Nvidia Revenue-Share Model Names 210,000 GPUs But Leaves Split Undisclosed
Nvidia is offering a revenue-sharing and credit-support model for AI cloud partners that use its infrastructure. Sharon AI and Firmus Technologies are named partners with up to 210,000 GPUs combined, while the public record still lacks the revenue-split percentages.
Nvidia Adds Revenue Share To AI Cloud Financing
Nvidia is offering a revenue-sharing and credit-support model for AI cloud companies that build services on its infrastructure.
The model lets participating AI clouds procure Nvidia systems and sell Nvidia-powered cloud services to end customers.
Nvidia would collect product revenue on the hardware and a percentage of income from cloud services running on that capacity.
Nvidia described the model as a way to widen compute access for startups that cannot finance large AI infrastructure deployments on their own.
Developers receive token credits in exchange for part of future sales, while the revenue split has not been disclosed.
Sharon AI And Firmus Name 210,000 GPUs
Nvidia listed Australia's Sharon AI and Singapore-based Firmus Technologies as partners in the model.
Their combined plans cover up to 210,000 GPUs.
Sharon AI disclosed in an 8-K filing dated June 12 that its agreement runs for six years.
The filing covers 72 MW of new Australian data-centre capacity built to Nvidia's DSX AI factory design and scaling to as many as 40,000 Grace Blackwell GB300 GPUs.
Sharon AI also has a revenue-share facility of up to $200 million with Digital Alpha, disclosed in its CY25 results.
Firmus is building a DSX-aligned campus in Batam, Indonesia, expected to scale to 360 MW and house up to 170,000 Nvidia GPUs.
Nvidia Revenue Depends On Cloud Usage
Under the model, Nvidia would receive hardware revenue and a percentage of income from cloud services running on the installed capacity.
If partner clouds fill racks and sell compute, Nvidia can receive recurring income from the cloud capacity as well as hardware revenue.
The model still depends on partner clouds keeping rented GPU capacity in use while the hardware depreciates.
The public record leaves the revenue-split percentages, customer contracts tied to the named GPU capacity, utilisation commitments or the full financing terms behind the model undisclosed.
