I want you to memorize two numbers: 42 and 35. The first number, 42, is the current Shiller CAPE ratio for the S&P 500. In 155 years of recorded market history, that ratio has been higher exactly once, at the absolute peak of the dot-com bubble in December 1999, right before the Nasdaq lost nearly 80% of its value. The second number, 35, is the percentage of the entire S&P 500 now controlled by just seven companies. Seven, not 70, not 700. Seven companies now carry more weight in the index than the previous concentration record set in March 2000, the month the dot-com bubble burst. Every financial network on earth is telling you the same story right now. They are telling you AI is a revolution, that the spending is justified, that the companies driving this technology are the most profitable enterprises in human history, and that anyone who questions the trajectory is a Luddite who does not understand exponential progress. That reading is not just incomplete, it is structurally dangerous. Because when you strip away the enthusiasm and actually follow where the money goes, not where it is announced, not where it is promised, but where it physically travels, you discover something that should alarm every serious investor on earth. Here is my thesis. Today, I am going to show you exactly three things. First, that the largest technology companies in the world are engaged in a circular financing architecture that artificially inflates demand and disguises the true health of the AI economy. Second, that the physical infrastructure required to sustain current AI spending projections is hitting a wall that no amount of capital can climb over in the time frame the market is pricing in. And third, that the gap between what AI companies are valued at and what they actually earn has reached a level that historically has preceded every major market correction of the past century. On this channel, we do not do panic and we do not do cheerleading. We follow the logic of power, we follow the incentives. And when you strip away the emotion and the headlines, the picture always becomes clear. Let me start with the facts, because you need the ground beneath your feet before I take you into the analysis. In 2026, the four largest hyperscalers, Alphabet, Amazon, Meta, and Microsoft, have committed to spending approximately $700 billion on AI infrastructure in a single year. Alphabet alone raised its capital spending forecast to as much as $205 billion. Amazon plans roughly 200 billion, Meta up to 145 billion. These are not projections from analysts. These are committed capital plans disclosed in earnings calls and SEC filings. Meanwhile, Nvidia delivered $215.9 billion in fiscal year 2026 revenue, a 65% year-over-year increase, with data center revenue alone reaching $193 billion. Nvidia commands somewhere between 80 and 90% of the AI accelerator market. Its market capitalization sits near $5 trillion. On the other side of the ledger, OpenAI, the company that ignited this entire era, burned $3.7 billion in the first quarter of 2026 against $5.7 billion in revenue. Its operating margin is negative 122%. It loses a dollar and 22 cents for every dollar it earns. Its valuation after closing a $122 billion funding round in March 2026 stands at $852 billion. That is a 34 times revenue multiple on a company that has never been profitable and projects $14 billion in losses for the full year. Now, the mainstream reading of these numbers is simple. AI spending is enormous because AI is transformative and the profits will come. But that explanation is too simple. To understand why, you first have to understand a concept I call the Ouroboros Engine. In ancient mythology, the Ouroboros is a serpent eating its own tail, a closed loop that appears to sustain itself, but is actually consuming itself. That is precisely the financial architecture powering the AI boom. Let me walk you through it. Nvidia invests in OpenAI. OpenAI uses that capital to buy Nvidia chips. OpenAI signs a $300 billion cloud contract with Oracle. Oracle uses that commitment to buy more Nvidia chips to fulfill the contract. Nvidia then reinvests in the same companies buying its products. Each company records these flows as revenue. Each company's stock price rises on the strength of that revenue, and the cycle repeats. By mid-2026, analysts tracking these interlocking deals estimate the total circular arrangements exceed $800 billion. Bloomberg called it circular financing. One prominent analyst called it a financial Ouroboros. Jensen Huang himself confirmed at the Morgan Stanley conference in March 2026 that Nvidia was pulling back from its planned $100 billion OpenAI investment, scaling it down to 30 billion, and signaled it would likely be Nvidia's last major equity investment in a customer. Think about what that means. The company at the center of the AI revolution just admitted between the lines that the circular arrangement had become more liability than asset. That is not a company investing in its ecosystem. That is a company quietly stepping away from a loop it helped create before the loop becomes indefensible. Does that sound like a healthy market? Or does that sound like a market where the largest participants are generating the appearance of demand by funding their own customers? Now let me take you deeper, because this is where it gets genuinely alarming. The Ouroboros Engine operates on three layers, and each one