FERC Pushes Grid Operators To Speed Power Access For AI Data Centers

The next phase of the AI race hinges on something far less glamorous than algorithms or GPUs: the electrical grid. On June 19, 2026, the Federal Energy Regulatory Commission (FERC) issued landmark orders directing grid operators across the United States to dramatically streamline interconnection processes for data centers serving artificial intelligence workloads. The move signals that Washington now views AI infrastructure as a national priority — and that electrical capacity, not just computational power, has become the critical constraint.

FERC’s orders target the slowest part of building an AI data center: getting permission to plug into the high-voltage transmission network. Interconnection queues at regional transmission organizations (RTOs) and independent system operators (ISOs) have swelled to nearly 2,000 pending projects nationally, with median wait times exceeding four years. Data centers — which can demand 100 to 500 megawatts each, comparable to a small city — have exacerbated the backlog. FERC’s directive mandates that grid operators use “expedited processing” for facilities that serve AI or high-performance computing, including simplified study requirements and shortened comment periods.

The commission also ordered RTOs and ISOs — including PJM, MISO, CAISO, and others — to submit compliance plans within 90 days. The plans must demonstrate how they will prioritize data center interconnections without undermining grid reliability or unfairly disadvantaging other generation projects. FERC Chair Amanda Smith said the agency is “ensuring that the regulatory framework keeps pace with the speed of innovation.” The order follows similar moves by the Department of Energy and the White House Office of Science and Technology Policy to fast-track AI data center permitting.

Behind the policy shift is a staggering demand forecast. The Electric Power Research Institute projects that AI data centers could consume 9% of U.S. electricity by 2030, up from less than 2% today. Hyperscalers like Microsoft, Amazon, and Google have announced renewable energy deals to offset their carbon footprints, but those purchases only help if the transmission lines exist to deliver the power. FERC’s order explicitly allows grid operators to consider “co-located generation” — including behind-the-meter solar, batteries, and even small modular nuclear reactors — as part of an expedited interconnection plan.

Critics warn that fast-tracking data center hookups could push other projects — especially wind and solar farms — to the back of the line, potentially slowing the clean energy transition. Environmental groups argue that the order lacks sufficient oversight on emissions and grid reliability. “FERC is essentially granting AI data centers a regulatory fast pass that could strain the grid and increase reliance on fossil fuel peaker plants,” said Rachel Stone of the Clean Grid Alliance. Proponents counter that AI data centers, often designed with battery backup and demand response capabilities, can actually improve grid flexibility if integrated properly.

The long-term implications extend beyond energy policy. By reducing interconnection delays from years to months, FERC is effectively lowering the cost and risk of building AI infrastructure in the United States. That could tilt the global AI race — tipping investment and innovation toward domestic hyperscalers rather than competitors in regions with slower regulatory environments. It also pressures utilities to accelerate transmission buildout, long considered the single greatest barrier to decarbonization.

What happens next is a rapid compliance sprint. Grid operators have until late September 2026 to submit their plans, with implementation expected by early 2027. A congressional hearing on data center energy demand is already scheduled for July. Meanwhile, developers are expected to flood interconnection queues with new data center applications, testing FERC’s promise that processing will actually speed up. The order may also spark similar actions in Europe and Asia, where regulators are watching the U.S. experiment closely. One thing is certain: the AI race now runs on electrons as much as algorithms — and FERC just turned on the power.