Meta CEO Mark Zuckerberg has announced plans to invest hundreds of billions of dollars in building large-scale artificial intelligence (AI) data centres across the United States, as part of the company’s long-term push to lead in the development of “superintelligence”—a form of AI that could surpass human capabilities.
In a post shared on Threads, Meta’s own social media platform, Zuckerberg revealed that the first of these next-generation facilities, named Prometheus, is set to go live in 2026 in New Albany, Ohio. Another mega-site, dubbed Hyperion, will be built in Louisiana and is expected to become fully operational by 2030. Both sites will be part of what Meta is calling “multi-gigawatt clusters” to support its rapidly growing AI infrastructure.
“We’re building multiple more titan clusters as well,” Zuckerberg wrote. “Just one of these covers a significant part of the footprint of Manhattan,” referring to the vast size of the facilities—some projected to span nearly 59 square kilometers.
Meta, which earned more than $160 billion in revenue in 2024—largely from its core online advertising business—is shifting its focus toward becoming a dominant force in AI. The company’s AI ambitions include creating powerful systems that can perform at levels beyond the smartest humans, a goal it refers to as “superintelligence.”
According to Zuckerberg, these vast data centres will be central to achieving that vision. “We’re going to invest hundreds of billions of dollars to build superintelligence and make it open source,” he said.
The scale of the investment has drawn industry attention. 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,” noting that Meta’s AI teams will have access to some of the best computing hardware available globally.
Following the announcement, Meta’s shares rose 1%, with the stock up over 20% for the year so far, according to Reuters.
However, the environmental impact of such developments is drawing scrutiny. AI-driven data centres are known to be highly resource-intensive. Studies estimate that by 2027, these centres could consume up to 1.7 trillion gallons of water annually. Each AI query, such as those made through ChatGPT or similar platforms, can use about as much water as a small bottled drink.
There are currently over 10,000 data centres worldwide, with the U.S. home to the majority. Meta’s latest announcement suggests the race to dominate the AI frontier is not only intensifying but reshaping the digital and physical infrastructure of the tech industry.
