The Hidden Regulators: Public Utility Commissions and AI Governance

View the full PDF report.

Executive Summary

Suddenly, talk of data centers is everywhere. As artificial intelligence (AI) companies announce hundreds of billions of dollars in investments, many AI companies and industry analysts are claiming that AI innovation and development are now bottlenecked by a lack of infrastructure. Similarly, tech companies, academics, and policymakers argue that for the United States to maintain its AI dominance, it must significantly increase the number and capacity of data centers. 

At the same time, recent research has examined the financial and environmental costs associated with new data center buildouts. Researchers and advocacy organizations are expressing concern that rapidly adding massive amounts of demand to the energy grid not only threatens grid stability but also risks increasing energy costs borne by other customers, harming the environment, and negatively impacting communities. Given these concerns, new data centers are finding growing opposition from advocacy groups, state policymakers, and even voters. 

Within this chaotic and contested landscape, state Public Utility Commissions (PUCs) are emerging as influential players in AI regulation through their significant, albeit indirect, oversight of data centers. 

PUCs are public bodies that oversee monopolistic utility companies that work to “ensure safe, reliable, affordable, and environmentally responsible utility service.” They set the prices that regulated utilities can charge customers, approve infrastructure upgrades, and make rules surrounding how utilities operate in states. 

PUCs do not directly regulate data centers; instead, they regulate the utilities that serve them. Yet, this role gives PUCs significant influence over how data centers operate and the costs they bear. But to understand and influence AI governance, we must consider not only the infrastructure — such as data centers — that enables AI, but also examine those who control and shape that infrastructure. 

This report examines how PUCs are becoming important AI regulators while introducing PUCs and the wider energy regulator system for AI policy audiences. 

Through interviews with commissioners and Commission employees across 12 states and a deep dive into how PUCs have been approaching data centers, we outline the specific powers that PUCs have to shape data center and AI policy, discuss key debates involving PUCs, and consider recent efforts to expand PUCs’ regulatory authority over data centers. 

We recognize five types of authority that PUCs have over data centers: rulemaking, rate setting, permitting, planning, and advising. Across these areas, PUCs have unique power to shape where, how, and if data centers are built. 

PUC’s indirect oversight of data centers provides an alternative route for the AI policy community to shape AI policy. AI policy professionals can bring a nuanced understanding of the trade-offs associated with data center development to PUC proceedings. That engagement, however, will require greater visibility into PUC processes and hearings, which, though often public, are hidden behind clunky websites, obtuse proceeding names, and arcane rules. 

PUCs were first established to be public advocates, protecting consumers and communities from the excesses of powerful utilities. As we navigate a path forward in innovating smart, fair, and safe AI, PUCs can be valuable resources and important partners for the AI policy community.