A new study published in Energy & Fuels estimates that U.S. data center power capacity will increase from 40 GW in 2025 to 169 GW by 2030, more than quadrupling within five years as AI computing requirements continue to accelerate.
The analysis, co authored by Hon Chung Lau, adjunct professor in Rice University’s Department of Chemical and Biomolecular Engineering and founder of Low Carbon Energies LLC, and Steve C. Tsai, evaluates whether carbon capture and storage could mitigate the emissions associated with the fossil fueled electricity likely to support much of that expansion.
The researchers estimate that without additional decarbonization measures, annual carbon dioxide emissions linked to electricity generation for U.S. data centers could increase from approximately 90 million metric tons in 2025 to more than 404 million metric tons by 2030. Those projections are based on publicly announced data center developments, projected electricity demand, and the current electricity generation mix in individual states.
The study reflects a growing recognition that the AI economy presents a distinct energy challenge. While many technology companies have committed to renewable electricity procurement, data centers require continuous, highly reliable power that intermittent renewable generation alone cannot always provide without significant storage or backup generation. This has renewed interest in natural gas generation, nuclear power, geothermal energy, and carbon capture technologies as potential complements to renewable energy.
The researchers argue that natural gas combined cycle power plants equipped with carbon capture and storage could provide one of the more practical near term solutions. Compared with coal fired generation, combined cycle gas plants produce lower carbon dioxide emissions while offering the operational flexibility and reliability required by hyperscale data centers operating around the clock.
A key finding is that geological storage capacity does not appear to be a limiting factor across much of the United States. According to the study, 34 states possess sufficient saline aquifer capacity to store more than 100 years of projected data center related carbon dioxide emissions beyond 2030.
Using only in state storage resources, the researchers estimate that saline aquifers could accommodate approximately 59 million metric tons of carbon dioxide in 2025, equivalent to about 66 percent of projected data center emissions. By 2030, available storage capacity could support sequestration of roughly 299 million metric tons, representing about 74 percent of projected emissions.
The mitigation potential increases further when carbon dioxide can be transported across state boundaries. Under that scenario, the study estimates that more than 90 percent of projected data center emissions could theoretically be captured and permanently stored underground.
The geographic distribution of future demand also plays an important role. The study identifies Texas, Virginia, Pennsylvania, Ohio, Arizona, Colorado, Utah, and Illinois among the states expected to experience substantial data center expansion. Texas alone may require an additional 25 GW of electricity generation capacity by 2030 to meet anticipated demand from new facilities.
The concentration of data center growth near regions with suitable geological storage creates opportunities to integrate carbon capture into future energy infrastructure. Many of these states already possess significant natural gas resources, existing pipeline networks, or geological formations suitable for long term carbon dioxide sequestration, potentially reducing infrastructure development costs compared with regions lacking those characteristics.
At the same time, the authors acknowledge several limitations. Their estimates include only projects with publicly disclosed power requirements, meaning total future electricity demand could ultimately exceed current projections. The analysis also assumes that electricity will be supplied from existing state grids when developers have not specified dedicated power sources and that each state’s electricity generation mix remains broadly unchanged through 2030.
Those assumptions make the findings more representative of current market conditions than long term decarbonization pathways. Continued growth in renewable generation, nuclear capacity, battery storage, or other low carbon technologies could reduce emissions relative to the study’s baseline projections. Conversely, AI adoption could expand more rapidly than anticipated, increasing electricity demand beyond current forecasts.
