The AI Employment Dividend Was Supposed to Be Bigger by Now

On the surface, the American labor market remains robust. Unemployment sits below four percent. Pay growth, while moderating, has not collapsed. The headlines write themselves: the economy is fine.

But look at the trend line rather than the snapshot, and a different picture emerges. So far in 2026, the economy has added an average of 68,000 jobs per month. That compares with 49,000 per month in 2025, 186,000 per month in 2024, and 251,000 per month in 2023, according to Unusual Whales data citing federal employment surveys. Three years ago, the economy was creating jobs at nearly four times the current rate. The deceleration is not a blip. It is a structural shift that the political class has not yet been forced to confront.

The conventional explanation for slower job growth is a cooling labor market — the natural result of an economy that created too many positions too quickly during the post-pandemic rebound. That reading is not wrong, exactly. But it is incomplete. The more uncomfortable question is whether the technologies that were supposed to supercharge productivity are simultaneously displacing the workers those productivity gains were supposed to benefit.

The Cloudflare Precedent

The announcement that Cloudflare was cutting staff for the first time in its sixteen-year history landed with unusual force on the business press. The company, which built much of its brand on the idea that it was a different kind of infrastructure company — employee-owned, mission-driven, built to last — framed the move as a strategic realignment driven by AI-driven efficiency. According to reporting by Unusual Whales, this was the first mass layoff in the company's history.

That framing — mass layoffs as an AI efficiency story rather than a failure story — has become the industry's preferred vocabulary. When companies that spent years hiring aggressively announce cuts, the communications playbook now runs through a single phrase: we are becoming more efficient with AI. The implication is that the workers being cut were redundant, that the technology has rendered their functions obsolete, and that the company is simply acknowledging an arithmetic reality.

The problem with that framing is that it is doing enormous ideological work that the evidence does not fully support. The job numbers suggest something more gradual and more ambiguous than a clean technological substitution. If AI were simply automating the tedious work and leaving human judgment intact, one would expect to see that judgment-intensive sectors — law, finance, medicine, consulting — continue to hire at elevated rates even as routine-back-office functions contract. What the data shows instead is broad-based deceleration across nearly every sector.

The Numbers Are Not Noise

It is worth being precise about what the employment statistics actually show. The 2023 pace of 251,000 jobs per month was itself inflated by post-pandemic catch-up dynamics — a hospitality sector restaffing after shutdowns, a consumer goods supply chain normalizing, a public sector adding back pandemic-era cuts. Strip out that rebound effect, and the pre-pandemic pace of job creation was closer to 150,000 to 180,000 per month. By that baseline, 68,000 jobs per month represents something closer to a labor market in stagnation than a labor market in equilibrium.

The political salience of these numbers is obvious. An administration that staked its economic credibility on AI-driven growth miracles is receiving data that points in the opposite direction. The jobs are not disappearing in a dramatic wave; they are simply not being created at a rate that can absorb labor force growth, immigration-driven expansion, or the normal churn of workers transitioning between jobs. The headline unemployment rate, which lags hiring data by months, has not yet registered the implications.

There is a plausible alternative reading. The 2024 and 2025 figures reflect an economy absorbing the aftereffects of aggressive Federal Reserve tightening; the slowdown is monetary rather than technological. This publication finds that reading insufficient on its own terms. Monetary tightening does not explain the specific sector composition of the slowdown, the fact that professional and business services — the sector most exposed to AI adoption — has weakened disproportionately, or the growing gap between corporate profit margins (which remain elevated) and hiring rates (which do not).

The Story the Tech Sector Does Not Want Told

For a decade, the technology industry cultivated a self-image as an engine of inclusive prosperity. The rhetoric around AI productivity gains — a phrase that appears in nearly every major company's earnings call — was supposed to translate into something tangible for ordinary workers: higher wages, more jobs, a broader distribution of the gains from automation. What the employment data is actually showing is a sector that has extracted the productivity gains and returned them almost entirely to shareholders.

This is not a new dynamic. Corporate automation has, in most historical cycles, concentrated gains among capital owners while displacing or stagnating wages for the workers it was supposed to augment. The difference now is the speed of adoption and the scale of the claims made in advance. The AI industry spent years telling policymakers, investors, and the public that this time would be different — that the technology was general-purpose, that it would create more jobs than it displaced, that the productivity dividend would be broadly shared. The jobs numbers, three years into the commercial AI boom, do not bear that out.

None of this means AI has failed as a technology. The productivity gains are real; the companies deploying AI are generating more output per worker than they were before. What the data suggests is that the gains are accruing to a narrower set of actors than the public-relations campaign promised. The dividend, such as it is, has been distributed primarily upward.

Cloudflare's first mass layoff is a data point, not a verdict. But data points accumulate. And the pattern they are forming — slower hiring, concentrated displacement, elevated profit margins alongside stagnant wages — is one that an honest accounting of the AI economy would have to acknowledge rather than explain away.