NVIDIA Unveils Vera CPU with Olympus Cores for Agentic AI Workloads

NVIDIA announced the Vera CPU, built around the Olympus core architecture, designed specifically for agentic AI workloads that require strong single-thread performance and efficient handling of irregular control flow. The CPU integrates NVIDIA Spatial Multithreading for flexible resource partitioning, the Scalable Coherency Fabric delivering 3.4 TB/s on-die bandwidth, and SOCAMM2 LPDDR5X memory modules providing up to 1.2 TB/s aggregate bandwidth. The architecture supports coherent NVLink-C2C, single-NUMA dual-socket configuration, PCIe 6.4, CXL 3.1, and Confidential Computing for secure data movement and VM isolation. NVIDIA also unveiled the Rubin GPU as part of the Vera Rubin platform, featuring 336 billion transistors, 224 SMs, 896 Tensor Cores, and 288 GB HBM4 memory with 22 TB/s bandwidth, delivering up to 10x agentic throughput per unit energy compared to Blackwell. The Vera Rubin NVL72 rack-scale system integrates liquid cooling, cable-free MGX architecture, and hot-swappable NVLink switch trays, enabling up to 40% more GPUs within the same power envelope.