Meta’s blitz to lure star AI scientists with multi-million-dollar paydays has backfired, sparking resentment among longtime staff and a wave of high-profile defections — even from some of the recruits Mark Zuckerberg personally courted, according to a report.
Meta has offered extraordinarily lucrative pay packages to lure top AI researchers, with some deals reaching up to $250–$300 million over four years, and select researchers reportedly receiving as much as $100 million in the first year through a mix of cash, stock, and other incentives.
Inside the company’s Menlo Park, Calif. headquarters, some of the star hires that make up the new “TBD Lab” sit in a badge-restricted zone near Zuckerberg’s desk, its members’ names hidden from the internal directory.
The secrecy, plus the cash lavished on outsiders, has fueled grumbling across the rank-and-file as employees fight for limited resources under a hiring freeze, the Wall Street Journal reported.
“This is yet another series of false, exaggerated or mischaracterized claims,” a Meta spokesperson told The Post.
“We’d have thought this level of navel-gazing was beneath the Wall Street Journal, but apparently we were wrong.”
The Post has sought comment from the Journal. News Corp, parent company of The Post, also owns Dow Jones and its subsidiary, the Journal.
At least one recruit granted millions quit anyway, complaining others were still paid “multiples more.” Others, including hires from blue chip firms such as OpenAI, xAI and Scale AI, have already left, the Journal reported.
One top scientist nearly bolted back to OpenAI just a week after joining before Meta scrambled to keep him with a lofty title and a massive pay boost, according to the Journal.
Among the casualties are Avi Verma and Ethan Knight, new hires from OpenAI and xAI, jumped ship back to OpenAI, the Journal reported.
Rishabh Agarwal, a Toronto-based researcher hired in April after directly appealing to Zuckerberg, resigned when told he had to work in-person and instead joined Periodic Labs, a startup launched by a former OpenAI executive.
Ruben Mayer, a product manager poached from Scale AI, also left, citing personal reasons, according to the Journal.
The biggest drama centered on Shengjia Zhao, a co-creator of ChatGPT. Within a week of joining Meta, Zhao had signed paperwork to return to OpenAI. Only a last-minute intervention — Meta handing him the title of chief scientist and what the Journal reported was a tripling of his pay — kept him in Menlo Park.
In a statement to the Journal, Meta disputed that his compensation was tripled but admits his role was formalized once the team took shape.
“Shengjia co-founded MSL and has been our scientific lead since day one,” a Meta spokesperson told The Post.
“We formalized his role once our recruiting had ramped and the team had taken shape.”
Meta has showered recruits with huge packages, luring more than 50 AI experts this summer alone. Of those, 21 of the new hires were poached from OpenAI, more than a dozen from Google, plus others from Apple and Elon Musk’s xAI.
But the open checkbook created a backlash inside Meta’s existing workforce, according to the Journal.
Longtime engineers began leveraging competing offers to wring raises and promotions. In July, a handful of staffers from Meta’s AI infrastructure team snagged offers from former OpenAI executive Mira Murati’s startup, Thinking Machines Lab.
When they brought the offers to Meta, their compensation was bumped and they were transferred to the TBD Lab. Meta insists those moves were planned regardless of outside offers, but the timing raised eyebrows.
“It’s classic talent management 101—if you don’t lay the cultural groundwork, you’ll burn these stars out and waste millions,” said Laszlo Bock, former head of HR at Google.
The TBD Lab’s exclusivity has become a flashpoint. Employees need special badge access to enter, and unlike every other group in the AI division, its roster isn’t visible on the company’s internal organization chart.
That secrecy, paired with reports of inflated pay packages, created the impression of a privileged caste inside Meta.
Other teams complain they are starved of computing power, and managers say hiring is so tightly controlled that even filling vacant roles requires personal sign-off from Meta’s top AI leadership.
The dysfunction at Meta mirrors the AI sector’s broader frenzy. In August, OpenAI handed out one-time bonuses worth millions to keep staff from leaving. Apple lost a string of key AI developers this summer to rival labs.
At Safe Superintelligence, the startup co-founded by OpenAI’s Ilya Sutskever, employees are reportedly discouraged from even listing the company on LinkedIn, to avoid becoming recruitment targets.
For researchers, the poaching wars have made it easier to bounce between labs, driving up compensation and creating a revolving door of talent.
Still, even with Zuckerberg spending millions to create a fortress-like AI lab, insiders question whether Meta can keep its stars from bolting again.
