Nvidia unveils DSX revenue-share model for AI cloud growth

Nvidia’s new DSX revenue-sharing model is a financial innovation wrapped around its AI factory stack. Instead of only selling GPUs and DSX software to well-capitalized hyperscalers, Nvidia is effectively saying to smaller AI clouds and data-center operators: we’ll help you stand up capacity now, and get paid later as your workloads generate revenue. That’s a classic vendor-financing move, but applied to the hottest bottleneck in the AI economy—compute.

This matters because access to large-scale training and inference clusters is still the main gating factor between “toy” models and frontier systems. By lowering the upfront capex required to join the ecosystem, Nvidia is broadening the field of serious model builders and inference providers, especially in emerging markets and second-tier clouds. The DSX “AI factory” framing also reinforces Nvidia’s ambition to define not just the chip, but the full reference architecture for how industrial-scale AI infrastructure is designed and monetized.

For the race to AGI, the net effect is to accelerate diffusion of very large compute footprints beyond today’s hyperscaler oligopoly. That could speed up experimentation with larger, more agentic models while further entrenching Nvidia as the central supplier whose economics every other player must live with.