Nvidia, the company that has profited more than any other from the artificial intelligence boom and controls an estimated 80% or more of the market for AI accelerators, is now turning to Wall Street to ensure the money keeps flowing.

On Monday, Nvidia announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to establish financing platforms designed to mobilize more than $500 billion for AI infrastructure. The capital will come primarily from third-party investors, enabling Nvidia's customers to finance chips and data centers while keeping Nvidia's own exposure limited and off its balance sheet.

The specific terms of each arrangement remain undisclosed. However, analysts have long anticipated a deal of this nature—one that reclassifies AI compute as an infrastructure asset comparable to a toll road or power plant, generating steady cash flows capable of supporting debt. Until now, a prevailing concern has been that GPUs function more like a rapidly depreciating stockpile of processors requiring ever-increasing capital infusions.

"We have moved from an era in which companies bought chips and built data centers project by project to one in which AI factories can be financed as productive infrastructure," Nvidia CEO Jensen Huang wrote on Tuesday. His formulation is succinct: "In AI, compute is revenue."

Beneath this shift lies another fundamental change: who is actually bankrolling the AI buildout.

A year ago, most of Big Tech could credibly claim it was funding AI expansion through enormous cash flows accumulated over decades of high-margin software businesses and massive profits. Debt has since taken center stage. Goldman Sachs estimates that AI-related financing now accounts for nearly one-quarter of all gross U.S. investment-grade bond issuance, while total AI investment is approaching $600 billion this year alone. The financing push also comes as the industry confronts physical constraints beyond capital: AI data centers demand unprecedented amounts of electricity, and grid capacity has emerged as a bottleneck for expansion in regions from Northern Virginia to Ireland, with the International Energy Agency projecting that global data center electricity consumption could reach double its 2022 level by 2026.

Nvidia is moving proactively to identify the next deep pool of capital. The mechanism is straightforward: an independent financing vehicle raises funds to purchase Nvidia GPUs and data-center infrastructure. An AI company then leases or commits to using that compute capacity, generating a payment stream against which the vehicle can borrow. Firms like Apollo and KKR structure and manage that debt, placing it with the vast institutional pools they oversee—primarily insurance and retirement capital.

Bloomberg columnist Matt Levine distilled the long-term vision into three steps: channel more private investments into ordinary people's retirement accounts, raise enormous sums of private-credit and infrastructure funds, and deploy that money to build the data centers that AI companies will rent.

The appeal of these capital pools is structural. Data centers are expensive, long-lived projects requiring multi-year financing. Insurers and pension funds, in turn, carry long-dated obligations—annuities that may pay out for decades, or retirement benefits owed far in the future. They naturally seek long-duration assets whose cash flows can be matched against those liabilities. Private-credit and infrastructure managers serve as intermediaries, transforming projects like data centers into debt instruments these institutions can hold. The model has historical precedent in the 1990s telecom buildout, when fiber-optic networks were similarly securitized and placed with institutional investors—an era that produced rapid national infrastructure expansion before overcapacity contributed to a wave of defaults and bankruptcies.

Nvidia has indicated it is contributing its own resources to the effort. Huang said the company may provide residual-value support of up to 25% for certain projects—effectively offering partial protection against the risk that the chips underpinning a financing deal lose significantly more value than lenders anticipated.

Ben Thompson, author of the technology strategy publication Stratechery, characterized this as "in a certain sense, a price cut": Nvidia is leveraging its own profits to lower customers' cost of capital and make Nvidia-based data centers easier to finance.

The evolution is notable. The AI boom began with some of the wealthiest corporations in history spending their own cash. Then came bonds. Now Nvidia is helping Wall Street transform compute itself into an investable asset class capable of attracting insurance floats, pension funds, and other long-duration savings. The strategy also arrives amid intensifying competition, with AMD ramping production of its MI300 series accelerators and major cloud providers including Google, Amazon, and Meta investing in custom silicon designed to reduce their dependence on Nvidia's platform.

Wall Street analysts view the arrangement favorably. Morgan Stanley's Joseph Moore said it alleviates concerns about circular financing, since third-party investors will provide the bulk of the capital while Nvidia participates only to a limited degree. Bank of America's Vivek Arya largely concurred, adding that Nvidia's chips are unusually financeable because GPUs can be redeployed among operators and the company's CUDA software can extend their useful lives.

Thompson, however, raised a broader concern. The institutional capital pools being tapped are fundamentally different from venture capital or technology equities: they are designed, at least in part, to pursue safety.

"It's one thing to spend all of your free cash flow; it's another thing to tap the debt markets," he wrote. "And, beyond that, it's a completely new nerve-racking thing to bring safety-seeking assets to bear."