Meta CEO Mark Zuckerberg has been directly courting some of the top minds in artificial intelligence, personally leading a high-stakes recruitment drive to build a new superintelligence-focused lab and reboot Meta’s Llama model. According to Bloomberg, this unusual hands-on approach has researchers across the industry asking: “Is that really Zuck?”

Zuckerberg is said to be contacting AI researchers himself—often via email or WhatsApp—referencing their work and requesting a quick 15-minute conversation. At Google alone, dozens of researchers have received such messages. For those who agree to take the call, the Meta CEO outlines an ambitious vision: high-risk innovation, access to Meta’s massive infrastructure, and significant autonomy for the new AI team. These individuals will reportedly work directly alongside Zuckerberg at Meta’s headquarters, where office space is already being rearranged for the incoming talent.

Much of the buzz has centered around the staggering compensation packages Meta is offering—some reaching into eight-figure territory. “Hiring the best AI researcher is like hiring a star basketball player,” one source noted. “There are very few of them, and you have to pay up.”

A prime example is Meta’s acquisition of Scale AI CEO Alexandr Wang, a move estimated to be worth $14.3 billion—roughly the equivalent of 14 Instagram acquisitions. This jaw-dropping figure makes it arguably the most expensive tech hire ever. Wang addressed the impact in a note to employees: “Opportunities of this magnitude often come at a cost. In this instance, that cost is my departure.”

Wang is expected to take on the role of “Chief AI Officer” at Meta, while Jack Rae, a principal researcher from DeepMind, has also joined and will lead pre-training efforts in the new lab. Sources suggest that the team will be formally unveiled once Zuckerberg finalizes key hires.

Zuckerberg’s aggressive recruiting tactics are shaking up the AI industry. Just before a deadline to hire senior OpenAI employees, OpenAI CEO Sam Altman published a pointed essay emphasizing their core mission: “Before anything else, we are a superintelligence research company.” In another instance, after Zuckerberg attempted to lure DeepMind CTO Koray Kavukcuoglu, Google responded by promoting him to SVP with a direct reporting line to CEO Sundar Pichai.

Internally, Meta is feeling the pressure. Llama, Meta’s flagship large language model, has reportedly lagged so much that product teams have floated the idea of using models from other providers—a move deemed unlikely but telling. Meanwhile, Meta’s engineers have already adopted Claude, an external AI model, in their internal coding tools.

While some existing Meta researchers may feel uneasy, others—particularly former Scale AI employees—are celebrating. With the $14.3 billion deal, many “Scaliens” are seeing massive financial windfalls and were reportedly celebrating with champagne.

Wang, in his emotional final all-hands meeting, didn’t disclose his role at Meta—but his influence will likely shape the future of AI there.