On July 25, 2025, Meta CEO Mark Zuckerberg announced that Shengjia Zhao would be the new Chief Scientist of Meta Superintelligence Labs (MSL). This move shows how fierce the competition in artificial intelligence is getting. This high-profile hire is one of Meta’s biggest wins in the AI talent war. They hired a key architect of OpenAI’s most cutting-edge technologies, such as ChatGPT and GPT-4.
The Man Who Made the Models
Shengjia Zhao’s path to Meta’s superintelligence lab is marked by his important work on some of the most advanced language models in the world. Zhao was a well-known research scientist at OpenAI and played a big role in making ChatGPT, the conversational AI that changed how people think about what AI can do. His work goes far beyond the viral chatbot. He has done important work on GPT-4, the advanced language model that powers many AI apps today.
Zhao is also an expert on OpenAI’s lighter-weight model variations, such as GPT-4.1 and the very advanced o3 model series. Because of these contributions, he is now one of the most knowledgeable people in the world about large language model architecture and training methods. His move from OpenAI to Meta in June 2025, where he quickly rose to the position of chief scientist, shows both how talented he is and how committed Meta is to hiring the best AI experts in the world.
Meta Superintelligence Labs: A New Step in the Growth of AI
Zhao’s hiring is part of Meta’s larger plan to reorganize its business around AI. Zuckerberg announced the creation of Meta Superintelligence Labs in late June 2025. This new unit would bring together all of the company’s different AI projects under one roof. This strategic reorganization puts together Meta’s foundation models team, which includes the open-source Llama software, product development teams, and Fundamental Artificial Intelligence Research (FAIR) projects.
MSL is Zuckerberg’s idea of how to make “personal superintelligence” available to everyone. This lofty goal is beyond what AI can do right now. It aims to make systems that can understand, reason, and help users in ways that come close to or exceed human cognitive abilities across a wide range of tasks. The lab’s goal is to create next-generation models that could completely change the way people use technology and get information.
Alexandr Wang, who is 28 years old and was the CEO of Scale AI, is in charge of the superintelligence unit. He is now Meta’s Chief AI Officer. Wang was hired after Meta bought a 49% stake in Scale AI for $14.8 billion. Scale AI is a data labeling company that is very important for training large-scale AI models. This partnership gives Meta better data processing tools that are necessary for building advanced AI systems.
Strategic Importance in the AI Arms Race
Zhao’s hiring shows how hard it is for big tech companies to find the best AI talent. The move is especially important because Meta and OpenAI are still competing to make better AI systems. By hiring one of OpenAI’s top researchers, Meta has not only improved its own abilities, but it may have also made it harder for a main competitor to do research.
Zuckerberg’s decision to hire shows that he is willing to spend a lot of money on AI talent and infrastructure. Meta’s dedication to AI is clear from its record-breaking plans for capital expenditures. In 2025 alone, the company will spend between $64 billion and $72 billion on building data centers and managing AI workloads. This is a huge jump from the $28 billion spent on capital expenditures in 2023, which shows that the company thinks AI is the future of technology and how people interact with each other.
Research Goals and Technical Vision
Meta Superintelligence Labs is expected to push the limits of what is currently possible in artificial intelligence under Zhao’s leadership. He is uniquely qualified to lead research into multimodal AI systems, better reasoning abilities, and more efficient training methods because he has worked with large language models before. The lab’s research goals probably include making models that can better understand context, remember conversations for longer periods of time, and combine different types of data and media.
Zhao’s background suggests that he is interested in scaling methodologies, which are the methods used to train AI models that are bigger and better. Not only did he make individual models at OpenAI, but he also worked on the methods that make it possible to train systems with hundreds of billions or even trillions of parameters. As Meta tries to make its own frontier models that can compete with the best systems from OpenAI, Google, and Anthropic, this knowledge will be very important.
The appointment also shows that Meta is dedicated to improving the theoretical foundations of AI. Zhao has done both practical model development and basic research on how large neural networks learn and apply what they learn. A superintelligence lab needs people who are good at both short-term product needs and long-term research goals.
Part of Meta’s larger AI strategy
Zhao’s job as chief scientist isn’t just to do research; he also has to figure out how to combine Meta’s current AI projects with his own. The AI community has really taken to the company’s open-source Llama models, and Zhao’s leadership will probably have an effect on how future Llama models are made. His work on both proprietary and open-source model development could help Meta improve how it balances community engagement with competitive advantage.
The work of the superintelligence lab will also have to work with Meta’s products, such as Facebook, Instagram, WhatsApp, and the company’s metaverse projects. As Meta tries to add superintelligence features to its different platforms and services, Zhao’s knowledge of how to use advanced AI systems on a large scale will be useful.
Problems and chances are on the way
Zhao has a lot of opportunities and a lot of problems to deal with as the head of Meta’s superintelligence efforts. On the plus side, he will have access to Meta’s huge computing power, a lot of user data, and a platform that reaches people all over the world. The fact that the company is willing to spend billions on AI infrastructure makes it possible for ambitious research projects that might not be possible anywhere else.
But the problems are just as big. Creating artificial superintelligence is still one of the hardest technical problems humans have ever faced. It will take breakthroughs in many fields of computer science and cognitive science. Zhao has to deal with more than just technical problems; he also has to think about the moral issues surrounding AI safety, fairness, and the effects of AI systems that are getting stronger.
The competitive environment makes things even more complicated. OpenAI, Google DeepMind, and Anthropic are all well-funded competitors of Meta, and each is working on its own way to make advanced AI. Zhao’s success will be measured not only by his scientific achievements but also by how well Meta can turn research breakthroughs into advantages in the market.
The Path to Superintelligence
Zhao’s hiring is a big step forward for Meta on its way to creating artificial superintelligence, but it’s not the end of the road; it’s just the beginning. To get to where we want to go, we need to keep coming up with new ideas in many areas, such as making training algorithms more efficient, coming up with better ways to evaluate AI, making sure AI is safe and aligned, and finally building systems that can really improve human intelligence.
Shengjia Zhao is the chief scientist at Meta Superintelligence Labs and is at the forefront of one of the most important technological efforts of our time. His leadership will not only help Meta’s place in the AI world, but it could also affect the direction that AI development takes in the future. In the next few years, we’ll see if this bold idea of personal superintelligence can go from being an ambitious goal to a life-changing reality.
