For decades, weather forecasting has been the exclusive domain of giant government supercomputers running complex physics-based equations. NVIDIA has broken that barrier with the release of Earth-2 family of open models and tools AI for weather and climate prediction is accessible to almost anyone, from tech startups to national meteorological agencies.
In a step that democratizes climate science, NVIDIA unveiled 3 groundbreaking new models powered by innovative architecture: atlas, stormscopeAnd Hilda. These tools promise to accelerate the speed of forecasting by orders of magnitude while providing accuracy that rivals or even exceeds traditional methods.
Democratization of weather intelligence
Historically, running high-fidelity weather models required infrastructure that only a few countries could afford. NVIDIA’s Earth-2 transforms calculus by offering an ‘open stack’, a collection of pre-trained models, estimation libraries, and optimization recipes available on platforms like GitHub and Hugging Face.
Mike Pritchard, director of climate simulation at NVIDIA, emphasized that NVIDIA is not becoming a weather service provider. Instead, they are creating “foundational building blocks” that allow nations and companies to create their own sovereign forecasting systems.
“Sovereignty matters. Weather is a national security issue… That’s why we built Earth-2, the world’s first fully open production-ready AI weather stack.” – Mike Pritchard, NVIDIA
Meet the new giants: Atlas, Stormscope and Hilda
The announcement introduces 3 specific models that address different stages of the forecasting pipeline, from processing dirty data to predicting storms weeks in advance.
1. Earth-2 Medium Range (Powered by Atlas)
Targeting a 15-day forecast window, this model uses a new architecture called atlas. It predicts over 70 weather variables including wind, humidity and pressure at high accuracy.
- Display: On standard industry benchmarks, Atlas has been shown to outperform gencastCurrent leading open model, across most variables.
- the shift: It represents a return towards “simple, scalable Transformer architectures”, moving away from niche, hand-crafted AI design.
2. Earth-2 Nowcasting (Powered by StormScope)
This is a game-changer for immediate disaster response. Powered by stormscopeThis generative AI model focuses on a 0 to 6 hour window, providing kilometer-scale resolution of local storms.
- why it matters: It is the first AI model to outperform traditional physics-based methods for short-term rainfall forecasting.
- pace: It predicts dangerous weather within minutes, giving emergency responders critical time to take action.
- Sovereignty: Because it is trained directly on geostationary satellite imagery rather than on region-specific physics output, it can be deployed by any country with good satellite coverage.
3. Earth-2 Global Data Assimilation (Powered by HealDA)
Often the unsung hero of forecasting, “data assimilation” is the process of combining messy satellite and balloon data into coherent snapshots of the atmosphere to generate a forecast.
- the breakthrough: Traditional assimilation consumes about 50% of supercomputing cycles. nvidia Hilda Architecture fulfills this function minutes on gpu Instead of hours on a supercomputer.
- Result: When combined with medium range models, it produces the most efficient predictions ever seen from a completely AI-based pipeline.
Real-world impacts: from solar power to hurricane risk
The Earth-2 stack is already in use by major global players, proving that AI weather forecasting is ready for commercial and operational prime time.
- Renewable energy: total energy And gcl (a major solar materials producer) is using Earth-2 to predict solar and wind variability. For solar farms, accurate cloud cover predictions can have a significant impact on energy market trading.
- Israel Meteorological Service: Using the CorrDiff model (part of the Prithvi-2 family), they have acquired a 90% reduction in computation time Generating high-resolution forecasts up to eight times per day.
- Insurance and Risk: axa And S&P Global Energy Taking advantage of Earth-2’s speed to run thousands of “counterfactual” scenarios. By simulating thousands of years of hypothetical storm data, they can better understand rare, high-impact climate events that haven’t happened yet but could happen.
- daily operations: brightbandThe AI weather tool provider is already integrating Earth-2 Medium Range to issue daily global forecasts.
bottom line
NVIDIA Earth-2 isn’t just a technological upgrade; This is a structural change in the way humans interact with the climate. By lowering the barrier to entry, moving from multimillion-dollar supercomputers to accessible GPU-accelerated AI, NVIDIA is enabling a future where hyper-local, high-accuracy weather forecasting is ubiquitous.
As extreme weather events become more frequent, tools like stormscope And atlas Will likely become essential infrastructure for governments and industries around the world.
Earth-2 Medium Range and Nowcasting are available on GitHub, Hugging Face, and NVIDIA Earth2Studio. Earth-2 global data assimilation is expected to be released later this year.
To learn more about getting started with these models, developers can visit the NVIDIA Earth-2 technical blog. Prithvi-2 medium range [Read the research paper], Earth-2 Nowcasting [Read the research paper]And Earth-2 Global Data Assimilation [Read the research paper].
Jean-Marc is a successful AI business executive. He leads and accelerates development of AI driven solutions and started a computer vision company in 2006. He is a recognized speaker at AI conferences and holds an MBA from Stanford.
