AI to Write Majority of Code at Meta by 2026, Says Zuckerberg
Meta CEO Mark Zuckerberg has made a bold prediction about the future of software development: artificial intelligence will soon outperform even top-level engineers in writing code. Speaking on a podcast hosted by Dwarkesh Patel, Zuckerberg said that within the next 12 to 18 months, most of the code driving Meta’s Llama AI projects will be written by AI agents—not humans.
Zuckerberg emphasized that AI is no longer just a tool for autocomplete but is evolving into a full-fledged software engineer capable of understanding goals, running tests, finding bugs, and producing high-quality, production-ready code.
“I don’t mean autocomplete,” Zuckerberg said. “If you give it a goal, it can run tests, it can find issues. It already writes higher-quality code than an average very good person on the team.”
The Meta chief further explained that the company is currently building specialized coding and research agents to work specifically on the Llama project. These AIs are being designed for internal use, fully integrated with Meta’s development systems—not as general-purpose developer tools.
“We’re not building a general developer tool. We’re creating a coding agent and a research agent aimed at advancing Llama research. It’s deeply integrated into our toolchain,” he said.
Zuckerberg’s view aligns with a growing consensus in the tech world. Anthropic CEO Dario Amodei recently stated that AI could be responsible for generating 90% of code within six months, and possibly all of it by year-end. Google CEO Sundar Pichai revealed that 25% of Google’s code is now AI-generated, while OpenAI CEO Sam Altman noted that AI already writes half of the code in some companies.
Earlier this year, Zuckerberg also commented that Meta’s future apps and the AI systems within them would be almost entirely developed by AI engineers.
As AI’s capabilities continue to grow rapidly, the role of human software developers may soon evolve from hands-on coding to overseeing, guiding, and refining AI-generated output, marking a historic shift in how the tech industry builds its products.
