Wall Street just endorsed Jensen Huang's 'big concept' for AI. What now?
By Ray Dombrowski ·
The whole thing stinks of financial engineering dressed up in silicon varnish.
The Asset Class is a Spreadsheet, Not a Factory Floor
The whole thing stinks of financial engineering dressed up in silicon varnish. What we saw last week—the grand pitch from Jensen Huang and his consortium of Wall Street titans—is not an economic forecast; it’s a structured capital play. Nvidia, through its memos of understanding with Goldman Sachs, BlackRock, Blackstone, KKR, Apollo, and Brookfield, is essentially selling the idea of future compute power. They are arguing that GPUs are now "productive, long-lived, fungible" assets, comparable to toll roads or power plants.
I’ve spent my career looking at spreadsheets: the BLS county employment series, the payroll reports from Delphi, the numbers on a local manufacturing floor. I know what real value looks like—a paycheck that lands in someone's account. What Huang and his partners are selling is something far more abstract: collateralized future revenue streams. As reported by valueaddvc.com, they are proposing to underwrite GPUs using "contracted revenue as collateral," turning the physical act of building a data center into a financial derivative.
The Illusion of Infrastructure Scale
The sheer scale—$500 billion, potentially more, for AI factories—is designed to make you forget that this is still just a collection of non-binding MOUs. When Larry Fink spoke about it being the "next future for financial engineering," he wasn't talking about workers; he was talking about yield curves.
The core deception lies in the leap from physical asset to pure finance. The consortium’s pitch, which finance.yahoo.com detailed, is that they are solving a financing bottleneck. But what does "financing" mean when you don't have actual committed capital flowing through payroll? It means pension funds and insurers—the ultimate source of retirement money—are being indirectly exposed to the volatility of chip sales cycles.
The comparison to infrastructure is where this whole house of cards wobbles. Toll roads generate revenue for decades; they are physical choke points that require continuous maintenance and human operation. GPUs, by contrast, have a useful life of three to five years before meaningful obsolescence, according to 247wallst.com. You cannot build an investment thesis based on something that depreciates this fast while simultaneously claiming it has the permanence of a power grid.
The Echo of Speculative Mania
This entire spectacle—the massive influx of structured capital promising future revenue streams before those revenues are proven—is not new. It is merely a highly polished echo of the dot-com bubble.
The shared mechanism is undeniable: speculative overvaluation of nascent technology, fueled by an enormous injection of financial promise. Back in the late 1990s, investments poured into the World Wide Web, promising untold future wealth that far outpaced actual commercial revenue. The difference now is simply the medium—chips instead of fiber optics. But the mechanism remains identical: a massive capital influx based on potential, not proven payroll.
The only things guaranteed in this whole arrangement are trillions in assets under management and an endless appetite for narrative. This isn't about building America’s industrial base; it's about creating the next great financial instrument.
What we are watching is not the buildout of a productive economy; it is the crowning moment of asset-based financial speculation, where the promise of future compute power gets mistaken for guaranteed cash flow. The real measure of progress is always who gets hired and at what wage, and right now, all I see are memos of understanding.