AI
Nvidia's $500B collateral guarantee could reshape the used GPU market
By backing the resale value of older chips, Nvidia is betting on long-term infrastructure stability and a thriving secondary market for aging hardware.
TechCrunch — AI
Nvidia announced this week that a coalition of heavyweight financial firms—Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR—are prepared to channel up to $500 billion into building AI data centers. While the headline figure has drawn widespread attention, the deeper story centers on Nvidia's ambitious push to establish a functioning secondary market for aging graphics processors.
To secure buy-in from these institutional investors, Nvidia has taken an unusual step: the company is pledging its own capital to backstop the resale value of its chips when they serve as collateral in these financing deals.
The Guarantee and Its Mechanics#
Under the arrangement, Nvidia commits to covering up to 25 percent of any shortfall if GPUs used to secure loans drop below their book value. Should a data center operator default and a lender be forced to liquidate the hardware at distressed prices, Nvidia will absorb part of the loss. Observers have noted both the ingenuity and the peril embedded in this arrangement.
The move has rattled bond markets enough that CEO Jensen Huang took to social media and cable television to clarify the company's actual risk exposure and explain that independent capital, rather than Nvidia's own money, would shoulder the bulk of the burden.
Building on Established Patterns#
The new $500 billion plan builds on financing schemes Nvidia has already been pursuing. The company has previously funneled billions toward customers including frontier AI labs OpenAI and Anthropic, as well as emerging cloud operators like CoreWeave, Nebius, Firmus, and Lambda. Research suggests Nvidia has also engineered up to $750 billion in additional circular financing arrangements over the summer months.
Critics have drawn parallels to Lucent Technologies, the telecommunications equipment maker that famously collapsed after lending to its own customers during the dot-com bubble. Huang has defended the current structure by emphasizing that institutional investors—not Nvidia—are providing the primary capital and taking on most of the risk.
The Broader Bet on Durable Infrastructure#
What makes Nvidia's approach distinctive is the underlying philosophy: the company is positioning AI computing hardware not as a depreciating asset, but as persistent infrastructure akin to railroads or telecommunications networks. By guaranteeing that older chips retain utility and value, Nvidia aims to nurture an ecosystem in which enterprises and startups have confidence in using pre-owned equipment alongside the latest generation.

