The researchers tried to add up all the motherboards, graphics, chips, cooling, batteries and other electronic components necessary for the global use of artificial intelligence.
Photo: AI
The expansion of artificial intelligence may dramatically increase the amount of e-waste by 2030.
The hardware used by generative artificial intelligence may end up as worn-out e-waste by the end of the decade, a new study suggests. However, one aspect that is often overlooked is the hardware necessary to use them, as they are power and energy intensive.
However, hardware has a certain shelf life – when it wears out or becomes obsolete, it must be replaced. Old hardware thus becomes e-waste. Therefore, scientists from the Institute of Urban Environment of the Chinese Academy of Sciences in Beijing and Reichman University in Tel Aviv tried to add up all the motherboards, graphics, chips, cooling, accumulators and other electronic components needed to use GenAI. They published the results of their research in the form of a study in the scientific journal Nature Computational Science.
They used the amount of hardware typically needed to run a particular GenAI application in a standard data center or server farm and the average lifespan of individual components to make the estimate. Then they counted the individual data centers. Based on the available data, they estimated the expected interest in individual applications and their services in the coming years. Finally, they used the data to create a computer model to estimate the amount of hardware.
The model calculated that if nothing changes and everything continues at the current pace, the AI industry could produce between 1.2 and five million tons of e-waste by 2030. In the past year, it was only 2,600 tons. Researchers point out that this can be avoided if hardware manufacturers adopt a circular economy and recycle hardware.
Source: vat.pravda.sk
