From AI to emissions꞉ Aligning ASEAN's digital growth with energy transition goals

Supporting materials

Methodology

Projection of data centre power consumption

The projected power consumption is calculated differently for each country, depending on the data availability. All the conversions from data centre capacity to power consumption, otherwise available, use Power Usage Effectiveness values based on a study for Singapore, Indonesia and Thailand and estimates for Malaysia and the Philippines.

For Singapore, the projection value is derived from the data centre power consumption share of total electricity demand stated in the news and calculated using Ember’s national power consumption data and Compound Annual Growth Rate (CAGR) approach.

Malaysia’s value also uses a CAGR approach, but using existing capacity in 2024 and absolute estimates of data centre power use in 2030 and 2040, calculating the consumption by multiplying the capacity by 24 hours and 365 days. Indonesia’s data follows the same approach as Malaysia, calculating the projection based on existing capacities in 2021 and 2024, and the capacity estimates in 2030. Thailand and the Philippines’ projections are calculated with the same methodology, using existing capacity data in 2021 and power use projection in 2030 for Thailand and capacity projection in 2028 for the Philippines.

Solar and wind investment ballpark estimates

We calculated the investment estimates for the data centre power consumption in the five countries using publicly available investment values for solar and wind power projects. Some data are obtained from specific projects, such as Indonesia’s Cirata Floating PV, Thailand’s Bangchak Solar Farm and the Philippines’ Bacolod Solar Farm. The rest comes from high-level estimates on Malaysia Renewable Energy Roadmap (MyRER) and Singapore’s Energy Market Authority.

The data is extrapolated to the present year using average inflation rates with a CAGR approach, also taking into account the solar and wind cost decline rates over the years, divided by the original capacity size and multiplied by the potential data centre demand in GW. The proportion of solar and wind capacity follows the proportion set in the national policy documents.

Power sector emissions in ASEAN data centres

Ember estimates power sector emissions from the data centre industry in the five countries by calculating the grid emission intensity projection with the estimated data centre power consumption up to 2030.

Viet Nam’s data centre power consumption has a minimal impact on overall electricity demand compared to other sectors and hence, is not included. This scope is maintained throughout the analysis of this report.

Grid emission intensity projection

The projection is derived from dividing the national and regional grids’ power generation for each fuel by the total emissions for each fuel. The total emission is calculated by multiplying the power generation per fuel by the emission factors. Power generation data is based on Ember’s national data (if the national grid is interconnected) and the historical and target data on the government websites for the specific grids, such as Peninsular Malaysia, Java-Madura-Bali and Luzon grids.

Generation data and renewable targets specific to Batam are not publicly available, only province-level data for Riau Islands can be found and therefore excluded from the analysis.

Percentage of data centre power demand met by solar, wind and batteries

The analysis comes from an Excel modelling tool, extracting the solar and wind profiles from Renewables.ninja for each country and the estimated data centre demand in 2030. The battery duration is assumed to be 4 hours with a round-trip efficiency of 90%. The assumption for solar and wind proportion is made based on the ratio of solar and wind for the respective countries and grids.

Limitation

Key industry information—such as facility locations, capacity and land use—is largely confined to commercial databases, not publicly accessible and not derived from official government sources. This restricts the depth of analysis for ASEAN’s data centre research, which remains limited.

This study primarily relies on data from consultancy firms, market research organisations, news articles and financial institutions. However, different approaches across these sources, particularly in projections, pose challenges in assessing the future landscape and providing accurate recommendations to support energy transition efforts.

Acknowledgements

Contributors

We would like to thank Pritesh Swarmy from Cushman & Wakefield for his valuable input, data sharing and quotes. Our sincere gratitude to other external reviewers who have contributed but cannot be publicly mentioned.

We further extend our appreciation to Aditya Lolla for his guidance and reviews; Neshwin Rodriguez for sharing insights on the capacity mix modelling tool; Rini Sucahyo, Shiyao Zhang and Ardhi Arsala Rahmani for editing and communications support; Matt Ewen for his contributions to data collection and validation; Jivan Zhen Thiru and Reynaldo Dizon for data visualisation support; and to Elisabeth Cremona and Duttatreya Das for their valuable reviews and discussions.