Data Center Growth in the Southeast: Impacts for the Region

The Southeast United States is experiencing a data center boom - or is it? While utilities across the region are planning massive infrastructure expansions to meet projected demand, a groundbreaking new study suggests they may be betting on growth that has less than a 0.2% chance of actually materializing.

In our latest report, Data Center Growth Impact on Southeast U.S., developed in partnership with Science for Georgia, Greenlink Analytics brings clarity to a landscape challenged by inconsistent data and inflated claims. What we found challenges the prevailing narrative and raises critical questions about billions of dollars in infrastructure investments.

The Southeast's Data Center Moment

The Southeast continues to emerge as one of the nation's most important data center markets - and one of its most uncertain. According to recent CBRE market assessments, Atlanta and Charlotte/Raleigh now account for 13% of current U.S. capacity across high-growth regions. The region offers attractive features for data center developers: available land, relatively lower energy costs, and proximity to major metropolitan areas.

But here's where the story gets complicated. The Southeast is currently a rapidly growing market, but the story is far more nuanced than headline projections suggest. There is no inherent reason for data centers to be sited in the Southeast, and indicators suggest that future growth may be dampened. Regional transparency remains remarkably low; Georgia, for instance, has no comprehensive data center registry, forcing analysts to piece together information from scattered public sources.

What the Data Actually Show

Our analysis synthesized forecasts from leading sources, including the International Energy Agency, Boston Consulting Group, S&P Global, McKinsey, and Enverus, to create an evidence-based baseline for data center load growth. Using Monte Carlo simulations, we developed the first rigorously constructed probability distribution for Southeastern data center growth.

The findings reveal three critical insights:

1. Industry Forecasts Show Growth Through 2030, Slower Thereafter

Industry forecasts consistently show strong growth through 2030, with annual increases ranging from 8% to 22%. However, credible models show this growth falling sharply after 2030, typically to just 1-7% annually.

2. Southeast Data Center Load Could Double or Triple by 2035

Starting from today's baseline of approximately 4 GW, our modeling suggests the most likely outcomes range from 7.5 to 9.5 GW by 2031, representing growth of 3.5 to 5.5 GW. Even accounting for considerable variability, the 95% probability range extends from 2.2 to 8.7 GW of additional load.

When we factor in technological improvements such as advances in liquid cooling, compute-in-memory, and AI algorithmic efficiencies, the picture becomes even more modest. These estimates, when adjusted for technology, suggest a lower level of demand growth, with a most likely range of 1.4 to 3.6 GW by 2031.

3. Utilities' Projections Appear to Represent Extreme Outlier Growth

Perhaps most strikingly, utility estimates consistently exceeded what our model considers a reasonable upper bound. Across nearly all simulations, utility estimates landed above what the model considers a reasonable upper bound. Excluding the technological advancement model, the likelihood of utilities' predicted growth is approximately 1 in 500, or 0.2%. When technological advancement is factored in, utility forecasts exceeded our modeled range only 13 times in 100,000 simulations.

Innovation Analysis: Technology Could Flatten the Curve

One of our most important findings is that technological innovation could fundamentally reshape energy demand. The best available evidence suggests that technology improvements could keep total regional data center load essentially flat for a decade, even as AI workloads increase.

This isn't science fiction; these improvements are already happening. Recent gains in cooling systems, hardware design, and algorithm efficiency are actively lowering projected energy use. From liquid cooling systems that dramatically reduce energy waste to emerging compute-in-memory architectures that process data more efficiently, the data center industry is innovating rapidly.

This finding is urgent and important for regional planning. It demonstrates that forward-looking, innovation-inclusive modeling is essential to avoid overbuilding costly infrastructure that may never be needed.

Why This Matters: Billions of Dollars at Stake

The consequences of getting these projections wrong are enormous. Utilities across the Southeast are proposing massive expansions of methane gas infrastructure and new power plants based on extreme growth scenarios. If these facilities are built and demand doesn't materialize, ratepayers could be left holding the bag for billions in stranded assets.

Consider what may happen when infrastructure is built to meet demand that never arrives: customers pay higher rates for unnecessary capacity, utilities lock in decades of fossil-fuel dependence, clean energy alternatives get crowded out, and economic resources are misallocated away from more productive investments.

Major Conclusions

The report is one of the first to demonstrate quantitatively what many have expected: utility data center demand projections inherently contain uncertainty and systematic upward bias. We have a few major takeaways for energy policy:

1. Planning Around Extreme Growth Forecasts Leads to Overstated Resource Needs

The study makes it clear that the highest growth projections, often used in recent utility filings, fall well outside what the modeling shows as realistic. Planning documents should incorporate probability distributions, not just worst-case scenarios, and should explain why they deviate from expert consensus when they do.

2. Technology Is Already Reshaping Demand

Recent gains in cooling, hardware design, and algorithm efficiency are already lowering projected energy use, a trend likely to continue. Encouraging investment in efficiency improvements through incentives, standards, or procurement requirements will enable smart growth without straining power and water supplies.

3. Stronger, More Transparent Data Reporting Is Crucial

Without clearer data on where load exists today and what's actually in the pipeline, infrastructure plans can easily overshoot. States should establish comprehensive data center registries. Utilities should be required to show their work, including the assumptions behind their forecasts and sensitivity analyses showing what happens if those assumptions prove wrong.

A Path Forward

This research represents more than just number-crunching. It's about ensuring that critical infrastructure decisions affecting millions of people are based on sound evidence rather than inflated projections. By bringing together Greenlink's nationally recognized modeling expertise with Science for Georgia's commitment to evidence-driven policy, we're providing the Southeast with something it desperately needs: a clear-eyed assessment grounded in the best available data.

The data center industry will continue to grow, but growth doesn't have to mean locking ourselves into outdated energy infrastructure. With smarter planning, technological innovation, and transparent data, the Southeast can accommodate this growth while advancing its clean energy goals and protecting ratepayers.

The question isn't whether data centers will expand in the Southeast. It's whether we'll make smart, evidence-based decisions about how to power that expansion, or whether we'll build expensive infrastructure for a future that may never arrive.

To read the full report, Data Center Growth Impact on Southeast U.S., visit our Publications page. For more information about how Greenlink Analytics can support evidence-based energy planning in your community, contact us.