Data centre emissions are soaring – it’s AI or the #climate
"A single query to an AI-powered chatbot can consume ten times as much energy as a traditional Google search.
Data centres are thirsty as well as power-hungry: millions of litres of water have to be pumped to keep them cool.
These enormous server warehouses are vying with people for an increasing share of power and #water, a situation which could prove deadly during a heatwave or drought."
https://theconversation.com/data-centre-emissions-are-soaring-its-ai-or-the-climate-240137
#GenerativeAI and #ClimateChange Are on a Collision Course
"[AI's} very significant costs to the environment range from generating colossal amounts of energy to power the data centers that run tools such as ChatGPT and Midjourney to the millions of gallons of freshwater that are pumped through these data centers to make sure they don’t overheat and the tons of rare earth metals needed to build the hardware they contain.
What we really need is more transparency."
https://www.wired.com/story/true-cost-generative-ai-data-centers-energy/
Michael Liebreich on the #energy rush for #AI datacenters, with numbers.
The numbers are mind numbing.
"What the tech industry really wants is AI-training data centers between 1GW and 2GW.
For the next few years the majority of new AI data center capacity will be built in the US. Semi Analysis estimates that US data center power demand will surge 250% by 2030, absorbing nearly 15% of total power and requiring the building of 76GW of new dispatchable supply."
https://about.bnef.com/blog/liebreich-generative-ai-the-power-and-the-glory/
"AI’s growing electricity needs could exacerbate competition for limited renewable energy resources. This competition risks increasing reliance on fossil fuels, especially during periods of peak energy demand. If additional renewable capacity cannot be deployed quickly enough, the UK might face a scenario where AI-driven electricity demand increases overall emissions rather than reducing them."
"Electricity demand on the largest US grid is poised to rise so fast thanks to the #AI boom that it’ll be the equivalent of installing two New England networks in the coming decade.
PJM expects peak summer demand to jump by nearly 58 gigawatts, to about 210 gigawatts in 2035 from last year’s high, according to a report on Friday. That addition is more than double the peak demand record on the New England grid."
Not so fast.
How a top Chinese AI model overcame US sanctions
"The AI community is abuzz over DeepSeek R1, a new open-source reasoning model.
The model was developed by the Chinese AI startup #DeepSeek which claims that R1 matches or even surpasses OpenAI’s ChatGPT o1 on multiple key benchmarks but operates at a fraction of the cost."
https://www.technologyreview.com/2025/01/24/1110526/china-deepseek-top-ai-despite-sanctions/
"Fraction of the cost" = fraction of the energy use.
"Power companies that are most exposed to the tech sector’s data center boom plunged early Monday, as the debut of China’s DeepSeek open source AI laboratory led investors to question how much energy artificial intelligence applications will actually consume."