Meta’s Big Bet on AI Talent
In June 2025, Meta launched its Superintelligence Lab (MSL) with a clear goal: to leapfrog rivals in the race toward Artificial General Intelligence (AGI). But just two months later, headlines are less about breakthroughs and more about exits.
According to WIRED, at least three researchers have already quit the lab, some returning to OpenAI despite Meta’s competitive offers.
Notable departures include:
- Avi Verma and Ethan Knight — who rejoined OpenAI after only weeks at Meta.
- Rishabh Agarwal, recruited from DeepMind on a reported $1M package, who wrote on X that he wanted “a different kind of risk.”
- Chaya Nayak, longtime head of Meta’s generative AI products, who has since joined OpenAI.
The Money Didn’t Work
Meta was reportedly offering $1M+ salaries, with some contracts reaching $300M over four years.
But the exodus shows that money doesn’t buy alignment.
- Internal reorganizations at MSL left teams unsettled, according to the Financial Times.
- Long-time AI staff were reportedly demoralized by the focus on new hires, as Business Insider observed.
Culture Over Compute
What’s emerging is a clear lesson for enterprises of all sizes:
Mission Matters
Researchers are choosing labs where governance, values, and long-term direction are coherent.
Stability Beats Spectacle
Constant reshuffling and secrecy erode trust.
Culture Scales AI
It’s not just about algorithms; it’s about how people align around them.
Or, to put it simply: culture eats compute for breakfast.
The Enterprise Takeaway
Most businesses don’t have Meta’s budget, but that’s not the barrier to AI adoption. The real challenge is building systems where people feel aligned, safe, and clear on purpose.
If billion-dollar labs can lose talent this quickly, enterprises should take note: frameworks for trust, compliance, and culture are as vital as any GPU cluster.
So what for enterprises?
When you think about your AI strategy, don’t just ask “What tools should we buy?” Ask instead:
- How do we align teams on risk, compliance, and mission?
- What guardrails give people confidence to use AI responsibly?
- How do we make our AI adoption not just powerful, but purposeful?
