At the 2026 OCP Global Summit, Lenovo Group unveiled that it has started the global rollout of NVIDIA’s cutting-edge Vera Rubin NVL72 rack-level AI system. Simultaneously, Lenovo is integrating the entire Vera Rubin platform into its AI Cloud Superfactory solution. This strategic initiative is designed to bolster the construction and delivery capabilities of AI infrastructure for cloud service providers across the globe.

The Vera Rubin NVL72 system is a powerhouse, incorporating 72 Rubin GPUs and 36 Vera CPUs. Leveraging NVLink 6 high-speed interconnection technology, it achieves a remarkable throughput increase of up to 30 times per megawatt compared to its predecessor. Additionally, it slashes inference costs per million tokens by as much as 45 times, marking a significant leap in efficiency and cost-effectiveness.

Lenovo Group also showcased five rack-level systems, each tailored for computing, inference, storage, and networking. Together, these systems form a comprehensive AI infrastructure architecture, providing a holistic solution for AI deployment.

Presently, Lenovo’s Vera Rubin NVL72 is already being shipped to customers worldwide, marking a new era in AI infrastructure development.