Key Moments
- Nvidia introduced the Earth-2 family of open AI weather models and tools at the American Meteorological Society’s Annual Meeting.
- The Earth-2 stack accelerates the full forecasting workflow, from ingesting observational data to generating 15-day global and localized storm forecasts.
- Energy, financial, and meteorological organizations are already deploying or testing Earth-2 models, with some reporting sharply reduced computing time.
Comprehensive AI Stack for Weather and Climate Modeling
Nvidia unveiled a new suite of open models and software tools for AI-driven weather prediction under its Earth-2 platform at the American Meteorological Society’s Annual Meeting on Monday.
The Earth-2 family is described as the world’s first fully open, accelerated weather AI software stack. It is designed to broaden access to advanced forecasting capabilities for researchers, startups, software developers, corporations, and public-sector agencies around the world.
The technology is built to speed up every step of the prediction process – from the initial handling of observational inputs to the creation of 15-day global outlooks and localized severe weather projections. According to Nvidia, this AI-based process can substantially cut computing time and expense relative to traditional physics-based numerical models that typically run on supercomputers.
New Models Within the Earth-2 Suite
Nvidia introduced several additions to the Earth-2 model lineup, each focused on a different aspect of the forecasting pipeline. These new tools complement existing components within the platform.
| Model | Architecture / Engine | Focus / Role |
|---|---|---|
| Earth-2 Medium Range | Powered by Atlas architecture | Supports medium-range forecasts, including up to 15-day global outlooks |
| Earth-2 Nowcasting | Powered by StormScope | Enables short-term, high-frequency weather prediction, including local storm forecasts |
| Earth-2 Global Data Assimilation | Powered by HealDA | Ingests and integrates global observational data for model initialization |
These new offerings are integrated alongside existing models in the Earth-2 stack, including Earth-2 CorrDiff and Earth-2 FourCastNet3.
Early Adoption Across Sectors
Multiple organizations are already working with Earth-2 technology. AI weather platform Brightband is using Earth-2 Medium Range to generate daily global forecasts.
The Israel Meteorological Service has adopted Earth-2 CorrDiff in operational settings and has reported a 90% reduction in computing time when compared with conventional approaches.
Within the energy sector, companies such as TotalEnergies, Eni, GCL, and Southwest Power Pool are either assessing or deploying Earth-2 models to enhance the accuracy and efficiency of their weather-dependent operations.
Financial firms are also turning to the platform. S&P Global Energy and AXA are utilizing Earth-2 technology for applications such as risk analysis and scenario modeling.
Access and Availability
Earth-2 Medium Range and Earth-2 Nowcasting are currently accessible through NVIDIA Earth2Studio, Hugging Face, and GitHub. Nvidia expects Earth-2 Global Data Assimilation to become available later this year.
TradingPedia.com is a financial media specialized in providing daily news and education covering Forex, equities and commodities. Our academies for traders cover 
