US Air Force Awards WindBorne $49M Weather Data Deal

The U.S. Air Force has awarded WindBorne Systems a sole-source contract valued at up to $49 million to supply real-time atmospheric data. Under the agreement, the military pays as it orders data for specific locations and times, with $1.7 million ordered to date. The startup, which raised a $37 million Series B round in August, will use the contract to support its hardware research and expand its global balloon fleet.

Founded in 2015 and based in Redwood City, California, WindBorne operates roughly 700 autonomous balloons worldwide. The unpowered vehicles navigate global winds solely by adjusting altitude through bleeding air or dropping sand ballast. In August, the company became the first organization to steer an unpowered flight system directly into the eye of a typhoon, using its deep-learning models to guide operational navigation decisions in real time.

WindBorne feeds its empirical atmospheric readings directly into WeatherMesh 6, its proprietary deep-learning forecasting model. According to the company, WeatherMesh 6 yields more accurate forecasts than traditional physical simulation models run by the European Centre for Medium-Range Weather Forecasts. The broader meteorology field has shifted toward end-to-end deep learning following the arrival of transformer architectures and open weather datasets, enabling AI systems to issue rapid predictions that bypass expensive physics equations on supercomputers.

While organizations such as Google DeepMind provide advanced weather and cyclone tools to the U.S. government, effective AI forecasting still relies heavily on fresh physical data. For practitioners, this contract highlights a growing shift toward pairing autonomous sensor constellations with machine learning models, offering operational teams fast, empirical weather intelligence during critical missions and severe weather conditions.