There's a classic scene in The Wolf of Wall Street where Jordan Belfort, already neck-deep in shit, looks straight at the camera and says everything's under control. His eyes say something else. The numbers say something else. But his mouth — oh, his mouth — keeps selling the dream.
That's pretty much what happened Friday with Mike Intrator, CEO of CoreWeave.
What Happened
CoreWeave, that New Jersey company that basically buys Nvidia GPUs on the corporate credit card and rents them out to giants like Microsoft and OpenAI, dropped its numbers and the market responded with all the subtlety of a punch to the gut: shares cratered nearly 22%.
The reason? Twofold. First, revenue guidance disappointed — meaning the company signaled it's going to bring in less than Wall Street expected. Second, and maybe more importantly, CoreWeave announced plans to spend between $30 billion and $35 billion in capex (infrastructure investment) in 2026. The FactSet consensus was $26.9 billion. The gap is nearly $8 billion more than the market had priced in.
Eight billion dollars. To put that in perspective, that's more than the GDP of about 40 countries.
The CEO Takes the Podium
Intrator went on CNBC to defend the plan. The pitch? The usual one when the house is on fire: "We're accelerating intentionally. Demand is massive. The backlog is enormous. This is a once-in-a-generation moment."
Where have we heard this before? In every bubble in history. Every. Damn. Time.
Look, I'm not saying demand for AI infrastructure isn't real. It is. Nvidia is printing money. Data centers are packed to the rafters. The hype has substance — up to a point. But when the CEO of a highly leveraged company tells you the debt narrative is "unfounded," you need to stop and think.
Intrator argued that CoreWeave's cost of capital dropped 300 basis points over the past 12 months, representing $700 million in savings. Over the past two years, a 600-basis-point decline. Looks great on a PowerPoint.
But here's the question nobody in the CNBC studio asked with the proper ruthlessness: what if the credit market tightens?
The Business Model That Should Give You Chills
Let's get down to basics, in plain English:
CoreWeave doesn't make anything. It buys absurdly expensive Nvidia chips, builds data centers, and rents out computing capacity to a handful of giant clients. It finances all of this with debt. Revenue depends on a tiny group of hyperscalers — Microsoft and OpenAI at the front of the line.
This is what Nassim Taleb would call an extremely fragile business. Revenue concentrated in a few clients. Dependence on a single hardware supplier. A debt-financed model in a sector where the technology changes every 18 months. If Microsoft sneezes, CoreWeave catches pneumonia.
The folks at JPMorgan cut straight to the chase: "If there's increased economic volatility, CRWV shares would likely suffer disproportionately due to risk-off positioning."
Translation: when shit hits the fan, this stock will be one of the first ones tossed out the window.
Barclays was a bit more diplomatic, saying it expects shares to "pause at these levels" while investors digest the new spending numbers. "Pause" is a polite euphemism for "nobody wants to catch this falling knife right now."
The Hard Truth Nobody Wants to Hear
Investing in AI infrastructure can be a valid thesis. But a valid thesis and a specific company are very different things. Cisco was a valid thesis in the '90s. Buying Cisco at the top of the dot-com bubble in 2000? Twenty-five years later the stock still hasn't returned to that level adjusted for inflation.
CoreWeave might be right about demand. It might even be right about declining capital costs. But when you're spending $35 billion a year bankrolled by debt, your margin for error is zero. And zero margin for error doesn't exist in the real world — it only exists in spreadsheets from CEOs trying to prop up their stock price.
Mike Intrator may genuinely believe in what he's selling. But the market doesn't run on faith. It runs on cash flow, cost of debt, and above all, fear.
And when 22% of your company's value evaporates in a single day, maybe — just maybe — the "narrative" about debt isn't so unfounded after all.
The question that lingers: would you lend $35 billion to someone who depends on two clients and chips that go obsolete faster than milk in the fridge?
