Meta Hires Scale AI Founder for New Superintelligence Labs

Mark Zuckerberg just redrew Meta’s org chart—and the global AI arms race—by folding every frontier-model skunkworks into a single unit called Meta Superintelligence Labs (MSL). The freshly minted division will be run by Alexandr Wang, the 28-year-old prodigy who built data-labelling powerhouse Scale AI and now becomes Meta’s first-ever chief AI officer. Zuckerberg’s internal memo, obtained by Business Insider and Bloomberg, calls Wang “the most impressive founder of his generation” and confirms that former GitHub CEO Nat Friedman will co-lead research and product strategy alongside him. 

MSL unites the teams behind Meta AI, the open-source Llama-4 family, FAIR’s long-horizon research, and the Reality Labs group experimenting with AI wearables. The goal: move from today’s billion-parameter chatbots to what Zuckerberg calls “personal superintelligence”—systems that can reason, plan and act autonomously across text, vision and robotics. Wang will control model research, data pipelines, and the compute budget that already tops 600,000 H100-class GPUs, giving Meta one of the world’s three largest AI clusters.

Meta began courting Scale AI last year, ultimately investing $14.3 billion for preferred equity and first-look access to Scale’s data-generation tools. Wang’s jump from CEO to C-suite insider cements that relationship and instantly plugs Meta into a labelling workforce spanning the US, UK, Germany, Norway—and a fast-growing operation in Lagos, Nigeria, that annotates everything from Hausa street signs to Yoruba inflection for speech models. The hire also yanks a coveted talent magnet away from Silicon Valley rivals: Scale was a feeder for Google DeepMind, Anthropic and Tesla’s Autopilot teams. 

Within hours of the announcement, Meta confirmed 11 marquee hires from OpenAI, Anthropic and Google DeepMind. Insiders say the first mandate is to merge Llama’s open-weight lineage with novel agent architectures Friedman has championed in open-source circles. The pair will also explore tighter silicon-to-software integration—leveraging Meta’s in-house MTIA accelerator and next-gen Arm server chips now that Wang’s data-ops expertise is on tap. This has prompted competitors like OpenAI to up their salary game in response to the “poaching” by Meta. 

With OpenAI shipping GPT-4o and Google unveiling Gemini 2, Meta’s open-source Llama strategy looked risky: great for community mindshare, but weaker for moat-building. Superintelligence Labs signals a pivot—keep the open weights coming, but pair them with proprietary agent layers, long-context memory and hardware optimizations only Meta can fund. Analysts at Bernstein say the reorg “puts a CEO-level founder in charge of Meta’s core growth engine,” betting that Wang can do for AGI what Sheryl Sandberg once did for ads.

Zuckerberg closed his memo with characteristic bravado: “We have the compute, the talent and the product reach to make superintelligence available to billions.” If MSL delivers, AI won’t just feel smarter on Facebook, Instagram, WhatsApp and Ray-Ban smart glasses—it could reshape how every person on the planet, from San Francisco coders to Oslo surgeons to Lagos traders, taps digital intelligence. For now, one thing is certain: the race to artificial general intelligence just got a new pit crew—and they’re wearing Meta blue.