by Wu Yangyu

Xiaomi is constructing a GPU cluster comprising 10,000 GPUs, underscoring its significant investment in artificial intelligence (AI) big models. The company’s big model team, established earlier this year with access to 6,500 GPUs, has rapidly scaled its infrastructure to support these efforts.

Insiders revealed that the project has been underway for several months, with Xiaomi CEO Lei Jun playing a key leadership role. An insider shared that Lei emphasized the strategic importance of focusing on mobile phones over other hardware like glasses, adding that Xiaomi is going "all in" on this field.

In China’s rapidly evolving AI landscape, computing clusters with at least 10,000 GPUs have become a fundamental requirement for developing cutting-edge generative AI large language models (LLMs). The massive computational needs of training and deploying LLMs have driven companies to significantly invest in infrastructure, ensuring competitive performance and scalability. By building a 10,000-GPU cluster, Xiaomi is positioning itself not just to meet this baseline but to gain a strategic edge in the AI race.

Public information shows that since forming its AI team in 2016, Xiaomi has expanded the team six times over seven years, growing its workforce in related fields to over 3,000 employees. Xiaomi's AI expertise now spans areas such as computer vision, acoustics, speech, NLP, knowledge graphs, machine learning, large models, and multimodal technologies. These capabilities are progressively being integrated into key business segments, including smartphones, automobiles, AIoT, and robotics.

Jiemian News exclusively learned of Xiaomi’s GPU cluster initiative through insider sources and reached out to the company for confirmation. As of press time, Xiaomi has not commented on the development.