Artificial intelligence has quietly crossed a remarkable threshold inside one of the world’s most important technology companies. Google revealed on Wednesday that three-quarters of all new code being created at the company is now generated by AI — reviewed by human engineers, but written by machines.
The figure represents a breathtaking acceleration. As recently as October 2024, roughly a quarter of Google’s code was AI-generated. By last autumn, that had climbed to 50%. Now, just months later, it has reached 75% — and there is little indication the trend is slowing down.
The Shift to Agentic Workflows
Google CEO Sundar Pichai framed Wednesday’s announcement as part of a broader transformation in how the company’s engineers work. In a blog post, he described a shift toward “truly agentic workflows” — a model in which AI systems operate with greater autonomy, handling complex tasks that previously required sustained human effort.
The real-world impact is already being felt. Pichai highlighted one particularly striking example: a complex code migration that, when completed by agents and engineers working in tandem, was finished six times faster than the same task would have taken engineers working alone just a year ago.
“Six times faster” is not a marginal efficiency gain. It is a fundamental reimagining of what software engineering looks like — and of how much human effort is required to build and maintain large-scale technology systems.
Gemini Powers the Change — and So Does Claude
Google’s engineers are using the company’s own Gemini AI models for the bulk of their code generation work. The company has been actively encouraging adoption, with some engineers now assigned specific AI-usage goals that will factor into their performance reviews this year — a clear signal that AI-assisted coding is no longer optional but expected.
In an interesting twist, some employees at Google DeepMind have been permitted to use Anthropic‘s Claude Code in recent months — a development that has reportedly created some internal tension, as it involves Google engineers using a competitor’s AI tools within one of the world’s most closely watched AI companies.
Google Is Not Alone — This Is an Industry-Wide Shift
What is happening at Google is not an isolated phenomenon. Across the technology industry, the proportion of AI-generated code is rising rapidly — and the ambitions being set are extraordinary.
Microsoft CEO Satya Nadella revealed last April that between 20 and 30 percent of the code for some of the company’s projects was already AI-written. Microsoft’s CTO, Kevin Scott, went further, predicting that 95 percent of all code would be AI-generated within five years — a statement that seemed bold when made and looks increasingly prescient now.
Meta has set aggressive internal targets of its own. By the first half of 2026, 65 percent of engineers in its creation organisation are expected to write more than 75 percent of their committed code using AI assistance. Snap has also announced that at least 65 percent of its new code is now AI-generated under its revised operating model.
The message from across Silicon Valley is consistent and unmistakable: AI-generated code is no longer a novelty or an experiment. It is rapidly becoming the default.
What This Means for the Future of Software Engineering
The implications of this shift are profound — and not entirely straightforward. On one hand, the productivity gains are real and significant. Tasks that took weeks can now be completed in days. Complex migrations that required armies of engineers can be handled by a fraction of the team. Companies can move faster, ship more, and scale their software capabilities without proportionally increasing headcount.
On the other hand, questions are multiplying about the long-term role of human engineers in this new landscape. If 75 percent of code is already AI-generated today, what does that number look like in two years? In five? And what skills will the engineers reviewing and directing that AI-generated code need to possess?
For now, the human element remains central — the review, the judgement, the creative direction. But the balance is shifting, and shifting quickly.
Google’s announcement on Wednesday is not just a milestone for one company. It is a signal about where the entire industry — and the nature of software development itself — is headed.
📩 Stay ahead of the biggest technology stories. Subscribe to the JournalTodays Newsletter for daily updates straight to your inbox.
