From Soot to Servers: Paving the way for Digital Age by using US Coal Power Plant Brownfields

Partnering with Roland Berger, this Capstone workshop evaluated converting retired U.S. coal plant brownfields into data centers. The project addressed rising demand from AI and cloud computing amid power and land constraints. Such sites offer cost efficiencies, scope 2 and 3 carbon savings, and accelerated grid interconnection, presenting a viable solution for data center expansion.

To pinpoint viable locations for converting decommissioned coal plants into data centers, the team established a formal evaluation rubric. Starting with the Global Energy Monitor database, the team applied foundational criteria to narrow the U.S. coal facility list to 14 initial candidates. Through an integrated quantitative and qualitative analysis, the team narrowed these down to four priority sites, ultimately performing a comparison between two finalists before determining the optimal location.

To test the broader thesis, the team conducted a deep-dive assessment of the final site selection: the retired Ameren Meredosia Coal Power Station in Meredosia, Illinois. Analysis included examining site characteristics, data center potential, regulatory conditions, and site-specific risks. To show the potential for revenue of the proposed 200 MW data center, the team also built a business model and a technical and financial model, including power supply configurations, capital expenditure, operating assumptions, and sensitivity analysis. The findings showed that brownfield sites can provide a speed-to-power advantage due to existing grid access and infrastructure as well as potential cost savings compared to greenfield sites.

In conclusion, conversions are financially optimal under specific conditions: viable business models, strong grid connectivity, supportive policies, and alignment with community needs. Former coal plant brownfields represent an underutilized asset class that should be leveraged to meet data center demand while mitigating environmental impacts from new AI needs.