WASHINGTON (TND) — A new study from Goldman Sachs Research says electricity demand from data centers will increase by 160% by 2030, driven primarily by artificial intelligence.
Energy lawyer Krista Castaneda said the bandwidth required to run an AI search is much larger than people realize.
“There are more and more computers and it takes more and more power to run them,” Castaneda said.
According to the International Energy Agency, one question on ChatGPT requires 2.9 watt-hours of power, compared with 0.3 watt-hours for a Google search.
“Many challenges in power management will remain over the next decade or so as we adapt to new realities,” Castaneda said.
Michael Nowatkowski, a cyberscience professor at Augusta University, said using all this extra power for AI could create some problems.
“Depending on the demand on the grid, this could lead to more severe brownouts and blackouts,” Nowatkowski said.
Fariba Mamagani of the Tulane University Business School said prices in the electricity market are determined by demand: when demand rises, so do prices.
“Electricity prices are going to go up and they are going to see an increase in their electricity bills,” Mamagani said.
The impact will be felt most by people who live near data centers in places like Northern Virginia, Dallas and Silicon Valley.
This is a concern as in such a scenario it will impact end users,” Mamagani said.
According to Goldman Sachs Research, by 2028, AI is expected to account for approximately 19% of total data center electricity demand.
