In 2024, UT Austin launched the Center for Generative AI with a Texas-sized GPU computing cluster hailed as one of the largest in academia. The university recently experienced another first—the addition of a next-generation NVIDIA GB200-NVL72 system and the first set of Blackwells to be up and running in academia.
The new hardware will be a game-changer to expand cutting-edge AI research at UT Austin, supporting a wide range of research areas, such as natural language processing to multimodal applications that integrate language, vision, and speech. The architecture is purpose-built for complex tasks like protein folding predictions and advanced drug discovery. The system will support research out of UT’s NSF AI Institute for Foundations of Machine Learning (IFML).
“The Blackwell system represents a quantum leap in computational power,” says Adam Klivans, Director of IFML. “This is a next-level system designed for large-scale generative AI and accelerated computing that is an important first for a university campus.”
This acquisition of the Blackwell system, combined with the recent renewal of an NSF grant for IFML’s work, solidifies UT as a powerhouse for AI research in academia. NSF’s renewed support is a testament to the importance of foundational research in a rapidly evolving landscape – and the synergy between foundational research and practical application is what sets UT Austin apart.
“These processors will help us scale up the advanced multimodal diffusion models that we are developing for images and language,” says Sanjay Shakkottai, Director of the Center for Generative AI. “They will provide us with exciting new capabilities that can immediately impact industry practice.”
Housed at the Texas Advanced Computing Center (TACC), the Blackwell system is part of a phased expansion to support cutting-edge AI research at UT Austin. The platform consists of 72 interconnected NVIDIA Blackwell GPUs, which undergird groundbreaking advancements for generative AI at UT. The partnership between the Center for Generative AI and TACC ensures that UT faculty and student researchers have access to the state-of-the-art resources needed to push the boundaries of AI innovation.
Those innovations have included 25 groundbreaking research projects spanning generative AI, large language models, computer vision, and AI-driven scientific discovery. These projects are driving the next wave of AI breakthroughs with real-world impact, from revolutionizing healthcare and scientific modeling to redefining human-AI collaboration. These research projects span a diverse range of domains, from computer vision and generative AI, to AI for healthcare and scientific discovery. These efforts tackle fundamental challenges in language models, video and image synthesis, reinforcement learning, and AI efficiency. They include:
- RGB-D Video Generation and Zero-Shot Layout-to-Image Generation, which push the creative boundaries of generative AI
- The Surgeon-AI Collaboration Platform, an example of AI for medical decision-making, which has the potential to save lives by improving organ transplantation outcomes
- Research in memory-efficient LLMs and low-energy inference ensures that AI remains accessible, scalable, and environmentally responsible
These interdisciplinary initiatives showcase the power of AI to revolutionize industries, enhance human-AI collaboration, and shape the future of intelligent systems.