75% Of New Code At Google Is AI-Generated: Sundar Pichai Reveals New Secrets Of The Company—Know How AI Is Changing The Tech Industry

Artificial Intelligence is no longer just a tool that helps you write emails or answer questions. It is now writing code at one of the biggest technology companies in the world. Google CEO Sundar Pichai has revealed that three out of every four lines of new code at Google are now being written by AI, and human engineers are simply reviewing and approving it. This is a massive moment for the entire technology industry, and people are paying very close attention.

The Big Number

Sundar Pichai shared the update at Google Cloud Next 2026, revealing that 75 percent of all new code at Google is now AI-generated and approved by engineers, up from 50 percent just last fall. To put that in perspective, when Pichai first mentioned this publicly back in October 2024, the number was just over 25 percent. By early 2025, it had grown to around 30 percent. The jump to 75 percent in such a short time is quite remarkable and shows just how fast things are changing inside Google.

What Is Actually Happening

Before anyone starts worrying about robots taking over, it is important to understand what this really means. Pichai has been clear that AI-generated code refers to code that is suggested by AI and then accepted or edited by humans. No code goes directly to production without being reviewed. The number describes who wrote the first draft, not who is accountable for it.

Engineers are no longer writing every line themselves but are instead focusing more on reviewing, guiding, and improving the outputs that AI produces. Think of it like having a very fast assistant who does the rough work while the experienced person checks and polishes it.

Agentic Workflows: The Next Big Thing

Pichai said Google is now shifting to truly agentic workflows, where engineers are orchestrating fully autonomous digital task forces, firing off AI agents to accomplish complex tasks together.

The results have been striking. A particularly complex code migration done by agents and engineers working together was completed six times faster than was possible a year ago with engineers alone. That kind of speed improvement is genuinely hard to ignore.

Another example was the Gemini app on macOS, where a team used Google’s internal agentic development platform called Antigravity to go from an idea to a working native app prototype in just a few days. Something that would have taken weeks before now takes days.

What Tool Are Engineers Using?

Engineers at Google primarily use the company’s own Gemini models for code generation. Google has also integrated AI usage into its performance evaluation system, with some engineers receiving specific AI-related goals that will be factored into their performance reviews for the current year. This shows that using AI tools at work is no longer optional at Google. It is becoming part of the job description.

What Does This Mean for Software Engineers?

This is the question everyone wants answered. The demand is shifting from people who can write code to people who can direct AI to write correct code and verify it. That means fundamentals like data structures, algorithms, and systems knowledge matter more, not less, because you need them to evaluate what AI produces.

The real risk for workers is not being replaced by AI directly. It is being replaced by someone who has learned to work alongside it effectively.

The Bigger Picture

Google is not alone in this shift. Sam Altman of OpenAI has also noted how quickly software development is evolving, with AI systems now performing tasks in seconds.

The message from Silicon Valley is clear. AI is not coming for your job tomorrow. But it is already changing what that job looks like, and the engineers who adapt the fastest will have the strongest edge going forward.

Also Read: Elon Musk Launches XChat: Ad-Free Encrypted Messaging App To Compete With WhatsApp—Check Eligible Devices