Jevon's Paradox in motion in the AI space

Jevons' Paradox is an economic principle that states that as technological improvements increase the efficiency of resource use, the overall consumption of that resource may increase instead of decrease. This counterintuitive effect was first described by the 19th-century economist William Stanley Jevons in his 1865 book The Coal Question. He observed that as coal-fired steam engines became more efficient, total coal consumption in England actually increased because cheaper coal encouraged more industries and consumers to use it.

Increased Efficiency --> Lower Cost per Unit of Output
- As technology makes resource use more efficient, the cost per unit of output drops.

Lower Costs --> Higher Demand
- With lower costs, businesses and consumers find new applications and increase their usage.

Higher Demand --> Increased Total Resource Consumption

- The overall increase in demand can outweigh the savings from efficiency, leading to net higher resource consumption.
This is a rebound effect that, when extreme, leads to Jevons' Paradox—where efficiency gains drive total resource consumption up rather than down.

Examples of Jevons’ Paradox

1. Coal and the Steam Engine
- James Watt’s improvements to the steam engine made it much more
fuel-efficient than earlier designs.
-Instead of reducing coal consumption, the lower cost of steam power
spurred industrial growth, increasing coal usage dramatically.

2. Automobile Fuel Efficiency
- Cars today are more fuel-efficient than ever due to advancements in
engine design, hybrid technology, and lightweight materials.
However, total fuel consumption continues to rise because:
-People drive more when fuel costs drop.
-More people can afford cars in developing economies.
-Larger and more powerful vehicles like SUVs become popular
despite efficiency gains.

3. AI Hardware and Energy Consumption
- AI chips (e.g., GPUs, TPUs, custom accelerators like NVIDIA’s H100 and Google's TPUv5) have dramatically improved computational efficiency.
- Instead of reducing power consumption, AI workloads have exploded due to:
-Increasingly complex AI models (e.g., GPT-4, GPT-5, and beyond).
-More AI applications in industries from finance to medicine.
-A growing number of AI datacenters consuming vast amounts of
energy.

Net effect: AI-related electricity consumption is rising sharply, despite more efficient hardware.