There’s a new economic term making the rounds in tech circles. Microsoft CEO Satya Nadella spotlighted it in his January 2025 post on X. Then, a senior leader brought it up in a recent conversation. Having heard it more than twice in the past year, I got curious: why is Jevons’ Paradox suddenly everywhere in tech?
In 1865, English economist William Stanley Jevons observed something counterintuitive about coal consumption in Britain. As steam engines became more efficient at burning coal, he expected demand to fall. Instead, it soared. This phenomenon, known as Jevons’ Paradox, reveals a fundamental truth: making something cheaper and more efficient often increases rather than decreases its demand and total consumption.
Today, the same logic is being applied to artificial intelligence, with tech leaders predicting that as AI becomes more efficient, our demand for it (and for AI-driven work) will only rise.
The Paradox Explained
When a resource becomes more efficient to use, its cost drops. This triggers three effects: existing users consume more, new users enter the market, and entirely new applications emerge. While originally applied to coal, Jevons’ paradox has held true with fuel efficiency leading to more driving, LED lights increasing total lighting usage, and faster computing creating higher demand for data processing.
Take fuel-efficient cars. We made them more efficient hoping to help the environment through lower fuel consumption and less pollution. But that didn’t happen. Instead, people drove more, bought more cars, consuming just as much fuel if not more, cancelling out any environmental gains.
The paradox challenges our intuition because we naturally focus on the direct effect while ignoring the systemic response. We see the efficiency improvement but miss how it reshapes the entire economic landscape around that resource.
AI & the Paradox
Gen and Agentic AI have brought this dynamic into sharper focus. Tasks that once required hours take minutes. Content that previously demanded specialized skills can now be generated by anyone, flooding the internet with blogs, images, video and music. Hallucinations aside, analysis that required teams of experts can be performed by algorithms. GitHub Copilot and similar AI coding assistants enable generation of more lines of code, often dramatically so.
Instead of generating the same amount of content faster, we’re seeing an explosion of content. Businesses that once published weekly blog posts now publish daily. Marketing teams that created dozens of variations now create thousands. Companies can build more features, tackle more ambitious projects, and enter markets that weren’t previously viable.
Like with coal, this expansion isn’t just about doing more of the same things. AI efficiency enables entirely new categories of products and services. With AI now at the hands of everyone at increasingly cheaper costs, Jevons’ Paradox insinuates that each efficiency gain unlocks new applications, and each new application drives more AI consumption. It is no wonder that everyone is in a hurry to invest and win this new race.
Jobs, Skills and Transition
The hope is that impact on employment follows an equally paradoxical path. Jevons’ Paradox suggests that making cognitive labour more efficient should increase the total demand for it, even if it is only to burn developer bandwidth checking the gazillion lines of code AI has retched out.
Take, for example, radiologists who are worried that AI will read X-rays and eliminate their jobs. Instead, AI-assisted diagnosis has made medical imaging cheaper and faster, leading to more scans being ordered, more conditions being detected earlier, and more demand for radiologists to interpret complex cases and make treatment decisions. The nature of the work has shifted, but the demand has grown.
The paradox also extends to entirely new job categories. AI hasn’t just made existing jobs more efficient; it’s created new ones. Prompt engineers, AI trainers, model evaluators, AI ethics specialists, and AI integration consultants are roles that barely existed three years ago. As AI becomes more powerful and pervasive, the ecosystem of jobs around developing, deploying, managing, and governing it expands.
The Limits and Complications
But Jevons’ Paradox doesn’t mean AI will never displace workers or that all efficiency gains lead to increased consumption. The paradox operates at the system level over time, but individuals and specific roles can still be disrupted in the short term.
A radiologist whose specific skill becomes less valuable needs time and resources to develop new ones. A content writer whose niche becomes automated needs pathways to higher-value work. The paradox suggests that aggregate demand will grow, but it doesn’t guarantee that every individual will benefit or that the transition will be smooth.
There are also natural limits to consumption, even in an age of abundance. We only have so many hours to consume content or use apps, and only so many products we truly need. While we haven’t hit those limits yet, Jevons’ Paradox isn’t infinite – it operates within human and market boundaries.
In summary…
Understanding Jevons’ Paradox helps us understand why the tech leaders are talking about it again. The Paradox reminds us that technological change is fundamentally about abundance, not scarcity. The steam engine made mechanical power abundant. Electricity made energy abundant. Computers made information processing abundant. AI is making cognitive work abundant
The goal is rarely to do the same things with fewer resources, but to do dramatically more things or entirely new things. AI efficiency allows for market expansion, not just cost reduction. Companies that use AI primarily for headcount reduction are missing the larger opportunity to expand their addressable market and create new value.
Maybe Elon Musk’s vision of not needing to save for retirement may come true someday but right now it’s a pipe dream; one I won’t sell. Ultimately, whether Jevons’ paradox is an apt reference when it comes to AI is something time will tell. Until then, at least you know what the term means next time. And hey – if companies start saying they need fewer people thanks to AI, maybe introduce them to Jevons’ paradox.
The HR Business Partner Story 