El moment més arriscat de la bombolla de la IA: quan la prudència sembla absurda

After 40 years of investing, I've learned that the riskiest moment in any market is never the one that feels the most dangerous. It's never the actual crash or the scary headline because by the time those arrive, the money's already long gone. The truly risky moment is the one that feels like a party when everyone around you is celebrating and getting so rich that you feel stupid for being cautious. And I think we could be standing in one of those moments right now because behind the celebration, the definition of a safe investment is being rewritten without your permission. And the bill for this whole AI bubble is about to land on the people who don't even know they're holding it. By the end of this video, you'll understand exactly what's happening, why it's happening, and most importantly, how to make sure you are not the one left holding a bag right at the top of the bubble. Imagine you are at a party. The music's great, everyone's having a time of their lives, and then you notice the people who own the house, the ones who actually know what's going on, are slipping out the back door and calling cabs. They're smiling, telling you to stay, have another drink. The night is young, all the while they're leaving. To me, that raises the question, what do they know that I don't? And that's exactly what I'm watching in the AI world right now. SpaceX going public on the stock market, has brought a lot of attention to the AI world, but more than 600 current and former open AI employees sold roughly $6.6 billion worth of their own stock on the private market. Yes, I said 6.6 billion with a B. These are the engineers, the early employees, the people on the inside who can see the actual numbers from behind the curtain, and they sold a fortune of it to outside investors. Now, when the people who built the thing are cashing out their chips and handing the stock to strangers, who exactly do you think those strangers are? Because somebody has to be on the other side of what the insiders are selling, and it's worth asking whether that somebody is eventually gonna be you because it isn't just SpaceX open. AI is lining up to go public later this year, and so is anthropic both at valuations, rumored to be near a trillion dollars each. All of these companies, the crown jewels of the artificial intelligence revolution, if you will, suddenly racing for the exit of the private market and into the public one at the exact same moment. And when a whole wave of companies rushes

to issue shares at the same time near the top of a boom, it's very often been a warning sign. The research firm, capital economics put it well, a sudden surg in companies issuing new shares has historically signaled that the end of an equity boom is months away. It's not a guarantee, but it is a pattern that has shown up again and again. But I do wanna be fair here because I always try to be, this is not proof of a scam. SpaceX is a real company that launches real rockets and open AI has real revenue and a product that hundreds of millions of people actually use. The insider selling might just be ordinary people who've worked hard for a decade and want to finally buy a house or take some chips off the table. That's just being human and completely normal, but an honest seller cashing out and a smart insider quietly heading for the door at the top of a bubble can look absolutely identical from the outside. The only way to really tell the difference between the two is to stop watching the celebration and start following the money to fully understand what is particular moment is different from every other scary headline you've rightly ignored for the last three years. You first need to understand a magic trick that's been running underneath this entire boom since the beginning. We often talk about the stock market as if it's some giant vault of money, but in reality it's just a confidence machine that runs on agreement, belief, and the price of the last trade. A company can gain a trillion dollars of value without a single extra trillion dollars actually existing anywhere on the planet. It works this way because a company has, let's say, a billion shares, and if someone buys just one of those shares at a higher price than the last person paid, every single other share gets repriced upwards to match on paper. The company is suddenly worth a lot more money. But did a lot of money actually exchange hands? Absolutely not. A simple way to picture it would be to think about the houses on your street. Imagine a place three doors down sells for a record price and way more than anyone expected. Suddenly every homeowner on that street feels richer because the estate agents will tell you the whole street is now worth more. And in a sense it is. But notice how nobody handed you or your neighbors a check. Everyone just feels wealthier because of the price of somebody else's house. The wealth is real and kind of unreal at the same time.

It exists for now right up until everybody tries to sell at once and then they find out what the street is actually worth. That right there is what the AI market has been doing for the last three years. The value has been believed marked to the price of the last trade for three years. This AI boom never once had to prove that anyone would actually pay up. It only had to be believed in for the price to go up, which in all fairness did work a tree. But starting this month with the wave of giant AI public listings, the boom doesn't just have to be believed in. It had to be funded by real investors with real money. That's the line we've just stepped across. So naturally the question changes. It stops being do people believe in ai because clearly we do and becomes, is there actually enough real cash sitting in the system to pay for everything that everybody believes? And when you sit down to add up the bill in real dollars, they have to come from somewhere. I've gotta be honest, it's worse than I expected. Let's start with the IPOs themselves. SpaceX, OpenAI and anthropic are expected to race somewhere in the region of $200 billion when they all go public. That will be a record breaking year for new listings, but that money has to actually come from somewhere and that somewhere is real investors moving billions of dollars into these companies buying their shares and providing the capital they need. But that's the small bit, the appetizer if you like, because sitting behind the IPOs is the real monster, which is the data centers. Chips and power stations needed to actually run all this artificial intelligence. Let me walk you through the numbers because they're almost hard to believe for 2026 alone, the biggest cloud companies are committing well over 700 billion in capital spending. But I get that sounds made up. So let's break it down even further. Amazon around 200 billion alphabet, somewhere between 175 and 185 billion meta between 115 and 135 billion Microsoft, 190 billion and Oracle around 50 billion. That's nearly double what they spent last year. And if we now put the two bills together, that's hundreds of billions to fund the new public companies and hundreds of billions more to build the infrastructure,

