NVIDIA Opens Source Run ai for Kubernetes GPU Management

NVIDIA Lets Go of Run:ai

NVIDIA has released its Run:ai platform as open source software under the Apache 2.0 license. The announcement came in October 2026 and makes the company's GPU orchestration tool available to anyone who wants to use it.¹

Run:ai was originally sold as an enterprise product that helps organizations manage graphics processing units across Kubernetes clusters. GPU management is critical for AI workloads, since training large models and running inference requires careful scheduling of expensive hardware resources.

What Is Inside the Release

The open source version includes several key components. The KAI Scheduler handles how GPU tasks get assigned to nodes in a cluster. The release also brings CUDA 13.4 GPU fractioning, which lets a single physical GPU be split into smaller pieces so multiple tasks can share it.¹

Quota management tools are included as well. These let administrators set limits on how much GPU capacity different teams or projects can use. A local development environment based on Kind, a tool for running Kubernetes locally, is also part of the package.

Why This Matters

Making Run:ai open source lowers the barrier for smaller teams and individual developers who want to run AI models efficiently. Previously, organizations needed to purchase enterprise licenses to access these orchestration features. Now they can set up the same kind of GPU management on their own infrastructure.

The Apache 2.0 license is permissive, meaning companies can use, modify, and distribute the software without strict requirements like those found in copyleft licenses. This should encourage adoption across both commercial and research settings.

The Bigger Picture

GPU orchestration on Kubernetes has become increasingly important as AI applications grow more demanding. Large language models and generative AI systems require significant computing power, and managing that power across a cluster is complex. Tools like Run:ai address that complexity by providing scheduling, resource allocation, and monitoring capabilities.

By open sourcing Run:ai, NVIDIA is making its infrastructure software available to the broader community. This could accelerate the development of AI applications by giving more people access to professional-grade GPU management tools.

The release comes at a time when demand for GPU computing continues to rise. Companies and research labs are investing heavily in AI infrastructure, and efficient resource management is essential to keeping costs down while maximizing throughput.¹