A new industry survey finds that AI coding agents now contribute the majority of code in production environments at major technology companies, marking a shift from experimental adoption to mainstream reliance.

The survey, released Monday by software analytics firm SlashData, polled 4,200 professional developers across 12 countries. Sixty-one percent reported that AI-generated code now accounts for more than half of their production codebase, up from 28 percent a year ago. The findings suggest that tools like GitHub Copilot, Cursor, and Claude Code have moved well beyond autocomplete into genuine autonomous contribution.

Engineering leaders at several large companies confirmed the trend. A senior vice president at a Fortune 500 financial services firm said her teams now use AI agents for routine feature implementation, bug fixes, and test generation, freeing human engineers to focus on architecture and code review. "The ratio has flipped," she said. "We review more than we write."

The shift has not been without friction. Developers reported that AI-generated code often requires significant refactoring, particularly around edge cases, error handling, and security-sensitive logic. Nearly 40 percent of respondents said they had encountered at least one production incident traceable to AI-written code in the past six months.

Despite those concerns, adoption continues to accelerate. Venture capital investment in AI coding startups reached $8.4 billion in the first half of 2026, more than triple the amount from the same period last year. The market is betting that the productivity gains outweigh the risks, at least for now.

Image source: i.ibb.co