AI-generated code at Microsoft now contributes to 20–30% of the company’s internal software development, CEO Satya Nadella revealed during a fireside chat with Meta CEO Mark Zuckerberg at the LlamaCon conference on Tuesday. The estimate reflects Microsoft’s increasing reliance on AI tools like GitHub Copilot and large language models (LLMs) to streamline coding workflows.

Progress in Python, Challenges in C++

When Zuckerberg asked how much of Microsoft’s code is currently written by AI, Nadella responded that the results vary by programming language. Python shows the most promise for AI-assisted development, while C++ still poses significant challenges due to its complexity and performance constraints.

This isn’t the first time a Microsoft executive has spoken about the rise of AI in coding. Chief Technology Officer Kevin Scott previously predicted that 95% of all code would be AI-generated by 2030, a bold projection that underscores the transformative impact of generative AI on the software engineering landscape.

Meta and Google Also Relying Heavily on AI Code Generation

When Nadella flipped the question back to Zuckerberg, the Meta CEO admitted he wasn’t sure how much AI contributes to Meta’s codebase. Still, the conversation highlighted how major tech companies are actively embedding AI into their development pipelines.

Meanwhile, Google CEO Sundar Pichai offered his own estimate during last week’s earnings call, saying more than 30% of Google’s code is now generated by AI. However, the methodologies behind these figures are not fully disclosed, and experts caution that definitions of “AI-generated code” can vary widely.

What It Means for the Future of Programming

The rise of AI-generated code at Microsoft and other tech giants is rapidly reshaping the developer experience. Engineers now frequently act as code reviewers or prompt engineers rather than sole authors of software. AI tools assist with everything from boilerplate generation to complex problem-solving, accelerating productivity while raising new questions about quality, maintainability, and security.

With generative AI models becoming increasingly capable, the software industry may be approaching a tipping point where human oversight is still essential, but large portions of code are machine-authored by default.

Get the Latest AI News on AI Content Minds Blog