Gordon Bell Winner: Forecasting Tsunamis in Real Time With Digital Twins | NVIDIA GTC

The Cascadia subduction zone is capable of unleashing a magnitude 9 earthquake with tsunamis as high as 30 meters, endangering millions in the Pacific Northwest. While it has been quiet since 1700, scientists place a 37% probability of an 8.2+ earthquake in the next 50 years. In response, the 2025 Gordon Bell Prize was awarded to UT-Austin, UC San Diego, and Lawrence Livermore National Laboratory for the world’s first digital twin for real-time, full-physics-based tsunami early warning. Leveraging exascale-class HPC systems, it delivers unprecedented accuracy and speed, achieving a ten-billion-fold performance improvement over existing state-of-the-art methods. This enables rapid, data-driven hazard assessment and actionable insights during critical windows. The work not only redefines the boundaries of real-time tsunami forecasting, but also sets a new standard for disaster resilience. Omar Ghattas | Professor and Cockrell Endowed Chair in Engineering | The University of Texas at Austin Stefan Henneking | Research Associate | The University of Texas at Austin Key Takeaways: With advanced algorithms and exascale-class computing, digital twins can achieve both full physics accuracy and real-time response. Billion-parameter inverse problems (inferring large-scale model parameters from observational data) can be solved in a fraction of a second. Industry: Academia / Higher Education Topic: Simulation / Modeling / Design - Climate / Weather / Ocean Modeling Technical Level: Technical - Beginner Intended Audience: Research: Non-Academic NVIDIA Technology: Grace CPU, Hopper #nvidiagtc