Artificial intelligence is consuming staggering amounts of energy and water, yet universities are uniquely positioned to teach the next generation how to build greener AI systems. A single large language model can generate a carbon footprint equivalent to five times a car's lifetime emissions, while data centers globally may soon consume six times more water than Denmark. At the same time, higher education institutions are embedding sustainability into AI curricula and challenging students to solve this paradox.

How Much Energy Does AI Actually Use?

The numbers are sobering. According to the International Energy Agency, global data center power usage reached 240 to 340 terawatt hours in 2022, and forecasts suggest AI could consume 20% of the global electricity supply by 2030 if current growth trends continue. Training a single large AI model like BERT, a natural language processing system, generates approximately 626,000 pounds of carbon dioxide, equivalent to five times the lifetime emissions of one car.

The water footprint is equally concerning. Data centers use water during construction and ongoing operations to cool electrical components. AI-related infrastructure may soon consume six times more water than Denmark, a country of 6 million people, at a time when roughly 26% of the global population lacks access to clean or safely managed drinking water.

In the United States, the concentration of data centers in Northern Virginia illustrates the scale of the problem. Loudoun County alone has more than 200 data centers built, with another 117 in the planning stages, providing 12,000 jobs and serving 3,500 companies. However, data centers are now the only growing source of energy demand in Virginia, and they are expected to double the load on the state's electricity grid, which is primarily powered by natural gas, by 2040.

What Are Universities Teaching About AI's Environmental Cost?

Higher education institutions are taking the problem seriously by integrating AI sustainability into their teaching. Universities are embedding the UN Sustainable Development Goals into AI courses, ensuring students understand both the promise and the price of the technology they are learning to build. This approach goes beyond simply listing environmental concerns; it connects AI development to real-world sustainability challenges.

One example comes from a high school student's independent research project. Owen Ahern, a first-year finance student at the University of Tampa, investigated how AI's expansion has affected energy sources and consumption, examining everything from data centers to emerging technologies that could make AI more sustainable. His work demonstrates how students, when given the freedom to investigate pressing questions, can develop critical thinking skills that extend beyond their discipline.

"IB taught me how important it is to understand the perspectives of others," said Ahern, reflecting on his research process. "That is something I'll carry with me well beyond high school."

Owen Ahern, First-year student at the University of Tampa

Universities are also emphasizing interdisciplinary collaboration as essential to solving the climate challenge. The Sustainable Development Goals do not fit neatly into different disciplines, so institutions are facilitating project collaboration among students from engineering, business, environmental science, and other fields to make sustainability relevant across all majors.

How to Build More Sustainable AI Systems

• Neuromorphic Computing: Unlike conventional processors that rely on sequential execution, neuromorphic processors can operate in parallel, processing multiple tasks simultaneously. The neuromorphic chip created by the Korea Advanced Institute of Science and Technology in 2024 uses just 1/625th of the power of NVIDIA graphics processing units while maintaining impressive AI capabilities.
• Model Optimization Techniques: Methods such as model pruning and knowledge distillation reduce the size and computational requirements of AI models. Model pruning selectively removes redundant parts of neural networks, while knowledge distillation trains a smaller AI model to replicate the behavior of a larger one, saving energy that would otherwise be wasted.
• Renewable Energy Integration: Companies are exploring power from solar and wind farms as a way to drastically cut emissions from the energy required to operate AI systems, addressing the grid's reliance on natural gas and fossil fuels.

Why University Leadership on AI Sustainability Matters

Universities have a unique responsibility to lead by example in promoting sustainability and environmental awareness. As centers of knowledge and innovation, they can influence how the next generation of engineers, data scientists, and business leaders approach AI development. This includes not only teaching the technical solutions but also fostering critical thinking about the ethical dimensions of AI's environmental impact.

The challenge is significant. Investment in AI is accelerating globally. In 2023, the United States invested $50.6 billion in AI, followed by China at $11.2 billion and the European Union at $6.1 billion. With 63% of organizations globally planning to adopt AI within the next three years, AI's market size is expected to grow by at least 120% year over year, and AI is projected to contribute $15.7 trillion to the global economy by 2030.

Yet the environmental cost cannot be ignored. Making a 2-kilogram computer requires approximately 800 kilograms of raw materials, according to the United Nations Environment Programme. The microchips that power AI also need rare earth elements, which are often mined in environmentally destructive ways.

Universities are responding by creating opportunities for students to contribute to institutional sustainability efforts and develop initiatives of their own, giving them agency to translate their knowledge into action. Some institutions have engaged hundreds of student volunteers in climate accountability projects, teaching them net-zero literacy while building their employability. Others are teaming up with students to design sustainability education, involving them as co-creators and mentors in sustainability courses to address differences in understanding and issues such as eco-anxiety.

The path forward requires both technological innovation and cultural change. Neuromorphic AI chips, renewable energy integration, and model optimization techniques offer concrete solutions to reduce AI's carbon footprint. But equally important is ensuring that the next generation of AI developers, deployed across industries and governments, understands the environmental stakes of their work and is equipped to prioritize sustainability alongside performance.