At first, everyone thought AI competition was about large models, then gradually realized it’s about computing power. Later, they discovered the bottleneck for computing power is electricity. Looking at electricity, both power supply (small nuclear power) and grid transmission (the U.S. doesn’t have China’s ultra-high voltage transmission technology) are issues. Finally, it turns out we need to focus on copper, lithium, and other non-ferrous metals. This is basically the path for AI computing power hype in the U.S. stock market.