Unscarcity
Sign in for free: Preamble (PDF, ebook & audiobook) + Forum access + Direct purchases Sign In

Unscarcity Research

Anticipatory Displacement: Losing Jobs to an AI Strategy

Monday.com cut 620 jobs for an 'AI-driven growth strategy,' then said it wasn't replacing people with AI. Both statements are true. That's the problem.

9 min read 2051 words Updated July 2026 /a/anticipatory-displacement

Note: This is a research note supplementing the book Unscarcity, now available for purchase. These notes expand on concepts from the main text. Start here or get the book.

Anticipatory Displacement: Losing Your Job to an AI Strategy

On July 21, 2026, monday.com filed a Form 6-K with the SEC announcing it would cut about 620 people, roughly 20% of its workforce. The filing gave the reason in the flat language securities lawyers use when they want to be accurate: the restructuring would “support a leaner, more focused operating model as the Company continues to invest in its AI-driven growth strategy.”

Days later, co-founder Eran Zinman told the departing employees something that sounded like the opposite. The decision, he wrote, “was not made to reduce costs or replace people with AI.”

The press covered this as corporate doublespeak, a company caught between the story it tells Wall Street and the story it tells its own staff. That reading is too easy, and it misses what actually happened. Both statements are true. No AI system at monday.com learned to do those 620 jobs. Nobody was automated. What happened is that the company decided to become an AI-first company, and becoming one required money, a different org chart, and a headcount profile that reads as AI-native to investors. The people were removed to make room for the bet.

That is a distinct thing from automation, and it needs its own name. Call it anticipatory displacement: jobs eliminated in anticipation of a capability rather than because of one.

The two curves have come apart

For most of the last three years it was reasonable to treat “AI-attributed layoffs” and “AI-caused layoffs” as the same series with noise in it. In 2026 the two separated far enough to measure.

On the announcement side, the numbers are enormous. Challenger, Gray & Christmas counted 139,156 job cuts in the technology sector through June 2026, up 83% from the 76,214 announced over the same months of 2025. Tech now accounts for nearly a third of all US job cuts. AI has been the single most-cited reason for workforce reductions for four consecutive months, named explicitly in 101,743 announcements this year, about 23% of every cut Challenger tracks.

On the measurement side, the picture refuses to match. Anthropic’s economists Maxim Massenkoff and Peter McCrory published a study in March 2026 building a task-level exposure measure from actual usage data rather than survey guesses. They found no systematic rise in unemployment among highly exposed workers since late 2022. Computer programmers came out as the most exposed occupation in the economy at 75% observed task coverage, followed by data entry keyers at 67% and customer service representatives around 65%. Those workers are not showing up in the unemployment statistics as a displaced class. The most exposed group also turns out to be richer and more credentialed than the unexposed one, with graduate degrees making up 17.4% of it versus 4.5% of the unexposed.

So: record AI-attributed cuts, and no matching displacement of the people whose tasks AI can actually do. The announcements are running ahead of the automation.

The one number that shows where the damage lands

There is a single place in the Anthropic data where something clearly breaks, and it is worth sitting with. Workers aged 22 to 25 entering high-exposure occupations are finding jobs 14% less often than peers entering low-exposure roles. Above age 25, the effect vanishes.

If AI were eating work, incumbents would be pushed out of it. Instead the door is closing on the people trying to get in. That is not what automation looks like. It is what a hiring freeze on the junior rung looks like, which is the cheapest cut a company can make and the one that generates no severance line, no WARN notice, and no press cycle. You simply don’t open the requisition, and if anyone asks, you say AI.

This is the same mechanism the solo unicorn runs on. The Labor Cliff was never going to arrive mainly as mass firings. It arrives as jobs that never get posted, which is why it stays invisible in the employment statistics that only count people who lost something they already had.

Why a company would do this on purpose

The incentive is not subtle once you look at it from the CFO’s chair.

Announce that you are cutting 20% because growth slowed and margins are under pressure, and the market treats it as a distress signal. Announce that you are cutting 20% to fund an AI-first transformation, and the same cut becomes a growth story. monday.com lifted its 2026 margin outlook in the same filing that disclosed the layoffs, and said it expects to keep hiring in strategic areas through the year. The restructuring charge runs $45–55 million, split between severance and office-space impairments.

