This article was first published in the forum, The Edge Malaysia Weekly, from June 8, 2026 to June 14, 2026.
Artificial intelligence (AI) has become one of the most powerful applications of cloud computing. From writing code to analyzing documents, AI is now embedded in everyday business workflows. As adoption accelerates, so does the environmental footprint of the infrastructure required to run these systems.
A common misconception is that AI’s greatest impact on the environment lies in training language models at scale. In reality, the majority of emissions come from inference, the process of using trained models to generate output. Inference occurs every time a user submits a prompt. While model training is controlled by a small number of AI providers, inference is driven by how organizations use AI in their daily operations. This allows companies to directly influence both the cost and environmental impact of using AI.
The scale of this impact is significant. The International Energy Agency predicts that the world’s data center power demand will more than double by 2030, driven primarily by generative AI (Gen AI). The World Economic Forum estimates that accelerated adoption of AI could lead to an additional 4.2 billion cubic meters to 6.6 billion cubic meters of water withdrawals by 2027. Data centers consume large amounts of electricity and water, especially for cooling, putting a strain on the world’s energy and water supplies.
In many regions, electricity grids are already under pressure as demand grows faster than new capacity can be built. Water scarcity is also becoming a significant risk as data center cooling competes with residential, agricultural, and industrial needs. In response, markets such as Malaysia and Singapore have moved away from unlimited data center expansion to more stringent, sustainability-focused planning frameworks that place greater emphasis on energy efficiency, carbon intensity, and water use efficiency.
Not all AI workloads are the same. Traditional AI typically involves machine learning models trained on structured data for tasks such as classification, regression, anomaly detection, and recommendations. These models are generally small, often run on central processing unit (CPU) or modest graphics processing unit (GPU) configurations, and consume relatively little energy per prediction. In contrast, Gen AI relies on large language models with billions of parameters to generate text, images, code, and other content. Because each request requires a large amount of GPU computing, Gen AI consumes significantly more energy with each use.
Environmental impact ultimately comes from the infrastructure activated by each AI request. Every prompt triggers coordinated computations across GPUs, CPUs, memory, networking, and storage within your data center. Gen AI usage is typically measured in tokens that represent units of text, images, audio, or other data processed by the model. When both input and output are included, a single prompt can generate hundreds or thousands of tokens. Although tokens are not a direct proxy for energy consumption, increasing the amount of tokens generally increases computing demand and increases electricity and water usage.
There are practical steps available to organizations. It is most effective to reduce the amount of unnecessary token usage. Instead of submitting the entire document, users can limit input to relevant sections. Clear, well-constructed prompts often yield better results with fewer resources. You can also batch non-urgent workloads to process them more efficiently if you don’t need real-time response.
Model selection is also important. Not every business task requires the most advanced model available. Smaller models, often with billions rather than tens of billions of parameters, can meet operational needs at a fraction of the environmental cost.
As AI becomes integrated into everyday business activities, its impact on the environment can no longer be treated as an abstract or upstream issue. Every quick, automated workflow and deployed model gradually contributes to energy usage, water consumption, and infrastructure demands. For organizations committed to environmental, social, and governance (ESG), the responsible use of AI is therefore not only a technology but also a governance issue. By increasing visibility into your inference usage and making disciplined choices about prompts, workloads, and model selection, you can ensure AI innovation advances without compromising sustainability and carbon transition goals.
Francis Xaviour Joe is Director of Client Relations at Sustainlaterre PLT, a sustainability software solutions provider.
Save by subscribing to us for your print and/or digital copy.
P.S.: The Edge is also available on Apple’s App Store and Android’s Google Play.
