NVIDIA Donates Dynamic Resource Allocation GPU Driver to Kubernetes Community
NVIDIA AI Blog·medium signal
NVIDIA is donating its Dynamic Resource Allocation (DRA) driver for GPUs to the Kubernetes community as open source, announced at KubeCon 2026. DRA enables fine-grained GPU scheduling and sharing across containerized AI workloads — a critical infrastructure primitive for enterprises running mixed AI inference on shared Kubernetes clusters. This moves GPU resource management from proprietary to community-owned.