When Mark Zuckerberg declared 2025 the year of “AI-native” at Meta, most people assumed it was a talking point. Internal documents obtained by Reuters and other outlets reveal it was a deadline.
Meta has set specific, quantified AI coding targets for its engineering teams — and the numbers are staggering. Some teams are expected to write over 75% of their code using AI tools like Metamate and Google Gemini. The Scalable Machine Learning team aims for 50–80% AI-assisted code. A company-wide goal for central products calls for 55% of software engineers’ code changes to be “agent-assisted.”
And if you think these are aspirational targets, Meta’s head of people Janelle Gale made the stakes clear in an internal memo: “AI-driven impact” is now a core expectation for employee performance evaluations starting 2026.
The Quotas, By Team
The mandates aren’t uniform — they’re targeted by function:
- Creation organisation: 65% of engineers writing over 75% AI-assisted code by mid-2026
- Scalable Machine Learning: 50–80% AI-assisted code by February 2026
- Central products: 55% of code changes agent-assisted by Q4 2025
- Company-wide: 80% of mid-to-senior engineers adopting AI tools (DevMate, Metamate, Gemini)
Beyond raw code percentages, Meta also introduced an “AI Performance Assistant” — an internal tool that helps engineers draft self-reviews highlighting their AI achievements. A company spokesperson told Business Insider the programme rewards impact from AI tools, not mere usage. But when your performance review asks you to demonstrate AI-driven value, the distinction feels academic.
Reality Labs: AI Pods and New Job Titles
The structural changes at Reality Labs are equally dramatic. A 1,000-person division has been reorganised into small “AI pods” — flat, cross-disciplinary teams that eliminate traditional hierarchy.
Every employee in the restructured division gets one of three new titles:
- AI Builder — hands-on execution
- AI Pod Lead — team coordination
- AI Org Lead — strategic oversight
Meta’s CTO Andrew Bosworth is personally overseeing the “AI for Work” initiative. This isn’t a pilot programme or an experiment. It’s a top-down restructuring of how engineering work happens at one of the world’s largest tech companies.
700 Gone. 15,000 More Coming.
The AI mandates arrive alongside brutal headcount reduction. Approximately 700 Meta employees have been cut across Reality Labs, recruitment, and other divisions. Reuters reports that over 15,000 more layoffs are expected this quarter as part of the company’s aggressive AI restructuring.
Meta claims the layoffs are unrelated to the restructuring. The timing tells a different story.
Zuckerberg himself has said that by 2026, AI could replace tasks that previously required large teams — potentially needing only one “very talented” person. When the CEO publicly suggests that AI makes most of your team redundant, the performance mandates and the layoffs aren’t separate stories. They’re the same one.
Reality Labs lost approximately 1,000 roles as funds redirect toward a $100 billion AI data centre build through 2028. The message is unambiguous: infrastructure gets funded. People get laid off.
The Gamified AI Quota App
Perhaps the most revealing detail: Meta has deployed a gamified internal app that tracks AI quota compliance across teams. Engineers can see how their AI-assisted code percentages compare to peers. Leaderboards. Progress bars. Targets.
It’s the kind of tool that makes AI adoption feel less like professional development and more like surveillance. When your employer gamifies how much of your job you’ve outsourced to AI, the question isn’t whether you’ll hit the target — it’s what happens to you after you do.
What This Means for Every Engineer
Meta is the canary in the coal mine, not the exception. The company’s approach — quotas, performance tie-ins, structural reorganisation, and simultaneous layoffs — is a blueprint that other big tech firms will study and replicate.
If you’re a software engineer anywhere in the industry, the signals are clear:
- AI proficiency is becoming non-negotiable. Not “nice to have” — job requirement.
- Your performance will be measured by AI leverage, not just output. Companies want to see you amplify your work with AI tools.
- Headcount is being traded for AI infrastructure. The math is brutal: one engineer with AI tools can replace multiple engineers without them.
- Cross-disciplinary skills matter more. The “AI Builder” model collapses traditional role boundaries.
Meta’s internal slogan for 2026 might as well be: adapt to AI, or be replaced by it. The tragedy is that even those who adapt may find the goalposts moving faster than they can run.
SOURCES
- Reuters (exclusive: Meta planning sweeping layoffs)
- Economic Times TechMonk (AI coding targets)
- Business Insider (performance evaluation programme)
- Republic World (Zuckerberg AI targets post-layoffs)
