NVIDIA AI Updates: April 2, 2026

1. NVIDIA Donates GPU Dynamic Resource Allocation Driver to Kubernetes Community

NVIDIA. NVIDIA announced it is donating its Dynamic Resource Allocation (DRA) driver for GPUs to the Kubernetes community as an open-source contribution, lowering the barrier for GPU-accelerated workloads in cloud-native environments. The DRA driver allows Kubernetes to allocate GPU resources dynamically at the pod level rather than requiring static partition planning, enabling more efficient use of expensive GPU capacity in multi-tenant clusters. The contribution was announced at KubeCon 2026 and aligns with NVIDIA’s strategy of embedding its infrastructure stack into the open-source ecosystem that enterprise AI teams rely on. Source

2. NVIDIA Dynamo 1.0 Released as Open-Source Inference Software for Agentic AI

NVIDIA. NVIDIA released Dynamo 1.0, open-source software for generative and agentic inference at scale, with widespread global adoption reported at launch. Dynamo is designed to optimize multi-GPU and multi-node inference serving, addressing the throughput and latency demands of agentic workflows where models must call tools and manage long-horizon tasks. The software complements NVIDIA’s Rubin GPU hardware platform, which entered full production and is slated for partner availability in the second half of 2026. Source