Navigating data center sustainability with GRESB: Trends and recommendations for 2025 GRESB participants
This report examines sustainability trends for data centres within the GRESB framework. It highlights significant energy intensity compared to traditional real estate and outlines criteria for choosing between Real Estate and Infrastructure assessments in 2025. Future developments involve a partnership with iMasons to create tailored sector-specific metrics by 2026.
Please login or join for free to read more.
OVERVIEW
Introduction
Data centres are rapidly expanding facilities essential to modern life, yet they remain significant consumers of energy, water, and land. McKinsey forecasts that global demand for data centre services will triple by the end of the decade, with a midrange scenario suggesting a compound annual growth rate (CAGR) of 22%. Developments in generative AI are expected to further intensify these energy demands. Over the last decade, the power demand of a typical new data centre asset has grown from 30 megawatts to more than 200 megawatts. This expansion involves substantial financial costs, typically ranging from USD 9-to-15 million per megawatt, resulting in new facility costs exceeding USD 1 billion. Data centres are categorised as hyperscale, colocation, edge, or enterprise based on their ownership and operations.
GRESB assessments
Currently, data centres can participate in either the GRESB Real Estate or Infrastructure Assessments. GRESB Real Estate classifies data centres as a ‘Technology/Science’ property type. These entities usually consist of a portfolio of assets, with assessment results combining aggregated asset-level performance, portfolio-level indicators, and management practices. If an entity owns multiple property types, GRESB weights the data centre contribution based on gross asset value (GAV). Conversely, GRESB Infrastructure categorises data centres under ‘Data Infrastructure: Data Storage, Data Centres’. This track includes three assessments for funds, operational assets, and assets in development. Data centre assets in this track are directly benchmarked against other data centre companies and utilize a built-in materiality approach to focus on relevant sustainability topics.
Participation
In 2024, a total of 41 entities reported data centres to GRESB, marking a 28% increase from 2023. This included 17 reporting under the Real Estate Assessment and 24 under the Infrastructure Asset Assessment, representing 842 sites globally. The geographic distribution shows significant concentration in the Americas (428 real estate assets and 81 infrastructure facilities) and Europe (168 real estate assets and 46 infrastructure facilities), followed by Asia (47 assets; 34 facilities), Oceania (10 assets; 15 facilities), and Africa (13 assets; 0 facilities).
Understanding energy use in data centres
Data centres are among the most energy-intensive real assets, significantly outpacing traditional buildings. GRESB REAL Statistics show that the median energy consumption for a data centre reporting under real estate ranged from 17.10 to 39.15 million kWh per year, compared to 1.46 to 1.62 million kWh for office buildings. Energy Use Intensity (EUI) for these data centres ranged from 1184.00 to 3095.00 kWh/m², while offices ranged from 137.00 to 159.00 kWh/m². EUI alone is not a perfect efficiency indicator as it does not account for operational load or utilisation factors.
Variability in energy demand is significant. The highest 10% in EUI consumed 199.0 million kWh/year, whereas the lowest 10% consumed 4.8 million kWh/year. The average energy use per asset has risen from 59.6 million kWh in 2020 to 76.0 million kWh in 2024. High consumption was particularly noted in the United States, Singapore, Germany, China, and the Netherlands. The 23.0 TWh of electricity consumed by GRESB-reporting data centres in 2023 was equivalent to approximately 19.2% of the total electricity generation of the Netherlands in 2022.
Example of a GRESB data centre participant
K-square Data Centre PFV Co. Ltd. (Koramco Asset Management Co, Ltd.) was recognised as a GRESB Real Estate Sector Leader and a 5-Star participant in the 2024 Real Estate Development Assessment. The K-Square Data Centre Gasan facility in Seoul prioritises renewable energy and efficiency features. These include solar power, fuel cells, and both indoor and outdoor water recycling systems, demonstrating how high-impact facilities can mitigate environmental impacts through thoughtful design and operation.
2025 considerations for GRESB participation
In 2025, operators must choose the assessment that best fits their investment approach. Firstly, participants should select the assessment matching the subscription of their target GRESB Investor Members to ensure efficient data sharing. Secondly, data centre REITs often choose Real Estate, while infrastructure funds typically select the Infrastructure Assessment. Thirdly, data granularity differs; Real Estate emphasises building-level data, while Infrastructure focuses on fund or portfolio company-level reporting.
Materiality and normalisation also differ between tracks. GRESB Infrastructure uses a materiality-based approach to determine indicator weightings, whereas Real Estate applies uniform requirements to all sub-sectors. For normalisation, Infrastructure uses GAV, revenue, and output (defined as data stored), while Real Estate normalises performance by floor area. Finally, Real Estate compares entities against similar property types and the broader market, whereas Infrastructure allows comparisons against both other data centre companies and various infrastructure sectors like energy and transport.
The future of data centres at GRESB
GRESB has launched a Data Centre Working Group in partnership with iMasons to develop a tailored assessment solution, as current frameworks do not fully capture the sector’s unique operational and sustainability characteristics. By 2026, GRESB and iMasons expect to introduce new assessment options with sector-specific indicators, metrics, and weightings to provide more accurate benchmarking for investors and managers based on their specific operational circumstances.