Data centres (DCs) are the heartbeat of our digital lives. They enable streaming, online payments, and cloud storage. But behind seamless connectivity lies a critical question: can the world’s digital backbone be sustainable?
In August 2025, a report I co-authored with the UNECE Task Force on Digitalisation in Energy was published.
Titled, “Intersection between large-scale digitalization and environmental sustainability: an overview of use and impacts of data centres“, the report explores DC energy and environmental impacts. It also proposes recommendations to address key sustainability challenges. Below are the report’s main highlights.
Definition & global trends of data centres
- Data centre is indispensable in the current digital economy era. It supports e-commerce, cloud services, and artificial intelligence (AI) through mass data storage and processing.
- A DC typically includes servers, storage, network systems, power supply, cooling, monitoring software, and physical facilities.
- Type of DC based on its typology:
- Enterprise: large and centralised, privately owned and managed by individual organisations located on-premises.
- Colocation: owned by a third party who leases physical space, power, and cooling to multiple clients who host their own servers and equipment.
- Hyperscale: owned and managed by major technology companies to house thousands of servers to support large-scale cloud services.
- Edge: small and distributed located closer to end users or data sources to minimise communication delays and support real-time data processing.
- There is also a term called Modular DC, which is prefabricated, scalable, and easy to deploy. Modular units can be used for any DC types above.
- Global DC deployment grew 4% annually between 2010 and 2020, with the US hosting 46% of facilities as of 2025.
- The installed capacity number is projected to keep increasing, from 40 GW in 2023 to nearly 80 GW in 2030, or about 15% per year.
- The market value of the DCs industry was estimated at USD 352.9 billion in 2024, dominated by the US, China, and Japan. It is then predicted to grow at a compound annual growth rate of 8.4% between 2025 and 2029.
Impacts of data centres
- With their growth trends and reliability demands to keep 24/7 uptime, DCs consume more energy than before. Global DC electricity use increased from 100 TWh in 2005 to 400 TWh in 2024, or about 1.5% of global demand. Current policies indicate this share will reach 3% by 2030. A single ChatGPT query uses about 2.9 Wh, compared to 0.3 Wh for one Google search.
- DCs can finish construction in under two years, while power transmission and generation infrastructure needs five to ten years, creating grid bottlenecks. Singapore provides a clear example of this. Facing land and energy constraints, the government imposed a moratorium on new DC developments in 2019. The moratorium ended in 2022 with stricter rules on energy efficiency and alignment with national decarbonisation goals.
- Aligned with their electricity consumption, DCs emit greenhouse gas (GHG) emissions. In 2024, they accounted for about 0.5% of global carbon emissions. Current policy trends indicate their share will rise to 1% by 2030.
- DCs use large volumes of water directly (for on-site cooling, humidification, maintenance) and indirectly (for electricity generation and semiconductor manufacturing). A 100 MW hyperscale facility can use about 2 million litres of water per day, equal to 6,500 households. Globally, DCs consume an estimated 560 billion litres of water per year, which could more than double by 2030.
- From a full life-cycle perspective, DCs rely on extensive extraction, processing, and assembly of raw and manufactured materials, such as cobalt, lithium, and rare earth elements. They also generate growing volumes of e-waste. The entire process raises concerns about environmental impacts, supply chain vulnerabilities, and people’s well-being.
Mitigation strategies
- Enhancing energy performance is critical for reducing data centers’ electricity consumption. As cooling is the most energy-intensive part of data centres, some key strategies include adopting liquid and free-air cooling. These methods reduce energy use and enable facilities to recover waste heat for district heating. In addition, deploying energy-efficient hardware, virtualisation, AI- and Internet of Things can optimise server utilisation and cooling loads.
- Although the rising electricity demand from DCs can strain grids, they also offer opportunities to support system stability through smart integration. Strategies such as demand response, load shifting, on-site storage, and strategic siting enable DCs to act as flexible grid assets, reducing peak loads and utilising renewable generation. Effective integration requires early coordination between developers and utilities, grid modernisation, and supportive policies, including incentives for flexibility.
- Integrating more renewable energy for powering DCs through captive plants, off-grid, or power purchase agreements (PPAs) can reduce the GHG emissions.
- Closed-loop water systems or air-cooled technologies can improve resilience and resource efficiency in water consumption.
- Circular economy approaches, such as low-carbon and locally sourced materials, modular and prefabricated construction, durable and recyclable IT hardware, refurbishment programs, and hardware-as-a-service models, can reduce emissions, waste, and resource depletion.
Policy implications & recommendation
- Develop strong sustainability key performance indicators that cover the various metrics, such as Power Usage Effectiveness (PUE), Carbon Usage Effectiveness (CUE), and Water Usage Effectiveness (WUE). A holistic assessment framework should account for full lifecycle emissions and prioritise circularity metrics that measure component reuse, recycling rates, and landfill diversion.
- Develop new efficiency indices that track data utilisation, measuring the proportion of actively used data relative to total storage. These metrics can guide strategies to minimise unnecessary data replication and support smarter data lifecycle management.
- Align data centre operations with grid flexibility goals. Public agencies should encourage utilities to view DCs as interactive grid resources rather than passive loads.
- Mandate comprehensive reporting, support open modeling tools, and create national repositories to enable evidence-based policy and accountability. Effective policymaking requires transparent, standardised data on energy use, emissions, and material flows across the full lifecycle of data center infrastructure.
- Provide financial incentives, such as tax credits, green bonds, discounted tariffs, and accelerated depreciation, to encourage energy-efficient and low-impact technologies. Public-private partnerships and instruments like PPAs can mobilise investment in shared infrastructure, complemented by workforce and research developments.
- Upgrade the voluntary actions. Policymakers should establish minimum performance standards and mandating extended producer responsibility.
In the end, sustainable data centres aren’t just about cutting energy bills or reducing emissions. They’re about powering our digital lives without costing the planet. Every step toward greener technology helps ensure that the ‘cloud’ we rely on daily can be as clean as the blue skies we hope to protect. The choices made today will decide whether our digital future is not only fast and connected, but also sustainable and responsible.
For more details, please access and read the article in this link.