How can AI help build a more sustainable environment when it requires an immense amount of water and energy to operate? | Insights with M. Chairul Ridjal

Artificial intelligence has become a central force in global sustainability efforts, yet it also brings environmental concerns that cannot be ignored. Training and operating large AI models requires vast amounts of electricity and water, and recent analyses show that some frontier models have consumed more than 50 GWh of electricity just to be developed, roughly equal to the annual energy use of several thousand households. Data centers, which enable AI to run, already consume about 1 to 1.5 percent of global electricity, and projections suggest that this share could increase significantly as AI adoption expands. Water usage is also a major challenge, as cooling systems rely heavily on water, sometimes amounting to millions of liters per data center per year. These concerns have prompted a growing debate about AI’s true environmental cost.

Even so, many experts argue that when designed responsibly, AI’s benefits for sustainability can outweigh its resource burden. One of them is M. Chairul Ridjal, a student at IPB University involved in the AYC program, who emphasizes that the key is whether the impact created by AI exceeds the cost of running it. He highlights that if AI systems are powered by renewable energy, and if models are intentionally optimized for efficiency rather than endlessly scaled, then the technology can become a net-positive force for the environment. AI already assists in renewable energy forecasting, particularly in predicting solar output, enhancing grid stability, modeling climate risks, and improving disaster prediction accuracy. Ridjal notes that in community-based solar energy discussions, AI-based forecasting tools allow for faster, smarter resource planning, ultimately reducing emissions and strengthening climate resilience.

However, the seriousness of AI’s environmental footprint depends largely on development practices. If companies pursue maximum performance without considering compute efficiency, cooling demands, or the source of electricity, AI becomes environmentally costly. But this is not an inevitability. Developers can significantly reduce impact by using renewable-powered data centers, improving model efficiency, compressing models rather than expanding them unnecessarily, and reducing reliance on massive centralized compute systems through edge computing. A major gap Ridjal points out is the lack of transparency: in Indonesia, for example, companies are not required to report how much water or electricity their data centers consume, making it difficult to evaluate environmental impacts or push for greener practices.

Whether AI’s environmental cost is justified eventually comes down to how the technology is directed. When renewable energy powers data centers and when models are built with efficiency principles, AI can offer benefits far greater than its footprint. AI is already contributing to wildfire prediction, precision agriculture, waste sorting, and flood forecasting, and smart energy management systems are helping balance demand and improve renewable integration. These applications show that AI is capable of addressing environmental challenges in ways that traditional methods cannot match.

Ridjal’s experience with youth sustainability initiatives also reflects how AI can meaningfully support sustainability outcomes in practice. Tools that project solar irradiance, predict climate risks, or guide energy planning make community projects more resilient and data-driven. Within the ASEAN Youth Council, he and others promote responsible, impact-driven AI development, encouraging students and innovators to build solutions for real problems while paying attention to ethics and interdisciplinary collaboration.

This highlights the need for clearer regulations and environmental standards for AI. Governments should require transparency in energy and water reporting, incentivize renewable-powered infrastructure, and encourage efficiency-first model design. If these frameworks are put into place, AI can evolve into one of the most important tools for building a sustainable future rather than an additional source of environmental strain. Ultimately, the question is not whether AI is environmentally costly, it is whether we choose to build it responsibly enough to ensure the benefits outweigh the burdens.