Mark Zuckerberg is creating new Applied AI engineering company, reorganises key teams

Meta CEO Mark Zuckerberg is establishing a new applied AI engineering organization to accelerate the company's push toward superintelligence. The move splits Meta's AI work into specialized teams, each with distinct mandates designed to move faster and stay agile as the company scales its artificial intelligence ambitions.The new organization will be led by Maher Saba, a long-time Reality Labs executive, and will report directly to Chief Technology Officer Andrew Bosworth. According to The Wall Street Journal, the group will have an unusually flat structure—up to 50 individual contributors for every manager—to speed decision-making and reduce bureaucracy.Saba's team will build what the company internally calls "the data engine that helps our models get better, faster." The organization will focus on data processing, tooling, and model evaluations that feed into Meta's broader AI research efforts. The approach reflects Zuckerberg's emphasis on infrastructure as a competitive advantage in the AI race.This comes months after Meta's chief AI officer Alexandr Wang announced that the company is laying off approximately 600 positions from its Meta Superintelligence Labs division. In a memo sent to staff members at that time, Wang wrote, "By reducing the size of our team, fewer conversations will be required to make a decision, and each person will be more load-bearing and have more scope and impact." New unit joins Superintelligence Labs in restructured approachThis restructuring marks a shift from the consolidated AI model Meta adopted when it hired Alexandr Wang in June to lead Meta Superintelligence Labs. Now, different teams—Wang's research lab, Saba's applied engineering organization, and Bosworth's broader technology strategy—will share responsibility for developing Meta's next-generation AI models, including projects code-named Avocado and Mango.The flat structure Saba is building aligns with a philosophy Zuckerberg outlined on recent earnings calls: "elevating individual contributors and flattening teams" to move faster and keep agility across the organization. Only two researchers left Wang's roughly 100-person team when their equity vested in November, suggesting stability within that division.Building redundancy into Meta's AI betBy distributing AI work across multiple specialized teams and leaders, Meta is building organizational redundancy into its superintelligence push. The approach ensures that if one team hits friction, others can continue advancing. It mirrors how Zuckerberg has structured other major company initiatives—spreading risk across multiple leaders rather than concentrating all authority in a single executive. The strategy signals Meta's commitment to scaling its AI operations beyond a single leader's oversight.