“AI Can Replace 80% of Human Work,” How Many Companies That Believe This Regret Their Layoffs?

Companies are betting on a future that hasn't arrived

It has been reported that an internal chat message—“Fire fast!”—was widely forwarded across Chinese social platforms, supposedly from a major internet firm urging rapid cuts. NetEase (网易) was recently accused of sweeping out game contractors and temporary staff; the company denied using AI to replace all outsourced workers but reportedly said it is “phasing out” some basic outsourced roles as tasks are automated. The headline is simple: across China and the West, firms from Bilibili (哔哩哔哩) and iFlytek (科大讯飞) to Meta and Block are citing AI as the rationale for fresh rounds of layoffs. Why now? Because investors reward the appearance of future-ready efficiency — and because executives are racing to signal AI readiness in a heated global tech race.

The data: many cuts were driven by expectation, not capability

Research suggests most of these moves are anticipatory. A survey discussed in Harvard Business Review by Thomas Davenport and Rakesh Srinivasan found 89% of 1,006 executives had cut staff or slowed hiring due to AI’s expected impact — but just 2% said the layoffs were because AI had already taken over the work. An Oxford Economics study flagged “profit-driven” layoffs: packaging cuts as an AI-led efficiency play often boosts stock prices. Visier data across 142 firms and 2.4 million employees shows only 5.3% of laid-off workers were later rehired with generous raises, while Careerminds reported that over 30% of companies that made AI-related cuts later rehired 25–50% of those roles. In short: many firms are cutting first and validating later.

Real-world reversals and hidden costs

The reversals are already visible. Block’s large layoffs prompted a partial rehiring and public admissions of error by CEO Jack Dorsey; Salesforce has publicly acknowledged overestimating AI’s ability to handle complex client work and is rehiring some of the consultants and engineers it previously let go. The reason is not technical nuance alone — it is organizational memory. Customer relationships, crisis response, and tacit knowledge accumulated over years are hard to encode in models. When systems fail or clients demand nuance, people still step in. And there’s a morale cost: employees at firms like Meta have described a “suffocating” atmosphere where weekly layoff lists create chronic insecurity.

Bigger picture: incentive misalignment in the AI era

So what are we watching? Not merely automation replacing tasks, but a market-driven theater where AI is used as cover for rapid restructuring. This dynamic matters in China as well as the West, especially amid geopolitics — US–China tech tensions, export controls on advanced chips, and competition for AI talent mean firms face both technical constraints and investor pressure to appear competitive. Will companies learn to calibrate when automation truly lowers costs versus when it simply signals futurism to markets? For workers, the lesson is stark: adoption of AI tools may improve productivity, but history shows that preemptive layoffs often need to be undone — and rebuilding the tacit capabilities lost in the meantime is far costlier than boardroom slide decks implied.