All the energy that artificial intelligence consumes

The evolution of artificial intelligence is bringing to light a big problem for our planet:the energy impact is becoming unsustainable.

A search carried out on ChatGpt would consume ten times the amount of energy needed for a search on a traditional search engine like Google.This is what the International Energy Agency says.The incredible rise of generative artificial intelligence in recent years has often distracted from some relevant and disturbing factors:among all, the environmental impact of generative AI, which requires large quantities of electricity, but also of water, to work.

This type of technology is in fact based on the "work" of data centers, i.e. powerful and sophisticated processing centers, which analyze and process enormous amounts of data to develop answers.Every banal interaction with ChatGpt or other AI capable of generating images and videos, such as those presented by OpenAI, Google and other companies in recent months, therefore requires great computational power, and therefore electricity.

The US electricity grid is in trouble due to artificial intelligence

The increase in interest in similar services has written the Washington Post, is already putting the US electricity grid in difficulty, forcing these centers to suck even more energy, even from dirty sources, the return of coal, in short.

One fact above all: a single Meta data center consumes as much as seven million laptops consume every day which are used for eight hours a day for an entire year.

This happens right in a sector dominated by a few giants who, over the years, have invested heavily, and with good results, to make these data centers more sustainable.Google has for example adopted a “transparent approach” on its water consumption, necessary to cool its continuously processing machinery, and promised to use clean energy 24 hours a day to power its data centers, by 2030.

Where does all this energy come from?

But the AI ​​race is on reviewing their priorities and therefore also these announcements, and is favoring a network of smaller entities, there are currently 2,700 processing centers in the United States, which do not always have the environment at heart, nor do they feel the social and political pressure to use clean sources.“They're starting to think like the chemical and cement industries,” he explained to the Washington Post Ganesh Sakshi, manager of Mountain V Oil & Gas, a natural gas company:“Those who have contacted us have an agnostic approach to where the energy comes from.”

To make the chatbots on which practically every technological company is betting everything work, therefore, we are increasingly resorting to fossil fuels, "dirtying" one of the few sectors which, at least superficially, had always focused heavily on the use of sustainable alternatives.

At this point, Silicon Valley might stop for a second and consider what has been done so far, the billion-dollar investments in a still experimental technology which, beyond the surprising results, according to some, has created a speculative bubble of pure hype.As written recently by the Wall Street Journal, “the AI ​​revolution is already losing steam,” with many of the early adopters of tools like Microsoft's AI they lift the first doubts, and the persistent fear that this technology is not economically sustainable.Also because, as seen, to make it work you need energy, but also powerful, sought-after and very expensive technological tools, such as Nvidia chips.

However, the Big Tech sector seems destined to choose another path, far from the caution just described, and seems to want to solve the energy problem by pursuing a miraculous source of clean and abundant energy: nuclear fusion.

The dream?Nuclear fusion by 2028

Sam Altman, co-founder and head of OpenAI, has invested in Helion Energy Inc., a startup he aims to build a nuclear fusion power plant by 2028.To be clear, this is the same nuclear phenomenon that powers the Sun and every other star in the universe:For decades, research centers and companies have been investing to bring this process to Earth, on a tiny scale, which could produce enormous quantities ofclean energy.

Recently the European Eurofusion consortium, which includes 31 countries including Italy with Enea, he announced to have concluded successfully the largest nuclear fusion experiment in the world, producing 69 megajoules of energy with just 0.2 milligrams of fuel, in a process that joins the nuclei of two isotopes until they fuse, releasing an enormous amount of energy.The goal is to make this process less extraordinary, and therefore more continuous and above all safe.

Helion Energy is pursuing a similar path, also thanks to funding from Altman, which sees the startup as the key to solving the AI ​​energy problem.AND Microsoft too, a company that has a well-established alliance with OpenAI, invested in the project, promising to buy energy produced in this way by the company.But we are, as they say, at vaporwave.To the fluff.Of course, it would be nice to be able to produce clean and safe energy by simulating stellar processes but are we sure that this turning point is just around the corner?And above all, if we succeeded, should the first application really be powering data centers for AI?

Altman himself had declared in recent months that money was needed to solve the technological and environmental problems of the sector by producing very powerful and less energy-intensive chips.Many.How many? Seven trillion dollars, that's how much, a figure just under double the German GDP, to give you an idea.In short, an economically unsustainable sector which is already causing environmental damage, all without mentioning the controversial problems these companies have with copyright, the relationship with artists and the public itself.But, of course, seven trillion dollars should be enough to solve everything.