Conference Agenda

The rapid expansion of hyperscale datacenters is generating a new class of public-policy problems at the intersection of electric utility regulation, environmental governance, and computing infrastructure design. Datacenters underpin modern communication systems and constitute the core infrastructure enabling today’s AI capabilities. States across the U.S. are increasingly moving beyond treating datacenters as economic-development projects and are developing distinct regulatory frameworks for large-scale datacenter facilities, including special tariffs, interconnection requirements, cost-allocation protections, environmental conditions, and reporting obligations. This shift reflects a structural governance mismatch: the technical characteristics of modern datacenters, such as concentrated and rapidly ramping electricity demand, stringent reliability requirements, backup-power coordination needs, and cooling-system tradeoffs, do not fit comfortably within regulatory frameworks designed for conventional commercial or industrial loads.

This paper asks: given the distinctive technical properties of datacenters, how should policy govern their integration into electricity systems and local resource environments? We answer this question through a comparative analysis of emerging regulatory approaches across multiple states. The methodology combines regulatory document analysis, legislative review, and technical assessment of datacenter operational characteristics to identify policy designs that are technically coherent and equitable to communities in which these datacenters are deployed.

On the electricity side, the analysis examines emerging large-load tariff and service-agreement practices across states, focusing on approaches that address risk mitigation, minimum load commitments, staged energization, clean-energy alignment, and ratepayer protection. We find that the most technically grounded approaches treat datacenters not as passive large loads, but as infrastructure capable of active grid participation when appropriate regulatory structures are in place. The DCFlex initiative, launched by EPRI with utilities, grid operators, and major datacenter operators, illustrates this potential: it aims to demonstrate how datacenters can support grid reliability and operational flexibility, but its central finding is that meaningful cooperation requires more than goodwill. It requires reference architectures, shared definitions of flexibility, and formal communication mechanisms between utilities and datacenter operators around load behavior and interconnection constraints. Texas's SB 6 illustrates the legislative parallel: the law establishes a framework for very large loads in ERCOT that includes cost-sharing, disclosure requirements, transmission-study fees, backup-generation reporting, and both mandatory and voluntary demand-management mechanisms. Together, these examples support the paper's central claim that better policy requires new abstractions and interfaces that formalize utility-operator coordination around load commitments, flexibility options, and reliability obligations.

The paper extends the same analytical logic to water consumption and cooling technology. Cooling-system design is not merely an internal engineering decision: it directly determines the burdens that datacenter growth imposes on local water supplies. Federal guidance emphasizes that cooling choices can significantly affect community water availability, while recent legal and water-sector analyses document growing state attention to datacenter water regulation and the need to assess alternative cooling approaches. The paper identifies a comparative best practice: cooling technologies should be evaluated against community resource availability, with policy frameworks requiring greater disclosure, heightened scrutiny of high-consumption methods in water-stressed areas, and stronger linkage between actual consumption and reclamation efforts and stated goals.

We show that regulation is essential not because it substitutes for cooperation between utility companies and datacenter operators, but because it gives cooperation clear public objectives. Across states, the most salient electricity-policy questions concern who bears the cost of new transmission and generation capacity, how stranded-cost risk is handled, and whether new datacenter demand is aligned with affordability and decarbonization goals. Minnesota’s 2025 law is especially instructive in requiring very large customers to bear attributable and stranded costs while tying service to renewable, solar, and carbon-free standards, while other states have pursued special rate classes, formal studies, or cost-protection mechanisms aimed at preventing cross-subsidization from households and smaller businesses. The comparative lesson is that durable utility–operator coordination depends on explicit rules for cost sharing, transparency, and environmental performance rather than ad hoc negotiation alone.

The paper's contribution is to connect the technical design features of datacenters, such as load shape, interconnection behavior, backup coordination, and cooling architecture, to the emerging state policy instruments that govern them. Rather than framing datacenter policy as a binary conflict between digital-economy growth and community protection, the paper identifies a comparative set of best practices: formalized utility-operator interfaces, stranded-cost protections, clean-energy and environmental performance requirements, and resource-aware cooling and siting rules. We conclude that states are beginning to assemble a more technically informed model of datacenter governance, but that these elements remain fragmented across jurisdictions; a comparative framework can help identify which policy designs are most promising for aligning infrastructure expansion with public goals, and which represent transferable models for federal or multi-state coordination.