Reports have suggested that Meta’s aggressive push into artificial intelligence has created unexpected fault lines inside the company.
According to The Wall Street Journal, after a lavish hiring spree that brought in dozens of researchers from rivals like OpenAI, the firm established a superintelligence unit known as TBD Lab.
Members of this group were said to have been placed in a separate, restricted area near chief executive Mark Zuckerberg, with special badge access. They were even left off Meta’s internal organisation chart creating what insiders described as an aura of exclusivity.
The secrecy and privilege surrounding this unit reportedly unsettled other Meta employees. While the company dismissed such accounts as exaggerated or mischaracterised, critics argued that the optics made it look as though an elite circle of “millionaire malcontents” had been airlifted into the organisation, Business Times reported.
Are coders wealthier than athletes?
The controversy was fuelled by the staggering sums involved. Industry chatter placed some offers at as much as $300 million over four years, with one researcher allegedly securing a deal worth $250 million — a package richer than the contract of basketball star Stephen Curry, CNBC-TV18 reported.
Meta reportedly dangled US$100-million bonuses to lure top talent from OpenAI, Google and Anthropic, though Anthropic’s chief executive Dario Amodei claimed his team had turned down such offers, MarketWatch said.
The broader market reflected similar trends. Median salaries for AI engineers rose from about $220,000 in 2022 to $280,000 in early 2024, while some OpenAI candidates were said to have received offers nearing $925,000 when equity and bonuses were included.
These figures eclipsed historical precedents: Robert Oppenheimer’s compensation during the Manhattan Project, IBM CEO Thomas Watson’s pay in the 1940s, and even Claude Shannon’s earnings when he pioneered information theory were dwarfed by these modern contracts.
Impact Shorts
More ShortsWhat’s happening in Meta’s ranks?
Such disparities appear to have unsettled longer-serving Meta staff. Employees were said to be lobbying for higher pay or transfers into the prized AI lab. One individual, despite receiving a grant worth millions, reportedly quit after concluding that newcomers were earning multiples more, The Wall Street Journal reported.
For many observers, this highlighted the dangers of introducing what one commentator likened to a “better-paid invasive species” into an existing workplace ecosystem.
While Meta insisted that the The Wall Street Journal’s reporting was overblown, the company has quietly frozen new hiring for its AI unit after the initial recruitment frenzy. This pause suggested a recalibration in strategy after months of lavish spending.
Are these echoes of the dotcom bubble?
Critics outside Meta have drawn parallels between today’s AI arms race and the excesses of the dotcom era. Back then, firms measured success not by profits but by their ability to burn through investor money. Extravagant offices, lavish parties and high-priced perks were commonplace, with one founder declaring that turning a profit was “so old-economy”.
Observers have argued that the current “AI bubble” reflects a similar mindset, where valuations of money-losing companies have soared into the hundreds of billions and tech giants are competing to outspend one another in talent wars.
Risk of overpaying for a mirage
Sceptics also pointed to the scientific uncertainties underpinning these wagers. Meta’s own chief scientist Yann LeCun has argued that current AI remains “dumber than a cat” and far from delivering artificial general intelligence (AGI). Large language models, according to researchers, cannot reliably link words to real-world meaning, making them prone to errors in high-stakes scenarios. A recent case study even documented a man nearly dying after following medical advice generated by ChatGPT.
If these concerns prove correct, Meta’s billion-dollar bet on superstar hires may not deliver the expected breakthroughs. Historical precedent offers a cautionary tale: transformative advances such as the transistor, the laser and the foundations of the internet emerged from collaborative environments like Bell Labs, where modestly paid engineers worked in relative anonymity rather than under the glare of multimillion-dollar contracts.
Did Meta overrate pay as incentive?
For now, Meta finds itself grappling not only with the technical challenges of building superintelligence but also with internal tensions sparked by the way it has chosen to pursue that goal.
By poaching talent at unprecedented prices and creating a cloistered elite inside its own headquarters, the company may have inadvertently weakened morale among its existing workforce. Whether this experiment in high-stakes hiring will pay off — or prove another example of Silicon Valley hubris — remains an open question.