Meta Reorganizes AI Division: A New Structure to Balance Product Speed and AGI Research

In the spring of 2025, Meta (formerly Facebook) implemented a bold organizational restructuring in the field of artificial intelligence (AI). According to reports from Axios, Meta has split its previously unified AI division into two teams: one specialized in consumer product development and the other in foundational research. This move is seen as a strategic step for Meta to maintain its unique presence while balancing technology and speed, amidst intense competition from rivals such as OpenAI, Google, and Anthropic.

In this article, we will decode the significance of this reorganization in the AI industry by examining the background and objectives of Meta's AI restructuring, the roles of each new team, and a strategic comparison with competitors.

1. The Full Picture of the Reorganization: Restructuring into Two Teams

Meta has reorganized its AI division into the following two teams:

1-1. AI Products Team: Accelerating Consumer Experience

This team is responsible for the integration and improvement of AI features across existing massive platforms such as Facebook, Instagram, and WhatsApp. This team also leads the development of standalone AI apps that Meta has been focusing on in recent years. For example, this area includes features that directly impact daily user experience, such as image generation, AI assistants, and AI-driven content suggestions.

It is worth noting that Meta undertook this reorganization with an emphasis on "product speed." The intention to compete with rapidly growing apps like TikTok by having the flexibility to quickly deploy features for Gen Z is clearly visible.

1-2. AGI Foundations Team: Foundational Research and Llama Development

The other team is in charge of more fundamental technological innovation. Its mission is the improvement of the "Llama" series of large language models, which Meta has been focusing on, and the long-term development of Artificial General Intelligence (AGI).

AGI is an ambitious field aiming to realize "AI with human-level intelligence," and currently, OpenAI's GPT-4 and Anthropic's Claude series hold the lead. How Meta's AGI division evolves Llama and addresses ethical and technical challenges will be the key to the future.

2. Competition with Other Companies and the "Llama for Startups" Strategy

Meta is moving not only in technical competition but also in forming an ecosystem. At "LlamaCon," held in April 2025, the support program for startups, "Llama for Startups," was announced.

Through this program, Meta provides API access, technical support, and joint marketing to help early-stage companies easily develop products using Meta's generative AI models. It can be called Meta's version of a "lock-in strategy" to counter competition from OpenAI's "ChatGPT Enterprise" and Google's Vertex AI.

Meta CTO Andrew Bosworth stated the following at LlamaCon:

3. Why Split Now? — Three Underlying Factors

3-1. The Complexity of Technology and the Need for Specialization

Due to the evolution of AI technology, it is becoming inefficient for a single team to handle both foundational research and product deployment. Since research and implementation move at different speeds, an optimal structure for each is required.

3-2. The "Time Race" with Rivals

With OpenAI and Anthropic releasing new models one after another, and Google showing its presence with the Gemini series, Meta needs to increase its development speed and deployment frequency to avoid being relegated to a "chasing position."

3-3. Message to Investors

There are no job cuts associated with this reorganization. By emphasizing the "clarification and focus of functions," Meta has signaled to investors its commitment to balancing long-term research with short-term profitability.

4. Future Outlook: The Intersection of AGI and Products

Meta's reorganization is not merely a "division of labor." It is built on the premise that technology developed by the AGI team will be delivered to the public through the AI Products team. This is also a mechanism to streamline the loop of "research → application → market feedback → further research."

If this structure succeeds, Meta could potentially gain an advantage in the next-generation market as the "only AI company that possesses both product development capabilities and fundamental research strength."

Meta's AI division reorganization is not an organizational separation, but rather a division for acceleration. In the fiercely competitive AI field, such structural transformation is essential to balance "product-driven responsiveness" with "AGI-oriented depth."

"Does the future of AGI lie with Meta?" The answer is not yet clear, but at the very least, the stage is being set. We look forward to seeing how Llama evolves and how it reaches our hands.

Related Articles