AI might account for nearly half of global datacentre power by end of 2025, expert claims

Alex de Vries-Gao, who founded Digiconomist, believes artificial intelligence (AI) could account for nearly half of global datacentre power by the end of 2025.

Artificial intelligence (AI) might account for nearly half of global datacentre power by the end of 2025, according to new analysis.

Alex de Vries-Gao, who founded Digiconomist – a platform dedicated to exposing the unintended consequences of digital trends – believes up to 49 per cent of datacentre energy could be represented by AI by the close of this year.

His claims come in a study entitled ‘Artificial intelligence: Supply chain constraints and energy implications’, published in energy journal Joule.

Alex’s research has discovered that AI-specific energy consumption may increase to 23 gigawatts (GW), which is twice the total power usage of the Netherlands.

He also admitted future research could be carried out to “examine where AI hardware production output is ultimately deployed”, which could be a “crucial” finding.

Alex writes: “Future research may also examine where AI hardware production output is ultimately deployed, as this is crucial for assessing the environmental impact of the electricity consumed by these devices.

“The characteristics of the relevant power grids will provide insights into the carbon and water intensity of the electricity generated to power AI hardware.

“A significant portion of this hardware may end up in the United States, as OpenAI has partnered with several others in a joint venture called Stargate to invest up to $500 billion over 4 years in new data center infrastructure across the country. There are early indications that these data centers could exacerbate dependence on fossil fuels: oil and gas company Crusoe has reportedly secured 4.5 GW of natural gas power capacity for AI data centers, with Stargate as one of its potential customers.

“However, while a growing reliance on fossil fuels threatens to undermine climate goals, effective policy responses first require urgent transparency.”