Meta poaches three OpenAI researchers with $100M bonuses as Zuckerberg escalates AI talent war

Meta isn’t slowing down in its effort to dominate AI. Just days after dropping $15 billion into Scale AI and hiring its CEO Alexandr Wang to lead a new “superintelligence” division, Meta has now lured away three top researchers from OpenAI.

According to the Wall Street Journal (WSJ), Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai—who helped set up OpenAI’s Zurich office last year—are now part of Meta’s AI team. OpenAI confirmed the departures, and Beyer himself acknowledged the move on X, saying he’s “excited about what’s ahead” but dismissed the rumored $100 million signing bonuses as “fake news.”

“Meta Chief Executive Mark Zuckerberg has poached three OpenAI researchers to join his superintelligence efforts, a recruiting coup intended to help steer Meta Platforms out of an AI crisis,” the WSJ reported.

“The social-media giant hired Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai, who were all working in OpenAI’s Zurich office, according to people familiar with the matter. The three set up OpenAI’s office in Zurich late last year. Before that, they worked together at Google DeepMind, the company’s AI unit.”

Zuckerberg’s AI Power Play: Meta Hires OpenAI’s Elite to Fuel AGI Race

Still, the timing couldn’t be more pointed. Meta is aggressively building a dream team, and Zuckerberg is personally involved in the recruiting effort. According to the report, he’s been reaching out to researchers over WhatsApp, hosting them at his homes, and coordinating everything through a group chat reportedly called “Recruiting Party.”

The Zurich trio isn’t just any group of engineers. Before OpenAI, they were at Google DeepMind, where they contributed to some of the most important breakthroughs in machine learning and computer vision, most notably, the Vision Transformer (ViT), which helped machines process visual data far more efficiently.

And they weren’t the only ones to jump ship.

Meta also hired Trapit Bansal, a key contributor to OpenAI’s o1 reasoning model, who previously worked with OpenAI co-founder Ilya Sutskever. Bansal’s move signals that Meta’s recruiting isn’t just about volume—it’s targeting the people building the next-gen reasoning frameworks that could define AGI.

Last week, OpenAI CEO Sam Altman claimed that Meta was offering $100 million bonuses in an attempt to lure his top talent. He downplayed the outcome, saying “none of OpenAI’s best people” took the offer. Meta’s CTO, Andrew Bosworth, called that claim “dishonest” in an internal meeting and added that offers of that size are limited to a very small number of senior roles. He also hinted that more OpenAI staff may be on the way to Meta.

Behind all this is a clear shift. Meta used to lag behind in AGI efforts, with its Llama 4 model falling short in key reasoning and math benchmarks. That’s likely part of what’s fueling this hiring spree. The company has already committed $65 billion to AI infrastructure, including plans for a facility packed with over 1.3 million Nvidia GPUs. And with Wang now at the helm of its AGI push, Meta is throwing everything it has at building systems that can reason, learn, and think at a higher level.

Losing researchers like Beyer, Kolesnikov, Zhai, and Bansal stings for OpenAI, not just because of their individual contributions, but because they were helping lead the charge in Europe. Still, Altman is putting on a confident front, insisting the core team is intact and that Meta’s poaching hasn’t disrupted operations.

What’s clear is that the AI talent war is heating up. These aren’t just hires—they’re strategic plays. And as the biggest names in tech try to get a step closer to AGI, the fight for the best minds isn’t slowing down.

Why It Matters

After the panic sparked in January by DeepSeek’s high-performing open-source model, Meta’s superintelligence team is quickly coming together. With four OpenAI researchers now jumping ship, Meta looks to have both the talent and resources to go head-to-head with the world’s leading AI company—and possibly dominate the space the same way it once did with social media.