Tencent says AI now writes most of its code this year

At a conference on June 5, Tencent’s senior executive vice president Tang Daosheng dropped a number that signals a quiet revolution inside the Chinese tech giant: this year, artificial intelligence now writes the majority of the company’s code. The admission, delivered without fanfare, marks a turning point not just for Tencent but for the broader engineering culture that has long defined China’s internet sector. Tang’s disclosure is less a boast about efficiency than a confession of strategic transformation.

For years, Tencent’s sprawling product empire—from WeChat to Honor of Kings to its cloud services—relied on armies of developers churning out lines of code. That model is now being upended. Engineers are no longer primarily coders; they are becoming supervisors and architects of AI-generated software. The shift is already reshaping Tencent’s hiring and training pipelines.

Junior developers who once spent months learning syntax and debugging are now being asked to master prompt engineering and system-level oversight. The company is effectively telling its workforce: write less, think more. Those who cannot adapt face obsolescence. What casual observers might miss is the implication for code quality and security. AI-generated code, while fast, introduces new risks: subtle bugs, hidden dependencies, and the potential for adversarial manipulation.

Tencent’s engineers are now tasked not just with reviewing logic but with auditing the AI’s decision-making process—a skill set that barely existed two years ago. This is not a pilot program or a limited experiment. Tang’s remarks suggest the transition is already baked into Tencent’s production workflows. The company’s massive user base—over a billion across its ecosystem—means that AI-generated code is now running in real-time on everything from messaging to payments to gaming.

The broader industry is watching closely. Tencent’s move could accelerate a trend already visible at global peers like Google and Microsoft, where AI-assisted coding has moved from novelty to necessity. But Tencent’s scale and its deep integration of AI into consumer products make its experience a bellwether for the region. For the engineers themselves, the change is existential. The romanticized image of the lone coder building features from scratch is fading.

In its place is a more collaborative, more abstract role: managing AI agents, defining architectural constraints, and ensuring that the machine’s output aligns with business goals and regulatory demands. Tencent is not merely adopting a tool; it is redefining what it means to be a software engineer. The next few quarters will reveal whether this bet on AI-generated code pays off in speed and innovation—or introduces vulnerabilities that only human intuition can catch.