Boundary cases and objections
34.1 What if intelligence remains permanently scarce?
Then the thesis weakens substantially. If frontier capability remains economically indispensable, alternatives remain far behind, compute or algorithms are controlled by a small number of actors, and the frontier continually opens monetizable tasks faster than old tasks commoditize, extraordinary rents can persist. This is the strongest bullish case and should be stated plainly.
34.2 What if network effects dominate everything?
Then application functionality can commoditize while the network owner remains powerful. The analysis must show that participation itself creates non-replicable value and that AI cannot move the interface or make interoperability sufficient.
34.3 What if regulation protects incumbents?
Regulation can create durable scarcity. Certification, liability, data rights and licensing can make entry expensive even when technology is cheap. But the resulting moat is institutional. A valuation should identify it as such rather than attributing the rent solely to superior intelligence.
34.4 What if new wants are effectively unbounded?
The paper does not require human wants to saturate. New wants can create new markets indefinitely. The claim is only that each new market still has prices, substitutes, budgets, time constraints and competitive supply. Unbounded imagination does not imply unbounded provider rent.
34.5 What if AI creates entirely new factors of production?
That can enlarge real output dramatically. The abundance analysis still applies: the monetary value captured by a factor depends on its scarcity relative to substitutes. A new factor of production can be foundational and cheap.
34.6 What if mass unemployment really occurs?
Then the distribution and demand questions become urgent, but the employment inversion does not disappear. The outcome would imply that complementary assets—capital, distribution, law, physical resources or access to AI—prevent displaced people from converting newly available intelligence into independent productive capability. Those bottlenecks become the real objects of analysis.
34.7 What if AI providers vertically integrate?
Vertical integration can move capture downstream and preserve rent. But it also converts the provider from a seller of intelligence into an owner/operator of the businesses being transformed. The valuation must then be justified by the economics of those downstream activities rather than by an intelligence tax on the entire economy.
34.8 Isn't this simply normal capitalism?
In one sense, yes. Competition, substitution and falling costs are ordinary economics. What is unusual is the breadth of the technology: AI is being valued partly because it can lower the cost of the very activities—research, software creation, coordination, organization and discovery—that normally protect high returns. The ordinary mechanisms therefore act recursively on the technology's own moat.