NVIDIA's new compute model expands cloud AI access with revenue-sharing

NVIDIA is expanding its AI infrastructure business with a new compute model designed to accelerate the deployment of computing capacity across global cloud providers. The framework allows cloud partners to deploy NVIDIA's advanced hardware using revenue-sharing and credit-support agreements, shifting away from traditional upfront purchasing. This model is intended to reduce infrastructure barriers for AI startups, enterprises, and model developers who need access to high-end GPU clusters but face high capital costs. By offering flexible financial arrangements, NVIDIA aims to increase utilization rates of its hardware and make advanced computing resources more accessible to a broader range of AI innovators.

The move comes as demand for AI compute continues to outpace supply, with startups and even established companies struggling to secure GPU capacity. NVIDIA's revenue-sharing approach could allow cloud providers to scale up their deployments more quickly, passing on flexible pricing to end users. This is particularly beneficial for AI startups that need sporadic bursts of high compute power for training and inference, but cannot commit to long-term leases. With credit-support agreements, NVIDIA essentially underwrites some of the risk for cloud partners, accelerating hardware rollout. The initiative positions NVIDIA not just as a chip supplier but as an infrastructure enabler, potentially reshaping how AI compute is accessed and priced in the cloud.