The Contract at the Edge of the Grid

The Federal Energy Regulatory Commission’s June orders do not answer that question once and for all. They open it. On 18 June 2026, FERC issued six tailored show-cause orders to the regional grid operators under its jurisdiction, asking each to justify or reform tariff provisions for data centres, manufacturing facilities and other large energy users seeking connection. The orders address several different problems, including study processes, cost transparency, co-location and flexible loads. They are regional proceedings, not a completed national allocation rule.

That distinction matters. A forecast of new demand can influence a decision about shared infrastructure before the demand exists as actual use. The question is not simply whether a project should be connected. It is when an expectation becomes an obligation: who must demonstrate that the projected load is real enough to plan around, and who carries the exposure if it takes less service than expected? FERC has put that question before six systems, but has not imposed one answer on all of them.

The New York order makes the issue unusually concrete. FERC preliminarily found that NYISO, the relevant transmission owner and the eligible customer taking transmission service on behalf of a large load should enter a cost-recovery agreement. It proposed a minimum contribution toward the transmission owner’s revenue requirement if the eligible customer ultimately takes less transmission service than anticipated because the large load does not materialize as planned or is not developed at all; the agreement should include sufficient credit support. NYISO must either justify why such an agreement is unnecessary or propose one. These are preliminary findings in a show-cause proceeding, not a final rule. The order expressly says this finding does not call existing regional network-upgrade cost-allocation methods into question.

The named counterparty is not necessarily the data-centre operator. FERC’s preliminary formulation identifies the eligible customer taking transmission service on the large load’s behalf, together with NYISO and the transmission owner; the order separately asks whether a direct agreement with the large-load customer could provide comparable or greater protection. That is a legal distinction, not a footnote. The proposed agreement can identify who owes a contribution to the transmission owner. It does not, by itself, tell us who ultimately bears that obligation after contracts, prices or other arrangements are considered. The first question is answered by tariff language. The second requires evidence the order does not supply.

Confusing those questions produces a convenient fiction: that naming a payer settles the distribution of cost. It does not. A cost-recovery agreement may make one obligation more legible to the parties who sign it; it cannot establish, without further evidence, whether another party absorbs, shares or passes along that cost. Nor does the NYISO order say that households have already paid for a failed project, or that any particular load has failed. The careful account is less dramatic and more useful: regulators are deciding how to assign financial exposure when projected use and actual service diverge.

The broader orders also resist a single abstraction. FERC asked about application and study processes, cost shifting and transparency, behind-the-meter arrangements, flexible service, and how to study generation serving nearby or co-located loads. The orders vary by region. That variation is not proof that local discretion is either fair or unfair. It does make the governing principle harder to see. If every region writes different terms, the public still needs to know what those terms make visible: which demand is treated as credible, which party must provide security, and which costs remain shared.

Planning need and payment responsibility must also be kept apart. The Department of Energy’s draft National Transmission Needs Study identifies a need for additional transmission based on load growth, interconnection and congestion, drawing on public information and more than 120 recent reports. It is intended to inform planning; it does not choose particular projects or decide who should pay for them. A national system can need more capacity while a regional tariff still has to decide how a particular risk is assigned. “The grid needs investment” does not answer “who owes when a forecast is wrong.”

The important test for any eventual agreement is therefore not whether it uses the language of a fair share. It is whether the rule discloses the chain of responsibility: whose forecast triggers planning, which customer is legally accountable, what happens when actual service falls short, and what remains unresolved about ultimate incidence. The June orders begin that inquiry. They do not settle it. Before we call a cost private or collective, we should ask what the contract actually assigns—and what it leaves outside the frame.