Mark Zuckerberg has launched an aggressive campaign to catch up in artificial intelligence after Meta’s AI products underperformed at a company conference in April. The Facebook parent company is now spending billions and offering unprecedented compensation packages to compete with OpenAI and Google in developing “superintelligence.”
According to reporting by Mike Isaac and Cade Metz in The New York Times, Zuckerberg realized Meta was falling behind when the company’s new AI model failed to impress developers at its April conference. Features like voice interactions were not ready, leaving attendees underwhelmed despite earlier promises that the technology would be a “beast.”
The setback triggered what sources describe as a frenzy of activity. Zuckerberg demoted Meta’s vice president in charge of generative AI and invested $14.3 billion in the startup Scale AI, hiring its 28-year-old founder Alexandr Wang. The company has also approached other AI startups, including search engine Perplexity, about potential deals.
Meta’s recruitment efforts have reached extraordinary levels. The company contacted more than 45 AI researchers at OpenAI alone this month, with some receiving formal offers as high as $100 million. At least four OpenAI researchers have accepted Meta’s proposals. Other tech giants are responding in kind, with CEOs from Google, Microsoft, and OpenAI personally involved in recruiting battles.
“The market is setting a rate here for a level of talent which is really incredible, and kind of unprecedented in my 20-year career as a technology executive,” said Meta’s chief technology officer Andrew Bosworth in a recent interview.
The competition centers on developing superintelligence, a hypothetical form of AI more powerful than the human brain. Currently, only OpenAI, Anthropic, and Google are considered to have the expertise for this technology. Zuckerberg wants Meta included in this exclusive group.
Meta’s struggles stem partly from its open-source strategy with its Llama AI model, where the underlying technology is shared publicly. While initially seen as an advantage, this approach backfired when Chinese startup DeepSeek created more advanced models based on Llama’s foundation.
Internal problems have compounded Meta’s challenges. The company’s AI team discovered that benchmarks were designed to make Meta’s models appear more advanced than they actually were. Zuckerberg was reportedly upset when he learned about these misleading tests after publicly promoting the results.
Meta’s AI division has grown to over 1,000 people this year from a few hundred two years ago. However, rapid expansion has led to management conflicts and burnout among engineers struggling under Zuckerberg’s intensive oversight style.
The company has established a new superintelligence lab and continues pursuing high-profile AI researchers. Zuckerberg personally reached out to Ilya Sutskever, OpenAI’s former chief scientist, though Sutskever declined to join.
Whether Meta can close the gap remains uncertain, but the AI talent war has permanently changed Silicon Valley’s compensation landscape.