NVIDIA open sources NVKE and NVKR for Kubernetes on GPU clusters

NVIDIA open sourced two Kubernetes projects this week at KubeCon North America in Salt Lake City. The company released NVIDIA Kubernetes Engine (NVKE) and NVIDIA Kubernetes Recipes (NVKR) to help operators run managed Kubernetes services on GPU infrastructure.

Every major Kubernetes distribution already supports NVIDIA hardware. NVIDIA decided that gap remains anyway. The company is targeting the cluster work that surrounds the hardware, not the hardware support itself.

What does NVKE actually automate?

Platform operators must provision bare-metal machines, confirm node identity, upgrade clusters and replace failed nodes. NVKE automates that lifecycle. NVKR builds and signs the software that NVKE deploys.

The stated goal is eliminating handoffs between server hardware teams and platform management teams. "NVKE natively uses the Kubernetes Cluster API to provision infrastructure through a standard interface," said Warren Barkley, vice president of AI factories at NVIDIA. "We built and open-sourced a Cluster API provider for NICo, our bare-metal infrastructure controller for GPUs."

According to Barkley, new and replacement capacity goes from the rack into a customer's cluster without a hand-off between teams. That claim covers provisioning speed. What NVIDIA does not quantify: how much faster, or what utilization gains operators should expect once workloads run.

Barkley said NVKR gives operators validated recipes to build their own signed, version-pinned images. He framed NVKE as one piece of a larger open stack where operators run the service and own their customer relationships.

Is NVKE a competitor to OpenShift and Rancher?

There appears to be overlap with Red Hat OpenShift and SUSE Rancher. NVIDIA officials insist there is none.

"NVKE isn't meant to replace OpenShift or Rancher," Barkley said. "NVKE and NVKR are an open source reference stack for AI factories. Reference architectures have traditionally been documents, but we saw an opportunity to go further by building a reference that operators and partners can deploy, test, and build on."

The vendors themselves agree, at least publicly. Joe Fernandes, vice president and general manager of the AI business unit at Red Hat, said his company is collaborating with NVIDIA on leveraging NVKE and NVKR alongside OpenShift and the broader Red Hat AI platform. He called ongoing ecosystem investment in upstream Kubernetes tooling complementary to enterprise AI adoption.

SUSE struck the same note. "SUSE builds one Kubernetes, but manages all, and where customers want to bring NVKE-provisioned clusters into the wider estate, we will extend that management through SUSE AI Factory and SUSE Rancher Prime," said Rhys Oxenham, vice president and general manager for AI at SUSE.

Eight partners, including Mirantis, Rafay and vCluster, have said they intend to adopt the projects. Barkley described three adoption patterns: some partners take the complete stack; others keep their mature components and add NVIDIA pieces such as validated configuration recipes, telemetry pipelines and GPU fault remediation. "A third group will borrow our patterns to improve their own offerings, which is one reason we released NVKE and NVKR as open source software," he said.

Where do analysts place it?

IDC analyst Arnal Dayaratna expects NVKE to complement established platforms like OpenShift and Rancher Prime, which give enterprises consistent deployment and policy enforcement across clusters. "NVKE sits closer to the hardware and automates how operators provision, update, and maintain Kubernetes clusters on bare-metal GPU servers," Dayaratna said. He expects Kubernetes vendors to bundle NVKE or integrate parts of it into their existing platforms.

Not every analyst accepts the complement framing. "It is an integrated software stack play, it won't unseat existing distributions on the traditional use cases but will be the #1 distribution for Nvidia centric workloads," said Holger Mueller, principal analyst and vice president at Constellation Research. "That is what it is built for - not to compete with generic, all purpose Kubernetes distros."

The NVKE project page on GitHub states plainly that it is not a Kubernetes distribution. Mueller's read effectively overrides that disclaimer for GPU-heavy estates.

What is the unstated advantage?

NVIDIA builds both the GPUs and the software that runs on them. Barkley claimed a validated Kubernetes stack can be ready for customers on the day a new GPU generation ships. That is a real edge for operators racing to stand up capacity. It is also a deeper lock-in path: the more of the cluster lifecycle NVIDIA automates, the more the stack depends on NVIDIA's release cadence.