Artificial intelligence is consuming enormous amounts of electricity, water, and critical minerals, but the solution isn't to stop building AI systems, it's to fundamentally rethink how we build them. Training a single large language model can use as much electricity as hundreds of households consume in a year, according to research highlighted by experts in the field. Yet the real environmental cost comes after deployment, when billions of daily prompts, searches, and predictions run continuously on energy-hungry data centers worldwide .

The stakes are higher than most people realize. Data centers already account for a significant share of global electricity demand, and this share is expected to rise sharply toward 2030 as AI becomes embedded across every sector of society. If that electricity comes from burning fossil fuels, AI risks becoming a climate-change accelerator rather than a solution .

What Is the Real Environmental Cost of AI Systems?

The environmental footprint of AI extends far beyond electricity consumption. Cooling data centers requires enormous amounts of water, creating new pressure on ecosystems and communities already experiencing drought and water stress. Before an AI system ever answers a single question, it may have already accumulated a significant ecological debt through the extraction of critical minerals needed for hardware .

The hardware for running AI systems depends on access to rare minerals extracted from vulnerable natural environments. This extraction often causes biodiversity loss, pollution, and serious human rights challenges in global supply chains. Additionally, major oil and gas companies are among the most sophisticated adopters of AI, using machine learning to analyze seismic data, identify new reserves, and optimize extraction from mature wells. By lowering costs and improving efficiency, AI can make fossil fuels more competitive compared to renewables, potentially extending the age of oil rather than accelerating the green transition .

How Can AI Become More Energy Efficient?

• Shift to Smaller Models: The dominant trend in AI has been "bigger is better," but this must change. Moving from large models to smaller, specialized models, from brute-force computing to knowledge-informed computational efficiency, has the potential to dramatically reduce energy use without sacrificing performance.
• Deploy Edge Intelligence: Running AI on local devices rather than at large, centralized data centers will reduce both energy demand and latency. This means processing happens closer to where it's needed, eliminating the need to send data back and forth to distant servers.
• Embrace Open Innovation: International collaboration and open-source software development can prevent redundant training of similar large models, saving enormous amounts of resources that must be harvested from vulnerable natural environments.
• Invest in Emerging Hardware: Neuromorphic computing inspired by the human brain, optical computing using photons rather than electrons for data transport, and quantum computing promise significant energy gains, signaling that AI does not have to remain permanently tied to today's energy-intensive architectures.
• Redesign Data Centers: Future data centers must be circular systems where waste heat supports district heating, backup power relies on green hydrogen instead of diesel generators, and facilities become integrated components of local energy ecosystems rather than isolated energy sinks.

These technological innovations alone, however, are not enough. Governance and policy must work hand in hand with engineering advances to ensure AI becomes truly sustainable .

Can AI Actually Help Solve the Climate Crisis?

Despite its environmental challenges, AI holds enormous potential as a catalyst for the green transition. The technology can optimize renewable energy systems by forecasting weather-driven generation patterns, managing grid stability, and coordinating distributed energy resources in real time. This is critical because wind and solar are variable and decentralized resources that require predictive modeling and adaptive control systems to balance supply and demand .

Beyond energy, AI can drive sustainability across multiple sectors. In agriculture, AI-powered sensors, drones, and predictive analytics allow farmers to apply water, fertilizer, and pesticides exactly where needed, sometimes down to individual plants. This results in higher yields, lower chemical runoff, and reduced emissions. In infrastructure and mobility, AI can optimize design to reduce material use, enable predictive maintenance to extend asset lifetimes, and improve traffic flow and shipping routes to reduce operational waste .

Healthcare systems can also benefit. AI enables earlier diagnosis, predictive analytics, and telemedicine, preventing disease escalation and reducing the need for high-intensity treatments and hospital stays. This lowers both costs and carbon footprints. Additionally, AI is accelerating scientific discovery itself, with machine learning models simulating ecological systems, forecasting biodiversity loss, and speeding up the discovery of high-capacity batteries and carbon capture materials essential for a post-carbon economy .

"Energy diversification is a strategic foresight for Geely Auto. A company's true strategic focus is ensuring every path leads to the future, which tests the technological depth and powerful energy resilience of our entire system," said Jerry Gan, CEO of Geely Auto Group.

Jerry Gan, CEO of Geely Auto Group

Real-world examples are already emerging. Geely Auto's new i-HEV Intelligent Hybrid technology integrates AI with advanced hybrid powertrains to achieve 2.22 liters per 100 kilometers fuel consumption and thermal efficiency of 48.41%, setting new world records. The system uses "AI Cloud Power," an innovation that enables real-time, self-optimizing petrol-electric energy management by sensing real-time exterior data such as temperature, humidity, and altitude. This intelligent approach improves overall energy efficiency by more than 10% compared to conventional engine-motor combinations .

What Conditions Must Be Met for AI to Support Sustainability?

For AI to truly decouple economic growth from the overuse of Earth's resources, three essential conditions must be met. First, intelligence must move to the edge, away from centralized data centers, and AI infrastructure must run on less energy while transitioning to renewable power sources. Second, policy direction must create incentives that reward AI applications reducing emissions and resource overuse, likely through tax regimes on digital resource consumption. Third, measurement and global cooperation are necessary to quantify AI's systemic impact, both negative and positive, to understand what actually works .

Environmental impact must become a formal part of AI accountability. Environmental, Social and Governance (ESG) frameworks should include mandatory reporting of energy use, water consumption, and material footprint. Standardized metrics, labeling schemes, and easy-to-use key performance indicators can empower users to choose greener AI providers. One approach is progressive taxation of digital resource consumption, keeping basic access affordable while discouraging wasteful overuse .

The ultimate test of AI may not be whether it writes better text or predicts markets more accurately. It may be whether it helps humanity achieve sustained prosperity without escalating ecological collapse. If AI can help decouple economic growth from the overuse of Earth's resources, it will not only be a technological revolution, it will be a turning point for civilization itself .