Nvidia has unveiled a paradigm-shifting business model for AI infrastructure, moving beyond traditional hardware sales to embrace revenue-sharing and credit-support mechanisms. By partnering with AI cloud service providers to build large-scale, multi-tenant 'AI Factories,' the Silicon Valley giant is positioning itself as a central pillar of the global computing utility. This new framework allows Nvidia to generate recurring income tied directly to compute usage while accelerating the deployment of its most advanced hardware.
Under this initiative, partners will deploy the DSX AI Factory architecture, specifically optimized for the demands of modern generative AI. Early adopters include Sharon AI, which is slated to deploy up to 40,000 NVIDIA Grace Blackwell GB300 GPUs, and Firmus, which is developing a massive 360-megawatt campus in Batam, Indonesia. The Firmus project represents one of the largest infrastructure commitments to date, with plans to house up to 170,000 Nvidia GPUs to serve the surging demand in the Indo-Pacific region.
The revenue-sharing aspect of this model is particularly significant, as it lowers the capital expenditure barriers for smaller, specialized cloud providers while giving Nvidia a direct stake in the AI services market. By providing credit support, Nvidia is effectively financing the expansion of its own ecosystem, ensuring that infrastructure is built even in environments where traditional financing might be slow to move. This strategy bypasses the typical bottlenecks of data center site selection and power procurement, offering clients immediate access to massive scale for model training and 'Agent AI' inference.
This shift reflects a broader trend where hardware leaders seek to capture more value from the software and services layer. For startups and research institutions, the model provides a faster route to high-performance compute without the traditional lead times associated with hardware procurement and facility construction. As Nvidia deepens its integration with regional players like Firmus in Indonesia, it is effectively decentralizing the AI landscape, moving compute power closer to emerging markets and diverse regulatory jurisdictions.