NVIDIA is launching its Vera Rubin architecture, which combines the Rubin GPU with the Vera CPU, Groq 3 LPX inference accelerator, and other units such as storage and networking racks. The Vera CPU focuses on data orchestration, aiming to improve data center efficiency rather than solely relying on more processor cycles. According to Jason Hardy, NVIDIA's Vice President of Storage Technology, the Vera CPU achieves up to a 3x performance improvement in these operations, thereby unleashing the full potential of flash memory and addressing data transfer bottlenecks. Analysis indicates that as AI computing scales to gigawatt levels, data orchestration becomes increasingly complex. NVIDIA's advantage in GPU-surrounding systems is becoming more prominent, and even with intensifying competition in GPUs themselves, the company is expected to maintain its lead in this new competitive arena.