Red Hat has announced support for Red Hat OpenShift on NVIDIA BlueField data processing units (DPUs), a collaboration aimed at delivering a more secure, high-performance platform for AI-driven data centres and next-generation “AI factories.”
The new capability enables organisations to deploy demanding AI workloads efficiently by combining Red Hat’s enterprise-grade Kubernetes platform with NVIDIA’s DPU-powered acceleration. Modern enterprises, particularly those running AI and data-intensive applications, often face performance bottlenecks as applications compete with infrastructure services for computing resources. Red Hat OpenShift on BlueField seeks to mitigate this by isolating application and infrastructure workloads across CPUs and DPUs, thereby enhancing performance, network efficiency, and security.
Key features of this integration include optimised resource utilisation, achieved by offloading networking functions from CPUs to DPUs; accelerated data plane and storage traffic, powered by NVMe-over-Fabrics (NVMe-oF) and an enhanced OVS data path; and advanced cloud networking with distributed routing to improve secure multi-tenancy and scalability across clusters. Additionally, the architecture supports a layered security approach, reducing the attack surface by moving key infrastructure services to the DPU layer.
Also Read: CS TECH Ai Drives Major Leap in Digitizing India’s Mining Sector
Support for NVIDIA BlueField on Red Hat OpenShift will begin as a technical preview in the coming weeks. The two companies are also working to extend integration with NVIDIA’s DOCA software framework and additional third-party network functions.
Looking forward, the next-generation NVIDIA BlueField-4 is expected to expand these capabilities with deeper DOCA integration and advanced acceleration features, while support for NVIDIA Spectrum-X Ethernet networking will enable seamless, high-performance connectivity across distributed cloud environments.
Ryan King, Vice President, AI and Infrastructure, Partner Ecosystem Success at Red Hat, said the collaboration “offers customers a more reliable, secure and high-performance platform to meet the growing demand for AI infrastructure.”
Justin Boitano, Vice President, Enterprise Products at NVIDIA, added that the partnership “builds on years of joint innovation to deliver the performance and scalability that next-generation AI reasoning workloads demand.”