The Jevons Paradox
Why Making Technology More Efficient Often Makes Us Use More of It
In 1865, a British economist proved that making coal engines more efficient would not save coal — it would cause Britain to burn dramatically more of it.
The Idea
The instinct behind green technology is seductive in its simplicity: if we can do the same thing using less energy, we'll consume less energy overall. It sounds like arithmetic. It isn't. William Stanley Jevons noticed something troubling while studying Britain's coal industry. As steam engines became more efficient — more work extracted per unit of coal — coal consumption didn't fall. It exploded. Efficiency had lowered the effective cost of using coal-powered machinery, which made it economical to deploy that machinery far more widely, across more industries, more locations, more hours of the day. The efficiency gain triggered a demand surge that swallowed the savings whole. This is the Jevons paradox: improvements in the efficiency of resource use tend to increase, not decrease, total resource consumption. It operates at every scale. More fuel-efficient cars make driving cheaper per kilometre, so people drive further and buy more vehicles. Better building insulation reduces heating costs, so people heat larger homes to higher temperatures. LEDs use a fraction of the electricity of incandescent bulbs — and so we now light up spaces we never would have lit before: garden paths, decorative strips behind televisions, the undersides of kitchen cabinets. The paradox doesn't mean efficiency is worthless. It means efficiency alone is not a climate strategy. Without a countervailing force — a price on carbon, a cap on emissions, a deliberate cultural shift — efficiency improvements are as likely to accelerate resource consumption as to reduce it.
In the World
Consider what happened to internet data use after broadband speeds improved. In the early 2000s, a faster connection was meant to make the same browsing experience less frustrating — same content, less waiting. Instead, faster connections enabled entirely new categories of content: streaming video, always-on cloud sync, high-resolution imagery, video calls. Netflix alone, by the mid-2010s, was consuming around a third of peak North American internet bandwidth. Each generation of infrastructure improvement — 3G, 4G, 5G — has been met not with stable or falling data demand but with consumption that consistently outruns the efficiency gain. The same dynamic is now playing out in artificial intelligence. Newer AI models are becoming more computationally efficient — producing better outputs per unit of processing power. OpenAI, Google, and others have celebrated these gains as progress toward sustainability. But the efficiency improvements are simultaneously making AI economically viable for vastly more use cases: customer service bots, real-time translation, code generation, image creation, drug discovery. The result is that the total energy demand of global AI infrastructure is rising sharply even as individual models become leaner. The International Energy Agency estimated in 2024 that data centre electricity consumption could double within a few years, driven largely by AI workloads. More efficient AI is not less AI. It is more AI, everywhere.
Why It Matters
The Jevons paradox should sit somewhere in the back of your mind whenever you hear a technology company announce that its new product or process is dramatically more efficient. That claim may be entirely true and still tell you almost nothing about whether resource consumption will fall. The question that matters is not 'how efficient is this?' but 'what does the efficiency make newly possible, and what will people do with that?' This reframe is useful beyond environmental debates. It applies to time: productivity software that saves you two hours a week rarely results in two hours of rest — it results in two hours of additional work becoming feasible. It applies to money: cheaper access to something tends to mean more of it gets consumed, not the same amount for less outlay. Recognising the paradox won't make you a pessimist about technology. It will make you a more precise optimist — one who understands that the value of efficiency gains depends entirely on whether they're paired with constraints, incentives, or norms that prevent rebound consumption from erasing them.
A Question to Ponder
Is there something in your own life where a genuine efficiency improvement has quietly led you to consume more — of time, energy, attention, or resources — than you did before?
Get a new one of these every morning.
Start learning with Thinkable