Microsoft cut about 4,800 jobs last week. It is profitable and spending heavily on AI. Cisco announced cuts of nearly 5% of its workforce in May, in the same breath as record quarterly revenue. Cloudflare eliminated more than 20% of its staff while growing more than 30%, which chief executive Matthew Prince, writing in the Wall Street Journal, said he believed no other US public company had matched. His prediction: "what we did is likely going to become the norm over the next year."
The obvious explanation is AI. It is also the one the companies keep declining to give. Microsoft said its latest cuts were not AI-related. Amazon, which has shed 57,000 corporate roles since 2022, about 16% of that workforce, has said AI was not the reason for the vast majority of them. Meta pointed to a statement noting the changes varied by team and included moving thousands of workers to other priorities.
Meanwhile, on earnings calls, mentions of layoffs alongside AI have gone from fewer than five a quarter in 2022, when ChatGPT launched, to more than 100 a quarter this year, according to an AlphaSense analysis. Both things are true at once: executives are talking about AI and job cuts in the same sentence more than ever, and when asked directly whether AI caused a specific round, they say no.
That contradiction is worth sitting with, because it points at what a layoff announcement actually is. It is a message to shareholders. "We are restructuring around AI" is a story about discipline and the future. "We over-hired during the pandemic and demand softened" is a story about management error. Cisco's Chuck Robbins framed his cuts as the mark of firms with the discipline to "continuously shift investment" toward long-term potential. The same cut, described two ways, produces two different stock reactions.
What is genuinely new is the rhythm. Business Insider's reporting describes cuts becoming a recurring ritual rather than a recession event: firms are making waves of reductions while they figure out what AI does to their business, rather than waiting for certainty. For employees, that turns layoffs from a downturn risk into a standing feature of the job. The human cost is not abstract. One former Amazon engineer of ten years described his identity as tied to the job; he sent more than 250 applications and got four responses, all rejections. Postings now routinely draw 200 to 300 applicants.
So is AI destroying jobs or not? The honest answer is that the evidence is thinner than the rhetoric. Stanford's Digital Economy Lab found entry-level roles most exposed to AI have declined 13% since late 2022, which is suggestive but not conclusive; other factors could explain it. The IMF projects roughly 40% of jobs worldwide are exposed, which is a measure of exposure, not of losses. A May 2026 survey of 12,000 executives found 99% expect AI to bring at least some headcount reduction within two years, which tells you about expectations rather than outcomes. As MoneyWeek notes, the loudest doom forecasts come from the people selling the technology: Dario Amodei on unemployment reaching 10 to 20%, Ford's Jim Farley on half of white-collar jobs, Sam Altman on the one-person billion-dollar company. Predictions of catastrophe from a new technology are ordinary. Predictions of catastrophe from its own makers are not, and it is worth asking what function they serve.
And there is a whole side of the ledger the layoff coverage mostly misses. Between November 2025 and January 2026, Americans filed 1.56 million new business applications, the largest three-month stretch since the Census Bureau's series began in 2004. January alone saw 532,319 filings, up 36.8% year over year. The number of people listing "founder" on LinkedIn is up 69%. Writing in Forbes, Lisa Curtis describes the shape of these firms: one founder who built a company solo in four months with AI coding tools and sold it to Wix for $80 million, another who described her compliance software business, serving nearly 300 customers, as "me, my dog, and about 40 AI agents." Payroll: one person.
That is not a refutation of the layoff story. A displaced Amazon engineer is not consoled by someone else's solo startup, and a company of one person and forty agents is, arithmetically, a company that did not hire thirty-nine people. But it does change what the story is. The economists warning about "large-scale job displacement" and the founders quietly building 40-agent companies are describing the same event from opposite ends: work is not simply vanishing, it is moving, from large payrolls to small ones. Whether that turns out to be a boom or a hollowing depends entirely on how many people can make the crossing, and nobody is measuring that yet.