Key points
- Nvidia is in early talks with insurers
- The coverage would protect lenders to neoclouds
- Used-chip prices are the open question
Nvidia (NVDA) has held talks with insurance companies about taking on part of the financing risk tied to its chips, according to people familiar with the discussions. The talks are at an early stage and may not lead to any deals.
The coverage would protect lenders to smaller cloud companies, often called neoclouds. If one of those borrowers defaults and the Nvidia chips pledged as collateral can't be resold for enough to cover the loan, an insurer would take part of the loss.
In our column on the Fed's "too big to fail" question last week, I kept coming back to who carries the risk in the AI buildout. These talks could bring insurers into that chain. They don't yet tell us whether Nvidia would carry less risk itself.
How the insurance would work
Nvidia has shared data on chip depreciation and the expected future value of computing power with at least one insurer, according to the same people. It's also working with the broker Howden Re on a structure involving insurers. Howden declined to comment.
The deals might not stop at traditional insurers. Nvidia has explored having insurers pass risk on to hedge funds and other alternative investors, and the potential deals could be bigger than even large insurers' balance sheets. It has also looked at joining groups of insurers, hedge funds, and asset managers.
The insurance talks are the third step in about seven weeks. On August 10, Nvidia signed memorandums of understanding with six financial firms, including Goldman Sachs (GS), Apollo (APO), and Blackstone (BX), aimed at mobilizing more than $500 billion of outside capital for AI computing.
Jensen Huang laid out Nvidia's side of it in a blog post the next day. "In some cases, NVIDIA may provide a residual-value support mechanism for up to 25% of an opportunity, assessed carefully on a project-by-project basis," he wrote. Residual value is what the hardware is worth at the end of the financing period. Nvidia could agree to cover part of the gap if that value falls below a specified amount.
A week later, Nvidia agreed to residual value guarantees capped at $105 billion on data center leases tied to OpenAI, starting with a site in Ohio. On Monday, it also added $150 billion to its buyback, leaving $235 billion available for repurchases.
Nvidia is expanding its buyback authorization while exploring insurance for chip-backed lending. What I'd want to know is whether insurers would replace any of Nvidia's guarantees or support more borrowing alongside them.
What a used AI chip is worth
For lenders, a chip's useful life matters because it helps determine what they could recover if a borrower defaults. Huang wrote that the A100, introduced in 2020, "remains in active commercial use" six years later, with customers "extending A100's economic life toward a decade."
He also cited one-year H100 rental prices, which rose from about $1.70 per GPU-hour in October 2025 to about $2.35 in March 2026. "NVIDIA compute is an investable asset," Huang said in the August announcement.
Those rental prices show demand for the chips. They don't establish what a lender could get by selling them after a default. That's the gap I'd want insurers to explain how they're pricing.
For neoclouds borrowing against their hardware, the answer could affect access to financing. We compared CoreWeave's (CRWV) debt with Oracle's earlier today. CoreWeave closed Tuesday at $85.91, up 1%, while Nvidia closed at $227.30, down 0.7%.
The Brookings paper we cited examined AI exposure through banks, insurers, and private credit funds. Chip-loan insurance could connect more of those firms to the same borrowers and equipment.
I'd watch whether a deal gets signed, how much risk the insurer accepts, and what the policy costs and excludes. Those terms would tell us how much protection lenders are actually buying. Our AI Bubble Index tracks circular financing as one of five warning signs drawn from the dot-com era.
