NVIDIA's DSX AI Factory model lets cloud partners build AI factories with revenue sharing

NVIDIA is rethinking its AI infrastructure strategy with a new collaborative partnership model that includes revenue-sharing and credit support mechanisms. Rather than simply selling GPUs, NVIDIA will now enable AI cloud service providers to jointly build large-scale, multi-tenant AI factories built on the DSX AI Factory architecture. This distributed, modular design allows multiple providers to share resources efficiently, lowering the barrier for smaller players to access high-performance compute. The revenue-sharing component means NVIDIA gets recurring income tied directly to compute usage, aligning its incentives with cloud partners and end customers.

For startups, developers, and research institutions, this could significantly reduce the cost and friction of obtaining cutting-edge AI compute. By pooling demand across multiple tenants, DSX factories can achieve higher utilization and pass savings downstream. The model also includes credit support to help new customers get started without large upfront commitments. NVIDIA is positioning this as a way to scale AI infrastructure faster while building a more predictable revenue stream. If successful, this approach could reshape how AI compute is provisioned, shifting from a hardware sales model to a service-oriented partnership ecosystem.