Article By Jenn Gidman
WindBorne Systems says its WeatherMesh-6 delivers faster forecasts than current top models
A weather startup powered by artificial intelligence says it’s now beating one of the world’s top forecasting centers, all by floating hundreds of “smart” balloons. WindBorne Systems, founded by Stanford students, has rolled out WeatherMesh-6, an AI model it claims can predict surface temperatures five days in advance, with about the same accuracy many traditional systems reach just one day out, per TechCrunch. The tool updates hourly, whereas conventional models tend to update every six hours.
WindBorne execs say this AI system’s edge comes from directly feeding data from some 400 in-flight balloons directly into a transformer-based model, reducing reliance on European and US government datasets. WindBorne, which was backed with $25 million in funding and valued at about $85 million in 2024, sells data to the National Oceanic and Atmospheric Administration and the US military, while also marketing forecasts to investors and commodity traders. One of the Stanford co-founders, John Dean, recently penned an explainer on how his company is able to create such accurate forecasts.
Be the first to comment