Boni Karanja, Jr.

Sam Altman just ran out of buyers at his price, and the AI bubble is GAME OVER. The money that carried this entire boom is pulling back, and the tech bros buying this dip are about to learn what every homebuyer learned in 2006. Watch this TERRIFIC video from @JayMartinBC comparing AI to the housing bubble. His point is that the housing bust really started in 2006, while prices were still at record highs, because borrowers paid off their old loans by refinancing against a higher price. When the gains shrank from the mid-teens to single digits, the refinancing stopped working and the loans started going bad. OpenAI runs on the same engine. It loses billions every year and pays yesterday's bills by raising money at a higher price, so the only number that matters is how much higher each new price comes in. Every time OpenAI has been priced since early 2024, the new mark came in 67% to 91% above the one before it. This month they went out to raise $30B at $1.4T, a 64% markup, and investors pushed back on the price, so it came down to $1.25T. Then on Friday we learned some of them cut what they were willing to put in, so the round shrank by a third to about $20B, and the first close is expected to come in below even that. One person familiar with the round said every earlier raise was a frenzy and this one landed with a thud. The markup just fell to 47% with a third less money behind it. The valuation is still going up while the speed is breaking, and that's how housing looked in 2006. What worries me most is who they go after next: OpenAI first pitched $1.4T to private wealth clients, and even they pushed back. Now the reports say the gap could be filled later with another offering to wealthy individuals at $1.25T. In 45 years I've watched the individual investor end up as the last buyer in every bubble, and that's who they're coming for now. OpenAI is also the customer at the end of the entire AI chain. Oracle's backlog leans on a $300B OpenAI contract and CoreWeave has up to $22.4B, and both borrowed heavily to build for it. When the end buyer struggles to raise money, the suppliers who borrowed against its promises feel it first. And the collateral is WORSE this time. A house in 2006 was still a house 10 years later, but the chips inside these data centers are outdated in about 3 years while the debt used to buy them runs far longer. The market is already pricing it. On Friday the Nasdaq 100 fell 3.5% for its worst day since June, and the chip index dropped 7.2%. This is why we shorted CoreWeave at $107 and Oracle at $189.97, and why we pressed the CoreWeave short at $81.58 earlier this month. CoreWeave is now down 49% from our entry and Oracle is down 31%. All 12 of our scored shorts are winners, and our long/short spread is 70%. And don't buy this dip. I'd trim anything that depends on OpenAI paying its bills, and I'd own gold and businesses that generate real cash today. If you want the names I'm investing in, join The Noble Update. IT'S GAME OVER Are you listening?