Introducing a Standardized Engineering-Economic Unit of Useful Artificial Intelligence — and a New Field: Artificial Intelligence Energy Economics
By Roger S. Hutchison, Ph.D. — President & Chief Technology Officer, CD ROM, Inc.
Artificial intelligence cannot be systematically optimized until intelligence itself can be systematically measured.
CD ROM, Inc. is pleased to announce the publication of a foundational monograph by Dr. Roger S. Hutchison: The Artificial Intelligence Work Unit (AIWU): A Foundational Monograph (Version 1.0, July 4, 2026).
Why This Paper Matters
Artificial intelligence has become one of the defining technologies of our era, yet a fundamental gap remains at the heart of the industry: there is no standardized engineering-economic unit of useful AI output.
Electricity is measured in kilowatt-hours. Petroleum is measured in barrels. Digital information is measured in bits and bytes. But artificial intelligence — despite consuming enormous and rapidly growing quantities of energy, capital, and critical materials — has no equivalent unit for the useful intelligence it produces. Tokens measure text processing, not usefulness. Benchmark scores measure capability, not production. Without a standardized unit of output, AI productivity cannot be rigorously measured, and the energy required to produce intelligence cannot be systematically optimized.
What the Monograph Proposes
This foundational work introduces three closely related concepts:
- The Artificial Intelligence Work Unit (AIWU) — a proposed standardized measure of useful artificial intelligence output.
- The Energy Intensity of Intelligence (Eii) — the quantity of energy required to produce one AIWU.
- The Fundamental Equation of Artificial Intelligence Energy Economics — a first-order engineering-economic relationship expressing the total energy cost of producing AI as the product of AI output, Energy Intensity of Intelligence, and the unit cost of electrical energy.
Together, these concepts establish Artificial Intelligence Energy Economics (AIEE): a proposed new branch of engineering economics dedicated to measuring, understanding, and reducing the resources required to produce useful artificial intelligence.
Built on a Proven Engineering-Economic Tradition
AIEE extends the well-established theory of Intensity of Use — long applied to metals, energy, and industrial materials — into the domain of artificial intelligence. The framework draws on Dr. Hutchison’s doctoral research in mineral economics at the Colorado School of Mines, applying decades of engineering-economic principles to one of the most pressing questions of the AI era:
How much energy does it take to produce useful intelligence — and how do we reduce it?
Who Should Read This
This monograph is written for an interdisciplinary audience:
- Engineers and computer scientists working on AI systems, hardware, and data-center infrastructure
- Economists and analysts studying AI productivity and energy markets
- Energy researchers and utilities planning for AI-driven demand growth
- Policymakers addressing sustainability, infrastructure, and national competitiveness
Download the White Paper
The complete foundational monograph (Version 1.0, Archival Edition) is available for download below.