Episode Details

On Friday morning, the Bureau of Labor Statistics reported that the American economy shed 92,000 jobs in February. Markets had expected a gain of 50,000. The miss was 142,000 jobs. In a single data release, the question that has been building for months became impossible to ignore: Is AI destroying jobs faster than anyone admitted? Or is something else happening entirely?

I want to give you a precise answer. Not a reassuring one. A precise one.

Because the 92,000 number is not a single story. It is three stories running simultaneously — three separate mechanisms that produce the same headline but require entirely different responses. Conflating them is how the people responsible avoid accountability. Separating them is how the rest of us start to see clearly.

• • •

The first story is the one you have been sold.

Artificial intelligence has arrived. It is replacing human work at speed. Companies that once needed a hundred people now need ten. This is either the most exciting productivity revolution in history or an unfolding catastrophe for working people, depending on whom you ask — but everyone agrees it is real, it is happening now, and it is accelerating.

Except that is not what the data shows.

According to Challenger, Gray & Christmas, approximately 55,000 layoffs in all of 2025 were attributed directly to AI. That is less than one percent of all job separations for the year. A paper from the National Bureau of Economic Research found that ninety percent of executives — nine out of ten people actually making hiring and firing decisions — said AI has had no material impact on workplace employment over the past three years.

Brad Conger, chief investment officer at Hirtle Callaghan, manages twenty-five billion dollars on behalf of institutional clients. He is not a skeptic of AI. He has purchased five AI software products in the past six months. And he is blunt: “A job does 100 things in a day, and that’s a lot more than a single AI workflow can perform. It replaces activities that are just pieces of jobs.” At his firm, he says, not one position has been eliminated because of AI.

This is the first face: the Displacement Story. It is, for the most part, a fiction — or at minimum, a dramatic exaggeration of a real but limited phenomenon. AI is improving workflows at the margins. It is not yet dismantling workforces at scale. The companies claiming otherwise are, in Altman’s own word, washing.

But if AI is not actually doing the work of elimination, something is. Which brings us to the second story.

• • •

When Amazon announced 30,000 job cuts between October and January, its capital expenditures were on a trajectory from fifty-three billion dollars in 2023 to a projected two hundred billion in 2026. When Workday laid off 1,700 employees — eight and a half percent of its workforce — in February, CEO Carl Eschenbach declared the cuts were necessary to “prioritize AI investment and free up resources.” When Block laid off 4,000 people in early March, the company was simultaneously accelerating its investment in AI tooling.

Conger connects the dots directly: “AI’s not replacing jobs, but job cuts are funding AI expenditures.”

This is the second face: the Capex Trade. It is more honest than the Displacement Story, but only slightly — because it obscures what is actually being traded away. When a company eliminates experienced people to fund infrastructure bets, it is not making a neutral financial reallocation. It is liquidating a specific kind of asset — accumulated human knowledge, judgment, and institutional memory — to purchase a speculative one.

In manufacturing, we have watched this calculation fail for thirty years. A plant eliminates its senior machinists to cut costs. Five years later, nobody can diagnose the recurring vibration problem in line three, because the three people who understood it have bee