AI industry could consume as much energy as a country the size of the Netherlands by 2027, a new study warns.

Following ChatGPT’s launch last year, large tech companies have scrambled to add AI-powered services.

Online usage is much more energy-intensive due to the fact that they consume far more power than conventional applications.

In addition, the study said that if AI’s current growth slowed, its environmental impact might be less than feared.

According to many experts, including the author of the report, such research is speculative since tech companies do not divulge enough data to make an accurate prediction.

Although AI requires a wider range of hardware than conventional computing, there is no doubt that it requires a higher level of performance.

As Alex De Vries, PhD candidate at the VU Amsterdam School of Business and Economics, points out, the study is based on the assumption that some parameters remain unchanged, such as the rate at which AI grows, the availability of AI chips, and servers working continuously.

It is estimated that Nvidia supplies 95% of the AI processing kit needed by the sector, according to Mr De Vries.

As a result of looking at the amount of computers expected to be delivered by 2027, he approximates the energy consumption of AI as 85-134 terrawatt-hours (TWh).

An average small country consumes roughly that amount of power each year.

Using artificial intelligence only when it is necessary, according to Mr De Vries, is a good idea.

It takes warehouses full of specialist computers – called data centres – to operate AI systems like OpenAI’s ChatGPT and Google’s Bard, which power popular chatbots.