Jensen Huang actually explained the investment logic of AI a long time ago. I broke this logic down into 5 layers: Layer 1: Energy No electricity, no AI. Power generation, nuclear energy, grid upgrades, energy storage—these are the foundational supports for AI's expansion. Layer 2: Chips GPUs, wafer manufacturing, semiconductor equipment, optical communication, storage—this is the 'heart' of AI. Layer 3: Infrastructure Data centers, servers, liquid cooling, power supply systems, compute leasing—this is the middle layer for scaling AI deployment. Layer 4: Models Whoever trains, deploys, and operates large models holds the core capabilities of AI. Layer 5: Applications The ones making the big money might not be the models themselves, but the software and application companies turning AI into real revenue.