In June, Mark Zuckerberg sometimes went to the ultimat Ole Mary in the Intensifing AI Race: He performed a brand-new Meta AI Lab after acquiring Scale AI $ 14.3 billion $ 14.3 billion-and then the billions spent more to hire some popular researchers and engineers of the industry.
In a few months, fast forward, and Zuckerberg may have recruited talent, but now the question is whether he will be able to avoid slipping through his fingers.
For that, we have to wait and see. Eventually, Meta re -designed its entire AI Division as a division of thousands of people as “Meta Superintendent Labs”. Within the division, we are hearing about a team-the team we are listening to by the industry-component AI researchers. To receive Superintendent – is called “TBD Lab”. In that team, only one active employee, Ethan Knight, has left so far, Meta spokesman Dave Arnold confirmed RuckusNight left less than a month after joining.
Of the other two, who allegedly leave TBD Lab, AV Verma and Rishabh Aggarwal, they actually started in roles with the lab, Arnold confirmed. Other employees who are Announced His departure from Meta was not part of the TBD lab, but a full superintendent organization, which included the Director of Product Management AI, Chaya Nayak, director of Rohan Verma and Meta Product Management of Mata. hero Left for Openai.
Meta’s hiring spright raised speed for its superintending team, with a 49 percent stake in Scale AI – an industry giant who provided training data to companies such as OpenAI, Google, Microsoft and Meta. As part of the deal, Scale AI CEO Alexandra Wang joined the meta to head the lab, and Meta hired employees of openi, deepmind, anthropic and other companies. Most of the rent came from Openai, of which Meta allegedly poisoned 10 as 10 Employees open some salary package and equity, By some reportsAdd up to $ 300 million in four years. (Meta dispute that number.)
But in recent months, AI industry employees told Ruckus That money was not enough to attract some talents, and some have rejected the meta reach and offer, especially to the TBD lab. In an industry with job safety and large salary, for at least one certain level engineers and researchers, those employees prefer to work somewhere that align with their values, no matter what those values may be (think: AI safety, social impact, acceleration, etc.).
Last week, meta Headlines made To send a memo about reorganization for a hiring freeze and its AI division. The company has said that both are natural followers for its recruitment processes in recent months. And although Meta is almost certainly tight on some aspects of the budget hiring its work, it is not necessary that the company is lagging or changing its strategy. When a publicly trading tech veteran spends billions to rent billions and wants to keep all its eggs in the basket, it is bound to stop and organized-either if it is sometimes trying to catch with the world’s leading laboratories in the AI Arms Race. For example, the company is Allegedly It is no longer working on its Bhamoth AI model, which failed to meet the expectations internal in favor of the new model.
So what is happening with working and restructuring? A memorandum Viewed by Ruckus Superintending freezing freeze tells that since Meta is starting to plan its budget for 2026, and “With such significant changes and growth in our ORG in 2025,” Company “is preventing hiring temporarily in all MSL (META Superintendent” Laboratories teams, with the exception to trading important roles. “
Memo states that it is “normal practice in several orgs across the meta over the last two years and will allow the leadership to plan our 2026 headcon development development, as we work through our strategy.” It states that Wang’s employees will meet weekly to evaluate business-matured fare “on a case based on the case.”
Last week, Wang Posted On X, “We are actually investing more and more in the Meta Superintendent Laboratories as only a company. Any reporting for the contrary is clearly wrong.”
Viewed by another memorandum RuckusPosted by Wang, it is said that the Superintendent of Meta would restructured the division and focus on three areas: research, products and infrastructure.
Those three areas will be covered by four teams. The first is the TBD lab, which Wang said, “A small team focused on training and scaling the larger models to receive the supervision during pre-training, logic and post-training, and discovered new directions such as the Omni model.”
Then Meta’s original AI Research Lab, or Fair, which is around for some time and the company feels Deproes in recent monthsIts leader, Jole Pineau, departed in May. Now, Wang wrote in the memo, the fair “will become an innovation engine for MSL, and we will aim to integrate and score several research ideas and projects from the fair in the large model run organized by TBD Lab.”
There is also “Product and Applied Research” team, which will bring “our product-focused research efforts closer to product development” and “AI Tech includes the first working teams on the voice, media, trust, avatar and developer columns,” Wang wrote. Finally, “MSL Infra” team enhances AI research and production by constructing advanced infrastructure, GPU clusters, comprehensive environment, data infrastructure, and developer tools to support state -of -the -art research, products and AI development in meta, “Wang wrote that Wang will also write that Meta will also disturb his” AGI foundation “organization and he wrote that Manta will also disturb the organization and Will take one of the teams.
One in Post On X, Meta spokesperson Andy Stone wrote that hiring freeze and restructuring were “worldly” updates because “all this is happening: a temporary break when we plan and forecast a temporary break when we plan and forecast, and a structure for our new superintendent efforts, we after bringing many new ones on the board.”
Update on 29 August: The article was updated to reflect Meta’s AI plans with specific models.
12 Information
