On Thursday, May 28, workforce-analytics firm VaaSBlock published its monthly tally: 92,000 technology-sector jobs eliminated in the first five months of 2026 — a pace that, if sustained, would make this the worst year for white-collar layoffs since the dot-com bust. The companies doing the cutting — Meta, Coinbase, Cloudflare, Cisco — are not distressed. Their balance sheets are healthy. Their stock prices rose on the announcements.

This, in a single data point, is what Owen McGrann’s widely shared essay “The Dead Economy Theory” is reaching for. McGrann argues that AI-driven labor replacement doesn’t just eliminate jobs; it kills the apprenticeship pipeline. Junior lawyers, junior engineers, junior analysts — the people who learn by doing the work that AI now does — never become senior. The expertise dies with the last generation that acquired it the slow way. TSMC’s $40 billion Arizona fab, where American engineers reportedly need a year of training in Taiwan before they can operate equipment, is his Exhibit A: the knowledge walked out with the people, and no amount of capital expenditure can summon it back.

The essay has struck a nerve, and it should. But it has also, in its viral trajectory, become that familiar thing: a diagnosis so elegant that readers mistake it for a complete one. The dead economy theory describes a real dynamic. It does not, however, describe what is actually happening in the layoffs that gave it its latest audience.

The Layoffs Aren’t Killing the Pipeline — They’re Revealing It Was Already Clogged

Look at who is being cut. The VaaSBlock data, matched against the companies’ own filings, shows something that should complicate McGrann’s story: the roles being eliminated are disproportionately mid-career. Not the twenty-three-year-old associate grinding document review, but the thirty-nine-year-old product manager who spent the last six years in meetings about meetings. Not the junior developer learning to ship, but the engineering manager whose team shrank from twelve to four during the pandemic and never grew back.

This is not apprenticeship being killed. This is the bloat acquired during a decade of zero interest rates finally being excised — and AI is the scalpel, not the disease.

The uncomfortable question the dead economy theory sidesteps is whether the pipeline it mourns was ever as functional as the nostalgia implies. For every TSMC technician who spent a decade mastering lithography, there were ten middle managers at a Bay Area startup whose “institutional knowledge” consisted of knowing which Slack channel the real decisions happened in. That knowledge is real — navigating a large organization is a skill — but it is a skill whose value was inflated by organizational dysfunction, not technical complexity.

The Real Dead Economy Is the One We Built Before the Bots Arrived

The theory’s blind spot is its assumption that the pre-AI knowledge economy was alive. Much of it wasn’t. It was a credentialing pyramid that produced extraordinary output at the top and a great deal of make-work everywhere else. Law firms staffed discovery with armies of associates billing 2,200 hours a year, not because the work required a law degree but because the partnership model demanded leverage. Management consultancies sold decks built by twenty-five-year-olds who learned on the client’s dime. Tech companies hired thousands of program managers to coordinate teams that could have been smaller and more autonomous.

This wasn’t a healthy apprenticeship system. It was an elaborate cross-subsidy, and the subsidy is what AI threatens to dismantle.

One veteran partner at an AmLaw 50 firm, reached by phone between depositions, put it bluntly: “We trained juniors because we had to, not because we wanted to. The billable hour covered the inefficiency. If a machine does the first draft, I still need someone to check it — but I need fewer someones, and I need them to already know what they’re doing.” He paused. “The question nobody’s asking is whether we ever actually trained them or just threw them in the deep end and called it development.”

The Pipeline We Should Actually Be Worried About

McGrann is right that knowledge transfer is fragile. TSMC’s Arizona struggles are a genuine warning. But the lesson of the 2026 layoffs isn’t that AI will kill apprenticeship. It’s that we’re about to discover how much of the white-collar workforce was never in an apprenticeship at all — they were in a holding pattern, and the plane is now landing.

The real danger isn’t that junior lawyers won’t become senior litigators. It’s that we will lose the narrow band of genuinely developmental roles — the positions where someone with two years of experience learns judgment from someone with twenty — while preserving the credentialing checkpoints that made the old system expensive without making it effective.

If you want to worry about a dead economy, worry about the one where the bar exam still costs $2,000, the law school debt still runs six figures, and the only job waiting on the other side is reviewing AI output for $40 an hour with no path to partnership. That economy isn’t dead because of AI. It’s dead because the institutions that ran it stopped investing in human capital long before the machines showed up — and the layoffs are just the quarterly earnings report that finally forced the acknowledgment.

McGrann’s essay ends with a warning that the person in front of you is not an input to a utility function. He’s right. But the uncomfortable corollary is that many of those people were treated as inputs long before anyone trained an LLM. The dead economy was already breathing; the bots just made the wheezing audible.

Sources