Nvidia launches revenue-sharing AI cloud financing

Nvidia has introduced a revenue-sharing financing model to help AI cloud providers buy its computing infrastructure, broadening access to large-scale AI compute for startups, enterprises, researchers and regional providers.

The model is aimed at a market shifting from model development to production inference, where computing systems run continuously to generate AI output at scale. Under it, AI cloud operators procure Nvidia infrastructure with credit support and revenue sharing tied to the cloud services sold on that capacity.

Nvidia will receive its usual product revenue from infrastructure sales, along with a share of cloud revenue generated from the supported capacity. The structure gives AI cloud operators a way to finance large installations that have often been difficult to fund, especially for emerging companies without the balance sheet strength typically required for capital-intensive projects.

The move reflects a broader change in the AI market. Demand is no longer focused only on training large models, but also on running them in production for enterprises, software providers and AI-native businesses that need constant access to computing resources for inference, fine-tuning and related workloads.

Early adopters

Two companies are among the first to work with Nvidia under the model. Sharon AI is deploying up to 40,000 Nvidia Grace Blackwell GB300 GPUs, while Firmus is developing a DSX AI factory campus in Batam, Indonesia.

The Batam campus is expected to scale to 360 megawatts and up to 170,000 Nvidia GPUs. The projects show the size of the infrastructure Nvidia wants to make easier to finance and bring online through a structure linked to future usage, rather than relying solely on conventional long-term commitments.

Nvidia described these facilities as DSX AI factories built around customer demand in different regions. For AI infrastructure users, the pitch is faster access to computing capacity instead of waiting for site selection, power procurement, construction and hardware installation through traditional buildouts.

Access to compute has been a persistent constraint across the AI sector. Many young companies building models, inference services or agent platforms need large amounts of hardware but have struggled to secure the capital required to lock in supply, even when they can demonstrate likely demand.

Nvidia is trying to address that bottleneck by aligning its economics more closely with cloud providers that serve those customers. In effect, it is moving beyond a straightforward equipment sale and taking an ongoing financial interest in the use of the systems it helps place.

Market shift

The initiative also shows how suppliers are adapting to the economics of AI services. As more companies move from experiments to commercial products, they want computing supply that can expand with usage while preserving flexibility in how they pay for capacity during the shift from pilot projects to full deployment.

Nvidia highlighted model builders, inference providers and enterprises that need immediate cloud access for training, post-training, fine-tuning and high-volume agentic inference. It also pointed to AI-native businesses such as Baseten, Fireworks AI and Together AI as examples of where demand is heading.

These groups depend on reliable access to accelerated computing as customer usage rises. At the same time, they often need more adaptable commercial arrangements than traditional data centre buildouts or fixed infrastructure commitments allow.

For Nvidia, the model creates a recurring income stream tied to usage of supported cloud capacity, rather than limiting returns to the initial hardware sale. That could deepen its role in the AI cloud market as competition grows among providers seeking to serve developers, digital-native companies and established enterprises.

Sharon AI described the agreement as a milestone for its plans. "This strategic collaboration with NVIDIA marks a pivotal moment in Sharon AI's mission to deliver sovereign, large-scale AI compute infrastructure," said James Manning, co-founder and CEO of Sharon AI.

Firmus cast the issue in terms of infrastructure access for internationally competitive AI businesses. "AI-native companies need access to scalable, energy- and cost-efficient compute infrastructure to compete globally," said Tim Rosenfield, co-CEO of Firmus Technologies.

"Firmus AI cloud is building a NVIDIA DSX-aligned AI factory, which will enable our cloud to help more customers access the compute they need to build and scale AI," Rosenfield said.