---
id: "contrarian-efficiency-increases-demand"
type: "contrarian-insight"
source_timestamps: ["§ The Great Value Loop", "¶9"]
tags: ["efficiency", "energy-demand", "economics"]
related: ["concept-ai-jevons-paradox", "claim-efficiency-increases-demand"]
challenges: "The conventional view that software and hardware efficiency improvements will solve AI's energy consumption problem."
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 efficiency will increase, not decrease, total energy demand

## Contrarian Insight
Conventional thinking assumes that as AI models become more efficient (cheaper to train and run), the energy crisis will naturally subside. The authors argue the **opposite**: due to the Jevons paradox, making intelligence cheaper vastly expands the number of economically viable use cases, driving total aggregate energy demand **up**.

**Challenges:** the conventional view that software and hardware efficiency improvements will solve AI's energy consumption problem.

## Supporting apparatus
- Mechanism: [[concept-ai-jevons-paradox]]
- Formal claim: [[claim-efficiency-increases-demand]]
- Empirical anchor: [[entity-deepseek-d2]]'s 2025 cost drop

## Counter to the counter (from enrichment)
The rebound effect is **context-dependent**. Under strong policy constraints, carbon pricing, or hard caps on compute, efficiency gains *could* reduce total energy use. Separately, AI itself may **mitigate** grid strain — an IEA-linked estimate (via Brookings) suggests AI grid optimization could free up to **175 GW** of transmission capacity. So the pessimistic reading is not inevitable; net impact depends on policy, design choices, and market structure.


## Related across articles
- [[concept-induced-demand]]
- [[contrarian-inefficiency-is-good]]