WindBorne Wins US Air Force Contract Worth Up to $49 Million for Balloon-Collected Weather Data

WindBorne Systems, a California startup that flies fleets of autonomous weather balloons and builds AI forecasting models, has won a contract with the US Air Force worth up to $49 million, the largest in the company’s history.

Under the agreement, reported by TechCrunch, the Air Force will buy atmospheric data collected by WindBorne’s balloons, which drift around the globe gathering sensor readings at altitudes that ground stations and satellites struggle to measure directly. The Air Force pays only as it needs data, placing orders for specific regions at specific times. So far it has ordered about $1.7 million worth, and company executives expect more orders to follow.

The contract was awarded without competitive bidding, which WindBorne says reflects the uniqueness of its capability. It comes as governments and militaries increasingly turn to private companies and AI to improve forecasting in an era of more frequent extreme weather.

A constellation in the sky

WindBorne currently has about 700 weather balloons aloft. Founded in 2015 and based in Redwood City, California, the company designed long-duration balloons that can adjust their altitude to ride different wind currents, allowing them to stay airborne for extended periods and steer towards areas where data is scarce.

Traditional radiosondes, the instrument packages carried by weather balloons released from hundreds of stations worldwide twice a day, burst after a couple of hours and cover mostly land. Vast stretches of ocean and remote regions remain poorly observed, creating blind spots in the data that feed weather models. WindBorne’s balloons are designed to fill those gaps.

The company has worked closely with US government agencies, including the National Weather Service, the Air Force and the Navy. Its balloons have flown into extreme weather systems, including the eye of Typhoon Dolphin, gathering data from conditions that are difficult and dangerous to observe by other means.

Why the Air Force wants it

Accurate weather data is critical to military operations, from planning flights and missile tests to deploying ships and troops. The Air Force’s weather wing supports operations around the world, often in regions where observation networks are thin.

WindBorne co-founder and chief technology officer Andrey Sushko said the contract would support further growth of the balloon constellation as well as research and development to improve the hardware and performance of the balloons. “This is largely a vehicle through which continued growth of the constellation can be supported,” he told TechCrunch.

He described the sole-source award as “a determination by the government that for this capability of collecting global data, we were the only company able to provide it in the way that met their needs for these operations.”

Data meets deep learning

WindBorne is not only a data provider. It also builds its own AI forecasting model, WeatherMesh, and was one of the first organisations to feed empirical observations directly into a deep-learning weather model. Earlier this year it released WeatherMesh 6, which the company says delivers more accurate forecasts than the physics-based model run by the European Centre for Medium-Range Weather Forecasts (ECMWF), long regarded as the global gold standard.

AI weather forecasting has become one of the most productive applications of deep learning. Traditional forecasting relies on supercomputers solving equations that approximate atmospheric physics. AI models trained on decades of historical weather data can produce forecasts far faster and, increasingly, with comparable or better accuracy for many variables. Google DeepMind, Nvidia, Microsoft and national meteorological agencies have all released AI weather models.

But AI models are only as good as the data that initialises them. WindBorne’s strategy is to own both the observations and the model, giving it a vertically integrated position that few competitors can match.

Funding and scale

WindBorne chief executive John Dean told TechCrunch in August that the company had raised a $37 million Series B round to continue scaling. The Air Force contract adds a significant, recurring government revenue stream, which can make commercial weather companies more resilient. Customers for weather data and forecasts range from airlines, shipping companies and energy traders to insurers and agricultural businesses.

The contract structure, an umbrella agreement with orders placed as needed, means actual revenue will depend on how much data the Air Force buys. But the ceiling signals substantial confidence in the technology.

Why observation still matters

AI forecasting models have improved rapidly, but they remain dependent on the quality of the initial snapshot of the atmosphere they are given. Forecast errors often originate in regions where few observations exist, particularly over oceans, polar areas and parts of the developing world, and then propagate through the model as it projects forward in time.

Balloons that can stay aloft for long periods and navigate towards data-sparse regions offer a way to fill those gaps at lower cost than aircraft reconnaissance or new satellites. Data gathered inside tropical cyclones is particularly valuable because storm intensity and track forecasts carry high stakes for coastal populations, shipping and militaries.

TechCrunch noted that models built by frontier labs and government agencies are proving more accurate than traditional forecasts in many tests, although it remains a jagged frontier, with both types of forecasts still needed. That combination, physics-based and AI-based models working alongside each other and fed with richer observations, is where most meteorologists expect the field to settle in the near term.

The Air Force’s flexible ordering model also suits the technology. Instead of paying for a fixed network, it can direct demand to specific regions at specific times, for example ahead of exercises, deployments or approaching storms, and pay only for the data it uses.

The wider stakes

The deal underscores a broader shift in how governments obtain weather data. Public agencies have historically operated their own observation networks and models. Budget pressures and the pace of private innovation are pushing them to buy data and services from companies instead, a model already familiar in satellite imagery and launch services.

That shift raises questions about data access and the public good. Weather data has traditionally been shared freely among nations through the World Meteorological Organization. Commercial models could create tensions if valuable observations become proprietary.

For India, which faces intense monsoons, cyclones along both coasts and increasingly frequent heatwaves, advances in AI forecasting and observation are highly relevant. The India Meteorological Department has been investing in AI-based forecasting and expanded observation networks, and Indian startups are exploring weather and climate analytics for agriculture and insurance. WindBorne’s Air Force contract is a reminder that better data, collected in places that have never been well observed, may prove as valuable as better algorithms in the race to predict an increasingly volatile climate.