FERC Large Load Interconnection Show-Cause Order Ignites Federal-State Battle Over Data Center Cost-Shifting
The physical bottleneck of electrical grid integration for gigawatt-scale AI data centers has triggered an intense legal and regulatory battle between federal regulators, state ratepayer advocates, and grid operators. Following FERC's landmark rejection of the Talen-Amazon Susquehanna nuclear co-location deal, the Federal Energy Regulatory Commission (FERC) and state utility advocates have clashed over who should bear the multi-billion-dollar transmission and network upgrade costs required to support soaring data center loads.
FERC's Show-Cause Order on Large Load Interconnection
In late June 2026, FERC launched an aggressive, targeted action, issuing a "show cause" order to the PJM Interconnection and five other major regional transmission organizations (RTOs). FERC declared that existing grid operator rules for interconnecting large loads (such as data centers exceeding tens of megawatts) appear to be "unjust and unreasonable" under the Federal Power Act.
The commission's primary concern is that existing tariffs do not adequately prevent speculative large loads from shifting their massive transmission network upgrade costs onto existing residential and commercial ratepayers.1 FERC ordered grid operators to promptly revise or justify their interconnection rules, acknowledging that resolving these issues:
"...will require the involvement of both federal and state actors, and that states maintain the responsibility to prevent retail cost shifting by, for example, creating new rate classes and properly assigning upgrade costs in retail tariffs."
State Ratepayer Advocates File Rehearing Demands
The federal framework, however, has faced immediate legal challenges from state ratepayer advocates who argue it does not go far enough. In late July 2026, ratepayer advocates from five major PJM states—Delaware, Illinois, Maryland, Ohio, and Pennsylvania—filed formal requests at FERC for rehearing and clarification, asserting that the show-cause order fails to shield consumers from data center transmission costs.
The advocates contend that PJM's regional cost allocation rules guarantee that network upgrade costs caused by data centers are embedded into transmission owners' revenue requirements, making existing consumers foot the bill. In their joint filing, the advocates from Delaware, Illinois, Maryland, and Ohio stated:
“Each network upgrade added to a transmission owner’s revenue requirement while these questions remain open embeds another cost shift, and every cost-recovery agreement negotiated against an unsettled standard invites the disputes the Commission could resolve today... It is arbitrary and capricious and contrary to the [Federal Power Act] for the Commission to conclude that it ‘has a duty to address the risk of cost shifting among transmission customers’ while failing to remedy the actual cost shifts the PJM tariff guarantees due to network upgrades for large loads.”
The Pennsylvania Office of Consumer Advocate filed a separate petition, warning that cost-recovery agreements are insufficient because "unjust and unreasonable wholesale transmission cost shifts from large loads in PJM can manifest through unjust and unreasonable cost allocation of both [Regional Transmission Expansion Plan] projects and supplemental projects."
The "Bring Your Own Power" and Microgrid Pivot
As regulatory hurdles and cost-allocation disputes threaten to delay grid interconnection timelines for years, data center operators are increasingly bypassing the grid altogether. This has accelerated a massive pivot toward "Bring Your Own Power" (BYOP) architectures—such as on-site natural gas microgrids, solid oxide fuel cells (including Oracle's 2.8 GW partnership with Bloom Energy), and behind-the-meter small modular reactors (SMRs)—to achieve grid-independent AI data center deployments.
-
An instance of High-density computing can no longer socialize its grid infrastructure costs. — FERC's show-cause order targets regional transmission operators to prevent massive AI loads from socializing their grid connection costs. ↩︎