabbit hole in this video. But the bottom line is there's only so much money out there. There's a limited pool of capital, so to speak. So if you have just this deluge of supply of bonds, debt, debt, debt, debt, debt, there's only so many dollars that are out there that are willing to buy that. So at a certain point, you have supply overwhelming demand. And what happens when supply overwhelms demand? The price goes down, or in this case, the yields go up. Which takes us down to the next chart, where we see that this is the yield premium of corporate bonds over treasuries. So just as a proxy, let's take a look at meta, or beta, as I always call them, because Zuckerberg is such a beta dude. But anyway, you see that recently, the interest rate that they're having to pay has gone straight up. And this is relative to, say, maturity treasury. So they've gone from a premium, I'll call it 0.9, up to over 1%. So let's just call it a 15 basis point bump higher, which is very, very significant, especially when you see the other, quote, unquote, hyperscalers. Their bonds are doing the exact same thing. This tells you that there's too much supply, and there's not enough demand. But unfortunately, for these tech companies, and I would argue for the broader US economy, they have to continue to spend. They don't have a choice. So the question becomes, where is the money going to come from, and at what price? So if they have to continue to go to the bond market, the bond market is going to charge them a higher, and higher, and higher, and higher interest rate, which means they have to make more, and more, and more money with this CapEx spend in order to prevent this whole house of cards from crashing down. But there's another big, big problem. And this gentleman, I believe his name is Travis King. He is the head of investment grade corporates at Voya Investment Management. He says, everyone knows there's a lot more coming. In other words, they have to issue more and more debt. They don't have a choice because it's not coming out of cash flow. And like I said, they have no more money, but they need a lot more money. So if you're someone that's considering buying one of these bonds right now, why would you do it at the going interest rate when those interest rates will likely go higher, and higher, and higher? So as an example, let's say the meta bond is trading at 5% or 6%. Well, you know the deluge, the tidal wave of more debt from beta or meta is coming. So why on earth would you buy at 6%? Why wouldn't you just wait three months? Why wouldn't you just wait six months when you know the interest rates will most likely be 8%, 10%, maybe even higher? The fact of the matter is you wouldn't buy now. You would wait, which makes the inevitable a self-fulfilling prophecy, which gets here sooner, and sooner, and sooner, and makes the death spiral worse, and worse, and worse. Let's get right back to the article. But the hyperscalers are in such a heated race for computing power, they appear prepared to issue tens of billions of dollars of bonds. I would say hundreds of billions in bonds at any moment, regardless of market conditions and whether they might need to pay a higher interest rate. They're absolutely going to have to pay a higher interest rate. And this is basically exactly what I was just talking about. As borrowing costs rise, tech giants such as Amazon, Alphabet, Meta Platform should be able to keep spending, but they could be pushed to issue more equity. In other words, the interest rates go too high to where they can't afford it at all. Okay, the only thing that we can do to generate the cash that we have to spend is to sell more common equity, in other words, shares. And what does that do? That dilutes the shareholders, likely lowering the price of the shares. Well, wait a minute here. The entire economy, I would argue, is propped up by asset prices. So what we're talking about are those asset prices coming down. Well, what does that do to aggregate demand? And that creates a massive headwind for the economy, potentially taking us into a recession. And just to give you some scale or to understand the magnitude of what we're talking about, six companies that bond investors consider hyperscalers have already issued around $244 billion in debt just this year. Why? Because they ain't got no more money and they need a lot more money. Everyone can understand that problem. And look at this. Many investors were especially caught off guard. Yeah, I was one of them by Nvidia, this mega, mega, mega company of AI, really the poster child for AI, trillions and trillions of dollars of market cap. But even Nvidia and Jensen and his spiffy leather jacket has to go to the bond market and borrow $25 billion. That happened in June. And this is a mind-blowing statistic from the Wall Street Journal. This isn't George Gammon. It's coming straight from this article. So I want to make sure that all of you are sitting down right now, because if you're not sitting down, you're going to collapse. Check this out. High end estimates coming in north of $10 trillion over the next several years. Let me repeat. They're going to need north of $10 trillion just in the next several years. So the big question there is, okay, assume they have to invest $10 trillion or north of $10 trillion. Number one, is there a return on that investment? Can they possibly hope to make more than $10 trillion in profit? And number two, does this cap expend turn into overhead? That's the key component here. And