全球数据中心为满足AI技术发展的电力需求而扩张，正在成为铜消费的重要驱动力。[1]随着超大规模AI训练设施的部署，电力配送基础设施的铜需求将大幅增加，尽管光纤通信技术的兴起曾被认为可能削弱铜在信息传输中的作用，但这一新兴需求足以抵消相关影响。[1]
单个超大规模AI训练设施的铜消耗量可达5万吨，是传统数据中心的5至10倍。[1]全球数据中心电力容量预计从77吉瓦增长至334吉瓦，到2030年仅数据中心和电网基础设施就需增加33万至42万吨铜。[1]彭博社估计到2030年，AI相关数据中心扩张可增加全球铜消费200万吨。[1]这一增长符合全球铜消费结构的特点——近70%的铜消费与电气应用相关。[1]从更长期的角度看，国际能源署预测全球铜需求将从2024年的2670万吨增至2040年的3400万吨。[1]
The rapid expansion of artificial intelligence infrastructure is set to significantly boost global copper consumption, offsetting concerns that fiber optic technology might diminish the metal's importance in communications. As tech giants including Meta, Microsoft, and OpenAI scale up data center operations worldwide, the massive electrical demands of these facilities are emerging as a major new driver for copper usage.[1]
Approximately 70 percent of global copper consumption is tied to electrical applications, and hyperscale AI training facilities are particularly copper-intensive.[1] A single large-scale AI training facility can consume up to 50,000 tons of copper—five to ten times more than traditional data centers.[1] Global data center power capacity is projected to expand from 77 gigawatts to 334 gigawatts by 2030, requiring between 330,000 and 420,000 additional tons of copper for data centers and electrical grid infrastructure alone during this period.[1] Bloomberg estimates that AI-related data center expansion could add 2 million tons to global copper consumption by 2030.[1] Looking further ahead, the International Energy Agency forecasts global copper demand will grow from 26.7 million tons in 2024 to 34 million tons by 2040.[1]