Can you all imagine what Tommy Hearns or Ray Leonard woulda done to the two mfers last night?? Actually scary to think of. As the self anointed kings of today like to remind us, there are "levels to this shit". They aren't at THAT level of #boxing ... nowhere close. Ain't even a discussion.
@SquawkStreet@USTreasury@SecScottBessent There it is -- the pivot. Bessent just said an actual way out of this mess. Economic isolation as the path instead of another round of bombing
@Stocktwits Love to see it. What I don't love to see is seemingly off all fin X taking the other side of Burry. Makes me think bulls AND bears are about to get it
“Let me get this straight, you’re mobilizing hundreds of billions with GPUs as the underlying asset?”
“Not just GPUs. NVIDIA compute. Full-stack AI factories. Vera Rubin, CUDA ecosystem, the whole thing.”
“So you’re packaging this into independent compute financing platforms with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, creating dedicated capital pools that get layered into more structures?”
“Yeah, the original capacity is backed by actual NVIDIA AI factories -- productive infrastructure with long useful life, usage-linked revenue, residual value. Some racks are brand new -- some are already generating tokens. Millions of GPUs across the ecosystem. But the real product is credit backed by NVIDIA compute. Compute as an investable asset class. AI-factory infrastructure squared if you will.”
“What if utilization drops or a new architecture makes the residual value collapse?”
“We’ll structure residual-value support mechanisms. Up to 25% in some cases. And keep improving the software stack so the economics get better over time.”
“C’mon! How much bigger is the market for synthetic compute financing and credit products than the actual physical chips and data centers people can use to train models tonight?”
“At least the $500 billion+ they’re aiming to mobilize — and that’s just the third-party capital these platforms are designed to unlock over time.”
“Ok, let’s say we have an underlying pool of real NVIDIA compute that customers need. How much institutional money could end up allocated against these financing platforms?”
“Over $500 billion.”
“That is fucking crazy.”
“No, it’s awesome. In AI, compute is revenue.”
“Let me get this straight, you’re lending billions with cheese as collateral?”
“Not cheese. Parmesan. Parmigiano-Reggiano to be precise.“
“So, you finance this by packaging the Parmesan loans into Parmesan CDO A, which has part of Parmesan CDO B and both get put into Parmesan CDO C?”
“Yeah. The original loans are backed by wheels of Parmesan. Minimum age of 12 months. Some are 24 months. 36 months. 48 month. Hell, up to 120 months. Delicious. Millions of wheels. But Parmesan CDO C is a synthetic Parmesan CDO. A CDO of Parmesan CDOs. Parmesan Squared if you will.”
“What if it gets too hot in the summer and the wheels of Parmesan melt?”
“We’ll sell Parmesan credit default swaps.”
“C’mon! How much bigger is the market for Parmesan-backed sythentic loan and CDS products than actual edible real-life Parmesan that I can grate on top of my homemade spaghetti bolognese tonight?”
“At least 10,000x.”
“Ok, let’s say we have an underlying pool of $10 million in Parmesan wheels. How much money could be out there betting on your synthetic Parmesan financial products?”
“Probably $100 billion.”
“That is fucking crazy.”
“No, it’s awesome.”