that's the better part of a trillion dollars of real cash that this industry needs to find just to keep the engine running at full speed. So you might be thinking, where does all this money actually come from? And that's the thing, there's no secret reservoir of fresh money just sitting there waiting to be poured into ai. For investors to buy hundreds of billions of dollars of new AI shares, they have to free up the cash first. And to do that, you have to sell something you already own. So you sell some Apple, some Tesla, some Microsoft, and maybe even a slice of your index funds. Essentially the money to buy the new winners has to come in large part, at least from selling the old winners. Think about that for a second. The AI boom has grown so large that it has to cannibalize itself just to feed itself. That's not a growing pie where everybody gets a bigger slice. That's essentially a game of musical chairs. And right now, a record number of players have all stood up at the exact same time reaching for the same chair. Maybe demand for these shares is so colossal that the whole thing just gets absorbed without anybody really feeling it. Maybe there's enough money sloshing around the world to soak it all up. But what I want you to notice is that even inside the ball case, which is the best possible version of events, it still requires people to sell an absolute fortune of existing stock to make room for the new stuff. So the question becomes who's gonna be the one selling and who's gonna be the one buying these insane prices? That's really where this whole thing gets clever because let's be honest, the industry isn't naive. They know the cash problem is real, and so over the last couple of years, they've been building two solutions and it's so genius. I almost admire it. Okay, so let me show you the first solution because it's brilliant the way a magic trick is right up until you spot how it's done. Nvidia, the chip company invests around $30 billion into open ai. So far pretty straightforward open AI now backed by that cash and billions more commits enormous sums to buying computing power. Roughly 300 billion with Oracle, 90 billion with a MD and 38 billion with Amazon's cloud infrastructure, eye watering amounts of money, all spent on the computing capacity needed to train and run AI models. But what do Oracle a MD

and Amazon do with a significant portion of that money? Well, they buy chips, and right now the most sought after AI chips in the world come from Nvidia and just like magic, the money circles back. It's a powerful flywheel invest in the company's driving AI demand then benefit again when that demand ultimately comes back to your products. Do you see what I'm seeing? The money goes in a circle and every single lap of that circle pumps AI valuations up a little higher. It's a closed loop and it's been spinning perfectly for a couple of years. Now, I've even spoken about it before on my other channel, but let me be fair, because there is a genuine argument on the other side of this. The companies themselves call this a virtuous circle, and they're not entirely wrong. You see, by committing all this money up front, they lock in a scarce supply. They guarantee themselves the chips, the builders, and the customers all at once in a world where everyone's fighting over the same limited resources. So while it is all smoke and mirrors, there is also a real strategy to it, but there is a pretty big vulnerability here because a closed loop of money that pays itself can look exactly like roaring growth for a very, very long time, even long after real outside demand has actually stalled, as long as the money keeps circling between the same handful of insiders, the numbers keep going up and everyone stays happy. The loop works perfectly until the moment it needs fresh cash from outside the circle to keep it spinning. And what do you think this wave of IPOs actually is? Well, I'll tell you exactly that moment. For the first time, the circle is reaching outside of itself and opening a door to the public market asking you to put money in, which brings us to the second solution. And honestly, this one is incredibly important for everyday investors like you and me because the AI industry doesn't just need to find new buyers for all this stock, as plenty of people will willingly buy. What they really need is a way to force people to buy without even realizing it. And it turns out those forced buyers might actually be you. This part literally stopped me in my tracks when I really understood it, and a lot of great YouTubers like Damien Talks money have already covered it. But I think it's worth discussing as it has a direct impact on your money investing strategy and overall financial future.

