Key Takeaways
AI data centers are huge consumers of electricity and are expected to consume increasing amounts of electricity in the future.
Nationally, data centers are expected to consume between 11% and 12% of total U.S. power demand by 2030 — higher than the 3% or 4% currently.
Virginia and Texas are expected to be the states with the largest data center demand, each with its own issues in adjusting to the demand requirements.
Dominion Energy in Virginia, to meet the growing energy demands, cannot retire its fossil fuel plants, which the Virginia climate law requires by 2045.
In Texas, AI data center projects are requesting to be added to the grid in astounding numbers that the grid operator believes are not only speculative but cannot be met.
Virginia is the hottest state for artificial intelligence (AI) data centers, and Dominion Energy is trying to meet AI electricity demands. However, Virginia’s Clean Economy Act (VCEA), which established a Renewable Portfolio Standard and forces Dominion Energy to generate 100% renewable electricity by 2045, is in conflict. According to the American Action Forum (AAF), “the company indicated in its recent filings that it does not appear viable to fully retire fossil fuel generation to meet VCEA compliance by 2045—a conclusion supported by a separate study commissioned by the commonwealth.” AAF adds that “Dominion’s projected summer peak load [is expected to grow] by 70 percent between 2022–2045.”
The rapid growth of AI data center demand makes meeting carbon-free targets extremely difficult, if not impossible. Climate goals are achievable only if demand is stifled, which would result in industry going abroad, as Europe is experiencing, leaving China to win the AI race.
Texas is a Booming AI Data Center State
Texas, as a red state with cheap land, relatively inexpensive energy, and a business-friendly tax and regulatory environment, is also attracting data center developers to the state. The expected demand is so large that many believe it will be impossible to meet by the end of the decade, according to CNBC. More than 220 gigawatts have asked to connect to the Texas electric grid by 2030, according to data from the Electric Reliability Council of Texas, and more than 70% of those projects are data centers. That demand is more than twice the state’s peak summer demand this year of around 85 gigawatts, and its total available power generation for the season of around 103 gigawatts. CNBC reports that ERCOT worries that speculative projects are clogging up the pipeline to connect to the electric grid, making it difficult to see how much demand will actually materialize. State legislation in 2023 requires projects that have not signed electric connection agreements to be considered in power demand forecasts.
Via CNBC, requests from large projects for connections to Texas’s electric grid have nearly quadrupled this year, highlighting the gap between speculative demand and what is likely to materialize. More than half of the proposed capacity — about 128 gigawatts — has not yet submitted the required studies for review by ERCOT. Another roughly 90 gigawatts is either under review or has received planning-study approval. By contrast, only about 7.5 gigawatts of projects have actually connected to the grid or secured full approval from ERCOT, a level of demand the state can readily accommodate. Texas could realistically supply 20 to 30 gigawatts of data-center demand by 2030, grid planners say, but not the far larger volumes implied by the speculative interconnection requests now on file.
Furthermore, CNBC reports that Texas is trying to reduce speculative requests for connections through a law passed in May that requires developers to pay $100,000 for the initial study of their project, show that a site is secured through an ownership interest or lease, and disclose whether they have outlined the same project anywhere else in Texas. Fortunately for residents, the cost of building new power plants to serve the Texas electric market is generally borne by investors, providing some protection to households from higher electricity prices if too much capacity is built.
Analysis
Because data centers are expected to consume between 11% and 12% of total U.S. power demand by 2030, states are scrambling to find ways to mitigate this load without raising prices for consumers. As Virginia is realizing, climate mandates can’t coexist with a functioning digital economy because AI requires reliable generation to operate. Entrenched utilities will have a difficult time working through the regulatory process to build enough firm generation to meet growing demand. Therefore, policymakers should allow firms to bypass public utilities and build new, independent power infrastructure. As the Cato Institute’s Travis Fisher and Jennifer Huddleston argue, “If the United States wants to lead the world in AI, policymakers must abandon the illusion that a centrally planned grid can power a decentralized technological revolution. Energy abundance and technological progress thrive when they are built on the same foundation: an open-ended market process with less government intervention.”
For inquiries, please contact [email protected].
