FERC Just Gave AI Data Centers a Fast Lane to the Grid

On June 18, the Federal Energy Regulatory Commission gave six regional grid operators 60 days to prove their interconnection rules can handle AI data centers — or rewrite them. The unanimous show-cause orders target every major U.S. market except Texas, covering roughly 200 million people and two-thirds of FERC-jurisdictional electricity demand. For the first time, federal regulators treated AI power demand as a national competitiveness emergency — not a local permitting spat.

Why now: Energy Secretary Chris Wright directed FERC in October 2025 to fix large-load interconnection, calling delays an "urgent" threat to U.S. AI leadership. Nine months and 3,500 pages of comments later, Chair Laura Swett issued six tailored orders under Section 206 of the Federal Power Act — a faster legal lever than slow rulemaking. Swett called it "historic action to push our country's electric markets and economy into the future — a future of fair cost allocation, unprecedented transparency for the American ratepayer, respect for states' rights, efficient markets and speed to power."

The bottleneck: Lawrence Berkeley National Laboratory found that by the end of 2023, projects in U.S. interconnection queues totaled roughly 2.6 terawatts — more than twice existing generating capacity. Data centers were turning to expensive behind-the-meter power out of desperation. PJM, the nation's largest grid operator, saw capacity prices jump from $28.92 to $329.17 per megawatt-day in two years, adding an estimated $9.4 billion in costs — driven primarily by data center load growth, per its market monitor.

What FERC ordered:

• PJM, MISO, Southwest Power Pool, CAISO, ISO New England, and NYISO must defend or revise tariffs within 60 days.
• Each must file a resource-adequacy report within 30 days explaining how enough generation will serve new large loads.
• Five reform areas: faster study processes, anti-cost-shifting rules, co-location and behind-the-meter generation, flexible large-load service options, and proximate generation studies.
• Existing signed deals are meant to stay protected; reforms apply prospectively.

Who's cheering: Nvidia called FERC's action "a win for ratepayers, grid reliability, and American competitiveness" — arguing faster large-load integration will help new customers shoulder infrastructure costs. TD Cowen analyst John Miller said the show-cause approach "accelerated the timeline in which reforms can be deployed" versus slow rulemaking. Former FERC Chair Neil Chatterjee, a Republican, called it "faster and likely more legally defensible than a rule."

What's missing: FERC ordered faster connections. It did not order more power plants online. TechCrunch noted the directive "did not address the shortage of generating capacity." The same week, the Trump administration paid $765 million to cancel offshore wind leases near California, Maine, and New York — money Invenergy said it would redirect toward Midwest natural gas and Western geothermal. Texas's ERCOT grid, home to explosive data center growth, sits outside federal jurisdiction entirely.

The politics: Public backlash over double-digit utility bill increases has made ratepayer protection bipartisan — from Trump to Pennsylvania Gov. Josh Shapiro. FERC's orders explicitly guard against cost-shifting to residential customers. Commissioner David Rosner said data centers that trigger transmission builds but fail to show up must not leave consumers "on the hook." Intel utilities segment leader Prithpal Khajuria put the deeper problem plainly: "Load growth is now outpacing the grid's ability to plan, price, and integrate it under the existing framework."

Convina's view: FERC didn't solve the AI power crisis. It nationalized the blame game. Hyperscalers get a procedural fast lane; grid operators get 60 days to prove their rules aren't "unjust and unreasonable"; ratepayers get promises that someone else will pay for the wires. But wires without watts is just a longer queue with better paperwork. The AI infrastructure trade has spent two years pricing GPUs, chips, and capex cycles — and barely pricing the electron. That pricing gap is now a federal problem. Boards evaluating AI vendors should ask not only about model latency but about where workloads sit on a grid whose regulators just admitted the old rules cannot handle gigawatt-scale intelligence. Speed to power just became as strategic as speed to compute.