Business professor and utility expert breaks down the future of energy markets and consumption.
As data centers dedicated solely to artificial intelligence are built at an increasingly rapid pace across the United States, an unprecedented demand for electricity is beginning to strain an already tight power grid, raising concerns about reliability and higher costs for everyday consumers.
“AI’s demand for power is an order of magnitude larger than any other type of data center or server warehouse,” said Richard Michelfelder, professor of professional practice in finance at Rutgers School of Business–Camden.
Michelfelder has over four decades of experience in the electric, gas, water, and wastewater utility industries, including executive roles at Atlantic Energy Inc. and in energy-efficiency and utility-technology ventures. He has published widely and testified before state and federal regulators on utility finance and regulation.
“Demand is outpacing supply, prices are rising faster than inflation, and it will take time to develop new power supply sources, even with policy and industry advocates to push for an increase in reliable energy options,” said Michelfelder. “It is a simple problem with a slow solution, and little net new generation has been developed due to flat forecasts.”
Industry experts estimate that AI queries require about 10 times the energy of a traditional web search, and as more businesses leverage the advantages of AI, demand for energy is expected to increase exponentially. According to a November 2025 Brookings Institution report, data centers consumed around 4.4% of the nation’s electrical power in 2023 and are expected to increase to as much as 12% by 2028. Globally, it is projected that AI’s energy needs could account for up to 21% of total electricity consumption by 2030.
The country’s aging electrical grid, already taxed by routine demand, may not be equipped to absorb the anticipated changes driven by the growth of AI. Michelfelder points to the 17-state region managed by PJM Interconnection, which stretches from the Midwest to the coastal Mid-Atlantic, and serves more than 20% of the nation, both in terms of population and gross domestic product.
“Historically, the PJM grid has always been robust in reliability, with about a 20% reserve margin,” said Michelfelder. “The reserve margin has been shrinking quickly in recent years, however, as supply has been much slower to grow than demand. Last year's annual PJM auction to supply power to the PJM grid resulted in prices that were multiple times higher than recent auctions.”
Michelfelder noted that the result of limited grid capacity could very likely be higher prices and electricity rationing during peak usage periods. There are ways to address these cost-related challenges, however, that Michelfelder believes could ease the burden in the near term as the capacity for additional electricity generation is built.
“Cost-of-service studies are a common practice within the utility ratemaking industry and can provide an understanding of the cost of providing service to different customer classes to ensure fair and equitable rate structures,” Michelfelder said.
Still, Michelfelder cautioned that regulatory fixes alone may not fully address the scale of the challenge. The surge in electricity demand driven by AI development and increased electrification in general raises larger questions about how innovation should be integrated into society without undermining the resources that support overall growth.
“AI shows great promise as a new technology that encourages 'creative destruction, fostering new industries, economic growth, and productivity, even as it makes other industries obsolete,” Michelfelder said. “When it comes to AI, however, it will be essential to address the issue of energy demand and supply, because electricity is key to economic growth in any market or economy.”
