The AI Revolution in Weather Forecasting
Deep learning techniques that power large language models (LLMs) have also revolutionized meteorology. AI-driven weather simulations can now run on laptops instead of supercomputers, making advanced forecasting more accessible than ever. But the more significant challenge lies in helping people and organizations harness these forecasts effectively.
WindBorne Systems: A New Era of Data Collection
WindBorne Systems, a startup deploying the world's longest-flying weather balloons and integrating their data into a powerful forecasting model, has secured a $37 million Series B round to tackle this challenge, CEO John Dean told TechCrunch. The round, co-led by Khosla Ventures and Galvanize, with participation from TransLink Capital, Lux Capital, and existing investors, values the company at $250 million post-funding.
From Data Acquisition to AI-Powered Forecasts
Founded in 2019, WindBorne initially focused on acquiring novel weather data through low-cost sensors and endurance balloons. The emergence of AI forecasting models over the past four years enabled them to generate their own predictions—a feat previously impossible for most private firms due to the prohibitive cost of supercomputers required for atmospheric simulation.
Today, WindBorne operates 20 launch sites globally and maintains roughly 600 balloons in the air at any time, collecting data in hard-to-reach regions, such as the eye of a typhoon. The company is now deploying aerial sensor packages that descend into the ocean and continue gathering measurements as floating buoys.
This proprietary data network—what Dean calls a "planetary nervous system"—creates a competitive moat for their model, which also ingests public datasets from government weather agencies worldwide. "We've shown that adding balloons to forecasts improves accuracy, and each data point delivers stronger value than satellites," Dean explained. "We've grown revenue alongside this progress, which has de-risked the demand signal for VCs."
Government-First, Commercial Next
WindBorne's primary customers are currently government agencies. The U.S. National Weather Service purchases their data, while the U.S. Air Force and Navy fund research partnerships, including an effort to deploy forecasting models on ships with intermittent connectivity.
The next frontier is commercial expansion, initially targeting investment funds that use weather data to predict commodity prices and other economic outcomes. The Series B funds will support compute scaling, replacing the balloon network's satellite communications with a mesh radio system, and building a go-to-market team to grow private-sector clientele.
Navigating the Private Sector Challenge
Expanding commercially isn't straightforward. Over the past decade, startups attempting to scale sensing businesses—like earth-observing satellite networks—have struggled to penetrate private markets. Extracting actionable value from such data requires specialized expertise and established workflows. Consequently, many have pivoted to government clients already accustomed to utilizing this data.
While private weather forecasters exist, WindBorne's success will hinge on bridging the gap between data generation and real-world application. With AI lowering barriers to entry and their unique data collection approach, the startup is well-positioned to make weather intelligence both impactful and profitable.
via TechCrunch AI