The Company Kill

She checks the stock price before she checks her email.

Not because she's a trader. Because she's a project manager at a consulting firm that bills by the hour, and the number on her phone screen tells her something the Slack channels won't say for weeks. She's watched it fall — not crash, not plummet, just deflate — every morning for eleven days. Like air leaving a tire you can't find the hole in.

She's been there nine years. She onboarded fresh out of grad school, learned to speak the language of legacy systems: migration timelines, client dependencies, the particular patience required to walk a Fortune 500 CFO through why his mainframe is not, in fact, fine. She was good at it. She is good at it. But the thing she's good at — the thing the whole firm is built around — just showed up as a feature demo from a company that didn't exist five years ago.

Nobody's been let go yet. But the parking lot is quieter. People leave earlier. The coffee machine conversations have a new texture — not panic, something softer. A shared uncertainty that nobody names out loud because naming it makes it negotiable, and this doesn't feel negotiable.

Her husband asks her at dinner if things are okay at work. She says yes. But she's already updated her résumé. Not because she knows what's coming. Because the stock price keeps telling her something, and she's run out of reasons to not listen.

The market moved before the memo did.

In late February 2026, Anthropic demonstrated that its AI could modernize COBOL code — the decades-old programming language that still runs banking, insurance, and government systems worldwide. IBM's consulting arm has built an empire around COBOL maintenance and migration. Within hours of the announcement, IBM shares suffered their worst single-day drop in twenty-five years.^[1] Not because IBM lost a contract. Because the market decided that IBM's reason to exist had just been compressed into an API call.

The same week, Block — the payments company formerly known as Square — saw its stock surge more than 15% after announcing it had cut roughly 4,000 employees, about 40% of its workforce.^[2] The message from investors was clean and unsentimental: fewer humans, higher valuation.

Anton Korinek, an AI economist at the University of Virginia, frames it as "ten times bigger than the 1990s internet disruption" — not because the technology is ten times better, but because AI disrupts "cognitive production at large," touching an economic surface area the internet never reached.^[3] The internet disrupted distribution. AI disrupts the thing being distributed.

Keum's research on corporate earnings transcripts reveals a measurable linguistic pivot: firms that adopt AI language in their calls are far more likely to frame headcount as expense rather than capacity.^[4] This isn't an accident. It's a signal to shareholders that the firm understands where the incentives now point. The language is doing real work — it's giving the market permission to reward contraction.

The macro numbers are moving.

U.S. labor productivity has averaged 2.8% since 2023 — double the prior decade's average of approximately 1.4%.^[5] That gap is the statistical signature of firms doing more with fewer people. It is not, in itself, a crisis. Productivity growth is how living standards rise. But when the gains concentrate at the top of the capital stack and the losses distribute across the labor force, you get a divergence that eventually shows up as political pressure, not just economic data.

Federal Reserve Bank of Richmond President Tom Barkin used the phrase "creative destruction" to describe the current landscape — a Schumpeterian framing that Wall Street has adopted with visible enthusiasm.^[1] The enthusiasm is the tell. When central bankers and equity analysts reach for the same metaphor, it means the pattern has passed from theory into pricing.

Citrini Research published a scenario paper in early March 2026 modeling what happens if AI capability gains continue at current pace. It triggered a brief S&P nosedive — not because the scenarios were extreme, but because they were plausible.^[6] The most exposed sectors: firms whose value proposition is human cognitive labor sold at scale. Consulting. Content. Support. Analysis. Code. The very companies that were supposed to be the safe, knowledge-economy employers of the future.

The displacement isn't sudden. It's structural, cumulative, and moving from the labor market into the capital market. The woman checking her stock price every morning isn't watching a company fail. She's watching a category of company become unnecessary. That's not a recession. It's a reclassification.

And the market, as always, noticed first.