Most new computer code at Google is now created by artificial intelligence, according to the company’s chief executive.
Speaking at the Google Cloud Next 2026 conference in Las Vegas, Sundar Pichai revealed that nearly 75 per cent of code is AI generated and approved by engineers.
This figure is up from 25 per cent in 2024, and 50 per cent in 2025.
“We’ve been using AI to generate code internally at Google for a while,” he said.
“We are now shifting to truly agentic workflows. Our engineers are orchestrating fully autonomous digital task forces, firing off agents, and accomplishing incredible things.”
The Google boss gave an example of how AI agents were able to complete a complex code migration task six-times faster than human engineers.
It follows a broader trend among tech companies to integrate AI into their workflows on a massive scale.
A report this week revealed that Meta is introducing tracking software to capture employees’ mouse movements, clicks and keystrokes in order to train new AI systems to do their work.
An internal memo from the Facebook and Instagram owner described a new set of tools called the Model Capability Initiative (MCI) that aim to replicate how humans interact with computers.
“The vision we are building towards is one where our agents primarily do the work and our role is to direct, review and help them improve,” Meta CTO Andrew Bosworth told employees in a separate memo reported by Reuters.
Meta is among several major tech firms that are also encouraging employees to use as much AI as possible in their daily work.
Internal leaderboards at Meta and ChatGPT creator OpenAI reportedly rank employees based on their monthly token consumption – referring to the amount of AI processing power they use.
This so-called “tokenmaxxing” trend has faced criticism for the financial and environmental cost of such heavy AI useage, as well its actual value as a performance metric.
“Ranking engineers by token spend is like me ranking my marketing team by who spent the most money,” Christina Cordova, chief operating officer at Linear, wrote on X. “Don’t mistake a high burn rate for a high success rate.”
Meta to surveil staff to teach its AI to work, report says
Apple’s new CEO is the safe choice. That might be the biggest risk of all
Yelp adds a new AI feature to help users sift through local recommendations
Samsung faces mass strike threat as workers demand share of AI profits
Robot beats elite table tennis players at their own game
Microsoft being sued for £1.7 billion over ‘abusing position’ in cloud computing
