NVIDIA unveils Vera CPU for agents at GTC Taipei, shipping from fall 2026

At GTC Taipei NVIDIA introduced Vera, a CPU it positions for agentic AI workloads, and said it is in full production. Vera systems reach customers from system builders and cloud partners starting in fall 2026, and standalone Vera servers are planned from Dell, HPE, Lenovo and Supermicro. The performance comparison is NVIDIA's own claim.

• 88 custom Olympus cores with LPDDR5X bandwidth up to 1.2 TB/s (NVIDIA spec)
• NVLink-C2C up to 1.8 TB/s coherent bandwidth to Vera Rubin GPUs
• Up to 1.8x faster task completion vs x86 CPUs (NVIDIA claim)
• Availability from fall 2026; Anthropic only evaluating, OpenAI and SpaceXAI named as adopters