Meta to Pour Hundreds of Billions into AI Data Centres Across the U.S.
Meta is preparing to spend hundreds of billions of dollars on vast artificial intelligence data centres across the United States, founder Mark Zuckerberg announced, marking one of the tech industry’s largest infrastructure bets to date.
The company behind Facebook and Instagram is building multiple multi-gigawatt facilities, each capable of powering massive AI workloads. The first of these sites, named Prometheus, will be constructed in New Albany, Ohio, and is expected to go live in 2026.
Another centre, called Hyperion, is planned for Louisiana and could scale to five gigawatts over several years. It is expected to become fully operational by 2030.
Zuckerberg, posting on Threads, described the scope of the projects as unprecedented: “Just one of these covers a significant part of the footprint of Manhattan.” Manhattan spans about 59 square kilometres (22.8 square miles).
Meta’s AI ambitions are centred around what the company calls superintelligence – a concept referring to systems that surpass the cognitive capabilities of the smartest humans. To support this vision, the company says it will make a generational investment in infrastructure to train and deploy advanced AI models.
“We’re building multiple more titan clusters as well,” Zuckerberg said, adding that the data centres had been given “names befitting their scale and impact.”
The announcement drew swift attention from industry observers. Karl Freund, principal analyst at Cambrian AI Research, told the BBC that Zuckerberg “intends to spend his way to the top of the AI heap,” pointing to Meta’s aggressive hiring and hardware acquisition.
Meta, which made over $160 billion in revenue last year primarily through online advertising, is leaning heavily into AI as the next frontier of growth. The company’s shares climbed 1% after the news, continuing a strong performance that has seen Meta’s stock rise more than 20% in 2025.
The global boom in AI development has sparked a parallel race to expand computing infrastructure. There are currently around 10,000 data centres worldwide, with the majority based in the U.S., followed by the UK and Germany.
However, AI infrastructure comes with a hefty environmental footprint. Research suggests AI data centres could consume up to 1.7 trillion gallons of water annually by 2027. Even a single AI query may require as much water as a small bottled drink, raising sustainability concerns as investment in the sector surges.
Meta has not yet detailed the environmental mitigation strategies it plans to adopt for its new projects.
