The AI Bottleneck No One Is Talking About Bitcoin Miners Already Own It

So everyone's arguing about whether AI is a bubble. Nobody's asking where the electricity comes from. It's the most important aspect. The demand for AI is overwhelming. Everyone's using it. Let's just be straightforward. It is increasing productivity on a serious level to the point where we have to have a whole bunch of AI data centers and it's causing supply chain issues with the power grid, right? There's a whole bunch of issues happening. And obviously when things like this happen where we have something that is really needed And there's a whole bunch of problems, the money-making opportunity is substantial. And so that's what I want to talk about in this video. We need to understand what's happening with AI. It's important to crypto. It's important to finance in general. And right now, a lot of people are calling it a bubble. If it is a bubble, it's going to affect everything, right? We need to comprehend the details behind AI that nobody else is willing to uncover. That's what we're going to do in this video. By the way, I've used AI so much in my life that I have actually replaced my entire trading strategy with an automatic AI buy and sell and research. It does it automatically, an entire We give that away for $1. Go to the link in the description below. Take advantage completely. I am not joking with you when I say that we give away the bot for $1. We really do that. Go to the description below and take advantage. We use AI a lot. I spend thousands and thousands of dollars on AI systems per month. I make sure all my employees have multiple Claude accounts. We use the heck out of it because it is substantially better. Than what humans can do. It can really brute force a lot of tasks. It's not smarter, it doesn't have more wisdom, but it could definitely brute force way more than a human can do. Our research papers are verified by like 300 AI agents, right? You just can't do that with humans. You can't scale to that capacity. It would be extremely expensive for me to have 300 humans trying to research altcoins. So the AIs do that for us and we give it away for $1. Go to the link in the description below. So one of the first things you probably already heard about is a Chinese company releasing a new AI model. The market actually kind of panicked and traders decided AI was about to get really cheap and cheap is bad if you sell the expensive stuff. So everybody's thinking that Anthropic and all these places were gonna get really, really like beat by just having a cheaper AI. And chip stocks actually got crushed. $3.3 trillion in value was erased in a few weeks, but the market got it completely wrong because they don't even understand how this works. Again, we have the market reacting to perception. I've said this so many times. If you've been with me for a while, you know I am a big advocate off of narrative economics. Right? I don't believe that the market moves off of pure truth. It does not. It moves off of what people think. So the market actually reacted in the exact wrong way it should have. And I'm gonna show you this video, but it doesn't matter anyways because if you know how to trade, you can make money off of that. Perception's more important than— we don't live in a math-based reality. Math does explain a lot of things, but it's not definitive truth. Sometimes there is things that are actually not true, But the market reacts in the exact opposite way. So what makes it true, right? The perception is true because the market, the price reflected that. And this is why, you know, I've made so much money off of these perception coins is understanding this concept. So if you're a new trader, investor, whatever the case is, remember, just because something's true doesn't mean that the market will perceive it that way. And that's what just happened with AI stocks. Let me show you. So chip stocks are getting crushed, losing a lot of value. And the exact same window, a 3-year-old computer chip actually got more expensive to rent. And I've seen this before, right? Mining Ethereum at $3 early on, there was a whole bunch of GPUs, which is basically the part of the computer that renders graphics, like for your computer games and stuff like that. The GPUs, literally the older ones, were getting more expensive because they were more profitable in crypto mining. And so those were going on eBay for premium prices, and they're extremely expensive. So Right now there's like a little supply chain issue, right? And the supply chain issue is pricing older chips at a higher price because they're actually useful for AI right now, not in the future. Like, they're needed right now. Simple supply and demand. And again, the economics behind this are extremely important. It affects every human. AI is like— I'm not joking with you guys— one of the biggest changes to our life that we will ever experience. Like, crypto is a big quantum leap. So is AI. The intersection between crypto and AI is the most lucrative big leap in innovation that we will ever experience. Our entire finance is changing. The way that money and value is transferred in the world is changing. And at the same time, the way that we perceive the internet itself is completely changing. And so we need to understand this. Older chips are getting more expensive, not cheaper, because of AI demand. This is very important. I'm going to say it straight up. AI is not a chip story. The market is pricing chips, right? They're all— everybody's fighting over chips and the price of chips, but that's actually an electricity story. We should know this by now. And almost every dollar in this trade is still priced like it's a chip story. That's called an opportunity. And I'm gonna actually prove all of this research to you in about 4 steps. Listen up. So let's talk about the Chinese model. So there's a company called Moonshot, and the model is called KIMI-3. You've probably heard of it before. And pretty much the market is like, you know, China builds a really good AI model and gives it out for cheap or free, and American AI companies lose their pricing power. So that story would be worth trillions if it were true, the people perceived it as true, but let's actually check it and look at what's going on. So AI companies charge by the token. I'm sure you guys know this. A token is basically a chunk of text the model reads or writes. They price it per million tokens. Moonshot priced their new model at $3 for a million tokens coming in, $15 for a million going out. Now here's the thing, that's the exact same price as American-made models, like literally the exact same. They're trying to compete, and it's about 5 times more than Moonshot's own version, so it is not cheaper. So the Chinese company didn't undercut anybody, they actually raised their own prices to match people in the West. The market erased trillions of dollars based on a story that the actual price disagrees with completely. So why did the price actually go up for KIMI-3? Because the model actually got a lot bigger, a lot more, a lot bigger, roughly 2.8 trillion parameters. Parameters are just the adjustable settings inside the actual model's brain. More settings means more hardware, more electricity, and more cost. And this matters way more than it sounds like it does. For 3 years, every single forecast in this industry has assumed one thing: that models get smarter and cheaper at the same time forever. This one got smarter and actually more expensive. Now, if that holds, a whole lot of the bearish math on AI just completely shattered. But here's the part that actually changed something permanently in this AI industry forever. 