that's what really puts them into a death spiral. We're going to go over that right now. Step number three. Now let's go over this death spiral I've been talking about and what I think the end game will be. And again, there are no certainties, only probabilities. But we have to start off by understanding a basic concept about AI cap expend and how it is different. I've done a lot of research on this one for this video, ironically enough, with AI itself. Like I said, there is no doubt the demand for AI is going to continue to go straight up. But it's all about the economic model. Here's what I'm referring to. Here's a car. As you can hopefully tell, like I always say, my artistic ability is absolutely infinite. It knows no limits. But this is just, let's say, your average everyday car. Okay. And let's say it has 50 horsepower. Okay, cool. That's kind of slow, kind of dull, but it gets you from point A to point B. And let's say the cost of this car is going down. Why? Well, because of technology. So at the beginning, let's say you buy the car right here. It's going to cost you $200. We've got the price over here. We've got the horsepower over here. So if you buy it right here, it's going to cost you $200. But if you wait time, time goes on and on and on. The longer you wait, most likely the cheaper it gets. Even if this car is new because we have advances in technology. And as you guys know, technology is always deflationary. Brings the prices of stuff down. But with AI, it isn't the exact same thing. Let me show you what I'm referring to. So we've got another chart here, the exact same thing. But instead of just a boring kind of everyday car that gets you from point A to point B. Now, all of a sudden, this technology that's bringing the cost of this car down is also creating cars with more and more horsepower, more bells and whistles, let's say. So that's why I've got this speedy car. Kind of looks like a UFO, but let's just assume that it's actually a car. This baby doesn't have 50 horsepower. This thing can fly. No pun intended. It's got 100 horsepower. So the horsepower, because of the advances in technology, goes higher and higher and higher. So everybody wants this car that looks like the spaceship. Nobody really wants this old car anymore. But the problem is the cost to build a car with an additional 50 horsepower isn't just twice as much. It's exponentially more. So you see over here, we've got a different scale. So let's say back here in time, you could buy this 100 horsepower car for 200 bucks if it existed, which remember it didn't. But now that it exists, you can buy it for not $200, not $400, not $600, but it's going to cost you 800 bucks. You see, you get 2x the output, but the cost to produce that output doesn't go up by 2x. It goes up exponentially. So taking this back to AI, we have these new models that come out that, I don't know what they're calling them, but CHAT-GPT. You had, what was it? 3, then 4, then 5. You guys probably know better than I do. But let's just say that we've gone from model 4 to model 5, and it gets 2x better from 4 to 5. Well, that's good news. But remember what we showed in this car example, that the compute needed for two times better a model, this new model that everyone wants, isn't 2x. As far as the additional compute, it's 10x. Now, this isn't an exact number. I'm just using it for the sake of the example. But you guys can now see the problem we run into, and that is the cost of compute for the hyperscalers. In other words, this AI CAPX spend isn't going to go down like you would typically see with a new technology. It's most likely going to go higher and higher and higher. And I know a lot of you right about now are saying, OK, George, OK. But if they buy this old car that gets you from point A to point B, they can go ahead and rent that out, using that as an example. They can rent out all of their old compute, let's say. Yeah, but that doesn't even come close to covering the cost to keep up with the Joneses, because everyone now wants this 100-horsepower car. And next year, they're going to want a 200-horsepower car. And the year after that, they're going to want a 400-horsepower car. You see, and if you're spending a trillion dollars a year this year, you're going to be spending 50 trillion dollars a year in two years to get the 400-horsepower car that everyone wants. And yes, you're going to be able to rent out the old cars, but you're not going to be able to rent them out for 50 trillion. I assure you of that. You might be able to rent them out for maybe 500 billion. And again, these are not accurate numbers, but the concept is accurate. I'm just using these basic oversimplified numbers so we can all get our head around what the problem is and why these hyperscalers that are so important to the S&P 500 or the stock market and so important to the economy are in a death spiral. Okay, so then the question becomes, oh, George, I get what you're saying, but what is the end game here? Well, let's try to think this through. First and foremost, you could see these giants, let's just use Google as an example, using AI or this service as a loss leader. So as an example, way back in the day, we created these things called sales funnels. I use one for Rebel Capitalist Pro. And you guys have seen them online. It's basically where you pay a dollar for a lead. And let's just say that that dollar turns into $2 of sales. Well, if you're spending a dollar to get that lead, if you're making two bucks, you