Some worry about an AI bubble, others fear missing out on investment opportunities in the AI grand narrative, but I've always emphasized that the best investment opportunity for retail investors is non-ferrous metals. In the California Gold Rush of 1849, the ultimate winners weren't the gold miners, but rather: hardware store owners selling shovels, merchants selling tents, and tavern service providers. In the 2026 AI gold rush, the most stable investments won't be large-scale model companies, but rather: ✅ Power stocks (AI consumes electricity; whoever provides electricity collects taxes) ✅ Copper and aluminum non-ferrous metals (data center infrastructure, demand surges by 170%) ✅ Data center REITs (AI companies are all tenants) Looking at the US-China AI competition, on the surface you see ByteDance, Alibaba, Google, and OpenAI competing at the application layer. But let's look at the strategic deployments at the national level: 🇨🇳 China's strategic deployment: • Silver export controls (effective January 1, 2026, accounting for 23.4% of global trade) • Comprehensive control over rare earths, gallium, germanium, and antimony (accounting for 60-90% of global trade) What is the role of silver? Core Materials for Photovoltaics + AI Data Centers 🇺🇸 US Response: • Copper inventory of 1 million tons (equivalent to the annual output of the world's largest copper mine, a multi-decade high) • Copper listed on the "Critical Minerals List" The Role of Copper? The lifeblood of AI computing infrastructure (electricity transmission) The underlying logic of great power competition: The physical bottleneck of AI computing power = energy + critical mineral supply capacity. Whoever controls the upstream controls the final outcome of AI. This is also why Musk is optimistic about China's AI industry, because China's infrastructure is simply too powerful.

The existing power grid simply cannot support the expansion of AI computing power. Ultra-high voltage and smart grid upgrades are a necessity, as AI data centers consume 3-5 times more electricity than traditional data centers.