The rapid expansion of artificial intelligence is transforming economies, industries and daily life across the world. From advanced chatbots and virtual assistants to healthcare innovations and scientific research, AI has become one of the most influential technologies of the modern era. However, a new United Nations University report has raised serious concerns about the environmental consequences of this technological revolution.

Released around World Environment Day, the report highlights the growing environmental burden associated with the infrastructure required to support AI systems. According to the study, AI data centers electricity consumption is expected to rise dramatically over the next few years, creating significant challenges related to energy demand, water usage, carbon emissions, land occupation and electronic waste.

One of the most striking findings of the report is the projection that AI data centers could consume nearly three times the combined annual electricity usage of Pakistan, Bangladesh and Nigeria by the year 2030. This warning has sparked global discussions about the need to ensure that the future of artificial intelligence remains environmentally sustainable.

The report was published by the United Nations University Institute for Water, Environment and Health (UNU-INWEH) and provides one of the most comprehensive assessments yet of the environmental footprint associated with AI infrastructure. Researchers argue that discussions about AI sustainability have often focused only on carbon emissions while ignoring other critical environmental impacts.

According to the study, evaluating the artificial intelligence environmental impact requires a broader approach that includes three key dimensions: carbon footprint, water footprint and land footprint. Looking at only one of these factors can create a misleading picture of the actual environmental cost of AI development.

The findings suggest that the world is approaching a crucial moment in determining how AI technologies will evolve. While artificial intelligence offers enormous benefits, its supporting infrastructure is becoming increasingly resource-intensive. As demand for AI services grows, so does the need for massive computing power housed in data centres around the world.

These facilities serve as the backbone of modern digital systems. Every AI query, machine learning operation, cloud computing task and data-processing activity relies on data centres equipped with powerful servers and advanced cooling systems. Although users often interact with AI through software applications, the physical infrastructure behind these services requires enormous amounts of electricity and water.

The report estimates that data center power consumption globally reached 448 terawatt-hours (TWh) during the previous year. To put this figure into perspective, it exceeds the total annual electricity consumption of entire countries such as Saudi Arabia. Artificial intelligence alone accounted for approximately one-fifth of this electricity demand.

However, the situation is expected to become much more significant by 2030. Researchers project that annual electricity consumption by data centres will increase to approximately 945 TWh, effectively doubling current levels. This amount of electricity would be roughly equivalent to the total annual power consumption of Japan, one of the world's largest economies.

A particularly concerning aspect of the projection is the growing role of artificial intelligence within this energy demand. By 2030, AI applications are expected to account for around 40 percent of total data centre electricity consumption. This means that AI will become one of the dominant drivers of global digital energy usage.

The report also highlights the substantial AI water footprint associated with data centre operations. Many people are unaware that data centres require vast quantities of water for cooling purposes. Servers generate significant heat during operation, and cooling systems help maintain safe temperatures to prevent equipment failure.

According to the study, data centres consumed approximately 4.5 trillion litres of water globally during the previous year. This volume of water is sufficient to meet the needs of more than 600 million people in Sub-Saharan Africa. The researchers warn that water demand will continue rising as AI infrastructure expands.

By 2030, annual water consumption by data centres could reach 9.3 trillion litres. This staggering amount is roughly equivalent to the basic annual domestic water requirements of all 1.3 billion people living in Sub-Saharan Africa. Such projections raise concerns about competition for water resources, particularly in regions already facing water scarcity challenges.

The environmental concerns extend beyond electricity and water. The report also examines the AI carbon footprint generated by expanding digital infrastructure. Data centres currently produce approximately 189 million tonnes of carbon dioxide emissions annually. By 2030, this figure could increase to 399 million tonnes, more than doubling within a decade.

Carbon emissions remain an important component of environmental sustainability because they contribute directly to climate change. However, researchers caution against viewing carbon emissions as the sole indicator of sustainability. Some solutions that reduce carbon emissions may simultaneously increase water consumption or require additional land use.

This observation forms one of the report's most important conclusions. According to lead researcher Dr Miriam Aczel, sustainability assessments based solely on carbon metrics can create unintended consequences. She explained that energy sources appearing environmentally friendly from a carbon perspective may generate greater pressure on water resources or land availability.

The report argues that policymakers, technology companies and investors need to adopt a more comprehensive understanding of AI sustainability. Decisions should consider multiple environmental dimensions rather than focusing exclusively on greenhouse gas emissions.

Land use represents another major challenge identified by the study. The AI infrastructure impact extends beyond the buildings that house servers. Data centres require extensive supporting infrastructure, including electricity generation facilities, transmission networks, cooling systems and transportation connections.

