INTERCONNECT SEQUENCING AS THE PACING FUNCTION: Substation Topology, Queue Position, and Site Viability for AI Infrastructure at Hyperscale
A Reference White Paper for Utility Coordination, Regulatory Engagement, and Capital Discipline in Interconnect-Constrained AI Infrastructure
Abstract
This paper addresses interconnect sequencing as the pacing function for hyperscale AI infrastructure deployment. The central proposition is that the binding schedule constraint on a hyperscale program is now most often outside the operator’s fence rather than inside it. Interconnection studies, transmission expansion, substation upgrade, and multi-party regulatory coordination operate on multi-year clocks that exceed the multi-quarter clocks of operator-side procurement, construction, and commissioning. The asymmetry has the consequence that the program schedule is paced by the utility-domain clock rather than by operator planning. Operators that treat the program as paced by the operator-domain clock are exposed to schedule slip that the asymmetry produces.
The paper develops the proposition in eight parts. Part I establishes the foundation: the pacing function inversion, why interconnection now determines site viability, and the substation topology reframe that shifts site-selection criteria. Part II addresses the utility domain including utility planning and transmission expansion, queue position as strategic asset, multi-party coordination across utilities and regulators, and the NERC, FERC, and reliability framework. Part III addresses site selection under interconnect constraints. Part IV addresses pre-application engagement and the interconnection study sequence. Part V addresses capital and schedule implications including behind-the-meter mitigation. Part VI addresses portfolio strategy across multiple sites and regions. Part VII addresses governance integration. Part VIII synthesizes recommendations and industry implications.
The paper is positioned for FCG advisory and governance use. It is calibrated to programs operating at densities at or projected to exceed two megawatts per rack with campus capacities in the 50-megawatt to 500-megawatt range during the 2026 through 2028 deployment window. The paper distinguishes verified facts, considered analysis, structured inference, and explicit assumption throughout, and includes appendices with engagement templates, study-process references, site-selection scoring, and references to the regulatory and reliability frameworks discussed.
Executive Summary
Interconnection has become the pacing function for AI infrastructure deployment at hyperscale. The shift is structural rather than cyclical. Pending interconnection capacity in major United States Independent System Operator and Regional Transmission Organization queues has grown roughly an order of magnitude in five years. Transmission expansion timelines now routinely exceed construction timelines for operator-side facilities. Substation upgrade and capacity expansion are more often the binding constraint on site viability than land cost, climate suitability, or labor availability. The traditional sequence in which an operator selected a site, designed an architecture, and requested utility interconnection at a late stage no longer matches the operating reality of the AI infrastructure transition.
The current operating reality inverts the sequence. Operators with discipline begin with interconnection intelligence — utility queue position surveys, substation capacity assessments, transmission expansion schedules — and select sites from the filtered set of locations where interconnection is feasible on the program’s required schedule. The inversion is not optional for operators that intend to deploy at hyperscale-relevant volumes during the current cycle. Operators that operate on the historical sequence will encounter feasibility surprises late in the program when reversibility is most limited and capital exposure is greatest.
The paper develops the inversion across four operational dimensions. The first dimension is queue position as strategic asset. Operators that hold queue position in regions with active demand have a structural advantage that operators without queue position cannot replicate without entering the queue and waiting. The asset is measurable, transferable in some structures, and tradeable through portfolio reallocation. The second dimension is substation topology. The substation that serves a hyperscale campus is no longer a passive utility-side facility; it is a co-engineered asset whose topology, capacity, and expansion path determine the campus’s near-term and long-term viability. The third dimension is multi-party coordination. The hyperscale interconnection involves the operator, the utility, the ISO or RTO, the state public utility commission, the Federal Energy Regulatory Commission, the local authority having jurisdiction, environmental authorities, the land owner, and the surrounding community. Each party operates on its own schedule and authority, and misalignment between parties produces the schedule slip that the operator absorbs. The fourth dimension is the regulatory and reliability framework. NERC reliability standards, FERC orders, and the tariff structures of individual ISOs and RTOs establish the rules within which the interconnection process operates. Operators who do not engage these frameworks explicitly will encounter rule-driven outcomes that explicit engagement could have anticipated.
Five strategic findings follow. First, the pacing function for AI infrastructure deployment is the utility domain rather than the operator domain in nearly all hyperscale-relevant programs. Second, queue position is a structural strategic asset that operators should secure ahead of architectural commitment. Third, substation topology selection is a joint operator-utility decision that should be engaged at the strategy gate of the broader governance framework rather than accepted as a default. Fourth, multi-party coordination is itself a program activity that requires named ownership, scheduled cadence, and documented engagement protocols. Fifth, behind-the-meter capacity is a meaningful pacing-function mitigation that accelerates partial operating capacity ahead of utility energization but does not eliminate the utility-domain pacing.
Six recommendations follow. First, treat interconnection intelligence as upstream capital governance with named ownership, documented protocols, and review at the strategy gate. Second, secure queue position in regions with active demand before architectural commitment, with portfolio-level coordination across multiple regions. Third, engage utilities on substation topology at the architecture gate rather than at the procurement gate; operators that engage early will shape the topology to the program’s needs while operators that engage late will accept the topology the utility offers. Fourth, structure multi-party coordination as a program function with a named coordinator, defined cadence, and documented engagement records. Fifth, integrate behind-the-meter capacity into program planning at the strategy gate when site conditions and regulatory environment support it. Sixth, integrate the interconnect framework with the architecture governance framework, the standards-alignment framework, and the operating-model framework discussed in companion papers; the four frameworks interact and should be applied together.
The paper is intended as a working reference for executive sponsors, capital committee members, chief architects, supply-chain leaders, and operations directors with responsibility for AI infrastructure programs. Government, regulator, and investor audiences will also find the paper directly relevant because the topic addressed is at the intersection of operator, utility, regulator, and capital-market interests.
Full white paper below
