Google has launched WeatherNext 3, an AI weather forecasting model designed for the energy sector, improving accuracy and response time.
Google has launched WeatherNext 3, an advanced AI weather forecasting model designed specifically for the energy sector.
This new iteration introduces critical variables that affect power generation assessments and electrical grid risks, including wind speed at 100 meters, cloud cover, and sunlight exposure.
Unlike its predecessor, WeatherNext 2, which utilized a 25 km grid and updated every six hours, WeatherNext 3 offers a more granular and timely service. It features global forecasts updated hourly, with a 5 km resolution for ground variables such as temperature and humidity, and an 11 km resolution for precipitation.
According to Google, WeatherNext 3 improves upon previous models by learning directly from meteorological station data and recent satellite imagery, which has decreased information latency from seven hours to a range of three to four hours, as highlighted by Ilan Price, a senior researcher at Google DeepMind.
The model has shown notable improvements in precipitation estimation accuracy, achieving:
- up to 60% better accuracy compared to NASA IMERG;
- up to 30% compared to MRMS;
- up to 10% when compared to rain gauges in early forecast periods.
Despite its advancements, there are concerns that training solely on historical data may struggle with unprecedented weather events, whereas traditional models rely on established atmospheric principles.
WeatherNext 3 is integrated into various Google services, including Google Search, Gemini, Google Maps, and the Weather API. Corporate clients can access it through BigQuery, Earth Engine, and Google Cloud Storage, simplifying deployment.
Google has not disclosed the pricing for corporate subscriptions to WeatherNext 3.
Weather Data Market Insights
S&P Global Market Intelligence's US Grid Outlook for 2026 predicts that solar generation and storage will lead new capacity additions in the U.S., with expected contributions of 51.2 GW and 25.7 GW, respectively, from a total of over 90 GW planned.
The demand for accurate weather information is heightened by the financial risks associated with procurement errors. For example, underestimating wind generation necessitates last-minute electricity purchases, typically from gas plants, while overestimating can lead to curtailed wind and solar outputs due to grid limitations.
Key competitors in the energy weather data market include Vaisala, Solcast, DNV's WindGEMINI, IBM HyperWatch, and Swiss firm Jua.
For Google, the introduction of WeatherNext 3 not only expands its product offerings but also aids in optimizing energy procurement costs, as accurate generation forecasts significantly impact its financial outcomes.
In March 2026, a new AI model named Aardvark Weather was developed by researchers from the UK and Canada for weather forecasting.
