Data centers—the backbone of AI and cloud computing—are reshaping electricity markets in ways that go beyond higher demand. This study shows that when these data centers plug in, the congestion and line losses drive up wholesale power prices by about 7%. Co-locating with fossil plants eases the jam but adds emissions; pairing with renewables helps the climate but worsens grid strain. The authors, led by Georgia Tech visiting scholar and EPIcenter affiliate Jamal Mamkhezri, call for smarter planning: expand transmission, rethink incentives, and align private investment with public welfare so the digital boom doesn’t derail grid reliability.
To learn more, read the paper (The Hidden Cost of the Cloud: Data Centers and Electricity Market Inefficiency by Jamal Mamkhezri, Xiaochen Sun, Yuting Yang :: SSRN) or contact us at epicenter@gatech.edu.
KEY TAKEAWAYS
1. Data Center Growth Creates Localized Market Inefficiencies
The study shows that new data center (DC) interconnections raise wholesale electricity prices by 7.3% through congestion and marginal loss components alone (Pages 12–13). This means that DC-driven demand shocks contribute to higher prices via congestion and losses, besides increasing overall load. These effects distort price formation at specific nodes and reveal inefficiencies in grid operations.
2. Transmission Constraints Are the Bottleneck
DCs amplify congestion because transmission lines cannot accommodate concentrated, inelastic loads (Pages 6–8, 15–16). Marginal loss prices also rise as power travels longer distances to meet demand. These findings underscore that grid strain is not only about generation scarcity but also about network limitations, making transmission expansion a critical policy priority.
3. Co-Location Decisions Involve Trade-Offs
Locating DCs near fossil fuel plants reduces congestion but increases emissions; pairing with renewables supports decarbonization but worsens inefficiencies due to intermittency (Pages 13–14). This tension suggests that planning for DC growth must integrate both environmental and network efficiency objectives rather than treating them separately.
4. Behavioral Economics Validates Policy Design
The study reinforces key behavioral economics theories—bounded rationality, rational inattention, and mental accounting—by showing that consumers do not always respond to price signals in predictable, rational ways. This suggests that traditional economic models may underestimate the complexity of household decision-making, and that behaviorally informed policies (e.g., nudges, feedback loops) are essential for effective energy management.
5. Current Incentives Ignore Externalities
States often offer tax exemptions to attract DC investment, but these policies fail to internalize congestion and loss costs that are eventually passed on to consumers (Page 16). This misalignment between private and social welfare calls for coordinated policy design—such as pricing mechanisms or infrastructure funds tied to DC development.
Glossary of Technical Terms
| Term | Definition |
| Locational Marginal Price (LMP) | The price of electricity at a specific grid location, reflecting system energy, congestion, and marginal loss costs. |
| System Energy Price | The base cost of producing one extra megawatt-hour (MWh), uniform across all nodes in the market. |
| Congestion Price | Extra cost when transmission lines are overloaded, preventing power from flowing along the cheapest route. |
| Marginal Loss Price | Cost of additional electricity needed to offset losses during transmission, which rise with distance and load. |
| Propensity Score Matching (PSM) | A statistical method to create comparable groups by matching on similar characteristics, reducing bias in causal analysis. |
| Difference-in-Differences (DiD) | An econometric technique estimating treatment effects by comparing changes over time between treated and untreated groups. |
| Staggered DiD | A variation of DiD used when treatments occur at different times across units (e.g., census tracts). |
| Merit-Order Dispatch | The process of meeting electricity demand by calling on the cheapest available generators first. |
| Event Study | A statistical approach to examine how an outcome changes between treated and untreated group before and after a specific event, validating assumptions like parallel trends. |
| PJM Interconnection | A regional transmission organization (RTO) coordinating electricity movement across multiple states, including Virginia. |
This summary was written with the assistance of Microsoft Copilot on November 12, 2025. Its content was edited and verified by EPIcenter staff and affiliates.