For years, economists have squinted at productivity charts looking for evidence that artificial intelligence was doing something—anything—to justify the hundreds of billions being poured into its development. The evidence has finally arrived, and it is exactly as unsettling as the optimists hoped and the pessimists feared: technology workers are becoming dramatically more productive while their ranks thin out.
The pattern emerging from recent labour data represents the first statistically significant signal that AI tools are changing how work gets done, not merely how work gets discussed. In the technology sector specifically, output per worker has climbed even as headcounts have contracted—a combination that suggests companies are genuinely substituting software for people rather than simply trimming fat accumulated during the zero-interest-rate hiring binge.
The productivity puzzle, partially solved
The technology industry has always been an outlier in productivity statistics, but the current divergence is striking. Companies that adopted AI coding assistants, automated testing frameworks, and machine-learning-powered operations tools are reporting that smaller teams can maintain or expand the same codebases and product portfolios. The math is uncomfortable but straightforward: if five engineers with AI tools can do what seven did before, two engineers become redundant regardless of how talented they are.
This is not the dystopian mass unemployment scenario that dominates popular imagination. The job losses remain concentrated in specific roles—junior developers, quality assurance testers, certain categories of data analysts—while demand for AI specialists and senior engineers who can effectively wield these tools has intensified. The labour market is reshuffling rather than collapsing.
Beyond the tech bubble
The more interesting question is whether this pattern will propagate. So far, the evidence outside technology is muddier. Healthcare, finance, and professional services have all invested heavily in AI tools, yet their productivity statistics remain frustratingly flat. The gap suggests that technology companies, being both the creators and earliest adopters of these tools, enjoy advantages that may not transfer easily.
Part of the explanation is cultural. Software engineers are accustomed to having their work automated—it is, after all, what they do to everyone else. They adapted to AI pair programmers with less friction than, say, lawyers adapting to AI contract reviewers or doctors adapting to AI diagnostic assistants. The tools may be equally capable, but the humans are not equally willing.
Our take
We are witnessing the beginning of a genuine economic transformation, not the end of one. The technology sector is the canary, and the canary is singing a complicated song: productivity gains are real, job displacement is real, and both are happening simultaneously in ways that will take years to fully understand. The companies that figure out how to redeploy displaced workers into higher-value roles will thrive; those that simply pocket the savings will find themselves with smaller teams and smaller ideas. The AI revolution is not coming—it is here, measurable at last, and as morally ambiguous as every industrial revolution before it.
