US Air Force Awards WindBorne Up to $49M for Weather-Balloon Data in Startup's Biggest Deal

The U.S. Air Force is buying weather data from balloons - autonomous, AI-guided balloons, to be precise. WindBorne Systems has signed a sole-source contract worth up to $49 million to supply atmospheric readings collected by its fleet of about 700 weather balloons drifting around the globe, TechCrunch reported this week.

The deal is the largest the Redwood City, California startup has ever landed. It also signals how seriously the Pentagon is taking a new approach to weather forecasting: rather than launching sensors from the ground, gather readings continuously from a fleet of steerable balloons and feed them straight into AI models.

The Air Force is not spending the full $49 million at once. The contract works on a pay-as-you-go basis: the service places orders for atmospheric data only when it needs it, for specific regions and time windows. So far, $1.7 million in orders have been placed, with executives expecting more to follow.

Notably, the contract was awarded without a competitive bidding process. Co-founder and chief technology officer Andrey Sushko said the government determined that WindBorne was the only company able to provide this kind of global data collection in the way the Air Force's operations require.

Sushko described the contract as a vehicle for the continued growth of the balloon constellation, with research and development attached to improve the balloons' hardware and performance. WindBorne already works with the National Weather Service and the U.S. Navy; this agreement is the company's biggest government partnership yet.

WindBorne's balloons have no engines. They steer by changing altitude - venting air and dropping ballast sand - while software decides where to send them. In August, the company became the first operator to fly an unpowered aircraft into the eye of a typhoon to collect data, an industry first.

That mission, Sushko said, was one of the first times a deep-learning model was used to make actual operational decisions - a glimpse of how AI is moving from the lab into live flight control.

Data from the fleet feeds directly into WindBorne's own forecasting model, WeatherMesh. The latest version, WeatherMesh 6, launched this summer, and the company claims it produces more accurate forecasts than the traditional physics-based models run by the European Centre for Medium-Range Weather Forecasts.

AI-driven weather prediction has become one of deep learning's strongest applications since the rise of transformer architectures: models train end-to-end on huge historical datasets, such as those released by the ECMWF, and produce forecasts far faster than supercomputer simulations of atmospheric physics.

In August, WindBorne closed a $37 million Series B round to scale its business. Founded in 2015, the startup is now pairing venture backing with government demand - and the Air Force's $49 million ceiling suggests its balloons have become a serious piece of military weather infrastructure.