Done getting your 80s photos ? Now comes the less glamorous part: What does it take to create them?
One AI image may have a relatively small footprint on its own. The bigger concern is what happens when millions of people generate images, try multiple versions and jump on the same viral trend. The scale is striking. Energy use also varies widely depending on the task.
Research from Carnegie Mellon University and Hugging Face estimated that generating an AI image can use between 0.01 and 0.29 kilowatt-hours of electricity, depending on the system and image-generation process, Firstpost reported. Inference — the day-to-day use of AI models — accounts for roughly 80% to 90% of their total energy demand, according to the United Nations University. A popular AI platform like ChatGPT is estimated to process around 2.5 billion prompts per day, consuming hundreds of gigawatt-hours of electricity each year. A typical conversational AI query is around 200 times more energy-intensive than basic text classification. Generating a single AI image can require around 1,450 times that baseline, while a single short AI-generated video can consume as much electricity as 200,000 spam classifications, the UN said.
Not to alarm you, but every AI-generated image requires computing power, which means electricity and, indirectly, water.
The study projects that AI’s global water footprint could reach 4.2–6.6 billion cubic metres annually by 2027. It also notes that two-thirds of post-2022 data centres are located in water-stressed regions. The electricity required to generate a typical AI image is enough to power a 10-watt LED bulb for about 17 minutes. For a high-complexity AI video, the equivalent energy could power the same bulb for 42 hours. The estimated electricity-associated water footprint of a single AI image is around 29 ml, or roughly two tablespoons. For a complex AI video, the estimated electricity-associated water footprint rises to about 4.1 litres. Data centres consumed 448 terawatt-hours of electricity globally last year, with AI accounting for about a fifth of the total. They also consumed 4.5 trillion litres of water and generated 189 million tonnes of carbon dioxide emissions, Reuters reported, citing researchers. By 2030, annual power consumption from data centres could almost double to 945 TWh, with AI accounting for 40% of the total. Water consumption could reach 9.3 trillion litres, while CO₂ emissions could rise to 399 million tonnes, the report added. The United Nations University research also warns that AI infrastructure could generate up to 2.5 million tonnes of electronic waste a year by 2030. It says the land footprint of data centres could rise from around 6,900 sq km to more than 14,500 sq km by 2030. Google’s planned $15 billion data-centre hub in Andhra Pradesh is facing opposition from environmentalists over concerns about water supplies and wildlife, Reuters reported in August.
Reuters reported that the project has also faced legal challenges over water supplies and its proximity to the wildlife sanctuary. Model choice, prompt length, output format and resolution all materially shape the footprint. Yet most of these decisions are made invisibly through product defaults that users never see. The UNU also warns that efficiency improvements alone are unlikely to offset rising demand. A study published in Water Research, titled ‘The water footprint of artificial intelligence: Emerging solutions and governance imperatives’, notes that AI infrastructure consumes freshwater through evaporative cooling, indirect water use in electricity generation and water-intensive semiconductor manufacturing. The United Nations University has put the footprint of individual AI tasks into more relatable terms. This refers to the water associated with generating the electricity required for the task, rather than water directly consumed by the image generation itself. The project is being built near a wildlife sanctuary, while activists have raised concerns about its potential impact on Visakhapatnam’s already strained water resources. The Andhra Pradesh government has denied allegations that the project was fast-tracked without assessing risks to water supplies and wildlife.
As AI becomes cheaper and more efficient, people may simply use it more — a so-called “rebound effect” that can ultimately increase total resource consumption.