Researchers estimate that the global land footprint associated with data centres covered approximately 6,900 square kilometres last year. By 2030, this area could expand to more than 14,500 square kilometres, more than doubling within a relatively short period.

The expansion of data centres often creates tensions between technological development and land conservation. Large facilities require suitable locations with access to reliable electricity, water supplies and network connectivity. As demand grows, communities may face difficult decisions regarding land allocation and environmental protection.

The report also draws attention to the growing problem of AI e-waste. Artificial intelligence systems depend on specialised hardware, including advanced processors, graphics chips and networking equipment. These components eventually become obsolete and must be replaced as technology advances.

By 2030, annual generation of AI-related electronic waste could reach approximately 2.5 million tonnes. Researchers compare this amount to the disposal of nearly 250 Eiffel Towers worth of electronic equipment every year. Managing this waste responsibly will require significant improvements in recycling and circular economy practices.

Electronic waste presents unique environmental challenges because it often contains valuable materials alongside hazardous substances. Improper disposal can result in pollution, resource loss and public health risks. As AI adoption accelerates, addressing e-waste management will become increasingly important.

Another notable finding concerns the geographical concentration of AI infrastructure. According to the report, more than 90 percent of global AI-specialised cloud computing capacity is concentrated in just two countries: the United States and China.

This concentration raises questions about the distribution of environmental burdens and benefits associated with AI development. While some regions host data centres and bear the environmental impacts, the economic and technological advantages may be distributed differently. Researchers suggest that future discussions should address issues of environmental justice and equitable resource management.

Professor Kaveh Madani, Director of UNU-INWEH and leader of the investigation team, emphasised that the report is not intended as criticism of artificial intelligence itself. Instead, he described it as a call for responsible innovation and proactive planning.

According to Madani, artificial intelligence has already improved the lives of billions of people worldwide. It supports medical research, enhances productivity, strengthens communication systems and contributes to scientific discoveries. The goal is not to slow technological progress but to ensure that it develops within environmental limits.

He stressed that society currently has a limited opportunity to shape the future trajectory of AI development. Decisions made today regarding energy sources, infrastructure design, resource management and recycling systems will influence the long-term sustainability of artificial intelligence.

The report arrives at a time when governments and businesses are investing heavily in AI technologies. Competition among nations and technology companies has accelerated the construction of new data centres and computing facilities. While these investments support innovation and economic growth, they also increase demand for natural resources.

The challenge facing policymakers is finding ways to balance technological advancement with environmental responsibility. This may involve improving energy efficiency, adopting renewable energy sources, developing water-saving cooling technologies and enhancing recycling systems for electronic equipment.

Technological innovation itself could play a role in addressing these challenges. Researchers and engineers are already exploring new approaches to data centre design, cooling methods and energy management. Advances in hardware efficiency could reduce electricity requirements, while alternative cooling technologies may lower water consumption.

Renewable energy adoption also offers potential benefits. However, as the report notes, renewable solutions must be evaluated carefully to ensure that reductions in carbon emissions do not create new pressures on water or land resources.

Public awareness will likely become increasingly important as AI continues expanding into everyday life. Many users view AI as purely digital technology without considering the physical infrastructure required to support it. Understanding the environmental dimensions of AI may encourage more informed discussions about sustainability and resource management.

The findings of the UN report AI data centers serve as a reminder that every technological revolution carries environmental consequences. Just as industrialisation transformed energy systems and resource consumption patterns, the AI revolution is reshaping global demand for electricity, water, land and materials.

At the same time, artificial intelligence itself may contribute to environmental solutions. AI technologies are already being used to optimise energy systems, improve water management, monitor ecosystems and support climate research. The challenge is ensuring that the benefits of AI outweigh its environmental costs.

The report ultimately calls for a balanced approach. Rather than viewing artificial intelligence as either entirely beneficial or entirely harmful, researchers advocate for responsible development that recognises both opportunities and risks. Sustainable AI requires careful planning, transparent assessment and ongoing efforts to minimise environmental impacts.

As the world moves toward an increasingly AI-driven future, the importance of addressing these issues will only grow. The projected rise in AI electricity demand 2030, water consumption, carbon emissions and electronic waste underscores the need for coordinated action by governments, technology companies and civil society.

The achievement of technological progress should not come at the expense of environmental sustainability. By recognising the full environmental footprint of AI today, policymakers and industry leaders have an opportunity to build a future where innovation and sustainability advance together rather than in conflict.

The message from the United Nations University report is clear: artificial intelligence offers tremendous promise, but its environmental costs must be managed responsibly. The choices made during the next few years will determine whether AI becomes a model of sustainable innovation or a source of growing environmental pressure. The window for action remains open, but researchers warn that it may not remain open for long.