Nasdaq-listed chipmaker Nvidia (NVDA) is urging Wall Street to reclassify AI computing power as an investable infrastructure asset—on par with commercial real estate, toll roads, and power plants—rather than a routine technology expense.

On Monday, Nvidia announced it has signed memorandums of understanding with six major financial institutions—Apollo Global Management, Blackstone, BlackRock, Brookfield Asset Management, Goldman Sachs, and KKR—to establish financing platforms capable of eventually tapping more than $500 billion in third-party capital. The six firms collectively manage trillions of dollars in assets and rank among the world's most active infrastructure investors, with portfolios spanning energy networks, transportation systems, and telecommunications. The memorandums are non-binding agreements signaling intent to explore collaboration rather than committed financing.

According to the company, the objective is to position AI compute as a bankable infrastructure asset, thereby incentivizing customers to expand AI data centre capacity while securing long-term demand for Nvidia's hardware. The initiative comes as the capital required to build and operate large-scale AI facilities has reached levels historically associated with traditional infrastructure, with major technology companies already committing tens of billions of dollars to individual data centre complexes.

"This is really the first time that technology chips have become an investable asset class. These are revenue-generating assets now. They're productive, they're long-lived, they're fungible, they're flexible," said Jensen Huang, Nvidia's founder and CEO.

"Fundamentally, what's different about this industry and this way of doing computing is that the computer is now part of the infrastructure, like electricity, like the internet, and so you have to think about it like it's infrastructure," Huang added.

Defining AI Compute

AI compute denotes the raw processing power required to train and operate artificial intelligence models. This workload is handled primarily by specialized chips—overwhelmingly Nvidia's high-end GPUs—which populate the large-scale data centres Nvidia designates as "AI factories."

These facilities consume electricity and data to produce functional intelligent systems: chatbots, image and video generators, drug discovery tools, robotic systems, and a broad array of other applications. As AI models grow more sophisticated, the demand for this specialized computing power increases correspondingly, intensifying pressure on companies to secure access to sufficient compute capacity.

The Paradigm Shift Nvidia Is Advocating

Under current accounting conventions, most enterprises treat the purchase or leasing of computing power as a straightforward technology expenditure recorded on the balance sheet—an asset that depreciates rapidly as newer chip generations reach the market.

Nvidia contends this perspective is outdated. The company argues that its systems are now widely deployed and can serve multiple customers over time, producing revenue streams that span years. On that basis, Nvidia maintains that an AI factory should be classified as a long-term investable asset.

Consider a company that currently requires high-performance AI chips. Under the existing model, it would typically spend millions of dollars from its own cash reserves or secure a conventional business loan to acquire a large cluster of Nvidia GPUs. That expenditure would appear on the balance sheet as equipment, which accountants depreciate over several years as successive chip iterations are released. The underlying assumption is that the AI product developed using those chips will generate sufficient revenue to more than offset their cost before they lose the bulk of their value.

Nvidia's initiative with the six financial firms seeks to replace that pattern with infrastructure-style financing, enabling companies to fund AI compute deployments through investment vehicles rather than traditional equipment purchases. Infrastructure financing of the kind these firms specialize in typically relies on assets with long useful lives and predictable, contracted cash flows—models used to finance power plants, pipelines, and telecommunications networks. Whether AI compute assets can meet those criteria over multi-decade horizons, particularly given the rapid pace of chip generation cycles, will likely shape how institutional investors approach the proposed asset class.