The study detailed the comprehensive environmental footprint of AI systems across five main dimensions.
A recent study issued by TRENDS Research & Advisory, a subsidiary of the TRENDS Group, confirmed that the massive surge witnessed in the artificial intelligence and advanced technology sector – despite being an exceptional economic driver that contributed to raising U.S. economic growth to 4.3 percent in the third quarter of 2025 – now imposes a critical and paradoxical equation, oscillating between direct environmental threats and opportunities to support climate sustainability.
The study, titled Artificial Intelligence: A New Environmental Burden or an Opportunity to Enhance Sustainability? stated that this rising power has two faces: the first is dark, characterized by the massive consumption of natural resources and the generation of toxic waste; the other is bright, characterized by harnessing the analytical and predictive capabilities of algorithms to find innovative solutions to resource scarcity and climate change.
The Five Depleting Footprints of Planet Earth
The study, prepared by Rana Al-Daqqaq, Senior Editor at TRENDS, detailed the comprehensive environmental footprint of AI systems across five main dimensions:
- Carbon Footprint: The study revealed that emissions from AI Agents applications alone now equate to 8% of the global aviation sector’s total emissions by the end of 2025.
- Water Footprint: Data center servers rely exclusively on freshwater to prevent critical failures. A single simple conversation (20–50 questions) with ChatGPT consumes about half a liter of water, with global AI water withdrawals expected to reach 6.6 billion cubic meters by 2027.
- Energy Footprint: The International Energy Agency projects that data center consumption will double by 2026 due to reliance on energy-intensive Graphics Processing Units (GPUs), threatening the targets of the Paris Agreement.
- Material Footprint and E-Waste: Resulting from the rapid acceleration of hardware upgrades and replacement every two to three years, leading to the accumulation of toxic waste and depletion of rare minerals such as cobalt and lithium.
- Land Footprint: Associated with the vast geographical areas required to build data centers, power stations, and mining sites.
Green AI: Efficiency and Solutions
In contrast, the study highlighted the concept of Green AI, which focuses on data efficiency, hardware efficiency, and transparency. It demonstrated the technology’s positive enabling impact: Google DeepMind algorithms have proven capable of reducing data center cooling energy by 40%, while initiatives like Climate TRACE contribute to pinpointing pollution sources with exceptional precision via satellites and reduce the time to discover new carbon- capture materials by 90 percent.
Governance Dilemmas and the ‘North-South’ Dichotomy
The study discussed three main dilemmas hindering digital
sustainability:
- The Rebound Effect: Where increased resource efficiency and reduced software costs lead to stimulating overall consumption on a broader scale.
- Climate Justice: Highlighting an unjust political geography where astronomical economic gains are concentrated in advanced Northern countries (such as the U.S. and China), while the environmental burden, toxic waste mountains, and mine depletion are exported to the Global South.
- The Black Box Dilemma: Represented by the lack of transparency from major tech companies regarding their actual consumption and reliance on purchasing paper carbon offsets. The study praised the EU AI Act as the first global legislative step mandating comprehensive environmental disclosure by companies.
Strategic Recommendations for the Future
Al-Daqqaq concluded her study by presenting an integrated framework for sustainable AI, including:
- Moving toward Lean AI software-wise.
- Localizing data centers in regions fully reliant on clean energy.
- Adopting circular economy strategies for digital resources and activating device take-back programs to recycle rare minerals.
- Formulating international legislation mandating a Five- Dimensional Environmental Impact Statement and combating institutional opacity to ensure the development of technical intelligence that serves humanity while respecting the environment that enabled its existence.
