AI Efficiency Gains Just Got a Dev Lead Ordered to Cut 20% of His Team
This is the real cost of AI efficiency that Western dev teams are starting to feel too: when productivity goes up and business doesn't, the next step is role consolidation and headcount cuts. It also puts tech leads in an impossible position, forced to measure their own team out of jobs.
A tech lead at a Chinese company spent six months rolling out an AI coding workflow, moving from local AI-assisted coding to cloud-based autonomous agents and multi-agent teams. His boss kept pushing him to quantify how much AI improved productivity, but he could never give a clean number, because too high or too low was equally dangerous. After the rollout, his reports said they were more exhausted than before and carried the same blame, while being pushed to keep creating new projects. "AI doesn't stop, so I can't stop either," he described.
The boss's math is blunt: one programmer costs nearly 200,000 RMB a year. If AI proves efficiency gains, the company can consolidate roles and cut staff. The tech lead now has orders to lay off 20% of his team. He feels the resentment daily, and the conversation with the author of this post ends with a grim conclusion: when the business pie isn't growing, AI efficiency gains lead to layoffs, not expansion.
AI efficiency gains are not neutral: when they land in a zero-growth business environment, they become a direct argument for headcount reduction, not for doing more.
A tech lead is caught between the boss's demand for measurable productivity gains and the team's resentment of being pushed harder, a structural conflict that no amount of tooling solves.
The push to measure AI's impact creates perverse incentives, because a low number makes the investment look useless and a high number makes people redundant.
The phrase 'AI doesn't stop, so I can't stop either' captures how autonomy in AI tools shifts pressure onto humans, who must keep pace with machines rather than being freed by them.
The article's premise draws both dismissal and dark recognition. Some view the story as incoherent or flat-out false, while others treat it as a faithful picture of enterprise IT departments. A direct equation emerges: more AI adoption means more layoffs. The 'leek' metaphor frames workers as a crop to be harvested, and a sarcastic push to fire the boss first redirects cuts toward management.
What on earth is he babbling about?
He's saying we're all leeks.
He never said anything about leeks.
Client enterprise IT departments are exactly in this situation.
Yes, the higher the AI adoption, the more layoffs.
Lay off the boss first.
Bro, that takes guts.