The dominant AI narrative centers on job destruction — how many roles will vanish, how many workers will be replaced. But this framing treats the total amount of work as fixed. Economic history tells a different story through a principle called Jevons Paradox: when the cost of a resource drops dramatically, total consumption does not shrink. It explodes.
AI is compressing the cost of cognitive work at unprecedented speed. Workers using AI tools report 66% productivity improvements, and the number of roles requiring AI fluency has grown sevenfold in just two years, from roughly one million in 2023 to around seven million in 2025. If the paradox holds — and structural indicators strongly suggest it does — total demand for human judgment, creativity, and domain expertise is poised to surge, not collapse.
The Pattern: Cheaper Resources Create Larger Markets
In 1865, English economist William Stanley Jevons published The Coal Question, observing that James Watt’s more efficient steam engine had not reduced Britain’s coal consumption. It had accelerated it, because cheaper energy made new industrial applications viable that were previously too expensive to attempt.
The pattern repeated with computing. Early computers were room-sized machines delivering slow calculations at enormous cost. Modern processors deliver roughly a trillion times more computation per unit of energy. Yet total computing energy consumption has soared, because cheap computation created personal computing, the internet, mobile platforms, and cloud infrastructure — entirely new industries employing hundreds of millions.
Even the paperless office prediction failed the same way. In the 1980s, experts forecast that computers would eliminate paper. Instead, because digital tools made document creation effortless, global paper consumption tripled between 1980 and 2000.
The mechanism is consistent: efficiency improvements do not shrink markets. They expand them by making previously unviable applications economically feasible.
AI Follows the Same Trajectory
Microsoft CEO Satya Nadella invoked Jevons Paradox directly in January 2025, responding to DeepSeek’s efficient R1 model by posting that cheaper AI would see usage “skyrocket.” The early data supports him.
GitHub Copilot now has over 20 million users and generates 46% of code written by developers on the platform. Users complete tasks 55% faster, with pull request cycles dropping from 9.6 days to 2.4 days. Yet Copilot adoption is associated with a small increase in software engineering hiring, not a decrease, because cheaper code production makes more software projects economically viable.
The Bureau of Labor Statistics still projects approximately 15% growth in software development jobs from 2024 to 2034. The software development market itself could grow at 20% annually, reaching $61 billion by 2029. However, the picture is not uniformly positive: employment among developers aged 22 to 25 fell nearly 20% between 2022 and 2025, suggesting that the expansion benefits experienced workers while entry-level pathways are shifting.
The Wrong Question vs. The Right Question
Most boardroom conversations ask: “How many fewer people do we need?” This assumes a fixed amount of work and optimizes for capturing savings. It is the wrong question.
The right question is: “What can we do now that was previously impossible?” This assumes the total opportunity was artificially constrained by execution cost. Removing that constraint creates a larger market than any headcount reduction could capture.
A January 2026 survey by EY-Parthenon found that 69% of CEOs believed AI investments would maintain or increase headcount. A February 2026 study of 12,000 European firms found that AI adopters saw 4% productivity gains without reducing staff. The companies treating AI as a growth engine, not a cost-cutting tool, are pulling ahead.
The Bottleneck Shifts to Human Capacities
When execution cost drops by orders of magnitude, the bottleneck moves from “can we build it?” to “should we build it?” — a fundamentally human question. Several capacities become dramatically more valuable:
- Hypothesis generation — spotting opportunities and framing the right experiments
- Customer intuition — understanding what users actually need versus what they say they want
- Domain expertise — the doctor, engineer, or logistics manager who knows what software their world requires
- Creative vision — imagining products and experiences that do not yet exist
- Contrarian judgment — seeing what everyone else is missing
Today, people with these skills spend most of their energy shepherding a single bet through an organization. In the new paradigm, they generate and evaluate ten bets a week. The human role is not smaller — it is bigger and qualitatively different.
