Nvidia CEO Jensen Huang is reframing how the world should think about graphics processing units (GPUs), describing them as productive infrastructure that generates revenue for years rather than hardware that simply depreciates. But this narrative is drawing scrutiny from one of Wall Street's most respected skeptics, raising questions about who truly benefits from the artificial intelligence infrastructure boom.

What Does Jensen Huang Mean by 'Highly Rentable' Chips?

Huang recently described Nvidia's compute as "fungible, durable and highly rentable" and positioned it as a "productive, revenue-generating asset." This framing represents a significant shift in how Nvidia markets its products. Rather than selling GPUs as one-time hardware purchases that lose value over time, the company is increasingly positioning them as infrastructure that can generate ongoing rental income for customers who buy them.

To support this thesis, Huang reshared data showing that rental prices for three-year-old H100 GPUs increased 22 percent over one month to $3.28 per hour. This defies the traditional assumption that aging hardware should steadily lose economic value. Nvidia has even partnered with major financial firms to create financing platforms around AI compute, describing it as an "investable asset".

Why Is Jim Chanos Questioning This Strategy?

Jim Chanos, the legendary short seller who famously predicted Enron's collapse and founded Kynikos Associates, responded to Huang's claims with a pointed question: "Then why not rent them out yourself? Or simply keep raising prices?". Chanos later clarified that his skepticism was not directed at Nvidia selling its GPUs, but rather at third-party companies buying Nvidia hardware specifically to rent it out to others.

The core of Chanos's argument touches on a fundamental economic question. If Nvidia's chips truly remain scarce, highly utilized, and capable of producing attractive rental returns, then third-party GPU clouds and so-called neoclouds could potentially capture some of the economic value that Nvidia leaves on the table by simply selling the hardware upfront. In other words, if the chips are as valuable and durable as Huang claims, shouldn't Nvidia be the one profiting from that rental income rather than allowing middlemen to capture it ?

How to Evaluate Nvidia's Productive Asset Thesis

• Scarcity Factor: If GPUs remain in short supply and maintain high utilization rates, they could theoretically generate consistent rental revenue over multiple years, supporting Huang's productive asset narrative.
• Price Stability: Monitor whether rental prices for older GPU models continue to hold their value or eventually decline as newer chips become available, which would test whether these are truly durable assets.
• Margin Capture: Assess whether Nvidia should be capturing rental economics directly through its own leasing programs rather than allowing customers to profit from reselling compute capacity.
• Market Competition: Watch whether other GPU manufacturers or cloud providers can undercut Nvidia's pricing by offering cheaper rental alternatives, which would indicate the market isn't as constrained as Huang suggests.

One commenter on X argued that Nvidia is already effectively "renting out" GPUs through investments, lease arrangements, and financing backstops. However, Chanos pushed back on this interpretation, emphasizing his skepticism about companies buying Nvidia GPUs to rent them out to third parties, not about Nvidia itself selling them.

The exchange highlights a growing debate about who ultimately captures the economics of the AI infrastructure boom. Nvidia's bullish narrative positions its chips as scarce, productive assets that will generate value for years. But Chanos's challenge suggests that if this thesis is truly correct, Nvidia may be leaving significant money on the table by not capturing more of that rental value itself.

On the trading platform Stocktwits, retail sentiment for Nvidia stock remained bearish as of the time of reporting, unchanged from the previous week. Nvidia shares dipped 2 percent on Tuesday amid a broader market selloff, though they were up 0.1 percent in overnight trading.