“Running agentic AI workloads for inference requires the latest generation of CPU servers.”