Brightband is developing machine-learning models trained on years of observational weather data to deliver faster, cheaper weather and climate forecasts than traditional physics-based systems. The company plans to ingest and release large, underused datasets (satellite and weather‑balloon observations) to improve model performance. Brightband aims to open source its core forecasting models, the training data, and evaluation metrics while offering paid, specialized services on top. The team is building on existing physics-based models and research, targeting state-of-the-art accuracy and much lower compute and latency. Use cases highlighted include renewables forecasting for energy, transportation risk avoidance, and agricultural planning. The product is early — nothing is live today — and Brightband expects a model that ingests observations and produces forecasts by the end of 2024. The company is organized as a public benefit corporation and has raised $10 million in funding.

Total raised

$10M

Funding rounds

1

Latest round

Series A

Latest activity

Sep 2024