Meta has restructured its AI division into four new groups under the name Meta Superintelligence Labs, according to an internal memo reported by several outlets.

Alexandr Wang, who joined Meta as chief AI officer in June, will oversee a new group called TBD Labs, focusing on foundation models such as the Llama series.

The other three groups will handle research, product integration, and infrastructure.

Meta has increased efforts to reorganize its AI teams amid competition from companies like OpenAI, Anthropic, and Google DeepMind.

CEO Mark Zuckerberg has been personally involved in recruitment for the new group.

🔗 Source: TechCrunch

🧠 Food for thought

1️⃣ Frequent reorganizations signal deeper operational challenges in AI development

Meta’s fourth AI restructuring in just six months reveals unusual organizational instability that goes beyond typical corporate adjustments.

This rapid succession of changes, from the original AI division to multiple reorganizations and now “Meta Superintelligence Labs,” suggests the company is struggling to find an effective structure for AI development 12.

Most successful tech organizations typically maintain organizational stability for 12-18 months to allow teams to execute on strategic initiatives, making Meta’s pattern particularly noteworthy.

The repeated restructuring may also indicate internal challenges with talent retention and operational efficiency, as confirmed by reports that approximately half of the modeling teams face elimination or reassignment 2.

This level of organizational churn can disrupt project continuity and make it difficult for teams to build momentum on complex AI initiatives that typically require sustained focus over extended periods.

2️⃣ Tech giants are converging on centralized AI organizational models

Meta’s consolidation mirrors a broader industry trend toward unified AI command structures, as Google recently made similar moves by folding more teams into DeepMind.

In January 2025, Google consolidated its AI Studio and Gemini development teams under DeepMind “to accelerate the research-to-developer pipeline,” showing that even established AI leaders see value in centralization 3.

Recent Meta developments