On July 29, the China Nonferrous Metals Industry Association released its latest data showing that in the first half of this year, the output of ten major nonferrous metals as tracked by the National Bureau of Statistics reached 41.513 million tons, a year-on-year increase of 3.3%. Meanwhile, the total profits of large-scale nonferrous metal enterprises hit 418.39 billion yuan, surging 94.0% year-on-year, ranking among the fastest-growing industrial sectors.
Within this impressive performance, demand for "computing power metals" has seen a concentrated explosion. Driven by AI computing infrastructure construction, prices of nonferrous metals such as copper, aluminum, tin, tantalum, and indium have risen sharply. Among them, tin prices increased by 40% within six months, tantalum prices soared by 158%, and indium prices rose by 60%.
From "industrial grain" to "cornerstone of computing power," these elements buried deep underground are now emerging as key resources supporting the development of the AI era.
Copper: Core Material for Computing Connectivity
Global data center copper consumption is expected to reach 740,000 tons in 2026.
China is the world's largest producer and consumer of copper products, accounting for more than half of global copper consumption. In recent years, the rapid development of emerging industries such as AI and new energy has driven strong demand, leading to generally low copper inventory levels, further supporting high copper prices.
At a copper product manufacturing plant in Ningbo, Zhejiang Province, copper ingots are repeatedly rolled, cleaned, and processed into copper strip coils. These copper strips are mainly used to produce high-speed connectors and other components widely applied in AI servers, data centers, communication equipment, and other fields. The company's head told reporters in May this year that since the beginning of the year, domestic intelligent computing centers have accelerated their layout and construction, resulting in orders exceeding supply.
Professional institutions predict that global data center copper consumption will reach 740,000 tons in 2026, and is forecast to hit 1.3 million tons by 2028.
Tin: AI Servers and Advanced Packaging Become the Largest Incremental Demand Drivers
The industry is laying out recycling loops.
Due to its good conductivity, low melting point, and strong welding stability, tin has become a key basic material in advanced semiconductor packaging processes. The stronger the computing power and the denser the chip stacking, the greater the consumption of tin. Hence it is also called "computing power metal." Traditionally, tin was mainly used in consumer electronics and tinplate steel. Now, AI servers, optical modules, and advanced semiconductor packaging have become the largest sources of incremental demand for tin.
Reporters visited an enterprise engaged in printed circuit board assembly in Shenzhen this May, where production lines were running at full capacity 24 hours a day to produce circuit board assemblies needed for AI servers and optical modules. The surge in AI-related orders directly drove a significant increase in their tin procurement. Industry research shows that a single AI server consumes more than three times the amount of tin used in traditional servers. Emerging high-end track products, with higher added value, have stronger price tolerance.
Currently, China's leading tin companies have fully established circular industrial chains, creating a closed loop from mining, smelting, waste recycling, to reuse. Through technological breakthroughs, they are improving recovery rates of waste circuit boards, tin slag, and tailings, continuously increasing the proportion of recycled tin supply.
Experts: Short-Term Supply-Demand Pattern for "Computing Power Metals" Unchanged, But Volatility Risk Noted
Duan Shaofu, Deputy Secretary-General of the China Nonferrous Metals Industry Association, said in an interview that the industry generally believes the short-term supply-demand pattern for "computing power metals" has not fundamentally changed, and price floors remain supported. However, current macroeconomic conditions create valuation pressure, geopolitical uncertainties persist, market sentiment is highly sensitive, and the risk of price volatility cannot be ignored.
Duan believes that from an industrial response perspective, adjustments need to be made on three levels:
1.Enhance product quality adaptability.
The computing power industry chain demands high purity and customization of materials, which mismatches traditional standardized supply. Enterprises should focus on improving product quality and consistency, refining their capabilities under capacity constraints, better meeting high-end manufacturing requirements for high-quality materials, and enhancing market competitiveness through differentiated supply.
2.Strengthen upstream-downstream supply-demand coordination.
Rising raw material prices have put cost pressure on midstream and downstream users, with some end applications seeking substitutes or adjusting inventories. Upstream and downstream players need to establish closer supply-demand coordination mechanisms, including upstream early involvement in downstream material selection and verification to provide material support, while downstream promptly feeds back demand changes to avoid information asymmetry that could lead to market misjudgment and overreaction.
3.Diversify resource security channels.
For computing-power-related varieties such as copper, aluminum, and tin, upstream mineral resources are heavily dependent on imports, and frequent overseas supply disruptions have become the new normal. Enterprises should form combined arrangements covering overseas equity mines, recycling, long-term procurement contracts, etc., to enhance supply chain resilience.
CCTV Finance Reports
