In the past two years, driven by the AI wave, the prices of cloud vendors, chips, and related resources have skyrocketed. Today, as AI narratives gradually enter the deep-water zone, the market is no longer satisfied with just trading forward expectations, but wants to see the actual profits brought by AI. This change has also caused “computing power metals,” which had previously had strong gains, to face re-examination. Global tech giants have revealed financial reports one after another. Microsoft, Meta, and Alphabet, the parent company of Google, is still investing huge sums of money to build AI infrastructure, but issues such as free cash flow pressure and capital expenditure growing faster than return on revenue have also made investors re-evaluate the sustainability of this AI competition. At the same time, new architectures such as optical interconnection and HVDC power supply are being implemented at an accelerated pace, which may also reduce copper consumption per unit of computing power in some areas. In addition to copper, more “computing power metals” such as tin and germanium also face similar problems. Is AI's drive to copper demand overestimated? Industry insiders believe that the answer is not a simple yes or no. The AI copper logic has integrated judgments on the actual scale of production, changes in technology routes, and the progress of power supply. The amount of copper used per unit in computer rooms may be declining, but demand for power, transformation, transmission and distribution, and cooling due to data center expansion is still growing.

In the past two years, driven by the AI wave, the prices of cloud vendors, chips, and related resources have skyrocketed. Today, as AI narratives gradually enter the deep-water zone, the market is no longer satisfied with just trading forward expectations, but wants to see the actual profits brought by AI. This change has also caused “computing power metals,” which had previously had strong gains, to face re-examination. Global tech giants have revealed financial reports one after another. Microsoft, Meta, and Alphabet, the parent company of Google, is still investing huge sums of money to build AI infrastructure, but issues such as free cash flow pressure and capital expenditure growing faster than return on revenue have also made investors re-evaluate the sustainability of this AI competition. At the same time, new architectures such as optical interconnection and HVDC power supply are being implemented at an accelerated pace, which may also reduce copper consumption per unit of computing power in some areas. In addition to copper, more “computing power metals” such as tin and germanium also face similar problems. Is AI's drive to copper demand overestimated? Industry insiders believe that the answer is not a simple yes or no. The AI copper logic has integrated judgments on the actual scale of production, changes in technology routes, and the progress of power supply. The amount of copper used per unit in computer rooms may be declining, but demand for power, transformation, transmission and distribution, and cooling due to data center expansion is still growing.