Santee Cooper partners with Google to improve load

Santee Cooper announced Thursday it is partnering with Google to modernize its financial planning and load forecasting using artificial intelligence tools. The utility’s CFO, Tami Wilson, said during a Wednesday press teleconference that Google’s WeatherNext 2 will provide more accurate medium-range weather forecasts to improve load forecasting, while the Gemini Enterprise Agent Platform will speed up financial forecasting.

Wilson described the initiative as a "major digital transformation" focused on shifting away from legacy, clunky processes and automating routine tasks without removing human accountability for risk management. She emphasized that human operators will retain responsibility for oversight.

The utility plans to use generative AI for low-cost initial drafting and agentic AI to execute multi-step processes autonomously. Santee Cooper did not provide details on the cost of the tools or which specific processes will be automated.

Raiford Smith, global director of power and energy industry at Google Cloud, said Gemini Enterprise was designed specifically as an intelligent operating layer for modern utilities, delivering at least three distinct capabilities built for complex infrastructure.

WeatherNext 2 delivers probabilistic forecasts across lead times of up to 15 days, according to a Google press release. Wilson said Santee Cooper previously relied on traditional meteorological models but expects WeatherNext 2 to dramatically improve forecasting ability, potentially saving millions of dollars annually.

She noted that Google is helping build a custom forecasting model that accounts for unique weather factors in the utility’s service area, including Lake Marion and Lake Moultrie, which are large enough to create their own microclimate.

Wilson highlighted the financial impact of forecast accuracy, stating that a one-degree variance in temperature can instantly swing power demand by about 100 MW per hour-equivalent to powering tens of thousands of homes. Being caught short on a freezing winter day or scorching afternoon could cost up to $100,000 per hour to procure emergency power.

Smith explained that WeatherNext 2 uses inference-based modeling, while traditional models are physics-based. Physics-based models perform well for 24 to 48 hours but lose accuracy beyond that as mathematical complexity increases, impairing medium-range forecasts.

Santee Cooper is also using Google AI tools to enable rapid, scenario-based financial planning that feeds into calculations for major projects, including its $10 billion grid-expansion budget, according to a Google release.

The utility is deploying several hundred Gemini Enterprise licenses across the organization to modernize workflow. Wilson said the rollout includes proper training for employees handling confidential and restricted data to ensure security and compliance.