As artificial intelligence (AI) language learning models proliferate across society and become more advanced and accessible, their draw upon the energy sector has become increasingly pronounced. AI models require steady and constant energy that is sufficient to maintain daily energy load and provide support for peak consumption times.[1] It is hard to overstate AI’s demands on the grid. Computers for AI models are housed in data centers, which require large amounts of electricity to keep running, day and night.[2] To illustrate the energy needs of an AI model, consider that if ChatGPT were to be utilized in the same way that Google searches are, every data center would experience a 10-fold increase in energy consumption.[3] Due to the extraordinary amount of electricity required to train and use AI models, concerns have risen that the growth of AI may impede the renewable energy transition.[4]

AI data centers are not spread out evenly across the United States. Instead, they are pocketed in large swaths of California, Texas, and Virginia, with the highest concentration of data centers in Northern Virginia’s Loudoun County.[5] Nicknamed “Data Center Alley,” this small geographic area twenty-five miles west of the nation’s capital sees more than 70% of the world’s internet traffic and has not seen a single day without data center development in over 14 years.[6] Although Virginia is a state with strong energy infrastructure, only 18.4% of energy consumed by Virginia in 2020 was imported, rather than domestically produced. Energy demands stemming from data centers based in the state are expected to grow in the coming years.[7] Currently, natural gas and nuclear power are Virginia’s largest sources of in-state electricity generation.[8] Nuclear power plants have the capacity to produce enormous amounts of energy with near- or zero-carbon emissions.[9] As a result, the AI industry has grown excited about the prospect of developing nuclear plants—specifically, small modular reactors—as a method to provide consistent electricity to Virginia’s AI data centers.[10]

Recognizing that Virginia’s economy will continue to rely heavily upon the construction and operation of data centers, in large part powered by nuclear-generated electricity, legislators have paved a pathway to zero-carbon, domestic energy production in the Virginia Clean Economy Act (VCEA). Passed in 2020, the VCEA mandates that public utility companies generate 100% of Virginia’s total electric energy from renewable sources by 2050, pursuant to a Renewable Portfolio Standard (RPS) program.[11] Although the VCEA’s definition of “renewable energy” does not include nuclear power, the RPS neatly sidesteps the issue by defining “total electric energy” to be the total amount of electric energy sold to Virginian customers, exclusive of any electric energy supplied from nuclear power plants located in the state.[12] Therefore, although the VCEA prioritizes the development of renewable energy sources in the state, it does so alongside the existing and future development and use of nuclear power plants.

Further statutory innovations in Virginia pave a path forward for utility companies to contract with third parties to build the energy infrastructure required by Virginia’s data center industry. During the 2024 legislative session, Virginia Governor Glenn Youngkin signed into law Senate Bill 454, allowing Dominion Energy to petition to recover costs from consumers associated with developing a small nuclear reactor project.[13] In a shift from the VCEA, the law allows up to 80% of the project development costs to be recovered from consumers prior to beginning construction.[14] Virginia’s statutory scheme has paid dividends already: Dominion Energy has announced plans to build at least one small nuclear reactor with Amazon, and more deals are likely to follow in the coming years.[15]

Despite the necessity of zero-carbon energy sources to power Virginia’s data centers, the cost-recovery mechanism provided for by the bill could prove troubling for Virginia residents. It is likely that even with heavy Amazon investment, consumers will see increased rates related to this project, raising questions of fairness. As the AI industry grows and requires more energy, Virginia electric grid customers will, in essence, be paying for its development.

However, Virginians also reap the benefits of data center concentration in their state, as data center-related activity is expected to add $9.1 billion annually to Virginia’s GDP, create 74,000 jobs, and increase tax revenues from the industry.[16] Although due consideration should be given to the fact that consumers will bear the higher rates, it should be weighed against the fact that data centers form the backbone of Northern Virginia’s economy, providing many benefits to those same consumers, and that generation facility construction is necessary to meet Virginia’s electricity demand and climate goals. To balance the aim of ensuring consumers pay just and reasonable rates with the goal of developing zero-carbon electricity generation sufficient to power the state’s data center industry, lawmakers should carefully consider how utilities are incentivized to construct nuclear power facilities.

 

[1] Energy Now, Why AI Consumes So Much Energy and What Might Be Done About It, Kleinman Ctr. for Energy Pol’y (Sept. 24, 2024), https://perma.cc/A8RR-SKQY.

[2] AI is poised to drive 160% increase in data center power demand, Goldman Sachs (May 14, 2024), https://perma.cc/A36Q-M6NZ.

[3] See Lauren Leffer, The AI Boom Could Use a Shocking Amount of Electricity, Scientific American (Oct. 13, 2023), https://perma.cc/9WJB-XGQF.

[4] See Va. Joint Legis. Audit and Rev. Comm’n, Report to the Governor and General Assemblies of Virginia: Data Centers in Virginia at 30 (2024), https://perma.cc/8XKB-2J2P.

[5] See USA Data Centers, Data Center Map (last visited Feb. 21, 2025), https://perma.cc/39R9-URVV.

[6] Data Centers, Loudon Cnty Econ. Dev. (last visited Feb. 21, 2025), https://perma.cc/YVR3-ZJK9.

[7] The Commonwealth of Virginia’s 2022 Energy Plan, Va. Dep’t of Energy, (Oct. 3, 2022), https://perma.cc/PVL5-KDL6.

[8] Virginia State Profile and Energy Estimates, U.S. Energy Info. Admin. (last visited Feb. 21, 2025), https://perma.cc/XK9R-FRPR.

[9] See Seokki Cha, The potential role of small modular reactors (SMRs) in addressing the increasing power demand of the artificial intelligence industry: A scenario-based analysis, Nuclear Eng’g and Tech. (Nov. 12, 2024).

[10] Id.

[11] Va. Code Ann. § 56-585.5.

[12] Va. Code Ann. § 56-576.

[13] 2024 Va. Acts; Senate Bill 454.

[14] Id.

[15] Dominion Energy and Amazon to explore advancement of Small Modular Reactor (SMR) nuclear development in Virginia, PR Newswire (Oct. 16, 2024, 9:00 AM), https://perma.cc/GA65-T5T2.

[16] See Va. Joint Legis. Audit and Rev. Comm’n, supra note 4.