AI is currently the only available narrative that converts a layoff into evidence of ambition. That creates something genuinely new: a financial incentive to displace workers that exists whether or not the technology works yet. The capability is not the trigger. The story about the capability is the trigger, and stories move at the speed of an earnings call.

This is shareholder primacy doing exactly what the book says it does. The firm is not optimizing for output. It is optimizing to be legible to capital, and right now the format capital wants to read is “AI-native.” Six hundred and twenty people are the cost of formatting.

“Isn’t this just layoffs with a new label?”

Partly. Companies have always dressed up cuts in whatever language the era rewarded, from “synergies” to “rightsizing” to “focus.” A skeptic is entitled to ask what makes this different from the usual euphemism cycle.

Two things do.

The first is timing. Ordinary restructuring is countercyclical. Revenue falls, so you cut. Anticipatory displacement is the reverse: monday.com is growing, raised its margin guidance, and is still cutting a fifth of its people. The cut is not a response to a bad quarter, it is an investment in a future capability. That inverts the entire logic labor forecasters rely on, because it means displacement is no longer downstream of business conditions or of what machines can do.

The second is that the money does not evaporate. It moves. The payroll that funded 620 salaries becomes budget for compute, models, and the smaller team meant to wield them. Anticipatory displacement finances the very automation it was announced in advance of, which is the bootstrap paradox running at the level of the income statement. The prophecy is not merely self-fulfilling. It is self-funding.

Which means the honest version of the skeptic’s question is worse than the question. Yes, the label came first. The label is also how the thing gets built.

Why this wrecks the forecast

Almost every serious model of AI and employment, including the timeline work in this project, treats displacement as a function of capability: identify the tasks, estimate when a machine performs them at acceptable quality and cost, and derive the labor effect. The substitution threshold is the hinge, and the potential timeline is built by walking it forward.

Anticipatory displacement is not a function of capability. It is a function of belief about capability, and belief has no upper bound imposed by whether the thing works. A benchmark score tells you when automation can happen. It tells you nothing about when a board decides to reorganize around the expectation that it will.

The practical consequence is that capability forecasts systematically underestimate near-term labor damage while remaining perfectly good at describing the long run. The gap between the two is not an error to be corrected. It is a real period, with real people in it, and we are in it now.

There is also a nastier asymmetry here. Genuine automation displacement at least arrives holding the productivity gain that could pay for the transition: the work still gets done, more cheaply, and the surplus exists somewhere to be captured or shared. Anticipatory displacement front-loads the human cost and back-loads the productivity. You get the unemployment first and the abundance later, if the bet pays off at all. Some of these bets will not pay off, and those firms will have destroyed the jobs anyway.

The Unscarcity Read

The book models the Transition as the dangerous interval between the old economy failing and the new one working, and argues that everything depends on getting a floor underneath people before the gap opens. Anticipatory displacement changes the schedule for that argument in a way Chapter 8 readers should take seriously.

The Labor Cliff arrives in two waves, and the first one is not automation. It is capital reallocating under an AI narrative, running years ahead of the machines. That means the social cliff opens before the technical one, and any plan calibrated to the capability curve arrives late by exactly the size of the gap. The Foundation has to be built on belief-time, not capability-time. Waiting to see whether AI can really do the jobs before establishing Universal High Income means waiting through the entire period in which people are losing work over a forecast.

It also sharpens the case in Chapter 9. A firm cutting a fifth of its workforce to look right to investors, while telling those same workers the decision had nothing to do with replacing them, is not a villain in this story. It is a rational actor inside an ownership structure that rewards narrative compliance over production. Mission Guilds exist because that structure cannot be argued out of its incentives, only exited. And Impact exists because a system that measures contribution directly is much harder to game with a press release.

The 620 people at monday.com were not replaced by artificial intelligence. They were replaced by a plan to acquire some. Whether that turns out to be better or worse than being automated depends entirely on whether anyone builds the floor before the second wave lands.

That is the argument of Unscarcity: the floor is not something we install after the disruption is confirmed. By then it is a relief operation. Built early, it is the thing that makes the transition survivable, and it has to be pointed at the wave that is already here rather than the one the benchmarks say is coming.

Sources

Share this article:

Share Share Share