Nvidia is asking banks to treat its AI accelerators the way the financial system treats commercial aircraft — as long-lived, asset-backed collateral that can be financed, leased and repossessed — and Wall Street is pushing back, according to a Times Square Chronicles report.

The airplane comparison is not incidental. Commercial aviation built a trillion-dollar financing industry on a simple premise: a $100 million-plus airframe has a durable secondary market, standardized maintenance records and an established repossession process. Lessors can value a 15-year-old 737 with confidence. Nvidia's pitch, as the report describes it, is that its data center GPUs deserve similar treatment — that an H-class AI accelerator holds predictable residual value and can underpin lending to cloud providers, neoclouds and enterprises that cannot pay cash for multi-hundred-million-dollar clusters.

Skeptics on Wall Street see the analogy breaking down in several places. Aircraft live for decades. GPUs are at the center of the fastest product cycle in the industry, with Nvidia itself refreshing its accelerator roadmap annually — a cadence that can gut the resale value of a previous generation within a few quarters. A jet that loses an engine can be repaired and returned to service; a GPU cluster that loses relevance to the frontier-model training market is, financially speaking, scrap.

The debate matters because of who buys these chips. A growing share of Nvidia's demand comes from capital-intensive infrastructure builders that finance their GPU purchases with debt. If banks accept the chips as airplane-grade collateral, credit flows more freely and the effective cost of building AI capacity falls. If they refuse, financing tightens and the burden shifts to equity investors and the balance sheets of the hyperscalers.

Wall Street's hesitation, as the report frames it, reflects uncertainty about what an AI chip is actually worth three years after purchase. Aircraft valuation rests on decades of transaction data. AI accelerators have been a mainstream asset class for barely two, and the secondary market is thin, opaque and concentrated among buyers whose own economics depend on continued model-training demand.

There is also the counterparty question. Airlines default, but airplanes keep flying for the next operator. Whether a repossessed GPU cluster finds a new home at a comparable price — or gets undercut by the next silicon generation — is precisely what lenders say they cannot yet model.

For Nvidia, the stakes are commercial, not cosmetic. Easing the financing channel supports demand from customers whose capex already stretches their ability to self-fund, and it would embed Nvidia's product families more deeply into the machinery of credit markets, much as Boeing and Airbus aircraft anchor aviation leasing. The company's leverage is considerable: its accelerators command the overwhelming share of AI training and inference deployments, and scarcity has kept prices and effective utilization high.

But scarcity cuts both ways in a lending discussion. Banks finance airplanes because the asset class survived multiple airline bankruptcies with collateral values intact. Nvidia's GPUs have not yet been through a comparable stress test, and lenders know it.

The outcome will turn on whether a liquid, price-transparent secondary market for AI accelerators emerges — and whether Nvidia's product cadence slows enough for any generation of chips to hold value through a full loan term. Until banks can answer the residual-value question, expect them to keep pricing AI-chip-backed credit with a discount the airplane analogy was designed to eliminate.