Capital accumulation and reproduction cost
Accumulated capital is valuable when competitors must reproduce comparable capital to compete. Rapid technological change can break that relationship.
Suppose an incumbent spent $100 billion over a decade to reach capability 100. A new entrant can now reach capability 90 for $10 billion, while customers require only capability 80.
Historical invested cost = $100B; current sufficient-entry cost = $10B
The incumbent's capital exists and may remain productive, but its historical cost is not the entrant's barrier. It can become legacy capital. The same logic applies to software codebases, data-center generations, manufacturing tooling and organizational structures.
Self-improving automation intensifies this because capital can participate in creating more capital. Robots can build components for robots; AI can design systems; software can generate software. But that does not imply the first accumulator owns all future productive capacity. It can imply faster diffusion of productive capacity if the reproduction process itself becomes accessible.
The relevant moat is therefore current cost of sufficient substitution, not cumulative historical expenditure.
High ROIC can be distributed without first falling
The accumulator's advantage can weaken even if project-level ROIC remains high. Economic profit has two multiplicands:
Economic profit = invested capital × (ROIC − WACC)
If AI reduces the invested capital required to reproduce a sufficient outcome, new entrants can access attractive returns on much smaller capital bases. High ROIC has been distributed as a capability. The original accumulator no longer owns the only capital base capable of earning it.
This does not guarantee that incumbent economic profit falls: the incumbent may reinvest, own another bottleneck or expand demand. It means cumulative historical capital no longer proves exclusive access to the spread.