New Jersey electricity rates are soaring largely because of power-sucking AI data centers, and states must work together to maintain pricing and grid reliability, energy experts said.
The Garden State has about 75 such warehouse-size centers, where computers process and deliver artificial intelligence. AI-powered applications have a range of uses, from simple internet searches to health-care breakthroughs to deep-space exploration. A 100-megawatt data center uses enough electricity to power 100,000 to 166,000 residences.
Because electricity grids extend over state lines, AI data center growth in one region has supply and price effects elsewhere, environmental activists and energy experts told news reporters during an online briefing on April 10.
“ChatGPT receives 29 million requests per day. When Google added the AI search results to every Google query, it added the energy demand of a small European nation to just do that,” said Allison McLeod, deputy executive director of the New Jersey League of Conservation Voters.
Starting June 1, New Jersey electricity rate increases will range from 17% to 20%, depending on customers’ service areas. By 2030, AI processing is expected to tap as much as 4% of global electricity supplies, roughly double the current rate, according to Goldman Sachs Group Inc. research.
While AI data centers place stress on the region’s power grid, the electricity supply hasn’t notably increased. New Jersey regulators are pressing PJM Interconnection, which moves wholesale electricity to local utility companies in 13 states, to rein in consumer costs. The utilities, which bid on electricity generation capacity via PJM auctions, pass on their costs to ratepayers.
“PJM’s recent capacity auction results are the main driver of these increases,” Christine Guhl-Sadovy, president of the state Board of Public Utilities, said in February, when the new rates were announced. “The Murphy administration will continue to aggressively push and hold PJM accountable to address rising costs by expediting interconnection and implementing additional market reforms, all of which will help drive down costs for ratepayers, which is a priority for the BPU.”
PJM has acknowledged a major interconnection project backlog to add energy sources to the grid. In a statement in December, amid a dispute with Pennsylvania over energy costs, the company said it “shares the concern about rising prices caused by this supply-demand imbalance, and we’ve taken multiple steps to mitigate it.”
Pennsylvania and PJM came to an agreement to maintain rates for two years.
McLeod, on the call with reporters, said the PJM backlog is critical.
“There are about 1,600 projects in the queue, about 95% of which are clean energy projects,” McLeod said. “If even 30% of those projects in the queue had been interconnected, the last capacity auction would have been 63% cheaper.”
Long-term improvement to PJM’s energy efficiency and interconnection queue “requires cooperation among the states,” said Michael Winka, a former clean energy director for the Board of Public Utilities.
New Jersey lawmakers have introduced several pieces of legislation to combat the demand and cost issues linked to artificial intelligence.
One bill, A-5267, sponsored by Assemblyman Wayne DeAngelo, a Democrat from Hamilton Township, would incentivize utilities to create energy storage systems that would send power to the grid when demand increases. Amended and approved by the Assembly telecommunications and utilities committee, the bill is awaiting consideration by the Assembly appropriations committee.
Another, S-4143, sponsored by Senator Bob Smith, a Democrat from Piscataway, would regulate AI energy use. Proposed data centers would be required to submit power plans and commit to renewable or nuclear sources. The bill was passed 3-2 by the Senate environment and energy committee and awaits potential amendments.
