Bridging AI ambition with Environmental Governance in National Policies

The surge of artificial intelligence(AI) promises a future of unparalleled economic growth and efficiency. Its core benefit lies in a profound ability to sift through mountains of data, detect hidden patterns, and predict future outcomes with remarkable accuracy. This power positions AI as a potentially invaluable tool for planetary stewardship, capable of monitoring environmental and climate changes and guiding governments, businesses, and individuals toward more sustainable choices.

Yet, beneath the surface of this computational intelligence lies a growing physical footprint that threatens to undermine its potential for good. Increasing analyses reveal startling environmental cost. In 2025 alone, the global operation of AI systems may carry a carbon footprint rivaling that of a megacity like New York. Even more striking is its water demand. The total water footprint of AI could reach hundreds of billions of litres, a volume that may surpass the entire world’s annual consumption of bottled water!

Understanding the full scale of this impact, however, is shrouded in challenge. While we can estimate global AI power demand, accurately quantifying its associated carbon and water footprints remains difficult. This opacity derives from the complex nature of AI infrastructure and a critical lack of corporate transparency.

AI water footprint and a policy imperative for transparency

The environmental disadvantages of AI are rooted in the data centers that power it. These facilities generate significant electronic waste, often laden with hazardous materials like mercury and lead. What's more, they are profoundly water-dependent, requiring vast quantities both for their initial construction and for the continuous cooling of their operational hardware.

The water footprint of AI is particularly murky to assess. A company’s total water consumption includes both direct and indirect use. Direct consumption refers to the water used on-site for cooling servers. Indirect consumption is the water embedded in the electricity that powers those servers, used at the power generation source. The central problem is that technology companies, at best, typically only disclose their direct water use. The vast majority of indirect water consumption remains hidden, as the embedded water in purchased electricity is rarely disclosed at any level. This lack of comprehensive, company-wide metrics, with some operators disclosing no environmental data at all, makes responsible management of AI’s growing impact nearly impossible.

This crisis of disclosure demands a policy response. The shortcomings in environmental reporting from data center operators could be remedied by new mandates requiring the disclosure of additional, standardized metrics. Given the rapid growth of the sector, the urgency for such transparency is escalating daily. By highlighting the gaps in available data and the staggering potential scale of AI’s environmental consequences, with carbon emissions estimated between 32.6 and 79.7 million tons and water use between 312.5 and 764.6 billion liters in 2025, policies can grasp both the urgency for action and the key areas of concern.

The United Nations Environment Programme (UNEP) offers a clear roadmap. Their recommendations include establishing standardized procedures for measuring AI’s impact, developing regulations that mandate corporate disclosure of direct environmental consequences, encouraging tech companies to create more efficient algorithms and recycle resources, greening data centers with renewable energy, and weaving AI policies into broader environmental governance frameworks.

Sustainable AI in Africa

This global conversation finds a critical focal point in Africa. As international tech companies like Microsoft announce plans to construct and expand data centers across the continent to tap into rising cloud computing demand, a crucial question emerges. While AI systems power global industries, who bears the environmental cost? The carbon and water footprints of these data centers will increasingly be felt in African nations.

Kenya has emerged as a leading voice in calling for accountability. At the United Nations Environment Assembly, Kenya submitted a resolution demanding the world account for the environmental costs of AI infrastructure. The resolution advocates for environmentally sustainable AI across Africa, emphasizing integrated renewable energy, reduced water consumption, and strict adherence to environmental and social governance practices. For Africa, the central challenge is scaling data centres without straining the environment. Any AI expansion must be sustainable, ensuring the technology is “green by design.”

Achieving this requires clear regulatory standards from the outset. New policies must prevent inefficient or unsustainable infrastructure from entering the market by setting benchmarks for energy use, cooling efficiency, and water management for all new AI and data centre projects. For existing facilities, a compliance timeline would allow operators to upgrade systems, adopt renewables, and improve efficiency without abrupt disruption.

However, a review of AI policy ambitions across Africa reveals a concerning gap. An analysis of 14 national AI strategies shows that while most countries plan to increase data processing for AI development, only six mention energy concerns. Even fewer—just three—acknowledge the environmental consequences. The African Union Continental AI Strategy notes the threat data centre cooling poses to water-scarce regions, and Kenya’s strategy highlights the long-term environmental impacts on natural resources. Yet concrete plans to address these issues are scarce, with Senegal being a notable exception for proposing to integrate AI’s environmental impact into its environmental code.

As these national strategies form the foundation for future regulatory roadmaps and policymaking, it is crucial that African countries keenly consider the full environmental lifecycle of AI. The next phase must definitively address how cloud service providers and data center operators will be guided toward sustainable AI development. The continent has the opportunity to leapfrog outdated, wasteful practices and build a digital future that is both innovative and inherently sustainable, setting a global standard for balancing technological ambition with planetary responsibility.