Nvidia Courts Insurers to Underwrite AI Chip Financing Risk

Nvidia has opened talks with insurance companies about underwriting some of the risks tied to lending against its artificial intelligence chips, a move that would extend the chipmaker’s push to draw outside capital into the financing of the AI build-out. The discussions, which are at an early stage and may not produce deals, are part of an effort by chief executive Jensen Huang to broaden the pool of customers who can buy Nvidia’s semiconductors beyond the handful of Big Tech groups with the balance sheets to fund vast data centre projects.

The structures under consideration would shift some of the risk of capital-intensive semiconductor financing to insurers and other investors, according to people familiar with the talks. One idea would insure lenders against losses on loans to “neoclouds,” the upstart cloud computing companies that lack the financial firepower of their larger rivals. If such a borrower defaulted and the Nvidia chips pledged against the debt could not be resold for enough to repay the loan, the insurer would cover the shortfall. Protection of that kind could encourage more capital to flow to these smaller customers.

Huang has argued that chips should be treated as an “investable asset class,” comparable to aeroplanes or other expensive, long-lived technology that supports complex financing structures to shift risks and costs between users and investors. The insurance push is the latest front in that effort. Nvidia has already offered to backstop a portion of financing deals intended to unlock $500bn of capital from Wall Street firms including Goldman Sachs and Apollo, and has guaranteed $105bn of leases to build a large data centre for OpenAI. The company has told investors it expects a quarter of its revenue next year to come from AI labs supported by its own balance sheet.

The insurance discussions mark a new direction. Nvidia has shared data on chip depreciation and expected future computing prices with at least one insurer, and is working with broker Howden Re on developing a structure involving insurers, according to people familiar with the matter. The effort is led by Ingemar Lanevi, Nvidia’s head of financial solutions. Nvidia has also explored using insurance groups to syndicate risk to hedge funds and other alternative investors, because the potential scale of the deals could overwhelm the balance sheets of even large insurers.

The move reflects a broader trend of insurers launching products aimed at the AI infrastructure build-out, including coverage for credit risk, falls in chip values, and contract breaches caused by power outages or cooling failures at data centres. For Nvidia, bringing in insurers serves a dual purpose: it reduces the risk that a default by a neocloud customer would force the company to absorb losses, and it signals to other capital providers that chips are a bankable asset.

The challenge is that the risks are largely untested. Chip values can be volatile, and the market for reselling high-end AI accelerators is thin and opaque. Insurers will need to price a risk that has little historical data, and they will be relying on Nvidia’s own projections for future computing prices. The company’s willingness to share that data suggests it sees insurers as a way to validate its claims about the long-term value of its products.

The outcome of the talks will be watched closely by the financial institutions that have already committed capital to the AI build-out. If insurers can be persuaded to take on some of the risk, it could unlock further investment in neoclouds and other AI infrastructure projects. If not, the burden will continue to fall on Nvidia’s own balance sheet, and on the handful of large firms that have so far shouldered the cost.

That leaves Nvidia in a delicate position. It is simultaneously the largest supplier of the technology underpinning the AI boom and a growing source of financing for that boom. Spreading the risk more widely, through insurers and other investors, would ease the strain on its own books and widen the market for its chips. But it depends on convincing a conservative industry to underwrite a technology whose value is still being proven.