The development of AI has two sides: semiconductor technology, which underpins the hardware foundation for computing and storage. On the other side is energy supply, from traditional thermal power and nuclear power to green energy, all of which are powering AI’s massive energy consumption. Among them, copper, as the “intersection,” is a core material.
From chip interconnects to data center wiring, and then to cables for energy transmission and components in renewable energy equipment, none of it can do without copper’s conductivity and corrosion resistance. Take data centers as an example: AI training and inference require massive amounts of electricity and efficient cooling systems, and copper is the preferred material for power transmission and heat exchangers.
The boundaries of technology are often constrained by energy and by physics. Once the energy problem is solved, AI’s main uptrend will arrive. Right now, it is still in the early stage.
On this front, Musk and xAI have an absolute advantage. Their efficiency and execution in dealing with tight power supply are textbook-level. By contrast, Microsoft bought a pile of GPUs but couldn’t even plug them in, which is too awkward. This shows that today’s standard of capability is completely different from before; what’s needed is full-chain system capability.