SANTA CLARA, Calif. — January 2026 — NVIDIA has released a set of open Earth-2 models, expanding access to AI-powered tools designed to improve climate, weather, and environmental forecasting.
The new models are part of Earth-2, NVIDIA’s initiative to build a high-resolution, digital twin of the planet using accelerated computing and artificial intelligence. By making these models openly available, NVIDIA aims to help researchers, governments, and enterprises develop more accurate and localized predictions of extreme weather and climate impacts.
The Earth-2 open models use deep learning to simulate atmospheric and environmental dynamics at far higher resolution than traditional numerical models, while running significantly faster. NVIDIA said this allows users to generate detailed forecasts for events such as floods, heatwaves, storms, and wind patterns with lower computational cost and energy use.
The models are designed to integrate with existing scientific workflows and can be customized for regional studies, infrastructure planning, and disaster preparedness. NVIDIA highlighted use cases ranging from early warning systems and renewable energy optimization to urban planning and climate risk assessment.
According to the company, open access is key to accelerating innovation in climate science. By sharing Earth-2 models with the broader ecosystem, NVIDIA said it hopes to enable collaboration across academia, industry, and the public sector, while lowering barriers to advanced climate modeling.
The release underscores NVIDIA’s broader strategy of applying AI and accelerated computing to some of the world’s most complex challenges. With Earth-2, the company positions digital twins and AI-driven simulation as foundational tools for understanding and responding to climate change at global and local scales.
Join the conversation! Your thoughts help the community grow.