Nvidia is in early-stage talks with insurance companies to cover losses on loans made to AI cloud providers, known as neoclouds, that buy its GPUs, the Financial Times reports. No agreements are in place yet.

Under the proposed structures, insurers would compensate lenders when a neocloud defaults and the GPUs securing the loan cannot be resold at a price high enough to repay the outstanding debt. The chips themselves increasingly serve as collateral in these financing agreements. Nvidia hopes the insurance wrapper will unlock expansion capital for some of its fastest-growing and riskiest customers, who currently struggle to borrow the billions they need.

A shift in Nvidia's customer base

The move reflects a structural change in who buys Nvidia chips. In the early days of the AI boom, demand came almost entirely from investment-grade tech giants — Amazon, Microsoft, Google and Meta — which could fund purchases from their own balance sheets.

Over the past 24 months, a growing share of demand has shifted to neoclouds: AI-specific cloud providers offering on-demand GPU computing. Major players include CoreWeave, Core Scientific, Nscale and Lambda. The sector posted 223% year-over-year revenue growth in the fourth quarter of 2025, according to Synergy Research Group, making it the fastest-growing segment of the data center market.

Unlike hyperscalers, neoclouds almost always need to finance their GPUs and the data centers that house them. That is a tall order for young companies with short track records and largely unproven business models. Lenders have grown wary.

The depreciation problem

Neocloud financing has rested partly on chips and data centers serving as collateral. Data centers hold value over time. Chips do not — they are regarded as short-lived assets that depreciate quickly.

Nvidia CEO Jensen Huang has argued that chips have long useful lives and should be regarded as an "investable asset class." Skeptics of Big Tech's AI spending frequently cite rapid GPU depreciation as a flaw in the economic assumptions underpinning the data center boom.

According to the FT, Nvidia has supplied at least one insurer with data on chip depreciation and on how the market is expected to value future computing capacity. It is working with insurance broker Howden Re to develop a structure for the deals.

The potential size of the transactions exceeds the balance sheet capacity of individual insurers. Nvidia is reportedly exploring structures in which insurers syndicate the risk to hedge funds and other alternative investors. The company has also considered joining these financing consortiums itself.

A pattern of financial engineering

If the insurance deals materialize, they would extend an established pattern. Nvidia has said it offered to backstop financing deals intended to unlock $500B in capital from institutional investors including Goldman Sachs and Apollo. It has provided major capital injections to large customers and increasingly used its balance sheet and credit rating to facilitate deals among cloud providers, neoclouds and digital infrastructure firms.

Huang describes this as building an "ecosystem" of companies that rely on Nvidia products.

Last month, Nvidia facilitated the development of Hut 8's 350-megawatt data center in Nueces County, Texas, for neocloud Lambda. It did so by backstopping a $35B cloud deal between Lambda and Anthropic, the maker of Claude — with Nvidia assuming all risk for Lambda's lease of a facility that will be filled with its own chips.

Growing unease

To some observers, the increasingly tangled financial relationships underpinning GPU demand look fragile. The growing frequency of such deals has raised red flags among investors, drawing accusations of circular financing and comparisons to the "vendor financing" common in the late-1990s dot-com bubble.

Data center industry leaders generally dispute that characterization. Still, the rising share of digital infrastructure investment tied to neoclouds rather than investment-grade tech giants is heightening concerns about the shockwaves a slowdown in AI computing demand would send through the economy.

As the fate of these upstart companies becomes linked to hyperscalers, institutional investors and debt markets, the potential blast radius grows. Bringing insurance companies into the mix is unlikely to ease skeptics' anxiety.