AI infrastructure has a math problem. Rent-a-GPU providers such as CoreWeave and Lambda need to borrow billions of dollars from venture capitalists and hedge funds to build datacenters, and they only stay profitable if revenue can clear interest payments. That financing constraint is why not every ambitious AI cloud idea can get off the ground, even if the demand hype is real.

Nvidia’s answer, in a blog post published this week, is a new program design that aims to make it easier for emerging AI cloud providers to access the financing they need. But Nvidia’s pitch is not just “we’ll help you raise money.” It also expects a cut of the revenues: “Through the partnership, AI clouds will sell Nvidia-powered cloud services, with Nvidia earning both standard product revenue and a share of the cloud revenue on the supported capacity,” the company explained. In plain terms, Nvidia wants to earn once when hardware is sold, and again later if the neocloud proves out and generates cloud revenue.

This is a subtle but consequential shift in incentives. Traditional GPU buyers have a straightforward relationship with Nvidia: you purchase GPUs, you build capacity, you monetize through your own customers. Nvidia’s proposed structure adds a second layer where its earnings become usage-linked, tied to the supported capacity being sold as cloud services. That matters because “supported capacity” and cloud revenue are closer to customer demand than the initial hardware transaction, which could change how both Nvidia and its partners think about adoption cycles.

It also creates a potential stabilizer for Nvidia if the market cools. The source notes that the approach could provide some insulation against a potential AI bust. If demand for new GPUs falls, Nvidia may still be able to earn a recurring revenue from the GPUs it has already sold, assuming customer demand remains high. In other words, Nvidia is trying to make part of its revenue stream behave more like subscription economics than pure hardware cycles.

Still, the practical details are thin, and that is where governance and capital allocation questions start to matter. Nvidia declined to offer details beyond the contents of its blog post. Importantly, the source says it is not clear whether Nvidia itself will provide the financing. It may instead be brokering deals with third-party lenders. That distinction changes the risk profile for AI cloud providers and for anyone in the ecosystem thinking about default risk, covenant pressure, and who ultimately bears the cost of a stalled rollout.

To stress-test the model, Nvidia says it has already signed up two customers. Sharon AI and Firmus are the early deployments. Sharon AI is a sovereign AI cloud provider founded in 2024 based out of Australia. It is looking to deploy as many as 40,000 Grace Blackwell GB300 GPUs in the land down under. Firmus plans to deploy as many as 170,000 Nvidia GPUs at a 360-megawatt facility in Batam, Indonesia. The facility is designed specifically to Nvidia’s DSX spec.

The numbers are big enough to make this more than a theoretical blog post. CoreWeave and Lambda’s experience, the source reminds us, is that building capacity requires deep financing and careful execution to keep revenues above interest payments. If Nvidia can structure partnerships so that emerging AI cloud providers get access to capital more easily, it potentially reduces a bottleneck across the whole supply chain, from GPU procurement to datacenter build outs to customer service launch timelines.

For executives at AI infrastructure firms, this raises a familiar but sharper question: what do you trade for financing? If Nvidia’s cut is tied to supported capacity, then the financing problem might shift from “can we afford interest payments” toward “can we afford to share upside tied to cloud revenue.” For investors and boards, it means diligence may need to expand beyond unit economics to include partner revenue-share structures, capacity definitions, and the likely behavior of lenders if the business plan depends on Nvidia-powered service adoption. And for Nvidia’s competitors and counterparties, the biggest stake is whether this creates a durable move toward recurring, usage-linked earnings that competitors cannot easily replicate, especially during periods when buyers get more cautious about scaling.