AI is not only consuming a lot of energy, it is also generating a large amount of electronic waste

We are witnessing firsthand how some of the largest companies on the planet invest billions of dollars in artificial intelligence (AI). It is a discipline whose advances promise to be integrated into all types of sectors: from transportation to education and medicine. And of course, the business opportunities are enormous.

But to stay at the forefront, a cost is being assumed that transcends the economic. The data centers on which AI language models are trained and run consume large amounts of energy and, depending on their cooling system, require a lot of water. In addition, they generate a lot of electronic waste.

AI and electronic waste

A recent study prepared by leading researchers suggests that the total electronic waste associated with AI could be between 1.2 million and 5 million tons by 2030. The document warns about the current trend, and encourages the different actors in this industry to manage waste proactively.

Data from the United Nations Global E-Waste Monitor They point out how the amount of e-waste are increasing five times faster than recorded recycling efforts. In 2019, around 53.6 million tons were generated, while by 2022 the figure had grown to 62 million tons.

We can see this scenario from several perspectives, but we can focus on the one we mentioned at the beginning of the article: investment. Spending on AI multiplied eight between 2022 and 2023, exceeding $25 billion. Much of this money has been allocated precisely to building and equipping AI data centers.

A key point of this type of infrastructure is that the replacement of components is not only limited by their useful life, but by their performance. Many companies are interested in incorporating the latest GPUs from NVIDIA and other manufacturers so that their data centers have greater computing capacity, even when the previous ones are still working.

Replacing components early is not something new in the business world. For some time now, the industry has preferred to destroy functional hard drives rather than recycle them for fear that certain private data They can fall into the wrong hands, although there are initiatives that seek to avoid these practices.

The researchers’ proposal has to do with reusing the components that are replaced. They understand that they won’t be suited to the most demanding tasks, but they envision them powering data centers that support less heavy tasks, such as website hosting and backups. They also propose donating them to educational centers.

Images | Xataka with Bing Image Creator | Meta AI

In Xataka | 01.AI is the new AI giant in China that competes with the best of OpenAI. Your problem is precisely China