55% of Leaders Who Cut Jobs and Cited AI Now Say It Was the Wrong Call

The AI layoff era is producing its first regret data, and the numbers are stark. In an Orgvue survey reported in coverage of the trend, 55 percent of leaders who cut jobs citing AI now say it was the wrong call. A Gartner study of 350 firms this spring, cited in the same reporting, found that the companies cutting headcount hardest showed no better returns than those that held steady, and some of the cautious ones outperformed.

The exhibit everyone points to is Klarna, which claimed its AI did the work of 700 agents, froze hiring, and ended up pulling staff from engineering, marketing, and legal onto support calls to rebuild the capacity it cut. But the pattern is broader than one fintech. Coverage of the trend notes that Amazon, Salesforce, Meta, and Cloudflare all cut jobs while citing AI, and the regret rate says the miscalculation was systemic rather than a single CEO's overreach. Markets have rendered their own verdict on the loudest example: Klarna's stock has fallen roughly by half since its 2025 IPO, a decline its support saga did nothing to slow.

The replacement math was wrong from the start

The error was treating AI as a substitution: one agent of software in, one human out. The evidence increasingly supports a different function entirely. A Ramp Economics Lab and Revelio Labs study of 21,559 firms found companies investing most heavily in AI grew headcount 10.2 percent over the two years after adoption, with entry-level hiring up 12 percent, while light adopters stayed flat. The organizations getting the most from AI are not shrinking around it. They are growing through it, with the technology absorbing scale work while people move to work that compounds. The same research exposes a timing trap that explains much of the regret: AI's productivity gains do not appear until six to twelve months after adoption, then build from there. A leader who cuts headcount in the same quarter the software arrives books the disruption immediately and the gains never, because the people who would have converted the tool into productivity are already gone.

The regretful 55 percent ran the opposite play: cut first, then discover which conversations, judgments, and relationships the software could not actually carry. The rebuild costs more than the savings, because it happens under pressure, with institutional knowledge already gone and customers already burned.

Redeployment is the model that survives

What separates the winners in this data is not enthusiasm for AI. Both groups bought it. The difference is what they did with the human hours it freed. The replacement camp booked the hours as savings and sent them out the door. The redeployment camp pointed them at the work machines do worst: the enterprise relationship, the escalation that decides a renewal, the judgment call that closes or saves a seven-figure account.

Gartner's buying research completes the argument from the customer's side: 69 percent of B2B buyers turn to human reps to validate AI-generated insights at decisive moments. The demand for humans did not disappear when the scale work automated. It concentrated, into fewer moments worth more money. Leaders who cut the humans discovered they had also cut the moments. The 55 percent regret rate is the market grading that decision, and the grade is in.