大宗商品的“AI含金量”（国金宏观陈瀚学）

Core Viewpoint - The article discusses the varying exposure of different commodities to AI, highlighting that metals like vanadium, copper, lithium, gallium, and aluminum significantly contribute to the value of data centers and electrical grid facilities, while gold serves as a hedge against the AI bubble due to its financial attributes. In contrast, crude oil stands in opposition to the AI narrative due to its inflation-suppressing political implications [2][4][55]. Group 1: Data Centers and Power Systems - AI technology is reshaping global metal demand, with data centers becoming critical infrastructure for national economic security. From 2024 to 2035, global data center capacity is expected to triple, with investments reaching $3-7 trillion by 2030. The electrical grid is projected to require $600 billion in investments by 2030, with a total length of 110 million kilometers by 2035 [7][4]. - Data centers and grid upgrades require substantial amounts of copper and aluminum for conductivity, while energy storage systems need lithium, cobalt, and nickel. High-end motors and transformers also require rare earth materials [7][4]. - Each megawatt (MW) of data center capacity corresponds to 60-75 tons of mineral metals, with cooling systems consuming the largest share (35-45%) of metals, primarily iron, aluminum, and copper [9][10][11]. Group 2: Contribution of Minerals to Data Center Value - The contribution of various minerals to the total value of data centers is significant, with gallium contributing 6.0%, germanium 2.4%, and tin 1.2%. Copper contributes only 0.9%, while iron, aluminum, and carbon contribute between 0.1% and 0.3% [12]. - The expansion of electrical demand from data centers is driving the construction of transmission and distribution (ET&D) infrastructure, which will also increase the demand for minerals and metals [13][18]. Group 3: AI-Related Commodity Rankings - The ranking of metals related to AI investments, based on their contribution to data center and electrical grid facilities, is as follows: vanadium, copper, lithium, gallium, and aluminum [20]. - Vanadium's current demand is primarily driven by the steel industry, but its potential in long-duration energy storage is expected to grow as the market evolves [21]. - Copper is essential for AI computing infrastructure due to its excellent conductivity and thermal properties, with a projected supply deficit expected by 2026 [25][22]. Group 4: Precious Metals and Oil - Gold serves as a hedge against the AI bubble, with significant inflows into gold ETFs indicating a growing demand for safe-haven assets amid economic uncertainty [45]. - Silver, with its industrial applications and financial attributes, has shown strong elasticity in demand, while platinum and palladium also have significant industrial uses [47][48]. - Crude oil currently opposes the AI narrative due to its association with inflation control and the political landscape, with potential for price rebounds if OPEC countries decide to cut production [52][55].