Im Ji-yong

AI frontier labs cannot achieve profitability or maintain super-gap dominance without continuous heavy infrastructure spending. As RSI (Recursive Self-Improvement) takes hold, AI development speed will outstrip human-led cycles, demanding ever-larger data centers, power, and chips. This secular trend makes the entire AI semiconductor/hardware sector a structural winner, with capex accelerating broadly across the supply chain.

Alphabet (Google) is the most comprehensive and stable AI investment because it possesses a full-stack AI value chain: in-house TPU chips, own cloud infrastructure, a leading frontier AI model (Gemini), massive user distribution via Search/Gmail/Calendar, strong free cash flow, and accumulated capital. This self-sustaining ecosystem offers greater stability compared to pure-play AI labs that are still loss-making.

AI infrastructure investment must accelerate because the token generation cost (C) will fall faster than token pricing (P), driven by hardware improvements. NVIDIA and similar chipset companies hold the key to this cost reduction, as they control the efficiency and speed of token production. As AI labs compete to lower costs and achieve super-gap advantages, capex into AI hardware will only increase, making NVIDIA the central enabler of the entire AI economy.