NVIDIA's Vera CPU is designed to boost AI factory throughput for agentic workloads, which involve multi-step workflows combining inference, tool use, code execution, and orchestration. The CPU is on the critical path for reasoning and response time, as it executes work between model steps. Sustained per-core performance under full socket load is a key metric, as faster cores determine how quickly each workflow advances. If CPU-side execution slows, the GPU fleet suffers: reinforcement learning (RL) receives fewer useful evaluations per cycle, leading to inflated time-to-train; the fleet takes longer to serve users; and KV-cache is evicted. For RL, a slower processor means fewer environment steps per second, degrading model policy updates. A baseline CPU might only complete 45% of its evaluations, leaving the GPU waiting. The Vera CPU design features 1.8x faster cores to address these challenges.