Meta Recruits Top AI Researchers From OpenAI Amid Superintelligence Race
Meta has quietly brought on board several prominent AI researchers from OpenAI, intensifying the battle for top talent in the industry.
Among the most notable hires is Trapit Bansal, a key contributor to OpenAI’s reinforcement learning programme and co-developer of its early reasoning model, known internally as o1.
His move was confirmed by OpenAI spokesperson Kayla Wood, who told TechCrunch that Bansal left the company in June.
Who Are The New Recruits?
Joining Bansal at Meta’s superintelligence lab are three other former OpenAI researchers—Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai.
The trio had previously worked at Google DeepMind before helping OpenAI set up its Zurich office in late 2024.
Their departures come amid broader concerns within OpenAI about losing talent to rivals, especially after reports emerged that Meta was offering sign-on bonuses as high as $100 million.
Bansal is expected to play a central role in Meta’s push to build a next-generation AI reasoning model, designed to compete with OpenAI’s o3 and DeepSeek’s R1.
While Meta has previewed its Llama 4 Behemoth model, delays over its performance have left the company without a publicly available reasoning model.
A Billion-Dollar Talent Hunt
Meta’s efforts to secure elite AI minds go well beyond OpenAI.
Earlier this year, Meta made headlines for investing $14.3 billion in Scale AI, acquiring a 49% stake and bringing its CEO, Alexandr Wang, into the fold.
In parallel, Meta is in advanced discussions to acquire PlayAI, a voice AI startup backed by Y Combinator, with plans to absorb its team and expand its voice agent capabilities.
The startup’s flagship technologies include two proprietary models: the lightweight Play AI 3.0 mini, built for hardware efficiency, and Dialog, a much larger system with ten times more parameters.
If the deal closes, Meta will add more specialised voice AI tools to its growing arsenal.
Personal Outreach And Failed Bids
Meta CEO Mark Zuckerberg has reportedly taken a hands-on approach to the hiring spree, personally reaching out to top researchers through a WhatsApp group dubbed “Recruiting party.”
Despite these efforts, some high-profile targets—including OpenAI co-founders Ilya Sutskever and John Schulman—have turned down offers to focus on their own ventures, such as Safe Superintelligence Inc.
A report by The Wall Street Journal suggests that while several Meta offers were accepted, others were rejected in favour of counter-offers from OpenAI, including more pay and broader responsibilities.
OpenAI CEO Sam Altman recently said on a podcast, dismissing Meta’s recruiting tactics,
“None of our best people have decided to take him up on that.”
What Is Meta Building With This Team?
The hires feed directly into Meta’s new AI superintelligence group—an internal lab with a mission to develop models that outperform humans in a wide range of tasks.
The initiative was launched after Meta’s struggles with Llama 4 Behemoth, which reportedly fell short of expectations and forced delays in deployment.
The lab is also drawing talent from other AI labs and startups.
Bloomberg reports that Meta has hired former DeepMind researcher Jack Rae and Johan Schalkwyk, a machine learning expert from AI startup Sesame.
Meanwhile, Meta is pursuing tech investor Daniel Gross and former GitHub CEO Nat Friedman, who currently lead Safe Superintelligence alongside Sutskever.
Massive Infrastructure Plans To Support AI Push
Behind these talent acquisitions is a vast investment in AI infrastructure.
Meta is on track to spend up to $65 billion this year on its data centres, including plans to build a massive new facility powered by more than 1.3 million Nvidia GPUs.
This scale of compute power is crucial to developing superintelligent models, particularly those focused on reasoning—an area where OpenAI, DeepMind, and DeepSeek have already made visible progress.
With several of OpenAI’s engineers now heading to Meta, the competition for AI supremacy is clearly heating up.
Whether Meta can catch up to the leaders remains to be seen, but the company is showing it’s willing to spend—and poach—to get there.