Zuckerberg Is Coding Again — Using Claude Code CLI
Zuckerberg Is Coding Again — Using Claude Code CLI
Mark Zuckerberg has returned to writing code after a two-decade hiatus, submitting diffs to Meta’s monorepo — and he’s doing it with Claude Code CLI, Anthropic’s terminal-based AI coding agent.
Last verified: April 4, 2026
What Happened
According to sources reported by Gergely Orosz (The Pragmatic Engineer):
- Zuckerberg submitted three diffs to Meta’s internal monorepo
- He’s described as a “heavy user” of Claude Code CLI
- This is his first hands-on coding in roughly 20 years
- Y Combinator president Garry Tan is also coding with AI tools
Why It Matters
This isn’t just a tech CEO doing a PR stunt. It signals something fundamental about how AI coding tools have changed software development:
- The barrier dropped — AI coding agents let anyone with technical understanding write production code, even if they haven’t coded in years
- Claude Code is dominant — Even the CEO of Meta, which builds Llama, reaches for Anthropic’s tool
- The “CEO-as-coder” trend — When executives can directly prototype and ship, it changes company dynamics
The Irony
Zuckerberg runs Meta, which builds the Llama family of AI models and has its own internal coding tools. The fact that he chose Claude Code — built by competitor Anthropic — speaks volumes about the current state of AI coding tools.
Claude Code hits ~80.8% on SWE-bench Verified with Opus 4.6, making it the top-performing autonomous coding agent. It processes up to 1M tokens of context, meaning it can understand large codebases holistically.
What Claude Code Offers CEOs (and Everyone)
- Natural language coding — Describe what you want, Claude Code writes it
- Codebase understanding — Reads your entire project, understands architecture
- Multi-file editing — Makes coordinated changes across many files
- Terminal-native — No IDE needed, just a terminal
- Git-aware — Understands branches, commits, PRs
The Bigger Trend
The Pragmatic Engineer notes this is part of a broader movement of “industry leaders returning to coding.” AI tools have made it possible for people who understand systems but haven’t written code in years to be productive again.
This has implications for:
- Hiring — Technical founders can do more with smaller teams
- Decision-making — Executives who code understand technical tradeoffs firsthand
- Speed — Skip the spec-to-implementation pipeline entirely
Last verified: April 4, 2026