Data Centers Drove a 76% Power Spike on America's Largest Grid

The cost of powering America's AI ambitions just landed on the doorsteps of 67 million people. Wholesale electricity prices on the PJM Interconnection, the largest electrical grid in the United States, climbed from $77.78 per megawatt-hour in the first quarter of 2025 to $136.53 in the same period of 2026, a 75.5 percent jump that a federal watchdog has explicitly traced to data center growth.

The Power Price Jump: From $77 to $136 Per Megawatt-Hour

Monitoring Analytics, the independent market monitor for the PJM Interconnection, published its annual state-of-the-market report in May 2026. Its central finding was blunt: power prices in the region have reached levels that cannot be reversed. "The price impacts on customers have been very large and are not reversible," the report stated.

PJM covers all or part of 13 states, including Delaware, Illinois, Indiana, Kentucky, Maryland, Michigan, New Jersey, North Carolina, Ohio, Pennsylvania, Tennessee, Virginia, and West Virginia, plus the District of Columbia. The grid serves roughly 67 million people and is home to the single largest concentration of data centers in the world, centered on the vast cluster that has grown over decades in Northern Virginia.

The price increase is not a short-term spike. Wholesale power costs climbed from $77.78 per megawatt-hour in the first three months of 2025 to $136.53 in the same period of 2026, an increase of $58.75. Monitoring Analytics converted that number into a direct consumer cost estimate: approximately $9.3 billion in additional electricity expenses will land on households, businesses, and public institutions across the region over the coming year.

Monitoring Analytics also cited operational failures inside PJM itself. The report flagged a lack of decision-making transparency and identified software upgrades that had been delayed for multiple years without firm completion timelines, pointing to an institution struggling to adapt its processes to a pace of demand growth it was never designed to manage.

Data Centers Now Drive 40 Percent of PJM Capacity Costs

Capacity auctions are how PJM ensures there will be enough generation available to meet peak demand in future years. Generators and demand-response providers bid into these forward auctions, and whatever price clears the auction flows directly into consumer electricity bills.

In PJM's most recent capacity auction, data centers accounted for approximately $6.5 billion of the $16.4 billion total cost, or about 40 percent of the entire bill. Across PJM's three most recent capacity auctions combined, data center load forecasts drove 45 percent of the combined $47.2 billion in costs. The clearing price in the latest auction hit the cap of $333.44 per megawatt-day, the third consecutive auction to set a record high.

Northern Virginia sits at the center of the demand surge. The region hosts the largest hyperscale data center cluster in the world, and its growth projections have consistently outrun every other demand segment on the PJM grid. The forecast peak load for the 2027/2028 delivery year is approximately 5,250 megawatts higher than the 2026/2027 forecast, and nearly all of that increase, about 5,100 megawatts, comes directly from data center expansion.

AI workloads have accelerated the underlying trend. Training large models and running inference at scale requires sustained, high-density power draws that conventional enterprise computing never placed on the grid. AI infrastructure providers have raced to build data center capacity near existing transmission lines, often faster than grid planners can accommodate new interconnection requests.

The capacity crunch is also a reliability problem. Data centers do not simply raise costs; they strain the physical ability of the grid to meet peak demand. PJM's target is to maintain a 20 percent installed reserve margin above its forecast peak demand. In its most recent auction, PJM procured 145,777 megawatts but fell approximately 6,625 megawatts short of that benchmark, missing the reliability target for the third consecutive year. FERC Chairman Laura Swett called the results "very concerning" and said the commission must "act to ensure that new supply is available to interconnect to PJM quickly enough to meet historically surging demand." The Federal Energy Regulatory Commission has since directed PJM to establish transparent interconnection rules specifically for AI-driven large loads. Cisco recently reported that its AI networking orders doubled its own forecast to $9 billion, illustrating how far up the supply chain that pressure now runs.

Monitoring Analytics Wants Tech Giants to Fund Their Own Power

The most contested element of the report is its proposed fix. Monitoring Analytics argued that data centers should be required to bring their own dedicated power generation to the grid rather than drawing on shared capacity paid for by existing ratepayers.

Under current rules, when a large load connects to PJM, its demand is pooled with all other customers when determining how much generation the region needs to procure. The cost of that procurement is then spread across every ratepayer in the region. Under the Monitoring Analytics proposal, large new loads would need to arrange or construct their own generation resources, removing them from the shared capacity pool and keeping the additional cost off the bills of the millions of households and businesses that have no direct connection to AI infrastructure.

Utilities have not uniformly welcomed the concept. AEP, one of PJM's largest member utilities, threatened to exit the grid over various proposed reforms, indicating how contentious the cost-allocation debate has become among the parties most directly affected.

FERC's directive to PJM stops short of mandating the Monitoring Analytics approach, but it requires PJM to develop clear and transparent rules for how AI-driven large loads are interconnected and priced. The specifics are still being worked out, and the outcome will determine how much of the financial burden falls on data center operators versus general ratepayers in the years ahead.

The structural gap between grid capacity and AI industry demand is not something any single rule change can fully close. According to TechCrunch, the U.S. power grid was not designed for the electricity demands of an AI-driven economy, and the distance between what the grid can deliver and what the AI industry expects keeps widening. Whether that gap closes through new generation investment, demand-side constraints, or regulatory reform will shape the economics of the AI buildout for years to come.