The 1980s are back and you need AI to get there.

A new social media trend has people uploading their photos to ChatGPT and asking it to reimagine them as if they lived in the 1980s.

Within seconds, modern photographs become grainy portraits with big hair, retro clothes and vintage backgrounds.

It is fun. It is also a perfect example of how casually we now use generative AI.

We ask it to write, draw, edit, summarise and generate images without thinking about what powers the technology.

That needs to change.

AI Is Not Weightless

An AI-generated image may appear on your screen as if it came from nowhere but that’s not the case.

The image is processed by powerful computers inside data centres. Those machines consume electricity and generate heat, requiring cooling systems to keep them running.

The infrastructure behind AI also requires water, land, hardware and vast networks of energy and connectivity.

The exact environmental cost of one AI image varies depending on the model, hardware, data centre and cooling system. That makes viral claims assigning one fixed amount of water or electricity to every prompt misleading.

It does not make the underlying problem disappear.

AI requires resources. More AI use requires more infrastructure. And Malaysia is building that infrastructure at speed.

Malaysia Is Powering the AI Boom

Malaysia has become one of Southeast Asia's fastest-growing data-centre hubs, attracting major investments from technology companies including Amazon and Microsoft.

The demand is already showing up on the power grid.

Data centres accounted for a record 9.3% of Malaysia's total electricity consumption in mid-August, up from an average of about 7% for the year, according to Reuters. If growth continues, they could consume as much as 31% of Peninsular Malaysia's electricity by 2035.

That surge is happening as temperatures rise and data centres require even more power for cooling.

Malaysia is also preparing for more AI infrastructure. On Sept 8, Nvidia-backed Firmus signed a multi-year agreement with OpenAI to provide computing capacity from two Malaysian data centres.

The message is clear. The country is preparing to support much more AI use.

The “Cloud” Is Physical

Every AI-generated photograph, video or block of text depends on physical infrastructure somewhere.

The servers need electricity. The electricity needs generation and transmission infrastructure. The servers produce heat. The heat needs to be removed. The facilities need land, equipment and water.

This is why the AI boom is increasingly becoming an energy and infrastructure issue rather than just a technology story.

Governments around the world are now confronting the same problem. In the US, data-centre expansion has become tied to concerns over electricity costs and who should pay for the infrastructure needed to support it.

Malaysia is already having its own version of that conversation.

In February, Prime Minister Anwar Ibrahim said the government had restricted new data-centre projects unrelated to AI because of their electricity and water consumption. He said Malaysia wanted to prioritise facilities that bring higher-value AI and advanced technology benefits.

Even our government recognises that data centres are competing for finite resources.

Yet the AI label makes that consumption easier to justify.

We Are Making AI Disposable

The technology has become so easy to use that we are applying it to increasingly trivial things.

Want to see yourself in 1985? Generate it.

Don't like the first result? Generate another. Change the outfit. Change the background. Change the hairstyle. Do it again.

The same pattern applies across AI-generated memes, advertisements, videos, avatars and endless variations of content.

The issue is not one photograph.

It is the scale of consumption created when computationally intensive technology becomes a casual, disposable service.

AI's biggest environmental challenge may not be that one prompt has an enormous footprint. It is that we are building an economy around generating millions of them.

Malaysia Wants The Future And It Needs To Pay For It

There are good reasons for Malaysia to pursue AI infrastructure.

Data centres bring investment. They can create jobs, attract technology companies and strengthen Malaysia's position in the regional digital economy.

But development has a price.

More data centres mean more electricity demand. More cooling means more resource demand. More infrastructure means more pressure on the systems that support everyone else.

That cost cannot simply disappear because the technology is labelled innovative.

Malaysia should absolutely pursue technological growth. It should also be asking harder questions about the resources required to sustain it.

Look Beyond The Picture

The irony of the 1980s trend is hard to miss.

We are using cutting-edge AI to recreate an era before smartphones, algorithms and generative technology existed.

The nostalgic image looks analogue. Everything behind it is anything but.

We see the finished photograph. We do not see the servers processing it, the electricity powering them or the infrastructure being built to support our growing appetite for AI.

Malaysia is betting heavily on that appetite.

So the next time you ask AI to turn your selfie into a 1980s portrait, remember what you are actually using.