Nvidia launches AI revenue-sharing model

Nvidia (NVDA) has launched a revenue-sharing and credit-support model intended to widen access to its AI computing infrastructure. The arrangement allows AI cloud providers to procure Nvidia hardware and sell Nvidia-powered cloud services, while Nvidia receives its usual product revenue plus a share of cloud revenue generated on supported capacity.

Chief Financial Officer Colette Kress outlined the approach in a company blog post, framing it as a mechanism to connect AI data-center operators with cloud service providers. The structure is designed to establish a recurring, usage-linked earnings stream for Nvidia tied directly to how much capacity customers consume.

The program is aimed at researchers and nascent companies that lack the capital to build or lease large-scale, continuously available AI infrastructure. By combining credit support with a revenue-sharing framework, Nvidia seeks to lower the up-front cost barrier these organizations face when accessing advanced GPU resources.

Under this model, cloud partners can extend Nvidia-powered capacity to a wider range of AI developers while sharing economic upside with Nvidia. The company positions the initiative as a way to accelerate adoption of its full-stack accelerated computing offerings among early-stage and experimental AI projects.

Nvidia and its partners are linking the new commercial structure to DSX-branded multi-tenant AI factories. These facilities are presented as large-scale environments where many customers can tap shared pools of Nvidia GPUs under flexible commercial terms supported by the new model.

Sharon AI is identified among the first participants, with plans cited for deployments of up to 40,000 Grace Blackwell GB300 GPUs. Firmus is another early collaborator, planning a DSX AI factory campus in Batam that is expected to scale to about 360 megawatts of capacity and up to 170,000 Nvidia GPUs.

These deployments illustrate the scale Nvidia envisions for DSX AI factories operating under the revenue-sharing framework. They also signal how the program could support substantial expansions of AI infrastructure while aligning economics among Nvidia, infrastructure operators, and cloud service providers.

By tying part of its returns to cloud revenue generated on Nvidia-powered capacity, the company is building a new recurring earnings stream. This usage-linked structure complements traditional hardware sales and is intended to track the ongoing intensity of AI workloads running on partner infrastructure.

Linking data-center operators, cloud platforms, and capital-constrained AI developers under one commercial umbrella may help standardize how large-scale GPU resources are financed and consumed. The early DSX partnerships suggest Nvidia is positioning this model as a template for future AI infrastructure expansions.