#USA #Sunnyvale #AI Lab #MBZUAI #Foundation Models

Expansion of MBZUAI in the Technological Frontier

On May 24, 2025, the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) made significant strides in its global presence by launching its Institute of Foundation Models (IFM) in Silicon Valley. Located in Sunnyvale, California, this new lab marks a pivotal expansion following previously announced facilities in Paris and Abu Dhabi.

Significance of the Launch

The inauguration was held at the Computer History Museum in Mountain View, California, symbolizing the establishment of a key hub in MBZUAI’s global research network. This development links the university with California's thriving ecosystem of AI researchers, startups, and tech enterprises, fostering collaboration and innovation.

For the United Arab Emirates (UAE) and MBZUAI, this initiative represents a strategic component of the nation's long-term economic diversification agenda. By investing in cutting-edge technologies such as advanced AI foundation models, the UAE continues to build knowledge-based sectors to support its overarching economic and social transformation efforts.

Professor Eric Xing, President and Professor of the University, commented, "The launch of the IFM is a significant step toward collaboration and the global advancement of state-of-the-art AI foundation models. Our expansion into Silicon Valley is essential for enhancing our presence in one of the world's most dynamic AI ecosystems. We are creating pathways for knowledge exchange with leading institutions while tapping into a talent pool equipped to translate research into practical applications.”

Engaging the Global AI Community

The launch event attracted representatives from top AI companies and academic institutions worldwide, underscoring the growing interest in MBZUAI's global research strategy. Central to MBZUAI's demonstrations was PAN, a world model that facilitates unlimited simulations of diverse realities, ranging from basic physical interactions to complex agent scenarios. Unlike prior systems primarily focused on generating text, audio, or images, PAN predicts comprehensive world states by incorporating multimodal inputs such as language, video, spatial data, and physical actions. This capability enables complex reasoning, strategic planning, and nuanced decision-making for applications spanning autonomous driving to robotics.

The innovative hierarchical architecture of PAN supports multi-layered reasoning and real-time interactions within simulations while ensuring high accuracy even in extensive scenarios. Its counterpart, the PAN Agent, showcases its utility in multimodal reasoning tasks, such as mathematics and programming, in dynamic simulated environments.

Development of Advanced Foundation Models

The IFM lab is also committed to developing two flagship AI systems—K2 and JAIS—that exemplify the university’s dedication to advancing pioneering-class foundation models. An upcoming update for K2-65B will focus on delivering groundbreaking reasoning capabilities with consistent performance. JAIS stands as the most advanced Arabic large-language model and will enhance its capabilities at the IFM by improving language support and adding more context, thereby preserving and promoting the cultures it serves.

Promoting Transparency and Open Development

MBZUAI has established one of the industry's most transparent approaches to AI development, offering not only models but entire developmental processes for public access. This positions the IFM as a leader in open development. The LLM360 initiative provides researchers with comprehensive materials, including training code, datasets, and model checkpoints. This openness is complemented by safety measures such as international advisory boards and peer-review processes, which uphold the integrity of the research. The structure of the IFM comprises specialized teams focusing on model architecture, training methodologies, evaluation frameworks, and security systems, blending the agility of a startup with the resources of a well-established research institution.

Conclusion

As the MBZUAI continues to innovate and expand its capabilities, it is poised to make substantial contributions to the global AI landscape. This initiative not only enhances the university's research capabilities but also plays a critical role in promoting AI development in the UAE, further solidifying its position as a global hub for technology and innovation.