Time to revisit Baumol Cost Disease

Anticipating and converting the cost disease into opportunities with Jevon's and PEMDAS!

AI success is a three-way coordination problem. Jevons determines whether abundance expands use. PEMDAS determines whether expansion precedes extraction. Baumol determines whether emerging scarcity becomes the next investment frontier.

Success occurs when cheaper intelligence multiplies productive participation, opens new value domains and expands the demand that finances further AI. Failure occurs when AI sprawls physically while firms optimize economically for subtraction, shrinking the human participation and purchasing power on which its own market ultimately depends.

The central race is therefore not AI versus humans. It is expansion and reinstatement versus displacement and premature harvest.

Jevons explains why cheaper AI drives more use. The risk is sprawl: more models, agents, compute, chips, energy, and infrastructure without proportional value creation.

PEMDAS is the sequencing rule: multiply capability and expand value before subtracting labor or dividing gains. If firms optimize too early for displacement, they may improve local margins while weakening distributed income, aggregate demand, and eventually the market that pays for AI.

Baumol reveals where value migrates next. As AI makes some capabilities abundant, scarcity shifts to complements such as integration, coordination, accountability, physical execution, trust, and other bottlenecks. Those should become the next investment frontier, not merely the next target for compression.

The success loop is:

cheaper AI → more experimentation → new capabilities → new products and markets → new scarce complements → investment in those complements → more productive participation → stronger demand → more economically justified AI.

The failure loop is:

cheaper AI → sprawl → rising infrastructure cost → pressure for ROI → labor substitution → weaker distributed income and demand → weaker revenues → more cost cutting → more substitution.

History suggests automation does not have one fixed outcome. Displacement can be offset by new tasks, new industries, lower prices, and demand expansion—but that compensation is not automatic.

Core thesis

AI succeeds when abundance expands the economy faster than substitution contracts participation. Jevons drives expansion, PEMDAS protects the order of operations, and Baumol points to the next scarce source of value.