Meta reshuffles teams for applied AI

Meta’s AI race is turning into an execution story as much as a research story. This reorg suggests Zuckerberg believes faster data engines, better evals, and flatter teams can compound model progress just as much as big-name hires.

• –The new org appears designed to feed Meta’s research teams with better data processing, tooling, and evaluation loops
• –Putting Maher Saba under Bosworth creates a separate execution lane alongside Meta’s Superintelligence Labs rather than folding everything into one AI org
• –The unusually flat structure points to a deliberate attempt to cut management drag and speed internal decision-making
• –For AI developers, the signal matters because future Meta models will likely be shaped by stronger infrastructure and iteration discipline, not just research breakthroughs