According to a recent study, the artificial intelligence (AI) sector may by 2027 use as much energy as the Netherlands as a nation.
Since ChatGPT exploded onto the market last year, major internet companies have been rushing to integrate AI-powered services.
They consume far more energy than traditional programmes, which increases the energy cost of using the internet.
The report did note that if AI’s present expansion slows down, its potential environmental impact may be less than initially thought.
Numerous experts, including the author of the report, argue that this kind of research is speculative because IT companies don’t release enough information to allow for precise forecasting.
But without a doubt, AI jobs demand more powerful hardware than conventional computer tasks.
The analysis by Alex De Vries, a PhD candidate at the VU Amsterdam School of Business and Economics, is predicated on the assumption that a few key variables stay the same, including the rate of advancement in AI, the accessibility of AI chips, and the continuous operation of servers.
According to Mr. De Vries, Nvidia is anticipated to provide roughly 95% of the AI processing equipment needed by the industry.
Based on the anticipated volume of these computers by 2027, he was able to estimate that AI will use between 85 and 134 terawatt-hours (TWh) of electricity annually.
At the upper end, that is approximately the annual power consumption of a small nation.
“In terms of electricity usage, you would be talking about the size of a nation like the Netherlands. You’re discussing roughly 0.5 percent of the world’s total electricity usage,” he said to BBC News.
Nvidia chose not to respond.
AI should only be applied in situations where it is truly necessary, according to Mr. De Vries’ research.
The journal Joule has published his peer-reviewed paper.
How much water and energy does AI require?
For AI systems to function, such as the massive language models behind well-known chatbots like Google’s Bard and OpenAI’s ChatGPT, warehouses full of specialised processors, or data centres, are needed.
This implies that the equipment requires more water-intensive systems to keep cool and is more power-hungry than standard kit.
A little manual to assist you in understanding AI
The energy needed for cooling was not included in the research. Many large tech companies don’t measure this particular water or energy use. Among those urging the industry to be more open about it is Mr. de Vries.
However, there is no denying that demand for the AI-powered computers is skyrocketing, as is the energy required to keep those servers cool.
At the beginning of 2023, Danny Quinn, the CEO of the Scottish data centre company DataVita, claimed his company was receiving “one or two enquiries a week” about using his facility to store AI kit. Now, he is receiving hundreds.
He also discussed how a rack with AI processors and one with regular servers used different amounts of energy.
“A typical rack filled with regular equipment uses roughly 4kWh of power, or the same amount as a family home. In contrast, an AI kit rack would require roughly 20 times that amount of electricity, or 8kWh. And a single data centre may have hundreds or even thousands of these.”
Inside DataVita’s Fortis data centre in Scotland’s central belt
He added that Scotland’s colder and wetter climate provided a natural advantage in helping the data centres with keeping equipment cool, but it is still a huge task.
In its latest sustainability report, Microsoft, which is investing heavily in AI development, revealed that its water consumption had jumped by 34% between 2021 and 2022, to 6.4 million cubic metres, around the size of 2,500 Olympic swimming pools.
Prof Kate Crawford, who wrote a book about AI and its impact on the environment, said the issue kept her awake at night.
Speaking to the BBC in July, she said: “These energy-intensive systems take enormous amounts of electricity and energy, but also enormous amounts of water to cool these gigantic AI supercomputers. So we are really looking at an enormous extractive industry for the 21st Century.”
-
Tech Life: Charting the true cost of AI
But there are also hopes that AI could help solve some of the environmental challenges facing the planet.
AI tools are being used to try to reduce the number of vapour trails left across our skies by aeroplanes
Recently, Google and American Airlines discovered that by utilising an experimental AI tool to choose altitude, pilots may reduce the number of contrails (vapour trails) produced by aircraft by half. It is well recognised that contrails exacerbate global warming.
Attempts to replicate nuclear fusion, the process by which the Sun generates energy, are costing millions of dollars, and the US government is one of those involved.
A green power source that is endless would be a big game changer if this project were to succeed. Artificial intelligence (AI) has the potential to accelerate research, which has advanced extremely slowly since the 1960s.
Science professor Brian Spears claimed in February of this year that he had employed artificial intelligence (AI) to forecast the outcome of an experiment that produced a breakthrough.
“We generated ten petawatts of power for one hundred trillionth of a second. It was the solar system’s brightest object.
