Hugging Face and NVIDIA launch Training Cluster as a Service for on-demand DGX Cloud Lepton GPU clusters

AI Impact Summary

Hugging Face and NVIDIA unveil Training Cluster as a Service to provide researchers with on-demand, scalable GPU clusters. The offering aggregates capacity from NVIDIA Cloud Partners and NVIDIA DGX Cloud Lepton across regional datacenters, with scheduling and monitoring handled via Hugging Face tools. Leveraging high-end GPUs like Hopper and GB200 and a pay-for-duration model lowers upfront costs and procurement barriers for universities and labs, enabling faster iteration on foundation models. Expect higher GPU utilization, accelerated research timelines, and a shift in enterprise buying toward a Hugging Face/NVIDIA-managed workflow.