NextEra, Santee Cooper Cite AI Savings for Grid and Market

NextEra Energy says an artificial intelligence tool for dispatch and outage scheduling has saved its customers more than $20 million so far this year. Santee Cooper expects a custom AI weather-forecasting model to help it avoid spot-market purchases that can cost $100,000 an hour during extreme weather.

Both utilities detailed their rapid AI deployments during a September 2 virtual media roundtable hosted by Google Cloud. The session highlighted the Gemini Enterprise platform as a foundation for moving AI beyond generic chatbots and into high-stakes operational decisions for the power sector.

Rich Argentieri, president of NextEra Analytics, said utilities are being asked to deliver greater reliability, lower costs, and a grid ready for the future. Santee Cooper's vice president and CFO, Tami Wilson, described a shift away from legacy processes toward AI-driven forecasting to save time and money.

NextEra's Grid Optimization Platform

NextEra built its Grid Composer platform on Google's Gemini Enterprise Agent Platform and rolled it out across Florida Power and Light's generating fleet. The platform integrates real-time telemetry, load data, and generation profiles-roughly half a trillion data points daily-into a single model that compares manual dispatch decisions against an AI-optimized alternative.

Argentieri said the first version of the tool was built in less than 12 weeks. The optimized dispatch and outage scheduling it enables have produced over $20 million in customer savings this year, according to the company. NextEra has since made the underlying tools available to other utilities through a Google Cloud marketplace product called Optos.

Optos Composer unifies generation, fuel, maintenance, trading, reserves, and storage decisions that were traditionally managed by separate, siloed teams. "All of those teams very much traditionally siloed. And not only the people, but all the data and the processes that go into that," Argentieri said. He credited the single platform with avoiding suboptimal decisions and providing visibility into how choices in one area affect system-wide costs.

Argentieri also pointed to field applications. He estimated 85% to 90% of FPL's workforce now uses voice-activated tools to pull specifications or guidance on site and can photograph a part for automatic catalog matching. He said these tools, along with AI-assisted troubleshooting, improve technician efficiency and safety, especially for newer employees during their learning curve. The company has also applied AI to driver-safety recommendations for its crews.

Santee Cooper's AI Weather Forecasting

Santee Cooper's AI push targets the financial exposure from weather forecasting errors. The utility serves more than 2 million people in South Carolina. Wilson explained that inaccurate forecasts lead to costly mistakes.

"If we overestimate how much electricity we will need, we spend money on fuel and operations we don't need." Wilson said. She noted that on the coldest winter day, one degree of temperature variance can cost up to $100,000 per hour on the spot market and swing power needs by roughly 100 megawatts an hour.

To address this, Santee Cooper is building a custom forecasting model with Google based on its WeatherNext platform. The model is tailored to local geography, including microclimates around two large lakes managed for hydro units. Wilson expects it to improve accuracy and help avoid costly market spikes, with savings flowing directly to customers.

On financial forecasting, Wilson said her team historically managed more than 150 reports in Excel, a process she called "extremely painful." Any change in assumptions could take weeks to process. While not yet in production, a new tool based on Gemini Enterprise is expected to cut those runtimes by about 75%. This will aid the finance team as it works through a $10 billion grid-expansion plan. The utility is also distributing several hundred Gemini Enterprise licenses for natural-language queries against internal systems.

AI Governance and Industry Impact

Both utilities emphasized formal governance structures for their AI rollouts. Wilson outlined a three-part framework of governance, training, and change management. Santee Cooper is establishing a cross-functional Innovation Council to oversee AI strategy and requires training for employees using Gemini Enterprise.

"We're not going to use this to replace jobs by any stretch," Wilson said, describing a "human in the loop" approach and a "creator-editor approval model" for generative AI outputs.

Raiford Smith, Google Cloud's global director of power and energy, connected the utilities' efficiency gains to improvements in Google's own data center operations. He cited a more than sixfold improvement in computing power per unit of electricity over five years. Smith described a "virtuous cycle of AI for energy and energy for AI," where efficiency gains in AI infrastructure feed back into tools utilities can use, such as security-constrained optimized power flow modeling.

Smith closed by recounting an anecdote about a longtime head of engineering, Barry Feldman, who found he could prototype ideas himself using natural-language tools rather than relying on specialized staff to translate his expert judgment into code.