Advertisement
The Unsolved Problems Are Enormous
The strongest evidence for the Jevons Paradox in AI comes from the scale of problems the world has not yet addressed:
- Personalized education — every student learning at their own pace with adaptive curriculum, still largely unrealized despite decades of research
- Clinical decision support — individual patient-level tools integrating medical history, genomics, and current research remain expensive and fragmented
- Financial inclusion — the World Bank’s Global Findex 2025 reports that 1.4 billion adults remain unbanked, locked out of the formal financial system
- Niche software — countless industries still run on spreadsheets because custom software was never economically viable
These are not unsolved technical problems. They are unsolved economic problems. The cost of building solutions has simply been too high. That constraint is now disappearing, and the World Economic Forum projects that 170 million new jobs will emerge by 2030 as these markets open up — a net gain of 78 million positions after accounting for 92 million displaced roles.
What This Means for Companies
Cost-Cutters vs. Builders
Companies are splitting into two camps. Cost-cutters pocket AI efficiency savings — fewer people doing the same work. Builders use the same efficiency to expand what is possible — the same people, or more, doing radically different work.
History’s verdict on this split is unambiguous. In every previous efficiency revolution, the builders won and the cutters lost their market position. McKinsey Global Institute estimates that AI-powered agents could unlock roughly $2.9 trillion in annual economic value in the United States alone by 2030 — but only if organizations redesign work around human-AI partnerships rather than automating tasks in isolation.
The Upskilling Imperative
The hardest challenge is not technical. It is redefining what upskilling means when the job is no longer “do the same thing faster” but “do something you have never been asked to do before.” The WEF reports that 40% of job skills are expected to change by 2030, with 63% of employers citing the skills gap as their primary transformation barrier.
Organizations that succeed will need leaders who give permission to experiment, teams that embrace rapid iteration, individuals who expand their sense of what is possible, and incentive structures that reward ambition alongside efficiency.
Conclusion
Displacement will happen in specific roles and industries — that reality should not be minimized. The entry-level software engineering market is already contracting even as the overall sector grows. But the net effect, if 160 years of economic history and current data are any guide, will be expansion. The companies, teams, and individuals who understand Jevons Paradox and position for the expanded pie will define the next decade. The ones stuck in the cost-reduction frame will wonder what happened.
FAQ
Does Jevons Paradox guarantee AI will create more jobs than it destroys?
Not automatically. The paradox describes a strong historical pattern, but it requires that cheaper intelligence opens genuinely new applications rather than simply replacing existing workers. The WEF projects a net gain of 78 million jobs by 2030, but displacement will be painful in specific roles and regions, particularly for entry-level positions.
Which skills become most valuable as AI gets cheaper?
Domain expertise, hypothesis generation, customer intuition, creative vision, and contrarian judgment. These are the human capacities that cannot be automated and that become the new bottleneck when execution cost drops. People who combine deep domain knowledge with AI fluency will command the highest premium.
How should companies decide between cost-cutting and expansion?
History consistently favors the builders. Companies that use AI efficiency gains to expand what is possible — entering new markets, serving underserved customers, building previously unviable products — outperform those that simply reduce headcount. McKinsey estimates $2.9 trillion in annual value is available in the US alone, but only through work redesign, not task automation in isolation.
Sources & Further Reading
- Future of Jobs Report 2025 — World Economic Forum
- Why the AI World Is Suddenly Obsessed with Jevons Paradox — NPR Planet Money
- AI in Software Development: Creating Jobs and Redefining Roles — Morgan Stanley
- Research: How AI Is Changing the Labor Market — Harvard Business Review
- Generative AI and the Future of Work in America — McKinsey Global Institute
- Evaluating the Impact of AI on the Labor Market — Yale Budget Lab
- What Is Jevons Paradox? And Why It May or May Not Predict AI’s Future — Northeastern University
- Global Findex Database 2025 — World Bank