Almost everyone in the world, including probably you believes that index funds are the safe, sensible grownup investing choice. Just buy the market. Don't try to be a hero, don't pick stocks. Just put your money in a nice boring index fund that owns a little slice of everything and you'll be protected from exactly this kind of wild AI speculation. That is the single most repeated piece of investing advice ever, including from me. And honestly, it's been great advice, but what you need to understand is that an index isn't actually the market. An index is a list, and that list is created and edited by companies like Vanguard, MSCI, NASDAQ and more. These are businesses that compete with one another and they make their money by staying relevant and having as many customers as possible track and invest in their list or in other words, index. And to stay relevant. In a world where the most exciting companies are these giant new AI listings, several of them just rewrote their own rules. So let me show you how the rules have changed because that in a nutshell is where the worry lives. Normally when a brand new company goes public, it has to wait. It has to sit in the market for months, sometimes a full year before it's allowed to join a major index. And that waiting period exists for a very good reason because it gives the wild opening price time to calm down and find something closer to reality. It's kind of like a calling off period, and it gives the market time for the hype to settle to a more normal baseline, but not anymore. NASDAQ changed its rule. So a giant new company can get fast tracked in after just 15 trading days, and the Russell 1000 and CRSP went even further with as little as just five trading days, and they relaxed the rules about how much of the company needs to be available to trade. This means a brand new, wildly speculative, barely tested stock can land inside your boring sensible index fund almost immediately, right at its most height overinflated price. According to estimates from Bloomberg Intelligence Index funds could be forced to absorb something like 19% of SpaceX's available public shares just for the s and p style funds and another 24% or so for the Russell and NASDAQ funds. So let me give credit where credit is clearly due. Not everyone caved. And on June the fourth s

and p Dow Jones indices held the line. They had every opportunity to weaken their rules and let SpaceX in early and they said no by keeping their 12 month waiting time, which in my opinion was a great move. But the problem is not all of them held firm and tens of millions of ordinary people hold NASDAQ and Russell Funds without ever reading the rules that sit underneath them. This is the handoff. The insiders sell their stock, the index rules get changed in the background and the speculative risk gets transferred automatically to the most conservative and trusted retirement accounts in the world. You were told you were buying safety, but you might actually be buying the very top of the most risky moment in the AI bubble. But look, maybe it's all completely fine. Maybe what you're being handed is a rock solid stock that's worth every penny. So let's take a look at the foundation underneath all of this and answer this question. Are the reported profits of these AI companies even real? I'm sure most of you have heard of Michael Burry, but if not, he's the investor who saw the 2008 housing crash coming when almost no one else did. Now, he's been wrong a few times since then and he's pretty controversial, but still I think he deserves to be heard. His argument is pretty interesting because on the surface it sounds almost a bit boring, but underneath it makes a lot of sense, and it all comes down to one accounting word depreciation. When one of these giant cloud companies buys billions of dollars of AI chips, it doesn't count that cost in just one year. Instead, it spreads the cost over the number of years. It expects the chips to be useful on its own. That's sensible and just normal accounting. But Barry's point is that spreading the cost of these chips over five or six years doesn't make any sense because those chips don't actually stay competitive for very long. Now, in the real world with how fast this technology is moving, a top chip might really only be cutting edge for two to three years before the next generation makes it look slow. So if the chip wears out competitively in three years, but you are pretending on paper that it lasts six, then you're essentially undercounting your true costs. And if your costs look smaller than they are, then your profits also look bigger than they are. This might not sound like a lot, but what makes it a big deal is the scale. Barry estimates this account in choice could be

underestimating the industry's cost by roughly $176 billion between 2026 and 2028, which would mean some of Wall Street's favorite earnings, the profits, everyone is pricing these stocks off could be overestimated by something like 20 to 30%. He pointed the finger specifically at Oracle, which he reckons could be overstating profit by around 27%, a meta by around 21%. But I wanna make it, this is just a thesis from somebody who has been right in a big way before, but also very wrong about a lot of things. This isn't a proven verdict, and there are plenty of very smart people who disagree with Barry. The companies have real revenue and the chips have real value. CMBC even said they can't independently confirm the practice. So I'm not sitting here telling you the books are cooked. I'm more so just floating the things I'm seeing by you so you can make your own decision. And if we're being honest, you don't need Michael Burry to be exactly right. You only need him to be slightly, right, because if he is even roughly correct, then the foundation underneath these record breaking valuations is softer than the price is currently assuming. So let's stack the whole thing up so far. Insiders are selling real cash is running short across the system. A circle of money is just paying itself over and over. Index rules change. So you become a forced buyer, and now there's a speculative question mark over whether the profits are even what they appear to be. Any single one of these on their own will be fine. It's not exactly worth losing sleepover, but all five of them lining up at exactly the same time, that's not a coincidence, and in my experience, that's what a bubble about to burst would look like. At this point, if you're anything like me, you're probably starting to think, hang on, is this just a.com bubble all over again? Is this whole AI thing a giant fraud that's about to be exposed? The answer is no. AI isn't fake and it's definitely not a fraud. But strangely, that turns out to be the most dangerous part of this entire story. A lot of people naturally think when it comes to AI that if a technology is real, then investing in it must be safe, and it does make sense. I mean, if the thing is genuinely life changing, then surely you can't lose by betting on it. But that's one of the most expensive and naive mistakes you can make,

