Idea
Efficiency: A Glutton in Disguise
Every industrial revolution has repeated the same promise — less energy, less material, less cost. And every one has ended in more of all three.
Abstract
Efficiency is capitalism's most innocent-sounding word. The essay takes Jevons's coal paradox as a template and follows the pattern through the LED bulb, the fuel-efficient engine, the low-power chip, and the AI model: whenever a technology reduces the input required per unit of output, the total input consumed rises, because the system uses the saving to do more of the thing. The result is that efficiency-led narratives — including the current AI-and-climate one — should be read as production strategies, not conservation strategies.
5-Second Answer
Efficiency lowers the cost of using more. Under normal market conditions, the system responds by using more. Presenting efficiency as conservation is either naive or convenient.
Key Arguments
- Jevons's coal paradox (1865) is not a curiosity — it is the general case for any productivity-improving technology in a growth-oriented system.
- Every subsequent efficiency cycle (LEDs, engines, refrigeration, compute) shows total consumption rising even as per-unit consumption falls.
- AI is on the same trajectory: efficient per query, ravenous at aggregate — with datacenter electricity demand projected to roughly double by 2030.
- Framing efficiency as a climate solution therefore requires either capping the total, or admitting that the pitch is about growth rather than reduction.
Analysis
The essay is structured as a pattern rather than a single case. In 1865, William Stanley Jevons noticed that as steam engines became more efficient, coal consumption rose rather than fell — the productivity gain was reinvested in more industrial activity. That is not a historical accident; it is what happens when the constraint on activity is cost, and the cost drops.
The piece walks the pattern through the twentieth century — Edison's lighting, the diesel engine, the modern refrigerator — and into the current one — the LED, the electric motor, the mobile chip, the large language model. In each case the per-unit metric improves and the aggregate metric worsens.
The conclusion is not that efficiency should be abandoned; it is that efficiency without a binding cap on total consumption is a production strategy dressed as an ecological one.
Counterarguments
Analysts sometimes argue that in specific sectors decoupling has been achieved and Jevons no longer applies. The essay concedes that decoupling is possible in principle, but points to the aggregate global data (energy, materials, emissions) as the test — and there, the pattern has not yet broken.
Implications
For policy: pair every efficiency programme with a cap on the underlying total, or expect the saving to be re-spent. For companies: publish absolute figures alongside intensity figures. For readers: treat ‘we did the same job with less’ as one half of a claim whose other half is ‘so we did more of the job.’
Related concepts
Efficiency Debt
The hidden liabilities — resource, cognitive, institutional — that accumulate when efficiency gains are booked without accounting for what they externalize.
The AI–Sustainability Paradox
The same technology that promises to accelerate sustainability transitions is also, at aggregate scale, one of their fastest-growing structural obstacles.
Related research
References
Signal, not noise.
A monthly briefing on AI, sustainability, and the future of human judgment — filtered from global reports, research, and emerging debates.
Join the BriefingMore ideas