NVIDIA unveils revenue-sharing AI compute model for cloud

NVIDIA has launched a new financial model designed to accelerate the global build-out of artificial intelligence infrastructure, moving beyond pure hardware sales to share in the revenue generated by cloud services. The initiative, announced on Wednesday, July 1, pairs revenue-sharing agreements with credit support for cloud providers, aiming to overcome the massive capital barrier facing many companies seeking to deploy advanced AI computing.

The strategy allows cloud providers to access NVIDIA's latest hardware, including Grace Blackwell GB300 GPUs, without bearing the full upfront cost. In return, NVIDIA earns its standard hardware revenue plus a recurring share of the cloud revenue generated from that capacity. The company stated the goal is to help startups, enterprises, and regional operators secure the infrastructure needed for training and, increasingly, for serving high-volume AI inference workloads.

Two companies have been named as initial partners, illustrating the program's scale. Sharon AI, an Australian cloud operator, plans to deploy up to 40,000 GPUs under a six-year agreement involving 72 megawatts of data center capacity. Firmus Technologies is developing an...