---
id: "concept-intelligence-per-watt"
type: "concept"
source_timestamps: ["§ The Great Value Loop", "¶9", "§ The Incumbent's Energy Playbook", "¶11"]
tags: ["metrics", "management", "efficiency"]
related: ["action-make-energy-visible", "action-reduce-demand", "quote-intelligence-per-watt-metric"]
definition: "A strategic management metric that quantifies the amount of AI output (e.g., tokens or inferences) generated per unit of electricity consumed."
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"
---
# Intelligence per Watt

## Definition
A strategic management metric that quantifies the amount of AI output (e.g., tokens or inferences) generated per unit of electricity consumed.

## The Idea
Rather than treating energy as an invisible engineering variable, leaders must quantify AI output — measured in workflow completions, tokens, or inferences — generated per unit of electricity consumed (kilowatt-hour). The authors are explicit that precision is not the point: [[quote-intelligence-per-watt-metric]].

## How to Optimize It
- Route simple tasks to smaller models rather than oversized frontier models.
- Cache repeated queries.
- Avoid using frontier models for basic tasks.

## Operationalized By
- [[action-make-energy-visible]] — build a quarterly dashboard that reports tokens/inferences per kWh.
- [[action-reduce-demand]] — the workload-optimization tactics that improve the ratio.

## Enrichment (external validation)
The *term* is proprietary to the authors, but energy-normalized performance metrics are established practice. Cloud providers already ship per-workload/per-instance energy and carbon reporting (Google Cloud Carbon Footprint, Azure Carbon Optimization), which effectively measure output-per-kWh even without the "intelligence per watt" label. Brookings notes disclosure is growing but non-interoperable across firms.