On June 12, a Meta employee hijacked a livestreamed internal presentation watched by thousands of colleagues, and used the microphone to call a top AI executive “a piece of sh*t,” per Mediaite’s reporting.
Hours later, Zuckerberg sent an internal memo acknowledging Meta had made “mistakes” in its AI restructuring and promising no further company-wide layoffs in 2026, according to Cryptopolitan’s coverage of the memo.
The same day, TechCrunch reported that engineers inside Meta’s three-month-old Applied AI organization — ~6,500 people, sitting under CTO Andrew Bosworth and led by 12-year Meta veteran Maher Saba — are calling the unit “the gulag.” Their assigned work: generating puzzles and coding problems to train AI agents on how humans complete computer tasks.
“It’s literally the gulag. Most people find the work soul-crushing.”
The Structural Read
The market reads this as a morale story. The actual story is a talent-task mismatch at scale.
In June 2025, Meta paid $14.3 billion for 49% of Scale AI and recruited founder Alexandr Wang to run Meta Superintelligence Labs, per CNBC’s coverage. The implicit pitch to recruits: come build frontier models, alongside the person who runs the world’s largest data-labeling operation.
What 6,500 engineers actually got is the data-labeling operation itself, repackaged in the language of agent training. “Generate puzzles and coding problems so models can learn how humans complete tasks” is a Scale AI workflow. It is operational data generation. It is not frontier research.
That gap — between the comp-and-mission profile of the hire and the day-to-day work — is what “gulag” means. It is not a vibe issue. It is a category error in org design.
The key insight:Meta paid frontier-research compensation for harness-adjacent work. The revolt is the price discovery of that mismatch.
Three Implications
IMPLICATION #1 — TALENT WILL CLEAR THE MARKET
Engineers told “join or quit,” per TechCrunch’s reporting, will quit. Anthropic, OpenAI, xAI, and Mistral are already-formed liquidity venues for senior Meta AI departures. The “no more layoffs” memo doesn’t address voluntary attrition — and voluntary attrition is the actual leak.
IMPLICATION #2 — THE SCALE AI THESIS IS NOW THE WHOLE STRATEGY
If 6,500 engineers are doing puzzle generation, Meta’s de facto AI strategy is “own the data pipeline.” That is defensible — Scale-style data ops are real moat — but it is a different bet than “build superintelligence.” Investors should reprice on the data-ops thesis, not the frontier-lab thesis Zuck originally sold.
“No more layoffs in 2026” is a stability signal aimed at retention. It does not change Applied AI’s work allocation, Saba’s mandate, or Wang’s leverage inside the org. Without restructuring the work itself, the same dynamic will produce a louder revolt in Q3.
Business Engineer Framework
Harness Theory
Companies win in AI by harnessing capability, not by trying to build everything in-house. Meta’s $14.3B Scale AI deal was a harness play — buy the data pipeline, deploy it across Reels/Ads/WhatsApp. But the comp packages and recruiting pitch were a build-the-frontier-lab play. The “gulag” revolt is what happens when those two motions collide inside one org chart.
Meta is not in chaos because of one viral incident. It is in chaos because $14.3 billion of talent is being deployed on operational data work, and the people doing it know the difference. The next move — whether Meta restructures the work or doubles down on the data-ops thesis — will tell you which company Zuckerberg is actually trying to build.
AI is fundamentally transforming Meta’s business model and operational scale under Mark Zuckerberg’s leadership, with massive investments reshaping the company’s trajectory. One concrete example is Meta’s development of its large language model LLaMA (Large Language Model Meta AI), which represents a $10+ billion annual investment in AI infrastructure and research. This initiative has required Meta to build enormous data centers filled with specialized AI chips, hire thousands of AI researchers, and pivot significant resources away from traditional social media features toward generative AI capabilities. The scale is unprecedented – Meta’s AI training clusters now consume enough electricity to power small cities, while LLaMA’s integration across Instagram, Facebook, and WhatsApp affects billions of users daily. This transformation demonstrates how AI is forcing tech giants to operate at previously unimaginable scales of computation and investment.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
