**Breaking Down Barriers: NVIDIA’s Earth-2 Revolutionizes Climate Tech with the World’s First Fully Open Accelerated AI Weather Stack**
I’m thrilled to announce that NVIDIA has just changed the game with the release of **Earth-2**, a suite of open models and tools that makes AI-driven climate and weather forecasting accessible to anyone from startups to national meteorological services. Gone are the days of requiring powerful supercomputers and complex physics-based models to predict the weather. With Earth-2, the field has just got a whole lot more interesting.
**The Democratization of Climate Intelligence**
Traditionally, running a high-fidelity climate model required infrastructure that only a handful of countries could afford. NVIDIA’s Earth-2 changes the calculus by providing an ‘open stack’, a set of pre-trained models, inference libraries, and customization recipes available on platforms like GitHub and Hugging Face. This is huge!
Mike Pritchard, Director of Climate Simulation at NVIDIA, emphasizes that NVIDIA is not becoming a climate service provider. Instead, they’re building the “foundational building blocks” that enable nations and companies to build their own sovereign forecasting systems. “Sovereignty matters. Climate is a national security issue… That’s why we’ve built Earth-2, the world’s first fully open production-ready AI climate stack.” – Mike Pritchard, NVIDIA
**Meet the New Heavyweights: Atlas, StormScope, and HealDA**
The announcement introduces three new models that tackle different stages of the forecasting pipeline, from processing messy data to predicting storms weeks in advance.
The **Earth-2 Medium Range** model, powered by **Atlas**, focuses on the 15-day forecast window and predicts over 70 climate variables with high accuracy. Business benchmarks show Atlas outperforms GenCast, the current leading open model, for most variables.
The **Earth-2 Nowcasting** model, powered by **StormScope**, is a game-changer for quick disaster response. It generates hazardous weather predictions in minutes, giving emergency responders crucial time to act.
The **Earth-2 Global Data Assimilation** model, powered by **HealDA**, is a breakthrough in assimilation, which is the process of combining messy satellite and balloon data into a coherent snapshot of the atmosphere. It accomplishes this task in **minutes on GPUs** instead of hours on supercomputers.
**Real-World Impact: From Solar Power to Hurricane Risk**
The Earth-2 stack is already in use by major global players, proving that AI climate forecasting is ready for business and operational prime time. Renewable energy companies, insurance companies, and government agencies are all using Earth-2 to predict solar and wind variability, assess hurricane risk, and more.
**The Bottom Line**
NVIDIA’s Earth-2 is not just a technical upgrade; it’s a structural shift in how humans interact with the climate. By lowering the barrier to entry, shifting from multimillion-dollar supercomputers to accessible GPU-accelerated AI, NVIDIA is enabling a future where hyper-local, high-accuracy climate prediction is ubiquitous.
To learn more about getting started with these models, developers can visit the NVIDIA Earth-2 technical blog. Read the research paper for Earth-2 Medium Range, Earth-2 Nowcasting, and Earth-2 Global Data Assimilation.
