Image by fabio on Unsplash
Meta hired top AI talent to supercharge its future. But weeks in, some are quitting, others are clashing, and the shake-ups just keep coming.
Meta’s bold bet on AI is hitting turbulence—fast.
Just days after joining the company, Shengjia Zhao, co-creator of ChatGPT, almost jumped ship back to OpenAI. He even signed rehire paperwork before being coaxed to stay. The band-aid? A shiny new title: Chief AI Scientist at Meta.
It’s been a dramatic few months for CEO Mark Zuckerberg’s personal superintelligence project—a moonshot initiative meant to create AI that surpasses human abilities. To make it work, he’s pulled in hotshot newcomers like Zhao, former Scale AI CEO Alexandr Wang, and ex-GitHub boss Nat Friedman.
But this high-stakes hiring spree has created more sparks than synergy.
Big hires, bigger exits
Meta’s new AI stars aren’t exactly easing into the corporate culture.
- Ethan Knight, a machine learning scientist, left just weeks in.
- Avi Verma, once at OpenAI, signed on with Meta but ghosted before day one.
- Rishabh Agarwal, a research scientist who joined in April, announced his exit this week with a cryptic post on X about taking on a “different kind of risk.”
Even long-timers are heading for the door. Chaya Nayak and Loredana Crisan, both veterans of Meta’s generative AI teams, announced their departures after nearly a decade each with the company.
One investor familiar with the internal dynamics summed it up this way: “There’s a lot of big men on campus.”
So, what exactly is going on?
Zuckerberg isn’t shy about making bold moves—and AI is the latest battleground. His plan: build what he calls “personal superintelligence.” Translation? AI that’s not just smart, but smarter than us.
To get there, he’s reshaped the org chart—hard. In just six months, Meta’s AI division, now called the Meta Superintelligence Lab (MSL), has already gone through four major restructurings. That’s a lot by any standard.
Image by BoliviaInteligente on Unsplash
One of those changes included forming a new, ultra-secretive team called TBD (yes, that’s really the name). It’s led by Alexandr Wang and staffed with big names lured in with nine-figure signing bonuses and promises of massive computing resources.
But the results are mixed. Their marquee model, Llama Behemoth, failed to live up to expectations and has been quietly shelved. The focus now? Starting over with newer models.
Culture clash in a $1.95 trillion company
Not everyone’s adjusting well.
Wang, just 28 and known for his fast-paced startup savvy, has been described as ambitious—but not used to Big Tech’s pace or politics. Inside sources say there are tensions over his leadership style and the shift from scrappy entrepreneurship to Meta’s sprawling corporate structure.
Some new hires are frustrated by red tape and stalled promises around compute power. One insider said they felt blocked by the very system that hired them.
Others point to Zuckerberg himself. While he’s deeply involved in TBD’s progress, some say he’s been too hands-on. The company flat-out denied claims of micromanagement, calling it “manufactured tension.”
Still, it’s clear something’s not clicking.
Ongoing shake-ups—and potential slowdowns
Meta hasn’t given up. It still believes the Meta Superintelligence Lab offers the most computing muscle per researcher in the industry.
But some roles remain undefined. Once-key players like Ahmad Al-Dahle and long-standing figures like Yann LeCun are now reporting to new leadership, or sitting outside of key decision-making circles. Even Chris Cox, Meta’s longtime product chief, has been partially cut out of the generative AI loop, with Wang now reporting directly to Zuckerberg.
Image by Christopher Gower on Unsplash
Looking ahead, the company is pressing pause on new hiring—unless the role is critical. Leaders issued a memo saying this hiring freeze would help “thoughtfully plan” for the team’s 2026 headcount strategy.
The bottom line
Meta’s AI ambitions are sky-high. But turning vision into reality is proving chaotic.
On paper, hiring top talent and investing billions sounds like a slam dunk. But in practice, it’s a culture clash: startup energy meets corporate sprawl, egos meet hierarchy, and urgency meets bureaucracy.
Whether Meta’s AI gamble pays off still remains to be seen. For now, the road to superintelligence is uneven—and surprisingly human.
Keywords: Meta AI, Mark Zuckerberg, Shengjia Zhao, personal superintelligence, Alexandr Wang, AI leadership, Meta Superintelligence Lab, big tech hiring, AI startup culture, OpenAI, AI turbulence, tech news
