Meta Restructures AI Teams Amid Talent Loss and Copyright Challenges, Stock Rises
May 27, 2025
Meta is reorganizing its artificial intelligence (AI) teams to accelerate the development and deployment of new products and features, reflecting a strategic shift in response to increasing competition in the AI sector.
This restructuring will not involve any job cuts, as Chief Product Officer Chris Cox emphasized the goal of promoting ownership within teams while clarifying dependencies.
This restructuring comes amid notable talent loss, with many original creators of the Llama models leaving for competitors, including the French startup Mistral.
Despite these challenges, Meta's stock has seen a positive reaction, recently increasing by 2.45%, indicating investor confidence in the company's strategic direction.
The restructuring aims to streamline operations and improve efficiency within the AI division, allowing for better alignment of talent and resources.
Meta continues to advance its AI ambitions, having recently previewed its Llama API and launched smaller versions of its Llama 4 models, despite facing delays.
To support its AI initiatives, Meta plans to deploy approximately 1.3 million GPUs by the end of 2025, underscoring its commitment to enhancing AI capabilities.
The reorganization follows the departure of Joelle Pineau, a prominent executive in AI research, marking a significant loss for Meta's leadership.
This reorganization mirrors previous efforts, such as the division of Reality Labs into Wearables and Metaverse divisions, aimed at showcasing successful products to investors.
The new structure is modeled after successful organizational strategies used by competitors like Google and OpenAI, separating product engineering from fundamental research.
Cox's memo highlights the importance of creating smaller, agile teams that can innovate rapidly while ensuring that deep-research scientists focus on complex AI challenges.
Legally, Meta is navigating a significant copyright case regarding allegations that its Llama models were trained on pirated works, adding pressure to its ongoing AI development efforts.
Summary based on 23 sources
Get a daily email with more Tech stories
Sources

TechCrunch • May 27, 2025
Meta reportedly splits its AI team to build products faster
DEV Community • May 28, 2025
Meta Splits AI Division to Accelerate Product Development and AGI Research
PYMNTS.com • May 28, 2025
Meta Reorganizes GenAI Team to Accelerate Product Rollouts