Meta is reorganizing a major part of its AI work under a new banner: Meta Superintelligence Labs. In a memo obtained by WIRED, Mark Zuckerberg introduced the group to Meta staff and laid out a roster of recently hired AI researchers and engineers from rival companies.
The move brings together existing Meta AI teams with a new lab focused on the next generation of models. It also places several high-profile hires into a single organization built around foundations, products, FAIR and applied AI research.
What Meta Superintelligence Labs Includes
Zuckerberg wrote in the memo on Monday that the overall organization will be called Meta Superintelligence Labs, or MSL. According to the memo, MSL includes Meta's foundations, product and FAIR teams, along with a new lab dedicated to developing the next generation of models.
The new structure follows a period in which Zuckerberg has been recruiting heavily for AI talent. The source article says Meta invested $14.3 billion in Scale AI and hired Alexandr Wang, Scale AI's CEO, to run Meta's Superintelligence Labs.
Wang was introduced as Meta's chief AI officer and the leader of MSL. Zuckerberg also introduced former GitHub CEO Nat Friedman, who will colead the new lab with Wang. Friedman's focus will be AI products and applied research.
Meta declined to comment, according to the source article. News of the memo was first reported by Bloomberg.
The Hiring Push Behind MSL
The memo lists recently hired employees, many of whom came from competing AI firms including OpenAI, Anthropic and Google. The list is notable because it shows Meta hiring people whose prior work touched reasoning, multimodal systems, post-training, image generation, voice mode, inference, coding and perception.
The memo's list does not include employees who joined from OpenAI's Zurich office. That means the names disclosed in the memo are not presented as the complete set of all AI hiring connected to Meta's recent push.
Still, the roster gives a clear picture of the technical areas Meta is trying to strengthen. Several hires are associated with model post-training and reasoning. Others are linked to multimodal systems, image generation, voice mode, inference infrastructure and perception models.
Who Zuckerberg Listed In The Memo
The hires named in the memo include people with backgrounds at OpenAI, Anthropic, Google, Google Research, Google Deepmind and DeepMind, as described in the source article.
- Trapit Bansal: worked on RL on chain of thought and was listed as a cocreator of o-series models at OpenAl.
- Shuchao Bi: was described as a cocreator of GPT-4o voice mode and o4-mini, and previously led multimodal post-training at OpenAl.
- Huiwen Chang: was listed as a cocreator of GPT-4o's image generation and previously invented MaskIT and Muse text-to-image architectures at Google Research.
- Ji Lin: helped build 03/o4-mini, GPT-4o, GPT-4.1, GPT-4.5, 40-imagegen and Operator reasoning stack.
- Joel Pobar: worked on inference at Anthropic and previously spent 11 years at Meta on HHVM, Hack, Flow, Redex, performance tooling and machine learning.
- Jack Rae: was pre-training tech lead for Gemini and reasoning for Gemini 2.5, and led Gopher and Chinchilla early LLM efforts at DeepMind.
The remaining names in the memo broaden the pattern. Hongyu Ren was listed as a cocreator of GPT-4o, 4o-mini, o1-mini, o3-mini, 03 and o4-mini, and previously led a group for post-training at OpenAl. Johan Schalkwyk was described as a former Google Fellow, early contributor to Sesame and technical lead for Maya.
Pei Sun's background includes post-training, coding and reasoning for Gemini at Google Deepmind, as well as creating the last two generations of Waymo's perception models. Jiahui Yu was listed as a cocreator of 03, 04-mini, GPT-4.1 and GPT-4o, and previously led the perception team at OpenAl and co-led multimodal at Gemini. Shengjia Zhao was described as a cocreator of ChatGPT, GPT-4, all mini models, 4.1 and 03, and previously led synthetic data at OpenAl.
Why The Structure Matters
The memo frames MSL as more than a hiring announcement. It describes an organizational home for multiple AI efforts inside Meta, including research, products and model development.
That matters because the named hires are not clustered around a single narrow specialty. The list spans model training, reasoning, post-training, multimodal work, image generation, voice mode, perception, inference and applied research. In practical terms, Meta is assembling people whose earlier work touched many layers of modern AI systems.
The leadership structure also shows two priorities side by side. Wang is positioned as chief AI officer and leader of MSL. Friedman is set to colead the new lab with an emphasis on AI products and applied research. Together, that suggests the new group is meant to connect model development with product-facing work, using the teams Zuckerberg named in the memo.
For now, the facts available from the memo are limited to the structure, leadership and listed hires. But the message to Meta staff makes clear that the company is putting its AI organization under the MSL name and building that organization with talent from some of the most prominent AI teams in the industry.