WindBorne Systems secured $37 million in Series B funding co-led by Khosla Ventures and Galvanize to scale its long duration weather balloon network and AI forecasting models while expanding into commercial markets.
WindBorne Systems closed a $37 million Series B funding round, co-led by Khosla Ventures and Galvanize, with participation from TransLink Capital, Lux Capital, and prior investors. The round values the company at $250 million post money.
What is WindBorne Systems’ technology?
Founded in 2019 (with roots among Stanford affiliates), WindBorne builds and operates Atlas, a global constellation of long duration Global Sounding Balloons (GSBs). These lightweight (~1.2–3 lb / under 2 kg) autonomous balloons control altitude by adjusting ballast and buoyancy, fly for multi week durations (often 50+ days), and collect repeated vertical atmospheric profiles from near surface into the stratosphere. Traditional radiosondes provide only a single short duration profile (hours). The company reports roughly 20 launch sites worldwide and about 600 balloons airborne at any time, with data collection focused on under observed regions such as oceans and remote areas (including typhoon eyes). It is also beginning to deploy sensor packages that descend into the ocean and operate as floating buoys. The long term goal is scaling to ~10,000 concurrent balloons for comprehensive global coverage.
This proprietary in-situ data feeds WeatherMesh, the company’s deep learning weather model (latest version WeatherMesh-6 / WM-6, released around June 2026). WM-6 operates at 0.25° (~25 km) global resolution with a higher resolution 3 km variant, produces hourly (or more frequent) updates via proprietary AI data assimilation that incorporates balloon observations, satellite data, and other sources, and outputs a large set of surface, soil, atmospheric, radiation, and cloud variables. Independent benchmarks and company evaluations claim superior skill versus leading operational systems such as ECMWF’s IFS and AIFS ensembles (e.g., lower RMSE across variables and lead times; for 2 m temperature, a ~4.5-day WM-6 forecast comparable in accuracy to a 1 day IFS forecast in tested periods). The model also supports ensemble forecasting and has been used to navigate the balloon fleet itself. Complementary products include MetaMesh (a multi model blend that incorporates WeatherMesh) and APIs for observations, gridded/point forecasts, and related data.
Earlier capital included a seed round (approximately $6 million in 2023, led by Footwork with Khosla Ventures, Pear VC, Ubiquity Ventures, Harvest/Humba Ventures, and others) and a $15 million Series A in May 2024 led by Khosla Ventures (with Footwork, Pear VC, and Convective Capital). Pre Series B totals reported in databases were in the mid $20 millions range; the new round brings cumulative equity funding substantially higher and reflects continued conviction from Khosla (an early backer that also took a board seat earlier).
Proceeds target:
- Expanded compute for model training and inference.
- Transition of the balloon network’s communications from satellite to a mesh radio system (cost and latency improvements).
- Go to market team build-out to accelerate private sector sales.
- Continued constellation scaling and product development (including ocean sensors).
Recommended: AegisAI Raises $36 Million In Series A Funding Round
WindBorne’s core thesis is that atmospheric data gaps (roughly 85% of the atmosphere under observed for forecasting purposes, per World Meteorological Organization framing) limit forecast skill, and low cost, long endurance in-situ sensing plus modern AI closes those gaps more effectively than satellites or sparse traditional radiosondes alone. Proprietary balloon data creates a differentiation/moat for WeatherMesh relative to pure AI models trained primarily on reanalysis or government data. AI has lowered the barrier to private forecasting (models can run far more efficiently than classical numerical weather prediction on supercomputers), enabling companies to move beyond data sales into full forecasting and decision support products.
Current commercial focus for expansion is investment funds and commodity related users that trade on weather driven outcomes (agriculture, energy, etc.). Broader private sector adoption has historically been constrained by the cost and complexity of integrating weather data into workflows; investors (including Galvanize partner Saloni Multani) argue that higher accuracy forecasts plus AI tooling make integration more practical and valuable. Existing private weather firms often refine or repackage government forecasts for media, aviation, shipping, or specialized use cases.
Risks and challenges implicit in the model include operational scaling of a physical balloon fleet (airspace coordination, recovery/sustainability goals of retrieving a high percentage of balloons, regulatory compliance), competition from other AI weather efforts (Google DeepMind and others), government budget cycles, and the need to convert superior accuracy into sticky private revenue. The company emphasizes growing revenue alongside technical progress as evidence of product market fit.
The Series B provides runway to enlarge the “planetary nervous system” of balloons, strengthen the AI forecasting stack, and shift the commercial mix from government heavy toward higher volume private applications in weather sensitive industries.
Please email us your feedback and news tips at hello(at)techcompanynews.com
