WindBorne Systems

WindBorne Systems is an AI weather-forecasting and atmospheric-data company founded in 2019 by Stanford Space Initiative alumni John Dean, Kai Marshland, Andrey Sushko, and Joan Creus-Costa. Headquartered in Palo Alto, California, the company has raised roughly $24.9 million — including a $15 million Series A led by Khosla Ventures, with Footwork, Pear VC, and Convective Capital participating — and runs with a team of around 48 people.

WindBorne pairs two tightly coupled systems. The first is a global constellation of long-duration smart weather balloons that launch from roughly 15 sites worldwide and stay aloft for weeks, gathering atmospheric readings from regions where conventional weather data is sparse, with hundreds operating simultaneously. The second is WeatherMesh, an AI forecasting model trained on that proprietary balloon data alongside public datasets, which lets the company assimilate observations directly rather than depending on outside agencies.

The company's June 2026 model, WeatherMesh-6, produces hourly forecasts at 3-kilometer resolution across Europe and the continental United States and, by WindBorne's account, is as accurate five days out as a traditional forecast is one day ahead — particularly for surface temperatures — outpacing the European Centre for Medium-Range Weather Forecasts. WindBorne sells both its raw balloon data and its forecasts to customers including the National Oceanic and Atmospheric Administration, the US Air Force, and the US Navy, as well as to investors and commodity traders. The company frames its work as infrastructure for climate adaptation and disaster preparedness — better forecasts where the world has the least data.