Tony Kim argues that AI is forcing a physical rebuild of the technology stack. Rising model complexity is making compute, memory, bandwidth, power, and data-center design strategic constraints, while market value shifts toward chips and hardware. He outlines an investment framework centered on the current AI buildout but leaves room for asymmetric bets converging around 2030, including quantum computing, robotics, orbital data centers, and advanced energy systems. He sees China’s manufacturing scale as a major robotics advantage, especially when paired with Western AI models, and describes a future enterprise organized around proprietary data, context, agents, and token flow.