Nvidia & Wall Street launch $500bn AI funding push

Nvidia has partnered with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to create financing platforms for AI infrastructure. The group said the platforms are intended to mobilise more than USD $500 billion of third-party capital over time.

The move is an effort by the chip designer to widen access to funding for AI-linked data centre buildouts, as demand for computing capacity outgrows the balance sheets of the largest cloud providers and technology groups.

The financing platforms will support what Nvidia calls AI factories, or large-scale computing installations built around its chips, networking, software and developer tools. It argues these systems should be treated as productive infrastructure rather than one-off equipment purchases.

Under the structure outlined by Nvidia, the financial institutions will independently assess each project, including the customer, expected demand, utilisation, cash flow and residual value. The more than USD $500 billion figure is not company revenue, a single fund or a commitment to one customer.

Funding model

The initiative reflects a broader shift in the AI market from research-led spending to industrial-scale deployment. Many AI companies, enterprises and specialist cloud operators have demand for compute but lack access to financing at the scale or cost needed to expand quickly, Nvidia said.

That creates an opening for infrastructure investors accustomed to backing long-lived assets with recurring income. Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR are among the world's largest alternative asset managers and financiers, with extensive exposure to energy, transport, digital infrastructure and private credit.

Nvidia's role will centre on providing the AI factory platform, while its capital partners provide long-term funding and underwriting expertise. It also sought to address concerns that such arrangements could amount to circular financing by stressing that investors will make independent decisions on each opportunity.

In some cases, Nvidia may provide a residual-value support mechanism for up to 25% of an opportunity on a project-by-project basis. It described that support as limited and tied to residual value, rather than a substitute for investor underwriting.

Asset case

A central part of Nvidia's argument is that AI compute should now be viewed as an investable infrastructure asset. The company said its systems can serve multiple customers and workloads, making them easier to redeploy if demand shifts between operators, cloud providers or end users.

It also argued that software updates extend the economic life of installed hardware. Nvidia pointed to its A100 processor, introduced in 2020, and said it remains in active commercial use for training, fine-tuning, inference and high-performance computing six years later, with customers still committing capacity for multi-year deployments.

To support its claim that demand for advanced AI hardware remains strong, Nvidia cited pricing data. One-year H100 rental pricing rose from about USD $1.70 per GPU-hour in October 2025 to about USD $2.35 per GPU-hour in March 2026, while cross-provider on-demand median pricing increased from roughly USD $2.00 per GPU-hour in October 2025 to USD $2.70 in June 2026.

Nvidia added that Blackwell capacity attracts higher rates, with reported B200 cloud pricing ranging from about USD $5.30 to USD $7.05 per GPU-hour. Those figures suggest newer chips continue to command a premium even as earlier generations remain in service.

Broader market

The financing push comes as investors and technology groups debate whether the AI boom will deliver sustainable returns on the heavy capital spending now under way. Nvidia argues that compute demand is tied to commercial uses of AI across software development, drug discovery, product design, customer service, industrial automation and new digital services.

The relevant question, Nvidia said, is not simply whether more data centres are being built, but whether those facilities can generate revenue from useful AI workloads. In its view, that depends on building around real customer economics and subjecting each project to discipline on demand, utilisation, cash flow and residual value.

Nvidia also cast the partnerships as the start of a broader capital market for AI infrastructure, in which institutional investors finance computing assets much as they do other major networks such as electricity, transport, communications and traditional computing infrastructure.

The announcement underlines how far Nvidia has moved beyond selling chips into a broader role in shaping the economics of AI deployment. It is seeking not only to supply the hardware and software behind the current buildout, but also to help define the financial structures that determine who can afford to take part.