Policy Recommendations: Accelerating AI Data Center Development in the US
GAIA Submission to the BASIS AI Policy Memo Competition on Data Centers and Community Impact
Vedant Srinivasan is a Sophomore at Georgetown University majoring in Science, Technology, and International Affairs. He is interested in studying the impact of AI on the labor market, how the technology will be diffused globally, and how “middle powers” can have a say in Responsible AI Development.
Bhumika Nebhnani is an MPP candidate at the McCourt School of Public Policy, Georgetown University. She currently works at the Center for Security and Emerging Technology (CSET) and Massive Data Institute (MDI) on AI policy and regulations in the US and globally.
On February 2nd, 2025, GAIA Members, Bhumika Nebhnani and Vedant Srinivasan, submitted a policy memo to The Berkeley AI Safety Student Initiative’s (BASIS) US AI Policy Hackathon Competition. This is their response to the prompt:
“How should the federal government accelerate AI data center development to maintain U.S. technological leadership while managing energy, environmental, and community concerns?”
The U.S. commitment to global AI dominance amid intensified competition with China has brought data centers to the forefront of federal policymaking. Data centers have been formally elevated to a national priority, with federal agencies directed to accelerate their buildout. While progress has been made, U.S. deployment remains constrained by a three-layered bottleneck stack: energy generation and electricity transmission, land allocation, and water concerns.
Energy and electricity:
Data centers consumed approximately 184 TWh - about 4% of total U.S. electricity use - in 2024 alone1, making it critical to solve the bottlenecks of energy generation and grid congestion2 to accelerate data centers’ speed-to-power3. To meet data centers’ energy demand without competing with other uses, the United States needs to ramp up its electricity generation. While the Federal Energy Regulatory Commission (FERC)’s July 2023 interconnection reforms reduced interconnection wait times4 for new power plants, they have not resolved the underlying capacity and construction timing issues. Grid congestion results in a loss of $6 Billion for American consumers annually. In addition, congestion also results in significant problems for the renewable sector whose providers supply intermittent energy5 leading to significant curtailment6 and in-turn losses. New transmission lines are necessary but expensive and take a minimum of a decade to build. Immediate Grid Enhancing Technologies7 such as dynamic line ratings and power flow controls, software and hardware interventions that optimize existing electrical transmission, can result in significant capacity gains while costing significantly less. While the federal government has stepped in through programs such as the Grid Resilience and Innovation Partnerships program, more effort needs to be placed on encouraging local utilities to enhance the existing grid.
Land allocation:
On roughly 60% of the US land under the states’ jurisdiction, siting of datacenters is primarily governed by state “police powers” delegated to county governments. While the states initially competed to offer hefty financial incentives, there has been a rising popular backlash leading to the approvals becoming time-consuming or in some cases, stalled. In this light, the executive push to identify federal, brownfields8, and superfund lands9 for siting data centers is pertinent. However, this process has been slow owing to the absence of clear timelines and a lack of site identification. Reusing brownfield and superfund sites may be feasible but requires careful cleanup and community engagement. The Enviornmental Protection Agency’s (EPA) budget cuts may also slow down the brownfields revitalization.
Water:
Hyperscale data centers can consume up to 1.8 billion gallons of water annually due to cooling requirements, equivalent to the annual consumption of a town with a population of approximately 10,000-50,000 people. This has become a flashpoint in water-stressed areas like Arizona where community skepticism about data centers is a significant barrier to the build -out. Several state assemblies have passed legislation demanding greater reporting on data center water-usage (though they have been vetoed by the state Governors).
Policy Recommendations
In the context of the highlighted problems with data center buildout, the following policy recommendations should be considered:
The Federal Energy Regulatory Commission (FERC) should initiate a notice of rule-making and finalize its reforms on incentivizing utilities’ adoption of Grid Enhancement Technologies (GETs):
Since utilities receive compensation as a percentage of their total assets, new lines are valued over increases in efficiency. FERC should exercise its rule-making power under Section 219 of the Federal Power Act to enable utility companies to recoup their investments in congestion reduction10. A new rule would give additional equity to utilities that improve existing lines using GETs. This unlocks previously inaccessible power, lowering the cost of electricity for consumers, enabling speed-to-power for data centers, and resulting in savings for renewable energy plants.
The Department of Energy (DOE) should clarify areas so that FERC can site interstate transmission lines:
While states control most of the ability to site transmission lines, interstate lines can be approved by FERC only when DOE designates them as part of a national energy corridor. DOE must begin this process, focusing on areas with a high data center concentration. By designating these areas, FERC can quickly approve interstate lines that may otherwise receive pushback. Exercising this power would require an extensive consultation process with affected communities which could delay the implementation.
Streamline procedures to maximize the use of Federal Lands:
DOE and the Department of Defense (DOD) should set clear timelines for finalizing the proposals received on federal lands. They should also continue to identify more sites. The Department of Interior (DOI) should develop and harmonize its sub-agencies’ siting-identification criteria11. Further, building the data centers on brownfield and superfund sites may face less opposition from communities as they redevelop used and contaminated lands. So, Congress should explicitly protect and restore EPA’s Brownfields and Superfund‑reuse funding. Also, it should set funding floors for EPA functions12 tied to national data center development goals. This will ensure that federal lands are used to their full potential.
Supporting a Data Center Transparency Act to Clarify the Information Landscape Surrounding Data Centers:
Public skepticism over data centers has grown due to a lack of information about their impacts. The proposed the Data Center Transparency Act before Congress mandates them to regularly report their water and energy consumption. This act would inform Congress of the ground-reality, and the resulting transparency will help to hold the companies accountable. Concern over data centers’ community impacts transcends partisan boundaries and the passage of this bill aligns with the spirit of cooperative federalism.
The demand is projected to grow to 9% by 2030.
The economic impact on the users of electricity that results in physical transmission constraints
The time it takes for data centers to access their needed supply of energy.
Prior to construction, energy suppliers are required to submit an interconnection request which enters them into a queue to study the feasibility of them joining the grid.
Energy sources for which supply is neither constant nor predictable.
The intentional reduction in energy supply due to grid complications
While the FERC attempted to propose such a rule in 2020, they failed to finalize it. If FERC does not exercise this power independently, Congress should step in and pass the Advancing GETS Act of 2025 which would mandate this change to occur.
This will ensure land under the Bureau of Land Management (BLM), U.S. Fish and Wildlife Service (FWS), and National Park Service (NPS) is also constructively assessed for the buildout.
Some functions are site assessment, cleanup decision‑making, and reuse technical assistance.