A Transparency-Based Approach to Regulating the Resource Footprint of U.S. Data Centers

Artificial intelligence systems increasingly rely on large-scale data center infrastructure that consumes substantial electricity and water, with impacts concentrated in rural and suburban regions where facilities cluster near land and transmission corridors. Yet in the United States, resource-consumption disclosures remain fragmented across voluntary corporate reporting and limited, uneven state requirements, leaving policymakers and communities without consistent facility-level data to evaluate cumulative grid stress, water scarcity risk, and emissions trajectories. This paper proposes a transparency-first federal policy that mandates standardized annual reporting of key resource metrics, including electricity consumption by source, total water use, renewable energy share, and reuse of waste heat and water, for covered data centers with capacity of at least 500 kilowatts operated by U.S.-listed firms and U.S.-based private corporations. Reports would be submitted through a joint Department of Energy and Securities and Exchange Commission portal and aggregated into a public database to enable third-party verification and reduce greenwashing. Because hyperscale facilities drive disproportionate demand, the proposal requires annual accredited third-party audits for hyperscale data centers, with unredacted audit copies retained by regulators and redacted versions published publicly. Enforcement relies on civil penalties calibrated to severity and firm size, leveraging existing SEC authority for public companies and Department of Energy enforcement for covered private firms, while minimizing disclosure of proprietary information. By establishing a uniform national reporting floor, the proposal enables evidence-based future regulation, improves investor and community oversight, and supports sustainable growth in U.S. AI infrastructure. This paper was submitted Dr. Cynthia Bailey's CS121 course: Equity and Governance for Artificial Intelligence and it won the 2025 Eric Roberts Prize for Tech Ethics and Policy.