Google says 75 percent of its new code is now written by AI

The 75 percent figure is striking, but the more consequential detail is what it implies about the role of human engineers: they are increasingly reviewers and validators rather than primary authors. That is a structural shift in how engineering headcount justifies itself, not just a productivity story.

This lands directly on top of the coding AI arms race we covered in mid-April, when OpenAI upgraded Codex with agentic capabilities as a direct shot at Anthropic's Claude Code (covered April 16 from both The Verge and TechCrunch). Those stories framed the competition as a race for developer adoption. Google's disclosure reframes the stakes: the real prize is not individual developer preference but enterprise-scale workflow capture, where a single customer like Google represents an enormous volume of generated code. The App Store boom story from April 18 adds a related data point — AI is lowering barriers to software creation broadly, not just inside large engineering orgs. Together, these signals suggest the coding AI market is bifurcating into consumer-facing tools and deep enterprise integration, with very different competitive dynamics in each lane.

Watch whether Microsoft discloses a comparable internal adoption figure for GitHub Copilot within the next two earnings cycles. If they do, it confirms enterprise-scale AI code generation is becoming a standard metric that large tech companies will report; if they stay silent, Google's disclosure may be a deliberate competitive signal rather than routine transparency.