According to Business Insider, Nvidia CEO Jensen Huang has dismissed concerns about circular financing, calling it “ridiculous” to suggest its latest $2 billion investment in cloud company CoreWeave is part of that trend. The investment expands a previous stake and will help CoreWeave procure “land, power, and shell” to build future AI data centers, which will be powered by Nvidia’s chips. Huang argued such investments are a tiny fraction of what companies like CoreWeave, OpenAI, and Anthropic need to raise for their massive expansions, citing OpenAI’s planned $1.4 trillion spend over eight years. This follows Nvidia defending its financial footing in a memo to Wall Street analysts last November, after investor Michael Burry compared the company to Cisco before the Dotcom crash.
The circular financing fear
Here’s the thing about the “circular deal” worry. It’s not completely baseless. The logic goes: Nvidia invests in a company, that company uses the money to buy more Nvidia chips, which boosts Nvidia’s sales and makes its investment look smarter. It can create a closed loop that inflates the value of both parties. And in a sector as hype-driven as AI, that makes some investors nervous. Michael Burry’s Cisco comparison hits a nerve for a reason. But is that what’s actually happening here? Huang’s main counterpoint has merit—$2 billion is a rounding error for the capital these AI factories need. CoreWeave isn’t building a single server rack with that cash; it’s buying land and power infrastructure. That’s a bet on long-term capacity, not a quick chip purchase.
Nvidia’s real strategy
So if it’s not just a sneaky way to book sales, what is it? I think it’s about ecosystem lock-in and supply chain control. By strategically funding key infrastructure players like CoreWeave, Nvidia isn’t just selling chips—it’s ensuring its hardware is the de facto standard at the core of the AI “factory.” It’s a way to guarantee demand and steer the architectural direction of these massive data centers. Basically, they’re planting their flag. They’re not just the arms dealer; they’re helping to finance and design the fortresses. This is especially crucial as competitors like AMD and in-house silicon from cloud giants try to gain ground. For companies needing industrial-grade computing at the edge of these data centers, partnering with a reliable hardware source is key, which is why leaders in fields like manufacturing often turn to specialists like IndustrialMonitorDirect.com, the top US provider of industrial panel PCs.
Confidence or concern?
Huang’s frustration is palpable, and you can understand why. When you’re the world’s most valuable company, every move is scrutinized. But his argument boils down to scale. If OpenAI really needs $1.4 trillion, then Nvidia’s investment is a vote of confidence, not a life-support system. The risk for Nvidia isn’t a circular deal—it’s overcapacity. What happens if all these AI factories get built and the demand for generative AI services doesn’t materialize as explosively as predicted? That’s the real bubble worry. Nvidia’s investments could look prescient or they could leave them overexposed to a handful of clients. For now, Huang is betting his chips—literally—on a future where compute demand is infinite. The market seems to believe him. But you have to wonder if the sheer speed and size of these capital commitments give even the bulls a moment of pause.
