Meta's Mark Zuckerberg is in vain! Spending hundreds of millions in salaries, three AI geniuses have left in two months.

Meta's investment of billions of dollars into Meta Superintelligence Labs (MSL), originally seen as the ultimate weapon in the race for artificial general intelligence (AGI), has lost several core researchers within just two months of its establishment. Even offering a staggering compensation package of up to 150 million dollars has not been able to retain top AI talent, highlighting the intense competition for AI talent globally and concerns regarding management.

Three core researchers lost in two months

According to a report by WIRED, MSL lost at least three core researchers just two months after its establishment in the first quarter of 2025:

Avi Verma, Ethan Knight: Both return to OpenAI

Rishabh Agarwal: Formerly worked at Google Brain and DeepMind, chose to leave MSL.

Agarwal stated on X (formerly Twitter) that although MSL has amazing resources and computing power, he wants to challenge different risks. The outside world interprets that his long-term stay in Canada, along with the geographical distance from the main team located in Menlo Park, California, may be the trigger for his resignation.

What is even more concerning is that Chaya Nayak, the Director of Product Management for Meta's Generative AI, has also moved to OpenAI, indicating that the issues are not just individual choices, but rather a structural challenge related to team composition and direction.

The Dilemma of High Salaries to Retain Talent: $150 Million is Not Enough

According to another report by WIRED, Zuckerberg is offering a total compensation package of up to $150 million to attract top AI talent. However, despite the high salary levels in the AI field, they cannot replace the talent's demand for a sense of mission, research freedom, and organizational stability.

The Wall Street Journal (WSJ) pointed out that Meta's AI department has undergone multiple reorganizations in recent years, and recently it was further divided into four small teams, leaving employees feeling confused about the future direction. OpenAI CEO Sam Altman has criticized Meta's poaching behavior as "extremely aggressive," but it has been proven that high salaries for poaching do not equate to long-term commitment.

Management and Cultural Challenges: Dream Teams Can Also Become 'Hell Teams'

MSL has gathered world-class AI researchers and engineers, but the Los Angeles Times describes that when a group of top minds collaborates, self-inflation, factional disputes, and conflicts over R&D priorities can lead to internal friction within the team.

In addition, Meta reportedly suspended hiring in its AI department in early August, shifting from "expansion sprint" to "internal integration." Although this can reduce personnel costs in the short term, the risk of talent loss remains if there is a lack of clear decision-making mechanisms and a culture of high trust.

External Cooperation and Future Variables

The new AI chief, Alexandr Wang, has reached a collaboration with Midjourney to accelerate video generation technology, attempting to fill technical gaps with external alliances. However, whether this strategy can be quickly integrated into Meta's product ecosystem remains to be seen.

Conclusion

Meta's MSL was originally highly anticipated, but within just two months, it lost three core AI researchers, exposing the fundamental problems that high salaries cannot resolve—culture, sense of mission, and stability. In the current heated global competition for AI talent, if Zuckerberg cannot find a balance in management and organizational culture, even with more financial investment, it may not lead to true long-term competitiveness.