NVIDIA Recruits Capital Partners and Supports Infrastructure Build-Out to Accelerate Large-Scale AI Computing

As the AI industry moves beyond model development into the large-scale Inference stage, computing demand centered on AI Factories capable of generating tokens at massive volume is surging. In response, NVIDIA is introducing a new strategy to help startups and enterprises solve the infrastructure problem that requires enormous capital.

Through Revenue-sharing and Credit-support models, NVIDIA is helping AI cloud companies provide services via NVIDIA DSX AI Factories. This allows cloud companies to scale capital-efficiently, while NVIDIA secures a new usage-based revenue stream.

Through this model, model builders and enterprise companies can gain immediate access to full-stack accelerated computing without going through the complex processes of site selection, power procurement, and construction.

The following companies are currently participating as initial partners in this strategy:

• Sharon AI: Planning to deploy up to 40,000 NVIDIA Grace Blackwell GB300 GPUs
• Firmus: Building a 360MW DSX AI Factory campus with up to 170,000 GPUs in Batam, Indonesia

AI-native companies such as Baseten, Fireworks AI, and Together AI urgently need immediate cloud capacity for everything from model training to high-volume agentic inference, and this collaboration is expected to meet that demand.