Critical Minerals Constrain AI Ambitions

The growth of artificial intelligence (AI) is being hindered by a largely overlooked factor: the availability of critical minerals. The International Energy Agency estimates that data centers will consume over 500,000 metric tons of copper and 75,000 tons of silicon each year by 2030, accounting for around 2% of global demand. This is particularly concerning given China's control of 80-90% of the world's refining capacity for silicon, gallium, and rare earths.

Beijing has imposed export restrictions on several key minerals since 2023, including gallium and germanium. These controls have led to sharp price increases, which in turn are driving up hardware manufacturing costs. The US, EU, Japan, and South Korea have all developed critical-mineral strategies, but experts warn that these efforts may not be enough to address the immediate supply chain issues.

Analyst Taggart cautions that while 'the cash will be there to build out those data centers,' investors are underestimating the physical constraints that could prevent data center expansion from keeping pace with market expectations. The future of AI is likely to be shaped as much by access to critical minerals such as copper, gallium, and silicon as it is by advances in algorithms.