amplifies the fragility of the system. Layer one is the demand illusion. When Nvidia invests in OpenAI, and OpenAI spends that money on Nvidia chips, and Nvidia reports that spending as revenue, an outside observer sees organic demand. But the money never left the loop. The actual external revenue, revenue from customers outside the small cohort of companies passing capital back and forth, is a fraction of the headline numbers. OpenAI generates roughly $25 billion in annualized revenue, mostly from ChatGPT subscriptions. That is real, but against committed infrastructure obligations reportedly reaching $1.4 trillion, the ratio of real customer revenue to committed capital expenditure is so imbalanced that the entire structure depends on a single assumption, that demand from actual humans and businesses will grow fast enough to replace the internal funding before the loop exhausts itself. Layer two is what almost nobody is talking about, the infrastructure wall. Sightline Climate's 2026 data center outlook found that roughly 50% of planned US data center capacity for 2026 faces delay or cancellation. Of approximately 16 gigawatts of planned additional capacity, only 5 gigawatts has entered active construction. The primary constraint is not money, it is electricity. Gartner projects 40% of AI data centers will be power constrained by 2027. Goldman Sachs estimates US data center power demand will more than double from 31 gigawatts in 2025 to 66 gigawatts in 2027, with data centers consuming 8.5% of summer peak demand. 75 data center projects worth $130 billion stalled in the summer of 2026 alone. I want you to hold that concept in your mind. The market is pricing in a future where AI spending accelerates indefinitely, but the physical grid cannot deliver the electricity to power the data centers that the spending requires. The money is writing checks the physics cannot cash. Geography is where the truth always lives, and in this case, the geography of power, literal electrical power, is telling you something the stock market refuses to hear. Layer three is the valuation gap. The Shiller CAPE ratio, which smooths earnings over 10 years to strip out cyclical distortions, currently sits at approximately 42. The long-run average since 1881 is 17.3. The only time in recorded history it exceeded the current level was December 1999, when it touched 44.2. Within two and a half years of that reading, the S&P 500 lost roughly half its value. The Magnificent Seven peaked at 35% of S&P 500 market capitalization in early June 2026, the highest concentration in the history of the index. The previous record was approximately 26%, set at the peak of the dot-com bubble. For context, the average weighting of the seven largest S&P 500 components since the index was created in 1957 has been roughly 17%. We are at double the historical norm. And here is the part that should keep you awake. Nvidia's market capitalization of $5 trillion rests on a forward price-to-earnings ratio of about 27. That sounds reasonable until you realize its entire revenue base depends on the capital budgets of a handful of customers, and capital budgets get revisited every year. OpenAI's $852 billion valuation at 34 times revenue is a bet that the company will not only stop losing money, but will generate profits large enough to justify a valuation exceeding the GDP of most nations, and it has to do this before the circular funding loop runs out of new capital to inject. That is not a market betting on technology, that is a market betting on perfection. Now before I continue, I want to present the strongest possible argument for the opposing side, because I owe you the real debate, not a simplified version. The bull case is genuinely strong. Unlike the dot-com era, the companies at the center of the AI boom are not profitless startups with catchy names. Nvidia generated $120 billion in net income in fiscal 2026. Its gross margins hover near 75%. The hyperscalers are the most cash-generative enterprises in human history. AI is not speculative technology, it is already embedded in search, advertising, cloud computing, autonomous systems, drug discovery, and military applications. The International Monetary Fund estimates 60% of jobs in developed economies are exposed to AI transformation. ChatGPT alone has over 900 million weekly active users. OpenAI's Ads Pilot reached $100 million in annualized revenue in under six weeks. The argument that AI is real is not wrong. AI is real. The argument that these companies will eventually monetize it is not wrong. They probably will. And the argument that this time is structurally different from the dot-com bubble has genuine merit. The underlying businesses are far healthier. I acknowledge all of that. But the steel man describes the next 12 months accurately. It does not describe the next 12 years. Because every bubble in history has been built on a technology that was genuinely transformative. The internet was real, radio was real, railways were real. The technology being real has never in the entire history of financial markets prevented the valuations from overshooting reality by a factor that eventually corrects violently. The question has never been whether AI is real. The question is whether the market has priced in a decade of flawless execution with no competitive disruption, no regulatory intervention, no infrastructure bottleneck, and no failure in the circular financing loop. And whether that pricing leaves any margin of safety if even one of those assumptions breaks. It does not. Here are my three predictions and I want you to write these down. Prediction one: Within the next 6 to 12 months, at least one major AI infrastructure deal, likely involving Oracle, CoreWeave, or a similar data center operator, will be restructured, delayed, or partially unwound due to power delivery failures. And the market will treat it as an isolated incident rather than a structural signal. Here is my reasoning. 