AI Job Cuts Reached 120,136 This Year. Your Rivals Have the Same AI.

AI job cuts reached 120,136 in the first nine months of 2026, about 21% of every job cut U.S. employers announced this year, and that share keeps climbing even as the overall number shrinks. Challenger, Gray & Christmas counted 573,195 announced cuts through September, down 39% from 946,426 over the same stretch of 2025, a decline steep enough to sound reassuring until you notice which portion of it keeps growing. Its August report had already named AI the leading stated reason for cuts for the fifth consecutive month, so September's figures arrived as confirmation of a pattern rather than as a surprise.

September looked calmer on its surface. Employers announced 43,281 cuts, fewer than August, with AI accounting for 3,961 of them, a smaller monthly share that still fits comfortably inside the larger trend. Holiday hiring told a quieter story underneath: Spirit Halloween and Michaels together announced 62,000 seasonal hires this September, compared with 100,800 announced by seasonal employers across the same stores a year earlier, the kind of restraint that shows up before a company explains itself at all.

Figures from the Challenger, Gray & Christmas September 2026 job-cuts report.

Andy Challenger, the firm's chief revenue officer, described the hiring side plainly:

"Hiring plans are up over the year, but we're not seeing the surge of hiring plans that come with the holiday season, which suggests a very cautious approach." Andy Challenger, Chief Revenue Officer, Challenger, Gray & Christmas

Companies have moved past asking whether AI is real

Cautious hiring and steady AI-attributed cuts point toward the same conclusion from two different directions. Executives have largely stopped debating whether the technology works and moved on to a harder question: how many people the business will actually need once the tools are fully in place. The monthly counts are where that question starts showing its answer.

Each of those 120,136 cuts is a kitchen-table conversation before it is ever a data point. Somewhere behind the number is a person who spent years learning which customer complaints mean real trouble and which ones blow over on their own, now explaining to their family that the company decided a piece of software could cover that judgment instead.

The direction has held all year. I keep a running view of the AI job loss statistics shaping 2026, and September fits the pattern: fewer layoffs overall, with a larger share of them pinned on AI.

Every competitor is buying the same AI

The headcount math leaves out one variable that matters more than any single quarter's numbers. When every company in a market rents its intelligence from the same small group of model providers, that intelligence raises everyone's floor at the same moment, rather than lifting the company that adopts it first above everyone else. Stanford's annual AI Index has tracked the distance between the leading models and their fast followers narrowing year after year, and whatever advantage your current tools give you, your competitors are closing that distance faster than most leadership teams assume.

Subtract the identical AI that every rival can buy, and what remains is what each company actually owns. Part of it is judgment built up over years of doing the work. Part of it is the people who can tell when the model's answer is wrong, and who know where your processes break before the dashboard shows it.

Those people are the hardest advantage any competitor could try to copy. A rival can match your software license in an afternoon with a purchase order and a signature, but matching a team that understands your customers the way yours does takes years it doesn't have, and the team walks out the door the day you cut it.

Everyone will have the same AI. Not everyone will have the same people.

Companies cutting their experienced staff to pay for AI are spending the asset that is slowest to rebuild to buy the one thing every rival will shortly have too.

Circuit City made a version of this trade in 2007

In March 2007, Circuit City fired 3,400 of its highest-paid, most experienced salespeople to cut costs, a decision the company's own leadership framed at the time as simple arithmetic. By November 2008, it was in Chapter 11. Many forces sank Circuit City beyond that single decision, and the layoff alone didn't cause the bankruptcy, but it certainly didn't save the company either.

The salespeople kept the judgment that had made them expensive in the first place, and they carried it with them across the street, where Best Buy, the competitor that survived, hired a meaningful number of the people Circuit City had just let go.

The 2026 version of this trade is larger and harder to see. When 120,136 jobs are cut with AI named as the reason, a great deal of hard-won knowledge changes hands, and some of it will end up training whichever rival hires it. Cutting takes a quarter. Rebuilding that kind of judgment takes a decade.

The relevance cliff applies to companies too

Most coverage of AI job cuts focuses on workers facing the relevance cliff, a central idea in the book I am writing now, The Great Relearning. Companies can walk off one too. The next AI transition will arrive in a few years, and a company that let go of everyone able to judge this one will be buying the next model with no one in-house to evaluate it.

When Meta cut 8,000 jobs in May, I wrote that the order of operations decides whether cuts like these make sense: redesign the work first, then decide who the redesigned work actually needs, rather than announcing the headcount number before the process underneath it has changed at all. The very next day, Samsung's workers raised the question that follows naturally from that one, when they turned down a one-time bonus and asked who owns the value AI creates instead of simply taking the payout and dropping the subject. September's numbers add a third question to that pile, and it belongs on the CEO's desk as much as it belongs with HR.

These private decisions add up. They shape which careers last and how fast a town recovers after a plant or office shrinks. State workforce agencies and community colleges do the hard work of retraining people after a layoff. Employers who bring those partners in early, while the experienced people are still on payroll, give that work a real head start.

You can adopt AI at full speed and still keep the people who can judge it. Your competitors are working through the same choice, with the same uncertainty you feel.

Here is the question I'm taking into my own meetings this quarter, and you're welcome to borrow it: if every competitor had our exact AI tomorrow, which of our people would we need to win anyway? If the answer includes anyone on this year's cut list, the list deserves a second look before it is final.

And if you're the one wondering about your own role, the same question works for you. What do you know about your work that a model can't check on its own? That knowledge is worth naming, and worth teaching to the people around you. I'm easy to find if you want to talk it through.

Here is what makes Alex a credible voice on this topic: he shaped Cisco's $1.1 billion innovation portfolio over 20 years and now advises leaders on which people their AI plans can't afford to cut.

Talk with Alex about your AI workforce plan →