NVIDIA introduced Vera, a new CPU architecture with 88 custom Olympus cores and 1.2 TB/s LPDDR5X memory bandwidth, specifically designed to accelerate AI agent workloads. The architecture achieves 1.8x higher sustained per-core performance than leading x86 CPUs on agentic tasks, with Perplexity reporting 1.5x faster coding workflows. NVIDIA also mentioned its upcoming Rosa CPU with the Rigel core, indicating a strategic focus on CPU evolution for autonomous agents.
NVIDIA is expanding its AI infrastructure dominance beyond GPUs into CPUs, creating a new bottleneck for agentic AI that x86 server CPU vendors must now address.