和讯·2025-03-07 10:04Core Viewpoint - The article emphasizes the importance of building a new AI computing ecology that is integrated, inclusive, and green, highlighting the challenges and innovations in AI computing as presented by ZTE Corporation's chairman, Li Zixue, at the World Internet Conference [1] Group 1: Challenges in AI Computing - Insufficient computing power is a long-term issue in the digital intelligence process, driven by the increasing demand for data processing due to the "connection + computing power" spiral development [2] - The rise of intelligent computing has led to two new challenges: the need for rapid expansion of inference computing power and the reinforcement of training computing power to support larger model training and breakthroughs in multimodal large models [2] Group 2: Green Energy Initiatives - ZTE proposes three advancements in green energy: compactness through technological innovation for energy-efficient cabinet designs, enhanced heat efficiency for high-density AI servers, and increased integration of renewable energy sources to reduce carbon emissions [3] - Innovations include intelligent micro-modules and elastic functional containers that shorten data center deployment cycles by over 40%, and a PUE reduction to 1.1 through integrated liquid cooling solutions [3] Group 3: AI Application Empowerment - ZTE is addressing challenges in AI application deployment, such as inapplicability, difficulty of use, and lack of knowledge [4] - The company offers the AiCube integrated machine for localized deployment, ensuring data security, and has developed an AI automotive design machine in collaboration with Dongfeng Motor and Hubei Mobile, enhancing design efficiency by 30% [5] - ZTE suggests a phased approach for enterprises to adopt AI applications, starting from specific scenarios and gradually expanding to operational, research, and marketing areas, achieving overall operational efficiency improvements of over 15% [5] Group 4: Future Outlook - The article concludes with a call for the ICT industry to seize opportunities in the transition to a new era of digital intelligence, advocating for an "open and decoupled" computing ecology and the acceleration of technological upgrades and application implementations [6]