and history proves it again and again. So let me make this very clear. I'm not telling you AI is fake. In fact, I'm saying quite the opposite. AI is very real. It will completely change the world, and it's definitely here to stay. But the greatest bubbles in all of financial history weren't built on lies. I mean, look at the railways back in the 18 hundreds, for example. I was there. It was great times. The railways genuinely did connect the world and changed civilization forever, but investing in them at the wrong moment still wipe people out. Look at the internet in the late 1990s, the internet did change everything exactly like the believer said it would, but that bubble still destroyed huge amounts of personal fortunes when it burst. The point I'm trying to make is that technology being real isn't your protection. If anything, the technology being real is the bait. A fake story con or obvious fraud is something you immediately dismiss because your guard stays up. But a true, genuinely world changing story is exactly what convinces a sensible person to pay any price, because deep down, you know it's the future. A bubble doesn't burst when people stop believing in the technology. I mean, people believed in the internet the whole way down. What actually causes a bubble to burst is when the cost of continuing to fund it gets too high in every single one of these historical bubbles, the people who got destroyed weren't the people who believed in the technology like the railways, the internet, and now ai. The people who got destroyed were the late buyers, the ones who showed up right at the very peak, handing over cash in exchange for shares just before the money got tight and the market started selling off. So who do you think is being carefully maneuvered into the role of the late buyer, potentially right at the peak through their boring, sensible, and safe index funds? Go and look at your portfolio and ask yourself honestly whether you might be unintentionally holding a little slice of exactly that. So after years of investing in the stock market and living through more than one of these bubbles, you've probably got one question left for me. What am I personally doing about this? So let me give you my honest answer. There are two very obvious moves people often want to make at a moment like this, and I think they're both traps. The first trap is the bet against AI and short the stock, similar to what Michael Bowie did with the housing market back in the day.

I won't be doing that, and I wouldn't suggest you do it either, because as I've mentioned, the technology is real and betting against it is most likely the fastest way to lose all your money. The second trap is the complete opposite. To pile into these IPOs such as SpaceX and OpenAI at any price because you're terrified of missing out, I won't be doing that either because the price is, well, quite frankly, ridiculous and chasing insane prices is the other fastest way to lose your money. The real move isn't about predicting the top because let's be honest, no one can do that, including myself. But what you can do is make sure you are not the forced buyer holding the bag at the top. So here's what that actually looks like in simple terms. Feel free to write it down somewhere and keep it to hand first. Know what you actually own the phrase just by the market. Stop being a free pass. The moment those index rules got rewritten. So go and have a look under the bonnet of the funds you own. Find out what's actually in them and how quickly new unproven companies get added. You might be perfectly happy with what you find, but just make sure you look and don't assume. Second, don't let a rule change. Choose your risk level for you. If you want exposure to these wild new AI companies, that's a completely valid choice, but it's your choice to make on purpose with your eyes wide open. Third, let new companies prove themselves. This might be the most important one, and it's the one that saved me most money over the years. A great company is still a terrible investment at the wrong price and time is often the solution. And fourth, keep your costs low and your head clear. When everyone around you is cheering and it feels like you are the only one being cautious and not getting rich, that feeling is usually all the data you need to know that we are near the top of a bubble. I've lived through this kind of thing more than once now, and every single time the lesson is the same one that nobody wants to hear while the party's still going. The riskiest moment never feels like the riskiest moment. In fact, it often feels like the complete opposite, which is exactly why you have to take the time to do your research and watch videos just like this one. Otherwise, you'll just get taken for a ride. If you wanna understand why the US economy hasn't collapsed yet, then I'm gonna leave that video right up there. But don't click on it just yet. Make sure to subscribe if you wanna stay ahead of everyone else. Okay, I'll see you over there.