NextFin news, the Gordon Bell Prize, awarded annually by the Association for Computing Machinery (ACM) to recognize outstanding achievements in high-performance computing (HPC), was awarded in November 2025 to a research team spearheaded by the Max Planck Institute for Meteorology (MPI-M) and the German Climate Computing Center (DKRZ) for their ICON Earth system simulation. The award ceremony took place at the Supercomputing Conference (SC25) held in St. Louis, Missouri.
The winning project achieved an unprecedented full-Earth simulation at a horizontal grid resolution of 1.25 kilometers, capturing intricate interactions across atmosphere, oceans, land, and the carbon cycle. Leveraging NVIDIA GH200 Grace Hopper Superchips, the simulation ran on Europe's leading exascale supercomputers, JUPITER in Germany and Alps in Switzerland. These systems provided the computational throughput to simulate approximately 146 days of Earth system dynamics every day of runtime, utilizing 85% of JUPITER’s resources with high energy efficiency, reducing consumption by a factor of four compared to conventional CPU-only setups.
Key contributors to the project included representatives from MPI-M, DKRZ, the Jülich Supercomputing Centre, ETH Zurich, the Swiss National Supercomputing Centre (CSCS), the University of Hamburg, and NVIDIA. Their collaboration integrated advanced data-centric optimizations and a heterogeneous CPU-GPU architecture, critical for balancing the diverse Earth system components modeled by ICON.
The ICON model's capability to depict climate processes at kilometer scale addresses a long-standing HPC challenge, previously considered computationally prohibitive. This leap in spatial resolution enables the accurate representation of localized climatic phenomena, including energy, water, and carbon fluxes, crucial for understanding future climate impacts on ecosystems and human societies.
On a parallel track, other finalists in the Gordon Bell Prize competition also utilized NVIDIA-powered supercomputers to drive frontier research in climate modeling, fluid dynamics, nanoscale electronic device simulation, and real-time disaster prediction, underscoring NVIDIA's integral role in advancing HPC-enabled scientific discovery.
From a technical perspective, the ICON project exemplifies how exascale computing platforms—powered by NVIDIA's unified virtual memory, CUDA-X libraries, and GPU acceleration—can empower simulations that compress decades of climate data into manageable computational timelines. The realized throughput and energy efficiency demonstrate not only raw performance gains but also a strategic approach towards sustainable HPC practices.
Scientifically, this milestone holds profound implications. Detailed simulations at kilometer scale yield unprecedented granularity in prognostic climate data, facilitating refined policy models addressing climate adaptation and mitigation. Integration of atmospheric, oceanic, terrestrial, and biogeochemical subsystems in real-time encourages interdisciplinary research and decision-making based on comprehensive Earth system dynamics.
Economically and geopolitically, accelerated and accurate climate modeling bolsters national and international preparedness. Enhanced predictive capabilities offer governments, such as the United States under President Donald Trump's administration, improved tools for resource allocation, disaster risk management, and infrastructure planning amid increasing climate change pressures.
The trajectory set by ICON and similar HPC climate initiatives suggests a rapidly intensifying race toward kilometer-scale and sub-kilometer environmental simulations. Continual hardware advancements, particularly in GPU architectures, alongside breakthroughs in AI-driven HPC optimization, indicate that future models will achieve even higher fidelity with lower operational costs, broadening accessibility beyond elite research labs.
This evolving HPC-climate nexus refines the competitive landscape for supercomputing vendors, where NVIDIA’s leadership in GPU-accelerated exascale systems cements its critical position. Research teams keen on addressing climate crises will likely pursue collaborations leveraging these HPC capabilities to unlock actionable insights faster and more efficiently.
In conclusion, the awarding of the 2025 Gordon Bell Prize to the ICON team signifies an inflection point—where cutting-edge supercomputing not only transcends performance frontiers but also fundamentally strengthens humanity’s ability to understand and respond to global climate change with unprecedented precision and speed.
According to HPCwire, the full Earth system simulation by the ICON team achieved a world record throughput and has paved the way for next-generation climate science powered by NVIDIA technology.
Explore more exclusive insights at nextfin.ai.
