Nvidia funding plan raising some eyebrows

US semiconductor company Nvidia's bold proposal to mobilize $500 billion from Wall Street for artificial intelligence infrastructure has drawn skepticism, with industry observers questioning whether the plan reflects genuine demand, or a financial move to prop up sales.

At the center of the debate is a simple but troubling question: if AI computing power is truly in such short supply, why does the chipmaker need to help its customers arrange financing? Some analysts suggest the arrangement may be designed to unlock orders that would otherwise remain on hold due to tight budgets.

The plan came into sharper focus on Monday, when Nvidia CEO Jensen Huang joined the chiefs of six major financial institutions — Goldman Sachs, BlackRock, Blackstone, KKR, Apollo Global Management and Brookfield — in a CNBC interview. Each firm has signed a separate memorandum of understanding to set up its own funding vehicle, channeling money from third-party investors into AI data centers and related hardware. The target is $500 billion, with the potential to grow even larger.

Huang called the initiative the "big idea", positioning AI computing capacity as a new asset class in its own right.

"These systems are not like our PCs, not like our phones. They are now revenue-generating assets — productive, long-lived, fungible and flexible," he said.

David Solomon, CEO of Goldman Sachs, said that asset-backed lending for such infrastructure is a logical step, precisely because the underlying facilities are tangible and hold genuine economic value.

Under the agreements, each lender will make its own credit decisions, while Nvidia will act as a matchmaker between customers and financiers. The company may also take on up to 25 percent of the risk per loan, though Huang later clarified on social media that such support is based on residual value and is meant to complement, not replace, independent underwriting.

Xiang Ligang, director-general of the Information Consumption Alliance, a telecom industry association, pointed to the divergent market reactions as evidence of underlying unease. While asset managers saw their shares rise on expectations of fee income, major tech counters continued to fall after the news.

That divergence, Xiang said, suggests that investors are not fully convinced by the AI demand narrative. If cloud giants such as Google and Microsoft were the clearest beneficiaries of a genuine computing shortage, their stocks would have rallied. Instead, Nvidia's push to engineer financing solutions raises a red flag that potential buyers may be struggling with balance-sheet constraints, he added.

Nvidia itself has argued that the financing platform addresses a new bottleneck: capital. Emerging AI labs, smaller cloud providers and sovereign AI projects have real needs but lack the deep pockets of hyperscalers. However, Xiang cautioned that the entire model rests on an untested assumption — the long-term salvage value of GPUs used as collateral.

Nvidia now refreshes its chips annually rather than every two years, meaning each new generation quickly erodes the resale value of older chips, experts said.

Meanwhile, the announcement on Monday only involved nonbinding MOUs, offering no details on borrowers, interest rates, project locations or timelines. That lack of specificity has left many investors waiting on the sidelines, seeking concrete evidence that the "big idea" can withstand real-world economic scrutiny, experts added.

masi@chinadaily.com.cn