Your AI workloads already run on Kubernetes — now NVIDIA is handing the scheduling logic to the community.
At KubeCon Europe in Amsterdam this week, NVIDIA donated a critical piece of software to the Cloud Native Computing Foundation, a nonprofit that oversees open-source infrastructure tools used by companies worldwide.
The software in question handles how GPU chips — the hardware that powers AI — get divided and assigned inside cloud servers. Until now, NVIDIA controlled that logic. From today, the community does.
The practical change is straightforward. Companies running AI on shared cloud infrastructure previously had to manage GPU assignments manually, which was time-consuming and inefficient.
The donated driver automates that process: it lets multiple workloads share a single GPU chip, scales across interconnected machines for large AI model training, and lets engineers adjust resource allocation on the fly without restarting servers.
Chris Aniszczyk, the top technology officer at the Cloud Native Computing Foundation, called it a turning point:
"NVIDIA's deep collaboration with the Kubernetes and CNCF community to upstream the NVIDIA DRA Driver for GPUs marks a major milestone for open source Kubernetes and AI infrastructure. By aligning its hardware innovations with upstream Kubernetes and AI conformance efforts, NVIDIA is making high-performance GPU orchestration seamless and accessible to all."
Alongside the driver, NVIDIA announced that AI workloads can now run inside more secure, isolated environments on the same hardware useful for organizations handling sensitive data that cannot mix with other computing jobs. A scheduling tool called KAI Scheduler also became part of the same open-source foundation and is available for use today.
Eight major technology companies — AWS, Broadcom, Canonical, Google Cloud, Microsoft, Nutanix, Red Hat, and SUSE — are contributing to the effort.
Red Hat's top technology officer, Chris Wright, put the business logic plainly:
"Open source will be at the core of every successful enterprise AI strategy, bringing standardization to the high-performance infrastructure components that fuel production AI workloads."
Ricardo Rocha, who oversees computing infrastructure at CERN, the particle physics research center, added the scientific dimension:
"For organizations like CERN, where efficiently analyzing petabytes of data is essential to discovery, community-driven innovation helps accelerate the pace of science."
All software mentioned is available today at no cost.
Source: Nvidia