Artificial intelligence is increasingly promoted as a catalyst for climate action, with applications ranging from grid optimisation to industrial efficiency. However, new research and policy reports warn that AI itself is emerging as a significant and poorly regulated environmental burden, driven by rapidly rising electricity demand, water consumption and carbon emissions from data centres.

A recent report by Arthur D Little, The Hidden Cost of AI, warns that the rapid expansion of generative AI could dramatically reshape global energy demand. The study estimates that AI-related computing could account for up to 3 percent of global electricity consumption by 2030, a level comparable to the power demand of major industrial economies if left unmanaged.

The report highlights that data centres supporting large language models are being built at unprecedented speed, often outpacing the development of renewable generation and grid infrastructure. As a result, AI growth risks reinforcing dependence on fossil-fuel-based electricity in regions where clean energy capacity is not expanding fast enough.

A major concern identified across multiple studies is the lack of transparency and measurement. An industry analysis cited by Manufacturing.net found that 97 percent of companies fail to measure the environmental footprint of their AI systems, meaning emissions, energy intensity and water use frequently go unreported in sustainability disclosures (Manufacturing.net, 2025). This reporting gap makes it difficult for regulators and investors to assess AI’s true climate impact.

Water consumption is emerging as another critical issue. Research from the Massachusetts Institute of Technology shows that large AI data centres can consume millions of litres of fresh water annually for cooling, intensifying pressure on already stressed water systems (MIT, 2025). This is particularly relevant for arid regions, including the Middle East, where water scarcity and energy planning are closely intertwined.

Beyond operational impacts, the United Nations Environment Programme has drawn attention to AI’s material footprint. In its recent digital sustainability assessments, UNEP warns that rising demand for AI hardware is increasing reliance on critical minerals, accelerating mining activity, and contributing to electronic waste unless circular-economy frameworks are strengthened (UNEP, 2024).

International energy institutions are also raising red flags. The International Energy Agency, in its latest electricity and data-centre outlooks, identifies data centres as one of the fastest-growing sources of electricity demand globally. The agency cautions that without firm efficiency standards and clean-power mandates, digital infrastructure growth could undermine national decarbonisation targets (IEA, 2024).

These findings challenge the widely held assumption that AI is inherently climate-positive. While targeted AI applications can improve energy efficiency and system management, experts stress that benefits are highly context-dependent. Recent investigative reporting argues that many climate claims associated with generative AI remain unproven when weighed against the technology’s escalating resource demands (The Guardian, 2026).

Policy responses are now beginning to surface. Analysts increasingly call for mandatory disclosure of AI-related energy and water use, tighter efficiency requirements for data centres, and alignment between AI expansion and renewable-energy availability. Without such measures, AI risks becoming a hidden driver of emissions growth rather than a tool for sustainability.

As governments and industries accelerate digital transformation, the emerging consensus is clear: AI must be governed as critical infrastructure, with the same environmental scrutiny applied to power generation, heavy industry and transport. Whether artificial intelligence supports or undermines climate goals will depend less on its potential, and more on how transparently and responsibly it is deployed.

2026 © All right reserved for Oman Establishment for Press, Publication and Advertising (OEPPA) Provided by SyndiGate Media Inc. (Syndigate.info).