AI Writes All the Code, Promotion Passes Him Over: A Chinese Frontend Dev's 2026 Pivot
When AI generates all the code, a promotion that rewards relationships and business impact rather than technical excellence becomes the new normal. Western developers chasing titles inside large orgs should take note that coding proficiency alone won't carry a career once AI commoditizes it.
After three years with consecutive excellent performance reviews, a frontend engineer asked for a promotion. The review in May came back negative: the list had three backends, three implementation engineers, two product managers, and zero frontends. It took more than a month to accept that promotions aren't a pure technical contest; relationships with leadership, current grade, and company need all weigh in, and code quality is just one variable.
The same year, the side business started with a WeChat account and a traffic mini-program, then expanded to Xiaohongshu, Douyin, Xianyu, WeChat Video, and Juejin, with small paid promotion. The platforms clamped down hard: Xiaohongshu banned him several times and WeChat limited his account. By year-end the total was just over ¥4,000. His own conclusion: writing code is the easy part of the monetization chain; attracting users is the hard part. The experiment pulled him out of a pure programmer mindset and taught him how business models actually run.
AI's share of his code went from 70% last year to essentially all of it this year. With coding no longer a daily drain, he's now reading business books, doing hands-on business work, and treating business understanding as a moat. The company stack is Nest + Vue3 with a Java backend, and he plans to accumulate Java experience while working on Agent plugin features for automatic modeling. Long-term, he's aiming for full-stack and Agent development roles.
The empty frontend slot on the promotion list arrives in the same year AI takes over all code writing; that pairing suggests frontend engineering is being devalued faster than other disciplines as AI tooling improves.
Earning ¥4,000 after repeated platform bans teaches a sharper lesson than a larger payout: distribution and user acquisition, not software quality, are what determine whether a side project makes money.
The shift from infrastructure obsession to business reading and hands-on practice matches the economics of AI commoditizing code: a developer's unique value gravitates to the commercial and product domain.
The comments break along two lines: whether business know-how and side-hustle income beat a stable paycheck, and which language stack belongs in enterprise work. Business understanding is called the hard part, while one reply counters that personal monetization is even harder than earning a salary. A separate view insists enterprise development still depends on Java, dismissing NestJS as fit only for personal projects, with a reply noting that Java will need to be learned regardless.
Understanding the business is so hard, how do you generate income?
Personal monetization is even harder; taking a salary is more comfortable.
You're giving up just like that?
Better to use Java. Enterprise development is definitely Java. Using NestJS is unrealistic; it's fine for your own side projects.
Our company has this layer. It's not my own project; I'll need to learn Java later anyway.