A striking revelation that underscores the accelerating shift in software development, Microsoft CEO Satya Nadella disclosed that artificial intelligence is now responsible for generating as much as 30% of the code within the company’s internal repositories. Speaking at Meta’s inaugural LlamaCon AI developer summit in Menlo Park, California, Nadella emphasized that this figure is steadily rising — a clear signal that generative AI is becoming deeply embedded in Microsoft’s engineering workflows.
Nadella made this statement during a candid conversation with Meta CEO Mark Zuckerberg, where the two tech giants discussed the growing role of AI in shaping their companies’ futures. “I’d say maybe 20%, 30% of the code that is inside of our repos today and some of our projects are probably all written by software,” Nadella noted before the live audience, hinting at a not-so-distant future where machines shoulder the bulk of code production.
Zuckerberg, while not quoting exact figures for Meta, echoed the sentiment. He revealed that Meta is currently developing AI systems capable of designing and evolving future iterations of its Llama models. “Our bet is sort of that in the next year probably … maybe half the development is going to be done by AI, as opposed to people,” Zuckerberg said, outlining a future where AI becomes the primary architect of digital infrastructure.
These insights are not isolated. They reflect a wider movement sweeping through the tech industry. Since the launch of ChatGPT in 2022, companies have increasingly turned to AI not just for customer interaction or content generation, but for core engineering functions. Google CEO Sundar Pichai recently said that over 25% of the company’s new code is now generated by AI tools. Shopify CEO Tobi Lutke went a step further, stating that employees must now demonstrate a task cannot be done by AI before requesting additional manpower. Meanwhile, Duolingo CEO Luis von Ahn announced a transition toward AI in place of some human contractors.
The implications go beyond operational efficiency. The dream now is software written faster, with fewer bugs, and better adaptability — a scenario that AI-powered development promises to bring closer to reality. Startups like Windsurf, reportedly in acquisition talks with OpenAI, are pushing the boundaries by offering “vibe coding” software that can generate entire applications from just a few lines of human input.
As Nadella and Zuckerberg continue to lead organizations that both create and adopt frontier AI models, their insights offer more than just a glimpse into internal operations — they signal a profound redefinition of how software itself will be imagined, designed, and deployed in the years to come.
