This AI weather startup is out-forecasting government agencies

WindBorne Systems, an AI weather startup, has launched WeatherMesh-6, an advanced forecasting tool. This new system offers more frequent and accurate predictions of key variables compared to the European Centre for Medium-Range Weather Forecasts (ECMWF), a leading intergovernmental organization in weather prediction. The startup traces its origins to Stanford in 2019, initially focusing on creating superior weather balloons for data collection before pivoting to AI model development in 2022.

WeatherMesh-6 produces hourly forecasts, a significant improvement over traditional models that typically generate forecasts every six hours. Its resolution has reached 3 km in Europe and the continental U.S., areas with optimal data quality. WindBorne’s chief product officer, Kai Marshland, highlights the system's accuracy, stating that WeatherMesh-6 is "as accurate five days out as a traditional forecast is the day before," particularly for surface temperature measurements.

The company’s approach is unique due to its integration of model-building and data collection. WindBorne operates approximately 400 weather balloons globally, continuously gathering sensor readings. These advancements stem from improved methods of integrating the collected balloon data into their AI models. Unlike traditional weather forecasts that rely on complex physics models and supercomputers, AI models, such as WeatherMesh-6, process information more rapidly.

WindBorne has garnered $25 million in venture funding. It sells its balloon data to entities like NOAA, the U.S. Air Force, and the Navy, and provides forecasts to investors and commodity traders. Despite these commercial activities, CEO John Dean emphasizes the company's primary focus on enhancing its model and data infrastructure, anticipating future information consumption trends that may move beyond conventional software as a service (SaaS) products.