New study reveals AI’s massive energy use barely impacts global emissions

Overview

A major peer-reviewed study published by researchers at the University of Waterloo and Georgia Tech has challenged the prevailing narrative that artificial intelligence is a significant driver of climate change. While AI’s electricity consumption in the United States now rivals that of some small countries, the study found that when measured against total global greenhouse gas emissions, AI’s direct contribution remains remarkably small. This finding has important implications for environmental consultants, sustainability reporting professionals, and those involved in environmental impact assessments for data centre developments across Australia.

Key details

The study’s central finding is the distinction between absolute energy consumption and proportional emissions impact. Key technical points include:

• Massive energy draw, modest emissions share: While the electricity consumed by AI workloads is enormous in absolute terms, it represents a small fraction of total global greenhouse gas emissions when the carbon intensity of the electricity source is accounted for.
• Grid mix dependency: The actual greenhouse gas impact of AI operations depends heavily on the carbon intensity of the regional electricity grid supplying the data centre. Facilities powered by high-renewable grids produce significantly lower emissions per unit of computation than those on coal-heavy grids.
• Power Purchase Agreements (PPAs): Many hyperscale data centre operators have entered into PPAs for renewable energy, further reducing their attributable emissions. The study highlights the importance of distinguishing between location-based and market-based Scope 2 emissions accounting.
• Water consumption: Beyond energy, the study notes the significant water demand of large data centres, particularly those using evaporative cooling systems. This is a resource consumption issue distinct from greenhouse gas emissions but increasingly relevant to environmental approvals.

The research does not dismiss AI’s environmental footprint. Rather, it argues for a more nuanced assessment that separates energy consumption from climate impact and considers the specific circumstances of each facility.

Australian context

Australia is experiencing a rapid expansion of data centre development, driven by growing demand for cloud computing, AI services, and digital infrastructure. This growth intersects with several environmental regulatory considerations:

• Environmental Impact Assessment (EIA): State planning authorities increasingly require detailed environmental assessments for proposed data centre developments. These assessments must address energy consumption, greenhouse gas emissions, water use, noise, and local ecological impacts. The findings of this study suggest that the emissions component of these assessments needs careful, context-specific analysis rather than blanket assumptions about AI’s climate impact.
• Water-energy nexus: In water-stressed regions of Australia, the significant water requirements of evaporative cooling systems in hyperscale data centres are a growing concern. Groundwater extraction for cooling purposes may trigger requirements under state water management legislation and potentially require hydrogeological assessment.
• Corporate sustainability reporting: Under the Australian Government’s mandatory climate-related financial disclosure requirements (aligned with the International Sustainability Standards Board framework), companies operating or hosting services in data centres must accurately report their Scope 2 and Scope 3 emissions. This study highlights the complexity of these calculations and the importance of using appropriate emissions factors and accounting methodologies.
• Beneficial AI applications: The study implicitly supports the case for applying AI to environmental management, including automated analysis of contaminated site data, contaminant transport modelling, and remote sensing for ecological monitoring. If AI’s net emissions contribution is small, the environmental benefits of applying these tools to improve environmental outcomes may well outweigh the energy costs.

Practical implications

Environmental professionals should consider the following takeaways:

• Nuanced emissions assessment: When conducting EIAs or sustainability assessments for data centre projects, avoid simplistic assumptions. Assess the specific grid mix, PPA arrangements, and cooling technology to determine actual emissions impact.
• Water resource assessment: Ensure that data centre proposals in water-sensitive areas include thorough assessment of water consumption, particularly where groundwater sources are involved. Hydrogeological assessments may be required under state legislation.
• Scope 2 and 3 reporting: Advise corporate clients on the appropriate methodology for reporting data centre-related emissions, distinguishing between location-based and market-based approaches as recommended by the Greenhouse Gas Protocol.
• AI adoption in consulting: Consider the potential for AI tools to enhance environmental consulting workflows, particularly in data analysis, predictive modelling, and monitoring system optimisation, recognising that the net environmental cost of these tools is relatively modest.
• Stakeholder communication: Be prepared to communicate nuanced findings about AI’s environmental impact to clients, regulators, and community stakeholders who may hold strong preconceptions about the technology’s climate footprint.

References and related sources

Original source: ScienceDaily
Source published: 21 March 2026
Added to Enviro News: 21 March 2026

Read the primary source article

Join the discussion on LinkedIn

Related references:

• GHG Protocol Scope 2 Guidance
• International Sustainability Standards Board (ISSB)

How iEnvi can help

iEnvi provides environmental assessment and consulting services relevant to data centre developments and corporate sustainability reporting. Our contaminated land team can conduct site assessments for proposed development locations, while our broader environmental consulting capabilities extend to environmental impact assessment support, groundwater assessment, and regulatory compliance. For projects involving complex environmental or technical disputes, our expert witness team can provide independent, authoritative technical opinions. Contact iEnvi to discuss how we can support your project.