---
id: "concept-ai-jevons-paradox"
type: "concept"
source_timestamps: ["§ The Great Value Loop", "¶9"]
tags: ["economics", "energy-demand", "efficiency"]
related: ["claim-efficiency-increases-demand", "contrarian-efficiency-increases-demand", "entity-deepseek"]
definition: "The phenomenon where increases in AI efficiency lower the cost of intelligence, thereby expanding economically viable use cases and driving total energy demand up rather than down."
sources: ["futures"]
sourceVaultSlug: "hbr-seg-futures"
originDay: 2
articleStem: "hbr-nm-101-energy-strategy-ai"
sourceUrl: "https://hbr.org/2026/06/your-company-needs-an-energy-strategy-for-ais-next-phase"
sourceTitle: "Your Company Needs an Energy Strategy for AI’s Next Phase"
---
# AI Jevons Paradox

## Definition
The phenomenon where increases in AI efficiency lower the cost of intelligence, thereby expanding economically viable use cases and driving total energy demand up rather than down.

## The Mechanism
An application of the classic economic Jevons paradox (originally about coal) to artificial intelligence. Algorithmic and hardware efficiency improvements — such as those seen in [[entity-deepseek-d2]]'s 2025 release — drastically lower the reported training and inference costs of AI. But cheaper *intelligence* expands the number of economically viable use cases. Consequently, **total aggregate demand for AI compute — and therefore electricity — goes up rather than down.**

The strategic implication: efficiency alone will **not** eliminate the impending energy bottleneck. See the claim it grounds — [[claim-efficiency-increases-demand]] — and the contrarian framing that makes it counterintuitive — [[contrarian-efficiency-increases-demand]].

## Enrichment (external validation)
The label "AI Jevons paradox" appears to be the authors' novel framing, but the underlying rebound-effect mechanism is well documented in ICT/energy literature. Americans for Prosperity cites forecasts where AI alone could drive up to a **165% increase in power demand by 2030**; WEF notes AI data-center investment is outpacing grid build-out despite ongoing efficiency gains.

## Nuance
Jevons effects are **context-dependent**. Under binding caps (strict emissions limits or hard power ceilings), efficiency *can* reduce total energy use. But in unconstrained competitive cloud-AI markets, rebound effects dominate, so the expectation of rising aggregate demand holds.


## Related across articles
- [[concept-induced-demand]]
- [[claim-post-covid-downshift]]