Google has rolled out a major upgrade to its AI-powered weather forecasting system, integrating the new WeatherNext 2 model into Search, Gemini, Google Maps, and Pixel phones. The company says the upgraded model delivers forecasts eight times faster and with higher accuracy across 99.9% of weather variables.
The advanced AI system can generate predictions in under a minute using a single Google TPU chip, compared to traditional models that require hours of processing on supercomputers. WeatherNext 2 also provides up to 15-day forecasts with detailed hourly updates, producing hundreds of scenario variations from a single starting point.
According to Google DeepMind’s senior research director Peter Battaglia, the company is moving its weather technology from experimentation to full consumer deployment, citing strong performance in real-world testing. The technology’s underlying Functional Generative Network (FGN) allows the model to introduce controlled randomness and generate multiple outcomes in one pass.
Industries such as energy, agriculture, transportation, and logistics are expected to benefit from more granular predictions. Google is also launching an early access program for businesses needing customized weather models, while developers can access forecast data through Google Earth Engine and BigQuery.
