Core Insights - The integration of digital and energy revolutions is driving a significant increase in demand for computing power, with projections indicating an average annual growth rate of around 20% in China's data center electricity consumption from 2024 to 2030, outpacing the overall electricity consumption growth rate [1] Group 1: Industry Developments - The 2025 China International Big Data Industry Expo highlighted discussions on promoting the collaborative development of electricity and carbon computing [1] - The Deputy Director of the National Data Bureau emphasized that electricity supply is a critical factor influencing computing power development, and the scale of computing power directly affects data processing efficiency [1] - The rapid advancement of AI is leading to an explosive growth in computing power demand, necessitating the establishment of a new infrastructure that integrates data, computing, networks, and energy [1] Group 2: Southern Power Grid's Innovations - Southern Power Grid showcased its developed Electric-Carbon-Computing Collaborative Operation System, which is based on power infrastructure and promotes the economic and green flow of computing resources [2] - The system features three core capabilities: real-time monitoring of power consumption, dynamic scheduling to reduce operational costs by over 10%, and nationwide collaboration to connect computing resources with the integrated national computing network [2] - The system aims to break down information barriers between electricity, computing power, and carbon emissions, achieving precise matching and efficient scheduling [2] Group 3: Practical Applications and Future Outlook - The Electric-Carbon-Computing Collaborative Operation System has shown initial success, with a case where a computing task was migrated to a data center, resulting in a 10% reduction in overall computing costs and the consumption of 113.4 kWh of green electricity [3] - The flexibility of computing power scheduling is identified as a crucial foundation for collaborative operations, requiring solutions for application, data, and model coordination [3] - Future expectations include making the use of computing resources as convenient and quantifiable as water and electricity, necessitating the establishment of a precise measurement system for computing power [3]

昨天，我国首套电碳算协同运营系统上线试运行，实现电力、算力与碳排放的实时联动和优化调度，将 助力数据中心节能降碳、降本增效。 南方电网数字化部副总经理 禤亮：预计到2030年，中国算力能耗一年达到4800亿度电，碳排放超过2亿 吨。如果算力用电综合成本降低1毛钱，每年可为算力产业节约480亿元，为广大终端用户提供更普惠的 算力。如果每提升1%的绿电使用，每年可减少碳排放约200万吨，相当于种植1.1亿棵树。 这套电碳算协同运营系统实现"电网、算网、通信网"三网融合。当西部地区出现更便宜电价时，系统会 自动将3D渲染、大模型训练等计算任务调度至西部的算力节点；当风电、光伏等新能源发电充裕时， 系统则会集中使用绿色电力支撑计算任务。 来源：央视新闻客户端 ...