NVIDIA Donates Its GPU Resource Driver to the CNCF as KubeCon Europe Draws a Wave of Open Source AI Infrastructure Contributions
NVIDIA transferred its Dynamic Resource Allocation driver for GPUs to community ownership under the Kubernetes project at KubeCon Europe 2026, while Google open-sourced the GKE Cluster Autoscaler and Broadcom donated Velero to the CNCF Sandbox.
NVIDIA announced on March 24 that it is donating its Dynamic Resource Allocation driver for GPUs to the Cloud Native Computing Foundation, placing one of the most critical pieces of GPU orchestration software under community governance within the Kubernetes project. The move was the headline contribution in a broader wave of open source AI infrastructure donations at KubeCon + CloudNativeCon Europe 2026 in Amsterdam, where Google, Broadcom, and Microsoft also transferred key projects to vendor-neutral foundations.
NVIDIA Hands Over GPU Scheduling to the Community
The donated DRA driver enables Kubernetes clusters to dynamically configure and share GPU devices, supporting NVIDIA technologies including Multi-Process Service for concurrent workload sharing and Multi-Instance GPU for partitioning a single GPU across multiple tenants. The driver also provides native support for Multi-Node NVLink interconnects, which are essential for distributing large AI training jobs across Grace Blackwell systems, according to NVIDIA’s announcement.
Chris Aniszczyk, chief technology officer of the CNCF, described the donation as “a major milestone for open source Kubernetes and AI infrastructure,” according to NVIDIA. The code is publicly available on GitHub under the k8s-dra-driver-gpu repository, and NVIDIA stated it will enjoy “full community ownership” going forward.
The donation arrives at a pivotal moment. Kubernetes 1.36, which reached code freeze on March 18, graduates four DRA-related features to general availability, establishing the framework as the standard mechanism for managing specialized hardware in cloud native environments. With the NVIDIA driver now under community control, third-party schedulers and orchestration platforms can integrate GPU-aware resource management without vendor lock-in.
KAI Scheduler and New Open Source Projects
Alongside the DRA driver, NVIDIA announced that its KAI Scheduler has been accepted as a CNCF Sandbox project. KAI is a high-performance workload scheduler designed specifically for AI training and inference jobs on GPU clusters, according to NVIDIA.
NVIDIA also introduced several new open source projects at the conference. Grove provides a Kubernetes API for orchestrating AI workloads on GPU clusters. NVSentinel offers automated GPU fault detection and remediation. The company also unveiled GPU support for Kata Containers, enabling confidential computing for AI workloads through hardware-isolated virtual machines, as detailed by NVIDIA.
Amazon Web Services, Broadcom, Canonical, Google Cloud, Microsoft, Nutanix, Red Hat, and SUSE were listed as collaborators on the initiatives.
Google Open-Sources GKE Cluster Autoscaler
Google used KubeCon Europe to announce the open source release of the GKE Cluster Autoscaler, one of the core components that manages infrastructure provisioning for Google Kubernetes Engine customers. The company stated its goal is to provide a vendor-neutral autoscaling platform that the broader community can build upon, according to Google Cloud.
Google also released its DRA driver for Tensor Processing Units as open source, extending the DRA ecosystem beyond GPUs to include custom AI accelerators. Google and NVIDIA said they coordinated these releases to establish a unified resource management standard for AI hardware in Kubernetes, according to Google Cloud.
Broadcom Donates Velero to CNCF
Broadcom contributed Velero, its Kubernetes backup and disaster recovery tool, to the CNCF Sandbox. Velero addresses a gap in Kubernetes itself, which does not provide built-in cluster-level backup or recovery capabilities, according to SiliconAngle. The donation transitions the widely adopted project from Broadcom’s governance to a vendor-neutral foundation model.
Broadcom also announced version 3.6 of VMware vSphere Kubernetes Service, which adds declarative tuning profiles for AI and data-intensive workloads, and expanded partnerships with F5 Networks, Kong, and Tigera for validated networking and security integrations, as reported by SiliconAngle.
A Broader Pattern
The concentration of major open source donations at a single conference reflects growing industry consensus that AI infrastructure tooling benefits from community governance. By placing GPU drivers, schedulers, autoscalers, and backup tools under foundation oversight, the contributing companies are collectively establishing a vendor-neutral stack for running AI workloads on Kubernetes. For platform engineering teams evaluating their AI infrastructure strategies, the practical consequence is that core GPU orchestration capabilities are now accessible without commercial licensing from any single vendor.