On January 6th local time, at the Lenovo Innovation Technology Conference (Tech World), Chairman and CEO of Lenovo Group Yang Yuanqing and CEO and founder of NVIDIA Huang Jinxun jointly announced that the two companies will deeply integrate their core strengths and officially launch the "Lenovo AI Cloud Super Factory." This initiative marks another upgrade in the strategic cooperation between the two giants in the era of hybrid AI.
According to the information released by both parties, NVIDIA's latest next-generation computing platform, Rubin Platform, will be a core component integrated into this plan. With this cutting-edge architecture, the "AI Cloud Super Factory" aims to address key pain points in AI deployment for global cloud service providers: not only significantly reducing the "Time to First Token," greatly improving model response speed, but also offering strong scalability, supporting rapid expansion to a massive scale of 100,000 GPUs. This extreme computing density will directly empower the development and operation of trillion-parameter intelligent agents (Agents) and large language models.
At the event, Huang Jinxun highly praised the cornerstone of this collaboration. He pointed out that Lenovo, as the manufacturer of most supercomputers worldwide, is one of the few companies that can fully integrate self-designed, large-scale manufacturing, and global service capabilities. To address the heat dissipation challenges brought by high-performance computing, Yang Yuanqing revealed that Lenovo's exclusive "Neptune" liquid cooling technology will also be deeply integrated into the Super Factory plan, ensuring efficient and sustainable computing through green empowerment.
This collaboration is seen as an important milestone in driving AI from experimental stages toward large-scale industrial applications. The two companies will jointly build a standardized AI production line from hardware infrastructure to software services, helping global enterprises seize opportunities in the era of trillion-parameter models.
