Study Maps Carbon Storage for Data Centers
For ML infrastructure teams, electricity sourcing and emissions accounting are becoming material constraints alongside compute availability. A study in ACS' Energy & Fuels estimates that US data center power demand could rise from 40 GW in 2025 to 169 GW in 2030, while annual CO2 emissions from fossil-fueled data center electricity could increase from 90 million to 404 million metric tons, according to ACS and Futurity. Researchers Hon Chung Lau and Steve C. Tsai mapped announced US data centers against saline aquifers and estimated that underground carbon capture and storage could store up to 90% of emissions. ACS reports that 34 states have enough saline-aquifer capacity for more than a century of storage. Lau told Futurity that data-center power is "one of the defining energy challenges of the AI era."
Compute growth meets power-system constraints
What the mapping found
The researchers mapped data centers against deep saline aquifers, underground rock formations containing salt water that can retain injected CO2. ACS reports that the analysis found storage capacity sufficient for more than a century in 34 US states, and estimates that up to 90% of data-center CO2 emissions could be stored underground.
Futurity reports projected data center growth in Texas, Virginia, Pennsylvania, Ohio, Arizona, Colorado, Utah, and Illinois. It also reports that Texas alone could require approximately 25 GW of additional power capacity by 2030 under the study's projections.
Lau told Futurity: "Data centers are becoming one of the defining energy challenges of the AI era. The question is not only whether we can build enough computing infrastructure, but whether we can power it in a way that is reliable, affordable, and compatible with decarbonization goals."
ACS quotes Lau describing natural-gas combined-cycle plants equipped with CCS as "the best way" to provide data-center power while preventing CO2 emissions. That is the study author's assessment, not a demonstrated deployment outcome.
Implications and boundaries
Editorial analysis
The paper provides a location-based scenario analysis, not evidence that CCS has been built for specific data centers or that its estimated storage potential is economically or operationally available. The distinction matters for infrastructure teams comparing procurement options, because geological capacity alone does not establish capture equipment availability, CO2 transport arrangements, permitting, financing, or plant-level capture performance.
For practitioners
AI capacity planning increasingly intersects with grid reliability, power procurement, and carbon accounting, rather than GPU availability alone. Industry context: Comparable large-load deployments often require teams to evaluate the emissions profile of marginal electricity generation, especially where new demand is served by fossil generation rather than existing low-carbon supply.
A study published in ACS' Energy & Fuels by Hon Chung Lau and Steve C. Tsai examines whether carbon capture and storage (CCS) in deep saline aquifers could reduce emissions associated with US data center electricity demand. The researchers analyzed publicly available information on announced US data centers, including projected capacity, energy sources, and locations, Futurity reports.
According to ACS and Futurity, the study projects US data center power requirements to grow from 40 GW in 2025 to 169 GW in 2030. The authors estimate that, if fossil fuels powered all US data centers, annual emissions would rise from 90 million metric tons of CO2 in 2025 to 404 million metric tons in 2030.
The study reinforces the value of incorporating location and grid mix into AI workload and capacity planning. In markets where incremental generation is fossil-based, operational carbon metrics can differ substantially from metrics based on annual renewable-energy matching. CCS is therefore relevant as one possible power-system variable in data-center emissions models, alongside generation source, load shape, grid geography, and transmission constraints.
Key Points
- 1Researchers project US data center demand at 169 GW by 2030, making electricity sourcing central to AI infrastructure planning.
- 2The study estimates saline aquifers could store up to 90% of relevant emissions, but storage capacity is not deployment evidence.
- 3For practitioners, location-aware carbon accounting becomes more important when new AI load drives marginal fossil electricity generation.
Scoring Rationale
The study quantifies a significant potential emissions pathway for rapidly expanding US AI and data center power demand. It is a scenario analysis rather than a deployed technical product or policy change, but it is relevant to infrastructure, sustainability, and capacity-planning teams.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
