Part 4 concludes The Architecture of Capital by examining when dependency stops looking cheap and optionality begins to carry a premium. Through Europe, the United States, China, and South Korea, it shows how capital architectures reveal their hidden liabilities under stress — and why strategic freedom depends on preserving alternatives before they are needed.

AI is not one industry but a layered system of energy, silicon, compute, connectivity, and models — and each layer requires a different form of capital. Part 3 of *The Architecture of Capital* examines why AI leadership depends not simply on how much capital a system has, but on whether its architecture can produce, regenerate, and sustain the right capital at the right duration.

Five systems regenerate capital in five different ways. This essay compares the United States, Europe, Japan, South Korea, and China to show how allocation architecture determines what each system can finance, where dependency emerges, and ultimately who builds AI—and who rents it.

Most discussions about AI focus on capabilities, adoption, and market size. This essay explores a different question: when the ecosystem matures, who will everyone depend on? From NVIDIA and TSMC to the lessons of IBM, it argues that dependency creates leverage, leverage determines value capture, and architecture ultimately determines where power accumulates.

Europe cannot buy its way out of an architectural dependency. Investment can purchase infrastructure, but it cannot purchase sovereignty. This analysis examines why optionality—not scale, self-sufficiency, or consolidation—is the true foundation of resilient AI architecture and long-term strategic autonomy.

AI is no longer primarily a software story. Energy, compute, semiconductors, connectivity, and governance are becoming the defining constraints of the AI era. Infrastructure—not applications—will determine who holds leverage in the decade ahead.