2025-07-17
Live updates: Celtics, Lakers facing off in Summer League
Utility companies are closely watching the demand the searing temperatures are putting on the power grid. Technology like artificial intelligence has also put more demand on the system. The Eversource power distribution operations center in Dorchester is filled with real-time information where crews monitor the power grid to assure it runs as designed. “The blue line is the forecasted energy use, and the lighter line is the actual,” said Doug Foley, the president of electric operations for Massachusetts for Eversource, while showing NBC10 Boston real-time energy consumption data. Amid scorching heat on Wednesday, Eversource called on its partners who voluntarily agreed to reduce their consumption during high-use days. “That was about 80-100 megawatts that our customers reduced in order to match generation,” Foley said.The balancing act has gotten trickier with AI tools. “A ChatGPT AI query actually uses 10 times the amount of energy as a basic search through Google or Yahoo,” said Mathew Carrara, president of Doble Engineering. The Boston-area company works with utilities like Eversource to help determine if their system needs tweaking to keep up with demand. “We basically monitor the equipment that is in play and say, ‘Hey, you are operating at a safe level. You should be okay with a certain amount of demand depending on what that demand could be,’” Carrara said.Vincent Lopopolo, of Holden, says he uses power and AI: “I wouldn’t have graduated college without AI. I had no idea it was in big demand. I thought it was just like using a google search.” Carrara says AI relies on supercomputing to be able to run a query in a matter of seconds. “Every time someone uses an AI format, the drain on the power grid is instant and because we don’t know when someone is going to do a ChatGPT search, the drain on the grid is unknown,” he said. “It can happen any time of the day, 24 hours, so AI and data centers as a whole are having a lasting impact on the power grid.”Carrera calls it one question that even AI may need a few years to figure out: “How do you generate this same type of intelligence but using less energy.” There are companies, Carrara said, trying to make the technology more efficient, but it is likely years away.