NVIDIA adds revenue-sharing model to expand AI cloud capacity

NVIDIA is introducing a business model aimed at widening access to large-scale accelerated computing as AI demand shifts from model development to production inference. The company says emerging AI companies have often struggled to secure capital-intensive infrastructure, even when long-term commitments were available.

The model lets AI clouds procure NVIDIA infrastructure for AI-native, enterprise and ISV customers through revenue sharing and credit support. AI cloud providers will sell NVIDIA-powered cloud services, while NVIDIA earns standard product revenue and a share of cloud revenue tied to supported capacity, creating a recurring, usage-linked earnings stream.

Sharon AI and Firmus are among the first companies working with NVIDIA under the model. Sharon AI is deploying up to 40,000 NVIDIA Grace Blackwell GB300 GPUs. Firmus is building a DSX AI factory campus in Batam, Indonesia, expected to scale to 360 megawatts and up to 170,000 NVIDIA GPUs.

NVIDIA frames the approach as a way for model builders, inference providers, agent platforms and enterprises to access full-stack compute without waiting on site selection, power procurement, construction and hardware deployment. Companies such as Baseten, Fireworks AI and Together AI are cited as examples of AI-native demand for training, post-training, fine-tuning and high-volume agentic inference capacity.