How NVIDIA's Earth-2 uses AI to Accurately Predict Weather

2026-02-10 · Source: AI Magazine · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Environmental Science & Earth Systems · Depth: Intermediate, quick

Summary

NVIDIA has launched Earth-2, an open-access, production-grade, and accelerated weather AI software stack designed to improve forecasting accuracy and speed. This initiative applies AI across the entire weather forecasting pipeline, from generating current atmospheric conditions to predicting weather weeks in advance. Earth-2 combines pretrained models, development frameworks, customization workflows, and inference libraries, making advanced forecasting tools accessible to researchers, public agencies, and enterprises. By leveraging GPUs instead of traditional CPU clusters, Earth-2 significantly reduces processing times from hours to seconds. The suite includes models like Earth-2 Medium Range for 15-day forecasts, Earth-2 Nowcasting for zero- to six-hour local storm predictions, and Earth-2 Global Data Assimilation for initial atmospheric conditions. Early adopters, including the Israel Meteorological Service and TotalEnergies, report substantial reductions in compute requirements and improved forecast accuracy.

Key takeaway

For research scientists and meteorological services focused on climate resilience and operational forecasting, Earth-2 offers a critical shift by providing an open, GPU-acceleraccelerated AI stack. You should explore integrating Earth-2 models like Medium Range and Nowcasting into your infrastructure to achieve significant reductions in compute time and enhance forecast accuracy, enabling faster, more informed decisions in an unpredictable climate.

Key insights

NVIDIA Earth-2 provides an open, GPU-accelerated AI stack for faster, more accurate weather and climate forecasting.

Principles

• AI accelerates weather prediction.
• Open access fosters collaboration.
• GPU processing reduces compute time.

Method

Earth-2 uses a modular AI architecture to accelerate every forecasting stage, from data assimilation to short- and medium-range prediction, running on GPUs for rapid processing.

In practice

• Deploy Earth-2 for 15-day atmospheric forecasts.
• Use Nowcasting for kilometer-scale storm predictions.
• Integrate CorrDiff to downscale global data faster.

Topics

• NVIDIA Earth-2
• Weather Forecasting AI
• AI Software Stack
• GPU Acceleration
• Climate Modeling

Best for: Research Scientist, AI Engineer, AI Scientist, Domain Expert

Related on AIssential

• NVIDIA Launches Earth-2 Family of Open Models — the World’s First Fully Open, Accelerated Set of Models and Tools for AI Weather — NVIDIA's Earth-2 provides an open, accelerated AI software stack for comprehensive weather and climate prediction.
• From Materials Simulation to Experimental Astronomy, New NVIDIA AI Software Unlocks Scientific Discoveries — NVIDIA's new GPU-accelerated software dramatically speeds scientific data processing and simulation across diverse research domains.
• Utilizing AMD Instinct GPU Accelerators for Weather and Precipitation Forecasting with NeuralGCM — NeuralGCM combines GCMs with ML for stable, accurate, and physically consistent weather and precipitation forecasts.
• AI Processing of Earth Images Can Now Run in Space — Onboard AI processing on satellites significantly reduces data latency for Earth observation, enabling real-time insights.
• Climate Physicists Face the Ghosts in Their Machines: Clouds — Cloud modeling is a critical bottleneck in climate prediction, driving new AI-physics hybrid approaches.
• Build Accelerated, Differentiable Computational Physics Code for AI with NVIDIA Warp — NVIDIA Warp accelerates GPU-native physics simulations and spatial computing with Python, enabling differentiable workflows.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI Magazine.