Part I · Money: Capital Cycles and the Physics of Profit

Part I is the financial foundation of the book. It asks a deliberately unromantic question: when money enters the AI industry, where does it become profit, where does it pass through, and where does it become someone else’s future write-down? Without this foundation, the later discussion of models, agents, power, law, and founders becomes too easy to misread as a technology ranking.

The first discipline of Part I is to separate growth from retention. A layer can grow quickly because it is necessary, visible, or strategically urgent. That does not mean it retains profit. The model layer can grow because every customer needs intelligence. Upstream can profit because every model needs compute and power. Downstream can profit because every customer needs workflow and trust. The visible growth and the durable profit may sit in different places.

The second discipline is to separate demand from financing. A signed order is not always evidence of terminal customer demand. It can be evidence of strategic investment, vendor financing, refinancing, or a purchase commitment that depends on future capital. This distinction is uncomfortable because it does not make the contract fake. It makes the contract lower quality as a demand signal.

The third discipline is to separate asset life from accounting life. AI infrastructure can be physically useful after it stops being economically frontier. The residual value may be real, but it may not justify the original purchase price. Accounting can smooth that mismatch for a while; markets eventually reprice it through depreciation changes, impairments, lower resale values, or lower utilization.

The fourth discipline is to separate depth from cleanliness. A profit pool can be deep and dirty, shallow and clean, deep and real, or shallow and inflated. The investor’s job is not to admire the deepest pool. It is to pay the right duration for the right quality of profit. The founder’s job is not to run toward the loudest pool. It is to find a clean, deepening pool where incumbents have not already captured the rent.

The four chapters in Part I build these distinctions in sequence. Chapter 1 explains why model-layer revenue can be spectacular while model-layer profit leaks. Chapter 2 shows why the capex wave is a depreciation race, not only an investment boom. Chapter 3 shows how circular financing damages price discovery. Chapter 4 turns those mechanisms into a profit map. The rest of the book depends on that map.