Meta’s founder Mark Zuckerberg has stated the social media large will spend masses of billions of greenbacks on development massive AI information centres in the USA.
The primary multi-gigawatt information centre, known as Prometheus, is predicted to return on-line in 2026, Zuckerberg stated.
He stated probably the most websites would duvet a space just about the scale of Long island (59.1 sq km/22.8 sq miles).
Meta has invested closely in efforts to expand what it known as “superintelligence” – era that it stated may just out-think the neatest people.
The corporate, which has made maximum of its cash from web advertising, generated greater than $160bn in earnings in 2024.
In a put up on his social media platform, Threads, Zuckerberg stated Meta used to be development a number of multi-gigawatt clusters, and that one cluster, known as Hyperion, may just scale as much as 5 gigawatts over a number of years.
“We are development more than one extra titan clusters as smartly. Simply this kind of covers a vital a part of the footprint of Long island,” he added.
Prometheus will probably be inbuilt New Albany, Ohio, whilst Hyperion will probably be inbuilt Louisiana and is predicted to be absolutely on-line via 2030, Zuckerberg stated.
He stated Meta would “make investments masses of billions of greenbacks… to construct superintelligence” and that the centres were given “names befitting their scale and have an effect on”.
Karl Freund, predominant analyst at Cambrian AI Analysis, advised the BBC, “obviously, Zuckerberg intends to spend his technique to the highest of the AI heap”.
“The skill he’s hiring could have get entry to to one of the vital perfect AI {Hardware} on the earth,” Freund added.
Meta stocks had been buying and selling 1% upper following the announcement, Reuters information company reported. The inventory has risen greater than 20% thus far this 12 months.
There are no less than 10,000 information centres all over the world internet hosting the cloud –Â far flung servers that retailer virtual data – with maximum of them situated in the USA, adopted via the United Kingdom and Germany.
AI-driven information centres are extraordinarily power and water in depth. One find out about estimates that those centres may just eat 1.7 trillion gallons of water globally via 2027. A unmarried AI question – as an example, a request to ChatGPT – can use about as a lot water as a small bottle you would purchase from the nook store.
