- Nvidia introduces three generative AI weather models capable of producing 15-day forecasts and short-term storm predictions with much lower computational costs.
- The new tools could reshape how insurers, energy firms and meteorological agencies model extreme weather, challenging traditional numerical simulation approaches.
What happened: AI at the front line of weather forecasting
In Houston on 26 January 2026, at the American Meteorological Society’s annual meeting, Nvidia – a leading US chipmaker known for its graphics processors and AI accelerators – announced three new open-source AI models designed to streamline weather forecasting.
The so-called Earth-2 family includes a model for 15-day global forecasts, an ultra-short-term “nowcasting” tool for severe storms up to six hours ahead, and a system to integrate diverse data sources such as satellite, radar and ground station observations.
According to Nvidia’s climate simulation director, once trained these models run about 1 000 times faster than traditional physics-based simulations, potentially allowing organisations to run 10 000-member ensemble forecasts that would be prohibitively costly with conventional methods.
Weather services and research agencies in regions from Taiwan to Israel are already testing Earth-2 tools for operational forecasting, while insurers and energy traders are exploring applications in risk modelling and resource planning.
Also Read: Nvidia Sparks Debate Over Chiller-Free Cooling in Data Centres
Also Read: Nvidia boss visits china amid regulatory headwinds and AI demand
Why it’s important
Nvidia’s move signals a paradigm shift in meteorology: from supercomputer-heavy numerical models to AI-driven systems that can handle complex, high-dimensional data more efficiently. Traditional forecasting has relied on solving physical equations over fine grids, a process demanding vast computational resources and time; AI has the potential to accelerate this dramatically.
The entry of a major compute-hardware company into public scientific modelling highlights how AI is blurring boundaries between commercial and research domains. Faster forecasts at lower cost may unlock new business models in insurance and energy markets, where predictive certainty directly affects pricing and capital allocation. A financial view suggests that firms offering real-time, high-resolution forecasts could gain a competitive edge in risk-adjusted returns.
However, academic work shows that some existing AI weather models can struggle with rare, record-breaking extremes, underscoring the need for rigorous validation before widespread operational deployment.
