Google Partners with National Hurricane Center for AI-powered Tropical Cyclone Forecasts

Google has unveiled a new artificial intelligence model designed to forecast tropical cyclones, developed in collaboration with the US National Hurricane Center (NHC). Through a new website called Weather Lab, Google aims to showcase this experimental AI technology, which can simulate up to 50 scenarios for a storm’s path, size, and intensity up to 15 days out. While still under evaluation, early results show that the AI model’s predictions are significantly closer to actual storm tracks than some leading traditional models, particularly in the North Atlantic and East Pacific regions.

Despite these advances, Google emphasizes that the tool is not meant for public forecasting use—yet. Instead, the project is focused on supporting the NHC’s efforts to improve lead times and accuracy, giving communities more time to prepare. The AI model is trained on the ERA5 dataset, which merges millions of global observations with traditional model outputs. This same data source helped power Google’s previous GenCast model, which outperformed conventional models in nearly all tests published in a 2024 Nature study.

The initiative also reflects growing interest in AI-enhanced forecasting tools amid staffing cuts and budget constraints at federal agencies like NOAA. While AI brings potential for more precise predictions, experts stress that it complements rather than replaces traditional physics-based weather modeling. Google’s partnership with academic institutions in the US, UK, and Japan underlines its commitment to collaborative, research-driven development of next-generation climate tools.

However, concerns remain about the future of public access to weather data and services, especially under proposals to privatize federal forecasting agencies. Google’s cautious language and partnership with public institutions suggest an awareness of the sensitive balance between innovation and accessibility in climate and weather forecasting.