Meta superintelligence research is already facing challenges, just months after Mark Zuckerberg unveiled the company’s flagship Superintelligence Labs (MSL).
Despite proposing some of the most lucrative packages in Silicon Valley history, the project is finding it difficult to hold on to its star hires.
The Superintelligence lab, launched in April 2025 with the aim of leapfrogging Meta competitors in artificial general intelligence (AGI), has been hit by high-profile departures.
Rishabh Agarwal, recruited from Google DeepMind earlier this year on a reported $1 million salary, announced in late August that he would be leaving after barely five months.
“It was a tough decision not to continue with the new Superintelligence TBD lab, especially given the talent and compute density. But after 7.5 years across Google Brain, DeepMind, and Meta, I felt the pull to take on a different kind of risk,” Agarwal wrote in a farewell post on X.
He also repeated Zuckerberg’s own words back at him: “In a world that’s changing so fast, the biggest risk you can take is not taking any risk.” The quote has since been widely interpreted as researchers using the Meta chief’s mantra to justify walking away.
Avi Verma and Ethan Knight, both previously with OpenAI, have returned to their former employer after brief stints at MSL. In a further blow, longtime Meta executive Chaya Nayak has also left, joining OpenAI to work on special initiatives.
These issues have led to uncomfortable questions for Meta. If billion-dollar offers cannot retain talent, what can? Insiders point to structural problems: frequent reorganisations, shifting goals, and reports of micromanagement at the top.
The company recently split its AI staff into four separate groups, creating suspense inside a lab already tasked with one of the most ambitious projects in tech.
Experts say money is not the ultimate driver for the best minds in the field. DeepMind cofounder Demis Hassabis once said frontier researchers want to “help influence how AGI plays out and steward the technology safely into the world” rather than simply chase paycheques.
Similarly, Anthropic’s cofounder Benjamin Mann said: “My best case at Anthropic is we affect the future of humanity. My best case at Meta is we make money.”
Meanwhile, rivals are capitalising. OpenAI has not only regained former staff but strengthened its bench at a time when it publicly criticised Meta’s aggressive poaching tactics.
Elon Musk’s xAI is also pulling engineers away from Zuckerberg’s company, with reports noting at least 14 defections this year alone. Unlike Meta’s cash-heavy approach, Musk promotes a performance-driven culture anchored in equity and speed.
Meta has invested heavily in leadership, hiring Scale AI founder Alexandr Wang and former GitHub CEO Nat Friedman to run its AI efforts. But reports of disagreements between Zuckerberg and Wang over timelines for superintelligence highlight deeper tensions.
Meta’s resources can buy time and talent, but not loyalty or mission alignment.
