Nvidia introduces revenue-sharing and credit support program to help AI startups access GPU compute

Nvidia is rolling out a program aimed at making it easier for AI startups to obtain large-scale computing resources, using a revenue-sharing structure tied to cloud and product economics rather than traditional upfront costs. The initiative is designed to address a recurring bottleneck for new AI businesses, which often need extensive GPU capacity but can struggle with liquidity and the high price of acquiring compute on demand.

Under the program, Nvidia said cloud-based AI firms and other enterprises can receive token credits to run and develop their services. In return, participants would share both product and cloud revenues with Nvidia, positioning the company less as a chip-only supplier and more as an intermediary that can align incentives across the AI stack.

Nvidia’s plan also builds around partnerships with AI infrastructure providers. CNBC reported that Nvidia named two initial Australian partners that will supply compute for the scheme, including Sharon AI and Firmus Technologies. The company described these partners as providing access to a combined pool of GPUs through the early rollout.

According to the reporting, Sharon AI plans to deploy up to 40,000 Nvidia Grace Blackwell GB300 GPUs as part of its contribution to the program. Firmus Technologies, meanwhile, is developing a data center in Batam, Indonesia, expected to scale to 360 megawatts and house up to 170,000 Nvidia GPUs, with the intention of expanding capacity for AI compute demand.

The new approach effectively ties Nvidia’s long-term revenue opportunities to whether customers succeed in commercializing their AI services. By taking a cut of future revenues while providing credits to access compute, Nvidia can potentially recapture more value than it would through hardware sales alone, while also reducing the upfront financial burden for startups that need expensive infrastructure to iterate models.

Nvidia’s announcement also reflects a broader sector shift toward revenue and equity-sharing arrangements between AI builders and infrastructure providers. As competition for GPUs has intensified and computing costs can be volatile, startups have increasingly sought structures that preserve cash early while still creating a pathway to monetize later.

Even with specific examples of partners and capacity targets, key terms were not fully disclosed in the public coverage referenced here. Details such as the exact credit amounts, eligibility rules, revenue-share percentages, duration of agreements, and how performance or usage limits are handled were not included in the cited reports, leaving uncertainty about how the economics will work for different types of AI workloads.

For Nvidia and its ecosystem, the next announcement to watch is whether additional cloud and data center partners join beyond the initial providers, and how quickly startups convert credits into deployable products. Observers will also look for signs that the model increases sustained demand for Nvidia-powered infrastructure, rather than just enabling short-term experimentation.