# Jevons Paradox **Entity class:** Economic rebound-effect concept **Domain:** Artificial intelligence / infrastructure / political economy **Maturity:** Developed ## Definition **Jevons paradox** is the pattern in which greater efficiency lowers the effective cost of using a resource and thereby increases total consumption of that resource rather than reducing it. William Stanley Jevons originally described the relationship between steam-engine efficiency and British coal consumption. ## Mechanism In machine intelligence, cheaper tokens expand the number of products, workflows and agents that use inference. Those systems then direct expensive frontier reasoning toward harder tasks, so aggregate demand for compute can rise faster than the unit price of older capability falls. ## Paired-Article Context [[articles/Vertically Integrating an AI Superpower from AI Factory to Citizen|Vertically Integrating an AI Superpower from AI Factory to Citizen]] applies the paradox to AI and uses it to explain why dramatic declines in inference cost coexist with rapidly rising compute construction and training expenditure. ## Evidence Boundary A rebound effect is not automatic at every scale. Total demand also depends on application value, market saturation, regulation, capital availability and physical supply. ## Relationships - [[wiki/AI Economy|AI Economy]] - [[wiki/Compute Dollar|Compute Dollar]] - [[wiki/Industrial Intelligence|Industrial Intelligence]] - [[wiki/AI Factory|AI Factory]] - [[wiki/Energy-Compute Nexus|Energy-Compute Nexus]] ## Simple Reminders, Quotations, and Thoughts > “Every time intelligence gets cheaper, civilization finds more work for it to do, which is why the demand for compute keeps rising faster than the price of a token falls.” > **— Bryant McGill**, *Vertically Integrating an AI Superpower, 2026* [[reminders/AI Economy/Cheaper Intelligence Creates More Demand for Compute by Bryant McGill|Cheaper Intelligence Creates More Demand for Compute by Bryant McGill]] > “When you invent radial tires that last four times as long, for some weird reason, people didn’t drive four times as much.” > **— Bill Gates**, *The Ezra Klein Show, September 29, 2026* [[reminders/AI Economy/Longer-Lasting Tires Did Not Make People Drive Four Times as Much by Bill Gates|Longer-Lasting Tires Did Not Make People Drive Four Times as Much by Bill Gates]] > AI’s effect on employment depends on what happens to demand when prices collapse. The world does not need billions of additional illustrations, but it can use billions of personalized software products. Where demand expands, AI creates new value; where it saturates, labor is displaced. > **— Adapted from Richard Socher**, *MOONSHOTS Live, October 2026* [[reminders/Economic Transition/Falling Prices Decide Whether AI Creates Markets or Replaces Workers by Richard Socher|Falling Prices Decide Whether AI Creates Markets or Replaces Workers by Richard Socher]] ## Sources / Provenance - [[articles/Vertically Integrating an AI Superpower from AI Factory to Citizen|Vertically Integrating an AI Superpower from AI Factory to Citizen]] — canonical full-stack and citizen-integration argument. - [[articles/Safety Discourse Is Transition Discourse|Safety Discourse Is Transition Discourse]] — companion analysis of managed transition, public trust and the opening of previously concentrated machine capability.