Ajay Pundhir, senior AI leader & founder of AskAjay.ai. Believes AI should amplify human expertise, not replace it.
getty
On May 14, Cisco told roughly 4,000 people their jobs were gone. The same day, it reported record quarterly revenue of $15.8 billion, up 12%. Six days later, Meta began cutting about 8,000 jobs—nearly one in ten employees—weeks after posting Q1 net income of $26.8 billion, up 61%. The same afternoon, Intuit cut 17% of its workforce, about 3,000 people, having just reported quarterly net income of $3.06 billion.
None of these companies is in trouble. All of them are cutting. And almost all of them are saying the same word: AI.
The comfortable reading is the efficiency story. The machine got good enough; the work it replaced went away; the headcount followed. It is tidy because it implies the savings are real and banked—the robot does the job now, so the salary is pure margin.
That story is mostly wrong. Not because AI can't do the work—sometimes it can—but because it is not what most of these cuts are paying for.
Listen to the executives when they are precise rather than promotional.
Cisco's CEO, Chuck Robbins, did not describe an efficiency gain. He described a transfer. "Given the speed at which the market is moving, we need to make a rapid reallocation of resources," he said, and went out of his way to call it "cost reallocation" rather than "cost reduction." His CFO, Mark Patterson, told analysts the restructuring was "not a savings-driven" exercise. In the company's own words: not banking savings, freeing capital to go elsewhere.
Where? Cisco raised its expected fiscal-2026 AI orders to $9 billion from $5 billion. Meta's finance chief, Susan Li, told analysts a "leaner operating model" would help "offset the substantial investments we are making"—investments that, on the capex line, run to as much as $145 billion in 2026 alone. The cut and the bet are named in the same breath.
That is what the efficiency story hides: the money saved by removing a profitable worker is not pocketed. It is fed into an AI build-out whose returns have yet to arrive.
Here is a test a board can run in one meeting. I call it the Backfill Test.
An efficiency layoff removes work that is already automated. The output still ships, the machine demonstrably does the job and you are cutting costs on something you no longer need. Capacity is automated before the headcount goes.
A reallocation layoff removes the worker but keeps the work, and bets AI will backfill it later. The output does not yet ship from a machine. You have removed the person and you are hoping the system arrives in time to cover what they did. Capacity is promised, not proven.
The two look identical on announcement day. They are opposite bets. Plot it on two axes: Is the AI in production doing this work today? Did you cut the people who did it? Three quadrants are ordinary management. The fourth—cut the people, AI not yet in production—is the Profitable Layoff: sold as efficiency, behaving like a wager.
And the wager is worse than the efficiency story admits, because its structure is upside down. The cost is front-loaded and certain: severance paid, institutional knowledge out the door, capacity gone this quarter. The return is back-loaded and speculative, depending on AI reliably reaching production at scale and doing work it is not yet doing. HCLTech's 2026 survey of senior AI leaders found nearly 43% of major AI initiatives are expected to fail—for the unglamorous reasons projects fail: unclear success criteria, brittle data access and no one accountable. You are spending a sure thing to buy a coin flip.
Not every company is saying the same thing.
Intuit's CEO, Sasan Goodarzi, rejected the AI framing outright. "None of it had to do with AI," he said. "Everything was about how do we become more effective"—flattening management, removing coordination-heavy roles after folding Credit Karma and TurboTax together. If true, Intuit's cut is not a Profitable Layoff. It is plain restructuring wearing an AI-shaped headline—the headline that sells in 2026.
That is the point, not the exception. The label "AI layoff" covers at least two distinct things—genuine restructuring and a capital reallocation bet—and tells you almost nothing about which one you face. The Backfill Test does.
There is a strong counterargument: reallocating capital from mature, lower-growth functions toward a strategic bet is not a mistake. It is the job. Cutting good people to fund the next thing is how companies avoid becoming the last thing.
True. Reallocation can be exactly the right call—which is precisely why it must be named as what it is. A bet gets governed like a bet: staged funding, falsifiable milestones, a point at which you admit the AI did not arrive and retrench. An "efficiency saving" is treated as a fact—no scrutiny at all, because everyone believes the money is in the bank. The danger is not the reallocation. It is calling the reallocation a saving, and then never checking whether the bet was paid.
Before you sign off on the next round, ask the one question that separates the two:
Are you cutting because the AI works—or to pay for AI that might?
If it is the first, show the production system doing the job. If it is the second, fine—but say so, fund it like the wager it is, and put a date on the board calendar to find out whether you were right. The companies cutting today will know in eighteen months. The ones that called it a saving will have spent the evidence before they ever collected it.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。