Meta Platforms Inc. is creating an artificial intelligence (AI) lab to pursue an AI system that surpasses human intelligence with “superintelligence.”
The secretive research lab, set to include Scale AI CEO Alexandr Wang, was first reported by the New York Times. Meta is also in advanced talks to invest more than $10 billion in Scale AI, according to a report from Bloomberg. (Other reports suggest an investment of up to $15 billion, or a 49% stake in Scale AI.)
Meta CEO Mark Zuckerberg is spearheading the effort of assembling a so-called “superintelligence group” of 50 top AI researchers and engineers in what amounts to a dramatic restructuring of Meta’s AI efforts. His ultimate goal, reports suggest, is to move beyond broader AI research to a more specialized focus on Artificial General Intelligence (AGI) and, ultimately, superintelligence. The idea is to weave AGI across Meta’s product suite of chatbots, augmented reality glasses, and other forthcoming projects.
Zuckerberg has been personally recruiting top AI talent and prioritizing the staffing and development of the new team, Bloomberg reports. To that end, he reportedly created a WhatsApp group with top executives called Recruiting Party to find talent. Potential employee targets include Scale AI employees, as well as researchers from Meta’s rivals.
The concept of a superintelligence lab signals not only a major reorganization but Zuckerberg’s mounting frustration with recent product releases such as Llama 4. Consequently, Meta delayed plans to release its next iteration, called Behemoth, reported the Wall Street Journal.
Meta and Scale AI had no comment.
“This is the next frontier from gaining 7% to 10% savings in productivity to being able to delegate tasks that would otherwise take you a week,” Amit Barave, vice president of product management for Webex applications at Cisco Systems Inc., said in an interview, explaining the benefits of superintelligence for large enterprises.
“Treating AGI as a ‘holy grail’ attracts outsized investment, often with unclear returns,” AI strategist Songyee Yoon said in an email. “With finite capital and natural resources, we must prioritize technologies with near-term impact.”
The grand notion of creating a hypothetical AI system that surpasses human cognitive capabilities across all domains signals Meta’s most-ambitious effort to date to compete in the rapidly expanding AI race with rivals OpenAI, Alphabet Inc.’s Google, and Apple Inc. Facebook’s parent company said it plans to spend up to $65 billion in capital expenditures this year on AI infrastructure, and it has several AI developments up its sleeve. On Tuesday, for example, OpenAI announced o3-pro, a new reasoning model, that it claims boasts better performance in “science, education, programming, data analysis, and writing.”
Meta also has plans to introduce a stand-alone Meta AI app this year, and it is testing a paid-subscription service that would compete with OpenAI’s ChatGPT. Meta claims its AI assistant is used by 1 billion monthly active users across its product line.