11 days after launch, Moonshot published the entire model file for free. Anyone on Earth can actually download everything for free, which means their price doesn't really matter anymore. If you can download the thing, right, you don't pay Moonshot anything. You just pay for the computers to actually run it. Do you see what I'm saying? You see how interesting it gets? Now it's a hardware issue, right? So let's follow one step further because this is where the market direction is like completely backwards. If anybody on Earth can now run a top-tier AI model, do we need fewer computers? No, we need a big surge of computers being made. Free models don't kill demand for computing power, they spread it out. They take money away from the companies selling access to models and they push it towards companies selling the ability to actually run them. AKA like NVIDIA and these AI computing companies, right? So the market actually sold the computing layer, but the layer that's actually in trouble is the model selling layer. They punished the wrong, the exact wrong industry for what just happened. I think it's crazy. So if the market is wrong about direction, let's actually look at the buildout and actually how it's getting paid for because that completely changed and it actually changed very quietly under people's noses, okay? For most of the last 2 years, big tech paid for AI out of pocket, real cash from real profits that were produced. That is completely over. 5 American companies are expected to spend roughly $700 billion this year on this buildout, and Goldman Sachs has pointed out that this is running close to 100% of their cash that their actual operations generate. I need you to listen to that one more time. They're spending basically every dollar that this business makes and actually borrowing for the rest. Morgan Stanley expects around $570 billion in AI-related debt to be issued this year alone. By the end of May, they'd already done about $236 billion of that projection. That is 4 times The pace of an exact year ago. So this debt is real with the AI bubble. It is. It's real. That's how bubbles are created. And I believe investors understand this. Here's how you know. Last week of July, Alphabet reports earnings. Cloud revenue up 82%, record profit per share. Stock fell 15%. Meta actually beat expectations But the stock actually fell 10%. Beating expectations is not enough anymore. The market is now grading these companies on how much they're actually spending, and people are reacting negatively even though they're meeting the expectations of their investors. And here's the trap, I believe. Investors have also said this out loud, that if any one of these companies actually cut their spending, it would actually tank the market. So they get punished for spending and they get punished harder for stopping. There's really no way out of that box where somebody doesn't eat a loss. Now, who's holding all of this risk? And this is the part that actually kind of reaches regular people so that you're aware. A lot of this debt is actually being bought by life insurance companies. The question why is because insurance companies actually own retirees' money decades from now, so they actually need long-term bonds to match that so they can pay them back. So I need you to listen to that for a second. Retirement money is helping pay for graphics cards right now in the American economy, and a lot of this financing isn't sitting on a public balance sheet anymore. It's actually side vehicles, private placements, asset-backed deals. You can't open a filing and see who's actually holding it. Bank of America surveys global fund managers every month As of this summer, about a third of them all named AI spending as the most likely sourcing of the next big credit blowup. That number doubled in 3 months. And here's the part that nobody draws out. The chipmaker sells chips to the cloud company themselves. The chipmaker also owns a piece of that cloud company. The chipmaker also agreed to buy back the cloud company's unused capacity if customers don't actually show up. Meanwhile, Google is guaranteeing another company's building leases so the landlords can actually borrow against them. So chipmaker backs the customer, the customer borrows against the chips, The hyperscaler guarantees the rent, and the rent supports the loan. That's not fraud; it's actually normal when things happen late in the actual spending cycle. We see this over and over again. This is like a bubble, but it means the demand looks a lot more independent than it actually is. Everybody's credit is everybody else's collateral. So what is backing all of this money? We have to. Figure that out. Listen, the chips themselves are backing all of this money. There's over $20 billion in loans right now where collateral is literally graphics cards. Collateral just means if you don't pay, the lender actually will go and take the equipment. That's the important part. The graphics cards right now are the value, right? So basically it's a heavy leverage bet secured by an asset nobody agrees on the lifespan of. Funded by a small circle of lenders who are holding the same collateral. That's essentially what's happening. If you're around in 2022, you've seen this movie. Same shape, different asset, much bigger numbers. We're talking about crypto and FTX exchange, and there is a wall of debt coming due between now and 2028. Now, that lifespan question is important. This is actually a big argument right now on Wall Street, and I want to show you that both sides, in my opinion, are completely missing it. Here's the setup. When a company buys equipment, they don't count the whole cost in one year. They spread it out over however long they think the equipment will actually last. That's called