At Meta, the last few months have been nothing short of being what can only be described as “action-packed.” Maybe that is what pressure does to you. The rise of AI has shown the landscape can change any moment, and those who fail to adapt could be left behind. Not long ago, Google issued code red at HQ, when it found out a new contender – OpenAI – had arrived to give it a reality check as well as a run for money. The tech giant that it is, Google, responded in kind, and swiftly moved to counter the challenge with Gemini. You could say Meta took a different route.

Instead of revamping an existing product or launching something new, it went on an all-out war with some of its biggest rivals in the AI space, including OpenAI and Google, to poach some of their best talent – scientists, engineers, you name it. Mark Zuckerberg personally took it upon himself to reach out and recruit people for his Meta Superintelligence Labs, a significant rebranding of its AI unit, with a goal – as you can probably tell from the naming – to achieve “superintelligence” which you can say is their version of AGI, basically the point where AI starts to behave on the lines of a human in terms of thinking and doing things.

The cornerstone of this huge marketing endeavour was a massive $14.3 billion investment in Scale AI which would give Meta a 49 percent stake in the company as well as bringing its 28-year-old CEO, Alexandr Wang, to lead Meta Superintelligence Labs. Until this point, it was widely understood – and communicated – that the illustrious Yann LeCun would continue to lead Meta’s Fundamental AI Research lab, or FAIR per usual even if it was becoming increasingly clear, Meta was on course for a complete revamp, which was only accentuated by the string of high-profile hires that followed.

Come and go

Meta basically went on a hiring spree with Zuckerberg offering candidates big pay cheques and hefty bonuses to make them jump ship, particularly from OpenAI – an act its CEO Sam Altman likened to a mercenary during a town hall, following which he revised the salaries of some top performers to boost morale and keep them from leaving. Some employees took the bait, many didn’t. But Meta was able to snag enough talent to come out and announce the new superintelligence labs initiative, with Wang posting an open letter to declare Meta was ready to get into the AI game, all guns blazing.

But as the dust started to settle, Meta found itself at the centre of another challenge: how to retain these new hires. Key researchers, including some lured with multi-million-dollar packages, have resigned in recent weeks, raising questions about the internal culture and the long-term viability of Zuckerberg’s latest venture.

Amongst the notable exits is IIT Bombay alumnus Rishabh Agarwal, a prominent AI scientist who joined Meta from Google DeepMind in April. Agarwal, who was reportedly offered a significant compensation package, announced his departure in August, stating he was seeking a “different kind of risk.” Adding to the talent drain, Chaya Nayak, a director of product management who played a key role in the development of Meta’s Llama models, has also left the company. She is joining OpenAI.

Furthermore, at least two researchers, Avi Verma and Ethan Knight, who had previously been at OpenAI before joining Meta’s superintelligence team, have returned to their former employer. Moreover, Knight left less than a month after joining, suggesting a big salary alone may not be enough to retain top talent. Research environment and culture may be as important, even more, given that this type of talent would presumably have many options today. In the same meeting he called Meta a mercenary, Altman called himself and OpenAI a missionary, meaning what they do probably has a bigger purpose.

All is well?

Meta and Wang, on their part, have downplayed the reverse brain drain, saying that such departures are normal within the industry even if the exit of such “precious” talent in a short period of time might seem like a major setback to its AI ambitions. Wang recently posted on X, “We are truly only investing more and more into Meta Superintelligence Labs as a company. Any reporting to the contrary of that is clearly mistaken.”

At the same time, ongoing reports speak of a hiring freeze and another “restructuring” of the newly founded AI division into four entities: TBD Lab, which is said to be “a small team focused on training and scaling large models to achieve superintelligence across pre-training, reasoning, and post-training, and explore new directions such as an omni model,” FAIR, now “an innovation engine for MSL,” Products and Applied Research for bringing Meta’s “product-focused research efforts closer to product development,” and finally MSL Infra, which will focus on “accelerating AI research and production by building advanced infrastructure, optimised GPU clusters, comprehensive environments, data infrastructure, and developer tools to support state-of-the-art research, products, and AI development across Meta.”

A lot is at stake. AI is like that. The way things stand, you can’t just stand and wait for things to happen on your own. You’ve got to make them happen. Meta, given the current state of things, must multiply every effort by two, to stabilise its superintelligence team and continue its push to the forefront of AI research where historically it is lagging.

– Ends