Google: 75% of Code Generated by AI, a Major Turning Point

Google and AI: A Silent Revolution in Software Development

Within Google, artificial intelligence is no longer just serving external users. It is now deeply embedded in the internal workings of the company. Developers, once the sole masters of their domain, now share control with automated systems. This transition, promising for some, also raises questions about the future of software development.

At the Google Cloud Next 2026 event in Las Vegas, Google made a resounding announcement. The company revealed that 75% of the new code produced internally is generated by AI. This impressive figure reignites the debate about the role of humans in software development. While engineers continue to play a crucial role, the balance between human and machine seems to be shifting.

An Expanding Human-Machine Collaboration

Under the leadership of Sundar Pichai, Google has adopted so-called agentic workflows, meaning systems capable of executing certain tasks autonomously. This approach has significantly improved productivity. For example, a complex code migration was completed six times faster than a year ago, thanks to close collaboration between engineers and AI.

Currently, 75% of the new code at Google is generated by AI, with final approval from engineers. This figure has significantly increased from last fall, when it was at 50%, and even more from the previous year, when it was only 25%. Engineers are now delegating repetitive or complex tasks to AI, focusing more on oversight and technical decisions, which explains the rapid adoption of these technologies.

However, this transition is not without tensions. Some employees at Google DeepMind have access to Claude Code, a tool developed by Anthropic, which creates internal friction.

AI: An Indispensable Partner but Not a Replacement

Although AI has accelerated code production at Google, the company emphasizes that every line of generated code is reviewed by engineers. This step is crucial to minimize errors and maintain control over the development process.

Google primarily relies on its Gemini models to generate this code. These tools are gradually becoming indispensable partners in software development. Developers must now integrate the use of AI into their daily routines, and this skill has even become a criterion for internal evaluation.

Google's announcement sends a strong message: AI is no longer just an assistant; it is now a central player in software production. The implications are significant. Development costs and timelines could be reduced, but the dependence on AI models is increasing.

This evolution raises questions about the future of developers. While it is difficult to predict with certainty what the future holds, it is clear that the profession is evolving towards a role of oversight and management, rather than mere code entry.