Meta's AI Strategy Evolves Despite 600-Person Cut: What the 'Slim & Impact' Structure Means
In October 2025, as part of a reorganization of its AI division, Meta (formerly Facebook) announced a reduction of approximately 600 personnel. The affected individuals were researchers and engineers belonging to the company's "Superintelligence Lab," and the move is intended to review the AI development structure. While ostensibly for "efficiency," the backdrop includes the intensifying generative AI competition and the difficulty of management decision-making.
1. What is the "Superintelligence" project being promoted by Meta?
Meta's Superintelligence Lab was established in 2023 as a next-generation AI model development hub to compete with rivals such as OpenAI and Anthropic. Chief AI Officer Alexandr Wang stated the following in an internal memo:
"By making the team smaller, the conversations required for decision-making are reduced, and each person's influence and responsibility increase."
This statement symbolizes the "Year of Efficiency" policy that Meta has upheld since 2023. It is directly linked to the philosophy "Leaner is better" previously expressed by CEO Mark Zuckerberg.
2. The talent war in the intensifying AI industry
Axios reported that this summer, Meta poached over 50 researchers with high compensation from OpenAI, Anthropic, Google DeepMind, and others. However, OpenAI CEO Sam Altman stated, "None of our top talent has moved," highlighting the ferocity of the talent acquisition competition in the AI industry.
Until now, Meta has treated generative AI and its metaverse strategy as a dual-track approach, but in line with the rapid growth of the AI sector, it is strengthening its focus on foundation models and inference technology. Meanwhile, coordination costs due to bloated team sizes have been increasing, and this reorganization is seen as aiming for
faster decision-making through organizational slimming.
3. A redeployment strategy, not just layoffs
According to TechCrunch, a Meta spokesperson explained, "The Axios report is accurate, and many affected employees are scheduled to be redeployed to other departments within the company." In other words, this measure is a
reorganization of the AI organization, focusing more on skill matching and structural redesign than on overall personnel reduction.
In recent years, Meta has announced the "Llama" series in the generative AI field and is promoting an open-source strategy. This reorganization is considered to be intended for personnel redeployment to priority areas such as strengthening Llama, AI agent development, and AI integration into the advertising business.
4. A future on the extension of the "Year of Efficiency"
Meta's "Year of Efficiency" was a slogan that symbolized the 10,000-person layoffs in 2023. However, now in 2025, the company's "efficiency" has evolved from mere labor cost reduction to optimization of the AI development process and improvement of decision-making speed.
Behind this is a "small, elite" approach to AI development. Since Anthropic and OpenAI are conducting AGI research with relatively small teams, it appears Meta is exploring a similar structure.
"By each person becoming more load-bearing, the entire organization becomes stronger."
This remark by Wang indicates that Meta is shifting from "distributed intelligence in a massive organization" to "maximizing individual capability."
Conclusion
The 600-person reduction in Meta's AI division is not just a layoff, but a structural reform in preparation for the next-generation AI competition. The company continues to expand its investment in the AI field, and this move is merely
part of that redeployment and centralization.
As OpenAI, Anthropic, Google, and others engage in an "AI arms race," Meta's true value depends not on "scale," but on "speed" and "direction." "Lean but smart"—that is Meta's next strategic keyword.