META CEO Mark Zuckerberg announced on Monday the creation of a new AI research group called Meta Superintelligence Labs (MSL), part of a drastic effort to reorganize the company's artificial intelligence initiative and build “personal superintelligence for everyone,” according to an internal memo obtained by Business Insider.
The new lab is led by former scale AI CEO Alexandr Wang, who joined Meta earlier this month after investing $14.3 billion in a data research startup. Former Github CEO and well-known AI investor Nat Friedman will partner with Wang to co-lead MSL's research and product initiatives.
“As AI progress accelerates, the development of super intelligence is starting to be seen,” Zuckerberg wrote in a memo, and Bloomberg first reported. “We believe this will mark the beginning of a new era for humanity, and we are completely committed to doing what Meta needs to lead.”
Zuckerberg emphasized that Meta's AI efforts are not only as capable as humans, but also aim to build “personal” systems, and that people are meant to integrate deeply into their interactions with technology.
This announcement is to compete with Openai and Google in the race to build the most advanced AI systems amid a wider and more aggressive push by meta. AI has become Zuckerberg's top priority this year, fostering large investments in infrastructure such as chips and data centers, and employing recruitment targeting researchers from rival labs.
In addition to the scale of the AI investment, Meta reportedly has been in confusion and buying talks with the runway. The company is also reportedly expected to acquire voice replication startup Playai as part of its push to AI personal assistants.
Read the full notes from Zuckerberg below:
As AI progresses accelerate, close developments are becoming apparent. I believe this will mark the beginning of a new age of humanity, and I am completely committed to doing what Meta needs to lead. Today, I would like to share some details about how we organize our AI efforts to build towards our vision.
Calls organization-wide Meta Superintelligence Labs (MSL). This includes all foundations, products, and fair teams, as well as new labs focusing on developing next-generation models.
Alexandr Wang joined Meta, is the Chief AI Officer and leads MSL. Alex and I have worked together for several years and consider him to be the most impressive founder of his generation. He has a clear sense of the historical importance of superintelligence, and as co-founder and CEO, he built Scaleai into a rapidly growing company involved in developing almost every major model across the industry.
Nat Friedman joins Meta and partners with Alex to lead MSL, leading the work on AI Products and Applied Research. Nat will work with Connor to define his role in the future. He runs Github for Microsoft and recently runs one of the leading AI investment companies. NAT worked for the MetaAdvisory Group last year, so he already has a good sense of our roadmap and what we have to do.
Also, there are some strong new team members participating today, or some have been there in the past few weeks.
- A boring conflict – We have developed the concept of Openai's O series model and the RL of the co-creators.
- shuchao bi – Co-creator of GPT-4o audio mode and O4-Mini. Previously, I was leading the post-training multimodals at Openai.
- Huiwen Chang – Co-creator of image generation for GPT-4o and previously invented the MaskGit and Muse Text-to-Image architectures at Google Research.
- Ji Lin – Helped to build inference stacks for O3/O4-MINI, GPT-4O, GPT-4.1, GPT-4.5, O4-ImageGen, and operators.
- Joel Pover – Human reasoning. Previously, he had 11 years at Meta with HHVM and 11 years at Hacks, Flows, Responses, Performance Tools, and Machine Learning.
- horny – Co-creator of GPT-4O, 4o-Mini, O1-Mini, O3-Mini, O3, O4-Mini. Previously, he led a post-training group at Openai.
- Johann Scharkwick – Former Google Fellow, early contributor to Sesame, and technical lead of Maya.
- Paysan – After Gemini training in Google Deepmind, coding, and reasoning. I previously created the last two generations of Waymo's perceptual model.
- Jiahui Yu – Co-creators of O3, O4-MINI, GPT-4.1 and GPT-4O. Previously, he led the recognition team at Openai and the co-led multimodal at Gemini.
- Shengjia Zhao – Co-creators of CHATGPT, GPT-4, all mini models, 4.1 and O3. Previously, I used to derive synthetic data in Openai.
I'm excited about the progress I've planned for Llama 4.1 and 4.2. These models are used by over 1 billion people active per month across the app, and are used by increasing agents across the meta that helps improve products and technology. We are committed to continuing to build these models.
In parallel, we will begin researching next-generation models to reach the frontier next year or so. Over the past few months I have met the top people of Meta, other AI labs and promising startups to bring together the founding group for this little talent strong effort. We are still forming this group. We ask several people across the AI organization to join this lab as well.
Meta is uniquely positioned to provide something close to the world. We have a strong business that supports building significantly more calculations than small labs. We are experienced in building and cultivating products that reach billions of people. We are pioneering and lead the very rapidly growing category of AI glasses and wearables. And our structure allows us to move with great confidence and boldness. I am optimistic that this influx of new talent and parallel approach to model development will be set up to fulfill the personal emergency promises for everyone.
Stay tuned as there are even more amazing people at all levels that will be taking part in this effort in the coming weeks. I'm excited to jump in and get into work.
Are you a current or former meta employee with more insights? Do you have a tip? Contact Hugh Langley safely at Hughlangley.01 signal, or hlangley@businessinsider.comor pranav dixit or on the signal +1408-905-9124 pranavdixit@protonmail.com. Use non-working devices.
