In a bold escalation of the artificial intelligence arms race, Meta CEO Mark Zuckerberg is reportedly taking personal charge of building a team dedicated to developing “superintelligence”—an advanced form of AI designed to surpass human capabilities.

According to a Bloomberg investigation corroborated by The New York Times, Zuckerberg is meeting AI researchers in private at his residences in Palo Alto and Lake Tahoe as part of a renewed push to accelerate Meta’s presence in the rapidly evolving AI ecosystem.

Zuckerberg’s Frustration Sparks Hands-On Leadership

The report reveals that Zuckerberg’s dissatisfaction with the progress of Meta’s Llama 4 model—the company’s large language model intended to rival OpenAI’s ChatGPT—has pushed him to take unprecedented steps.

Zuckerberg has begun personally recruiting approximately 50 AI experts, reorganizing Meta’s Menlo Park headquarters to keep the new AI team within his physical proximity, and is said to be holding high-level strategy sessions himself. This rare level of CEO involvement underscores the urgency Meta feels in keeping pace with OpenAI, Google DeepMind, and Anthropic.

Meta has not issued a comment regarding the report.

Meta’s Big Bet on Superintelligence

The concept of “superintelligence” goes beyond the buzz surrounding today’s generative AI tools. It refers to a hypothetical AI system that not only matches but exceeds human cognitive abilities across all domains—reasoning, creativity, problem-solving, and more.

Before achieving superintelligence, most experts agree that AI would need to reach Artificial General Intelligence (AGI)—a level of machine learning at which an AI system can perform any intellectual task a human can. While some researchers believe AGI could arrive within years, others argue it remains an elusive, undefined goal with no current pathway.

Despite the ambiguity, Zuckerberg’s ambitions are clear. And with Meta’s massive advertising business providing funding, the effort could soon draw in billions.

Strategic Partnerships and High-Stakes Rivalries

Notably, Zuckerberg has reportedly tapped Alexandr Wang, CEO of Scale AI, to play a key role in the initiative. Meta is also weighing a multi-billion-dollar investment in Wang’s company, according to the report.

The superintelligence project places Meta in direct competition with:

  • OpenAI, backed by Microsoft

  • Alphabet’s Google DeepMind

  • Elon Musk’s xAI

  • Anthropic, a rising star in safe AI development

  • And even Apple, which recently announced its own AI capabilities

Each of these companies is vying to define the future of artificial intelligence. And for Zuckerberg, the stakes are existential: failure to compete effectively in this domain could spell long-term irrelevance for Meta’s tech empire.

Why This Matters

Meta’s strategic approach diverges from rivals in one critical way: its commitment to open-source AI. While OpenAI and Anthropic keep their models mostly proprietary, Meta aims to position its Llama family of models as the “Android” of AI—a free and extensible platform that becomes the standard layer powering the world’s AI applications.

But open-source AI isn’t without controversy. It raises security concerns and complicates questions of accountability and bias. Meanwhile, Apple and Google are placing greater emphasis on controlled AI ecosystems embedded within their products.

As the industry races toward AGI, companies are grappling with existential questions: Who gets to build AI that rivals human intelligence? Who governs it? And how do we ensure it serves the public good?

The Superintelligence Stakes

Zuckerberg’s direct involvement in building a superintelligent AI reflects the intensity and urgency of the AI competition gripping Silicon Valley. It also reaffirms Meta’s belief that AI is not just a product enhancement—it is the battleground for the future of computing.

Whether Meta’s bet on open-source superintelligence pays off remains to be seen. But one thing is certain: Zuckerberg is all in.