The US government has spent the better part of a year trying to solve one specific problem: data centers need to connect to the electricity grid, and the process for doing so is badly broken.
The Federal Energy Regulatory Commission, which oversees interstate electricity, ordered the country's largest grid operator, PJM, to rewrite its rules in December 2025. PJM runs the electricity system for 65 million people across 13 states and Washington D.C. Before those rule changes, there were no clear procedures for how a data center could share a grid connection with a power plant or run its own backup generators. Different utilities were making different decisions. The result was delays, confusion, and rising costs for everyone.
In April 2026, FERC committed to issuing national standards covering all large electricity users by the end of June 2026. The rules target any facility consuming more than 20 megawatts, which includes virtually every large AI data center being built today.
The core idea behind the new rules is co-location: a data center sits next to a power plant and draws electricity directly from it, bypassing the congested main grid. FERC has also said plainly that whoever requests the connection pays for the infrastructure needed to make it happen. That is a meaningful shift. Previously, those costs had been quietly spread across all electricity customers.
But here is the part that gets overlooked in the policy announcements. Faster connections do not produce more electricity. The shortage is real. At the end of 2023, the queue of power plants waiting to connect to the grid was already longer than the entire existing grid capacity. Data centers are expected to consume nearly three times as much electricity by 2035 as they do today, rising from roughly 40 gigawatts to over 100 gigawatts. To put that in perspective, 100 gigawatts is about ten times what New York City uses at peak summer demand.
The grid is already feeling it. Wholesale electricity prices in the PJM region rose 56% in a single year, reaching $80.5 billion in total costs for 2025. The region's independent market monitor has said the price impacts are large and, in their words, "not reversible." Capacity costs, the payments made to power plants just to stay available, rose 262% in one year. PJM's most recent capacity auction also, for the first time in the grid operator's history, failed to meet its reliability target.
Those wholesale costs pass through to businesses and households. In Washington D.C., residential electricity bills rose by roughly $21 per month in mid-2025. States from New Jersey to Indiana have seen rate increases become a political issue cutting across party lines. An analysis by PJM's independent monitor estimated that data center demand drove $23.1 billion in additional market costs across the last three capacity auctions combined.
For any business that pays an electricity bill across the Mid-Atlantic, Midwest, or is expanding operations in those regions, this is not a distant technology story. It is a cost input that is already moving and is likely to keep moving. The new FERC rules may eventually reduce delays for data center developers, but they do not add a single watt of generating capacity to the grid. The supply problem remains open, and the costs of that problem are being distributed broadly.
The Trump administration's simultaneous decision to pay $765 million to cancel offshore wind projects, replacing that capacity with natural gas plants, adds a further layer. One of the cancelled offshore wind projects would have generated up to 2.4 gigawatts at peak, enough for roughly 1.8 million homes. The administration has now spent about $2.6 billion to cancel offshore wind developments in total. Whether the replacement gas plants arrive fast enough to ease the supply crunch is a separate question that the regulatory announcements do not answer.