The Environmental Cost Behind the AI-generated 1980s Photo Trend

The resurgence of 1980s-style AI-generated images has raised concerns about their environmental impact. While the images themselves do not produce visible pollution, the process of creating them demands significant energy and water resources from data centers. Factors that influence the environmental cost include the AI model used, image resolution, hardware efficiency, electricity source, and cooling systems. No single estimate can accurately represent the energy or water consumption for each AI-generated image.

The popularity of this trend stems from its simplicity: users upload a photo, input a prompt, and receive a past-decade version of themselves. However, each request necessitates computational operations to interpret, process, and generate new images. Although individual requests require minimal processing power, the collective demand from millions of users can be substantial. A 2025 study found considerable differences in energy efficiency among image-generation models, but broad estimates should be approached cautiously.

Estimates of global AI-generated image production vary, with some sources suggesting around 80 million per day. However, it is crucial to note that there is no global system to track every image. The growth of AI applications is occurring amid a broader increase in AI infrastructure. Data center electricity consumption rose by 17% in 2025, and consumption specifically related to artificial intelligence grew faster still.

Water consumption is another aspect of AI's environmental footprint. Data centers require cooling systems, which may use water directly in some cases. A study cited in research on AI's water footprint estimated that training the GPT-3 model required around 700,000 liters of freshwater, but this figure applies to model training and not subsequent image generation.

Water usage during operation depends on the infrastructure and the electricity powering it, with location also playing a role. A data center in a cooler region may rely more on outside air for cooling, while warmer areas may require different cooling systems. The water footprint of AI includes direct cooling, indirect electricity generation consumption, and semiconductor manufacturing.

Lastly, the hardware behind AI, including graphics processing units, servers, and semiconductors, requires extensive resources and energy-intensive industrial processes to manufacture. Therefore, the environmental impact of AI-generated images extends beyond the user's prompt, encompassing a broader technological chain. Global data center electricity consumption is projected to rise from 485 TWh in 2025 to approximately 950 TWh by 2030, with AI demand contributing significantly to this increase.