Copper may become the next major bottleneck for artificial intelligence hardware and the expansion of global data centers.

This potential shortage matters because AI models require massive power and interconnect capacity. Since copper is essential for wiring and power delivery, supply constraints could limit the growth of the industry in the U.S. and China.

Dan Dreyfus, host of the All-In Podcast, said that copper demand for AI over the next 18 years could be equivalent to 10,000 years of historic copper use [1]. This surge in demand is driven by the physical infrastructure required to support large-scale AI training.

Data from China highlights the intensity of this requirement. An AI training data center in China uses 47 metric tons of copper per megawatt installed [3]. By comparison, a cryptocurrency data center requires 21 metric tons of copper per megawatt [4].

However, industry analysts are divided on whether this will create a permanent crisis. While some reports suggest the AI boom is running into a copper problem, other analysts said AI may not be the demand booster that copper bulls expect [2].

Some companies are already working to bypass the material entirely. Nvidia has invested over $6.5 billion in photonics since March 2024 to replace copper with light in AI data centers [5]. This shift toward photonics could reduce the industry's reliance on traditional metal wiring.

Separate from the AI debate, Dreyfus said that U.S. debt totals about $140 trillion [2].

“Copper demand for AI over the next 18 years could be equivalent to 10,000 years of historic copper use.”

The tension between copper dependency and the development of photonics represents a critical pivot point for AI infrastructure. If copper supplies cannot scale with the energy demands of next-generation models, the pace of data center deployment may slow unless alternative light-based interconnects can be commercialized at scale.