Red Tape Shrouds Creative Data Center Power Arrangements
Last Friday, the Federal Energy Regulatory Commission (FERC) hosted a technical conference on co-locating electric loads with power plants. Hours later, the agency issued a decision rejecting an interconnection service agreement (ISA) to increase power at a data center sited at a nuclear plant. This ruling cast shade on the outlook for co-location proposals, catching the capital market by surprise and devaluing companies by billions of dollars; however, the stated rationale was rooted in a particular legal sufficiency concern rather than a desire to stymie co-location. The better policy indicator is the direction in which FERC commissioners steered conversation at the conference.
The language used in FERC’s ISA decision is vague and creates commercial uncertainty. Commissioner Mark Christie, who joined the majority’s opinion, said in a separate concurrence that the ISA filing is “rejected without prejudice” and that the “specific co-location arrangement proposed here may make sense and be acceptable under the Federal Power Act, but on this record that claim simply has not been proven.” Importantly, case law has shown that FERC sets a high legal bar for justifying deviations from standardized interconnection agreements—including the non-conforming ISA proposed in this case. For this reason, the agency’s decision largely hews with esoteric ISA precedent rather than setting policy regarding specific co-location configurations.
Although unintentional, FERC’s ISA decision shook investor confidence. Stocks of competitive generators interested in co-location opportunities plummeted after the decision; however, the way in which FERC ultimately pursues broad co-location policy will dictate long-term market confidence.
Commissioners hinted at their interests during the conference. Although some degree of appreciation for co-location as a creative market response to pressing conditions was evident, panelists and commissioners raised two main areas of concern with co-location: 1) inducing cost shifts to other customers and 2) undermining regional resource adequacy.
Islanded systems remain uneconomic or unreliable at this scale, meaning that co-location configurations will remain connected to the central grid for the foreseeable future. Because co-location uses the central network for transmission and services like balancing power, some are concerned that co-located customers may not pay their fair share for network service.
As noted in R Street testimony delivered at the FERC conference, consumers should pay for the portion of network service they use. If they do not, the proper policy response is to refine cost allocation and rate design—not to force customers to use a network in a defined manner. In fact, co-location likely reduces network costs compared to building new loads elsewhere. Customers who do not participate in co-location are better off letting new customers creatively avoid triggering higher network upgrade costs (which all customers share in paying) while ensuring cost recovery reflects cost causation. This was underscored in a letter representing customer classes of both stripes, which was filed in the conference record.
The resource adequacy complaint is that co-locating new demand at an existing power plant reduces supply to the central market; however, this is largely overstated. As we have written before, one must consider the counterfactual—that is, a bilateral agreement between an existing power plant and a consumer that has the same aggregate supply-demand effect on the central system as if the consumer had purchased power straight from the central market. The one valid consideration is timing. As a grid operator representative remarked at the conference, co-location could accelerate the arrival of new demand. Although noteworthy, accelerating the arrival of new demand by even a few years is not a huge concern. There are sufficient reserves until late this decade, which is when new supply responding to market signals would normally become available.
Indeed, data centers are pursuing co-location in part because contracting for new supply is otherwise very difficult. Some of this is due to market constraints, such as supply chain problems. But the main issue is that while markets are signaling for new power plant development, backlogged generator interconnection queues, a saturated transmission system, and permitting/siting restrictions are artificially stretching the lead time for new development from a few years to a decade. If FERC permits co-location gatekeeping, it will merely create red tape for new consumers in an attempt to offset a problem created by red tape for power plants.
Moving forward, FERC can restore market confidence and improve outcomes for all customers by reaffirming its commitment to lowering barriers to new supply and demand—including co-location—while ensuring any network service cost allocation occurs via a broad policy vehicle. States should do the same within their jurisdictions while reaffirming their commitment to markets as the best-demonstrated approach to securing reliable, least-cost service.