Mark Zuckerberg is on a spending spree to fix a major problem: Meta lost many of its top artificial intelligence experts over the past few years. Now, he’s paying unprecedented amounts to rebuild his AI team as the company races to develop what he calls “superintelligence.”
Meta once employed researchers who later founded successful AI companies like Perplexity, Mistral, and Fireworks AI. But a wave of departures left the company struggling to keep pace with rivals like OpenAI and Google.
“They already had the best people and lost them to OpenAI. This is Mark trying to undo the loss of talent,” one former Meta AI employee told Forbes.
The brain drain has been so severe that executives at competing AI companies considered Meta’s remaining talent pool subpar. “We might be interested in hiring some of the new people Mark is hiring now. But it’s been a while since we were particularly interested in the people who were already there,” said a senior executive at a major AI company.
Former employees describe Meta’s AI work environment as chaotic. Teams formed and disbanded within weeks, and researchers cycled through multiple managers. When Meta consolidated its AI research under a more product-focused team called GenAI in 2023, researchers faced grueling sprints to ship products, working late nights and weekends.
“We went like bananas, working our tails off for the entire year,” said a former senior researcher. The environment became increasingly disorganized with senior leaders disagreeing on technical approaches and teams fighting for credit.
Meta’s AI reputation took another hit in April when it released Llama 4, widely criticized for poor reasoning and coding abilities. “Llama 4 was a disaster,” a former researcher told Forbes.
Zuckerberg’s solution? Open his wallet – wide. He has reportedly offered compensation packages worth $100-300 million over four years to attract talent from OpenAI, Google DeepMind, and Anthropic. In June, Meta invested $14.3 billion in data labeling startup Scale AI, bringing in its 28-year-old CEO Alexandr Wang to lead a new “Meta Superintelligence Labs.”
Similar Posts
The new lab sits next to Zuckerberg’s office, highlighting its importance. In a memo to staff, Zuckerberg wrote: “As the pace of AI progress accelerates, developing superintelligence is coming into sight.”
Beyond talent acquisition, Meta is building massive AI infrastructure. Zuckerberg revealed the company is constructing “multi-gigawatt data centers” including one called Hyperion that “is going to scale up to five gigawatts” and covers “a significant portion of the footprint of Manhattan.”
The spending is enormous – Meta now expects total expenses in 2025 to reach between $114 billion and $118 billion, with $66-72 billion in capital spending.
Despite these efforts, people continue to leave. Anthropic recently hired Laurens van der Maaten, who co-led research strategy for Meta’s Llama models. And according to SignalFire, Meta has a 64% retention rate for technical talent – lower than Anthropic (80%), DeepMind (78%), and OpenAI (67%).
Some competitors claim Zuckerberg’s approach creates “mercenaries” rather than “missionaries.” OpenAI’s Sam Altman wrote to staff: “Missionaries will beat mercenaries,” adding “what Meta is doing will, in my opinion, lead to very deep cultural problems.”
Will Zuckerberg’s expensive gamble pay off? For now, Wall Street seems optimistic – Meta’s stock jumped 10% after recent earnings showed $47.52 billion in revenue, beating expectations despite the massive AI investments.