50% of planned data center capacity is already delayed. The companies most exposed are not the hyperscalers themselves, but the middlemen, the Oracles and CoreWeaves, whose entire business model depends on building capacity fast enough to fulfill contracts signed on optimistic timelines. When the power wall forces a contract restructuring, it will reveal the gap between committed capital and deliverable infrastructure. Prediction two: By mid-2027, Nvidia's revenue growth rate will decelerate meaningfully, not because AI demand disappears, but because the circular financing loop contracts as IPOs for OpenAI and Anthropic close the private investment window that funded the loop. Jensen Huang already signaled this in March 2026 when he capped Nvidia's OpenAI investment at 30 billion and called it likely the last. When the loop contracts, the demand that appeared organic will reveal itself as partially manufactured, and the market will reprice Nvidia's multiple accordingly. Prediction three: The correction, when it comes, will not look like a crash, it will look like a rotation. The Magnificent Seven's share of the S&P 500 has already begun declining from its June 2026 peak of 35% to approximately 32%. That 3 percentage point shift represents hundreds of billions in capital moving out of concentrated AI bets and into healthcare, financials, and industrials. Watch that number. When it falls below 28%, the rotation will have become self-reinforcing, and the investors who were overweight the Magnificent Seven will discover that concentration works in both directions. Come back to this video in 12 months and tell me whether I was right or wrong. Now let me tell you the three things to watch. First, watch the power grid approvals, not the spending announcements, not the press releases about new data center campuses. Watch the actual interconnection queue, the pipeline of projects that have secured grid connections and broken ground. That number is the real constraint. If it does not accelerate dramatically by the end of 2026, the infrastructure wall is binding, and every projection built on unlimited scaling is wrong. Second, watch OpenAI's S1. The company filed a confidential S1 with the SEC in June 2026 and is reportedly leaning toward a 2027 listing. When that document becomes public, it will be the first time the market sees audited financials for the company at the center of the AI revolution. If the operating losses look anything like the leaked figures, negative 122% operating margins, 14 billion in projected annual losses, the narrative will shift from when will they be profitable to can they survive long enough to get there. Third, watch your own exposure, if you own an S&P 500 index fund. You already have 32% of your money riding on seven companies, all of which are making massive bets on AI delivering returns within a narrow time frame. That is not diversification, that is concentration wearing the costume of diversification. And if you want the 47 specific strategies for hedging your wealth, cutting your inflation exposure, and reading the five signals that tell you the truth before any headline does, I put everything into the Crisis Preparation Blueprint. It is the personal finance chapter that the geopolitical analysis never includes. The link is in the description and the pinned comment. Most households recover the cost in the first week. Here is the question I want to leave you with. Every generation believes it has discovered the technology that changes everything. And every generation is right. The technology does change everything. The internet changed everything, radio changed everything, the railroad changed everything. But in every single case, the market's ability to price that transformation accurately lagged behind its enthusiasm by years, sometimes decades. The investors who recognized the technology was real, but the pricing was irrational, they were the ones who built durable wealth. The ones who confused the brilliance of the innovation with the rationality of the market, they were the ones who got wiped out. So the question is not whether AI is real. AI is real. The question is whether you are investing in the technology or investing in the enthusiasm. Because history does not punish people who believe in the wrong technology. History punishes people who pay the wrong price for the right one. The most dangerous bubble is not the one built on fraud, it is the one built on a genuine revolution, because that is the one nobody believes can burst. If this analysis was useful to you, subscribe, because this story is going to move fast over the next 12 months, and I will be tracking every development. Tell me in the comments, do you believe the AI correction will come as a crash or as a slow rotation? I read your responses and your questions shape the next analysis. I am Professor Jiang Shuechin. Thank you for watching.