NVIDIA Unveils Earth-2 Digital Twin to Accelerate Global Climate Simulation
NVIDIA launches the Earth-2 cloud platform to combat extreme weather economic losses through high-speed AI climate modeling. The system delivers interactive forecasts in seconds instead of hours using a breakthrough generative AI model.
NVIDIA introduces the Earth-2 climate digital twin cloud platform to help organizations combat the massive economic losses caused by extreme weather. Built on the NVIDIA DGX Cloud, this new system provides cloud APIs that allow virtually any user to create AI-powered emulations of global atmospheric conditions, local cloud cover, and typhoons at an unprecedented scale.
The platform features a groundbreaking generative AI model named CorrDiff, which produces images with 12.5x higher resolution than traditional numerical models. This diffusion model operates 1,000 times faster and 3,000 times more energy efficiently than legacy CPU-driven systems, enabling users to deliver critical weather warnings and updated forecasts in seconds rather than minutes or hours.
Organizations like the Central Weather Administration of Taiwan plan to leverage these diffusion models to predict precise typhoon landfall locations and improve early evacuation protocols. By combining Earth-2's advanced APIs with proprietary climate tech data, users globally gain the ability to rapidly synthesize fine-scale local weather metrics and better prepare for severe climate events.