Look, it’s not like we need more reasons to be dubious about the advent of AI.

If you did, though, you might bring up how much electrical power it’s using on a regular basis.

The data centers that power chatbots like OpenAI’s ChatGPT and Google’s Bard are massive, and the energy they demand is growing rapidly to unwieldy amounts.

Data scientist Alex de Vries in the Netherlands recently published his analysis, finding that by 2027 (three short years away) these server farms will be using between 85 and 134 terawatt hours of energy every year.

For reference, that’s around the same annual energy use as Argentina, a country with a population of over 45 million people. Currently, a country of that size has 0.5% of the energy demands of the entire world.

This is a complaint similar to those surrounding the consumption needs of the crypto industry, with no one having a plan for dealing with leaving such a massive carbon footprint.

Scientists say that alone should be enough to make people reconsider their investment in AI. Companies like OpenAI are secretive about how much energy they use, says de Vries, so no one thinks they’re going to be deterred by this argument.

He estimated the numbers by looking at sales of Nvidia A100 servers, which are about 95% of the servers used by the AI industry.

“Each of these Nvidia servers, they are power-hungry beasts.”

Experts like professor Roberto Verdecchia agree that taking time and space to reevaluate would be smart.

“Maybe we need to ideally slow a bit down to start applying solutions that we have. Let’s not make a new model to improve only its accuracy and speed. But also, let’s take a big breath and look at how much are we burning in terms of environmental resources.”

The companies that are headquartered in California, at least, will have to contend with recent climate disclosure laws signed by Governor Newson. They force companies to disclose how much carbon they produce by 2026.

Even so, right now AI companies are basically governing themselves, which is a recipe for disaster.

It seems likely they will burn through whatever energy is needed to keep their models ahead in the race.