The Most Expensive Hire In AI History Finally Talks

Alex Wang, head of Meta Superintelligence Labs, details the strategic shift toward building frontier AI models from scratch within a nine-month timeframe. The lab prioritizes high compute-per-researcher ratios and talent density to accelerate development, moving away from previous, less focused trajectories. Wang emphasizes the "personal superintelligence" mission, aiming to empower billions of users and businesses through agentic ecosystems that handle complex, long-running tasks. Despite external perceptions of a mercenary approach to recruiting, the team culture focuses on ambitious research bets and technical rigor. While the recently released MuseSpark model serves as an early data point in their scaling ladder, future iterations are expected to achieve state-of-the-art performance. The organization remains committed to open-source principles, provided safety guardrails are met, while simultaneously exploring physical intelligence and robotics to integrate AI into the real world.