Microsoft’s New Aurora AI Brings Smarter, Teamwork-Powered Weather and Climate Forecast

Microsoft announced a groundbreaking move to open up new possibilities for climate science and meteorology: “The Next Phase of Aurora.” This ambitious initiative aims to make Aurora—Microsoft’s large-scale AI model for weather and climate forecasting—more accessible, collaborative, and impactful. By opening up datasets, models, and tools to the global research community, Microsoft is not only advancing forecasting accuracy but also actively responding to the urgent challenges of climate change.​

What Is Aurora and Why Does It Matter?

Aurora is a highly sophisticated artificial intelligence system, with over 1.3 billion parameters, designed to model atmospheric processes across Earth. It serves as a “foundation model” for Earth system science, capable of learning from vast, diverse datasets of satellite images, radar data, and simulated forecasts accumulated over more than a million hours. Unlike traditional weather models, Aurora uses deep learning to rapidly generate high-resolution forecasts—making predictions in seconds that used to require hours on a supercomputer.​

• Extreme accuracy: Aurora has outperformed leading traditional and AI weather models on over 90% of standard forecasting targets, including predicting hurricanes, cyclones, and air pollution with improved accuracy.​

• Efficiency and accessibility: By leveraging cloud computing via Azure, Aurora delivers this performance at a fraction of the computational cost, opening the door for broader adoption—even in regions with limited resources.​

Breaking Down the Barrier: Open Science and Collaboration

One of the most exciting aspects of Aurora’s next phase is Microsoft’s commitment to openness and collaboration. According to Microsoft Research, the models and datasets that underpin Aurora are being made available to scientists, meteorologists, and developers worldwide. This means:​

• Open Access: Researchers can download Aurora’s AI code and model weights, run their own forecasts, or fine-tune the models for specific regions and use-cases.​

• Global Collaboration: Microsoft is partnering with major weather agencies and organizations—including NOAA, ECMWF, and the UK Met Office—to evaluate and refine Aurora in operational settings.​

• Community Innovation: By lowering barriers to entry, Microsoft hopes the broader scientific community will accelerate innovations in both weather and climate prediction, disaster preparedness, and environmental monitoring.​

How Aurora Works: The Power of Diverse Data and Flexible AI

Aurora’s breakthrough capabilities stem from its ability to learn from a broader, more diverse set of data than any previous model:

• Over one million hours of training data: Including climate simulations, observational reanalyses, and operational weather predictions, resulting in a uniquely robust understanding of atmospheric dynamics.​

• High spatial resolution: Aurora can generate forecasts down to 0.1º resolution (about 11 km at the equator), capturing details crucial for events like cyclones or flash floods.​

• Flexible architecture: The model can be fine-tuned for specialized tasks—such as forecasting rainfall in India, predicting air quality in California, or tracking ocean wave patterns—by training on smaller, task-specific datasets.​

Microsoft claims that Aurora produces a 10-day global weather forecast or a 5-day air pollution forecast in just seconds on a single GPU. This efficiency is a game-changer for scaling climate prediction and operational forecasting, especially in data-sparse regions where resources are limited.​

Real-World Impact: From Storm Prediction to Health and Agriculture

The real-world results speak for themselves. Aurora has:

• Predicted Typhoon Doksuri’s landfall in the Philippines four days in advance, outperforming expert forecasts.​

• Outperformed the U.S. National Hurricane Center in tracking tropical cyclone paths and proved better in forecasting severe sandstorms in the Middle East.​

• Delivered highly accurate air quality forecasts, which are now incorporated into Microsoft’s MSN Weather for consumer use.​

These advances mean improved disaster preparedness, enhanced resilience for communities at risk of extreme weather, and better planning for sectors like agriculture and energy.​

Why Open AI like Aurora Matters for the Climate Crisis

The global stakes of climate change make collaboration essential. By open-sourcing Aurora and inviting community input, Microsoft is aiming to:

• Democratize access to the best AI tools in atmospheric science, helping more regions benefit from cutting-edge research.​

• Foster global scientific cooperation, enabling rapid improvements and local adaptations of the model.

• Accelerate progress on the world’s biggest environmental challenges, from early warning systems for hurricanes and floods, to tracking long-range air pollution and supporting climate-resilient development.

This open-science approach is a shift away from closed, proprietary forecasting tools—one that reflects a broader trend in tech towards responsible, community-driven AI innovation.​

The Future: Aurora as a Foundation for Scientific Discovery

Aurora is part of Microsoft’s AI for Science initiative, which envisions even larger “foundation models” that could one day integrate not just atmospheric data, but also information about ocean currents, ecosystems, and the molecular composition of the environment.​

Microsoft sees this as only the beginning—a platform that will support breakthroughs in climate science, disaster readiness, and perhaps even the development of new pharmaceuticals by simulating biological as well as atmospheric systems.

Get Involved

Researchers, developers, and organizations interested in leveraging Aurora can access code, data, and resources via the Azure AI Foundry Labs. Microsoft encourages contributions, collaborations, and experimentation to help shape the future of climate science and weather forecasting with open, accessible AI.​ Don’t forget to check out the latest agent to be released, MMCTAgent, a new AI tool that understands videos and pictures like people do.

For more breaking Microsoft updates, subscribe to msftnewsnow.com for the latest on AI, cloud technology, and the future of science-driven innovation!