National Grid ESO, the UK electricity system operator that balances tens of gigawatts of electricity supply and demand every minute, partnered with Open Climate Fix and is now running an AI-powered solar forecasting model in its control room — combining real-time satellite imagery with weather and solar generation data to deliver three times the accuracy of the prior approach.
Lyndon Ruff, AI Centre of Excellence Manager at National Grid ESO, said: "Embedded solar is one of the biggest sources of uncertainty in our net demand forecast, but it's crucial to our operations." He added: "National Grid ESO has worked closely with Open Climate Fix to develop this AI solar forecasting solution... It helps control room engineers reduce balancing costs and emissions."
National Grid ESO manages the UK's electricity transmission grid and must continuously balance supply and demand. Embedded solar is one of the biggest sources of forecast uncertainty — even brief cloud cover can swing solar output and force operators to bring other generation online to keep the grid stable. A more accurate short-window solar forecast translates directly into lower balancing costs and lower emissions from over-firing fossil reserves.
The case study does not specify start dates, but it describes a partnership in which National Grid ESO and Open Climate Fix "combined forces to reduce the uncertainty through improved solar forecasting," moving from initial collaboration into active production deployment where the AI forecasts are now live in the control room and used by engineers for real-time balancing decisions.
The AI model combines real-time satellite imagery, weather data, and historical solar generation data into a single forecast pipeline. The case study does not publish the underlying model architecture in detail, but the deployment surfaces forecasts to control room engineers through the existing operational tooling, where the team uses them to refine real-time balancing decisions across the GB transmission network.
The forecasts sit inside the National Grid ESO control room — the same team that moves megawatts around the country — and are being used by control room engineers for live balancing decisions. The Quartz case study reports the model as delivering three times the accuracy of the previous forecasting approach, which translates directly into reduced balancing costs and lower emissions from over-firing reserve generation.
The case study does not lay out an explicit future roadmap for the deployment beyond noting that National Grid ESO is "currently using these forecasts in the control room"; the implicit next-stage opportunity in the source is broader rollout of the same model to other forecasting horizons or grid-management use cases.
Details