Weather balloons meet AI: WindBorne raises $37 million to sell smarter forecasts
The startup runs 600 balloons around the globe to collect data that even satellites miss. Now it wants to sell that edge to private businesses, not just governments.
Key points
- WindBorne Systems raised $37 million in a Series B round, valuing the company at $250 million.
- The startup operates roughly 600 weather balloons from 20 launch sites and sells forecast data to the U.S. National Weather Service, the Air Force and the Navy.
- AI forecasting tools now let WindBorne run atmosphere simulations that once required a supercomputer, cutting costs dramatically.
- The company is developing sensor packages that drop into the ocean and keep collecting data as floating buoys.
- WindBorne plans to expand sales to private-sector buyers, starting with investment funds that trade on weather data.
For most of history, predicting the weather accurately meant a room-sized supercomputer crunching billions of calculations. New AI models, the same family of technology behind chatbots, have changed that. A forecast that once demanded a national government's computing budget can now run on a laptop.
WindBorne Systems, founded in 2019, wants to turn that shift into a business. As first reported by TechCrunch AI, the company raised a $37 million Series B round, co-led by Khosla Ventures and Galvanize, with participation from TransLink Capital and Lux Capital. It values the company at $250 million.
What does WindBorne actually do?
It collects weather data from places that satellites and ground stations miss, using cheap, long-endurance balloons.
About 600 balloons stay aloft at any given time, launched from 20 sites worldwide, reaching places like the eye of a typhoon. CEO John Dean calls the network a "planetary nervous system." WindBorne is now testing sensor packages that parachute into the ocean and float on the surface, collecting readings as buoys.
All that raw data feeds into WindBorne's own AI forecasting model, which also pulls in publicly available data from government weather agencies. Dean says adding balloons to the forecast yields meaningfully more accurate results than satellite data alone, and that the value per data point from a balloon beats a satellite's.
Current paying customers are almost entirely government bodies. The National Weather Service buys WindBorne's data directly. The Air Force and Navy fund research partnerships, including work on forecasting models that can run aboard ships with intermittent connectivity, where downloading a full forecast from shore isn't always possible.
Why is the private market so hard to crack?
Selling weather data to businesses has tripped up many startups before. Extracting value from raw forecast data takes experience and existing workflows that most companies don't have.
That's why private weather companies have mostly made money repackaging government forecasts for narrow needs: news broadcasts, aircraft de-icing schedules, ship routing. The customers who want raw data tend to be investment funds that use weather patterns to predict crop yields or commodity prices. That's also where WindBorne is starting its commercial push.
Galvanize partner Saloni Multani, who co-led the round, put it plainly: "Integrating weather forecasts into broader business decision-making has traditionally been expensive and difficult. We think AI changes that equation."
The argument is that AI tools can now connect a forecast directly to a business decision without an army of specialist analysts in between. Worth watching: whether energy traders and logistics firms, which have real weather exposure but no established data workflows, follow the investment funds in.
What happens next?
WindBorne will use the new money on four things: computing costs, replacing the balloon network's satellite links with a mesh radio network (a web of radios that passes signals between balloons instead of routing everything through a satellite), building a sales team, and expanding into the private sector.
We first covered WindBorne on 5 August 2026. The bet the company is making now is that AI lowers the barrier enough that businesses outside finance start acting on weather data routinely. That's a reasonable thesis, but it's the same thesis several sensing-network startups have made before and not yet proved.