depreciation, right? So big tech will actually spread the cost of AI chips over 5 or 6 years. Critics say that's way too generous because a new and much better chip will actually come out every single year. We know this. If the critics are right, report profits across the industry are way too high. How big is this argument? Goldman ran the entire math scenario. If you shorten the assumed lifespan from 5 years to 3, the total depreciation across the industry goes from about $3 trillion to about $4 trillion. $1 trillion swing from one assumption. And here's the big tell. The companies can't agree with each other right now as we speak, and they're fighting over it. Amazon shortened their entire estimate and took a real hit to profits right now. Meta expanded theirs and added almost $3 billion to earnings. Microsoft and Google are still at 6 years. Same equipment, same year, same technology, completely different answers. When the biggest companies, listen, on earth can't agree on how long their own equipment lasts, that number is not a fact. It's actually an opinion, and it will be corrected by the market. And about a trillion dollars is riding on people's opinions. Now let me tell you what everybody in that argument, in my opinion, is actually missing. Go back to the very first thing I showed you. Okay, that three-year-old chip, the one that got more expensive to rent. If these things were becoming junk, that price would be collapsing. In my opinion, it went up 36%, and you have to ask the question why? Because you can't plug in the new ones. There's all these new chips coming out, but you can't even use them. Here's the actual situation on the ground so you understand what I'm saying. To connect the big facility to the power grid in America, you get in line. There's a lot of places. that this line is 5 to 7 years long. You have to understand this. In Virginia, the main utility has about 70 gigawatts of projects waiting right now. They can connect roughly 10 large customers a year. Average wait for their data center customer is 7 years. And after you get the permission, you need an actual transformer. That's the equipment that takes the high voltage power off the grid and turns it into something that actual buildings can use. The wait for a large transformer has been running over 2 years. So the shortage isn't chips anymore. The shortage is electricity and the gear that delivers it, which brings me to the point of the entire video. A chip doesn't lose its value when a better chip comes out, in my opinion. It loses the value the day that somebody can fill that same megawatt with something more profitable. Right now, nobody can, so the old chips keep earning. That's why the rent went up. But that protection is temporary. The moment real power capacity comes online, the old stuff gets kicked out of its megawatt, and not slowly, all at once, wherever the power switch comes on. So in my opinion, depreciation isn't a gentle slide downhill, it's an actual staircase, and the steps are the grid delivery dates, right? Almost every financial model out there assumes that there's a gentle slide. So where does the actual power for the power grid come from right now? Mostly gas on-site, skipping the grid entirely. There are companies literally trucking gas turbines onto property because the utility told them to wait One facility in Tennessee right now is running about 60 of them as we speak. But gas actually has a line too. The 3 companies that make big turbines are effectively sold out through 2030. What about nuclear? What about fusion? Real, but it's all late. The first units from big nuclear partnerships are targeted around 2032. Fusion is further out than that, and no other private company has even hit break-even yet. But here's the detail that settles everything. One of the leading fusion companies filed to connect their future plant to the grid this past April. That application process takes 4 to 6 years. So even if they crack fusion tomorrow, they still stand in the same line as everybody else. The bottleneck isn't invention, it's installation. And the debt, the debt comes due between now and 2028. Nuclear shows up in 2032. That gap is the entire thing that I'm trying to tell you guys. Now I want to show you something that almost nobody in this space is talking about. I told you that the scarcest thing right now is not chips, it's land with power already running to it. A site that's already connected, already permitted, already energized. So you have to ask the question, who owns a lot of those? And it happens to be Bitcoin mining companies. So we have Terawolf, Cipher, Hut 8, Core Scientific, and Applied Digital. For a decade, mining companies did exactly one thing extremely well. They found cheap electricity, got connected to the grid, and turned megawatts into money. And for a decade, everybody called it a massive waste. Then power became the scarcest thing in the most expensive buildout in modern history. And it turned out the miners already owned all of it. AI didn't replace crypto. Crypto is quietly building the floor that AI is actually currently standing on right now. And nobody wants to say that out loud in either industry because crypto has a bad consensus. So let's bring this all home. I'm not going to hand you a list of tickers. I'm going to hand you something even more useful, a filter. One question. Ask it about any company in this space. Does this company own something that can't be manufactured faster? That's a big question for AI companies because that question pretty much sorts everything out when it comes to you making money off the company. A shortage of manufactured goods goes away when factories expand. A shortage of permitted energized land does not. So when you look at a turbine maker with a huge backlog, Ask yourself, are they expanding production? Because if they are, they're actually dissolving their own advantage. When you look at a company that owns power plants already connected to the grid, ask, is it already in the price? Usually the answer is yes. And when you look at the actual company sitting on energized land, ask about debt, ask about customer concentration, ask whether the money's due versus when the power actually arrives. That last one is the failure mode for this entire cycle, by the way. And here's the thing I want you to keep after this video ends. Please write it down. From here on, every AI headline that crosses your feed sorts into one of two piles. Does this add megawatts, or does it add another claim on megawatts? Almost every single one is the second kind. That filter will still be working 2 years from now, and it will tell you more than any price prediction ever will. I'll see you in the next video. Thank you guys for watching.