There is a particular note in the May 2026 layoff press releases that an MIT labour economist heard before, and finally said out loud. On Sunday, Fortune ran a piece in which Paul Osterman, professor emeritus at MIT Sloan and author of Disposable Workers: The Transformation of Employment, called the AI-as-explanation framing what it has now plainly become: a perfectly serviceable cover story for decisions companies were going to make anyway.
“They’ve been saying that for 20 years,” Osterman told Fortune. “AI is a perfect excuse to justify big layoffs. It makes it seem as if it’s not our decision, our fault — it’s the technology.”
The quintet that prompted the interview is the May 2026 lineup. Wix’s Avishai Abrahami announced 1,000+ cuts on May 28 and pitched the move as the leaner, flatter, AI-era version of his company. Block’s Jack Dorsey opened the year with 4,000 cuts — 40% of staff — and the same “smaller and flatter” vocabulary. Snap CEO Evan Spiegel had already cited AI when announcing his own April round. Atlassian did the same with 1,600. Cisco eliminated 4,000 jobs on May 13 — the same day it reported a record $15.8 billion quarter — and watched its stock pop 13% in after-hours trading.
That is the part Osterman is pointing at. A record-revenue print, on the same day, with the same cuts.
The 20-year pattern, restated for the AI era
Osterman’s argument is not that AI does nothing. It is that the language CEOs are using — “leaner,” “flatter,” “smaller and more productive teams,” “adapting to a new way of working” — is the exact dialect that has been wheeled out, in his telling, since the dot-com correction, the post-2008 squeeze, the 2014 globalisation wave, and every subsequent restructuring cycle. The story changes. The framing does not. What is new in 2026 is that, for the first time, the cover story is also a positive earnings catalyst. The market rewards the cut precisely because the AI label is attached.
Cisco’s 13% stock pop on May 13 is the test case. The fundamentals — record revenue, AI hyperscaler orders revised up from $5B to $9B for FY26 — were already strong. The layoff did not unlock the revenue, but it did unlock the framing. A company posting record revenue and a 4,000-person cut is one story. A company posting record revenue and a 4,000-person “AI-era realignment” is a different story. Investors paid for the second one.
Once that pricing mechanism is established, every CFO with a quarter to defend has the same option in their kit. Take the cut you were going to take. Add the AI vocabulary. Brief analysts on the architecture, not the headcount. Watch the multiple.
The disposable-workforce argument running underneath
Osterman’s deeper claim — the one that gives the Disposable Workers book its title — is that the labour market has been quietly reorganising around contingent employment for two decades, and the AI wave is accelerating it rather than starting it. The conservative BLS count from 2023 put contingent workers (contractors, freelancers, temps) at 6.9 million, or 4.3% of the US workforce, up from 3.8% in 2017. Osterman’s broader estimate, which includes gig and adjacent categories, puts the real figure closer to 35% of working Americans.
Those workers are cheaper, take no benefits, and can be shed on a quarter’s notice. They also, in Osterman’s research, report lower wages, lower satisfaction, and lower discretionary effort than full-time employees. The economic case for hiring them rises with technological uncertainty, because the option value of being able to downsize at low friction goes up. AI uncertainty therefore acts as a contingent-hiring accelerant whether or not the AI itself replaces any specific role.
This reframes the May 2026 cuts in a way the press releases do not. The quintet are not, in Osterman’s reading, primarily a story about software replacing labour. They are a story about the standard employment relationship continuing its 30-year retreat, with AI as the most recent and most marketable explanation.
Why now, why this language
Three forces converge to make “AI-era restructuring” the May 2026 register of choice.
The first is investor incentive. Cisco proved on May 13 that the framing carries a premium. After May 13, no CFO with a layoff to announce had a reason not to attach the AI label, because the alternative framing now costs basis points.
The second is regulatory cover. California’s WARN Act reform memo from May 21 — Governor Newsom’s executive order on AI-driven displacement reporting — created a future compliance line item that companies want to be in front of. Naming AI as the driver, voluntarily, ahead of the disclosure mandate, is cheaper than being named by the disclosure mandate. The EU AI Act’s employment-impact provisions, already in force, push the same direction across European subsidiaries. Amdocs’ Hortig used the “AI-era” language explicitly on May 28; the geography was Israel but the regulatory frame was the same one.
The third is narrative ownership. Until last week, the AI-jobs apocalypse narrative was owned by Sam Altman and Dario Amodei, who both walked back their forecasts on May 26 ahead of their respective trillion-dollar IPO road shows. When the founders stop pushing the apocalypse, the CFOs inherit the vocabulary at a discount. That is the inversion Osterman is pointing at. The two people who built the doom narrative no longer need it. The hundred companies who can use it for cover do.
What the Osterman read does not explain
The piece is not a clean exoneration. Two things complicate it.
First, internal-usage numbers from the companies that are cutting do not look like cover stories. Cloudflare disclosed last week that internal AI usage grew 600% in three months alongside its 1,100-person cut. Wix’s prior pivot to its Base44 AI-first stack is documented in product-roadmap terms, not just in the layoff memo. These are companies whose operating mechanics have plausibly changed, and where the AI label, while convenient, is not pure marketing.
Second, the contingent-worker share has been ticking up structurally; what the May quintet adds is full-time employees moving into the contingent column under AI-tagged cover. The shift from full-time to disposable employment that Osterman has been tracking for 20 years now has a Q2 2026 vehicle that did not previously exist: voluntary AI-era reorgs that move whole functions to vendor contracts or freelancer pools while the corporate headcount line drops, the cost line drops, and the stock pops.
That is the thing the BLS will eventually pick up in 2027 data and Osterman is already writing about now. The AI label is doing the framing work. The contingent-worker structure underneath is doing the actual labour-market work. Both can be true in the same press release.
What to watch in June and Q2 earnings
The next two weeks are the test of the Cisco pricing mechanism. Salesforce reports on June 4; its “AI-rebalancing” line has been part of investor calls for three quarters now and the May 2026 data is the first time investors are demanding specific numbers behind it. Atlassian’s late-July guide should include either an updated AI-era headcount target or a meaningful walk-back. Block’s H1 print, expected in August, has to either show the 40% Dorsey took out actually generated the promised margin, or it does not.
The second test is the regulatory one. California’s first WARN-AI reports under the May 21 EO are due in early September. Companies that voluntarily AI-tagged their May cuts will get to point at the filing. Companies that did not — the ones, mostly, who quietly cut and called it “rightsizing” — will not. The asymmetry will shape which corporate vocabulary survives into Q4.
The third is the academic one. Disposable Workers has been on Osterman’s shelf since 2023. The Fortune piece is the first time the framing has been read into a tech-layoff cycle in real time, by name, in a magazine CFOs read. That puts the term “AI washing” into the lexicon. The next CEO who recycles the Cisco language word-for-word will encounter it in the analyst Q&A.
The thing Osterman gave the room on May 31 is not a defence of any specific worker, and not a denial that AI is changing labour demand. It is a labour economist’s seasoned reminder that when companies tell you the technology made the decision, the company is — every time, for 20 years — the one that actually made the decision. The technology gets the headline. The headcount line is where the action is. The stock pop is the receipt.