Key Highlights:

• Nvidia released three open-source AI models for weather forecasting.
• The models promise faster and cheaper predictions than traditional simulations.
• Nvidia says AI can run weather ensembles 1,000 times faster once trained.
• Insurance firms could use them to study floods, hurricanes, and rare extremes.

Nvidia has unveiled three open-source AI models designed to make weather forecasts faster, cheaper, and more detailed. The models, announced at the American Meteorological Society’s annual meeting in Houston, aim to replace slow and costly weather simulations with AI-driven systems that can match or exceed traditional accuracy.

Why it matters: modern forecasting depends on massive simulations that consume time, power, and money. Nvidia says its AI can change that balance.

The new models are part of Nvidia’s “Earth-2” initiative. They target long-range forecasts, short-term severe weather, and the integration of data from diverse sensors.

How Nvidia’s AI changes weather forecasting

Traditional weather prediction runs on “ensembles.” Each forecast includes many possible outcomes from the same starting conditions. Scientists use these to estimate risk. But every added “member” increases cost and compute time.

Nvidia’s approach flips that equation.

Mike Pritchard, Nvidia’s director of climate simulation research, said once trained, AI models can run forecasts 1,000 times faster. That speed allows researchers to run huge ensembles without prohibitive cost.

As a result, teams can test thousands of scenarios. They can search for rare but devastating outcomes, like flash floods or category-five hurricanes.

Why insurance firms care

One of the earliest business users may be insurance companies. Insurers need to model extreme events that happen rarely but cause massive losses.

Until now, simulating those outliers in detail was expensive. Firms often had to compromise on resolution or scale.

With AI, they can run 10,000-member ensembles. They can zoom in on whether a specific neighborhood might flood. They can price risk with finer detail.

This changes how climate risk is measured.

What the Earth-2 models do

Nvidia’s Earth-2 release includes three tools:

• A model for 15-day global weather forecasts.
• A short-range model focused on severe U.S. storms, up to six hours ahead.
• A data-integration model that merges inputs from satellites, radars, and sensors.

Together, they create a pipeline from raw data to usable forecasts.

Nvidia’s open-source bet

By releasing these models as open source, Nvidia wants developers, researchers, and institutions to build on them. The company is not just selling chips. It is shaping the software layer of climate science.

In the long run, Nvidia aims to make high-resolution forecasting accessible beyond national labs and supercomputers.

That shift could redefine how the world prepares for weather. And it places Nvidia at the center of climate-scale AI.