Jensen Huang’s plan faces China risk

Jensen Huang built the world’s most valuable company by pioneering the specialized computer chips behind the artificial intelligence boom.

To keep his vision for the future within reach, the Nvidia founder is now attempting a different kind of engineering: convincing Wall Street investors that those chips are long-term financial assets akin to commercial real estate or toll roads.

His bet hinges on outpacing AI developments in China.

This week, Nvidia unveiled agreements with six of the world’s largest asset managers, BlackRock, Blackstone, Apollo, KKR, Brookfield and Goldman Sachs. The goal was to assemble a $500 billion pipeline to finance the construction of data centers and GPU clusters for companies that lack the credit rating or cash to buy millions of dollars of silicon outright.

Key to his plan, which Huang announced during a CNBC segment flanked by the leaders of all six Wall Street firms, is one crucial assumption: that Nvidia’s graphics processing units will hold their value over time, behaving more like traditional hard assets than fast-depreciating consumer electronics.

“Nvidia’s AI factory platform is really an investable asset, an infrastructure asset,” Huang said. “The reason for that is because it’s productive, it’s revenue generating, it is fungible, it’s used by just about every cloud service provider, it runs every AI model.”

In standard asset-backed finance, a bank lends money because if a borrower defaults, the bank can repossess the asset — like a building, a warehouse or a cargo ship — and sell it to get their money back. Those physical assets have established secondary markets and can last decades.

But the productive lifespan of cutting-edge GPUs is far from settled.

While new chips power frontier model training, after a few years they are relegated to lower-margin inference work — a shift that directly impacts their resale and collateral value.

“Depreciation is the one key risk here,” said Ben Emons, founder of FedWatch Advisors, who structured similar asset-backed loans for IndyMac before joining Pimco as a portfolio manager. Nvidia chips “could depreciate faster than expected,” he said.

In particular, Emons said he believes the single biggest threat to Nvidia’s financing model comes from China, which is rapidly ramping up domestic compute capacity and could choose to flood the market with low-cost silicon in a price war.

If Chinese production pushes hardware prices into a freefall, the collateral backing hundreds of billions in private loans could erode far faster than the terms of the debt itself, leaving investors exposed to losses, according to Emons.

To compensate at least partly for that risk, Emons estimates investors will treat GPUs as high-depreciation equipment rather than real estate, demanding high-yield returns in the 11% to 17% range depending on where they sit in the capital structure.

On top of that, the borrowers are likely to be non-investment grade firms locked out of traditional debt markets, including AI startups and neoclouds, according to a Bank of America Securities note.

If those higher-risk borrowers go under, Wall Street fund managers will be forced to repossess and resell used chips into a potentially falling market.

Whatever risks China poses wouldn’t be realized anytime soon. Huawei, the dominant provider of Chinese AI chips, has been on the U.S. Commerce Department’s Entity List since 2019. And in May, the U.S. government said Huawei’s Ascend AI chips violate U.S. export controls, preventing any American company from using the chips.

In the meantime, Nvidia remains by far the leading supplier of AI chips in the U.S., with upwards of 75% market share by most estimates.

And for now, the economics are still moving in Huang’s favor. Driven by scarcity as hyperscalers race to build out capacity, rental rates for Nvidia’s H100 chips rose from roughly $1.70 per GPU-hour in late 2025 to about $2.35 per GPU-hour this year, Huang noted.

Crucially, Nvidia argues its CUDA software layer — which enables developers to run AI workloads on its GPUs — continuously improves hardware performance after deployment, allowing older chips to stay productive and generate yield longer than traditional accounting models predict.

The future of the AI buildout, and hundreds of billions of dollars in investor money, may depend on who is right.

— CNBC’s Ari Levy contributed to this report.