Nvidia just changed the economics of building AI infrastructure. The company announced a new partnership framework with specialized AI cloud providers, combining revenue-sharing arrangements with credit support to accelerate the construction of what it calls “AI factories,” massive data centers optimized for continuous AI workloads.
Think of it like a franchise model for compute. Nvidia provides the hardware and brand ecosystem, cloud partners provide the real estate and energy, and both sides split the upside. Nvidia still collects its standard hardware revenue on top of a cut from cloud service revenues tied to supported capacities.
The scale is staggering
Among the initial collaborators, Sharon AI is projected to deploy up to 40,000 Nvidia Grace Blackwell GB300 GPUs. That’s not a small pilot. That’s a serious commitment to next-generation silicon at production scale.
Then there’s Firmus, which is expanding its campus in Batam, Indonesia to accommodate 360 MW of energy and up to 170,000 Nvidia GPUs.