can spend up to $1.99 and still have a profit. But what happened is all of these competitors came in to buy that advertising, which led to that one lead, and they bumped the price up to $3, $5, $10, $20, $50. And you say, George, how on earth did they do that? Because on the back end, they had more products above and beyond just the initial $2 product, you see? So for them, let's say the lifetime value of the customer was $100, so they can afford to pay 50 bucks. They can afford to pay $99 for that one lead that the other guy can only afford to pay $1.99. Now, I wanna be very, very clear. In this world, if it plays out this way, we still get to a point where we're stuck with the old model and they can't afford to continue to bring us a new model year after year after year. So this is very unlikely in my view. Scenario B here in our end game thought experiment would be that we just run into constraints, limits like resources. Energy would be the most obvious. And in this scenario, again, we're stuck with the car that's 50 horsepower. It just gets us from point A to point B. Now, another argument you hear all the time is we're gonna go from just selling AI as a commodity where there really isn't an economic model due to what we're talking about here to where it becomes an agent. So this is basically replacing all of the workers. So you've heard Jensen on CNBC as an example in his spiffy black jacket trying to look really, really cool talking about the TAM, the total addressable market for these agents is trillions and trillions and trillions and trillions of dollars because in his view, what he's trying to pitch is that it's going to replace all of these workers. So we're not just using AI as a search engine where there's no economic model. We're actually replacing the global payroll. The problem with this is number one, AI isn't even close to that yet. As you guys know from using it and all of its hallucinations, if you get the thread too long, it just turns, it goes like full retard like Ben Stiller in Tropic Thunder. And do you really want simple Jack running your company? I don't think so. I can't believe I actually remember his name or the character name from that movie. But anyway, even if this did come true, what would happen to aggregate demand? In other words, what would happen to the economy? All right, great, you fire all the workers. Okay, who's going to pay for your stuff? Nobody, because nobody has any money. So that is a problem in and of itself. Now, the fourth option here, scenario, I think is the most likely. And what sucks is it's the worst because it involves your drunk, insolvent Uncle Sam. And I think that's why a lot of these companies like OpenAI are going to Washington DC. So often they're trying to get Trump to take a position in the company. Why? Because they know there isn't a realistic business model there or economic model. And they know that people are going to want faster and faster cars. And they kind of sell this as a national security issue. And therefore they get the government on board so they can continue to spend trillions and trillions and trillions and trillions of dollars. Even if they're only bringing in billions of dollars in revenue. Basically, they're going to the government for a bailout. And I also think this explains why you're seeing such an aggressive approach from all of these hyperscalers as far as creating this circular economy where one man's spending is another man's income. But then that man takes his income and spends it back with the original company. In fact, to explain this further, no one does it better than Jim Chanos himself. Editor, cut to the clip. When Tom buys chips for his data center from NVIDIA, NVIDIA recognizes that as revenue and profit. Tom capitalizes those expenses and writes them off over five to 10 years. And so you have a mismatch where the same dollar in a CapEx boom is recognized as profits by one entity and deferred by the same people expending the dollar. And that happened in 98, 99, 2000. From the middle of 98, S&P operating earnings rose 30% to the middle of 2000 over the two years. From the middle of 2000 to the middle of 2001 when order books got pulled, S&P earnings dropped 40%. So my base case, and again understanding there are no certainties, there are only probabilities, is these companies will become part of the government. There will be this private public partnership and effectively what they're doing now is they're trying to get to the point where they are too big to fail before they actually fail. Now, I know a lot of you right about now are looking at this and saying, okay, George, I get it. Makes a lot of sense. It's tough to argue with the math, especially when you're getting this math from AI itself. It's tough to argue with, but what do I do? How do I protect myself? How do I not only survive what's coming, but hopefully financially thrive from this new world that we're moving into? Well, the good news is I set up an online private investment community for retail investors just like you called Rebel Capitalist Pro. I did this with my good friends, Chris McIntosh, Brent Johnson, Patrick Srezna, Jason Hartman, just to name a few. And we specifically set this up for investors just like you guys watching to navigate these tumultuous times and actually profit from this insanity that we're seeing in today's markets. So if you wanna find out more about Rebel Capitalist Pro and how it can help you, we'll go ahead and put a link in the description below.