Red Hat and NVIDIA Power AI Datacenters with BlueField

Red Hat and NVIDIA partner to make AI datacenters faster by integrating Red Hat OpenShift with NVIDIA BlueField DPUs.

The Red Hat has reported about a significant partnership with NVIDIA to power AI data centers with the help of combining Red Hat OpenShift with NVIDIA BlueField Data Processing Units (DPUs). This alliance is a continuation of the open-source scalability into the hardware acceleration layer, and this could lead to quicker, safer, and larger AI settings.

The integration will allow AI factories and next-generation data center deployment to operate more efficiently, as it will separate AI and infrastructure workloads. By using Red Hat OpenShift on NVIDIA BlueField, organizations can enjoy the advantage of optimum utilization of resources, as networking service is transferred to DPUs instead of CPUs, leaving processing power free for AI tasks.

Other improvements are faster storage and data transfer with NVMe over Fabrics (NVMe-oF) capabilities, an integrated cloud network, and greater multi-tenancy infrastructure that is provided by distributed routing capabilities backed by BlueField. The outcome is efficiency, facilitated supplying, and better security.

Ryan King, the Vice President of AI and Infrastructure at Red Hat, emphasized that the partnership provides customers with high-performance and secure infrastructure that provides maximum hardware investments. As it was highlighted by NVIDIA's Justin Boitano, it is another milestone to their common mission of enabling enterprises with scalable AI infrastructure.

Red Hat OpenShift on NVIDIA BlueField should be available as a technical preview in the near future, and NVIDIA DOCA, BlueField-4, and Spectrum-X Ethernet networking support should be possible.