Mark Zuckerberg Says AI Will Write Most of Meta’s Code in the Next 18 Months

In a groundbreaking statement during a podcast interview with Dwarkesh Patel, Meta CEO Mark Zuckerberg predicted that, within the next 12 to 18 months, artificial intelligence will take over much of the coding work at Meta, particularly for the company's ambitious Llama project. Zuckerberg’s bold vision marks a significant shift in how software development could unfold in the coming years, with AI poised to surpass even the top human engineers.

According to Zuckerberg, AI is already performing at the level of a “very good person” on a software development team. However, the real leap will come soon when AI starts doing more than just helping with auto-completion of code. “In the next 12 to 18 months, most of the code will be written by AI. I don’t mean like autocomplete. I mean AI will be able to run tests, find issues, and write high-quality code independently,” Zuckerberg stated confidently. His vision is for AI to handle everything from bug fixing to writing clean, optimized code without much human intervention.

AI and Meta's Llama Project: A Powerful Collaboration

Meta’s Llama project, which has already garnered attention for its focus on AI research and development, will be one of the first beneficiaries of this shift. Zuckerberg revealed that AI is "fully plugged into our tool chain," which means it is integrated into Meta’s internal workflows and is already contributing to the coding and research efforts surrounding Llama. The company has developed a specialized AI research agent and coding agent to specifically aid in Llama’s development.

"We are building coding agents for ourselves," Zuckerberg explained. "We are not trying to build a general developer tool, but a coding agent that specifically advances Llama research." By targeting specific projects like Llama, Meta is ensuring that the AI agents are highly specialized and purpose-driven, designed to meet the company's unique needs and accelerate development in ways traditional coding teams cannot.

Zuckerberg’s Vision of AI Engineers Replacing Human Coders

Mark Zuckerberg’s bold statements about the future of AI in software development echo similar predictions from other leaders in the tech industry. Earlier this year, Zuckerberg shared his belief that all code in Meta’s apps and AI systems will eventually be written by AI engineers rather than human developers. "AI is already good enough to replace a mid-level software developer," he added.

This idea of AI-powered engineering is gaining traction across the tech world. In fact, Dario Amodei, CEO of Anthropic, recently echoed Zuckerberg’s sentiments. In an interview in March 2025, Amodei predicted that in the next three to six months, AI will be responsible for generating 90 percent of code, and by the end of 2025, AI will write all code across the industry.

The Growing Role of AI in the Tech Industry

Zuckerberg is not alone in this forecast. Google CEO Sundar Pichai recently revealed that 25 percent of the code at Google is already being written by AI, while OpenAI CEO Sam Altman shared that in some companies, AI is already generating 50 percent of the code. This increasing reliance on AI to handle coding tasks represents a fundamental shift in the way software is created and developed across the tech industry.

As AI continues to evolve, its capacity to automate complex tasks traditionally handled by skilled engineers will only grow. Zuckerberg’s estimate that AI will soon surpass human developers is not as far-fetched as it may seem, given the rapid advancements in AI research and the widespread adoption of AI tools within tech companies.

The Implications of AI in Coding: What’s Next?

Zuckerberg’s vision for AI-powered software development not only impacts Meta but sets the stage for a broader transformation within the tech industry. If AI can handle the majority of coding tasks, it will free up human developers to focus on higher-level design, strategy, and innovation. It could also lead to faster development cycles, more efficient software, and even new breakthroughs in AI research itself.

However, this shift raises important questions about the future of the software development workforce. Will there still be a place for human coders, or will they become obsolete in the face of increasingly capable AI? How will the role of engineers evolve as AI takes over more tasks traditionally associated with their work?

These are questions that the tech world will likely grapple with in the years to come. But one thing is certain: AI is already playing a major role in shaping the future of coding, and companies like Meta are at the forefront of this transformation.