Red Hat and NVIDIA Expand Collaboration to Support Rack-Scale AI with Rubin Platform

Red Hat has announced a significant expansion of its collaboration with NVIDIA to align open source technologies with the rapid evolution of rack-scale AI. This partnership aims to provide Day 0 support for the next-generation NVIDIA Rubin platform through Red Hat’s enterprise stack. By optimizing Red Hat Enterprise Linux and OpenShift for this new architecture, the companies seek to accelerate the transition from AI experimentation to full-scale production.

A central piece of this initiative is the introduction of Red Hat Enterprise Linux for NVIDIA. This specialized edition is designed to incorporate the latest platform advancements immediately upon their release. It simplifies lifecycle management by providing direct access to validated GPU drivers and the CUDA toolkit through official repositories. This ensures that organizations can deploy high-density AI systems with the stability required for enterprise environments.

The collaboration targets the NVIDIA Vera Rubin platform, which includes the energy-efficient Vera CPU and advanced Rubin GPUs. Red Hat OpenShift will offer automated deployment and configuration for these systems, specifically supporting the NVL72 rack-scale solution. This infrastructure is further enhanced by support for BlueField-4 data processors to optimize networking and resource utilization. More details are available at the Red Hat newsroom.

Security remains a top priority as organizations move sensitive workloads into AI factories. Red Hat is introducing support for NVIDIA Confidential Computing across its AI portfolio to protect model weights and data during processing. This provides cryptographic proof of data protection, allowing industries like finance and healthcare to scale their AI initiatives with confidence. The platform leverages a hardened foundation built on years of open source security leadership.

Red Hat and NVIDIA are also expanding support for distributed inference through the Red Hat AI Inference Server. This includes deeper integration with NVIDIA’s open models, such as the Nemotron family, and plans to support vision and robotics-specific models. These integrations allow developers to run complex AI agents across a consistent operational model throughout the hybrid cloud.

Support for the NVIDIA Vera Rubin platform is expected to coincide with its general availability in the second half of 2026. Customers will be able to access the necessary drivers and integration tools through the Red Hat Customer Portal. This timeline ensures that enterprises can begin planning their hardware refresh cycles with a validated software stack ready to go. Further partnership details can be found at redhat.com.