CoreWeave upgrades to CNCF Platinum membership to support open-source AI infrastructure

CoreWeave upgraded to Platinum membership in the Cloud Native Computing Foundation on November 10, 2026, formalizing support for an open-source community already central to its cloud. Announced at KubeCon + CloudNativeCon North America, the move comes with a stated goal: bring its experience running AI workloads into shared infrastructure and standards.

The cloud came before the membership

CoreWeave’s connection to the foundation starts with the software beneath its services. CNCF hosts Kubernetes, along with other open-source infrastructure projects. CoreWeave says it built its cloud on Kubernetes and uses it across the systems that manage customers’ computing clusters and the physical GPU servers running their workloads.

That dependence extends to how customers organize AI work. CoreWeave Kubernetes Service is its managed environment for building, training and deploying AI applications. Its Slurm on Kubernetes offering lets training jobs and Kubernetes services share the same pool of GPUs, rather than assigning each a separate pool.

The membership upgrade follows engineering work already underway. CNCF identifies CoreWeave as a contributor to Kubernetes and Kueue; CoreWeave also names CloudNativePG among the projects its engineers support. The company presents the higher membership tier as another way to support that community, building on those existing contributions.

Platinum membership is how we bring that experience back to the community, showing up in the working groups, helping define what AI on Kubernetes actually means, and making top-tier compute available to the open source projects the whole ecosystem depends on.

Shadi Saba, CoreWeave vice president of engineering, AI/ML, in CNCF’s announcement

From shared inference software to shared standards

One earlier contribution has a clear date. In May 2025, CoreWeave joined Red Hat, IBM, Google and NVIDIA as a founding contributor to llm-d. CNCF describes that effort as rooted in the view that production inference—the work of running models to generate answers—should be built in the open.

The llm-d project is now a CNCF Sandbox project. CoreWeave also participates in the Certified Kubernetes AI Conformance Program. The company describes that program as a step toward shared standards that make deploying AI at scale more consistent and reliable.

CNCF places the upgrade in the context of AI moving from research into production. Its announcement points to distributed inference and workflow orchestration—coordinating work across computing systems—as challenges CoreWeave believes should be addressed with the broader open-source community. That is the foundation’s stated rationale for the membership change, not a new performance result.

The next step is a resource commitment

Saba’s description of the next phase goes beyond maintaining software. He says CoreWeave wants to apply lessons from large-scale training, production inference and agent workloads to community working groups and standards. He also explicitly includes making computing resources available to open-source projects in the company’s intended contribution.

Neither announcement puts a quantity or delivery schedule on that computing support. The confirmed change is the membership upgrade; the resource allocation remains a stated intention without a disclosed amount. That leaves the practical scale of this part of CoreWeave’s commitment unresolved.