Meta Platforms Inc. Chief Executive Officer Mark Zuckerberg has initiated a bold restructuring of the company’s artificial intelligence (AI) ambitions with the formation of a specialized unit dubbed “Superintelligence Labs” aimed at achieving artificial general intelligence (AGI) and surpassing rivals like OpenAI and Alphabet Inc.’s Google.
According to people familiar with the development, Zuckerberg has taken personal charge of the recruitment and strategic formation of the elite group, which is internally tasked with building AI models capable of human-level cognitive performance across multiple domains.
The initiative follows internal dissatisfaction with the performance of Meta’s latest large language model (LLM), LLaMA 4, which has failed to meet internal benchmarks and external expectations.
Sources revealed that the Facebook founder has entered what insiders call “founder mode,” adopting a highly hands-on management approach.
Zuckerberg is spearheading efforts to hire approximately 50 experts for the new AGI-focused division, most of whom he is recruiting directly. The team will operate in close proximity to his office at Meta’s Menlo Park headquarters.
Zuckerberg’s AGI pivot is backed by a planned multibillion-dollar investment in Scale AI, a leading data infrastructure provider for AI applications.
Bloomberg News previously reported that the deal could exceed $10 billion, making it Meta’s largest external investment to date.
The founder of Scale AI, Alexandr Wang, is reportedly expected to join the Superintelligence Labs once the transaction is finalized.
The move comes amid mounting pressure from within Meta’s leadership to enhance the competitiveness of its AI capabilities.
Despite repeated internal promises of delivering the most advanced AI models in the market, Meta’s rollout of LLaMA 4 has been met with criticism over underperformance and a lack of meaningful advancement over prior iterations. The release of its most ambitious model, codenamed “Behemoth,” has been postponed.
Meta’s new superintelligence unit is expected to play a critical role in refining the company’s AI roadmap, with a focus on voice capabilities, personalization tools, and advanced applications integrated across Meta’s suite of platforms including WhatsApp, Instagram and its AI-powered Ray-Ban smart glasses.
The formation of Superintelligence Labs marks a strategic escalation in Meta’s broader AGI race against OpenAI and Google DeepMind. Meta has already committed tens of billions of dollars in capital expenditures toward AI infrastructure in 2025 with Zuckerberg projecting long-term investments into the hundreds of billions to stay competitive.
Over recent weeks, the CEO has hosted leading AI researchers and engineers at his residences in Lake Tahoe and Palo Alto, pitching Meta’s vision of AI dominance and emphasizing the company’s internal financing capacity, powered by its robust advertising revenues.
Zuckerberg reportedly argued that unlike competitors dependent on venture funding, Meta can fund multi-gigawatt data centers essential for next-generation AI development without external support.
Internally, a WhatsApp group named “Recruiting Party” has been created to accelerate the identification and onboarding of high-value AI talent. Sources said Zuckerberg personally manages outreach to top candidates and compiles recruitment lists, reinforcing the significance he attaches to the new unit’s success.
While it remains unclear how Superintelligence Labs will align with Meta’s existing AI divisions, internal reshuffling is already underway, with select staff expected to transition to the new group.
As the global tech ecosystem intensifies its race toward AGI, Zuckerberg’s aggressive restructuring positions Meta as a major contender, aiming to not only close the performance gap but to lead the next frontier of machine intelligence.
Leave a Comment