Federal grid regulators have issued binding orders to every major regional transmission organization in the US. Six operators. One directive. Sixty days to either defend the status quo or propose reforms. The underlying question FERC is forcing into the open: should the infrastructure costs of connecting AI data centers to the grid be shared across all ratepayers, or borne by the companies building those data centers?
According to Associated Press reporting, FERC issued the show-cause orders on June 18 under Section 206 of the Federal Power Act, citing large-load growth from AI data center development as the policy driver. The docket is RM26-4-000, a concrete identifier that allows independent verification through FERC’s public docket system even with the agency’s primary press release URL currently unavailable.
The affected operators, PJM, MISO, SPP, CAISO, ISO New England, and NYISO, collectively manage transmission for the vast majority of US electricity consumers. These aren’t isolated regional actors. Their responses will shape interconnection policy for essentially every enterprise data center planning to site or expand in the US.
The action reportedly followed a 2025 request by Energy Secretary Chris Wright, according to reporting on the orders. Wright’s role as DOE Secretary is confirmed; the specific 2025 request has not been independently verified from primary sources in as of publication.
FERC Large-Load Tariff Reform: Key Positions
What the orders require: Operators must either justify why their current tariffs appropriately allocate large-load interconnection costs, or propose specific reforms. Reporting on the orders indicates five reform categories under consideration, reportedly including co-location arrangements (where data centers connect behind a generator rather than at the grid boundary), behind-the-meter generation, and flexible transmission arrangements. The rulemaking reportedly targets loads above 20 MW, though this threshold hasn’t been confirmed from a primary source.
The ratepayer protection argument is the central tension. FERC’s stated aim, per reporting, is to prevent cost-shifting, ensuring that the grid upgrades required to serve 500-megawatt hyperscaler campuses don’t get distributed across residential and commercial utility customers who have no stake in AI infrastructure. Tech companies want fast, predictable interconnection. Utility consumers want cost isolation. State utility regulators, through NARUC, may push back on federal jurisdiction over what they see as state-level ratemaking authority.
Context: This is the federal layer landing on top of a legislative push that’s already underway. Congress has been considering bipartisan bills addressing AI data center power since at least June 18. The FERC action is administrative rather than legislative, it operates under existing statutory authority. That means it moves faster and doesn’t require Congressional approval, but it also means the reform scope is constrained to what the Federal Power Act permits.
What to Watch
What to watch: The August 17 deadline is real and specific. Watch for PJM first, it’s the largest and most complex operator, and its response will set a practical floor for what “reform proposal” means in this context. CAISO’s response matters differently: California has its own renewable integration rules and state utility commission oversight that may create jurisdictional friction. A legal challenge from state regulators is a plausible third path that delays everything.
TJS synthesis: For enterprise AI infrastructure teams, this is a 60-day window with direct implications for US data center site selection and interconnection planning. If FERC’s reform proposals include new large-load cost allocation rules, the economics of behind-the-meter generation versus grid-connected campuses shift materially. Track the operator filings in docket RM26-4-000 directly, that’s where the practical policy will be made, not in the press coverage.