Discovery Green builds an AI operating system for corporate renewable power

Discovery Green has launched EnergyOS, a proprietary operating system designed to help business customers decide how to use contracted renewable electricity against cost, coverage and risk. The system combines forecasting, allocation models and an AI layer that Discovery calls EnergyAI.

The South African angle is direct. Private electricity trading is becoming a larger part of how companies reduce exposure to Eskom tariffs and decarbonise operations, but buying renewable power is not the same as matching it efficiently to a company’s changing demand. EnergyOS is Discovery Green’s attempt to turn that optimisation problem into software.

What the system actually does

Discovery says EnergyOS is built on a shared data lake and combines forecasting, customisation and allocation models. Its stated job is to improve long-term savings, renewable-energy coverage and risk management for business customers. EnergyAI runs across the system and is intended to learn from operating outcomes over time.

That makes the product closer to an energy-management and decision engine than a consumer-facing AI assistant. The underlying problem is highly numerical: electricity demand moves through the day, renewable generation varies, market prices change and each customer has different constraints. The software has to decide how those variables should be balanced rather than merely produce a dashboard after the fact.

Discovery Green is a licensed electricity trader, so the platform also sits inside a changing South African power market where private generators, traders and large users are taking on roles that were historically concentrated around Eskom. For businesses, the practical measure will be whether the optimisation produces measurable savings and higher renewable coverage without adding unacceptable exposure to market volatility.

The company says EnergyOS powers every Discovery Green solution. It has not publicly disclosed enough operating data to independently judge how much EnergyAI improves outcomes against conventional forecasting and optimisation methods.

That is the unresolved part of the launch. The architecture is credible for the problem Discovery Green is trying to solve, but the useful benchmark will be customer-level performance: forecast accuracy, avoided cost, renewable coverage and how the system behaves when electricity prices or generation depart sharply from expected patterns.