Everyone’s Making AI 80s Photos. But What Does One Image Really Cost the Planet?

The 1980s are back on Instagram.

Users are turning their photos into retro AI portraits with big hair, vintage clothes and film-style effects.

The images may look like harmless nostalgia.

But the viral trend has also raised questions about the environmental cost of generating AI images.

And with millions joining in, that cost is becoming harder to ignore.

How Is Everyone Making These 80s Pictures?

The process is fairly simple for the user.

A clear photograph is uploaded to an AI image-generation tool such as ChatGPT. The user then gives instructions to preserve their recognisable facial features while changing the styling, surroundings and photography to match the 1980s.

Prompts can ask for things such as voluminous hairstyles, vintage clothing, denim jackets, statement accessories, old Bollywood-inspired settings, warm lighting, faded colours and a film-camera finish.

The model then generates a new image rather than simply applying a conventional Instagram filter. Users can keep refining it by asking the system to change the clothes, hairstyle, background or overall aesthetic. 

That is similar to how the earlier Ghibli trend worked. Instead of manually editing an image, users describe the desired visual style and let a generative AI model create a new version.

The Picture Is Instant. The Computing Is Not

The biggest environmental cost is hidden behind the word “generate”.

AI image generation requires computing power. That computing happens in data centres filled with specialised processors and other infrastructure.

Research cited by GQ India estimates that generating a single AI image can use between 0.01 and 0.29 kilowatt-hours of electricity, depending on the system and task. 

One image is not enough to make the environmental impact obvious.

The problem is scale.

A user may generate five versions before choosing one. Another may make ten. A viral trend can involve millions of image requests, with each request adding to the overall demand on AI infrastructure.

The International Energy Agency estimates that global data-centre electricity consumption could rise from around 460 TWh in 2024 to about 945 TWh by 2030. AI is expected to be one of the biggest drivers of that growth. 

So the question is not whether one retro portrait is going to “destroy the planet”. It is what happens when entertainment-scale AI use becomes routine for millions of people.

Then There Is The Water Bill

Data centres also produce large amounts of heat and need cooling.

Some cooling systems use water, while electricity generation itself can also have a water footprint.

A 2023 study by researchers from the University of California, Riverside and the University of Texas at Arlington estimated that training GPT-3 could directly consume around 700,000 litres of freshwater. The researchers also modelled the water used during the model's operation. 

But there is an important caveat.

The viral claim that “one ChatGPT prompt uses 500 ml of water” is not what the study actually found. Its estimate was roughly 500 ml for 10–50 medium-length responses, depending on location and operating conditions. It was a modelled estimate for GPT-3, not a universal measurement for every current AI tool. 

That distinction matters because AI's environmental footprint varies according to the model, hardware, data-centre location, cooling system, electricity source and type of task.

It Starts Before The Prompt

There is another part of the footprint that is easy to miss.

The processors used by AI data centres need semiconductors and other materials. Producing this hardware involves mining, manufacturing, energy use and chemical processing.

GQ India notes that modern computing hardware carries a substantial material footprint, while the supply chain for AI infrastructure depends on critical minerals and semiconductor manufacturing. 

In other words, the environmental cost of an AI image does not begin when someone presses “generate”.

It extends from the hardware supply chain to the data centre and the electricity and cooling required to run it.

From Bollywood To Parliament

The trend has also moved well beyond ordinary Instagram users.

Bollywood names including Kiara Advani, Sidharth Malhotra, Parineeti Chopra, Ameesha Patel, Tahira Kashyap and Saba Ali Khan have shared versions of the retro aesthetic. 

Politicians have joined too.

Piyush Goyal shared an AI-generated 1980s-style image, while Kiren Rijiju and Shashi Tharoor also participated in the retro wave. Reports have also featured Himanta Biswa Sarma and Anurag Thakur in connection with the trend. 

There is, however, a difference between some of these posts. Tharoor has also shared genuine photographs from the 1980s, rather than relying only on an AI recreation. 

The Bigger Question 

AI is becoming more efficient, with electricity use per task falling as technology improves. But growing usage and more energy-intensive applications are increasing overall demand.

That raises a bigger question around viral AI trends like 80s portraits.

The issue is not whether one image is harmful. It is how millions of low-stakes generations add to the physical infrastructure behind AI.

The trend may fade, but another will follow. The bigger question is whether we understand the environmental cost behind every viral AI moment.

TL;DR | News At a Glance

What is the AI 80s trend?
It turns regular photos into 1980s-style portraits using AI.

Why is it being discussed?
The trend has renewed debate about the environmental cost of generating AI images.

Does one AI image have a major environmental impact?
The concern is less about one image and more about the cumulative impact when millions of people generate images.

Is AI becoming more efficient?
Yes. But rising usage and more energy-intensive AI applications are also increasing overall demand.

What is the bigger concern?
Viral AI trends can make digital creations feel effortless while hiding the physical infrastructure needed to produce them.