The rapid expansion of artificial intelligence infrastructure is extending supply-chain pressure beyond advanced processors and high-bandwidth memory to electricity, transformers, cooling equipment, optical networks and the specialised minerals required across these systems. As AI data centres scale, demand is increasingly reaching into mining, refining and materials processing markets.
Copper is among the industrial metals directly exposed to this build-out. An AI training data centre requires approximately 47 tonnes of copper per megawatt, compared with about 21 tonnes per megawatt for a cryptocurrency facility. Global data-centre copper demand is projected to rise from roughly 1.1 million tonnes in 2025 to 2.5 million tonnes by 2040.
AI is not, however, the sole driver of copper demand. Grid expansion, electric transport and renewable power generation remain larger sources of consumption, while the expansion of data-centre infrastructure adds another significant demand stream for the metal.
Gallium and germanium emerge as tighter technology constraints
Gallium and germanium have much smaller markets than copper but occupy critical positions in high-frequency electronics and optical communications. The requirement for greater bandwidth within AI clusters is accelerating the adoption of silicon photonics and indium-phosphide lasers, increasing requirements for germanium, indium and gallium.
The shift toward optical connections can reduce the amount of copper required within server facilities, while simultaneously increasing demand for these specialised semiconductor and photonics materials.
China remains the central supplier across the strategic-mineral supply chain. It is the leading refiner for 19 of the 20 strategic minerals monitored closely by the IEA, with an average market share of approximately 70%. Export licensing has already affected gallium, germanium, graphite, antimony, tungsten and rare-earth products. In June 2026, China continued withholding several heavy rare earths from Japan and recorded no yttrium exports to the United States for a second consecutive month, demonstrating the significance of small-volume minerals within international supply chains.
Semiconductor growth diverges across materials and technologies
Semiconductor demand is producing different conditions for individual materials. Rapid expansion in advanced logic and high-bandwidth memory is supporting demand for electronic-grade silicon, copper, tantalum, tungsten, ruthenium and hafnium, alongside materials used in advanced packaging. By comparison, consumer electronics and conventional automotive semiconductor markets are expanding more slowly, resulting in weaker conditions for portions of mature-node manufacturing capacity.
Silicon carbide and gallium nitride remain strategically important, but their markets are no longer uniformly characterised by shortages. Investment in Chinese silicon-carbide production has created the possibility of wafer and device oversupply, putting pressure on prices and western producers.
Demand for SiC and GaN continues to expand across electric vehicles, charging infrastructure, renewable-energy inverters and data-centre power-conversion systems. At the same time, manufacturers face the requirement to reduce costs while transitioning from 150mm to 200mm wafers. That shift has turned competition among STMicroelectronics, Infineon, Wolfspeed, onsemi, Rohm and Chinese producers into a contest centred on manufacturing scale and yield rather than simply supplying markets facing shortages.
Solar materials face contrasting supply conditions
The solar industry is also divided between oversupplied and specialised material markets. Chinese capacity expansion has outpaced installations for polysilicon, wafers and conventional solar modules, leaving prices low and producer margins under pressure.
Silver remains an important input for solar manufacturing, while thin-film technologies generate specialised demand for tellurium, cadmium, indium and gallium. Solar manufacturers are continually reducing the quantity of silver used per watt. Despite that trend, continued global growth in installations constrains the decline in overall silver demand.
Advanced applications broaden the critical-material base
AI and advanced electronics are also increasing the relevance of materials that receive less attention in conventional mining markets. Rhenium is used in high-temperature turbine superalloys, while beryllium serves aerospace, satellite and precision-instrument applications. Niobium is used in superconducting and specialised alloys, titanium in aerospace structures, and high-purity alumina in semiconductor equipment, LEDs and battery separators. Other advanced materials are gaining roles in thermal management and high-performance electronics. These include synthetic diamond, silicon nitride and technical ceramics.
Electronic waste becomes a secondary source of technology minerals
Electronic waste is emerging as a major secondary resource for technology minerals. Materials that can be recovered include copper, gold, silver, palladium, tin, cobalt, tantalum and rare earths. The principal challenge is not the presence of metals in discarded equipment but the systems required to collect waste, identify individual components and separate materials economically.
Companies combining automated dismantling, hydrometallurgy and traceable refining can capture substantially more value from electronic waste than conventional scrap-processing operations. Technology-related mineral developments therefore span five interconnected markets: semiconductors and photonics; AI data-centre infrastructure; permanent magnets and robotics; solar and power electronics; and electronic-waste recovery. Increasingly, developments at the intersection of these technology chains, mining and refining are determining material demand and supply conditions across the sector.
