Mark Zuckerberg Is Handpicking A 'Superintelligence' Team To Lead Meta's AGI Race After Llama 4 Letdown: Report

Meta Platforms Inc. META CEO Mark Zuckerberg, unhappy with the company's setbacks in AI innovation, is reportedly handpicking a new team of top researchers and engineers to work on artificial general intelligence.

What Happened: Zuckerberg is personally steering efforts to recruit around 50 experts, a team internally known as the "superintelligence" group, to achieve artificial general intelligence (AGI), the point where machines can match human abilities across tasks, according to a Bloomberg report.

His goal is for Meta to overtake its rivals in this race, with plans to integrate AGI into chatbots, smart glasses and other products. Seating plans at Meta's Menlo Park campus have allegedly been redrawn so the new hires can work close to him.

Zuckerberg's superintelligence push is closely tied to Meta's planned multi-billion-dollar investment in Scale AI. Once finalized, the deal, which will be Meta's biggest external investment yet, is expected to bring Scale AI founder Alexandr Wang into the new AGI team.

Why It Matters: This new hands-on approach to recruiting reportedly originates from disappointment with Llama 4, Meta's latest large language model, which failed to meet his expectations for performance and adoption. These setbacks led Meta to delay its next major release, "Behemoth," amid concerns it featured little improvement over earlier versions.

The Llama models are the key to Meta's AI strategy. By open-sourcing Llama, Meta hopes to establish it as a global AI foundation, like how Android became the backbone of mobile software.

Last month, several engineers working on Llama reportedly left Meta to join Mistral AI, a Paris-based startup.

Price Action: Meta stock is currently trading at $700.02, down -0.34% pre-market.

Benzinga Edge scores the stock high on Growth (92.74), Momentum (86.41), and Quality (86.62), but low on Value (27.02). For more details, check out Benzinga Edge rankings.

Read Next: